Index: projects/cxl_iscsi/bin/rcp/Makefile =================================================================== --- projects/cxl_iscsi/bin/rcp/Makefile (revision 279716) +++ projects/cxl_iscsi/bin/rcp/Makefile (revision 279717) @@ -1,12 +1,11 @@ # @(#)Makefile 8.1 (Berkeley) 7/19/93 # $FreeBSD$ PROG= rcp SRCS= rcp.c util.c CFLAGS+=-DBINDIR=${BINDIR} BINOWN= root BINMODE=4555 -PRECIOUSPROG= .include Index: projects/cxl_iscsi/bin/sh/error.c =================================================================== --- projects/cxl_iscsi/bin/sh/error.c (revision 279716) +++ projects/cxl_iscsi/bin/sh/error.c (revision 279717) @@ -1,199 +1,199 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Kenneth Almquist. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef lint #if 0 static char sccsid[] = "@(#)error.c 8.2 (Berkeley) 5/4/95"; #endif #endif /* not lint */ #include __FBSDID("$FreeBSD$"); /* * Errors and exceptions. */ #include "shell.h" +#include "eval.h" #include "main.h" #include "options.h" #include "output.h" #include "error.h" #include "nodes.h" /* show.h needs nodes.h */ #include "show.h" #include "trap.h" #include #include #include #include /* * Code to handle exceptions in C. */ struct jmploc *handler; volatile sig_atomic_t exception; volatile sig_atomic_t suppressint; volatile sig_atomic_t intpending; -char *commandname; static void exverror(int, const char *, va_list) __printf0like(2, 0) __dead2; /* * Called to raise an exception. Since C doesn't include exceptions, we * just do a longjmp to the exception handler. The type of exception is * stored in the global variable "exception". * * Interrupts are disabled; they should be reenabled when the exception is * caught. */ void exraise(int e) { INTOFF; if (handler == NULL) abort(); exception = e; longjmp(handler->loc, 1); } /* * Called from trap.c when a SIGINT is received and not suppressed, or when * an interrupt is pending and interrupts are re-enabled using INTON. * (If the user specifies that SIGINT is to be trapped or ignored using the * trap builtin, then this routine is not called.) Suppressint is nonzero * when interrupts are held using the INTOFF macro. If SIGINTs are not * suppressed and the shell is not a root shell, then we want to be * terminated if we get here, as if we were terminated directly by a SIGINT. * Arrange for this here. */ void onint(void) { sigset_t sigs; intpending = 0; sigemptyset(&sigs); sigprocmask(SIG_SETMASK, &sigs, NULL); /* * This doesn't seem to be needed, since main() emits a newline. */ #if 0 if (tcgetpgrp(0) == getpid()) write(STDERR_FILENO, "\n", 1); #endif if (rootshell && iflag) exraise(EXINT); else { signal(SIGINT, SIG_DFL); kill(getpid(), SIGINT); _exit(128 + SIGINT); } } static void vwarning(const char *msg, va_list ap) { if (commandname) outfmt(out2, "%s: ", commandname); else if (arg0) outfmt(out2, "%s: ", arg0); doformat(out2, msg, ap); out2fmt_flush("\n"); } void warning(const char *msg, ...) { va_list ap; va_start(ap, msg); vwarning(msg, ap); va_end(ap); } /* * Exverror is called to raise the error exception. If the first argument * is not NULL then error prints an error message using printf style * formatting. It then raises the error exception. */ static void exverror(int cond, const char *msg, va_list ap) { /* * An interrupt trumps an error. Certain places catch error * exceptions or transform them to a plain nonzero exit code * in child processes, and if an error exception can be handled, * an interrupt can be handled as well. * * exraise() will disable interrupts for the exception handler. */ FORCEINTON; #ifdef DEBUG if (msg) TRACE(("exverror(%d, \"%s\") pid=%d\n", cond, msg, getpid())); else TRACE(("exverror(%d, NULL) pid=%d\n", cond, getpid())); #endif if (msg) vwarning(msg, ap); flushall(); exraise(cond); } void error(const char *msg, ...) { va_list ap; va_start(ap, msg); exverror(EXERROR, msg, ap); va_end(ap); } void exerror(int cond, const char *msg, ...) { va_list ap; va_start(ap, msg); exverror(cond, msg, ap); va_end(ap); } Index: projects/cxl_iscsi/bin/sh/mknodes.c =================================================================== --- projects/cxl_iscsi/bin/sh/mknodes.c (revision 279716) +++ projects/cxl_iscsi/bin/sh/mknodes.c (revision 279717) @@ -1,449 +1,449 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Kenneth Almquist. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #if 0 #ifndef lint static char const copyright[] = "@(#) Copyright (c) 1991, 1993\n\ The Regents of the University of California. All rights reserved.\n"; #endif /* not lint */ #ifndef lint static char sccsid[] = "@(#)mknodes.c 8.2 (Berkeley) 5/4/95"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); /* * This program reads the nodetypes file and nodes.c.pat file. It generates * the files nodes.h and nodes.c. */ #include #include #include #include #include #define MAXTYPES 50 /* max number of node types */ #define MAXFIELDS 20 /* max fields in a structure */ #define BUFLEN 100 /* size of character buffers */ /* field types */ #define T_NODE 1 /* union node *field */ #define T_NODELIST 2 /* struct nodelist *field */ #define T_STRING 3 #define T_INT 4 /* int field */ #define T_OTHER 5 /* other */ #define T_TEMP 6 /* don't copy this field */ struct field { /* a structure field */ char *name; /* name of field */ int type; /* type of field */ char *decl; /* declaration of field */ }; struct str { /* struct representing a node structure */ char *tag; /* structure tag */ int nfields; /* number of fields in the structure */ struct field field[MAXFIELDS]; /* the fields of the structure */ int done; /* set if fully parsed */ }; static int ntypes; /* number of node types */ static char *nodename[MAXTYPES]; /* names of the nodes */ static struct str *nodestr[MAXTYPES]; /* type of structure used by the node */ static int nstr; /* number of structures */ static struct str str[MAXTYPES]; /* the structures */ static struct str *curstr; /* current structure */ static FILE *infp; static char line[1024]; static int linno; static char *linep; static void parsenode(void); static void parsefield(void); static void output(char *); static void outsizes(FILE *); static void outfunc(FILE *, int); static void indent(int, FILE *); static int nextfield(char *); static void skipbl(void); static int readline(void); static void error(const char *, ...) __printf0like(1, 2) __dead2; static char *savestr(const char *); int main(int argc, char *argv[]) { if (argc != 3) error("usage: mknodes file"); infp = stdin; if ((infp = fopen(argv[1], "r")) == NULL) error("Can't open %s: %s", argv[1], strerror(errno)); while (readline()) { if (line[0] == ' ' || line[0] == '\t') parsefield(); else if (line[0] != '\0') parsenode(); } output(argv[2]); exit(0); } static void parsenode(void) { char name[BUFLEN]; char tag[BUFLEN]; struct str *sp; if (curstr && curstr->nfields > 0) curstr->done = 1; nextfield(name); if (! nextfield(tag)) error("Tag expected"); if (*linep != '\0') error("Garbage at end of line"); nodename[ntypes] = savestr(name); for (sp = str ; sp < str + nstr ; sp++) { if (strcmp(sp->tag, tag) == 0) break; } if (sp >= str + nstr) { sp->tag = savestr(tag); sp->nfields = 0; curstr = sp; nstr++; } nodestr[ntypes] = sp; ntypes++; } static void parsefield(void) { char name[BUFLEN]; char type[BUFLEN]; char decl[2 * BUFLEN]; struct field *fp; if (curstr == NULL || curstr->done) error("No current structure to add field to"); if (! nextfield(name)) error("No field name"); if (! nextfield(type)) error("No field type"); fp = &curstr->field[curstr->nfields]; fp->name = savestr(name); if (strcmp(type, "nodeptr") == 0) { fp->type = T_NODE; sprintf(decl, "union node *%s", name); } else if (strcmp(type, "nodelist") == 0) { fp->type = T_NODELIST; sprintf(decl, "struct nodelist *%s", name); } else if (strcmp(type, "string") == 0) { fp->type = T_STRING; sprintf(decl, "char *%s", name); } else if (strcmp(type, "int") == 0) { fp->type = T_INT; sprintf(decl, "int %s", name); } else if (strcmp(type, "other") == 0) { fp->type = T_OTHER; } else if (strcmp(type, "temp") == 0) { fp->type = T_TEMP; } else { error("Unknown type %s", type); } if (fp->type == T_OTHER || fp->type == T_TEMP) { skipbl(); fp->decl = savestr(linep); } else { if (*linep) error("Garbage at end of line"); fp->decl = savestr(decl); } curstr->nfields++; } -char writer[] = "\ +static const char writer[] = "\ /*\n\ * This file was generated by the mknodes program.\n\ */\n\ \n"; static void output(char *file) { FILE *hfile; FILE *cfile; FILE *patfile; int i; struct str *sp; struct field *fp; char *p; if ((patfile = fopen(file, "r")) == NULL) error("Can't open %s: %s", file, strerror(errno)); if ((hfile = fopen("nodes.h", "w")) == NULL) error("Can't create nodes.h: %s", strerror(errno)); if ((cfile = fopen("nodes.c", "w")) == NULL) error("Can't create nodes.c"); fputs(writer, hfile); for (i = 0 ; i < ntypes ; i++) fprintf(hfile, "#define %s %d\n", nodename[i], i); fputs("\n\n\n", hfile); for (sp = str ; sp < &str[nstr] ; sp++) { fprintf(hfile, "struct %s {\n", sp->tag); for (i = sp->nfields, fp = sp->field ; --i >= 0 ; fp++) { fprintf(hfile, " %s;\n", fp->decl); } fputs("};\n\n\n", hfile); } fputs("union node {\n", hfile); fprintf(hfile, " int type;\n"); for (sp = str ; sp < &str[nstr] ; sp++) { fprintf(hfile, " struct %s %s;\n", sp->tag, sp->tag); } fputs("};\n\n\n", hfile); fputs("struct nodelist {\n", hfile); fputs("\tstruct nodelist *next;\n", hfile); fputs("\tunion node *n;\n", hfile); fputs("};\n\n\n", hfile); fputs("struct funcdef;\n", hfile); fputs("struct funcdef *copyfunc(union node *);\n", hfile); fputs("union node *getfuncnode(struct funcdef *);\n", hfile); fputs("void reffunc(struct funcdef *);\n", hfile); fputs("void unreffunc(struct funcdef *);\n", hfile); fputs(writer, cfile); while (fgets(line, sizeof line, patfile) != NULL) { for (p = line ; *p == ' ' || *p == '\t' ; p++); if (strcmp(p, "%SIZES\n") == 0) outsizes(cfile); else if (strcmp(p, "%CALCSIZE\n") == 0) outfunc(cfile, 1); else if (strcmp(p, "%COPY\n") == 0) outfunc(cfile, 0); else fputs(line, cfile); } } static void outsizes(FILE *cfile) { int i; fprintf(cfile, "static const short nodesize[%d] = {\n", ntypes); for (i = 0 ; i < ntypes ; i++) { fprintf(cfile, " ALIGN(sizeof (struct %s)),\n", nodestr[i]->tag); } fprintf(cfile, "};\n"); } static void outfunc(FILE *cfile, int calcsize) { struct str *sp; struct field *fp; int i; fputs(" if (n == NULL)\n", cfile); if (calcsize) fputs(" return;\n", cfile); else fputs(" return NULL;\n", cfile); if (calcsize) fputs(" funcblocksize += nodesize[n->type];\n", cfile); else { fputs(" new = funcblock;\n", cfile); fputs(" funcblock = (char *)funcblock + nodesize[n->type];\n", cfile); } fputs(" switch (n->type) {\n", cfile); for (sp = str ; sp < &str[nstr] ; sp++) { for (i = 0 ; i < ntypes ; i++) { if (nodestr[i] == sp) fprintf(cfile, " case %s:\n", nodename[i]); } for (i = sp->nfields ; --i >= 1 ; ) { fp = &sp->field[i]; switch (fp->type) { case T_NODE: if (calcsize) { indent(12, cfile); fprintf(cfile, "calcsize(n->%s.%s);\n", sp->tag, fp->name); } else { indent(12, cfile); fprintf(cfile, "new->%s.%s = copynode(n->%s.%s);\n", sp->tag, fp->name, sp->tag, fp->name); } break; case T_NODELIST: if (calcsize) { indent(12, cfile); fprintf(cfile, "sizenodelist(n->%s.%s);\n", sp->tag, fp->name); } else { indent(12, cfile); fprintf(cfile, "new->%s.%s = copynodelist(n->%s.%s);\n", sp->tag, fp->name, sp->tag, fp->name); } break; case T_STRING: if (calcsize) { indent(12, cfile); fprintf(cfile, "funcstringsize += strlen(n->%s.%s) + 1;\n", sp->tag, fp->name); } else { indent(12, cfile); fprintf(cfile, "new->%s.%s = nodesavestr(n->%s.%s);\n", sp->tag, fp->name, sp->tag, fp->name); } break; case T_INT: case T_OTHER: if (! calcsize) { indent(12, cfile); fprintf(cfile, "new->%s.%s = n->%s.%s;\n", sp->tag, fp->name, sp->tag, fp->name); } break; } } indent(12, cfile); fputs("break;\n", cfile); } fputs(" };\n", cfile); if (! calcsize) fputs(" new->type = n->type;\n", cfile); } static void indent(int amount, FILE *fp) { while (amount >= 8) { putc('\t', fp); amount -= 8; } while (--amount >= 0) { putc(' ', fp); } } static int nextfield(char *buf) { char *p, *q; p = linep; while (*p == ' ' || *p == '\t') p++; q = buf; while (*p != ' ' && *p != '\t' && *p != '\0') *q++ = *p++; *q = '\0'; linep = p; return (q > buf); } static void skipbl(void) { while (*linep == ' ' || *linep == '\t') linep++; } static int readline(void) { char *p; if (fgets(line, 1024, infp) == NULL) return 0; for (p = line ; *p != '#' && *p != '\n' && *p != '\0' ; p++); while (p > line && (p[-1] == ' ' || p[-1] == '\t')) p--; *p = '\0'; linep = line; linno++; if (p - line > BUFLEN) error("Line too long"); return 1; } static void error(const char *msg, ...) { va_list va; va_start(va, msg); (void) fprintf(stderr, "line %d: ", linno); (void) vfprintf(stderr, msg, va); (void) fputc('\n', stderr); va_end(va); exit(2); } static char * savestr(const char *s) { char *p; if ((p = malloc(strlen(s) + 1)) == NULL) error("Out of space"); (void) strcpy(p, s); return p; } Index: projects/cxl_iscsi/bin/sh/mksyntax.c =================================================================== --- projects/cxl_iscsi/bin/sh/mksyntax.c (revision 279716) +++ projects/cxl_iscsi/bin/sh/mksyntax.c (revision 279717) @@ -1,329 +1,329 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Kenneth Almquist. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #if 0 #ifndef lint static char const copyright[] = "@(#) Copyright (c) 1991, 1993\n\ The Regents of the University of California. All rights reserved.\n"; #endif /* not lint */ #ifndef lint static char sccsid[] = "@(#)mksyntax.c 8.2 (Berkeley) 5/4/95"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); /* * This program creates syntax.h and syntax.c. */ #include #include #include #include "parser.h" struct synclass { const char *name; const char *comment; }; /* Syntax classes */ -struct synclass synclass[] = { +static const struct synclass synclass[] = { { "CWORD", "character is nothing special" }, { "CNL", "newline character" }, { "CBACK", "a backslash character" }, { "CSBACK", "a backslash character in single quotes" }, { "CSQUOTE", "single quote" }, { "CDQUOTE", "double quote" }, { "CENDQUOTE", "a terminating quote" }, { "CBQUOTE", "backwards single quote" }, { "CVAR", "a dollar sign" }, { "CENDVAR", "a '}' character" }, { "CLP", "a left paren in arithmetic" }, { "CRP", "a right paren in arithmetic" }, { "CEOF", "end of file" }, { "CCTL", "like CWORD, except it must be escaped" }, { "CSPCL", "these terminate a word" }, { "CIGN", "character should be ignored" }, { NULL, NULL } }; /* * Syntax classes for is_ functions. Warning: if you add new classes * you may have to change the definition of the is_in_name macro. */ -struct synclass is_entry[] = { +static const struct synclass is_entry[] = { { "ISDIGIT", "a digit" }, { "ISUPPER", "an upper case letter" }, { "ISLOWER", "a lower case letter" }, { "ISUNDER", "an underscore" }, { "ISSPECL", "the name of a special parameter" }, { NULL, NULL } }; -static char writer[] = "\ +static const char writer[] = "\ /*\n\ * This file was generated by the mksyntax program.\n\ */\n\ \n"; static FILE *cfile; static FILE *hfile; static void add_default(void); static void finish(void); static void init(const char *); static void add(const char *, const char *); static void output_type_macros(void); int main(int argc __unused, char **argv __unused) { int i; char buf[80]; int pos; /* Create output files */ if ((cfile = fopen("syntax.c", "w")) == NULL) { perror("syntax.c"); exit(2); } if ((hfile = fopen("syntax.h", "w")) == NULL) { perror("syntax.h"); exit(2); } fputs(writer, hfile); fputs(writer, cfile); fputs("#include \n", hfile); fputs("#include \n\n", hfile); /* Generate the #define statements in the header file */ fputs("/* Syntax classes */\n", hfile); for (i = 0 ; synclass[i].name ; i++) { sprintf(buf, "#define %s %d", synclass[i].name, i); fputs(buf, hfile); for (pos = strlen(buf) ; pos < 32 ; pos = (pos + 8) & ~07) putc('\t', hfile); fprintf(hfile, "/* %s */\n", synclass[i].comment); } putc('\n', hfile); fputs("/* Syntax classes for is_ functions */\n", hfile); for (i = 0 ; is_entry[i].name ; i++) { sprintf(buf, "#define %s %#o", is_entry[i].name, 1 << i); fputs(buf, hfile); for (pos = strlen(buf) ; pos < 32 ; pos = (pos + 8) & ~07) putc('\t', hfile); fprintf(hfile, "/* %s */\n", is_entry[i].comment); } putc('\n', hfile); fputs("#define SYNBASE (1 - CHAR_MIN)\n", hfile); fputs("#define PEOF -SYNBASE\n\n", hfile); putc('\n', hfile); fputs("#define BASESYNTAX (basesyntax + SYNBASE)\n", hfile); fputs("#define DQSYNTAX (dqsyntax + SYNBASE)\n", hfile); fputs("#define SQSYNTAX (sqsyntax + SYNBASE)\n", hfile); fputs("#define ARISYNTAX (arisyntax + SYNBASE)\n", hfile); putc('\n', hfile); output_type_macros(); /* is_digit, etc. */ putc('\n', hfile); /* Generate the syntax tables. */ fputs("#include \"parser.h\"\n", cfile); fputs("#include \"shell.h\"\n", cfile); fputs("#include \"syntax.h\"\n\n", cfile); fputs("/* syntax table used when not in quotes */\n", cfile); init("basesyntax"); add_default(); add("\n", "CNL"); add("\\", "CBACK"); add("'", "CSQUOTE"); add("\"", "CDQUOTE"); add("`", "CBQUOTE"); add("$", "CVAR"); add("}", "CENDVAR"); add("<>();&| \t", "CSPCL"); finish(); fputs("\n/* syntax table used when in double quotes */\n", cfile); init("dqsyntax"); add_default(); add("\n", "CNL"); add("\\", "CBACK"); add("\"", "CENDQUOTE"); add("`", "CBQUOTE"); add("$", "CVAR"); add("}", "CENDVAR"); /* ':/' for tilde expansion, '-^]' for [a\-x] pattern ranges */ add("!*?[]=~:/-^", "CCTL"); finish(); fputs("\n/* syntax table used when in single quotes */\n", cfile); init("sqsyntax"); add_default(); add("\n", "CNL"); add("\\", "CSBACK"); add("'", "CENDQUOTE"); /* ':/' for tilde expansion, '-^]' for [a\-x] pattern ranges */ add("!*?[]=~:/-^", "CCTL"); finish(); fputs("\n/* syntax table used when in arithmetic */\n", cfile); init("arisyntax"); add_default(); add("\n", "CNL"); add("\\", "CBACK"); add("`", "CBQUOTE"); add("\"", "CIGN"); add("$", "CVAR"); add("}", "CENDVAR"); add("(", "CLP"); add(")", "CRP"); finish(); fputs("\n/* character classification table */\n", cfile); init("is_type"); add("0123456789", "ISDIGIT"); add("abcdefghijklmnopqrstuvwxyz", "ISLOWER"); add("ABCDEFGHIJKLMNOPQRSTUVWXYZ", "ISUPPER"); add("_", "ISUNDER"); add("#?$!-*@", "ISSPECL"); finish(); exit(0); } /* * Output the header and declaration of a syntax table. */ static void init(const char *name) { fprintf(hfile, "extern const char %s[];\n", name); fprintf(cfile, "const char %s[SYNBASE + CHAR_MAX + 1] = {\n", name); } static void add_one(const char *key, const char *type) { fprintf(cfile, "\t[SYNBASE + %s] = %s,\n", key, type); } /* * Add default values to the syntax table. */ static void add_default(void) { add_one("PEOF", "CEOF"); add_one("CTLESC", "CCTL"); add_one("CTLVAR", "CCTL"); add_one("CTLENDVAR", "CCTL"); add_one("CTLBACKQ", "CCTL"); add_one("CTLBACKQ + CTLQUOTE", "CCTL"); add_one("CTLARI", "CCTL"); add_one("CTLENDARI", "CCTL"); add_one("CTLQUOTEMARK", "CCTL"); add_one("CTLQUOTEEND", "CCTL"); } /* * Output the footer of a syntax table. */ static void finish(void) { fputs("};\n", cfile); } /* * Add entries to the syntax table. */ static void add(const char *p, const char *type) { for (; *p; ++p) { char c = *p; switch (c) { case '\t': c = 't'; break; case '\n': c = 'n'; break; case '\'': c = '\''; break; case '\\': c = '\\'; break; default: fprintf(cfile, "\t[SYNBASE + '%c'] = %s,\n", c, type); continue; } fprintf(cfile, "\t[SYNBASE + '\\%c'] = %s,\n", c, type); } } /* * Output character classification macros (e.g. is_digit). If digits are * contiguous, we can test for them quickly. */ static const char *macro[] = { "#define is_digit(c)\t((unsigned int)((c) - '0') <= 9)", "#define is_eof(c)\t((c) == PEOF)", "#define is_alpha(c)\t((is_type+SYNBASE)[(int)c] & (ISUPPER|ISLOWER))", "#define is_name(c)\t((is_type+SYNBASE)[(int)c] & (ISUPPER|ISLOWER|ISUNDER))", "#define is_in_name(c)\t((is_type+SYNBASE)[(int)c] & (ISUPPER|ISLOWER|ISUNDER|ISDIGIT))", "#define is_special(c)\t((is_type+SYNBASE)[(int)c] & (ISSPECL|ISDIGIT))", "#define digit_val(c)\t((c) - '0')", NULL }; static void output_type_macros(void) { const char **pp; for (pp = macro ; *pp ; pp++) fprintf(hfile, "%s\n", *pp); } Index: projects/cxl_iscsi/bin/sh/var.c =================================================================== --- projects/cxl_iscsi/bin/sh/var.c (revision 279716) +++ projects/cxl_iscsi/bin/sh/var.c (revision 279717) @@ -1,955 +1,956 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Kenneth Almquist. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef lint #if 0 static char sccsid[] = "@(#)var.c 8.3 (Berkeley) 5/4/95"; #endif #endif /* not lint */ #include __FBSDID("$FreeBSD$"); #include #include #include /* * Shell variables. */ #include #include #include "shell.h" #include "output.h" #include "expand.h" #include "nodes.h" /* for other headers */ #include "eval.h" /* defines cmdenviron */ #include "exec.h" #include "syntax.h" #include "options.h" #include "mail.h" #include "var.h" #include "memalloc.h" #include "error.h" #include "mystring.h" #include "parser.h" #include "builtins.h" #ifndef NO_HISTORY #include "myhistedit.h" #endif #define VTABSIZE 39 struct varinit { struct var *var; int flags; const char *text; void (*func)(const char *); }; #ifndef NO_HISTORY struct var vhistsize; struct var vterm; #endif struct var vifs; struct var vmail; struct var vmpath; struct var vpath; struct var vps1; struct var vps2; struct var vps4; static struct var voptind; struct var vdisvfork; +struct localvar *localvars; int forcelocal; static const struct varinit varinit[] = { #ifndef NO_HISTORY { &vhistsize, VUNSET, "HISTSIZE=", sethistsize }, #endif { &vifs, 0, "IFS= \t\n", NULL }, { &vmail, VUNSET, "MAIL=", NULL }, { &vmpath, VUNSET, "MAILPATH=", NULL }, { &vpath, 0, "PATH=" _PATH_DEFPATH, changepath }, /* * vps1 depends on uid */ { &vps2, 0, "PS2=> ", NULL }, { &vps4, 0, "PS4=+ ", NULL }, #ifndef NO_HISTORY { &vterm, VUNSET, "TERM=", setterm }, #endif { &voptind, 0, "OPTIND=1", getoptsreset }, { &vdisvfork, VUNSET, "SH_DISABLE_VFORK=", NULL }, { NULL, 0, NULL, NULL } }; static struct var *vartab[VTABSIZE]; static const char *const locale_names[7] = { "LC_COLLATE", "LC_CTYPE", "LC_MONETARY", "LC_NUMERIC", "LC_TIME", "LC_MESSAGES", NULL }; static const int locale_categories[7] = { LC_COLLATE, LC_CTYPE, LC_MONETARY, LC_NUMERIC, LC_TIME, LC_MESSAGES, 0 }; static int varequal(const char *, const char *); static struct var *find_var(const char *, struct var ***, int *); static int localevar(const char *); static void setvareq_const(const char *s, int flags); extern char **environ; /* * This routine initializes the builtin variables and imports the environment. * It is called when the shell is initialized. */ void initvar(void) { char ppid[20]; const struct varinit *ip; struct var *vp; struct var **vpp; char **envp; for (ip = varinit ; (vp = ip->var) != NULL ; ip++) { if (find_var(ip->text, &vpp, &vp->name_len) != NULL) continue; vp->next = *vpp; *vpp = vp; vp->text = __DECONST(char *, ip->text); vp->flags = ip->flags | VSTRFIXED | VTEXTFIXED; vp->func = ip->func; } /* * PS1 depends on uid */ if (find_var("PS1", &vpp, &vps1.name_len) == NULL) { vps1.next = *vpp; *vpp = &vps1; vps1.text = __DECONST(char *, geteuid() ? "PS1=$ " : "PS1=# "); vps1.flags = VSTRFIXED|VTEXTFIXED; } fmtstr(ppid, sizeof(ppid), "%d", (int)getppid()); setvarsafe("PPID", ppid, 0); for (envp = environ ; *envp ; envp++) { if (strchr(*envp, '=')) { setvareq(*envp, VEXPORT|VTEXTFIXED); } } setvareq_const("OPTIND=1", 0); } /* * Safe version of setvar, returns 1 on success 0 on failure. */ int setvarsafe(const char *name, const char *val, int flags) { struct jmploc jmploc; struct jmploc *const savehandler = handler; int err = 0; int inton; inton = is_int_on(); if (setjmp(jmploc.loc)) err = 1; else { handler = &jmploc; setvar(name, val, flags); } handler = savehandler; SETINTON(inton); return err; } /* * Set the value of a variable. The flags argument is stored with the * flags of the variable. If val is NULL, the variable is unset. */ void setvar(const char *name, const char *val, int flags) { const char *p; size_t len; size_t namelen; size_t vallen; char *nameeq; int isbad; isbad = 0; p = name; if (! is_name(*p)) isbad = 1; p++; for (;;) { if (! is_in_name(*p)) { if (*p == '\0' || *p == '=') break; isbad = 1; } p++; } namelen = p - name; if (isbad) error("%.*s: bad variable name", (int)namelen, name); len = namelen + 2; /* 2 is space for '=' and '\0' */ if (val == NULL) { flags |= VUNSET; vallen = 0; } else { vallen = strlen(val); len += vallen; } INTOFF; nameeq = ckmalloc(len); memcpy(nameeq, name, namelen); nameeq[namelen] = '='; if (val) memcpy(nameeq + namelen + 1, val, vallen + 1); else nameeq[namelen + 1] = '\0'; setvareq(nameeq, flags); INTON; } static int localevar(const char *s) { const char *const *ss; if (*s != 'L') return 0; if (varequal(s + 1, "ANG")) return 1; if (strncmp(s + 1, "C_", 2) != 0) return 0; if (varequal(s + 3, "ALL")) return 1; for (ss = locale_names; *ss ; ss++) if (varequal(s + 3, *ss + 3)) return 1; return 0; } /* * Sets/unsets an environment variable from a pointer that may actually be a * pointer into environ where the string should not be manipulated. */ static void change_env(const char *s, int set) { char *eqp; char *ss; INTOFF; ss = savestr(s); if ((eqp = strchr(ss, '=')) != NULL) *eqp = '\0'; if (set && eqp != NULL) (void) setenv(ss, eqp + 1, 1); else (void) unsetenv(ss); ckfree(ss); INTON; return; } /* * Same as setvar except that the variable and value are passed in * the first argument as name=value. Since the first argument will * be actually stored in the table, it should not be a string that * will go away. */ void setvareq(char *s, int flags) { struct var *vp, **vpp; int nlen; if (aflag) flags |= VEXPORT; if (forcelocal && !(flags & (VNOSET | VNOLOCAL))) mklocal(s); vp = find_var(s, &vpp, &nlen); if (vp != NULL) { if (vp->flags & VREADONLY) { if ((flags & (VTEXTFIXED|VSTACK)) == 0) ckfree(s); error("%.*s: is read only", vp->name_len, s); } if (flags & VNOSET) { if ((flags & (VTEXTFIXED|VSTACK)) == 0) ckfree(s); return; } INTOFF; if (vp->func && (flags & VNOFUNC) == 0) (*vp->func)(s + vp->name_len + 1); if ((vp->flags & (VTEXTFIXED|VSTACK)) == 0) ckfree(vp->text); vp->flags &= ~(VTEXTFIXED|VSTACK|VUNSET); vp->flags |= flags; vp->text = s; /* * We could roll this to a function, to handle it as * a regular variable function callback, but why bother? * * Note: this assumes iflag is not set to 1 initially. * As part of initvar(), this is called before arguments * are looked at. */ if ((vp == &vmpath || (vp == &vmail && ! mpathset())) && iflag == 1) chkmail(1); if ((vp->flags & VEXPORT) && localevar(s)) { change_env(s, 1); (void) setlocale(LC_ALL, ""); updatecharset(); } INTON; return; } /* not found */ if (flags & VNOSET) { if ((flags & (VTEXTFIXED|VSTACK)) == 0) ckfree(s); return; } INTOFF; vp = ckmalloc(sizeof (*vp)); vp->flags = flags; vp->text = s; vp->name_len = nlen; vp->next = *vpp; vp->func = NULL; *vpp = vp; if ((vp->flags & VEXPORT) && localevar(s)) { change_env(s, 1); (void) setlocale(LC_ALL, ""); updatecharset(); } INTON; } static void setvareq_const(const char *s, int flags) { setvareq(__DECONST(char *, s), flags | VTEXTFIXED); } /* * Process a linked list of variable assignments. */ void listsetvar(struct strlist *list, int flags) { struct strlist *lp; INTOFF; for (lp = list ; lp ; lp = lp->next) { setvareq(savestr(lp->text), flags); } INTON; } /* * Find the value of a variable. Returns NULL if not set. */ char * lookupvar(const char *name) { struct var *v; v = find_var(name, NULL, NULL); if (v == NULL || v->flags & VUNSET) return NULL; return v->text + v->name_len + 1; } /* * Search the environment of a builtin command. If the second argument * is nonzero, return the value of a variable even if it hasn't been * exported. */ char * bltinlookup(const char *name, int doall) { struct strlist *sp; struct var *v; char *result; result = NULL; for (sp = cmdenviron ; sp ; sp = sp->next) { if (varequal(sp->text, name)) result = strchr(sp->text, '=') + 1; } if (result != NULL) return result; v = find_var(name, NULL, NULL); if (v == NULL || v->flags & VUNSET || (!doall && (v->flags & VEXPORT) == 0)) return NULL; return v->text + v->name_len + 1; } /* * Set up locale for a builtin (LANG/LC_* assignments). */ void bltinsetlocale(void) { struct strlist *lp; int act = 0; char *loc, *locdef; int i; for (lp = cmdenviron ; lp ; lp = lp->next) { if (localevar(lp->text)) { act = 1; break; } } if (!act) return; loc = bltinlookup("LC_ALL", 0); INTOFF; if (loc != NULL) { setlocale(LC_ALL, loc); INTON; updatecharset(); return; } locdef = bltinlookup("LANG", 0); for (i = 0; locale_names[i] != NULL; i++) { loc = bltinlookup(locale_names[i], 0); if (loc == NULL) loc = locdef; if (loc != NULL) setlocale(locale_categories[i], loc); } INTON; updatecharset(); } /* * Undo the effect of bltinlocaleset(). */ void bltinunsetlocale(void) { struct strlist *lp; INTOFF; for (lp = cmdenviron ; lp ; lp = lp->next) { if (localevar(lp->text)) { setlocale(LC_ALL, ""); updatecharset(); return; } } INTON; } /* * Update the localeisutf8 flag. */ void updatecharset(void) { char *charset; charset = nl_langinfo(CODESET); localeisutf8 = !strcmp(charset, "UTF-8"); } void initcharset(void) { updatecharset(); initial_localeisutf8 = localeisutf8; } /* * Generate a list of exported variables. This routine is used to construct * the third argument to execve when executing a program. */ char ** environment(void) { int nenv; struct var **vpp; struct var *vp; char **env, **ep; nenv = 0; for (vpp = vartab ; vpp < vartab + VTABSIZE ; vpp++) { for (vp = *vpp ; vp ; vp = vp->next) if (vp->flags & VEXPORT) nenv++; } ep = env = stalloc((nenv + 1) * sizeof *env); for (vpp = vartab ; vpp < vartab + VTABSIZE ; vpp++) { for (vp = *vpp ; vp ; vp = vp->next) if (vp->flags & VEXPORT) *ep++ = vp->text; } *ep = NULL; return env; } static int var_compare(const void *a, const void *b) { const char *const *sa, *const *sb; sa = a; sb = b; /* * This compares two var=value strings which creates a different * order from what you would probably expect. POSIX is somewhat * ambiguous on what should be sorted exactly. */ return strcoll(*sa, *sb); } /* * Command to list all variables which are set. This is invoked from the * set command when it is called without any options or operands. */ int showvarscmd(int argc __unused, char **argv __unused) { struct var **vpp; struct var *vp; const char *s; const char **vars; int i, n; /* * POSIX requires us to sort the variables. */ n = 0; for (vpp = vartab; vpp < vartab + VTABSIZE; vpp++) { for (vp = *vpp; vp; vp = vp->next) { if (!(vp->flags & VUNSET)) n++; } } INTOFF; vars = ckmalloc(n * sizeof(*vars)); i = 0; for (vpp = vartab; vpp < vartab + VTABSIZE; vpp++) { for (vp = *vpp; vp; vp = vp->next) { if (!(vp->flags & VUNSET)) vars[i++] = vp->text; } } qsort(vars, n, sizeof(*vars), var_compare); for (i = 0; i < n; i++) { /* * Skip improper variable names so the output remains usable as * shell input. */ if (!isassignment(vars[i])) continue; s = strchr(vars[i], '='); s++; outbin(vars[i], s - vars[i], out1); out1qstr(s); out1c('\n'); } ckfree(vars); INTON; return 0; } /* * The export and readonly commands. */ int exportcmd(int argc __unused, char **argv) { struct var **vpp; struct var *vp; char **ap; char *name; char *p; char *cmdname; int ch, values; int flag = argv[0][0] == 'r'? VREADONLY : VEXPORT; cmdname = argv[0]; values = 0; while ((ch = nextopt("p")) != '\0') { switch (ch) { case 'p': values = 1; break; } } if (values && *argptr != NULL) error("-p requires no arguments"); if (*argptr != NULL) { for (ap = argptr; (name = *ap) != NULL; ap++) { if ((p = strchr(name, '=')) != NULL) { p++; } else { vp = find_var(name, NULL, NULL); if (vp != NULL) { vp->flags |= flag; if ((vp->flags & VEXPORT) && localevar(vp->text)) { change_env(vp->text, 1); (void) setlocale(LC_ALL, ""); updatecharset(); } continue; } } setvar(name, p, flag); } } else { for (vpp = vartab ; vpp < vartab + VTABSIZE ; vpp++) { for (vp = *vpp ; vp ; vp = vp->next) { if (vp->flags & flag) { if (values) { /* * Skip improper variable names * so the output remains usable * as shell input. */ if (!isassignment(vp->text)) continue; out1str(cmdname); out1c(' '); } if (values && !(vp->flags & VUNSET)) { outbin(vp->text, vp->name_len + 1, out1); out1qstr(vp->text + vp->name_len + 1); } else outbin(vp->text, vp->name_len, out1); out1c('\n'); } } } } return 0; } /* * The "local" command. */ int localcmd(int argc __unused, char **argv __unused) { char *name; nextopt(""); if (! in_function()) error("Not in a function"); while ((name = *argptr++) != NULL) { mklocal(name); } return 0; } /* * Make a variable a local variable. When a variable is made local, it's * value and flags are saved in a localvar structure. The saved values * will be restored when the shell function returns. We handle the name * "-" as a special case. */ void mklocal(char *name) { struct localvar *lvp; struct var **vpp; struct var *vp; INTOFF; lvp = ckmalloc(sizeof (struct localvar)); if (name[0] == '-' && name[1] == '\0') { lvp->text = ckmalloc(sizeof optlist); memcpy(lvp->text, optlist, sizeof optlist); vp = NULL; } else { vp = find_var(name, &vpp, NULL); if (vp == NULL) { if (strchr(name, '=')) setvareq(savestr(name), VSTRFIXED | VNOLOCAL); else setvar(name, NULL, VSTRFIXED | VNOLOCAL); vp = *vpp; /* the new variable */ lvp->text = NULL; lvp->flags = VUNSET; } else { lvp->text = vp->text; lvp->flags = vp->flags; vp->flags |= VSTRFIXED|VTEXTFIXED; if (name[vp->name_len] == '=') setvareq(savestr(name), VNOLOCAL); } } lvp->vp = vp; lvp->next = localvars; localvars = lvp; INTON; } /* * Called after a function returns. */ void poplocalvars(void) { struct localvar *lvp; struct var *vp; INTOFF; while ((lvp = localvars) != NULL) { localvars = lvp->next; vp = lvp->vp; if (vp == NULL) { /* $- saved */ memcpy(optlist, lvp->text, sizeof optlist); ckfree(lvp->text); optschanged(); } else if ((lvp->flags & (VUNSET|VSTRFIXED)) == VUNSET) { (void)unsetvar(vp->text); } else { if ((vp->flags & VTEXTFIXED) == 0) ckfree(vp->text); vp->flags = lvp->flags; vp->text = lvp->text; } ckfree(lvp); } INTON; } int setvarcmd(int argc, char **argv) { if (argc <= 2) return unsetcmd(argc, argv); else if (argc == 3) setvar(argv[1], argv[2], 0); else error("too many arguments"); return 0; } /* * The unset builtin command. */ int unsetcmd(int argc __unused, char **argv __unused) { char **ap; int i; int flg_func = 0; int flg_var = 0; int ret = 0; while ((i = nextopt("vf")) != '\0') { if (i == 'f') flg_func = 1; else flg_var = 1; } if (flg_func == 0 && flg_var == 0) flg_var = 1; INTOFF; for (ap = argptr; *ap ; ap++) { if (flg_func) ret |= unsetfunc(*ap); if (flg_var) ret |= unsetvar(*ap); } INTON; return ret; } /* * Unset the specified variable. * Called with interrupts off. */ int unsetvar(const char *s) { struct var **vpp; struct var *vp; vp = find_var(s, &vpp, NULL); if (vp == NULL) return (0); if (vp->flags & VREADONLY) return (1); if (vp->text[vp->name_len + 1] != '\0') setvar(s, "", 0); if ((vp->flags & VEXPORT) && localevar(vp->text)) { change_env(s, 0); setlocale(LC_ALL, ""); updatecharset(); } vp->flags &= ~VEXPORT; vp->flags |= VUNSET; if ((vp->flags & VSTRFIXED) == 0) { if ((vp->flags & VTEXTFIXED) == 0) ckfree(vp->text); *vpp = vp->next; ckfree(vp); } return (0); } /* * Returns true if the two strings specify the same variable. The first * variable name is terminated by '='; the second may be terminated by * either '=' or '\0'. */ static int varequal(const char *p, const char *q) { while (*p == *q++) { if (*p++ == '=') return 1; } if (*p == '=' && *(q - 1) == '\0') return 1; return 0; } /* * Search for a variable. * 'name' may be terminated by '=' or a NUL. * vppp is set to the pointer to vp, or the list head if vp isn't found * lenp is set to the number of characters in 'name' */ static struct var * find_var(const char *name, struct var ***vppp, int *lenp) { unsigned int hashval; int len; struct var *vp, **vpp; const char *p = name; hashval = 0; while (*p && *p != '=') hashval = 2 * hashval + (unsigned char)*p++; len = p - name; if (lenp) *lenp = len; vpp = &vartab[hashval % VTABSIZE]; if (vppp) *vppp = vpp; for (vp = *vpp ; vp ; vpp = &vp->next, vp = *vpp) { if (vp->name_len != len) continue; if (memcmp(vp->text, name, len) != 0) continue; if (vppp) *vppp = vpp; return vp; } return NULL; } Index: projects/cxl_iscsi/bin/sh/var.h =================================================================== --- projects/cxl_iscsi/bin/sh/var.h (revision 279716) +++ projects/cxl_iscsi/bin/sh/var.h (revision 279717) @@ -1,130 +1,130 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Kenneth Almquist. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)var.h 8.2 (Berkeley) 5/4/95 * $FreeBSD$ */ /* * Shell variables. */ /* flags */ #define VEXPORT 0x01 /* variable is exported */ #define VREADONLY 0x02 /* variable cannot be modified */ #define VSTRFIXED 0x04 /* variable struct is statically allocated */ #define VTEXTFIXED 0x08 /* text is statically allocated */ #define VSTACK 0x10 /* text is allocated on the stack */ #define VUNSET 0x20 /* the variable is not set */ #define VNOFUNC 0x40 /* don't call the callback function */ #define VNOSET 0x80 /* do not set variable - just readonly test */ #define VNOLOCAL 0x100 /* ignore forcelocal */ struct var { struct var *next; /* next entry in hash list */ int flags; /* flags are defined above */ int name_len; /* length of name */ char *text; /* name=value */ void (*func)(const char *); /* function to be called when */ /* the variable gets set/unset */ }; struct localvar { struct localvar *next; /* next local variable in list */ struct var *vp; /* the variable that was made local */ int flags; /* saved flags */ char *text; /* saved text */ }; -struct localvar *localvars; +extern struct localvar *localvars; extern int forcelocal; extern struct var vifs; extern struct var vmail; extern struct var vmpath; extern struct var vpath; extern struct var vps1; extern struct var vps2; extern struct var vps4; extern struct var vdisvfork; #ifndef NO_HISTORY extern struct var vhistsize; extern struct var vterm; #endif extern int localeisutf8; /* The parser uses the locale that was in effect at startup. */ extern int initial_localeisutf8; /* * The following macros access the values of the above variables. * They have to skip over the name. They return the null string * for unset variables. */ #define ifsval() (vifs.text + 4) #define ifsset() ((vifs.flags & VUNSET) == 0) #define mailval() (vmail.text + 5) #define mpathval() (vmpath.text + 9) #define pathval() (vpath.text + 5) #define ps1val() (vps1.text + 4) #define ps2val() (vps2.text + 4) #define ps4val() (vps4.text + 4) #define optindval() (voptind.text + 7) #ifndef NO_HISTORY #define histsizeval() (vhistsize.text + 9) #define termval() (vterm.text + 5) #endif #define mpathset() ((vmpath.flags & VUNSET) == 0) #define disvforkset() ((vdisvfork.flags & VUNSET) == 0) void initvar(void); void setvar(const char *, const char *, int); void setvareq(char *, int); struct strlist; void listsetvar(struct strlist *, int); char *lookupvar(const char *); char *bltinlookup(const char *, int); void bltinsetlocale(void); void bltinunsetlocale(void); void updatecharset(void); void initcharset(void); char **environment(void); int showvarscmd(int, char **); void mklocal(char *); void poplocalvars(void); int unsetvar(const char *); int setvarsafe(const char *, const char *, int); Index: projects/cxl_iscsi/contrib/tzdata/antarctica =================================================================== --- projects/cxl_iscsi/contrib/tzdata/antarctica (revision 279716) +++ projects/cxl_iscsi/contrib/tzdata/antarctica (revision 279717) @@ -1,385 +1,386 @@ # This file is in the public domain, so clarified as of # 2009-05-17 by Arthur David Olson. # From Paul Eggert (1999-11-15): # To keep things manageable, we list only locations occupied year-round; see # COMNAP - Stations and Bases # http://www.comnap.aq/comnap/comnap.nsf/P/Stations/ # and # Summary of the Peri-Antarctic Islands (1998-07-23) # http://www.spri.cam.ac.uk/bob/periant.htm # for information. # Unless otherwise specified, we have no time zone information. # # Except for the French entries, # I made up all time zone abbreviations mentioned here; corrections welcome! # FORMAT is 'zzz' and GMTOFF is 0 for locations while uninhabited. # These rules are stolen from the 'southamerica' file. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule ArgAQ 1964 1966 - Mar 1 0:00 0 - Rule ArgAQ 1964 1966 - Oct 15 0:00 1:00 S Rule ArgAQ 1967 only - Apr 2 0:00 0 - Rule ArgAQ 1967 1968 - Oct Sun>=1 0:00 1:00 S Rule ArgAQ 1968 1969 - Apr Sun>=1 0:00 0 - Rule ArgAQ 1974 only - Jan 23 0:00 1:00 S Rule ArgAQ 1974 only - May 1 0:00 0 - Rule ChileAQ 1972 1986 - Mar Sun>=9 3:00u 0 - Rule ChileAQ 1974 1987 - Oct Sun>=9 4:00u 1:00 S Rule ChileAQ 1987 only - Apr 12 3:00u 0 - Rule ChileAQ 1988 1989 - Mar Sun>=9 3:00u 0 - Rule ChileAQ 1988 only - Oct Sun>=1 4:00u 1:00 S Rule ChileAQ 1989 only - Oct Sun>=9 4:00u 1:00 S Rule ChileAQ 1990 only - Mar 18 3:00u 0 - Rule ChileAQ 1990 only - Sep 16 4:00u 1:00 S Rule ChileAQ 1991 1996 - Mar Sun>=9 3:00u 0 - Rule ChileAQ 1991 1997 - Oct Sun>=9 4:00u 1:00 S Rule ChileAQ 1997 only - Mar 30 3:00u 0 - Rule ChileAQ 1998 only - Mar Sun>=9 3:00u 0 - Rule ChileAQ 1998 only - Sep 27 4:00u 1:00 S Rule ChileAQ 1999 only - Apr 4 3:00u 0 - Rule ChileAQ 1999 2010 - Oct Sun>=9 4:00u 1:00 S Rule ChileAQ 2000 2007 - Mar Sun>=9 3:00u 0 - # N.B.: the end of March 29 in Chile is March 30 in Universal time, # which is used below in specifying the transition. Rule ChileAQ 2008 only - Mar 30 3:00u 0 - Rule ChileAQ 2009 only - Mar Sun>=9 3:00u 0 - Rule ChileAQ 2010 only - Apr Sun>=1 3:00u 0 - Rule ChileAQ 2011 only - May Sun>=2 3:00u 0 - Rule ChileAQ 2011 only - Aug Sun>=16 4:00u 1:00 S -Rule ChileAQ 2012 max - Apr Sun>=23 3:00u 0 - -Rule ChileAQ 2012 max - Sep Sun>=2 4:00u 1:00 S +Rule ChileAQ 2012 2015 - Apr Sun>=23 3:00u 0 - +Rule ChileAQ 2012 2014 - Sep Sun>=2 4:00u 1:00 S # Argentina - year-round bases # Belgrano II, Confin Coast, -770227-0343737, since 1972-02-05 # Carlini, Potter Cove, King George Island, -6414-0602320, since 1982-01 # Esperanza, Hope Bay, -6323-05659, since 1952-12-17 # Marambio, -6414-05637, since 1969-10-29 # Orcadas, Laurie I, -6016-04444, since 1904-02-22 # San Martín, Barry I, -6808-06706, since 1951-03-21 # (except 1960-03 / 1976-03-21) # Australia - territories # Heard Island, McDonald Islands (uninhabited) # previously sealers and scientific personnel wintered # Margaret Turner reports # http://web.archive.org/web/20021204222245/http://www.dstc.qut.edu.au/DST/marg/daylight.html # (1999-09-30) that they're UTC+5, with no DST; # presumably this is when they have visitors. # # year-round bases # Casey, Bailey Peninsula, -6617+11032, since 1969 # Davis, Vestfold Hills, -6835+07759, since 1957-01-13 # (except 1964-11 - 1969-02) # Mawson, Holme Bay, -6736+06253, since 1954-02-13 # From Steffen Thorsen (2009-03-11): # Three Australian stations in Antarctica have changed their time zone: # Casey moved from UTC+8 to UTC+11 # Davis moved from UTC+7 to UTC+5 # Mawson moved from UTC+6 to UTC+5 # The changes occurred on 2009-10-18 at 02:00 (local times). # # Government source: (Australian Antarctic Division) # http://www.aad.gov.au/default.asp?casid=37079 # # We have more background information here: # http://www.timeanddate.com/news/time/antarctica-new-times.html # From Steffen Thorsen (2010-03-10): # We got these changes from the Australian Antarctic Division: ... # # - Casey station reverted to its normal time of UTC+8 on 5 March 2010. # The change to UTC+11 is being considered as a regular summer thing but # has not been decided yet. # # - Davis station will revert to its normal time of UTC+7 at 10 March 2010 # 20:00 UTC. # # - Mawson station stays on UTC+5. # # Background: # http://www.timeanddate.com/news/time/antartica-time-changes-2010.html # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Antarctica/Casey 0 - zzz 1969 8:00 - AWST 2009 Oct 18 2:00 # Australian Western Std Time 11:00 - CAST 2010 Mar 5 2:00 # Casey Time 8:00 - AWST 2011 Oct 28 2:00 11:00 - CAST 2012 Feb 21 17:00u 8:00 - AWST Zone Antarctica/Davis 0 - zzz 1957 Jan 13 7:00 - DAVT 1964 Nov # Davis Time 0 - zzz 1969 Feb 7:00 - DAVT 2009 Oct 18 2:00 5:00 - DAVT 2010 Mar 10 20:00u 7:00 - DAVT 2011 Oct 28 2:00 5:00 - DAVT 2012 Feb 21 20:00u 7:00 - DAVT Zone Antarctica/Mawson 0 - zzz 1954 Feb 13 6:00 - MAWT 2009 Oct 18 2:00 # Mawson Time 5:00 - MAWT # References: # Casey Weather (1998-02-26) # http://www.antdiv.gov.au/aad/exop/sfo/casey/casey_aws.html # Davis Station, Antarctica (1998-02-26) # http://www.antdiv.gov.au/aad/exop/sfo/davis/video.html # Mawson Station, Antarctica (1998-02-25) # http://www.antdiv.gov.au/aad/exop/sfo/mawson/video.html # Belgium - year-round base # Princess Elisabeth, Queen Maud Land, -713412+0231200, since 2007 # Brazil - year-round base # Ferraz, King George Island, -6205+05824, since 1983/4 # Bulgaria - year-round base # St. Kliment Ohridski, Livingston Island, -623829-0602153, since 1988 # Chile - year-round bases and towns # Escudero, South Shetland Is, -621157-0585735, since 1994 # Frei Montalva, King George Island, -6214-05848, since 1969-03-07 # O'Higgins, Antarctic Peninsula, -6319-05704, since 1948-02 # Prat, -6230-05941 # Villa Las Estrellas (a town), around the Frei base, since 1984-04-09 # These locations have always used Santiago time; use TZ='America/Santiago'. # China - year-round bases # Great Wall, King George Island, -6213-05858, since 1985-02-20 # Zhongshan, Larsemann Hills, Prydz Bay, -6922+07623, since 1989-02-26 # France - year-round bases (also see "France & Italy") # # From Antoine Leca (1997-01-20): # Time data entries are from Nicole Pailleau at the IFRTP # (French Institute for Polar Research and Technology). # She confirms that French Southern Territories and Terre Adélie bases # don't observe daylight saving time, even if Terre Adélie supplies came # from Tasmania. # # French Southern Territories with year-round inhabitants # # Alfred Faure, Possession Island, Crozet Islands, -462551+0515152, since 1964; # sealing & whaling stations operated variously 1802/1911+; # see Indian/Reunion. # # Martin-de-Viviès, Amsterdam Island, -374105+0773155, since 1950 # Port-aux-Français, Kerguelen Islands, -492110+0701303, since 1951; # whaling & sealing station operated 1908/1914, 1920/1929, and 1951/1956 # # St Paul Island - near Amsterdam, uninhabited # fishing stations operated variously 1819/1931 # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Indian/Kerguelen 0 - zzz 1950 # Port-aux-Français 5:00 - TFT # ISO code TF Time # # year-round base in the main continent # Dumont d'Urville, Île des Pétrels, -6640+14001, since 1956-11 # (2005-12-05) # # Another base at Port-Martin, 50km east, began operation in 1947. # It was destroyed by fire on 1952-01-14. # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Antarctica/DumontDUrville 0 - zzz 1947 10:00 - PMT 1952 Jan 14 # Port-Martin Time 0 - zzz 1956 Nov 10:00 - DDUT # Dumont-d'Urville Time # France & Italy - year-round base # Concordia, -750600+1232000, since 2005 # Germany - year-round base # Neumayer III, -704080-0081602, since 2009 # India - year-round bases # Bharati, -692428+0761114, since 2012 # Maitri, -704558+0114356, since 1989 # Italy - year-round base (also see "France & Italy") # Zuchelli, Terra Nova Bay, -744140+1640647, since 1986 # Japan - year-round bases # Syowa (also known as Showa), -690022+0393524, since 1957 # # From Hideyuki Suzuki (1999-02-06): # In all Japanese stations, +0300 is used as the standard time. # # Syowa station, which is the first antarctic station of Japan, # was established on 1957-01-29. Since Syowa station is still the main # station of Japan, it's appropriate for the principal location. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Antarctica/Syowa 0 - zzz 1957 Jan 29 3:00 - SYOT # Syowa Time # See: # NIPR Antarctic Research Activities (1999-08-17) # http://www.nipr.ac.jp/english/ara01.html # S Korea - year-round base # Jang Bogo, Terra Nova Bay, -743700+1641205 since 2014 # King Sejong, King George Island, -6213-05847, since 1988 # New Zealand - claims # Balleny Islands (never inhabited) # Scott Island (never inhabited) # # year-round base # Scott Base, Ross Island, since 1957-01. # See Pacific/Auckland. # Norway - territories # Bouvet (never inhabited) # # claims # Peter I Island (never inhabited) # # year-round base # Troll, Queen Maud Land, -720041+0023206, since 2005-02-12 # # From Paul-Inge Flakstad (2014-03-10): # I recently had a long dialog about this with the developer of timegenie.com. # In the absence of specific dates, he decided to choose some likely ones: # GMT +1 - From March 1 to the last Sunday in March # GMT +2 - From the last Sunday in March until the last Sunday in October # GMT +1 - From the last Sunday in October until November 7 # GMT +0 - From November 7 until March 1 # The dates for switching to and from UTC+0 will probably not be absolutely # correct, but they should be quite close to the actual dates. # # From Paul Eggert (2014-03-21): # The CET-switching Troll rules require zic from tzcode 2014b or later, so as # suggested by Bengt-Inge Larsson comment them out for now, and approximate # with only UTC and CEST. Uncomment them when 2014b is more prevalent. # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S #Rule Troll 2005 max - Mar 1 1:00u 1:00 CET Rule Troll 2005 max - Mar lastSun 1:00u 2:00 CEST #Rule Troll 2005 max - Oct lastSun 1:00u 1:00 CET #Rule Troll 2004 max - Nov 7 1:00u 0:00 UTC # Remove the following line when uncommenting the above '#Rule' lines. Rule Troll 2004 max - Oct lastSun 1:00u 0:00 UTC # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Antarctica/Troll 0 - zzz 2005 Feb 12 0:00 Troll %s # Poland - year-round base # Arctowski, King George Island, -620945-0582745, since 1977 # Romania - year-bound base # Law-Racoviță, Larsemann Hills, -692319+0762251, since 1986 # Russia - year-round bases # Bellingshausen, King George Island, -621159-0585337, since 1968-02-22 # Mirny, Davis coast, -6633+09301, since 1956-02 # Molodezhnaya, Alasheyev Bay, -6740+04551, # year-round from 1962-02 to 1999-07-01 # Novolazarevskaya, Queen Maud Land, -7046+01150, # year-round from 1960/61 to 1992 # Vostok, since 1957-12-16, temporarily closed 1994-02/1994-11 # From Craig Mundell (1994-12-15): # http://quest.arc.nasa.gov/antarctica/QA/computers/Directions,Time,ZIP # Vostok, which is one of the Russian stations, is set on the same # time as Moscow, Russia. # # From Lee Hotz (2001-03-08): # I queried the folks at Columbia who spent the summer at Vostok and this is # what they had to say about time there: # "in the US Camp (East Camp) we have been on New Zealand (McMurdo) # time, which is 12 hours ahead of GMT. The Russian Station Vostok was # 6 hours behind that (although only 2 miles away, i.e. 6 hours ahead # of GMT). This is a time zone I think two hours east of Moscow. The # natural time zone is in between the two: 8 hours ahead of GMT." # # From Paul Eggert (2001-05-04): # This seems to be hopelessly confusing, so I asked Lee Hotz about it # in person. He said that some Antarctic locations set their local # time so that noon is the warmest part of the day, and that this # changes during the year and does not necessarily correspond to mean # solar noon. So the Vostok time might have been whatever the clocks # happened to be during their visit. So we still don't really know what time # it is at Vostok. But we'll guess UTC+6. # Zone Antarctica/Vostok 0 - zzz 1957 Dec 16 6:00 - VOST # Vostok time # S Africa - year-round bases # Marion Island, -4653+03752 # SANAE IV, Vesleskarvet, Queen Maud Land, -714022-0025026, since 1997 # Ukraine - year-round base # Vernadsky (formerly Faraday), Galindez Island, -651445-0641526, since 1954 # United Kingdom # # British Antarctic Territories (BAT) claims # South Orkney Islands # scientific station from 1903 # whaling station at Signy I 1920/1926 # South Shetland Islands # # year-round bases # Bird Island, South Georgia, -5400-03803, since 1983 # Deception Island, -6259-06034, whaling station 1912/1931, # scientific station 1943/1967, # previously sealers and a scientific expedition wintered by accident, # and a garrison was deployed briefly # Halley, Coates Land, -7535-02604, since 1956-01-06 # Halley is on a moving ice shelf and is periodically relocated # so that it is never more than 10km from its nominal location. # Rothera, Adelaide Island, -6734-6808, since 1976-12-01 # # From Paul Eggert (2002-10-22) # says Rothera is -03 all year. # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Antarctica/Rothera 0 - zzz 1976 Dec 1 -3:00 - ROTT # Rothera time # Uruguay - year round base # Artigas, King George Island, -621104-0585107 # USA - year-round bases # # Palmer, Anvers Island, since 1965 (moved 2 miles in 1968) # # From Ethan Dicks (1996-10-06): # It keeps the same time as Punta Arenas, Chile, because, just like us # and the South Pole, that's the other end of their supply line.... # I verified with someone who was there that since 1980, # Palmer has followed Chile. Prior to that, before the Falklands War, # Palmer used to be supplied from Argentina. # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Antarctica/Palmer 0 - zzz 1965 - -4:00 ArgAQ AR%sT 1969 Oct 5 + -4:00 ArgAQ AR%sT 1969 Oct 5 -3:00 ArgAQ AR%sT 1982 May - -4:00 ChileAQ CL%sT + -4:00 ChileAQ CL%sT 2015 Apr 26 3:00u + -3:00 - CLT # # # McMurdo Station, Ross Island, since 1955-12 # Amundsen-Scott South Pole Station, continuously occupied since 1956-11-20 # # From Chris Carrier (1996-06-27): # Siple, the first commander of the South Pole station, # stated that he would have liked to have kept GMT at the station, # but that he found it more convenient to keep GMT+12 # as supplies for the station were coming from McMurdo Sound, # which was on GMT+12 because New Zealand was on GMT+12 all year # at that time (1957). (Source: Siple's book 90 Degrees South.) # # From Susan Smith # http://www.cybertours.com/whs/pole10.html # (1995-11-13 16:24:56 +1300, no longer available): # We use the same time as McMurdo does. # And they use the same time as Christchurch, NZ does.... # One last quirk about South Pole time. # All the electric clocks are usually wrong. # Something about the generators running at 60.1hertz or something # makes all of the clocks run fast. So every couple of days, # we have to go around and set them back 5 minutes or so. # Maybe if we let them run fast all of the time, we'd get to leave here sooner!! # # See 'australasia' for Antarctica/McMurdo. Index: projects/cxl_iscsi/contrib/tzdata/asia =================================================================== --- projects/cxl_iscsi/contrib/tzdata/asia (revision 279716) +++ projects/cxl_iscsi/contrib/tzdata/asia (revision 279717) @@ -1,2860 +1,2848 @@ # This file is in the public domain, so clarified as of # 2009-05-17 by Arthur David Olson. # This file is by no means authoritative; if you think you know better, # go ahead and edit the file (and please send any changes to # tz@iana.org for general use in the future). For more, please see # the file CONTRIBUTING in the tz distribution. # From Paul Eggert (2014-10-31): # # Unless otherwise specified, the source for data through 1990 is: # Thomas G. Shanks and Rique Pottenger, The International Atlas (6th edition), # San Diego: ACS Publications, Inc. (2003). # Unfortunately this book contains many errors and cites no sources. # # Gwillim Law writes that a good source # for recent time zone data is the International Air Transport # Association's Standard Schedules Information Manual (IATA SSIM), # published semiannually. Law sent in several helpful summaries # of the IATA's data after 1990. Except where otherwise noted, # IATA SSIM is the source for entries after 1990. # # Another source occasionally used is Edward W. Whitman, World Time Differences, # Whitman Publishing Co, 2 Niagara Av, Ealing, London (undated), which # I found in the UCLA library. # # For data circa 1899, a common source is: # Milne J. Civil time. Geogr J. 1899 Feb;13(2):173-94. # http://www.jstor.org/stable/1774359 # # For Russian data circa 1919, a source is: # Byalokoz EL. New Counting of Time in Russia since July 1, 1919. # (See the 'europe' file for a fuller citation.) # # A reliable and entertaining source about time zones is # Derek Howse, Greenwich time and longitude, Philip Wilson Publishers (1997). # # I invented the abbreviations marked '*' in the following table; # the rest are from earlier versions of this file, or from other sources. # Corrections are welcome! # std dst # LMT Local Mean Time # 2:00 EET EEST Eastern European Time # 2:00 IST IDT Israel # 3:00 AST ADT Arabia* # 3:30 IRST IRDT Iran # 4:00 GST Gulf* # 5:30 IST India # 7:00 ICT Indochina, most times and locations* # 7:00 WIB west Indonesia (Waktu Indonesia Barat) # 8:00 WITA central Indonesia (Waktu Indonesia Tengah) # 8:00 CST China # 8:00 IDT Indochina, 1943-45, 1947-55, 1960-75 (some locations)* # 8:00 JWST Western Standard Time (Japan, 1896/1937)* # 9:00 JCST Central Standard Time (Japan, 1896/1937) # 9:00 WIT east Indonesia (Waktu Indonesia Timur) # 9:00 JST JDT Japan # 9:00 KST KDT Korea # 9:30 ACST Australian Central Standard Time # # See the 'europe' file for Russia and Turkey in Asia. # From Guy Harris: # Incorporates data for Singapore from Robert Elz' asia 1.1, as well as # additional information from Tom Yap, Sun Microsystems Intercontinental # Technical Support (including a page from the Official Airline Guide - # Worldwide Edition). The names for time zones are guesses. ############################################################################### # These rules are stolen from the 'europe' file. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule EUAsia 1981 max - Mar lastSun 1:00u 1:00 S Rule EUAsia 1979 1995 - Sep lastSun 1:00u 0 - Rule EUAsia 1996 max - Oct lastSun 1:00u 0 - Rule E-EurAsia 1981 max - Mar lastSun 0:00 1:00 S Rule E-EurAsia 1979 1995 - Sep lastSun 0:00 0 - Rule E-EurAsia 1996 max - Oct lastSun 0:00 0 - Rule RussiaAsia 1981 1984 - Apr 1 0:00 1:00 S Rule RussiaAsia 1981 1983 - Oct 1 0:00 0 - Rule RussiaAsia 1984 1991 - Sep lastSun 2:00s 0 - Rule RussiaAsia 1985 1991 - Mar lastSun 2:00s 1:00 S Rule RussiaAsia 1992 only - Mar lastSat 23:00 1:00 S Rule RussiaAsia 1992 only - Sep lastSat 23:00 0 - Rule RussiaAsia 1993 max - Mar lastSun 2:00s 1:00 S Rule RussiaAsia 1993 1995 - Sep lastSun 2:00s 0 - Rule RussiaAsia 1996 max - Oct lastSun 2:00s 0 - # Afghanistan # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Kabul 4:36:48 - LMT 1890 4:00 - AFT 1945 4:30 - AFT # Armenia # From Paul Eggert (2006-03-22): # Shanks & Pottenger have Yerevan switching to 3:00 (with Russian DST) # in spring 1991, then to 4:00 with no DST in fall 1995, then # readopting Russian DST in 1997. Go with Shanks & Pottenger, even # when they disagree with others. Edgar Der-Danieliantz # reported (1996-05-04) that Yerevan probably wouldn't use DST # in 1996, though it did use DST in 1995. IATA SSIM (1991/1998) reports that # Armenia switched from 3:00 to 4:00 in 1998 and observed DST after 1991, # but started switching at 3:00s in 1998. # From Arthur David Olson (2011-06-15): # While Russia abandoned DST in 2011, Armenia may choose to # follow Russia's "old" rules. # From Alexander Krivenyshev (2012-02-10): # According to News Armenia, on Feb 9, 2012, # http://newsarmenia.ru/society/20120209/42609695.html # # The Armenia National Assembly adopted final reading of Amendments to the # Law "On procedure of calculation time on the territory of the Republic of # Armenia" according to which Armenia [is] abolishing Daylight Saving Time. # or # (brief) # http://www.worldtimezone.com/dst_news/dst_news_armenia03.html # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Yerevan 2:58:00 - LMT 1924 May 2 3:00 - YERT 1957 Mar # Yerevan Time 4:00 RussiaAsia YER%sT 1991 Mar 31 2:00s 3:00 1:00 YERST 1991 Sep 23 # independence 3:00 RussiaAsia AM%sT 1995 Sep 24 2:00s 4:00 - AMT 1997 4:00 RussiaAsia AM%sT 2012 Mar 25 2:00s 4:00 - AMT # Azerbaijan # From Rustam Aliyev of the Azerbaijan Internet Forum (2005-10-23): # According to the resolution of Cabinet of Ministers, 1997 # Resolution available at: http://aif.az/docs/daylight_res.pdf # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Azer 1997 max - Mar lastSun 4:00 1:00 S Rule Azer 1997 max - Oct lastSun 5:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Baku 3:19:24 - LMT 1924 May 2 3:00 - BAKT 1957 Mar # Baku Time 4:00 RussiaAsia BAK%sT 1991 Mar 31 2:00s 3:00 1:00 BAKST 1991 Aug 30 # independence 3:00 RussiaAsia AZ%sT 1992 Sep lastSat 23:00 4:00 - AZT 1996 # Azerbaijan Time 4:00 EUAsia AZ%sT 1997 4:00 Azer AZ%sT # Bahrain -# Zone NAME GMTOFF RULES FORMAT [UNTIL] -Zone Asia/Bahrain 3:22:20 - LMT 1920 # Manamah - 4:00 - GST 1972 Jun - 3:00 - AST +# See Asia/Qatar. # Bangladesh # From Alexander Krivenyshev (2009-05-13): # According to newspaper Asian Tribune (May 6, 2009) Bangladesh may introduce # Daylight Saving Time from June 16 to Sept 30 # # Bangladesh to introduce daylight saving time likely from June 16 # http://www.asiantribune.com/?q=node/17288 # http://www.worldtimezone.com/dst_news/dst_news_bangladesh02.html # # "... Bangladesh government has decided to switch daylight saving time from # June # 16 till September 30 in a bid to ensure maximum use of daylight to cope with # crippling power crisis. " # # The switch will remain in effect from June 16 to Sept 30 (2009) but if # implemented the next year, it will come in force from April 1, 2010 # From Steffen Thorsen (2009-06-02): # They have finally decided now, but changed the start date to midnight between # the 19th and 20th, and they have not set the end date yet. # # Some sources: # http://in.reuters.com/article/southAsiaNews/idINIndia-40017620090601 # http://bdnews24.com/details.php?id=85889&cid=2 # # Our wrap-up: # http://www.timeanddate.com/news/time/bangladesh-daylight-saving-2009.html # From A. N. M. Kamrus Saadat (2009-06-15): # Finally we've got the official mail regarding DST start time where DST start # time is mentioned as Jun 19 2009, 23:00 from BTRC (Bangladesh # Telecommunication Regulatory Commission). # # No DST end date has been announced yet. # From Alexander Krivenyshev (2009-09-25): # Bangladesh won't go back to Standard Time from October 1, 2009, # instead it will continue DST measure till the cabinet makes a fresh decision. # # Following report by same newspaper-"The Daily Star Friday": # "DST change awaits cabinet decision-Clock won't go back by 1-hr from Oct 1" # http://www.thedailystar.net/newDesign/news-details.php?nid=107021 # http://www.worldtimezone.com/dst_news/dst_news_bangladesh04.html # From Steffen Thorsen (2009-10-13): # IANS (Indo-Asian News Service) now reports: # Bangladesh has decided that the clock advanced by an hour to make # maximum use of daylight hours as an energy saving measure would # "continue for an indefinite period." # # One of many places where it is published: # http://www.thaindian.com/newsportal/business/bangladesh-to-continue-indefinitely-with-advanced-time_100259987.html # From Alexander Krivenyshev (2009-12-24): # According to Bangladesh newspaper "The Daily Star," # Bangladesh will change its clock back to Standard Time on Dec 31, 2009. # # Clock goes back 1-hr on Dec 31 night. # http://www.thedailystar.net/newDesign/news-details.php?nid=119228 # http://www.worldtimezone.com/dst_news/dst_news_bangladesh05.html # # "...The government yesterday decided to put the clock back by one hour # on December 31 midnight and the new time will continue until March 31, # 2010 midnight. The decision came at a cabinet meeting at the Prime # Minister's Office last night..." # From Alexander Krivenyshev (2010-03-22): # According to Bangladesh newspaper "The Daily Star," # Cabinet cancels Daylight Saving Time # http://www.thedailystar.net/newDesign/latest_news.php?nid=22817 # http://www.worldtimezone.com/dst_news/dst_news_bangladesh06.html # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Dhaka 2009 only - Jun 19 23:00 1:00 S Rule Dhaka 2009 only - Dec 31 24:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Dhaka 6:01:40 - LMT 1890 5:53:20 - HMT 1941 Oct # Howrah Mean Time? 6:30 - BURT 1942 May 15 # Burma Time 5:30 - IST 1942 Sep 6:30 - BURT 1951 Sep 30 6:00 - DACT 1971 Mar 26 # Dacca Time 6:00 - BDT 2009 6:00 Dhaka BD%sT # Bhutan # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Thimphu 5:58:36 - LMT 1947 Aug 15 # or Thimbu 5:30 - IST 1987 Oct 6:00 - BTT # Bhutan Time # British Indian Ocean Territory # Whitman and the 1995 CIA time zone map say 5:00, but the # 1997 and later maps say 6:00. Assume the switch occurred in 1996. # We have no information as to when standard time was introduced; # assume it occurred in 1907, the same year as Mauritius (which # then contained the Chagos Archipelago). # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Indian/Chagos 4:49:40 - LMT 1907 5:00 - IOT 1996 # BIOT Time 6:00 - IOT # Brunei # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Brunei 7:39:40 - LMT 1926 Mar # Bandar Seri Begawan 7:30 - BNT 1933 8:00 - BNT # Burma / Myanmar # Milne says 6:24:40 was the meridian of the time ball observatory at Rangoon. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Rangoon 6:24:40 - LMT 1880 # or Yangon 6:24:40 - RMT 1920 # Rangoon Mean Time? 6:30 - BURT 1942 May # Burma Time 9:00 - JST 1945 May 3 6:30 - MMT # Myanmar Time # Cambodia # See Asia/Bangkok. # China # From Guy Harris: # People's Republic of China. Yes, they really have only one time zone. # From Bob Devine (1988-01-28): # No they don't. See TIME mag, 1986-02-17 p.52. Even though # China is across 4 physical time zones, before Feb 1, 1986 only the # Peking (Beijing) time zone was recognized. Since that date, China # has two of 'em - Peking's and Ürümqi (named after the capital of # the Xinjiang Uyghur Autonomous Region). I don't know about DST for it. # # . . .I just deleted the DST table and this editor makes it too # painful to suck in another copy. So, here is what I have for # DST start/end dates for Peking's time zone (info from AP): # # 1986 May 4 - Sept 14 # 1987 mid-April - ?? # From U. S. Naval Observatory (1989-01-19): # CHINA 8 H AHEAD OF UTC ALL OF CHINA, INCL TAIWAN # CHINA 9 H AHEAD OF UTC APR 17 - SEP 10 # From Paul Eggert (2008-02-11): # Jim Mann, "A clumsy embrace for another western custom: China on daylight # time - sort of", Los Angeles Times, 1986-05-05 ... [says] that China began # observing daylight saving time in 1986. # From Paul Eggert (2014-06-30): # Shanks & Pottenger have China switching to a single time zone in 1980, but # this doesn't seem to be correct. They also write that China observed summer # DST from 1986 through 1991, which seems to match the above commentary, so # go with them for DST rules as follows: # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Shang 1940 only - Jun 3 0:00 1:00 D Rule Shang 1940 1941 - Oct 1 0:00 0 S Rule Shang 1941 only - Mar 16 0:00 1:00 D Rule PRC 1986 only - May 4 0:00 1:00 D Rule PRC 1986 1991 - Sep Sun>=11 0:00 0 S Rule PRC 1987 1991 - Apr Sun>=10 0:00 1:00 D # From Anthony Fok (2001-12-20): # BTW, I did some research on-line and found some info regarding these five # historic timezones from some Taiwan websites. And yes, there are official # Chinese names for these locales (before 1949). # # From Jesper Nørgaard Welen (2006-07-14): # I have investigated the timezones around 1970 on the # http://www.astro.com/atlas site [with provinces and county # boundaries summarized below].... A few other exceptions were two # counties on the Sichuan side of the Xizang-Sichuan border, # counties Dege and Baiyu which lies on the Sichuan side and are # therefore supposed to be GMT+7, Xizang region being GMT+6, but Dege # county is GMT+8 according to astro.com while Baiyu county is GMT+6 # (could be true), for the moment I am assuming that those two # counties are mistakes in the astro.com data. # From Paul Eggert (2014-06-30): # Alois Treindl kindly sent me translations of the following two sources: # # (1) # Guo Qingsheng (National Time-Service Center, CAS, Xi'an 710600, China) # Beijing Time at the Beginning of the PRC # China Historical Materials of Science and Technology # (Zhongguo ke ji shi liao, 中国科技史料), Vol. 24, No. 1 (2003) # It gives evidence that at the beginning of the PRC, Beijing time was # officially apparent solar time! However, Guo also says that the # evidence is dubious, as the relevant institute of astronomy had not # been taken over by the PRC yet. It's plausible that apparent solar # time was announced but never implemented, and that people continued # to use UT+8. As the Shanghai radio station (and I presume the # observatory) was still under control of French missionaries, it # could well have ignored any such mandate. # # (2) # Guo Qing-sheng (Shaanxi Astronomical Observatory, CAS, Xi'an 710600, China) # A Study on the Standard Time Changes for the Past 100 Years in China # [undated and unknown publication location] # It says several things: # * The Qing dynasty used local apparent solar time throughout China. # * The Republic of China instituted Beijing mean solar time effective # the official calendar book of 1914. # * The French Concession in Shanghai set up signal stations in # French docks in the 1890s, controlled by Xujiahui (Zikawei) # Observatory and set to local mean time. # * "From the end of the 19th century" it changed to UT+8. # * Chinese Customs (by then reduced to a tool of foreign powers) # eventually standardized on this time for all ports, and it # became used by railways as well. # * In 1918 the Central Observatory proposed dividing China into # five time zones (see below for details). This caught on # at first only in coastal areas observing UT+8. # * During WWII all of China was in theory was at UT+7. In practice # this was ignored in the west, and I presume was ignored in # Japanese-occupied territory. # * Japanese-occupied Manchuria was at UT+9, i.e., Japan time. # * The five-zone plan was resurrected after WWII and officially put into # place (with some modifications) in March 1948. It's not clear # how well it was observed in areas under Nationalist control. # * The People's Liberation Army used UT+8 during the civil war. # # An AP article "Shanghai Internat'l Area Little Changed" in the # Lewiston (ME) Daily Sun (1939-05-29), p 17, said "Even the time is # different - the occupied districts going by Tokyo time, an hour # ahead of that prevailing in the rest of Shanghai." Guess that the # Xujiahui Observatory was under French control and stuck with UT+8. # # In earlier versions of this file, China had many separate Zone entries, but # this was based on what were apparently incorrect data in Shanks & Pottenger. # This has now been simplified to the two entries Asia/Shanghai and # Asia/Urumqi, with the others being links for backward compatibility. # Proposed in 1918 and theoretically in effect until 1949 (although in practice # mainly observed in coastal areas), the five zones were: # # Changbai Time ("Long-white Time", Long-white = Heilongjiang area) UT+8.5 # Asia/Harbin (currently a link to Asia/Shanghai) # Heilongjiang (except Mohe county), Jilin # # Zhongyuan Time ("Central plain Time") UT+8 # Asia/Shanghai # most of China # This currently represents most other zones as well, # as apparently these regions have been the same since 1970. # Milne gives 8:05:43.2 for Xujiahui Observatory time; round to nearest. # Guo says Shanghai switched to UT+8 "from the end of the 19th century". # # Long-shu Time (probably due to Long and Shu being two names of that area) UT+7 # Asia/Chongqing (currently a link to Asia/Shanghai) # Guangxi, Guizhou, Hainan, Ningxia, Sichuan, Shaanxi, and Yunnan; # most of Gansu; west Inner Mongolia; west Qinghai; and the Guangdong # counties Deqing, Enping, Kaiping, Luoding, Taishan, Xinxing, # Yangchun, Yangjiang, Yu'nan, and Yunfu. # # Xin-zang Time ("Xinjiang-Tibet Time") UT+6 # Asia/Urumqi # This currently represents Kunlun Time as well, # as apparently the two regions have been the same since 1970. # The Gansu counties Aksay, Anxi, Dunhuang, Subei; west Qinghai; # the Guangdong counties Xuwen, Haikang, Suixi, Lianjiang, # Zhanjiang, Wuchuan, Huazhou, Gaozhou, Maoming, Dianbai, and Xinyi; # east Tibet, including Lhasa, Chamdo, Shigaise, Jimsar, Shawan and Hutubi; # east Xinjiang, including Ürümqi, Turpan, Karamay, Korla, Minfeng, Jinghe, # Wusu, Qiemo, Xinyan, Wulanwusu, Jinghe, Yumin, Tacheng, Tuoli, Emin, # Shihezi, Changji, Yanqi, Heshuo, Tuokexun, Tulufan, Shanshan, Hami, # Fukang, Kuitun, Kumukuli, Miquan, Qitai, and Turfan. # # Kunlun Time UT+5.5 # Asia/Kashgar (currently a link to Asia/Urumqi) # West Tibet, including Pulan, Aheqi, Shufu, Shule; # West Xinjiang, including Aksu, Atushi, Yining, Hetian, Cele, Luopu, Nileke, # Zhaosu, Tekesi, Gongliu, Chabuchaer, Huocheng, Bole, Pishan, Suiding, # and Yarkand. # From Luther Ma (2009-10-17): # Almost all (>99.9%) ethnic Chinese (properly ethnic Han) living in # Xinjiang use Chinese Standard Time. Some are aware of Xinjiang time, # but have no need of it. All planes, trains, and schools function on # what is called "Beijing time." When Han make an appointment in Chinese # they implicitly use Beijing time. # # On the other hand, ethnic Uyghurs, who make up about half the # population of Xinjiang, typically use "Xinjiang time" which is two # hours behind Beijing time, or UTC +0600. The government of the Xinjiang # Uyghur Autonomous Region, (XAUR, or just Xinjiang for short) as well as # local governments such as the Ürümqi city government use both times in # publications, referring to what is popularly called Xinjiang time as # "Ürümqi time." When Uyghurs make an appointment in the Uyghur language # they almost invariably use Xinjiang time. # # (Their ethnic Han compatriots would typically have no clue of its # widespread use, however, because so extremely few of them are fluent in # Uyghur, comparable to the number of Anglo-Americans fluent in Navajo.) # # (...As with the rest of China there was a brief interval ending in 1990 # or 1991 when summer time was in use. The confusion was severe, with # the province not having dual times but four times in use at the same # time. Some areas remained on standard Xinjiang time or Beijing time and # others moving their clocks ahead.) # From Luther Ma (2009-11-19): # With the risk of being redundant to previous answers these are the most common # English "transliterations" (w/o using non-English symbols): # # 1. Wulumuqi... # 2. Kashi... # 3. Urumqi... # 4. Kashgar... # ... # 5. It seems that Uyghurs in Ürümqi has been using Xinjiang since at least the # 1960's. I know of one Han, now over 50, who grew up in the surrounding # countryside and used Xinjiang time as a child. # # 6. Likewise for Kashgar and the rest of south Xinjiang I don't know of any # start date for Xinjiang time. # # Without having access to local historical records, nor the ability to legally # publish them, I would go with October 1, 1949, when Xinjiang became the Uyghur # Autonomous Region under the PRC. (Before that Uyghurs, of course, would also # not be using Beijing time, but some local time.) # From David Cochrane (2014-03-26): # Just a confirmation that Ürümqi time was implemented in Ürümqi on 1 Feb 1986: # http://content.time.com/time/magazine/article/0,9171,960684,00.html # From Luther Ma (2014-04-22): # I have interviewed numerous people of various nationalities and from # different localities in Xinjiang and can confirm the information in Guo's # report regarding Xinjiang, as well as the Time article reference by David # Cochrane. Whether officially recognized or not (and both are officially # recognized), two separate times have been in use in Xinjiang since at least # the Cultural Revolution: Xinjiang Time (XJT), aka Ürümqi Time or local time; # and Beijing Time. There is no confusion in Xinjiang as to which name refers # to which time. Both are widely used in the province, although in some # population groups might be use one to the exclusion of the other. The only # problem is that computers and smart phones list Ürümqi (or Kashgar) as # having the same time as Beijing. # From Paul Eggert (2014-06-30): # In the early days of the PRC, Tibet was given its own time zone (UT+6) but # this was withdrawn in 1959 and never reinstated; see Tubten Khétsun, # Memories of life in Lhasa under Chinese Rule, Columbia U Press, ISBN # 978-0231142861 (2008), translator's introduction by Matthew Akester, p x. # As this is before our 1970 cutoff, Tibet doesn't need a separate zone. # # Xinjiang Time is well-documented as being officially recognized. E.g., see # "The Working-Calendar for The Xinjiang Uygur Autonomous Region Government" # (2014-04-22). # Unfortunately, we have no good records of time in Xinjiang before 1986. # During the 20th century parts of Xinjiang were ruled by the Qing dynasty, # the Republic of China, various warlords, the First and Second East Turkestan # Republics, the Soviet Union, the Kuomintang, and the People's Republic of # China, and tracking down all these organizations' timekeeping rules would be # quite a trick. Approximate this lost history by a transition from LMT to # XJT at the start of 1928, the year of accession of the warlord Jin Shuren, # which happens to be the date given by Shanks & Pottenger (no doubt as a # guess) as the transition from LMT. Ignore the usage of UT+8 before # 1986-02-01 under the theory that the transition date to UT+8 is unknown and # that the sort of users who prefer Asia/Urumqi now typically ignored the # UT+8 mandate back then. # Zone NAME GMTOFF RULES FORMAT [UNTIL] # Beijing time, used throughout China; represented by Shanghai. Zone Asia/Shanghai 8:05:43 - LMT 1901 8:00 Shang C%sT 1949 8:00 PRC C%sT # Xinjiang time, used by many in western China; represented by Ürümqi / Ürümchi # / Wulumuqi. (Please use Asia/Shanghai if you prefer Beijing time.) Zone Asia/Urumqi 5:50:20 - LMT 1928 6:00 - XJT # Hong Kong (Xianggang) # Milne gives 7:36:41.7; round this. # From Lee Yiu Chung (2009-10-24): # I found there are some mistakes for the...DST rule for Hong # Kong. [According] to the DST record from Hong Kong Observatory (actually, # it is not [an] observatory, but the official meteorological agency of HK, # and also serves as the official timing agency), there are some missing # and incorrect rules. Although the exact switch over time is missing, I # think 3:30 is correct. The official DST record for Hong Kong can be # obtained from # http://www.hko.gov.hk/gts/time/Summertime.htm # From Arthur David Olson (2009-10-28): # Here are the dates given at # http://www.hko.gov.hk/gts/time/Summertime.htm # as of 2009-10-28: # Year Period # 1941 1 Apr to 30 Sep # 1942 Whole year # 1943 Whole year # 1944 Whole year # 1945 Whole year # 1946 20 Apr to 1 Dec # 1947 13 Apr to 30 Dec # 1948 2 May to 31 Oct # 1949 3 Apr to 30 Oct # 1950 2 Apr to 29 Oct # 1951 1 Apr to 28 Oct # 1952 6 Apr to 25 Oct # 1953 5 Apr to 1 Nov # 1954 21 Mar to 31 Oct # 1955 20 Mar to 6 Nov # 1956 18 Mar to 4 Nov # 1957 24 Mar to 3 Nov # 1958 23 Mar to 2 Nov # 1959 22 Mar to 1 Nov # 1960 20 Mar to 6 Nov # 1961 19 Mar to 5 Nov # 1962 18 Mar to 4 Nov # 1963 24 Mar to 3 Nov # 1964 22 Mar to 1 Nov # 1965 18 Apr to 17 Oct # 1966 17 Apr to 16 Oct # 1967 16 Apr to 22 Oct # 1968 21 Apr to 20 Oct # 1969 20 Apr to 19 Oct # 1970 19 Apr to 18 Oct # 1971 18 Apr to 17 Oct # 1972 16 Apr to 22 Oct # 1973 22 Apr to 21 Oct # 1973/74 30 Dec 73 to 20 Oct 74 # 1975 20 Apr to 19 Oct # 1976 18 Apr to 17 Oct # 1977 Nil # 1978 Nil # 1979 13 May to 21 Oct # 1980 to Now Nil # The page does not give start or end times of day. # The page does not give a start date for 1942. # The page does not givw an end date for 1945. # The Japanese occupation of Hong Kong began on 1941-12-25. # The Japanese surrender of Hong Kong was signed 1945-09-15. # For lack of anything better, use start of those days as the transition times. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule HK 1941 only - Apr 1 3:30 1:00 S Rule HK 1941 only - Sep 30 3:30 0 - Rule HK 1946 only - Apr 20 3:30 1:00 S Rule HK 1946 only - Dec 1 3:30 0 - Rule HK 1947 only - Apr 13 3:30 1:00 S Rule HK 1947 only - Dec 30 3:30 0 - Rule HK 1948 only - May 2 3:30 1:00 S Rule HK 1948 1951 - Oct lastSun 3:30 0 - Rule HK 1952 only - Oct 25 3:30 0 - Rule HK 1949 1953 - Apr Sun>=1 3:30 1:00 S Rule HK 1953 only - Nov 1 3:30 0 - Rule HK 1954 1964 - Mar Sun>=18 3:30 1:00 S Rule HK 1954 only - Oct 31 3:30 0 - Rule HK 1955 1964 - Nov Sun>=1 3:30 0 - Rule HK 1965 1976 - Apr Sun>=16 3:30 1:00 S Rule HK 1965 1976 - Oct Sun>=16 3:30 0 - Rule HK 1973 only - Dec 30 3:30 1:00 S Rule HK 1979 only - May Sun>=8 3:30 1:00 S Rule HK 1979 only - Oct Sun>=16 3:30 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Hong_Kong 7:36:42 - LMT 1904 Oct 30 8:00 HK HK%sT 1941 Dec 25 9:00 - JST 1945 Sep 15 8:00 HK HK%sT ############################################################################### # Taiwan # From smallufo (2010-04-03): # According to Taiwan's CWB [Central Weather Bureau], # http://www.cwb.gov.tw/V6/astronomy/cdata/summert.htm # Taipei has DST in 1979 between July 1st and Sep 30. # From Yu-Cheng Chuang (2013-07-12): # On Dec 28, 1895, the Meiji Emperor announced Ordinance No. 167 of # Meiji Year 28 "The clause about standard time", mentioned that # Taiwan and Penghu Islands, as well as Yaeyama and Miyako Islands # (both in Okinawa) adopt the Western Standard Time which is based on # 120E. The adoption began from Jan 1, 1896. The original text can be # found on Wikisource: # http://ja.wikisource.org/wiki/標準時ニ關スル件_(公布時) # ... This could be the first adoption of time zone in Taiwan, because # during the Qing Dynasty, it seems that there was no time zone # declared officially. # # Later, in the beginning of World War II, on Sep 25, 1937, the Showa # Emperor announced Ordinance No. 529 of Showa Year 12 "The clause of # revision in the ordinance No. 167 of Meiji year 28 about standard # time", in which abolished the adoption of Western Standard Time in # western islands (listed above), which means the whole Japan # territory, including later occupations, adopt Japan Central Time # (UTC+9). The adoption began on Oct 1, 1937. The original text can # be found on Wikisource: # http://ja.wikisource.org/wiki/明治二十八年勅令第百六十七號標準時ニ關スル件中改正ノ件 # # That is, the time zone of Taipei switched to UTC+9 on Oct 1, 1937. # From Yu-Cheng Chuang (2014-07-02): # I've found more evidence about when the time zone was switched from UTC+9 # back to UTC+8 after WW2. I believe it was on Sep 21, 1945. In a document # during Japanese era [1] in which the officer told the staff to change time # zone back to Western Standard Time (UTC+8) on Sep 21. And in another # history page of National Cheng Kung University [2], on Sep 21 there is a # note "from today, switch back to Western Standard Time". From these two # materials, I believe that the time zone change happened on Sep 21. And # today I have found another monthly journal called "The Astronomical Herald" # from The Astronomical Society of Japan [3] in which it mentioned the fact # that: # # 1. Standard Time of the Country (Japan) was adopted on Jan 1, 1888, using # the time at 135E (GMT+9) # # 2. Standard Time of the Country was renamed to Central Standard Time, on Jan # 1, 1898, and on the same day, the new territories Taiwan and Penghu islands, # as well as Yaeyama and Miyako islands, adopted a new time zone called # Western Standard Time, which is in GMT+8. # # 3. Western Standard Time was deprecated on Sep 30, 1937. From then all the # territories of Japan adopted the same time zone, which is Central Standard # Time. # # [1] Academica Historica, Taiwan: # http://163.29.208.22:8080/govsaleShowImage/connect_img.php?s=00101738900090036&e=00101738900090037 # [2] Nat'l Cheng Kung University 70th Anniversary Special Site: # http://www.ncku.edu.tw/~ncku70/menu/001/01_01.htm # [3] Yukio Niimi, The Standard Time in Japan (1997), p.475: # http://www.asj.or.jp/geppou/archive_open/1997/pdf/19971001c.pdf # Yu-Cheng Chuang (2014-07-03): # I finally have found the real official gazette about changing back to # Western Standard Time on Sep 21 in Taiwan. It's Taiwan Governor-General # Bulletin No. 386 in Showa 20 years (1945), published on Sep 19, 1945. [1] ... # [It] abolishes Bulletin No. 207 in Showa 12 years (1937), which is a local # bulletin in Taiwan for that Ordinance No. 529. It also mentioned that 1am on # Sep 21, 1945 will be 12am on Sep 21. I think this bulletin is much more # official than the one I mentioned in my first mail, because it's from the # top-level government in Taiwan. If you're going to quote any resource, this # would be a good one. # [1] Taiwan Governor-General Gazette, No. 1018, Sep 19, 1945: # http://db2.th.gov.tw/db2/view/viewImg.php?imgcode=0072031018a&num=19&bgn=019&end=019&otherImg=&type=gener # From Yu-Cheng Chuang (2014-07-02): # In 1946, DST in Taiwan was from May 15 and ended on Sep 30. The info from # Central Weather Bureau website was not correct. # # Original Bulletin: # http://subtpg.tpg.gov.tw/og/image2.asp?f=03502F0AKM1AF # http://subtpg.tpg.gov.tw/og/image2.asp?f=0350300AKM1B0 (cont.) # # In 1947, DST in Taiwan was expanded to Oct 31. There is a backup of that # telegram announcement from Taiwan Province Government: # # http://subtpg.tpg.gov.tw/og/image2.asp?f=0360310AKZ431 # # Here is a brief translation: # # The Summer Time this year is adopted from midnight Apr 15 until Sep 20 # midnight. To save (energy?) consumption, we're expanding Summer Time # adoption till Oct 31 midnight. # # The Central Weather Bureau website didn't mention that, however it can # be found from historical government announcement database. # From Paul Eggert (2014-07-03): # As per Yu-Cheng Chuang, say that Taiwan was at UT+9 from 1937-10-01 # until 1945-09-21 at 01:00, overriding Shanks & Pottenger. # Likewise, use Yu-Cheng Chuang's data for DST in Taiwan. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Taiwan 1946 only - May 15 0:00 1:00 D Rule Taiwan 1946 only - Oct 1 0:00 0 S Rule Taiwan 1947 only - Apr 15 0:00 1:00 D Rule Taiwan 1947 only - Nov 1 0:00 0 S Rule Taiwan 1948 1951 - May 1 0:00 1:00 D Rule Taiwan 1948 1951 - Oct 1 0:00 0 S Rule Taiwan 1952 only - Mar 1 0:00 1:00 D Rule Taiwan 1952 1954 - Nov 1 0:00 0 S Rule Taiwan 1953 1959 - Apr 1 0:00 1:00 D Rule Taiwan 1955 1961 - Oct 1 0:00 0 S Rule Taiwan 1960 1961 - Jun 1 0:00 1:00 D Rule Taiwan 1974 1975 - Apr 1 0:00 1:00 D Rule Taiwan 1974 1975 - Oct 1 0:00 0 S Rule Taiwan 1979 only - Jul 1 0:00 1:00 D Rule Taiwan 1979 only - Oct 1 0:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] # Taipei or Taibei or T'ai-pei Zone Asia/Taipei 8:06:00 - LMT 1896 Jan 1 8:00 - JWST 1937 Oct 1 9:00 - JST 1945 Sep 21 1:00 8:00 Taiwan C%sT # Macau (Macao, Aomen) # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Macau 1961 1962 - Mar Sun>=16 3:30 1:00 S Rule Macau 1961 1964 - Nov Sun>=1 3:30 0 - Rule Macau 1963 only - Mar Sun>=16 0:00 1:00 S Rule Macau 1964 only - Mar Sun>=16 3:30 1:00 S Rule Macau 1965 only - Mar Sun>=16 0:00 1:00 S Rule Macau 1965 only - Oct 31 0:00 0 - Rule Macau 1966 1971 - Apr Sun>=16 3:30 1:00 S Rule Macau 1966 1971 - Oct Sun>=16 3:30 0 - Rule Macau 1972 1974 - Apr Sun>=15 0:00 1:00 S Rule Macau 1972 1973 - Oct Sun>=15 0:00 0 - Rule Macau 1974 1977 - Oct Sun>=15 3:30 0 - Rule Macau 1975 1977 - Apr Sun>=15 3:30 1:00 S Rule Macau 1978 1980 - Apr Sun>=15 0:00 1:00 S Rule Macau 1978 1980 - Oct Sun>=15 0:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Macau 7:34:20 - LMT 1912 Jan 1 8:00 Macau MO%sT 1999 Dec 20 # return to China 8:00 PRC C%sT ############################################################################### # Cyprus # # Milne says the Eastern Telegraph Company used 2:14:00. Stick with LMT. # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Cyprus 1975 only - Apr 13 0:00 1:00 S Rule Cyprus 1975 only - Oct 12 0:00 0 - Rule Cyprus 1976 only - May 15 0:00 1:00 S Rule Cyprus 1976 only - Oct 11 0:00 0 - Rule Cyprus 1977 1980 - Apr Sun>=1 0:00 1:00 S Rule Cyprus 1977 only - Sep 25 0:00 0 - Rule Cyprus 1978 only - Oct 2 0:00 0 - Rule Cyprus 1979 1997 - Sep lastSun 0:00 0 - Rule Cyprus 1981 1998 - Mar lastSun 0:00 1:00 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Nicosia 2:13:28 - LMT 1921 Nov 14 2:00 Cyprus EE%sT 1998 Sep 2:00 EUAsia EE%sT # IATA SSIM (1998-09) has Cyprus using EU rules for the first time. # Classically, Cyprus belongs to Asia; e.g. see Herodotus, Histories, I.72. # However, for various reasons many users expect to find it under Europe. Link Asia/Nicosia Europe/Nicosia # Georgia # From Paul Eggert (1994-11-19): # Today's _Economist_ (p 60) reports that Georgia moved its clocks forward # an hour recently, due to a law proposed by Zurab Murvanidze, # an MP who went on a hunger strike for 11 days to force discussion about it! # We have no details, but we'll guess they didn't move the clocks back in fall. # # From Mathew Englander, quoting AP (1996-10-23 13:05-04): # Instead of putting back clocks at the end of October, Georgia # will stay on daylight savings time this winter to save energy, # President Eduard Shevardnadze decreed Wednesday. # # From the BBC via Joseph S. Myers (2004-06-27): # # Georgia moved closer to Western Europe on Sunday... The former Soviet # republic has changed its time zone back to that of Moscow. As a result it # is now just four hours ahead of Greenwich Mean Time, rather than five hours # ahead. The switch was decreed by the pro-Western president of Georgia, # Mikheil Saakashvili, who said the change was partly prompted by the process # of integration into Europe. # From Teimuraz Abashidze (2005-11-07): # Government of Georgia ... decided to NOT CHANGE daylight savings time on # [Oct.] 30, as it was done before during last more than 10 years. # Currently, we are in fact GMT +4:00, as before 30 October it was GMT # +3:00.... The problem is, there is NO FORMAL LAW or governmental document # about it. As far as I can find, I was told, that there is no document, # because we just DIDN'T ISSUE document about switching to winter time.... # I don't know what can be done, especially knowing that some years ago our # DST rules where changed THREE TIMES during one month. # Milne 1899 says Tbilisi (Tiflis) time was 2:59:05.7. # Byalokoz 1919 says Georgia was 2:59:11. # Go with Byalokoz. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Tbilisi 2:59:11 - LMT 1880 2:59:11 - TBMT 1924 May 2 # Tbilisi Mean Time 3:00 - TBIT 1957 Mar # Tbilisi Time 4:00 RussiaAsia TBI%sT 1991 Mar 31 2:00s 3:00 1:00 TBIST 1991 Apr 9 # independence 3:00 RussiaAsia GE%sT 1992 # Georgia Time 3:00 E-EurAsia GE%sT 1994 Sep lastSun 4:00 E-EurAsia GE%sT 1996 Oct lastSun 4:00 1:00 GEST 1997 Mar lastSun 4:00 E-EurAsia GE%sT 2004 Jun 27 3:00 RussiaAsia GE%sT 2005 Mar lastSun 2:00 4:00 - GET # East Timor # See Indonesia for the 1945 transition. # From João Carrascalão, brother of the former governor of East Timor, in # East Timor may be late for its millennium # (1999-12-26/31): # Portugal tried to change the time forward in 1974 because the sun # rises too early but the suggestion raised a lot of problems with the # Timorese and I still don't think it would work today because it # conflicts with their way of life. # From Paul Eggert (2000-12-04): # We don't have any record of the above attempt. # Most likely our records are incomplete, but we have no better data. # From Manoel de Almeida e Silva, Deputy Spokesman for the UN Secretary-General # http://www.hri.org/news/world/undh/2000/00-08-16.undh.html # (2000-08-16): # The Cabinet of the East Timor Transition Administration decided # today to advance East Timor's time by one hour. The time change, # which will be permanent, with no seasonal adjustment, will happen at # midnight on Saturday, September 16. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Dili 8:22:20 - LMT 1912 Jan 1 8:00 - TLT 1942 Feb 21 23:00 # E Timor Time 9:00 - JST 1945 Sep 23 9:00 - TLT 1976 May 3 8:00 - WITA 2000 Sep 17 0:00 9:00 - TLT # India # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Kolkata 5:53:28 - LMT 1880 # Kolkata 5:53:20 - HMT 1941 Oct # Howrah Mean Time? 6:30 - BURT 1942 May 15 # Burma Time 5:30 - IST 1942 Sep 5:30 1:00 IST 1945 Oct 15 5:30 - IST # The following are like Asia/Kolkata: # Andaman Is # Lakshadweep (Laccadive, Minicoy and Amindivi Is) # Nicobar Is # Indonesia # # From Paul Eggert (2014-09-06): # The 1876 Report of the Secretary of the [US] Navy, p 306 says that Batavia # civil time was 7:07:12.5; round to even for Jakarta. # # From Gwillim Law (2001-05-28), overriding Shanks & Pottenger: # http://www.sumatera-inc.com/go_to_invest/about_indonesia.asp#standtime # says that Indonesia's time zones changed on 1988-01-01. Looking at some # time zone maps, I think that must refer to Western Borneo (Kalimantan Barat # and Kalimantan Tengah) switching from UTC+8 to UTC+7. # # From Paul Eggert (2007-03-10): # Here is another correction to Shanks & Pottenger. # JohnTWB writes that Japanese forces did not surrender control in # Indonesia until 1945-09-01 00:00 at the earliest (in Jakarta) and # other formal surrender ceremonies were September 9, 11, and 13, plus # September 12 for the regional surrender to Mountbatten in Singapore. # These would be the earliest possible times for a change. # Régimes horaires pour le monde entier, by Henri Le Corre, (Éditions # Traditionnelles, 1987, Paris) says that Java and Madura switched # from JST to UTC+07:30 on 1945-09-23, and gives 1944-09-01 for Jayapura # (Hollandia). For now, assume all Indonesian locations other than Jayapura # switched on 1945-09-23. # # From Paul Eggert (2013-08-11): # Normally the tz database uses English-language abbreviations, but in # Indonesia it's typical to use Indonesian-language abbreviations even # when writing in English. For example, see the English-language # summary published by the Time and Frequency Laboratory of the # Research Center for Calibration, Instrumentation and Metrology, # Indonesia, (2006-09-29). # The abbreviations are: # # WIB - UTC+7 - Waktu Indonesia Barat (Indonesia western time) # WITA - UTC+8 - Waktu Indonesia Tengah (Indonesia central time) # WIT - UTC+9 - Waktu Indonesia Timur (Indonesia eastern time) # # Zone NAME GMTOFF RULES FORMAT [UNTIL] # Java, Sumatra Zone Asia/Jakarta 7:07:12 - LMT 1867 Aug 10 # Shanks & Pottenger say the next transition was at 1924 Jan 1 0:13, # but this must be a typo. 7:07:12 - BMT 1923 Dec 31 23:47:12 # Batavia 7:20 - JAVT 1932 Nov # Java Time 7:30 - WIB 1942 Mar 23 9:00 - JST 1945 Sep 23 7:30 - WIB 1948 May 8:00 - WIB 1950 May 7:30 - WIB 1964 7:00 - WIB # west and central Borneo Zone Asia/Pontianak 7:17:20 - LMT 1908 May 7:17:20 - PMT 1932 Nov # Pontianak MT 7:30 - WIB 1942 Jan 29 9:00 - JST 1945 Sep 23 7:30 - WIB 1948 May 8:00 - WIB 1950 May 7:30 - WIB 1964 8:00 - WITA 1988 Jan 1 7:00 - WIB # Sulawesi, Lesser Sundas, east and south Borneo Zone Asia/Makassar 7:57:36 - LMT 1920 7:57:36 - MMT 1932 Nov # Macassar MT 8:00 - WITA 1942 Feb 9 9:00 - JST 1945 Sep 23 8:00 - WITA # Maluku Islands, West Papua, Papua Zone Asia/Jayapura 9:22:48 - LMT 1932 Nov 9:00 - WIT 1944 Sep 1 9:30 - ACST 1964 9:00 - WIT # Iran # From Roozbeh Pournader (2003-03-15): # This is an English translation of what I just found (originally in Persian). # The Gregorian dates in brackets are mine: # # Official Newspaper No. 13548-1370/6/25 [1991-09-16] # No. 16760/T233 H 1370/6/10 [1991-09-01] # # The Rule About Change of the Official Time of the Country # # The Board of Ministers, in the meeting dated 1370/5/23 [1991-08-14], # based on the suggestion number 2221/D dated 1370/4/22 [1991-07-13] # of the Country's Organization for Official and Employment Affairs, # and referring to the law for equating the working hours of workers # and officers in the whole country dated 1359/4/23 [1980-07-14], and # for synchronizing the official times of the country, agreed that: # # The official time of the country will should move forward one hour # at the 24[:00] hours of the first day of Farvardin and should return # to its previous state at the 24[:00] hours of the 30th day of # Shahrivar. # # First Deputy to the President - Hassan Habibi # # From personal experience, that agrees with what has been followed # for at least the last 5 years. Before that, for a few years, the # date used was the first Thursday night of Farvardin and the last # Thursday night of Shahrivar, but I can't give exact dates.... # I have also changed the abbreviations to what is considered correct # here in Iran, IRST for regular time and IRDT for daylight saving time. # # From Roozbeh Pournader (2005-04-05): # The text of the Iranian law, in effect since 1925, clearly mentions # that the true solar year is the measure, and there is no arithmetic # leap year calculation involved. There has never been any serious # plan to change that law.... # # From Paul Eggert (2006-03-22): # Go with Shanks & Pottenger before Sept. 1991, and with Pournader thereafter. # I used Ed Reingold's cal-persia in GNU Emacs 21.2 to check Persian dates, # stopping after 2037 when 32-bit time_t's overflow. # That cal-persia used Birashk's approximation, which disagrees with the solar # calendar predictions for the year 2025, so I corrected those dates by hand. # # From Oscar van Vlijmen (2005-03-30), writing about future # discrepancies between cal-persia and the Iranian calendar: # For 2091 solar-longitude-after yields 2091-03-20 08:40:07.7 UT for # the vernal equinox and that gets so close to 12:00 some local # Iranian time that the definition of the correct location needs to be # known exactly, amongst other factors. 2157 is even closer: # 2157-03-20 08:37:15.5 UT. But the Gregorian year 2025 should give # no interpretation problem whatsoever. By the way, another instant # in the near future where there will be a discrepancy between # arithmetical and astronomical Iranian calendars will be in 2058: # vernal equinox on 2058-03-20 09:03:05.9 UT. The Java version of # Reingold's/Dershowitz' calculator gives correctly the Gregorian date # 2058-03-21 for 1 Farvardin 1437 (astronomical). # # From Steffen Thorsen (2006-03-22): # Several of my users have reported that Iran will not observe DST anymore: # http://www.irna.ir/en/news/view/line-17/0603193812164948.htm # # From Reuters (2007-09-16), with a heads-up from Jesper Nørgaard Welen: # ... the Guardian Council ... approved a law on Sunday to re-introduce # daylight saving time ... # http://uk.reuters.com/article/oilRpt/idUKBLA65048420070916 # # From Roozbeh Pournader (2007-11-05): # This is quoted from Official Gazette of the Islamic Republic of # Iran, Volume 63, Number 18242, dated Tuesday 1386/6/24 # [2007-10-16]. I am doing the best translation I can:... # The official time of the country will be moved forward for one hour # on the 24 hours of the first day of the month of Farvardin and will # be changed back to its previous state on the 24 hours of the # thirtieth day of Shahrivar. # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Iran 1978 1980 - Mar 21 0:00 1:00 D Rule Iran 1978 only - Oct 21 0:00 0 S Rule Iran 1979 only - Sep 19 0:00 0 S Rule Iran 1980 only - Sep 23 0:00 0 S Rule Iran 1991 only - May 3 0:00 1:00 D Rule Iran 1992 1995 - Mar 22 0:00 1:00 D Rule Iran 1991 1995 - Sep 22 0:00 0 S Rule Iran 1996 only - Mar 21 0:00 1:00 D Rule Iran 1996 only - Sep 21 0:00 0 S Rule Iran 1997 1999 - Mar 22 0:00 1:00 D Rule Iran 1997 1999 - Sep 22 0:00 0 S Rule Iran 2000 only - Mar 21 0:00 1:00 D Rule Iran 2000 only - Sep 21 0:00 0 S Rule Iran 2001 2003 - Mar 22 0:00 1:00 D Rule Iran 2001 2003 - Sep 22 0:00 0 S Rule Iran 2004 only - Mar 21 0:00 1:00 D Rule Iran 2004 only - Sep 21 0:00 0 S Rule Iran 2005 only - Mar 22 0:00 1:00 D Rule Iran 2005 only - Sep 22 0:00 0 S Rule Iran 2008 only - Mar 21 0:00 1:00 D Rule Iran 2008 only - Sep 21 0:00 0 S Rule Iran 2009 2011 - Mar 22 0:00 1:00 D Rule Iran 2009 2011 - Sep 22 0:00 0 S Rule Iran 2012 only - Mar 21 0:00 1:00 D Rule Iran 2012 only - Sep 21 0:00 0 S Rule Iran 2013 2015 - Mar 22 0:00 1:00 D Rule Iran 2013 2015 - Sep 22 0:00 0 S Rule Iran 2016 only - Mar 21 0:00 1:00 D Rule Iran 2016 only - Sep 21 0:00 0 S Rule Iran 2017 2019 - Mar 22 0:00 1:00 D Rule Iran 2017 2019 - Sep 22 0:00 0 S Rule Iran 2020 only - Mar 21 0:00 1:00 D Rule Iran 2020 only - Sep 21 0:00 0 S Rule Iran 2021 2023 - Mar 22 0:00 1:00 D Rule Iran 2021 2023 - Sep 22 0:00 0 S Rule Iran 2024 only - Mar 21 0:00 1:00 D Rule Iran 2024 only - Sep 21 0:00 0 S Rule Iran 2025 2027 - Mar 22 0:00 1:00 D Rule Iran 2025 2027 - Sep 22 0:00 0 S Rule Iran 2028 2029 - Mar 21 0:00 1:00 D Rule Iran 2028 2029 - Sep 21 0:00 0 S Rule Iran 2030 2031 - Mar 22 0:00 1:00 D Rule Iran 2030 2031 - Sep 22 0:00 0 S Rule Iran 2032 2033 - Mar 21 0:00 1:00 D Rule Iran 2032 2033 - Sep 21 0:00 0 S Rule Iran 2034 2035 - Mar 22 0:00 1:00 D Rule Iran 2034 2035 - Sep 22 0:00 0 S Rule Iran 2036 2037 - Mar 21 0:00 1:00 D Rule Iran 2036 2037 - Sep 21 0:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Tehran 3:25:44 - LMT 1916 3:25:44 - TMT 1946 # Tehran Mean Time 3:30 - IRST 1977 Nov 4:00 Iran IR%sT 1979 3:30 Iran IR%sT # Iraq # # From Jonathan Lennox (2000-06-12): # An article in this week's Economist ("Inside the Saddam-free zone", p. 50 in # the U.S. edition) on the Iraqi Kurds contains a paragraph: # "The three northern provinces ... switched their clocks this spring and # are an hour ahead of Baghdad." # # But Rives McDow (2000-06-18) quotes a contact in Iraqi-Kurdistan as follows: # In the past, some Kurdish nationalists, as a protest to the Iraqi # Government, did not adhere to daylight saving time. They referred # to daylight saving as Saddam time. But, as of today, the time zone # in Iraqi-Kurdistan is on standard time with Baghdad, Iraq. # # So we'll ignore the Economist's claim. # From Steffen Thorsen (2008-03-10): # The cabinet in Iraq abolished DST last week, according to the following # news sources (in Arabic): # http://www.aljeeran.net/wesima_articles/news-20080305-98602.html # http://www.aswataliraq.info/look/article.tpl?id=2047&IdLanguage=17&IdPublication=4&NrArticle=71743&NrIssue=1&NrSection=10 # # We have published a short article in English about the change: # http://www.timeanddate.com/news/time/iraq-dumps-daylight-saving.html # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Iraq 1982 only - May 1 0:00 1:00 D Rule Iraq 1982 1984 - Oct 1 0:00 0 S Rule Iraq 1983 only - Mar 31 0:00 1:00 D Rule Iraq 1984 1985 - Apr 1 0:00 1:00 D Rule Iraq 1985 1990 - Sep lastSun 1:00s 0 S Rule Iraq 1986 1990 - Mar lastSun 1:00s 1:00 D # IATA SSIM (1991/1996) says Apr 1 12:01am UTC; guess the ':01' is a typo. # Shanks & Pottenger say Iraq did not observe DST 1992/1997; ignore this. # Rule Iraq 1991 2007 - Apr 1 3:00s 1:00 D Rule Iraq 1991 2007 - Oct 1 3:00s 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Baghdad 2:57:40 - LMT 1890 2:57:36 - BMT 1918 # Baghdad Mean Time? 3:00 - AST 1982 May 3:00 Iraq A%sT ############################################################################### # Israel # From Ephraim Silverberg (2001-01-11): # # I coined "IST/IDT" circa 1988. Until then there were three # different abbreviations in use: # # JST Jerusalem Standard Time [Danny Braniss, Hebrew University] # IZT Israel Zonal (sic) Time [Prof. Haim Papo, Technion] # EEST Eastern Europe Standard Time [used by almost everyone else] # # Since timezones should be called by country and not capital cities, # I ruled out JST. As Israel is in Asia Minor and not Eastern Europe, # EEST was equally unacceptable. Since "zonal" was not compatible with # any other timezone abbreviation, I felt that 'IST' was the way to go # and, indeed, it has received almost universal acceptance in timezone # settings in Israeli computers. # # In any case, I am happy to share timezone abbreviations with India, # high on my favorite-country list (and not only because my wife's # family is from India). # From Shanks & Pottenger: # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Zion 1940 only - Jun 1 0:00 1:00 D Rule Zion 1942 1944 - Nov 1 0:00 0 S Rule Zion 1943 only - Apr 1 2:00 1:00 D Rule Zion 1944 only - Apr 1 0:00 1:00 D Rule Zion 1945 only - Apr 16 0:00 1:00 D Rule Zion 1945 only - Nov 1 2:00 0 S Rule Zion 1946 only - Apr 16 2:00 1:00 D Rule Zion 1946 only - Nov 1 0:00 0 S Rule Zion 1948 only - May 23 0:00 2:00 DD Rule Zion 1948 only - Sep 1 0:00 1:00 D Rule Zion 1948 1949 - Nov 1 2:00 0 S Rule Zion 1949 only - May 1 0:00 1:00 D Rule Zion 1950 only - Apr 16 0:00 1:00 D Rule Zion 1950 only - Sep 15 3:00 0 S Rule Zion 1951 only - Apr 1 0:00 1:00 D Rule Zion 1951 only - Nov 11 3:00 0 S Rule Zion 1952 only - Apr 20 2:00 1:00 D Rule Zion 1952 only - Oct 19 3:00 0 S Rule Zion 1953 only - Apr 12 2:00 1:00 D Rule Zion 1953 only - Sep 13 3:00 0 S Rule Zion 1954 only - Jun 13 0:00 1:00 D Rule Zion 1954 only - Sep 12 0:00 0 S Rule Zion 1955 only - Jun 11 2:00 1:00 D Rule Zion 1955 only - Sep 11 0:00 0 S Rule Zion 1956 only - Jun 3 0:00 1:00 D Rule Zion 1956 only - Sep 30 3:00 0 S Rule Zion 1957 only - Apr 29 2:00 1:00 D Rule Zion 1957 only - Sep 22 0:00 0 S Rule Zion 1974 only - Jul 7 0:00 1:00 D Rule Zion 1974 only - Oct 13 0:00 0 S Rule Zion 1975 only - Apr 20 0:00 1:00 D Rule Zion 1975 only - Aug 31 0:00 0 S Rule Zion 1985 only - Apr 14 0:00 1:00 D Rule Zion 1985 only - Sep 15 0:00 0 S Rule Zion 1986 only - May 18 0:00 1:00 D Rule Zion 1986 only - Sep 7 0:00 0 S Rule Zion 1987 only - Apr 15 0:00 1:00 D Rule Zion 1987 only - Sep 13 0:00 0 S # From Avigdor Finkelstein (2014-03-05): # I check the Parliament (Knesset) records and there it's stated that the # [1988] transition should take place on Saturday night, when the Sabbath # ends and changes to Sunday. Rule Zion 1988 only - Apr 10 0:00 1:00 D Rule Zion 1988 only - Sep 4 0:00 0 S # From Ephraim Silverberg # (1997-03-04, 1998-03-16, 1998-12-28, 2000-01-17, 2000-07-25, 2004-12-22, # and 2005-02-17): # According to the Office of the Secretary General of the Ministry of # Interior, there is NO set rule for Daylight-Savings/Standard time changes. # One thing is entrenched in law, however: that there must be at least 150 # days of daylight savings time annually. From 1993-1998, the change to # daylight savings time was on a Friday morning from midnight IST to # 1 a.m IDT; up until 1998, the change back to standard time was on a # Saturday night from midnight daylight savings time to 11 p.m. standard # time. 1996 is an exception to this rule where the change back to standard # time took place on Sunday night instead of Saturday night to avoid # conflicts with the Jewish New Year. In 1999, the change to # daylight savings time was still on a Friday morning but from # 2 a.m. IST to 3 a.m. IDT; furthermore, the change back to standard time # was also on a Friday morning from 2 a.m. IDT to 1 a.m. IST for # 1999 only. In the year 2000, the change to daylight savings time was # similar to 1999, but although the change back will be on a Friday, it # will take place from 1 a.m. IDT to midnight IST. Starting in 2001, all # changes to/from will take place at 1 a.m. old time, but now there is no # rule as to what day of the week it will take place in as the start date # (except in 2003) is the night after the Passover Seder (i.e. the eve # of the 16th of Nisan in the lunar Hebrew calendar) and the end date # (except in 2002) is three nights before Yom Kippur [Day of Atonement] # (the eve of the 7th of Tishrei in the lunar Hebrew calendar). # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Zion 1989 only - Apr 30 0:00 1:00 D Rule Zion 1989 only - Sep 3 0:00 0 S Rule Zion 1990 only - Mar 25 0:00 1:00 D Rule Zion 1990 only - Aug 26 0:00 0 S Rule Zion 1991 only - Mar 24 0:00 1:00 D Rule Zion 1991 only - Sep 1 0:00 0 S Rule Zion 1992 only - Mar 29 0:00 1:00 D Rule Zion 1992 only - Sep 6 0:00 0 S Rule Zion 1993 only - Apr 2 0:00 1:00 D Rule Zion 1993 only - Sep 5 0:00 0 S # The dates for 1994-1995 were obtained from Office of the Spokeswoman for the # Ministry of Interior, Jerusalem, Israel. The spokeswoman can be reached by # calling the office directly at 972-2-6701447 or 972-2-6701448. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Zion 1994 only - Apr 1 0:00 1:00 D Rule Zion 1994 only - Aug 28 0:00 0 S Rule Zion 1995 only - Mar 31 0:00 1:00 D Rule Zion 1995 only - Sep 3 0:00 0 S # The dates for 1996 were determined by the Minister of Interior of the # time, Haim Ramon. The official announcement regarding 1996-1998 # (with the dates for 1997-1998 no longer being relevant) can be viewed at: # # ftp://ftp.cs.huji.ac.il/pub/tz/announcements/1996-1998.ramon.ps.gz # # The dates for 1997-1998 were altered by his successor, Rabbi Eli Suissa. # # The official announcements for the years 1997-1999 can be viewed at: # # ftp://ftp.cs.huji.ac.il/pub/tz/announcements/YYYY.ps.gz # # where YYYY is the relevant year. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Zion 1996 only - Mar 15 0:00 1:00 D Rule Zion 1996 only - Sep 16 0:00 0 S Rule Zion 1997 only - Mar 21 0:00 1:00 D Rule Zion 1997 only - Sep 14 0:00 0 S Rule Zion 1998 only - Mar 20 0:00 1:00 D Rule Zion 1998 only - Sep 6 0:00 0 S Rule Zion 1999 only - Apr 2 2:00 1:00 D Rule Zion 1999 only - Sep 3 2:00 0 S # The Knesset Interior Committee has changed the dates for 2000 for # the third time in just over a year and have set new dates for the # years 2001-2004 as well. # # The official announcement for the start date of 2000 can be viewed at: # # ftp://ftp.cs.huji.ac.il/pub/tz/announcements/2000-start.ps.gz # # The official announcement for the end date of 2000 and the dates # for the years 2001-2004 can be viewed at: # # ftp://ftp.cs.huji.ac.il/pub/tz/announcements/2000-2004.ps.gz # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Zion 2000 only - Apr 14 2:00 1:00 D Rule Zion 2000 only - Oct 6 1:00 0 S Rule Zion 2001 only - Apr 9 1:00 1:00 D Rule Zion 2001 only - Sep 24 1:00 0 S Rule Zion 2002 only - Mar 29 1:00 1:00 D Rule Zion 2002 only - Oct 7 1:00 0 S Rule Zion 2003 only - Mar 28 1:00 1:00 D Rule Zion 2003 only - Oct 3 1:00 0 S Rule Zion 2004 only - Apr 7 1:00 1:00 D Rule Zion 2004 only - Sep 22 1:00 0 S # The proposed law agreed upon by the Knesset Interior Committee on # 2005-02-14 is that, for 2005 and beyond, DST starts at 02:00 the # last Friday before April 2nd (i.e. the last Friday in March or April # 1st itself if it falls on a Friday) and ends at 02:00 on the Saturday # night _before_ the fast of Yom Kippur. # # Those who can read Hebrew can view the announcement at: # # ftp://ftp.cs.huji.ac.il/pub/tz/announcements/2005+beyond.ps # From Paul Eggert (2012-10-26): # I used Ephraim Silverberg's dst-israel.el program # (2005-02-20) # along with Ed Reingold's cal-hebrew in GNU Emacs 21.4, # to generate the transitions from 2005 through 2012. # (I replaced "lastFri" with "Fri>=26" by hand.) # The spring transitions all correspond to the following Rule: # # Rule Zion 2005 2012 - Mar Fri>=26 2:00 1:00 D # # but older zic implementations (e.g., Solaris 8) do not support # "Fri>=26" to mean April 1 in years like 2005, so for now we list the # springtime transitions explicitly. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Zion 2005 only - Apr 1 2:00 1:00 D Rule Zion 2005 only - Oct 9 2:00 0 S Rule Zion 2006 2010 - Mar Fri>=26 2:00 1:00 D Rule Zion 2006 only - Oct 1 2:00 0 S Rule Zion 2007 only - Sep 16 2:00 0 S Rule Zion 2008 only - Oct 5 2:00 0 S Rule Zion 2009 only - Sep 27 2:00 0 S Rule Zion 2010 only - Sep 12 2:00 0 S Rule Zion 2011 only - Apr 1 2:00 1:00 D Rule Zion 2011 only - Oct 2 2:00 0 S Rule Zion 2012 only - Mar Fri>=26 2:00 1:00 D Rule Zion 2012 only - Sep 23 2:00 0 S # From Ephraim Silverberg (2013-06-27): # On June 23, 2013, the Israeli government approved changes to the # Time Decree Law. The next day, the changes passed the First Reading # in the Knesset. The law is expected to pass the Second and Third # (final) Readings by the beginning of September 2013. # # As of 2013, DST starts at 02:00 on the Friday before the last Sunday # in March. DST ends at 02:00 on the last Sunday of October. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Zion 2013 max - Mar Fri>=23 2:00 1:00 D Rule Zion 2013 max - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Jerusalem 2:20:54 - LMT 1880 2:20:40 - JMT 1918 # Jerusalem Mean Time? 2:00 Zion I%sT ############################################################################### # Japan # '9:00' and 'JST' is from Guy Harris. # From Paul Eggert (1995-03-06): # Today's _Asahi Evening News_ (page 4) reports that Japan had # daylight saving between 1948 and 1951, but "the system was discontinued # because the public believed it would lead to longer working hours." # From Mayumi Negishi in the 2005-08-10 Japan Times: # http://www.japantimes.co.jp/cgi-bin/getarticle.pl5?nn20050810f2.htm # Occupation authorities imposed daylight-saving time on Japan on # [1948-05-01].... But lack of prior debate and the execution of # daylight-saving time just three days after the bill was passed generated # deep hatred of the concept.... The Diet unceremoniously passed a bill to # dump the unpopular system in October 1951, less than a month after the San # Francisco Peace Treaty was signed. (A government poll in 1951 showed 53% # of the Japanese wanted to scrap daylight-saving time, as opposed to 30% who # wanted to keep it.) # From Paul Eggert (2006-03-22): # Shanks & Pottenger write that DST in Japan during those years was as follows: # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Japan 1948 only - May Sun>=1 2:00 1:00 D Rule Japan 1948 1951 - Sep Sat>=8 2:00 0 S Rule Japan 1949 only - Apr Sun>=1 2:00 1:00 D Rule Japan 1950 1951 - May Sun>=1 2:00 1:00 D # but the only locations using it (for birth certificates, presumably, since # their audience is astrologers) were US military bases. For now, assume # that for most purposes daylight-saving time was observed; otherwise, what # would have been the point of the 1951 poll? # From Hideyuki Suzuki (1998-11-09): # 'Tokyo' usually stands for the former location of Tokyo Astronomical # Observatory: 139 degrees 44' 40.90" E (9h 18m 58.727s), # 35 degrees 39' 16.0" N. # This data is from 'Rika Nenpyou (Chronological Scientific Tables) 1996' # edited by National Astronomical Observatory of Japan.... # JST (Japan Standard Time) has been used since 1888-01-01 00:00 (JST). # The law is enacted on 1886-07-07. # From Hideyuki Suzuki (1998-11-16): # The ordinance No. 51 (1886) established "standard time" in Japan, # which stands for the time on 135 degrees E. # In the ordinance No. 167 (1895), "standard time" was renamed to "central # standard time". And the same ordinance also established "western standard # time", which stands for the time on 120 degrees E.... But "western standard # time" was abolished in the ordinance No. 529 (1937). In the ordinance No. # 167, there is no mention regarding for what place western standard time is # standard.... # # I wrote "ordinance" above, but I don't know how to translate. # In Japanese it's "chokurei", which means ordinance from emperor. # From Yu-Cheng Chuang (2013-07-12): # ...the Meiji Emperor announced Ordinance No. 167 of Meiji Year 28 "The clause # about standard time" ... The adoption began from Jan 1, 1896. # http://ja.wikisource.org/wiki/標準時ニ關スル件_(公布時) # # ...the Showa Emperor announced Ordinance No. 529 of Showa Year 12 ... which # means the whole Japan territory, including later occupations, adopt Japan # Central Time (UTC+9). The adoption began on Oct 1, 1937. # http://ja.wikisource.org/wiki/明治二十八年勅令第百六十七號標準時ニ關スル件中改正ノ件 # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Tokyo 9:18:59 - LMT 1887 Dec 31 15:00u 9:00 - JST 1896 Jan 1 9:00 - JCST 1937 Oct 1 9:00 Japan J%sT # Since 1938, all Japanese possessions have been like Asia/Tokyo. # Jordan # # From # Jordan Week (1999-07-01) via Steffen Thorsen (1999-09-09): # Clocks in Jordan were forwarded one hour on Wednesday at midnight, # in accordance with the government's decision to implement summer time # all year round. # # From # Jordan Week (1999-09-30) via Steffen Thorsen (1999-11-09): # Winter time starts today Thursday, 30 September. Clocks will be turned back # by one hour. This is the latest government decision and it's final! # The decision was taken because of the increase in working hours in # government's departments from six to seven hours. # # From Paul Eggert (2005-11-22): # Starting 2003 transitions are from Steffen Thorsen's web site timeanddate.com. # # From Steffen Thorsen (2005-11-23): # For Jordan I have received multiple independent user reports every year # about DST end dates, as the end-rule is different every year. # # From Steffen Thorsen (2006-10-01), after a heads-up from Hilal Malawi: # http://www.petranews.gov.jo/nepras/2006/Sep/05/4000.htm # "Jordan will switch to winter time on Friday, October 27". # # From Steffen Thorsen (2009-04-02): # This single one might be good enough, (2009-03-24, Arabic): # http://petra.gov.jo/Artical.aspx?Lng=2&Section=8&Artical=95279 # # Google's translation: # # > The Council of Ministers decided in 2002 to adopt the principle of timely # > submission of the summer at 60 minutes as of midnight on the last Thursday # > of the month of March of each year. # # So - this means the midnight between Thursday and Friday since 2002. # From Arthur David Olson (2009-04-06): # We still have Jordan switching to DST on Thursdays in 2000 and 2001. # From Steffen Thorsen (2012-10-25): # Yesterday the government in Jordan announced that they will not # switch back to standard time this winter, so the will stay on DST # until about the same time next year (at least). # http://www.petra.gov.jo/Public_News/Nws_NewsDetails.aspx?NewsID=88950 # From Steffen Thorsen (2013-12-11): # Jordan Times and other sources say that Jordan is going back to # UTC+2 on 2013-12-19 at midnight: # http://jordantimes.com/govt-decides-to-switch-back-to-wintertime # Official, in Arabic: # http://www.petra.gov.jo/public_news/Nws_NewsDetails.aspx?Menu_ID=&Site_Id=2&lang=1&NewsID=133230&CatID=14 # ... Our background/permalink about it # http://www.timeanddate.com/news/time/jordan-reverses-dst-decision.html # ... # http://www.petra.gov.jo/Public_News/Nws_NewsDetails.aspx?lang=2&site_id=1&NewsID=133313&Type=P # ... says midnight for the coming one and 1:00 for the ones in the future # (and they will use DST again next year, using the normal schedule). # From Paul Eggert (2013-12-11): # As Steffen suggested, consider the past 21-month experiment to be DST. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Jordan 1973 only - Jun 6 0:00 1:00 S Rule Jordan 1973 1975 - Oct 1 0:00 0 - Rule Jordan 1974 1977 - May 1 0:00 1:00 S Rule Jordan 1976 only - Nov 1 0:00 0 - Rule Jordan 1977 only - Oct 1 0:00 0 - Rule Jordan 1978 only - Apr 30 0:00 1:00 S Rule Jordan 1978 only - Sep 30 0:00 0 - Rule Jordan 1985 only - Apr 1 0:00 1:00 S Rule Jordan 1985 only - Oct 1 0:00 0 - Rule Jordan 1986 1988 - Apr Fri>=1 0:00 1:00 S Rule Jordan 1986 1990 - Oct Fri>=1 0:00 0 - Rule Jordan 1989 only - May 8 0:00 1:00 S Rule Jordan 1990 only - Apr 27 0:00 1:00 S Rule Jordan 1991 only - Apr 17 0:00 1:00 S Rule Jordan 1991 only - Sep 27 0:00 0 - Rule Jordan 1992 only - Apr 10 0:00 1:00 S Rule Jordan 1992 1993 - Oct Fri>=1 0:00 0 - Rule Jordan 1993 1998 - Apr Fri>=1 0:00 1:00 S Rule Jordan 1994 only - Sep Fri>=15 0:00 0 - Rule Jordan 1995 1998 - Sep Fri>=15 0:00s 0 - Rule Jordan 1999 only - Jul 1 0:00s 1:00 S Rule Jordan 1999 2002 - Sep lastFri 0:00s 0 - Rule Jordan 2000 2001 - Mar lastThu 0:00s 1:00 S Rule Jordan 2002 2012 - Mar lastThu 24:00 1:00 S Rule Jordan 2003 only - Oct 24 0:00s 0 - Rule Jordan 2004 only - Oct 15 0:00s 0 - Rule Jordan 2005 only - Sep lastFri 0:00s 0 - Rule Jordan 2006 2011 - Oct lastFri 0:00s 0 - Rule Jordan 2013 only - Dec 20 0:00 0 - Rule Jordan 2014 max - Mar lastThu 24:00 1:00 S Rule Jordan 2014 max - Oct lastFri 0:00s 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Amman 2:23:44 - LMT 1931 2:00 Jordan EE%sT # Kazakhstan # From Paul Eggert (1996-11-22): # Andrew Evtichov (1996-04-13) writes that Kazakhstan # stayed in sync with Moscow after 1990, and that Aqtobe (formerly Aktyubinsk) # and Aqtau (formerly Shevchenko) are the largest cities in their zones. # Guess that Aqtau and Aqtobe diverged in 1995, since that's the first time # IATA SSIM mentions a third time zone in Kazakhstan. # From Paul Eggert (2006-03-22): # German Iofis, ELSI, Almaty (2001-10-09) reports that Kazakhstan uses # RussiaAsia rules, instead of switching at 00:00 as the IATA has it. # Go with Shanks & Pottenger, who have them always using RussiaAsia rules. # Also go with the following claims of Shanks & Pottenger: # # - Kazakhstan did not observe DST in 1991. # - Qyzylorda switched from +5:00 to +6:00 on 1992-01-19 02:00. # - Oral switched from +5:00 to +4:00 in spring 1989. # From Kazakhstan Embassy's News Bulletin #11 # (2005-03-21): # The Government of Kazakhstan passed a resolution March 15 abolishing # daylight saving time citing lack of economic benefits and health # complications coupled with a decrease in productivity. # # From Branislav Kojic (in Astana) via Gwillim Law (2005-06-28): # ... what happened was that the former Kazakhstan Eastern time zone # was "blended" with the Central zone. Therefore, Kazakhstan now has # two time zones, and difference between them is one hour. The zone # closer to UTC is the former Western zone (probably still called the # same), encompassing four provinces in the west: Aqtobe, Atyrau, # Mangghystau, and West Kazakhstan. The other zone encompasses # everything else.... I guess that would make Kazakhstan time zones # de jure UTC+5 and UTC+6 respectively. # # Zone NAME GMTOFF RULES FORMAT [UNTIL] # # Almaty (formerly Alma-Ata), representing most locations in Kazakhstan Zone Asia/Almaty 5:07:48 - LMT 1924 May 2 # or Alma-Ata 5:00 - ALMT 1930 Jun 21 # Alma-Ata Time 6:00 RussiaAsia ALM%sT 1991 6:00 - ALMT 1992 6:00 RussiaAsia ALM%sT 2005 Mar 15 6:00 - ALMT # Qyzylorda (aka Kyzylorda, Kizilorda, Kzyl-Orda, etc.) Zone Asia/Qyzylorda 4:21:52 - LMT 1924 May 2 4:00 - KIZT 1930 Jun 21 # Kizilorda Time 5:00 - KIZT 1981 Apr 1 5:00 1:00 KIZST 1981 Oct 1 6:00 - KIZT 1982 Apr 1 5:00 RussiaAsia KIZ%sT 1991 5:00 - KIZT 1991 Dec 16 # independence 5:00 - QYZT 1992 Jan 19 2:00 6:00 RussiaAsia QYZ%sT 2005 Mar 15 6:00 - QYZT # Aqtobe (aka Aktobe, formerly Aktyubinsk) Zone Asia/Aqtobe 3:48:40 - LMT 1924 May 2 4:00 - AKTT 1930 Jun 21 # Aktyubinsk Time 5:00 - AKTT 1981 Apr 1 5:00 1:00 AKTST 1981 Oct 1 6:00 - AKTT 1982 Apr 1 5:00 RussiaAsia AKT%sT 1991 5:00 - AKTT 1991 Dec 16 # independence 5:00 RussiaAsia AQT%sT 2005 Mar 15 # Aqtobe Time 5:00 - AQTT # Mangghystau # Aqtau was not founded until 1963, but it represents an inhabited region, # so include time stamps before 1963. Zone Asia/Aqtau 3:21:04 - LMT 1924 May 2 4:00 - FORT 1930 Jun 21 # Fort Shevchenko T 5:00 - FORT 1963 5:00 - SHET 1981 Oct 1 # Shevchenko Time 6:00 - SHET 1982 Apr 1 5:00 RussiaAsia SHE%sT 1991 5:00 - SHET 1991 Dec 16 # independence 5:00 RussiaAsia AQT%sT 1995 Mar lastSun 2:00 # Aqtau Time 4:00 RussiaAsia AQT%sT 2005 Mar 15 5:00 - AQTT # West Kazakhstan Zone Asia/Oral 3:25:24 - LMT 1924 May 2 # or Ural'sk 4:00 - URAT 1930 Jun 21 # Ural'sk time 5:00 - URAT 1981 Apr 1 5:00 1:00 URAST 1981 Oct 1 6:00 - URAT 1982 Apr 1 5:00 RussiaAsia URA%sT 1989 Mar 26 2:00 4:00 RussiaAsia URA%sT 1991 4:00 - URAT 1991 Dec 16 # independence 4:00 RussiaAsia ORA%sT 2005 Mar 15 # Oral Time 5:00 - ORAT # Kyrgyzstan (Kirgizstan) # Transitions through 1991 are from Shanks & Pottenger. # From Paul Eggert (2005-08-15): # According to an article dated today in the Kyrgyzstan Development Gateway # http://eng.gateway.kg/cgi-bin/page.pl?id=1&story_name=doc9979.shtml # Kyrgyzstan is canceling the daylight saving time system. I take the article # to mean that they will leave their clocks at 6 hours ahead of UTC. # From Malik Abdugaliev (2005-09-21): # Our government cancels daylight saving time 6th of August 2005. # From 2005-08-12 our GMT-offset is +6, w/o any daylight saving. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Kyrgyz 1992 1996 - Apr Sun>=7 0:00s 1:00 S Rule Kyrgyz 1992 1996 - Sep lastSun 0:00 0 - Rule Kyrgyz 1997 2005 - Mar lastSun 2:30 1:00 S Rule Kyrgyz 1997 2004 - Oct lastSun 2:30 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Bishkek 4:58:24 - LMT 1924 May 2 5:00 - FRUT 1930 Jun 21 # Frunze Time 6:00 RussiaAsia FRU%sT 1991 Mar 31 2:00s 5:00 1:00 FRUST 1991 Aug 31 2:00 # independence 5:00 Kyrgyz KG%sT 2005 Aug 12 # Kyrgyzstan Time 6:00 - KGT ############################################################################### # Korea (North and South) # From Annie I. Bang (2006-07-10): # http://www.koreaherald.com/view.php?ud=200607100012 # Korea ran a daylight saving program from 1949-61 but stopped it # during the 1950-53 Korean War. The system was temporarily enforced # between 1987 and 1988 ... # From Sanghyuk Jung (2014-10-29): # http://mm.icann.org/pipermail/tz/2014-October/021830.html # According to the Korean Wikipedia # http://ko.wikipedia.org/wiki/한국_표준시 # [oldid=12896437 2014-09-04 08:03 UTC] # DST in Republic of Korea was as follows.... And I checked old # newspapers in Korean, all articles correspond with data in Wikipedia. # For example, the article in 1948 (Korean Language) proved that DST # started at June 1 in that year. For another example, the article in # 1988 said that DST started at 2:00 AM in that year. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule ROK 1948 only - Jun 1 0:00 1:00 D Rule ROK 1948 only - Sep 13 0:00 0 S Rule ROK 1949 only - Apr 3 0:00 1:00 D Rule ROK 1949 1951 - Sep Sun>=8 0:00 0 S Rule ROK 1950 only - Apr 1 0:00 1:00 D Rule ROK 1951 only - May 6 0:00 1:00 D Rule ROK 1955 only - May 5 0:00 1:00 D Rule ROK 1955 only - Sep 9 0:00 0 S Rule ROK 1956 only - May 20 0:00 1:00 D Rule ROK 1956 only - Sep 30 0:00 0 S Rule ROK 1957 1960 - May Sun>=1 0:00 1:00 D Rule ROK 1957 1960 - Sep Sun>=18 0:00 0 S Rule ROK 1987 1988 - May Sun>=8 2:00 1:00 D Rule ROK 1987 1988 - Oct Sun>=8 3:00 0 S # From Paul Eggert (2014-10-30): # The Korean Wikipedia entry gives the following sources for UT offsets: # # 1908: Official Journal Article No. 3994 (Edict No. 5) # 1912: Governor-General of Korea Official Gazette Issue No. 367 # (Announcement No. 338) # 1954: Presidential Decree No. 876 (1954-03-17) # 1961: Law No. 676 (1961-08-07) # 1987: Law No. 3919 (1986-12-31) # # The Wikipedia entry also has confusing information about a change # to UT+9 in April 1910, but then what would be the point of the later change # to UT+9 on 1912-01-01? Omit the 1910 change for now. # # I guessed that time zone abbreviations through 1945 followed the same # rules as discussed under Taiwan, with nominal switches from JST to KST # when the respective cities were taken over by the Allies after WWII. # # For Pyongyang we have no information; guess no changes since World War II. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Seoul 8:27:52 - LMT 1908 Apr 1 8:30 - KST 1912 Jan 1 9:00 - JCST 1937 Oct 1 9:00 - JST 1945 Sep 8 9:00 - KST 1954 Mar 21 8:30 ROK K%sT 1961 Aug 10 9:00 ROK K%sT Zone Asia/Pyongyang 8:23:00 - LMT 1908 Apr 1 8:30 - KST 1912 Jan 1 9:00 - JCST 1937 Oct 1 9:00 - JST 1945 Aug 24 9:00 - KST ############################################################################### # Kuwait -# Zone NAME GMTOFF RULES FORMAT [UNTIL] -Zone Asia/Kuwait 3:11:56 - LMT 1950 - 3:00 - AST +# See Asia/Riyadh. # Laos # See Asia/Bangkok. # Lebanon # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Lebanon 1920 only - Mar 28 0:00 1:00 S Rule Lebanon 1920 only - Oct 25 0:00 0 - Rule Lebanon 1921 only - Apr 3 0:00 1:00 S Rule Lebanon 1921 only - Oct 3 0:00 0 - Rule Lebanon 1922 only - Mar 26 0:00 1:00 S Rule Lebanon 1922 only - Oct 8 0:00 0 - Rule Lebanon 1923 only - Apr 22 0:00 1:00 S Rule Lebanon 1923 only - Sep 16 0:00 0 - Rule Lebanon 1957 1961 - May 1 0:00 1:00 S Rule Lebanon 1957 1961 - Oct 1 0:00 0 - Rule Lebanon 1972 only - Jun 22 0:00 1:00 S Rule Lebanon 1972 1977 - Oct 1 0:00 0 - Rule Lebanon 1973 1977 - May 1 0:00 1:00 S Rule Lebanon 1978 only - Apr 30 0:00 1:00 S Rule Lebanon 1978 only - Sep 30 0:00 0 - Rule Lebanon 1984 1987 - May 1 0:00 1:00 S Rule Lebanon 1984 1991 - Oct 16 0:00 0 - Rule Lebanon 1988 only - Jun 1 0:00 1:00 S Rule Lebanon 1989 only - May 10 0:00 1:00 S Rule Lebanon 1990 1992 - May 1 0:00 1:00 S Rule Lebanon 1992 only - Oct 4 0:00 0 - Rule Lebanon 1993 max - Mar lastSun 0:00 1:00 S Rule Lebanon 1993 1998 - Sep lastSun 0:00 0 - Rule Lebanon 1999 max - Oct lastSun 0:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Beirut 2:22:00 - LMT 1880 2:00 Lebanon EE%sT # Malaysia # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule NBorneo 1935 1941 - Sep 14 0:00 0:20 TS # one-Third Summer Rule NBorneo 1935 1941 - Dec 14 0:00 0 - # # peninsular Malaysia # taken from Mok Ly Yng (2003-10-30) # http://www.math.nus.edu.sg/aslaksen/teaching/timezone.html # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Kuala_Lumpur 6:46:46 - LMT 1901 Jan 1 6:55:25 - SMT 1905 Jun 1 # Singapore M.T. 7:00 - MALT 1933 Jan 1 # Malaya Time 7:00 0:20 MALST 1936 Jan 1 7:20 - MALT 1941 Sep 1 7:30 - MALT 1942 Feb 16 9:00 - JST 1945 Sep 12 7:30 - MALT 1982 Jan 1 8:00 - MYT # Malaysia Time # Sabah & Sarawak # From Paul Eggert (2014-08-12): # The data entries here are mostly from Shanks & Pottenger, but the 1942, 1945 # and 1982 transition dates are from Mok Ly Yng. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Kuching 7:21:20 - LMT 1926 Mar 7:30 - BORT 1933 # Borneo Time 8:00 NBorneo BOR%sT 1942 Feb 16 9:00 - JST 1945 Sep 12 8:00 - BORT 1982 Jan 1 8:00 - MYT # Maldives # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Indian/Maldives 4:54:00 - LMT 1880 # Male 4:54:00 - MMT 1960 # Male Mean Time 5:00 - MVT # Maldives Time # Mongolia # Shanks & Pottenger say that Mongolia has three time zones, but # The USNO (1995-12-21) and the CIA map Standard Time Zones of the World # (2005-03) both say that it has just one. # From Oscar van Vlijmen (1999-12-11): # General Information Mongolia # (1999-09) # "Time: Mongolia has two time zones. Three westernmost provinces of # Bayan-Ölgii, Uvs, and Hovd are one hour earlier than the capital city, and # the rest of the country follows the Ulaanbaatar time, which is UTC/GMT plus # eight hours." # From Rives McDow (1999-12-13): # Mongolia discontinued the use of daylight savings time in 1999; 1998 # being the last year it was implemented. The dates of implementation I am # unsure of, but most probably it was similar to Russia, except for the time # of implementation may have been different.... # Some maps in the past have indicated that there was an additional time # zone in the eastern part of Mongolia, including the provinces of Dornod, # Sükhbaatar, and possibly Khentii. # From Paul Eggert (1999-12-15): # Naming and spelling is tricky in Mongolia. # We'll use Hovd (also spelled Chovd and Khovd) to represent the west zone; # the capital of the Hovd province is sometimes called Hovd, sometimes Dund-Us, # and sometimes Jirgalanta (with variant spellings), but the name Hovd # is good enough for our purposes. # From Rives McDow (2001-05-13): # In addition to Mongolia starting daylight savings as reported earlier # (adopted DST on 2001-04-27 02:00 local time, ending 2001-09-28), # there are three time zones. # # Provinces [at 7:00]: Bayan-Ölgii, Uvs, Khovd, Zavkhan, Govi-Altai # Provinces [at 8:00]: Khövsgöl, Bulgan, Arkhangai, Khentii, Töv, # Bayankhongor, Övörkhangai, Dundgovi, Dornogovi, Ömnögovi # Provinces [at 9:00]: Dornod, Sükhbaatar # # [The province of Selenge is omitted from the above lists.] # From Ganbold Ts., Ulaanbaatar (2004-04-17): # Daylight saving occurs at 02:00 local time last Saturday of March. # It will change back to normal at 02:00 local time last Saturday of # September.... As I remember this rule was changed in 2001. # # From Paul Eggert (2004-04-17): # For now, assume Rives McDow's informant got confused about Friday vs # Saturday, and that his 2001 dates should have 1 added to them. # From Paul Eggert (2005-07-26): # We have wildly conflicting information about Mongolia's time zones. # Bill Bonnet (2005-05-19) reports that the US Embassy in Ulaanbaatar says # there is only one time zone and that DST is observed, citing Microsoft # Windows XP as the source. Risto Nykänen (2005-05-16) reports that # travelmongolia.org says there are two time zones (UTC+7, UTC+8) with no DST. # Oscar van Vlijmen (2005-05-20) reports that the Mongolian Embassy in # Washington, DC says there are two time zones, with DST observed. # He also found # http://ubpost.mongolnews.mn/index.php?subaction=showcomments&id=1111634894&archive=&start_from=&ucat=1& # which also says that there is DST, and which has a comment by "Toddius" # (2005-03-31 06:05 +0700) saying "Mongolia actually has 3.5 time zones. # The West (OLGII) is +7 GMT, most of the country is ULAT is +8 GMT # and some Eastern provinces are +9 GMT but Sükhbaatar Aimag is SUHK +8.5 GMT. # The SUKH timezone is new this year, it is one of the few things the # parliament passed during the tumultuous winter session." # For now, let's ignore this information, until we have more confirmation. # From Ganbold Ts. (2007-02-26): # Parliament of Mongolia has just changed the daylight-saving rule in February. # They decided not to adopt daylight-saving time.... # http://www.mongolnews.mn/index.php?module=unuudur&sec=view&id=15742 # From Deborah Goldsmith (2008-03-30): # We received a bug report claiming that the tz database UTC offset for # Asia/Choibalsan (GMT+09:00) is incorrect, and that it should be GMT # +08:00 instead. Different sources appear to disagree with the tz # database on this, e.g.: # # http://www.timeanddate.com/worldclock/city.html?n=1026 # http://www.worldtimeserver.com/current_time_in_MN.aspx # # both say GMT+08:00. # From Steffen Thorsen (2008-03-31): # eznis airways, which operates several domestic flights, has a flight # schedule here: # http://www.eznis.com/Container.jsp?id=112 # (click the English flag for English) # # There it appears that flights between Choibalsan and Ulaanbaatar arrive # about 1:35 - 1:50 hours later in local clock time, no matter the # direction, while Ulaanbaatar-Khovd takes 2 hours in the Eastern # direction and 3:35 back, which indicates that Ulaanbaatar and Khovd are # in different time zones (like we know about), while Choibalsan and # Ulaanbaatar are in the same time zone (correction needed). # From Arthur David Olson (2008-05-19): # Assume that Choibalsan is indeed offset by 8:00. # XXX--in the absence of better information, assume that transition # was at the start of 2008-03-31 (the day of Steffen Thorsen's report); # this is almost surely wrong. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Mongol 1983 1984 - Apr 1 0:00 1:00 S Rule Mongol 1983 only - Oct 1 0:00 0 - # Shanks & Pottenger and IATA SSIM say 1990s switches occurred at 00:00, # but McDow says the 2001 switches occurred at 02:00. Also, IATA SSIM # (1996-09) says 1996-10-25. Go with Shanks & Pottenger through 1998. # # Shanks & Pottenger say that the Sept. 1984 through Sept. 1990 switches # in Choibalsan (more precisely, in Dornod and Sükhbaatar) took place # at 02:00 standard time, not at 00:00 local time as in the rest of # the country. That would be odd, and possibly is a result of their # correction of 02:00 (in the previous edition) not being done correctly # in the latest edition; so ignore it for now. Rule Mongol 1985 1998 - Mar lastSun 0:00 1:00 S Rule Mongol 1984 1998 - Sep lastSun 0:00 0 - # IATA SSIM (1999-09) says Mongolia no longer observes DST. Rule Mongol 2001 only - Apr lastSat 2:00 1:00 S Rule Mongol 2001 2006 - Sep lastSat 2:00 0 - Rule Mongol 2002 2006 - Mar lastSat 2:00 1:00 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] # Hovd, a.k.a. Chovd, Dund-Us, Dzhargalant, Khovd, Jirgalanta Zone Asia/Hovd 6:06:36 - LMT 1905 Aug 6:00 - HOVT 1978 # Hovd Time 7:00 Mongol HOV%sT # Ulaanbaatar, a.k.a. Ulan Bataar, Ulan Bator, Urga Zone Asia/Ulaanbaatar 7:07:32 - LMT 1905 Aug 7:00 - ULAT 1978 # Ulaanbaatar Time 8:00 Mongol ULA%sT # Choibalsan, a.k.a. Bajan Tümen, Bajan Tumen, Chojbalsan, # Choybalsan, Sanbejse, Tchoibalsan Zone Asia/Choibalsan 7:38:00 - LMT 1905 Aug 7:00 - ULAT 1978 8:00 - ULAT 1983 Apr 9:00 Mongol CHO%sT 2008 Mar 31 # Choibalsan Time 8:00 Mongol CHO%sT # Nepal # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Kathmandu 5:41:16 - LMT 1920 5:30 - IST 1986 5:45 - NPT # Nepal Time # Oman +# See Asia/Dubai. -# Milne says 3:54:24 was the meridian of the Muscat Tidal Observatory. - -# Zone NAME GMTOFF RULES FORMAT [UNTIL] -Zone Asia/Muscat 3:54:24 - LMT 1920 - 4:00 - GST - # Pakistan # From Rives McDow (2002-03-13): # I have been advised that Pakistan has decided to adopt dst on a # TRIAL basis for one year, starting 00:01 local time on April 7, 2002 # and ending at 00:01 local time October 6, 2002. This is what I was # told, but I believe that the actual time of change may be 00:00; the # 00:01 was to make it clear which day it was on. # From Paul Eggert (2002-03-15): # Jesper Nørgaard found this URL: # http://www.pak.gov.pk/public/news/app/app06_dec.htm # (dated 2001-12-06) which says that the Cabinet adopted a scheme "to # advance the clocks by one hour on the night between the first # Saturday and Sunday of April and revert to the original position on # 15th October each year". This agrees with McDow's 04-07 at 00:00, # but disagrees about the October transition, and makes it sound like # it's not on a trial basis. Also, the "between the first Saturday # and Sunday of April" phrase, if taken literally, means that the # transition takes place at 00:00 on the first Sunday on or after 04-02. # From Paul Eggert (2003-02-09): # DAWN reported on 2002-10-05 # that 2002 DST ended that day at midnight. Go with McDow for now. # From Steffen Thorsen (2003-03-14): # According to http://www.dawn.com/2003/03/07/top15.htm # there will be no DST in Pakistan this year: # # ISLAMABAD, March 6: Information and Media Development Minister Sheikh # Rashid Ahmed on Thursday said the cabinet had reversed a previous # decision to advance clocks by one hour in summer and put them back by # one hour in winter with the aim of saving light hours and energy. # # The minister told a news conference that the experiment had rather # shown 8 per cent higher consumption of electricity. # From Alex Krivenyshev (2008-05-15): # # Here is an article that Pakistan plan to introduce Daylight Saving Time # on June 1, 2008 for 3 months. # # "... The federal cabinet on Wednesday announced a new conservation plan to # help reduce load shedding by approving the closure of commercial centres at # 9pm and moving clocks forward by one hour for the next three months. ...." # # http://www.worldtimezone.net/dst_news/dst_news_pakistan01.html # http://www.dailytimes.com.pk/default.asp?page=2008%5C05%5C15%5Cstory_15-5-2008_pg1_4 # From Arthur David Olson (2008-05-19): # XXX--midnight transitions is a guess; 2008 only is a guess. # From Alexander Krivenyshev (2008-08-28): # Pakistan government has decided to keep the watches one-hour advanced # for another 2 months - plan to return to Standard Time on October 31 # instead of August 31. # # http://www.worldtimezone.com/dst_news/dst_news_pakistan02.html # http://dailymailnews.com/200808/28/news/dmbrn03.html # From Alexander Krivenyshev (2009-04-08): # Based on previous media reports that "... proposed plan to # advance clocks by one hour from May 1 will cause disturbance # to the working schedules rather than bringing discipline in # official working." # http://www.thenews.com.pk/daily_detail.asp?id=171280 # # recent news that instead of May 2009 - Pakistan plan to # introduce DST from April 15, 2009 # # FYI: Associated Press Of Pakistan # April 08, 2009 # Cabinet okays proposal to advance clocks by one hour from April 15 # http://www.app.com.pk/en_/index.php?option=com_content&task=view&id=73043&Itemid=1 # http://www.worldtimezone.com/dst_news/dst_news_pakistan05.html # # .... # The Federal Cabinet on Wednesday approved the proposal to # advance clocks in the country by one hour from April 15 to # conserve energy" # From Steffen Thorsen (2009-09-17): # "The News International," Pakistan reports that: "The Federal # Government has decided to restore the previous time by moving the # clocks backward by one hour from October 1. A formal announcement to # this effect will be made after the Prime Minister grants approval in # this regard." # http://www.thenews.com.pk/updates.asp?id=87168 # From Alexander Krivenyshev (2009-09-28): # According to Associated Press Of Pakistan, it is confirmed that # Pakistan clocks across the country would be turned back by an hour from # October 1, 2009. # # "Clocks to go back one hour from 1 Oct" # http://www.app.com.pk/en_/index.php?option=com_content&task=view&id=86715&Itemid=2 # http://www.worldtimezone.com/dst_news/dst_news_pakistan07.htm # # From Steffen Thorsen (2009-09-29): # Now they seem to have changed their mind, November 1 is the new date: # http://www.thenews.com.pk/top_story_detail.asp?Id=24742 # "The country's clocks will be reversed by one hour on November 1. # Officials of Federal Ministry for Interior told this to Geo News on # Monday." # # And more importantly, it seems that these dates will be kept every year: # "It has now been decided that clocks will be wound forward by one hour # on April 15 and reversed by an hour on November 1 every year without # obtaining prior approval, the officials added." # # We have confirmed this year's end date with both with the Ministry of # Water and Power and the Pakistan Electric Power Company: # http://www.timeanddate.com/news/time/pakistan-ends-dst09.html # From Christoph Göhre (2009-10-01): # [T]he German Consulate General in Karachi reported me today that Pakistan # will go back to standard time on 1st of November. # From Steffen Thorsen (2010-03-26): # Steffen Thorsen wrote: # > On Thursday (2010-03-25) it was announced that DST would start in # > Pakistan on 2010-04-01. # > # > Then today, the president said that they might have to revert the # > decision if it is not supported by the parliament. So at the time # > being, it seems unclear if DST will be actually observed or not - but # > April 1 could be a more likely date than April 15. # Now, it seems that the decision to not observe DST in final: # # "Govt Withdraws Plan To Advance Clocks" # http://www.apakistannews.com/govt-withdraws-plan-to-advance-clocks-172041 # # "People laud PM's announcement to end DST" # http://www.app.com.pk/en_/index.php?option=com_content&task=view&id=99374&Itemid=2 # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Pakistan 2002 only - Apr Sun>=2 0:01 1:00 S Rule Pakistan 2002 only - Oct Sun>=2 0:01 0 - Rule Pakistan 2008 only - Jun 1 0:00 1:00 S Rule Pakistan 2008 2009 - Nov 1 0:00 0 - Rule Pakistan 2009 only - Apr 15 0:00 1:00 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Karachi 4:28:12 - LMT 1907 5:30 - IST 1942 Sep 5:30 1:00 IST 1945 Oct 15 5:30 - IST 1951 Sep 30 5:00 - KART 1971 Mar 26 # Karachi Time 5:00 Pakistan PK%sT # Pakistan Time # Palestine # From Amos Shapir (1998-02-15): # # From 1917 until 1948-05-15, all of Palestine, including the parts now # known as the Gaza Strip and the West Bank, was under British rule. # Therefore the rules given for Israel for that period, apply there too... # # The Gaza Strip was under Egyptian rule between 1948-05-15 until 1967-06-05 # (except a short occupation by Israel from 1956-11 till 1957-03, but no # time zone was affected then). It was never formally annexed to Egypt, # though. # # The rest of Palestine was under Jordanian rule at that time, formally # annexed in 1950 as the West Bank (and the word "Trans" was dropped from # the country's previous name of "the Hashemite Kingdom of the # Trans-Jordan"). So the rules for Jordan for that time apply. Major # towns in that area are Nablus (Shchem), El-Halil (Hebron), Ramallah, and # East Jerusalem. # # Both areas were occupied by Israel in June 1967, but not annexed (except # for East Jerusalem). They were on Israel time since then; there might # have been a Military Governor's order about time zones, but I'm not aware # of any (such orders may have been issued semi-annually whenever summer # time was in effect, but maybe the legal aspect of time was just neglected). # # The Palestinian Authority was established in 1993, and got hold of most # towns in the West Bank and Gaza by 1995. I know that in order to # demonstrate...independence, they have been switching to # summer time and back on a different schedule than Israel's, but I don't # know when this was started, or what algorithm is used (most likely the # Jordanian one). # # To summarize, the table should probably look something like that: # # Area \ when | 1918-1947 | 1948-1967 | 1967-1995 | 1996- # ------------+-----------+-----------+-----------+----------- # Israel | Zion | Zion | Zion | Zion # West bank | Zion | Jordan | Zion | Jordan # Gaza | Zion | Egypt | Zion | Jordan # # I guess more info may be available from the PA's web page (if/when they # have one). # From Paul Eggert (2006-03-22): # Shanks & Pottenger write that Gaza did not observe DST until 1957, but go # with Shapir and assume that it observed DST from 1940 through 1947, # and that it used Jordanian rules starting in 1996. # We don't yet need a separate entry for the West Bank, since # the only differences between it and Gaza that we know about # occurred before our cutoff date of 1970. # However, as we get more information, we may need to add entries # for parts of the West Bank as they transitioned from Israel's rules # to Palestine's rules. # From IINS News Service - Israel - 1998-03-23 10:38:07 Israel time, # forwarded by Ephraim Silverberg: # # Despite the fact that Israel changed over to daylight savings time # last week, the PLO Authority (PA) has decided not to turn its clocks # one-hour forward at this time. As a sign of independence from Israeli rule, # the PA has decided to implement DST in April. # From Paul Eggert (1999-09-20): # Daoud Kuttab writes in Holiday havoc # http://www.jpost.com/com/Archive/22.Apr.1999/Opinion/Article-2.html # (Jerusalem Post, 1999-04-22) that # the Palestinian National Authority changed to DST on 1999-04-15. # I vaguely recall that they switch back in October (sorry, forgot the source). # For now, let's assume that the spring switch was at 24:00, # and that they switch at 0:00 on the 3rd Fridays of April and October. # From Paul Eggert (2005-11-22): # Starting 2004 transitions are from Steffen Thorsen's web site timeanddate.com. # From Steffen Thorsen (2005-11-23): # A user from Gaza reported that Gaza made the change early because of # the Ramadan. Next year Ramadan will be even earlier, so I think # there is a good chance next year's end date will be around two weeks # earlier - the same goes for Jordan. # From Steffen Thorsen (2006-08-17): # I was informed by a user in Bethlehem that in Bethlehem it started the # same day as Israel, and after checking with other users in the area, I # was informed that they started DST one day after Israel. I was not # able to find any authoritative sources at the time, nor details if # Gaza changed as well, but presumed Gaza to follow the same rules as # the West Bank. # From Steffen Thorsen (2006-09-26): # according to the Palestine News Network (2006-09-19): # http://english.pnn.ps/index.php?option=com_content&task=view&id=596&Itemid=5 # > The Council of Ministers announced that this year its winter schedule # > will begin early, as of midnight Thursday. It is also time to turn # > back the clocks for winter. Friday will begin an hour late this week. # I guess it is likely that next year's date will be moved as well, # because of the Ramadan. # From Jesper Nørgaard Welen (2007-09-18): # According to Steffen Thorsen's web site the Gaza Strip and the rest of the # Palestinian territories left DST early on 13.th. of September at 2:00. # From Paul Eggert (2007-09-20): # My understanding is that Gaza and the West Bank disagree even over when # the weekend is (Thursday+Friday versus Friday+Saturday), so I'd be a bit # surprised if they agreed about DST. But for now, assume they agree. # For lack of better information, predict that future changes will be # the 2nd Thursday of September at 02:00. # From Alexander Krivenyshev (2008-08-28): # Here is an article, that Mideast running on different clocks at Ramadan. # # Gaza Strip (as Egypt) ended DST at midnight Thursday (Aug 28, 2008), while # the West Bank will end Daylight Saving Time at midnight Sunday (Aug 31, 2008). # # http://www.guardian.co.uk/world/feedarticle/7759001 # http://www.abcnews.go.com/International/wireStory?id=5676087 # http://www.worldtimezone.com/dst_news/dst_news_gazastrip01.html # From Alexander Krivenyshev (2009-03-26): # According to the Palestine News Network (arabic.pnn.ps), Palestinian # government decided to start Daylight Time on Thursday night March # 26 and continue until the night of 27 September 2009. # # (in Arabic) # http://arabic.pnn.ps/index.php?option=com_content&task=view&id=50850 # # (English translation) # http://www.worldtimezone.com/dst_news/dst_news_westbank01.html # From Steffen Thorsen (2009-08-31): # Palestine's Council of Ministers announced that they will revert back to # winter time on Friday, 2009-09-04. # # One news source: # http://www.safa.ps/ara/?action=showdetail&seid=4158 # (Palestinian press agency, Arabic), # Google translate: "Decided that the Palestinian government in Ramallah # headed by Salam Fayyad, the start of work in time for the winter of # 2009, starting on Friday approved the fourth delay Sept. clock sixty # minutes per hour as of Friday morning." # # We are not sure if Gaza will do the same, last year they had a different # end date, we will keep this page updated: # http://www.timeanddate.com/news/time/westbank-gaza-dst-2009.html # From Alexander Krivenyshev (2009-09-02): # Seems that Gaza Strip will go back to Winter Time same date as West Bank. # # According to Palestinian Ministry Of Interior, West Bank and Gaza Strip plan # to change time back to Standard time on September 4, 2009. # # "Winter time unite the West Bank and Gaza" # (from Palestinian National Authority): # http://www.moi.gov.ps/en/?page=633167343250594025&nid=11505 # http://www.worldtimezone.com/dst_news/dst_news_gazastrip02.html # From Alexander Krivenyshev (2010-03-19): # According to Voice of Palestine DST will last for 191 days, from March # 26, 2010 till "the last Sunday before the tenth day of Tishri # (October), each year" (October 03, 2010?) # # http://palvoice.org/forums/showthread.php?t=245697 # (in Arabic) # http://www.worldtimezone.com/dst_news/dst_news_westbank03.html # From Steffen Thorsen (2010-03-24): # ...Ma'an News Agency reports that Hamas cabinet has decided it will # start one day later, at 12:01am. Not sure if they really mean 12:01am or # noon though: # # http://www.maannews.net/eng/ViewDetails.aspx?ID=271178 # (Ma'an News Agency) # "At 12:01am Friday, clocks in Israel and the West Bank will change to # 1:01am, while Gaza clocks will change at 12:01am Saturday morning." # From Steffen Thorsen (2010-08-11): # According to several sources, including # http://www.maannews.net/eng/ViewDetails.aspx?ID=306795 # the clocks were set back one hour at 2010-08-11 00:00:00 local time in # Gaza and the West Bank. # Some more background info: # http://www.timeanddate.com/news/time/westbank-gaza-end-dst-2010.html # From Steffen Thorsen (2011-08-26): # Gaza and the West Bank did go back to standard time in the beginning of # August, and will now enter daylight saving time again on 2011-08-30 # 00:00 (so two periods of DST in 2011). The pause was because of # Ramadan. # # http://www.maannews.net/eng/ViewDetails.aspx?ID=416217 # Additional info: # http://www.timeanddate.com/news/time/palestine-dst-2011.html # From Alexander Krivenyshev (2011-08-27): # According to the article in The Jerusalem Post: # "...Earlier this month, the Palestinian government in the West Bank decided to # move to standard time for 30 days, during Ramadan. The Palestinians in the # Gaza Strip accepted the change and also moved their clocks one hour back. # The Hamas government said on Saturday that it won't observe summertime after # the Muslim feast of Id al-Fitr, which begins on Tuesday..." # ... # http://www.jpost.com/MiddleEast/Article.aspx?id=235650 # http://www.worldtimezone.com/dst_news/dst_news_gazastrip05.html # The rules for Egypt are stolen from the 'africa' file. # From Steffen Thorsen (2011-09-30): # West Bank did end Daylight Saving Time this morning/midnight (2011-09-30 # 00:00). # So West Bank and Gaza now have the same time again. # # Many sources, including: # http://www.maannews.net/eng/ViewDetails.aspx?ID=424808 # From Steffen Thorsen (2012-03-26): # Palestinian news sources tell that both Gaza and West Bank will start DST # on Friday (Thursday midnight, 2012-03-29 24:00). # Some of many sources in Arabic: # http://www.samanews.com/index.php?act=Show&id=122638 # # http://safa.ps/details/news/74352/%D8%A8%D8%AF%D8%A1-%D8%A7%D9%84%D8%AA%D9%88%D9%82%D9%8A%D8%AA-%D8%A7%D9%84%D8%B5%D9%8A%D9%81%D9%8A-%D8%A8%D8%A7%D9%84%D8%B6%D9%81%D8%A9-%D9%88%D8%BA%D8%B2%D8%A9-%D9%84%D9%8A%D9%84%D8%A9-%D8%A7%D9%84%D8%AC%D9%85%D8%B9%D8%A9.html # # Our brief summary: # http://www.timeanddate.com/news/time/gaza-west-bank-dst-2012.html # From Steffen Thorsen (2013-03-26): # The following news sources tells that Palestine will "start daylight saving # time from midnight on Friday, March 29, 2013" (translated). # [These are in Arabic and are for Gaza and for Ramallah, respectively.] # http://www.samanews.com/index.php?act=Show&id=154120 # http://safa.ps/details/news/99844/%D8%B1%D8%A7%D9%85-%D8%A7%D9%84%D9%84%D9%87-%D8%A8%D8%AF%D8%A1-%D8%A7%D9%84%D8%AA%D9%88%D9%82%D9%8A%D8%AA-%D8%A7%D9%84%D8%B5%D9%8A%D9%81%D9%8A-29-%D8%A7%D9%84%D8%AC%D8%A7%D8%B1%D9%8A.html # From Steffen Thorsen (2013-09-24): # The Gaza and West Bank are ending DST Thursday at midnight # (2013-09-27 00:00:00) (one hour earlier than last year...). # This source in English, says "that winter time will go into effect # at midnight on Thursday in the West Bank and Gaza Strip": # http://english.wafa.ps/index.php?action=detail&id=23246 # official source...: # http://www.palestinecabinet.gov.ps/ar/Views/ViewDetails.aspx?pid=1252 # From Paul Eggert (2013-09-24): # For future dates, guess the last Thursday in March at 24:00 through # the first Friday on or after September 21 at 00:00. This is consistent with # the predictions in today's editions of the following URLs, # which are for Gaza and Hebron respectively: # http://www.timeanddate.com/worldclock/timezone.html?n=702 # http://www.timeanddate.com/worldclock/timezone.html?n=2364 # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule EgyptAsia 1957 only - May 10 0:00 1:00 S Rule EgyptAsia 1957 1958 - Oct 1 0:00 0 - Rule EgyptAsia 1958 only - May 1 0:00 1:00 S Rule EgyptAsia 1959 1967 - May 1 1:00 1:00 S Rule EgyptAsia 1959 1965 - Sep 30 3:00 0 - Rule EgyptAsia 1966 only - Oct 1 3:00 0 - Rule Palestine 1999 2005 - Apr Fri>=15 0:00 1:00 S Rule Palestine 1999 2003 - Oct Fri>=15 0:00 0 - Rule Palestine 2004 only - Oct 1 1:00 0 - Rule Palestine 2005 only - Oct 4 2:00 0 - Rule Palestine 2006 2007 - Apr 1 0:00 1:00 S Rule Palestine 2006 only - Sep 22 0:00 0 - Rule Palestine 2007 only - Sep Thu>=8 2:00 0 - Rule Palestine 2008 2009 - Mar lastFri 0:00 1:00 S Rule Palestine 2008 only - Sep 1 0:00 0 - Rule Palestine 2009 only - Sep Fri>=1 1:00 0 - Rule Palestine 2010 only - Mar 26 0:00 1:00 S Rule Palestine 2010 only - Aug 11 0:00 0 - Rule Palestine 2011 only - Apr 1 0:01 1:00 S Rule Palestine 2011 only - Aug 1 0:00 0 - Rule Palestine 2011 only - Aug 30 0:00 1:00 S Rule Palestine 2011 only - Sep 30 0:00 0 - Rule Palestine 2012 max - Mar lastThu 24:00 1:00 S Rule Palestine 2012 only - Sep 21 1:00 0 - Rule Palestine 2013 max - Sep Fri>=21 0:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Gaza 2:17:52 - LMT 1900 Oct 2:00 Zion EET 1948 May 15 2:00 EgyptAsia EE%sT 1967 Jun 5 2:00 Zion I%sT 1996 2:00 Jordan EE%sT 1999 2:00 Palestine EE%sT 2008 Aug 29 0:00 2:00 - EET 2008 Sep 2:00 Palestine EE%sT 2010 2:00 - EET 2010 Mar 27 0:01 2:00 Palestine EE%sT 2011 Aug 1 2:00 - EET 2012 2:00 Palestine EE%sT Zone Asia/Hebron 2:20:23 - LMT 1900 Oct 2:00 Zion EET 1948 May 15 2:00 EgyptAsia EE%sT 1967 Jun 5 2:00 Zion I%sT 1996 2:00 Jordan EE%sT 1999 2:00 Palestine EE%sT # Paracel Is # no information # Philippines # On 1844-08-16, Narciso Clavería, governor-general of the # Philippines, issued a proclamation announcing that 1844-12-30 was to # be immediately followed by 1845-01-01; see R.H. van Gent's # History of the International Date Line # http://www.staff.science.uu.nl/~gent0113/idl/idl_philippines.htm # The rest of the data entries are from Shanks & Pottenger. # From Jesper Nørgaard Welen (2006-04-26): # ... claims that Philippines had DST last time in 1990: # http://story.philippinetimes.com/p.x/ct/9/id/145be20cc6b121c0/cid/3e5bbccc730d258c/ # [a story dated 2006-04-25 by Cris Larano of Dow Jones Newswires, # but no details] # From Paul Eggert (2014-08-14): # The following source says DST may be instituted November-January and again # March-June, but this is not definite. It also says DST was last proclaimed # during the Ramos administration (1992-1998); but again, no details. # Carcamo D. PNoy urged to declare use of daylight saving time. # Philippine Star 2014-08-05 # http://www.philstar.com/headlines/2014/08/05/1354152/pnoy-urged-declare-use-daylight-saving-time # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Phil 1936 only - Nov 1 0:00 1:00 S Rule Phil 1937 only - Feb 1 0:00 0 - Rule Phil 1954 only - Apr 12 0:00 1:00 S Rule Phil 1954 only - Jul 1 0:00 0 - Rule Phil 1978 only - Mar 22 0:00 1:00 S Rule Phil 1978 only - Sep 21 0:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Manila -15:56:00 - LMT 1844 Dec 31 8:04:00 - LMT 1899 May 11 8:00 Phil PH%sT 1942 May 9:00 - JST 1944 Nov 8:00 Phil PH%sT # Qatar # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Qatar 3:26:08 - LMT 1920 # Al Dawhah / Doha 4:00 - GST 1972 Jun 3:00 - AST +Link Asia/Qatar Asia/Bahrain # Saudi Arabia # # From Paul Eggert (2014-07-15): # Time in Saudi Arabia and other countries in the Arabian peninsula was not # standardized until relatively recently; we don't know when, and possibly it # has never been made official. Richard P Hunt, in "Islam city yielding to # modern times", New York Times (1961-04-09), p 20, wrote that only airlines # observed standard time, and that people in Jeddah mostly observed quasi-solar # time, doing so by setting their watches at sunrise to 6 o'clock (or to 12 # o'clock for "Arab" time). # # The TZ database cannot represent quasi-solar time; airline time is the best # we can do. The 1946 foreign air news digest of the U.S. Civil Aeronautics # Board (OCLC 42299995) reported that the "... Arabian Government, inaugurated # a weekly Dhahran-Cairo service, via the Saudi Arabian cities of Riyadh and # Jidda, on March 14, 1947". Shanks & Pottenger guessed 1950; go with the # earlier date. # # Shanks & Pottenger also state that until 1968-05-01 Saudi Arabia had two # time zones; the other zone, at UTC+4, was in the far eastern part of # the country. Ignore this, as it's before our 1970 cutoff. # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Riyadh 3:06:52 - LMT 1947 Mar 14 3:00 - AST +Link Asia/Riyadh Asia/Aden # Yemen +Link Asia/Riyadh Asia/Kuwait # Singapore # taken from Mok Ly Yng (2003-10-30) # http://www.math.nus.edu.sg/aslaksen/teaching/timezone.html # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Singapore 6:55:25 - LMT 1901 Jan 1 6:55:25 - SMT 1905 Jun 1 # Singapore M.T. 7:00 - MALT 1933 Jan 1 # Malaya Time 7:00 0:20 MALST 1936 Jan 1 7:20 - MALT 1941 Sep 1 7:30 - MALT 1942 Feb 16 9:00 - JST 1945 Sep 12 7:30 - MALT 1965 Aug 9 # independence 7:30 - SGT 1982 Jan 1 # Singapore Time 8:00 - SGT # Spratly Is # no information # Sri Lanka # From Paul Eggert (2013-02-21): # Milne says "Madras mean time use from May 1, 1898. Prior to this Colombo # mean time, 5h. 4m. 21.9s. F., was used." But 5:04:21.9 differs considerably # from Colombo's meridian 5:19:24, so for now ignore Milne and stick with # Shanks and Pottenger. # From Paul Eggert (1996-09-03): # "Sri Lanka advances clock by an hour to avoid blackout" # (, 1996-05-24, # no longer available as of 1999-08-17) # reported "the country's standard time will be put forward by one hour at # midnight Friday (1830 GMT) 'in the light of the present power crisis'." # # From Dharmasiri Senanayake, Sri Lanka Media Minister (1996-10-24), as quoted # by Shamindra in Daily News - Hot News Section # (1996-10-26): # With effect from 12.30 a.m. on 26th October 1996 # Sri Lanka will be six (06) hours ahead of GMT. # From Jesper Nørgaard Welen (2006-04-14), quoting Sri Lanka News Online # (2006-04-13): # 0030 hrs on April 15, 2006 (midnight of April 14, 2006 +30 minutes) # at present, become 2400 hours of April 14, 2006 (midnight of April 14, 2006). # From Peter Apps and Ranga Sirila of Reuters (2006-04-12) in: # http://today.reuters.co.uk/news/newsArticle.aspx?type=scienceNews&storyID=2006-04-12T172228Z_01_COL295762_RTRIDST_0_SCIENCE-SRILANKA-TIME-DC.XML # [The Tamil Tigers] never accepted the original 1996 time change and simply # kept their clocks set five and a half hours ahead of Greenwich Mean # Time (GMT), in line with neighbor India. # From Paul Eggert (2006-04-18): # People who live in regions under Tamil control can use [TZ='Asia/Kolkata'], # as that zone has agreed with the Tamil areas since our cutoff date of 1970. # From K Sethu (2006-04-25): # I think the abbreviation LKT originated from the world of computers at # the time of or subsequent to the time zone changes by SL Government # twice in 1996 and probably SL Government or its standardization # agencies never declared an abbreviation as a national standard. # # I recollect before the recent change the government announcements # mentioning it as simply changing Sri Lanka Standard Time or Sri Lanka # Time and no mention was made about the abbreviation. # # If we look at Sri Lanka Department of Government's "Official News # Website of Sri Lanka" ... http://www.news.lk/ we can see that they # use SLT as abbreviation in time stamp at the beginning of each news # item.... # # Within Sri Lanka I think LKT is well known among computer users and # administrators. In my opinion SLT may not be a good choice because the # nation's largest telcom / internet operator Sri Lanka Telcom is well # known by that abbreviation - simply as SLT (there IP domains are # slt.lk and sltnet.lk). # # But if indeed our government has adopted SLT as standard abbreviation # (that we have not known so far) then it is better that it be used for # all computers. # From Paul Eggert (2006-04-25): # One possibility is that we wait for a bit for the dust to settle down # and then see what people actually say in practice. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Colombo 5:19:24 - LMT 1880 5:19:32 - MMT 1906 # Moratuwa Mean Time 5:30 - IST 1942 Jan 5 5:30 0:30 IHST 1942 Sep 5:30 1:00 IST 1945 Oct 16 2:00 5:30 - IST 1996 May 25 0:00 6:30 - LKT 1996 Oct 26 0:30 6:00 - LKT 2006 Apr 15 0:30 5:30 - IST # Syria # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Syria 1920 1923 - Apr Sun>=15 2:00 1:00 S Rule Syria 1920 1923 - Oct Sun>=1 2:00 0 - Rule Syria 1962 only - Apr 29 2:00 1:00 S Rule Syria 1962 only - Oct 1 2:00 0 - Rule Syria 1963 1965 - May 1 2:00 1:00 S Rule Syria 1963 only - Sep 30 2:00 0 - Rule Syria 1964 only - Oct 1 2:00 0 - Rule Syria 1965 only - Sep 30 2:00 0 - Rule Syria 1966 only - Apr 24 2:00 1:00 S Rule Syria 1966 1976 - Oct 1 2:00 0 - Rule Syria 1967 1978 - May 1 2:00 1:00 S Rule Syria 1977 1978 - Sep 1 2:00 0 - Rule Syria 1983 1984 - Apr 9 2:00 1:00 S Rule Syria 1983 1984 - Oct 1 2:00 0 - Rule Syria 1986 only - Feb 16 2:00 1:00 S Rule Syria 1986 only - Oct 9 2:00 0 - Rule Syria 1987 only - Mar 1 2:00 1:00 S Rule Syria 1987 1988 - Oct 31 2:00 0 - Rule Syria 1988 only - Mar 15 2:00 1:00 S Rule Syria 1989 only - Mar 31 2:00 1:00 S Rule Syria 1989 only - Oct 1 2:00 0 - Rule Syria 1990 only - Apr 1 2:00 1:00 S Rule Syria 1990 only - Sep 30 2:00 0 - Rule Syria 1991 only - Apr 1 0:00 1:00 S Rule Syria 1991 1992 - Oct 1 0:00 0 - Rule Syria 1992 only - Apr 8 0:00 1:00 S Rule Syria 1993 only - Mar 26 0:00 1:00 S Rule Syria 1993 only - Sep 25 0:00 0 - # IATA SSIM (1998-02) says 1998-04-02; # (1998-09) says 1999-03-29 and 1999-09-29; (1999-02) says 1999-04-02, # 2000-04-02, and 2001-04-02; (1999-09) says 2000-03-31 and 2001-03-31; # (2006) says 2006-03-31 and 2006-09-22; # for now ignore all these claims and go with Shanks & Pottenger, # except for the 2006-09-22 claim (which seems right for Ramadan). Rule Syria 1994 1996 - Apr 1 0:00 1:00 S Rule Syria 1994 2005 - Oct 1 0:00 0 - Rule Syria 1997 1998 - Mar lastMon 0:00 1:00 S Rule Syria 1999 2006 - Apr 1 0:00 1:00 S # From Stephen Colebourne (2006-09-18): # According to IATA data, Syria will change DST on 21st September [21:00 UTC] # this year [only].... This is probably related to Ramadan, like Egypt. Rule Syria 2006 only - Sep 22 0:00 0 - # From Paul Eggert (2007-03-29): # Today the AP reported "Syria will switch to summertime at midnight Thursday." # http://www.iht.com/articles/ap/2007/03/29/africa/ME-GEN-Syria-Time-Change.php Rule Syria 2007 only - Mar lastFri 0:00 1:00 S # From Jesper Nørgaard (2007-10-27): # The sister center ICARDA of my work CIMMYT is confirming that Syria DST will # not take place 1st November at 0:00 o'clock but 1st November at 24:00 or # rather Midnight between Thursday and Friday. This does make more sense than # having it between Wednesday and Thursday (two workdays in Syria) since the # weekend in Syria is not Saturday and Sunday, but Friday and Saturday. So now # it is implemented at midnight of the last workday before weekend... # # From Steffen Thorsen (2007-10-27): # Jesper Nørgaard Welen wrote: # # > "Winter local time in Syria will be observed at midnight of Thursday 1 # > November 2007, and the clock will be put back 1 hour." # # I found confirmation on this in this gov.sy-article (Arabic): # http://wehda.alwehda.gov.sy/_print_veiw.asp?FileName=12521710520070926111247 # # which using Google's translate tools says: # Council of Ministers also approved the commencement of work on # identifying the winter time as of Friday, 2/11/2007 where the 60th # minute delay at midnight Thursday 1/11/2007. Rule Syria 2007 only - Nov Fri>=1 0:00 0 - # From Stephen Colebourne (2008-03-17): # For everyone's info, I saw an IATA time zone change for [Syria] for # this month (March 2008) in the last day or so.... # Country Time Standard --- DST Start --- --- DST End --- DST # Name Zone Variation Time Date Time Date # Variation # Syrian Arab # Republic SY +0200 2200 03APR08 2100 30SEP08 +0300 # 2200 02APR09 2100 30SEP09 +0300 # 2200 01APR10 2100 30SEP10 +0300 # From Arthur David Olson (2008-03-17): # Here's a link to English-language coverage by the Syrian Arab News # Agency (SANA)... # http://www.sana.sy/eng/21/2008/03/11/165173.htm # ...which reads (in part) "The Cabinet approved the suggestion of the # Ministry of Electricity to begin daylight savings time on Friday April # 4th, advancing clocks one hour ahead on midnight of Thursday April 3rd." # Since Syria is two hours east of UTC, the 2200 and 2100 transition times # shown above match up with midnight in Syria. # From Arthur David Olson (2008-03-18): # My best guess at a Syrian rule is "the Friday nearest April 1"; # coding that involves either using a "Mar Fri>=29" construct that old time zone # compilers can't handle or having multiple Rules (a la Israel). # For now, use "Apr Fri>=1", and go with IATA on a uniform Sep 30 end. # From Steffen Thorsen (2008-10-07): # Syria has now officially decided to end DST on 2008-11-01 this year, # according to the following article in the Syrian Arab News Agency (SANA). # # The article is in Arabic, and seems to tell that they will go back to # winter time on 2008-11-01 at 00:00 local daylight time (delaying/setting # clocks back 60 minutes). # # http://sana.sy/ara/2/2008/10/07/195459.htm # From Steffen Thorsen (2009-03-19): # Syria will start DST on 2009-03-27 00:00 this year according to many sources, # two examples: # # http://www.sana.sy/eng/21/2009/03/17/217563.htm # (English, Syrian Arab News # Agency) # http://thawra.alwehda.gov.sy/_View_news2.asp?FileName=94459258720090318012209 # (Arabic, gov-site) # # We have not found any sources saying anything about when DST ends this year. # # Our summary # http://www.timeanddate.com/news/time/syria-dst-starts-march-27-2009.html # From Steffen Thorsen (2009-10-27): # The Syrian Arab News Network on 2009-09-29 reported that Syria will # revert back to winter (standard) time on midnight between Thursday # 2009-10-29 and Friday 2009-10-30: # http://www.sana.sy/ara/2/2009/09/29/247012.htm (Arabic) # From Arthur David Olson (2009-10-28): # We'll see if future DST switching times turn out to be end of the last # Thursday of the month or the start of the last Friday of the month or # something else. For now, use the start of the last Friday. # From Steffen Thorsen (2010-03-17): # The "Syrian News Station" reported on 2010-03-16 that the Council of # Ministers has decided that Syria will start DST on midnight Thursday # 2010-04-01: (midnight between Thursday and Friday): # http://sns.sy/sns/?path=news/read/11421 (Arabic) # From Steffen Thorsen (2012-03-26): # Today, Syria's government announced that they will start DST early on Friday # (00:00). This is a bit earlier than the past two years. # # From Syrian Arab News Agency, in Arabic: # http://www.sana.sy/ara/2/2012/03/26/408215.htm # # Our brief summary: # http://www.timeanddate.com/news/time/syria-dst-2012.html # From Arthur David Olson (2012-03-27): # Assume last Friday in March going forward XXX. Rule Syria 2008 only - Apr Fri>=1 0:00 1:00 S Rule Syria 2008 only - Nov 1 0:00 0 - Rule Syria 2009 only - Mar lastFri 0:00 1:00 S Rule Syria 2010 2011 - Apr Fri>=1 0:00 1:00 S Rule Syria 2012 max - Mar lastFri 0:00 1:00 S Rule Syria 2009 max - Oct lastFri 0:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Damascus 2:25:12 - LMT 1920 # Dimashq 2:00 Syria EE%sT # Tajikistan # From Shanks & Pottenger. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Dushanbe 4:35:12 - LMT 1924 May 2 5:00 - DUST 1930 Jun 21 # Dushanbe Time 6:00 RussiaAsia DUS%sT 1991 Mar 31 2:00s 5:00 1:00 DUSST 1991 Sep 9 2:00s 5:00 - TJT # Tajikistan Time # Thailand # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Bangkok 6:42:04 - LMT 1880 6:42:04 - BMT 1920 Apr # Bangkok Mean Time 7:00 - ICT Link Asia/Bangkok Asia/Phnom_Penh # Cambodia Link Asia/Bangkok Asia/Vientiane # Laos # Turkmenistan # From Shanks & Pottenger. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Ashgabat 3:53:32 - LMT 1924 May 2 # or Ashkhabad 4:00 - ASHT 1930 Jun 21 # Ashkhabad Time 5:00 RussiaAsia ASH%sT 1991 Mar 31 2:00 4:00 RussiaAsia ASH%sT 1991 Oct 27 # independence 4:00 RussiaAsia TM%sT 1992 Jan 19 2:00 5:00 - TMT # United Arab Emirates # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Dubai 3:41:12 - LMT 1920 4:00 - GST +Link Asia/Dubai Asia/Muscat # Oman # Uzbekistan # Byalokoz 1919 says Uzbekistan was 4:27:53. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Samarkand 4:27:53 - LMT 1924 May 2 4:00 - SAMT 1930 Jun 21 # Samarkand Time 5:00 - SAMT 1981 Apr 1 5:00 1:00 SAMST 1981 Oct 1 6:00 - TAST 1982 Apr 1 # Tashkent Time 5:00 RussiaAsia SAM%sT 1991 Sep 1 # independence 5:00 RussiaAsia UZ%sT 1992 5:00 - UZT # Milne says Tashkent was 4:37:10.8; round to nearest. Zone Asia/Tashkent 4:37:11 - LMT 1924 May 2 5:00 - TAST 1930 Jun 21 # Tashkent Time 6:00 RussiaAsia TAS%sT 1991 Mar 31 2:00 5:00 RussiaAsia TAS%sT 1991 Sep 1 # independence 5:00 RussiaAsia UZ%sT 1992 5:00 - UZT # Vietnam # From Paul Eggert (2014-10-04): # Milne gives 7:16:56 for the meridian of Saigon in 1899, as being # used in Lower Laos, Cambodia, and Annam. But this is quite a ways # from Saigon's location. For now, ignore this and stick with Shanks # and Pottenger for LMT before 1906. # From Arthur David Olson (2008-03-18): # The English-language name of Vietnam's most populous city is "Ho Chi Minh # City"; use Ho_Chi_Minh below to avoid a name of more than 14 characters. # From Paul Eggert (2014-10-21) after a heads-up from Trần Ngọc Quân: # Trần Tiến Bình's authoritative book "Lịch Việt Nam: thế kỷ XX-XXI (1901-2100)" # (Nhà xuất bản Văn Hoá - Thông Tin, Hanoi, 2005), pp 49-50, # is quoted verbatim in: # http://www.thoigian.com.vn/?mPage=P80D01 # is translated by Brian Inglis in: # http://mm.icann.org/pipermail/tz/2014-October/021654.html # and is the basis for the information below. # # The 1906 transition was effective July 1 and standardized Indochina to # Phù Liễn Observatory, legally 104 deg. 17'17" east of Paris. # It's unclear whether this meant legal Paris Mean Time (00:09:21) or # the Paris Meridian (2 deg. 20'14.03" E); the former yields 07:06:30.1333... # and the latter 07:06:29.333... so either way it rounds to 07:06:30, # which is used below even though the modern-day Phù Liễn Observatory # is closer to 07:06:31. Abbreviate Phù Liễn Mean Time as PLMT. # # The following transitions occurred in Indochina in general (before 1954) # and in South Vietnam in particular (after 1954): # To 07:00 on 1911-05-01. # To 08:00 on 1942-12-31 at 23:00. # To 09:00 in 1945-03-14 at 23:00. # To 07:00 on 1945-09-02 in Vietnam. # To 08:00 on 1947-04-01 in French-controlled Indochina. # To 07:00 on 1955-07-01 in South Vietnam. # To 08:00 on 1959-12-31 at 23:00 in South Vietnam. # To 07:00 on 1975-06-13 in South Vietnam. # # Trần cites the following sources; it's unclear which supplied the info above. # # Hoàng Xuân Hãn: "Lịch và lịch Việt Nam". Tập san Khoa học Xã hội, # No. 9, Paris, February 1982. # # Lê Thành Lân: "Lịch và niên biểu lịch sử hai mươi thế kỷ (0001-2010)", # NXB Thống kê, Hanoi, 2000. # # Lê Thành Lân: "Lịch hai thế kỷ (1802-2010) và các lịch vĩnh cửu", # NXB Thuận Hoá, Huế, 1995. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Asia/Ho_Chi_Minh 7:06:40 - LMT 1906 Jul 1 7:06:30 - PLMT 1911 May 1 7:00 - ICT 1942 Dec 31 23:00 8:00 - IDT 1945 Mar 14 23:00 9:00 - JST 1945 Sep 2 7:00 - ICT 1947 Apr 1 8:00 - IDT 1955 Jul 1 7:00 - ICT 1959 Dec 31 23:00 8:00 - IDT 1975 Jun 13 7:00 - ICT # Yemen - -# Milne says 2:59:54 was the meridian of the saluting battery at Aden, -# and that Yemen was at 1:55:56, the meridian of the Hagia Sophia. - -# Zone NAME GMTOFF RULES FORMAT [UNTIL] -Zone Asia/Aden 2:59:54 - LMT 1950 - 3:00 - AST +# See Asia/Riyadh. Index: projects/cxl_iscsi/contrib/tzdata/backward =================================================================== --- projects/cxl_iscsi/contrib/tzdata/backward (revision 279716) +++ projects/cxl_iscsi/contrib/tzdata/backward (revision 279717) @@ -1,122 +1,122 @@ # This file is in the public domain, so clarified as of # 2009-05-17 by Arthur David Olson. # This file provides links between current names for time zones # and their old names. Many names changed in late 1993. # Link TARGET LINK-NAME -Link Africa/Asmara Africa/Asmera +Link Africa/Nairobi Africa/Asmera Link Africa/Abidjan Africa/Timbuktu Link America/Argentina/Catamarca America/Argentina/ComodRivadavia Link America/Adak America/Atka Link America/Argentina/Buenos_Aires America/Buenos_Aires Link America/Argentina/Catamarca America/Catamarca Link America/Atikokan America/Coral_Harbour Link America/Argentina/Cordoba America/Cordoba Link America/Tijuana America/Ensenada Link America/Indiana/Indianapolis America/Fort_Wayne Link America/Indiana/Indianapolis America/Indianapolis Link America/Argentina/Jujuy America/Jujuy Link America/Indiana/Knox America/Knox_IN Link America/Kentucky/Louisville America/Louisville Link America/Argentina/Mendoza America/Mendoza Link America/Rio_Branco America/Porto_Acre Link America/Argentina/Cordoba America/Rosario Link America/Denver America/Shiprock Link America/Port_of_Spain America/Virgin Link Pacific/Auckland Antarctica/South_Pole Link Asia/Ashgabat Asia/Ashkhabad Link Asia/Kolkata Asia/Calcutta Link Asia/Shanghai Asia/Chongqing Link Asia/Shanghai Asia/Chungking Link Asia/Dhaka Asia/Dacca Link Asia/Shanghai Asia/Harbin Link Asia/Urumqi Asia/Kashgar Link Asia/Kathmandu Asia/Katmandu Link Asia/Macau Asia/Macao Link Asia/Ho_Chi_Minh Asia/Saigon Link Asia/Jerusalem Asia/Tel_Aviv Link Asia/Thimphu Asia/Thimbu Link Asia/Makassar Asia/Ujung_Pandang Link Asia/Ulaanbaatar Asia/Ulan_Bator Link Atlantic/Faroe Atlantic/Faeroe Link Europe/Oslo Atlantic/Jan_Mayen Link Australia/Sydney Australia/ACT Link Australia/Sydney Australia/Canberra Link Australia/Lord_Howe Australia/LHI Link Australia/Sydney Australia/NSW Link Australia/Darwin Australia/North Link Australia/Brisbane Australia/Queensland Link Australia/Adelaide Australia/South Link Australia/Hobart Australia/Tasmania Link Australia/Melbourne Australia/Victoria Link Australia/Perth Australia/West Link Australia/Broken_Hill Australia/Yancowinna Link America/Rio_Branco Brazil/Acre Link America/Noronha Brazil/DeNoronha Link America/Sao_Paulo Brazil/East Link America/Manaus Brazil/West Link America/Halifax Canada/Atlantic Link America/Winnipeg Canada/Central Link America/Regina Canada/East-Saskatchewan Link America/Toronto Canada/Eastern Link America/Edmonton Canada/Mountain Link America/St_Johns Canada/Newfoundland Link America/Vancouver Canada/Pacific Link America/Regina Canada/Saskatchewan Link America/Whitehorse Canada/Yukon Link America/Santiago Chile/Continental Link Pacific/Easter Chile/EasterIsland Link America/Havana Cuba Link Africa/Cairo Egypt Link Europe/Dublin Eire Link Europe/London Europe/Belfast Link Europe/Chisinau Europe/Tiraspol Link Europe/London GB Link Europe/London GB-Eire Link Etc/GMT GMT+0 Link Etc/GMT GMT-0 Link Etc/GMT GMT0 Link Etc/GMT Greenwich Link Asia/Hong_Kong Hongkong Link Atlantic/Reykjavik Iceland Link Asia/Tehran Iran Link Asia/Jerusalem Israel Link America/Jamaica Jamaica Link Asia/Tokyo Japan Link Pacific/Kwajalein Kwajalein Link Africa/Tripoli Libya Link America/Tijuana Mexico/BajaNorte Link America/Mazatlan Mexico/BajaSur Link America/Mexico_City Mexico/General Link Pacific/Auckland NZ Link Pacific/Chatham NZ-CHAT Link America/Denver Navajo Link Asia/Shanghai PRC Link Pacific/Pohnpei Pacific/Ponape Link Pacific/Pago_Pago Pacific/Samoa Link Pacific/Chuuk Pacific/Truk Link Pacific/Chuuk Pacific/Yap Link Europe/Warsaw Poland Link Europe/Lisbon Portugal Link Asia/Taipei ROC Link Asia/Seoul ROK Link Asia/Singapore Singapore Link Europe/Istanbul Turkey Link Etc/UCT UCT Link America/Anchorage US/Alaska Link America/Adak US/Aleutian Link America/Phoenix US/Arizona Link America/Chicago US/Central Link America/Indiana/Indianapolis US/East-Indiana Link America/New_York US/Eastern Link Pacific/Honolulu US/Hawaii Link America/Indiana/Knox US/Indiana-Starke Link America/Detroit US/Michigan Link America/Denver US/Mountain Link America/Los_Angeles US/Pacific Link Pacific/Pago_Pago US/Samoa Link Etc/UTC UTC Link Etc/UTC Universal Link Europe/Moscow W-SU Link Etc/UTC Zulu Index: projects/cxl_iscsi/contrib/tzdata/europe =================================================================== --- projects/cxl_iscsi/contrib/tzdata/europe (revision 279716) +++ projects/cxl_iscsi/contrib/tzdata/europe (revision 279717) @@ -1,3398 +1,3395 @@ # This file is in the public domain, so clarified as of # 2009-05-17 by Arthur David Olson. # This file is by no means authoritative; if you think you know better, # go ahead and edit the file (and please send any changes to # tz@iana.org for general use in the future). For more, please see # the file CONTRIBUTING in the tz distribution. # From Paul Eggert (2014-10-31): # # Unless otherwise specified, the source for data through 1990 is: # Thomas G. Shanks and Rique Pottenger, The International Atlas (6th edition), # San Diego: ACS Publications, Inc. (2003). # Unfortunately this book contains many errors and cites no sources. # # Gwillim Law writes that a good source # for recent time zone data is the International Air Transport # Association's Standard Schedules Information Manual (IATA SSIM), # published semiannually. Law sent in several helpful summaries # of the IATA's data after 1990. Except where otherwise noted, # IATA SSIM is the source for entries after 1990. # # A reliable and entertaining source about time zones is # Derek Howse, Greenwich time and longitude, Philip Wilson Publishers (1997). # # Except where otherwise noted, Shanks & Pottenger is the source for # entries through 1991, and IATA SSIM is the source for entries afterwards. # # Other sources occasionally used include: # # Edward W. Whitman, World Time Differences, # Whitman Publishing Co, 2 Niagara Av, Ealing, London (undated), # which I found in the UCLA library. # # William Willett, The Waste of Daylight, 19th edition # # [PDF] (1914-03) # # Milne J. Civil time. Geogr J. 1899 Feb;13(2):173-94 # . He writes: # "It is requested that corrections and additions to these tables # may be sent to Mr. John Milne, Royal Geographical Society, # Savile Row, London." Nowadays please email them to tz@iana.org. # # Byalokoz EL. New Counting of Time in Russia since July 1, 1919. # This Russian-language source was consulted by Vladimir Karpinsky; see # http://mm.icann.org/pipermail/tz/2014-August/021320.html # The full Russian citation is: # Бялокоз, Евгений Людвигович. Новый счет времени в течении суток # введенный декретом Совета народных комиссаров для всей России с 1-го # июля 1919 г. / Изд. 2-е Междуведомственной комиссии. - Петроград: # Десятая гос. тип., 1919. # http://resolver.gpntb.ru/purl?docushare/dsweb/Get/Resource-2011/Byalokoz__E.L.__Novyy__schet__vremeni__v__techenie__sutok__izd__2(1).pdf # # Brazil's Divisão Serviço da Hora (DSHO), # History of Summer Time # # (1998-09-21, in Portuguese) # # I invented the abbreviations marked '*' in the following table; # the rest are from earlier versions of this file, or from other sources. # Corrections are welcome! # std dst 2dst # LMT Local Mean Time # -4:00 AST ADT Atlantic # -3:00 WGT WGST Western Greenland* # -1:00 EGT EGST Eastern Greenland* # 0:00 GMT BST BDST Greenwich, British Summer # 0:00 GMT IST Greenwich, Irish Summer # 0:00 WET WEST WEMT Western Europe # 0:19:32.13 AMT NST Amsterdam, Netherlands Summer (1835-1937)* # 0:20 NET NEST Netherlands (1937-1940)* # 1:00 BST British Standard (1968-1971) # 1:00 CET CEST CEMT Central Europe # 1:00:14 SET Swedish (1879-1899)* # 2:00 EET EEST Eastern Europe # 3:00 FET Further-eastern Europe (2011-2014)* # 3:00 MSK MSD MSM* Moscow # From Peter Ilieve (1994-12-04), # The original six [EU members]: Belgium, France, (West) Germany, Italy, # Luxembourg, the Netherlands. # Plus, from 1 Jan 73: Denmark, Ireland, United Kingdom. # Plus, from 1 Jan 81: Greece. # Plus, from 1 Jan 86: Spain, Portugal. # Plus, from 1 Jan 95: Austria, Finland, Sweden. (Norway negotiated terms for # entry but in a referendum on 28 Nov 94 the people voted No by 52.2% to 47.8% # on a turnout of 88.6%. This was almost the same result as Norway's previous # referendum in 1972, they are the only country to have said No twice. # Referendums in the other three countries voted Yes.) # ... # Estonia ... uses EU dates but not at 01:00 GMT, they use midnight GMT. # I don't think they know yet what they will do from 1996 onwards. # ... # There shouldn't be any [current members who are not using EU rules]. # A Directive has the force of law, member states are obliged to enact # national law to implement it. The only contentious issue was the # different end date for the UK and Ireland, and this was always allowed # in the Directive. ############################################################################### # Britain (United Kingdom) and Ireland (Eire) # From Peter Ilieve (1994-07-06): # # On 17 Jan 1994 the Independent, a UK quality newspaper, had a piece about # historical vistas along the Thames in west London. There was a photo # and a sketch map showing some of the sightlines involved. One paragraph # of the text said: # # 'An old stone obelisk marking a forgotten terrestrial meridian stands # beside the river at Kew. In the 18th century, before time and longitude # was standardised by the Royal Observatory in Greenwich, scholars observed # this stone and the movement of stars from Kew Observatory nearby. They # made their calculations and set the time for the Horse Guards and Parliament, # but now the stone is obscured by scrubwood and can only be seen by walking # along the towpath within a few yards of it.' # # I have a one inch to one mile map of London and my estimate of the stone's # position is 51 degrees 28' 30" N, 0 degrees 18' 45" W. The longitude should # be within about +-2". The Ordnance Survey grid reference is TQ172761. # # [This yields GMTOFF = -0:01:15 for London LMT in the 18th century.] # From Paul Eggert (1993-11-18): # # Howse writes that Britain was the first country to use standard time. # The railways cared most about the inconsistencies of local mean time, # and it was they who forced a uniform time on the country. # The original idea was credited to Dr. William Hyde Wollaston (1766-1828) # and was popularized by Abraham Follett Osler (1808-1903). # The first railway to adopt London time was the Great Western Railway # in November 1840; other railways followed suit, and by 1847 most # (though not all) railways used London time. On 1847-09-22 the # Railway Clearing House, an industry standards body, recommended that GMT be # adopted at all stations as soon as the General Post Office permitted it. # The transition occurred on 12-01 for the L&NW, the Caledonian, # and presumably other railways; the January 1848 Bradshaw's lists many # railways as using GMT. By 1855 the vast majority of public # clocks in Britain were set to GMT (though some, like the great clock # on Tom Tower at Christ Church, Oxford, were fitted with two minute hands, # one for local time and one for GMT). The last major holdout was the legal # system, which stubbornly stuck to local time for many years, leading # to oddities like polls opening at 08:13 and closing at 16:13. # The legal system finally switched to GMT when the Statutes (Definition # of Time) Act took effect; it received the Royal Assent on 1880-08-02. # # In the tables below, we condense this complicated story into a single # transition date for London, namely 1847-12-01. We don't know as much # about Dublin, so we use 1880-08-02, the legal transition time. # From Paul Eggert (2014-07-19): # The ancients had no need for daylight saving, as they kept time # informally or via hours whose length depended on the time of year. # Daylight saving time in its modern sense was invented by the # New Zealand entomologist George Vernon Hudson (1867-1946), # whose day job as a postal clerk led him to value # after-hours daylight in which to pursue his research. # In 1895 he presented a paper to the Wellington Philosophical Society # that proposed a two-hour daylight-saving shift. See: # Hudson GV. On seasonal time-adjustment in countries south of lat. 30 deg. # Transactions and Proceedings of the New Zealand Institute. 1895;28:734 # http://rsnz.natlib.govt.nz/volume/rsnz_28/rsnz_28_00_006110.html # Although some interest was expressed in New Zealand, his proposal # did not find its way into law and eventually it was almost forgotten. # # In England, DST was independently reinvented by William Willett (1857-1915), # a London builder and member of the Royal Astronomical Society # who circulated a pamphlet "The Waste of Daylight" (1907) # that proposed advancing clocks 20 minutes on each of four Sundays in April, # and retarding them by the same amount on four Sundays in September. # A bill was drafted in 1909 and introduced in Parliament several times, # but it met with ridicule and opposition, especially from farming interests. # Later editions of the pamphlet proposed one-hour summer time, and # it was eventually adopted as a wartime measure in 1916. # See: Summer Time Arrives Early, The Times (2000-05-18). # A monument to Willett was unveiled on 1927-05-21, in an open space in # a 45-acre wood near Chislehurst, Kent that was purchased by popular # subscription and open to the public. On the south face of the monolith, # designed by G. W. Miller, is the William Willett Memorial Sundial, # which is permanently set to Summer Time. # From Winston Churchill (1934-04-28): # It is one of the paradoxes of history that we should owe the boon of # summer time, which gives every year to the people of this country # between 160 and 170 hours more daylight leisure, to a war which # plunged Europe into darkness for four years, and shook the # foundations of civilization throughout the world. # -- "A Silent Toast to William Willett", Pictorial Weekly; # republished in Finest Hour (Spring 2002) 1(114):26 # http://www.winstonchurchill.org/images/finesthour/Vol.01%20No.114.pdf # From Paul Eggert (1996-09-03): # The OED Supplement says that the English originally said "Daylight Saving" # when they were debating the adoption of DST in 1908; but by 1916 this # term appears only in quotes taken from DST's opponents, whereas the # proponents (who eventually won the argument) are quoted as using "Summer". # From Arthur David Olson (1989-01-19): # A source at the British Information Office in New York avers that it's # known as "British" Summer Time in all parts of the United Kingdom. # Date: 4 Jan 89 08:57:25 GMT (Wed) # From: Jonathan Leffler # [British Summer Time] is fixed annually by Act of Parliament. # If you can predict what Parliament will do, you should be in # politics making a fortune, not computing. # From Chris Carrier (1996-06-14): # I remember reading in various wartime issues of the London Times the # acronym BDST for British Double Summer Time. Look for the published # time of sunrise and sunset in The Times, when BDST was in effect, and # if you find a zone reference it will say, "All times B.D.S.T." # From Joseph S. Myers (1999-09-02): # ... some military cables (WO 219/4100 - this is a copy from the # main SHAEF archives held in the US National Archives, SHAEF/5252/8/516) # agree that the usage is BDST (this appears in a message dated 17 Feb 1945). # From Joseph S. Myers (2000-10-03): # On 18th April 1941, Sir Stephen Tallents of the BBC wrote to Sir # Alexander Maxwell of the Home Office asking whether there was any # official designation; the reply of the 21st was that there wasn't # but he couldn't think of anything better than the "Double British # Summer Time" that the BBC had been using informally. # http://www.polyomino.org.uk/british-time/bbc-19410418.png # http://www.polyomino.org.uk/british-time/ho-19410421.png # From Sir Alexander Maxwell in the above-mentioned letter (1941-04-21): # [N]o official designation has as far as I know been adopted for the time # which is to be introduced in May.... # I cannot think of anything better than "Double British Summer Time" # which could not be said to run counter to any official description. # From Paul Eggert (2000-10-02): # Howse writes (p 157) 'DBST' too, but 'BDST' seems to have been common # and follows the more usual convention of putting the location name first, # so we use 'BDST'. # Peter Ilieve (1998-04-19) described at length # the history of summer time legislation in the United Kingdom. # Since 1998 Joseph S. Myers has been updating # and extending this list, which can be found in # http://www.polyomino.org.uk/british-time/ # From Joseph S. Myers (1998-01-06): # # The legal time in the UK outside of summer time is definitely GMT, not UTC; # see Lord Tanlaw's speech # http://www.publications.parliament.uk/pa/ld199798/ldhansrd/vo970611/text/70611-10.htm#70611-10_head0 # (Lords Hansard 11 June 1997 columns 964 to 976). # From Paul Eggert (2006-03-22): # # For lack of other data, follow Shanks & Pottenger for Eire in 1940-1948. # # Given Ilieve and Myers's data, the following claims by Shanks & Pottenger # are incorrect: # * Wales did not switch from GMT to daylight saving time until # 1921 Apr 3, when they began to conform with the rest of Great Britain. # Actually, Wales was identical after 1880. # * Eire had two transitions on 1916 Oct 1. # It actually just had one transition. # * Northern Ireland used single daylight saving time throughout WW II. # Actually, it conformed to Britain. # * GB-Eire changed standard time to 1 hour ahead of GMT on 1968-02-18. # Actually, that date saw the usual switch to summer time. # Standard time was not changed until 1968-10-27 (the clocks didn't change). # # Here is another incorrect claim by Shanks & Pottenger: # * Jersey, Guernsey, and the Isle of Man did not switch from GMT # to daylight saving time until 1921 Apr 3, when they began to # conform with Great Britain. # S.R.&O. 1916, No. 382 and HO 45/10811/312364 (quoted above) say otherwise. # # The following claim by Shanks & Pottenger is possible though doubtful; # we'll ignore it for now. # * Dublin's 1971-10-31 switch was at 02:00, even though London's was 03:00. # # # Whitman says Dublin Mean Time was -0:25:21, which is more precise than # Shanks & Pottenger. # Perhaps this was Dunsink Observatory Time, as Dunsink Observatory # (8 km NW of Dublin's center) seemingly was to Dublin as Greenwich was # to London. For example: # # "Timeball on the ballast office is down. Dunsink time." # -- James Joyce, Ulysses # "Countess Markievicz ... claimed that the [1916] abolition of Dublin Mean Time # was among various actions undertaken by the 'English' government that # would 'put the whole country into the SF (Sinn Féin) camp'. She claimed # Irish 'public feeling (was) outraged by forcing of English time on us'." # -- Parsons M. Dublin lost its time zone - and 25 minutes - after 1916 Rising. # Irish Times 2014-10-27. # http://www.irishtimes.com/news/politics/dublin-lost-its-time-zone-and-25-minutes-after-1916-rising-1.1977411 # From Joseph S. Myers (2005-01-26): # Irish laws are available online at . # These include various relating to legal time, for example: # # ZZA13Y1923.html ZZA12Y1924.html ZZA8Y1925.html ZZSIV20PG1267.html # # ZZSI71Y1947.html ZZSI128Y1948.html ZZSI23Y1949.html ZZSI41Y1950.html # ZZSI27Y1951.html ZZSI73Y1952.html # # ZZSI11Y1961.html ZZSI232Y1961.html ZZSI182Y1962.html # ZZSI167Y1963.html ZZSI257Y1964.html ZZSI198Y1967.html # ZZA23Y1968.html ZZA17Y1971.html # # ZZSI67Y1981.html ZZSI212Y1982.html ZZSI45Y1986.html # ZZSI264Y1988.html ZZSI52Y1990.html ZZSI371Y1992.html # ZZSI395Y1994.html ZZSI484Y1997.html ZZSI506Y2001.html # # [These are all relative to the root, e.g., the first is # .] # # (These are those I found, but there could be more. In any case these # should allow various updates to the comments in the europe file to cover # the laws applicable in Ireland.) # # (Note that the time in the Republic of Ireland since 1968 has been defined # in terms of standard time being GMT+1 with a period of winter time when it # is GMT, rather than standard time being GMT with a period of summer time # being GMT+1.) # From Paul Eggert (1999-03-28): # Clive Feather (, 1997-03-31) # reports that Folkestone (Cheriton) Shuttle Terminal uses Concession Time # (CT), equivalent to French civil time. # Julian Hill (, 1998-09-30) reports that # trains between Dollands Moor (the freight facility next door) # and Frethun run in CT. # My admittedly uninformed guess is that the terminal has two authorities, # the French concession operators and the British civil authorities, # and that the time depends on who you're talking to. # If, say, the British police were called to the station for some reason, # I would expect the official police report to use GMT/BST and not CET/CEST. # This is a borderline case, but for now let's stick to GMT/BST. # From an anonymous contributor (1996-06-02): # The law governing time in Ireland is under Statutory Instrument SI 395/94, # which gives force to European Union 7th Council Directive # 94/21/EC. # Under this directive, the Minister for Justice in Ireland makes appropriate # regulations. I spoke this morning with the Secretary of the Department of # Justice (tel +353 1 678 9711) who confirmed to me that the correct name is # "Irish Summer Time", abbreviated to "IST". # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S # Summer Time Act, 1916 Rule GB-Eire 1916 only - May 21 2:00s 1:00 BST Rule GB-Eire 1916 only - Oct 1 2:00s 0 GMT # S.R.&O. 1917, No. 358 Rule GB-Eire 1917 only - Apr 8 2:00s 1:00 BST Rule GB-Eire 1917 only - Sep 17 2:00s 0 GMT # S.R.&O. 1918, No. 274 Rule GB-Eire 1918 only - Mar 24 2:00s 1:00 BST Rule GB-Eire 1918 only - Sep 30 2:00s 0 GMT # S.R.&O. 1919, No. 297 Rule GB-Eire 1919 only - Mar 30 2:00s 1:00 BST Rule GB-Eire 1919 only - Sep 29 2:00s 0 GMT # S.R.&O. 1920, No. 458 Rule GB-Eire 1920 only - Mar 28 2:00s 1:00 BST # S.R.&O. 1920, No. 1844 Rule GB-Eire 1920 only - Oct 25 2:00s 0 GMT # S.R.&O. 1921, No. 363 Rule GB-Eire 1921 only - Apr 3 2:00s 1:00 BST Rule GB-Eire 1921 only - Oct 3 2:00s 0 GMT # S.R.&O. 1922, No. 264 Rule GB-Eire 1922 only - Mar 26 2:00s 1:00 BST Rule GB-Eire 1922 only - Oct 8 2:00s 0 GMT # The Summer Time Act, 1922 Rule GB-Eire 1923 only - Apr Sun>=16 2:00s 1:00 BST Rule GB-Eire 1923 1924 - Sep Sun>=16 2:00s 0 GMT Rule GB-Eire 1924 only - Apr Sun>=9 2:00s 1:00 BST Rule GB-Eire 1925 1926 - Apr Sun>=16 2:00s 1:00 BST # The Summer Time Act, 1925 Rule GB-Eire 1925 1938 - Oct Sun>=2 2:00s 0 GMT Rule GB-Eire 1927 only - Apr Sun>=9 2:00s 1:00 BST Rule GB-Eire 1928 1929 - Apr Sun>=16 2:00s 1:00 BST Rule GB-Eire 1930 only - Apr Sun>=9 2:00s 1:00 BST Rule GB-Eire 1931 1932 - Apr Sun>=16 2:00s 1:00 BST Rule GB-Eire 1933 only - Apr Sun>=9 2:00s 1:00 BST Rule GB-Eire 1934 only - Apr Sun>=16 2:00s 1:00 BST Rule GB-Eire 1935 only - Apr Sun>=9 2:00s 1:00 BST Rule GB-Eire 1936 1937 - Apr Sun>=16 2:00s 1:00 BST Rule GB-Eire 1938 only - Apr Sun>=9 2:00s 1:00 BST Rule GB-Eire 1939 only - Apr Sun>=16 2:00s 1:00 BST # S.R.&O. 1939, No. 1379 Rule GB-Eire 1939 only - Nov Sun>=16 2:00s 0 GMT # S.R.&O. 1940, No. 172 and No. 1883 Rule GB-Eire 1940 only - Feb Sun>=23 2:00s 1:00 BST # S.R.&O. 1941, No. 476 Rule GB-Eire 1941 only - May Sun>=2 1:00s 2:00 BDST Rule GB-Eire 1941 1943 - Aug Sun>=9 1:00s 1:00 BST # S.R.&O. 1942, No. 506 Rule GB-Eire 1942 1944 - Apr Sun>=2 1:00s 2:00 BDST # S.R.&O. 1944, No. 932 Rule GB-Eire 1944 only - Sep Sun>=16 1:00s 1:00 BST # S.R.&O. 1945, No. 312 Rule GB-Eire 1945 only - Apr Mon>=2 1:00s 2:00 BDST Rule GB-Eire 1945 only - Jul Sun>=9 1:00s 1:00 BST # S.R.&O. 1945, No. 1208 Rule GB-Eire 1945 1946 - Oct Sun>=2 2:00s 0 GMT Rule GB-Eire 1946 only - Apr Sun>=9 2:00s 1:00 BST # The Summer Time Act, 1947 Rule GB-Eire 1947 only - Mar 16 2:00s 1:00 BST Rule GB-Eire 1947 only - Apr 13 1:00s 2:00 BDST Rule GB-Eire 1947 only - Aug 10 1:00s 1:00 BST Rule GB-Eire 1947 only - Nov 2 2:00s 0 GMT # Summer Time Order, 1948 (S.I. 1948/495) Rule GB-Eire 1948 only - Mar 14 2:00s 1:00 BST Rule GB-Eire 1948 only - Oct 31 2:00s 0 GMT # Summer Time Order, 1949 (S.I. 1949/373) Rule GB-Eire 1949 only - Apr 3 2:00s 1:00 BST Rule GB-Eire 1949 only - Oct 30 2:00s 0 GMT # Summer Time Order, 1950 (S.I. 1950/518) # Summer Time Order, 1951 (S.I. 1951/430) # Summer Time Order, 1952 (S.I. 1952/451) Rule GB-Eire 1950 1952 - Apr Sun>=14 2:00s 1:00 BST Rule GB-Eire 1950 1952 - Oct Sun>=21 2:00s 0 GMT # revert to the rules of the Summer Time Act, 1925 Rule GB-Eire 1953 only - Apr Sun>=16 2:00s 1:00 BST Rule GB-Eire 1953 1960 - Oct Sun>=2 2:00s 0 GMT Rule GB-Eire 1954 only - Apr Sun>=9 2:00s 1:00 BST Rule GB-Eire 1955 1956 - Apr Sun>=16 2:00s 1:00 BST Rule GB-Eire 1957 only - Apr Sun>=9 2:00s 1:00 BST Rule GB-Eire 1958 1959 - Apr Sun>=16 2:00s 1:00 BST Rule GB-Eire 1960 only - Apr Sun>=9 2:00s 1:00 BST # Summer Time Order, 1961 (S.I. 1961/71) # Summer Time (1962) Order, 1961 (S.I. 1961/2465) # Summer Time Order, 1963 (S.I. 1963/81) Rule GB-Eire 1961 1963 - Mar lastSun 2:00s 1:00 BST Rule GB-Eire 1961 1968 - Oct Sun>=23 2:00s 0 GMT # Summer Time (1964) Order, 1963 (S.I. 1963/2101) # Summer Time Order, 1964 (S.I. 1964/1201) # Summer Time Order, 1967 (S.I. 1967/1148) Rule GB-Eire 1964 1967 - Mar Sun>=19 2:00s 1:00 BST # Summer Time Order, 1968 (S.I. 1968/117) Rule GB-Eire 1968 only - Feb 18 2:00s 1:00 BST # The British Standard Time Act, 1968 # (no summer time) # The Summer Time Act, 1972 Rule GB-Eire 1972 1980 - Mar Sun>=16 2:00s 1:00 BST Rule GB-Eire 1972 1980 - Oct Sun>=23 2:00s 0 GMT # Summer Time Order, 1980 (S.I. 1980/1089) # Summer Time Order, 1982 (S.I. 1982/1673) # Summer Time Order, 1986 (S.I. 1986/223) # Summer Time Order, 1988 (S.I. 1988/931) Rule GB-Eire 1981 1995 - Mar lastSun 1:00u 1:00 BST Rule GB-Eire 1981 1989 - Oct Sun>=23 1:00u 0 GMT # Summer Time Order, 1989 (S.I. 1989/985) # Summer Time Order, 1992 (S.I. 1992/1729) # Summer Time Order 1994 (S.I. 1994/2798) Rule GB-Eire 1990 1995 - Oct Sun>=22 1:00u 0 GMT # Summer Time Order 1997 (S.I. 1997/2982) # See EU for rules starting in 1996. # # Use Europe/London for Jersey, Guernsey, and the Isle of Man. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/London -0:01:15 - LMT 1847 Dec 1 0:00s 0:00 GB-Eire %s 1968 Oct 27 1:00 - BST 1971 Oct 31 2:00u 0:00 GB-Eire %s 1996 0:00 EU GMT/BST Link Europe/London Europe/Jersey Link Europe/London Europe/Guernsey Link Europe/London Europe/Isle_of_Man # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Dublin -0:25:00 - LMT 1880 Aug 2 -0:25:21 - DMT 1916 May 21 2:00 -0:25:21 1:00 IST 1916 Oct 1 2:00s 0:00 GB-Eire %s 1921 Dec 6 # independence 0:00 GB-Eire GMT/IST 1940 Feb 25 2:00 0:00 1:00 IST 1946 Oct 6 2:00 0:00 - GMT 1947 Mar 16 2:00 0:00 1:00 IST 1947 Nov 2 2:00 0:00 - GMT 1948 Apr 18 2:00 0:00 GB-Eire GMT/IST 1968 Oct 27 1:00 - IST 1971 Oct 31 2:00u 0:00 GB-Eire GMT/IST 1996 0:00 EU GMT/IST ############################################################################### # Europe # EU rules are for the European Union, previously known as the EC, EEC, # Common Market, etc. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule EU 1977 1980 - Apr Sun>=1 1:00u 1:00 S Rule EU 1977 only - Sep lastSun 1:00u 0 - Rule EU 1978 only - Oct 1 1:00u 0 - Rule EU 1979 1995 - Sep lastSun 1:00u 0 - Rule EU 1981 max - Mar lastSun 1:00u 1:00 S Rule EU 1996 max - Oct lastSun 1:00u 0 - # The most recent directive covers the years starting in 2002. See: # Directive 2000/84/EC of the European Parliament and of the Council # of 19 January 2001 on summer-time arrangements. # http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:32000L0084:EN:NOT # W-Eur differs from EU only in that W-Eur uses standard time. Rule W-Eur 1977 1980 - Apr Sun>=1 1:00s 1:00 S Rule W-Eur 1977 only - Sep lastSun 1:00s 0 - Rule W-Eur 1978 only - Oct 1 1:00s 0 - Rule W-Eur 1979 1995 - Sep lastSun 1:00s 0 - Rule W-Eur 1981 max - Mar lastSun 1:00s 1:00 S Rule W-Eur 1996 max - Oct lastSun 1:00s 0 - # Older C-Eur rules are for convenience in the tables. # From 1977 on, C-Eur differs from EU only in that C-Eur uses standard time. Rule C-Eur 1916 only - Apr 30 23:00 1:00 S Rule C-Eur 1916 only - Oct 1 1:00 0 - Rule C-Eur 1917 1918 - Apr Mon>=15 2:00s 1:00 S Rule C-Eur 1917 1918 - Sep Mon>=15 2:00s 0 - Rule C-Eur 1940 only - Apr 1 2:00s 1:00 S Rule C-Eur 1942 only - Nov 2 2:00s 0 - Rule C-Eur 1943 only - Mar 29 2:00s 1:00 S Rule C-Eur 1943 only - Oct 4 2:00s 0 - Rule C-Eur 1944 1945 - Apr Mon>=1 2:00s 1:00 S # Whitman gives 1944 Oct 7; go with Shanks & Pottenger. Rule C-Eur 1944 only - Oct 2 2:00s 0 - # From Jesper Nørgaard Welen (2008-07-13): # # I found what is probably a typo of 2:00 which should perhaps be 2:00s # in the C-Eur rule from tz database version 2008d (this part was # corrected in version 2008d). The circumstantial evidence is simply the # tz database itself, as seen below: # # Zone Europe/Paris 0:09:21 - LMT 1891 Mar 15 0:01 # 0:00 France WE%sT 1945 Sep 16 3:00 # # Zone Europe/Monaco 0:29:32 - LMT 1891 Mar 15 # 0:00 France WE%sT 1945 Sep 16 3:00 # # Zone Europe/Belgrade 1:22:00 - LMT 1884 # 1:00 1:00 CEST 1945 Sep 16 2:00s # # Rule France 1945 only - Sep 16 3:00 0 - # Rule Belgium 1945 only - Sep 16 2:00s 0 - # Rule Neth 1945 only - Sep 16 2:00s 0 - # # The rule line to be changed is: # # Rule C-Eur 1945 only - Sep 16 2:00 0 - # # It seems that Paris, Monaco, Rule France, Rule Belgium all agree on # 2:00 standard time, e.g. 3:00 local time. However there are no # countries that use C-Eur rules in September 1945, so the only items # affected are apparently these fictitious zones that translate acronyms # CET and MET: # # Zone CET 1:00 C-Eur CE%sT # Zone MET 1:00 C-Eur ME%sT # # It this is right then the corrected version would look like: # # Rule C-Eur 1945 only - Sep 16 2:00s 0 - # # A small step for mankind though 8-) Rule C-Eur 1945 only - Sep 16 2:00s 0 - Rule C-Eur 1977 1980 - Apr Sun>=1 2:00s 1:00 S Rule C-Eur 1977 only - Sep lastSun 2:00s 0 - Rule C-Eur 1978 only - Oct 1 2:00s 0 - Rule C-Eur 1979 1995 - Sep lastSun 2:00s 0 - Rule C-Eur 1981 max - Mar lastSun 2:00s 1:00 S Rule C-Eur 1996 max - Oct lastSun 2:00s 0 - # E-Eur differs from EU only in that E-Eur switches at midnight local time. Rule E-Eur 1977 1980 - Apr Sun>=1 0:00 1:00 S Rule E-Eur 1977 only - Sep lastSun 0:00 0 - Rule E-Eur 1978 only - Oct 1 0:00 0 - Rule E-Eur 1979 1995 - Sep lastSun 0:00 0 - Rule E-Eur 1981 max - Mar lastSun 0:00 1:00 S Rule E-Eur 1996 max - Oct lastSun 0:00 0 - # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Russia 1917 only - Jul 1 23:00 1:00 MST # Moscow Summer Time Rule Russia 1917 only - Dec 28 0:00 0 MMT # Moscow Mean Time Rule Russia 1918 only - May 31 22:00 2:00 MDST # Moscow Double Summer Time Rule Russia 1918 only - Sep 16 1:00 1:00 MST Rule Russia 1919 only - May 31 23:00 2:00 MDST Rule Russia 1919 only - Jul 1 2:00 1:00 MSD Rule Russia 1919 only - Aug 16 0:00 0 MSK Rule Russia 1921 only - Feb 14 23:00 1:00 MSD Rule Russia 1921 only - Mar 20 23:00 2:00 MSM # Midsummer Rule Russia 1921 only - Sep 1 0:00 1:00 MSD Rule Russia 1921 only - Oct 1 0:00 0 - # Act No.925 of the Council of Ministers of the USSR (1980-10-24): Rule Russia 1981 1984 - Apr 1 0:00 1:00 S Rule Russia 1981 1983 - Oct 1 0:00 0 - # Act No.967 of the Council of Ministers of the USSR (1984-09-13), repeated in # Act No.227 of the Council of Ministers of the USSR (1989-03-14): Rule Russia 1984 1991 - Sep lastSun 2:00s 0 - Rule Russia 1985 1991 - Mar lastSun 2:00s 1:00 S # Rule Russia 1992 only - Mar lastSat 23:00 1:00 S Rule Russia 1992 only - Sep lastSat 23:00 0 - Rule Russia 1993 2010 - Mar lastSun 2:00s 1:00 S Rule Russia 1993 1995 - Sep lastSun 2:00s 0 - Rule Russia 1996 2010 - Oct lastSun 2:00s 0 - # As described below, Russia's 2014 change affects Zone data, not Rule data. # From Alexander Krivenyshev (2011-06-14): # According to Kremlin press service, Russian President Dmitry Medvedev # signed a federal law "On calculation of time" on June 9, 2011. # According to the law Russia is abolishing daylight saving time. # # Medvedev signed a law "On the Calculation of Time" (in russian): # http://bmockbe.ru/events/?ID=7583 # # Medvedev signed a law on the calculation of the time (in russian): # http://www.regnum.ru/news/polit/1413906.html # From Arthur David Olson (2011-06-15): # Take "abolishing daylight saving time" to mean that time is now considered # to be standard. # These are for backward compatibility with older versions. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone WET 0:00 EU WE%sT Zone CET 1:00 C-Eur CE%sT Zone MET 1:00 C-Eur ME%sT Zone EET 2:00 EU EE%sT # Previous editions of this database used abbreviations like MET DST # for Central European Summer Time, but this didn't agree with common usage. # From Markus Kuhn (1996-07-12): # The official German names ... are # # Mitteleuropäische Zeit (MEZ) = UTC+01:00 # Mitteleuropäische Sommerzeit (MESZ) = UTC+02:00 # # as defined in the German Time Act (Gesetz über die Zeitbestimmung (ZeitG), # 1978-07-25, Bundesgesetzblatt, Jahrgang 1978, Teil I, S. 1110-1111).... # I wrote ... to the German Federal Physical-Technical Institution # # Physikalisch-Technische Bundesanstalt (PTB) # Laboratorium 4.41 "Zeiteinheit" # Postfach 3345 # D-38023 Braunschweig # phone: +49 531 592-0 # # ... I received today an answer letter from Dr. Peter Hetzel, head of the PTB # department for time and frequency transmission. He explained that the # PTB translates MEZ and MESZ into English as # # Central European Time (CET) = UTC+01:00 # Central European Summer Time (CEST) = UTC+02:00 # Albania # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Albania 1940 only - Jun 16 0:00 1:00 S Rule Albania 1942 only - Nov 2 3:00 0 - Rule Albania 1943 only - Mar 29 2:00 1:00 S Rule Albania 1943 only - Apr 10 3:00 0 - Rule Albania 1974 only - May 4 0:00 1:00 S Rule Albania 1974 only - Oct 2 0:00 0 - Rule Albania 1975 only - May 1 0:00 1:00 S Rule Albania 1975 only - Oct 2 0:00 0 - Rule Albania 1976 only - May 2 0:00 1:00 S Rule Albania 1976 only - Oct 3 0:00 0 - Rule Albania 1977 only - May 8 0:00 1:00 S Rule Albania 1977 only - Oct 2 0:00 0 - Rule Albania 1978 only - May 6 0:00 1:00 S Rule Albania 1978 only - Oct 1 0:00 0 - Rule Albania 1979 only - May 5 0:00 1:00 S Rule Albania 1979 only - Sep 30 0:00 0 - Rule Albania 1980 only - May 3 0:00 1:00 S Rule Albania 1980 only - Oct 4 0:00 0 - Rule Albania 1981 only - Apr 26 0:00 1:00 S Rule Albania 1981 only - Sep 27 0:00 0 - Rule Albania 1982 only - May 2 0:00 1:00 S Rule Albania 1982 only - Oct 3 0:00 0 - Rule Albania 1983 only - Apr 18 0:00 1:00 S Rule Albania 1983 only - Oct 1 0:00 0 - Rule Albania 1984 only - Apr 1 0:00 1:00 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Tirane 1:19:20 - LMT 1914 1:00 - CET 1940 Jun 16 1:00 Albania CE%sT 1984 Jul 1:00 EU CE%sT # Andorra # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Andorra 0:06:04 - LMT 1901 0:00 - WET 1946 Sep 30 1:00 - CET 1985 Mar 31 2:00 1:00 EU CE%sT # Austria # Milne says Vienna time was 1:05:21. # From Paul Eggert (2006-03-22): Shanks & Pottenger give 1918-06-16 and # 1945-11-18, but the Austrian Federal Office of Metrology and # Surveying (BEV) gives 1918-09-16 and for Vienna gives the "alleged" # date of 1945-04-12 with no time. For the 1980-04-06 transition # Shanks & Pottenger give 02:00, the BEV 00:00. Go with the BEV, # and guess 02:00 for 1945-04-12. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Austria 1920 only - Apr 5 2:00s 1:00 S Rule Austria 1920 only - Sep 13 2:00s 0 - Rule Austria 1946 only - Apr 14 2:00s 1:00 S Rule Austria 1946 1948 - Oct Sun>=1 2:00s 0 - Rule Austria 1947 only - Apr 6 2:00s 1:00 S Rule Austria 1948 only - Apr 18 2:00s 1:00 S Rule Austria 1980 only - Apr 6 0:00 1:00 S Rule Austria 1980 only - Sep 28 0:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Vienna 1:05:21 - LMT 1893 Apr 1:00 C-Eur CE%sT 1920 1:00 Austria CE%sT 1940 Apr 1 2:00s 1:00 C-Eur CE%sT 1945 Apr 2 2:00s 1:00 1:00 CEST 1945 Apr 12 2:00s 1:00 - CET 1946 1:00 Austria CE%sT 1981 1:00 EU CE%sT # Belarus # From Yauhen Kharuzhy (2011-09-16): # By latest Belarus government act Europe/Minsk timezone was changed to # GMT+3 without DST (was GMT+2 with DST). # # Sources (Russian language): # http://www.belta.by/ru/all_news/society/V-Belarusi-otmenjaetsja-perexod-na-sezonnoe-vremja_i_572952.html # http://naviny.by/rubrics/society/2011/09/16/ic_articles_116_175144/ # http://news.tut.by/society/250578.html # # From Alexander Bokovoy (2014-10-09): # Belarussian government decided against changing to winter time.... # http://eng.belta.by/all_news/society/Belarus-decides-against-adjusting-time-in-Russias-wake_i_76335.html # From Paul Eggert (2014-10-08): # Hence Belarus can share time zone abbreviations with Moscow again. # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Minsk 1:50:16 - LMT 1880 1:50 - MMT 1924 May 2 # Minsk Mean Time 2:00 - EET 1930 Jun 21 3:00 - MSK 1941 Jun 28 1:00 C-Eur CE%sT 1944 Jul 3 3:00 Russia MSK/MSD 1990 3:00 - MSK 1991 Mar 31 2:00s 2:00 1:00 EEST 1991 Sep 29 2:00s 2:00 - EET 1992 Mar 29 0:00s 2:00 1:00 EEST 1992 Sep 27 0:00s 2:00 Russia EE%sT 2011 Mar 27 2:00s 3:00 - FET 2014 Oct 26 1:00s 3:00 - MSK # Belgium # # From Paul Eggert (1997-07-02): # Entries from 1918 through 1991 are taken from: # Annuaire de L'Observatoire Royal de Belgique, # Avenue Circulaire, 3, B-1180 BRUXELLES, CLVIIe année, 1991 # (Imprimerie HAYEZ, s.p.r.l., Rue Fin, 4, 1080 BRUXELLES, MCMXC), # pp 8-9. # LMT before 1892 was 0:17:30, according to the official journal of Belgium: # Moniteur Belge, Samedi 30 Avril 1892, N.121. # Thanks to Pascal Delmoitie for these references. # The 1918 rules are listed for completeness; they apply to unoccupied Belgium. # Assume Brussels switched to WET in 1918 when the armistice took effect. # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Belgium 1918 only - Mar 9 0:00s 1:00 S Rule Belgium 1918 1919 - Oct Sat>=1 23:00s 0 - Rule Belgium 1919 only - Mar 1 23:00s 1:00 S Rule Belgium 1920 only - Feb 14 23:00s 1:00 S Rule Belgium 1920 only - Oct 23 23:00s 0 - Rule Belgium 1921 only - Mar 14 23:00s 1:00 S Rule Belgium 1921 only - Oct 25 23:00s 0 - Rule Belgium 1922 only - Mar 25 23:00s 1:00 S Rule Belgium 1922 1927 - Oct Sat>=1 23:00s 0 - Rule Belgium 1923 only - Apr 21 23:00s 1:00 S Rule Belgium 1924 only - Mar 29 23:00s 1:00 S Rule Belgium 1925 only - Apr 4 23:00s 1:00 S # DSH writes that a royal decree of 1926-02-22 specified the Sun following 3rd # Sat in Apr (except if it's Easter, in which case it's one Sunday earlier), # to Sun following 1st Sat in Oct, and that a royal decree of 1928-09-15 # changed the transition times to 02:00 GMT. Rule Belgium 1926 only - Apr 17 23:00s 1:00 S Rule Belgium 1927 only - Apr 9 23:00s 1:00 S Rule Belgium 1928 only - Apr 14 23:00s 1:00 S Rule Belgium 1928 1938 - Oct Sun>=2 2:00s 0 - Rule Belgium 1929 only - Apr 21 2:00s 1:00 S Rule Belgium 1930 only - Apr 13 2:00s 1:00 S Rule Belgium 1931 only - Apr 19 2:00s 1:00 S Rule Belgium 1932 only - Apr 3 2:00s 1:00 S Rule Belgium 1933 only - Mar 26 2:00s 1:00 S Rule Belgium 1934 only - Apr 8 2:00s 1:00 S Rule Belgium 1935 only - Mar 31 2:00s 1:00 S Rule Belgium 1936 only - Apr 19 2:00s 1:00 S Rule Belgium 1937 only - Apr 4 2:00s 1:00 S Rule Belgium 1938 only - Mar 27 2:00s 1:00 S Rule Belgium 1939 only - Apr 16 2:00s 1:00 S Rule Belgium 1939 only - Nov 19 2:00s 0 - Rule Belgium 1940 only - Feb 25 2:00s 1:00 S Rule Belgium 1944 only - Sep 17 2:00s 0 - Rule Belgium 1945 only - Apr 2 2:00s 1:00 S Rule Belgium 1945 only - Sep 16 2:00s 0 - Rule Belgium 1946 only - May 19 2:00s 1:00 S Rule Belgium 1946 only - Oct 7 2:00s 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Brussels 0:17:30 - LMT 1880 0:17:30 - BMT 1892 May 1 12:00 # Brussels MT 0:00 - WET 1914 Nov 8 1:00 - CET 1916 May 1 0:00 1:00 C-Eur CE%sT 1918 Nov 11 11:00u 0:00 Belgium WE%sT 1940 May 20 2:00s 1:00 C-Eur CE%sT 1944 Sep 3 1:00 Belgium CE%sT 1977 1:00 EU CE%sT # Bosnia and Herzegovina # See Europe/Belgrade. # Bulgaria # # From Plamen Simenov via Steffen Thorsen (1999-09-09): # A document of Government of Bulgaria (No.94/1997) says: # EET -> EETDST is in 03:00 Local time in last Sunday of March ... # EETDST -> EET is in 04:00 Local time in last Sunday of October # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Bulg 1979 only - Mar 31 23:00 1:00 S Rule Bulg 1979 only - Oct 1 1:00 0 - Rule Bulg 1980 1982 - Apr Sat>=1 23:00 1:00 S Rule Bulg 1980 only - Sep 29 1:00 0 - Rule Bulg 1981 only - Sep 27 2:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Sofia 1:33:16 - LMT 1880 1:56:56 - IMT 1894 Nov 30 # Istanbul MT? 2:00 - EET 1942 Nov 2 3:00 1:00 C-Eur CE%sT 1945 1:00 - CET 1945 Apr 2 3:00 2:00 - EET 1979 Mar 31 23:00 2:00 Bulg EE%sT 1982 Sep 26 2:00 2:00 C-Eur EE%sT 1991 2:00 E-Eur EE%sT 1997 2:00 EU EE%sT # Croatia # See Europe/Belgrade. # Cyprus # Please see the 'asia' file for Asia/Nicosia. # Czech Republic # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Czech 1945 only - Apr 8 2:00s 1:00 S Rule Czech 1945 only - Nov 18 2:00s 0 - Rule Czech 1946 only - May 6 2:00s 1:00 S Rule Czech 1946 1949 - Oct Sun>=1 2:00s 0 - Rule Czech 1947 only - Apr 20 2:00s 1:00 S Rule Czech 1948 only - Apr 18 2:00s 1:00 S Rule Czech 1949 only - Apr 9 2:00s 1:00 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Prague 0:57:44 - LMT 1850 0:57:44 - PMT 1891 Oct # Prague Mean Time 1:00 C-Eur CE%sT 1944 Sep 17 2:00s 1:00 Czech CE%sT 1979 1:00 EU CE%sT # Use Europe/Prague also for Slovakia. # Denmark, Faroe Islands, and Greenland # From Jesper Nørgaard Welen (2005-04-26): # http://www.hum.aau.dk/~poe/tid/tine/DanskTid.htm says that the law # [introducing standard time] was in effect from 1894-01-01.... # The page http://www.retsinfo.dk/_GETDOCI_/ACCN/A18930008330-REGL # confirms this, and states that the law was put forth 1893-03-29. # # The EU treaty with effect from 1973: # http://www.retsinfo.dk/_GETDOCI_/ACCN/A19722110030-REGL # # This provoked a new law from 1974 to make possible summer time changes # in subsequent decrees with the law # http://www.retsinfo.dk/_GETDOCI_/ACCN/A19740022330-REGL # # It seems however that no decree was set forward until 1980. I have # not found any decree, but in another related law, the effecting DST # changes are stated explicitly to be from 1980-04-06 at 02:00 to # 1980-09-28 at 02:00. If this is true, this differs slightly from # the EU rule in that DST runs to 02:00, not 03:00. We don't know # when Denmark began using the EU rule correctly, but we have only # confirmation of the 1980-time, so I presume it was correct in 1981: # The law is about the management of the extra hour, concerning # working hours reported and effect on obligatory-rest rules (which # was suspended on that night): # http://www.retsinfo.dk/_GETDOCI_/ACCN/C19801120554-REGL # From Jesper Nørgaard Welen (2005-06-11): # The Herning Folkeblad (1980-09-26) reported that the night between # Saturday and Sunday the clock is set back from three to two. # From Paul Eggert (2005-06-11): # Hence the "02:00" of the 1980 law refers to standard time, not # wall-clock time, and so the EU rules were in effect in 1980. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Denmark 1916 only - May 14 23:00 1:00 S Rule Denmark 1916 only - Sep 30 23:00 0 - Rule Denmark 1940 only - May 15 0:00 1:00 S Rule Denmark 1945 only - Apr 2 2:00s 1:00 S Rule Denmark 1945 only - Aug 15 2:00s 0 - Rule Denmark 1946 only - May 1 2:00s 1:00 S Rule Denmark 1946 only - Sep 1 2:00s 0 - Rule Denmark 1947 only - May 4 2:00s 1:00 S Rule Denmark 1947 only - Aug 10 2:00s 0 - Rule Denmark 1948 only - May 9 2:00s 1:00 S Rule Denmark 1948 only - Aug 8 2:00s 0 - # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Copenhagen 0:50:20 - LMT 1890 0:50:20 - CMT 1894 Jan 1 # Copenhagen MT 1:00 Denmark CE%sT 1942 Nov 2 2:00s 1:00 C-Eur CE%sT 1945 Apr 2 2:00 1:00 Denmark CE%sT 1980 1:00 EU CE%sT Zone Atlantic/Faroe -0:27:04 - LMT 1908 Jan 11 # Tórshavn 0:00 - WET 1981 0:00 EU WE%sT # # From Paul Eggert (2004-10-31): # During World War II, Germany maintained secret manned weather stations in # East Greenland and Franz Josef Land, but we don't know their time zones. # My source for this is Wilhelm Dege's book mentioned under Svalbard. # # From Paul Eggert (2006-03-22): # Greenland joined the EU as part of Denmark, obtained home rule on 1979-05-01, # and left the EU on 1985-02-01. It therefore should have been using EU # rules at least through 1984. Shanks & Pottenger say Scoresbysund and Godthåb # used C-Eur rules after 1980, but IATA SSIM (1991/1996) says they use EU # rules since at least 1991. Assume EU rules since 1980. # From Gwillim Law (2001-06-06), citing # (2001-03-15), # and with translations corrected by Steffen Thorsen: # # Greenland has four local times, and the relation to UTC # is according to the following time line: # # The military zone near Thule UTC-4 # Standard Greenland time UTC-3 # Scoresbysund UTC-1 # Danmarkshavn UTC # # In the military area near Thule and in Danmarkshavn DST will not be # introduced. # From Rives McDow (2001-11-01): # # I correspond regularly with the Dansk Polarcenter, and wrote them at # the time to clarify the situation in Thule. Unfortunately, I have # not heard back from them regarding my recent letter. [But I have # info from earlier correspondence.] # # According to the center, a very small local time zone around Thule # Air Base keeps the time according to UTC-4, implementing daylight # savings using North America rules, changing the time at 02:00 local time.... # # The east coast of Greenland north of the community of Scoresbysund # uses UTC in the same way as in Iceland, year round, with no dst. # There are just a few stations on this coast, including the # Danmarkshavn ICAO weather station mentioned in your September 29th # email. The other stations are two sledge patrol stations in # Mestersvig and Daneborg, the air force base at Station Nord, and the # DPC research station at Zackenberg. # # Scoresbysund and two small villages nearby keep time UTC-1 and use # the same daylight savings time period as in West Greenland (Godthåb). # # The rest of Greenland, including Godthåb (this area, although it # includes central Greenland, is known as west Greenland), keeps time # UTC-3, with daylight savings methods according to European rules. # # It is common procedure to use UTC 0 in the wilderness of East and # North Greenland, because it is mainly Icelandic aircraft operators # maintaining traffic in these areas. However, the official status of # this area is that it sticks with Godthåb time. This area might be # considered a dual time zone in some respects because of this. # From Rives McDow (2001-11-19): # I heard back from someone stationed at Thule; the time change took place # there at 2:00 AM. # From Paul Eggert (2006-03-22): # From 1997 on the CIA map shows Danmarkshavn on GMT; # the 1995 map as like Godthåb. # For lack of better info, assume they were like Godthåb before 1996. # startkart.no says Thule does not observe DST, but this is clearly an error, # so go with Shanks & Pottenger for Thule transitions until this year. # For 2007 on assume Thule will stay in sync with US DST rules. # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Thule 1991 1992 - Mar lastSun 2:00 1:00 D Rule Thule 1991 1992 - Sep lastSun 2:00 0 S Rule Thule 1993 2006 - Apr Sun>=1 2:00 1:00 D Rule Thule 1993 2006 - Oct lastSun 2:00 0 S Rule Thule 2007 max - Mar Sun>=8 2:00 1:00 D Rule Thule 2007 max - Nov Sun>=1 2:00 0 S # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Danmarkshavn -1:14:40 - LMT 1916 Jul 28 -3:00 - WGT 1980 Apr 6 2:00 -3:00 EU WG%sT 1996 0:00 - GMT Zone America/Scoresbysund -1:27:52 - LMT 1916 Jul 28 # Ittoqqortoormiit -2:00 - CGT 1980 Apr 6 2:00 -2:00 C-Eur CG%sT 1981 Mar 29 -1:00 EU EG%sT Zone America/Godthab -3:26:56 - LMT 1916 Jul 28 # Nuuk -3:00 - WGT 1980 Apr 6 2:00 -3:00 EU WG%sT Zone America/Thule -4:35:08 - LMT 1916 Jul 28 # Pituffik air base -4:00 Thule A%sT # Estonia # From Peter Ilieve (1994-10-15): # A relative in Tallinn confirms the accuracy of the data for 1989 onwards # [through 1994] and gives the legal authority for it, # a regulation of the Government of Estonia, No. 111 of 1989.... # # From Peter Ilieve (1996-10-28): # [IATA SSIM (1992/1996) claims that the Baltic republics switch at 01:00s, # but a relative confirms that Estonia still switches at 02:00s, writing:] # "I do not [know] exactly but there are some little different # (confusing) rules for International Air and Railway Transport Schedules # conversion in Sunday connected with end of summer time in Estonia.... # A discussion is running about the summer time efficiency and effect on # human physiology. It seems that Estonia maybe will not change to # summer time next spring." # From Peter Ilieve (1998-11-04), heavily edited: # The 1998-09-22 Estonian time law # http://trip.rk.ee/cgi-bin/thw?${BASE}=akt&${OOHTML}=rtd&TA=1998&TO=1&AN=1390 # refers to the Eighth Directive and cites the association agreement between # the EU and Estonia, ratified by the Estonian law (RT II 1995, 22-27, 120). # # I also asked [my relative] whether they use any standard abbreviation # for their standard and summer times. He says no, they use "suveaeg" # (summer time) and "talveaeg" (winter time). # From The Baltic Times (1999-09-09) # via Steffen Thorsen: # This year will mark the last time Estonia shifts to summer time, # a council of the ruling coalition announced Sept. 6.... # But what this could mean for Estonia's chances of joining the European # Union are still unclear. In 1994, the EU declared summer time compulsory # for all member states until 2001. Brussels has yet to decide what to do # after that. # From Mart Oruaas (2000-01-29): # Regulation no. 301 (1999-10-12) obsoletes previous regulation # no. 206 (1998-09-22) and thus sticks Estonia to +02:00 GMT for all # the year round. The regulation is effective 1999-11-01. # From Toomas Soome (2002-02-21): # The Estonian government has changed once again timezone politics. # Now we are using again EU rules. # # From Urmet Jänes (2002-03-28): # The legislative reference is Government decree No. 84 on 2002-02-21. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Tallinn 1:39:00 - LMT 1880 1:39:00 - TMT 1918 Feb # Tallinn Mean Time 1:00 C-Eur CE%sT 1919 Jul 1:39:00 - TMT 1921 May 2:00 - EET 1940 Aug 6 3:00 - MSK 1941 Sep 15 1:00 C-Eur CE%sT 1944 Sep 22 3:00 Russia MSK/MSD 1989 Mar 26 2:00s 2:00 1:00 EEST 1989 Sep 24 2:00s 2:00 C-Eur EE%sT 1998 Sep 22 2:00 EU EE%sT 1999 Nov 1 2:00 - EET 2002 Feb 21 2:00 EU EE%sT # Finland # From Hannu Strang (1994-09-25 06:03:37 UTC): # Well, here in Helsinki we're just changing from summer time to regular one, # and it's supposed to change at 4am... # From Janne Snabb (2010-07-15): # # I noticed that the Finland data is not accurate for years 1981 and 1982. # During these two first trial years the DST adjustment was made one hour # earlier than in forthcoming years. Starting 1983 the adjustment was made # according to the central European standards. # # This is documented in Heikki Oja: Aikakirja 2007, published by The Almanac # Office of University of Helsinki, ISBN 952-10-3221-9, available online (in # Finnish) at # http://almanakka.helsinki.fi/aikakirja/Aikakirja2007kokonaan.pdf # # Page 105 (56 in PDF version) has a handy table of all past daylight savings # transitions. It is easy enough to interpret without Finnish skills. # # This is also confirmed by Finnish Broadcasting Company's archive at: # http://www.yle.fi/elavaarkisto/?s=s&g=1&ag=5&t=&a=3401 # # The news clip from 1981 says that "the time between 2 and 3 o'clock does not # exist tonight." # From Konstantin Hyppönen (2014-06-13): # [Heikki Oja's book Aikakirja 2013] # http://almanakka.helsinki.fi/images/aikakirja/Aikakirja2013kokonaan.pdf # pages 104-105, including a scan from a newspaper published on Apr 2 1942 # say that ... [o]n Apr 2 1942, 24 o'clock (which means Apr 3 1942, # 00:00), clocks were moved one hour forward. The newspaper # mentions "on the night from Thursday to Friday".... # On Oct 4 1942, clocks were moved at 1:00 one hour backwards. # # From Paul Eggert (2014-06-14): # Go with Oja over Shanks. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Finland 1942 only - Apr 2 24:00 1:00 S Rule Finland 1942 only - Oct 4 1:00 0 - Rule Finland 1981 1982 - Mar lastSun 2:00 1:00 S Rule Finland 1981 1982 - Sep lastSun 3:00 0 - # Milne says Helsinki (Helsingfors) time was 1:39:49.2 (official document); # round to nearest. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Helsinki 1:39:49 - LMT 1878 May 31 1:39:49 - HMT 1921 May # Helsinki Mean Time 2:00 Finland EE%sT 1983 2:00 EU EE%sT # Åland Is Link Europe/Helsinki Europe/Mariehamn # France # From Ciro Discepolo (2000-12-20): # # Henri Le Corre, Régimes horaires pour le monde entier, Éditions # Traditionnelles - Paris 2 books, 1993 # # Gabriel, Traité de l'heure dans le monde, Guy Trédaniel, # Paris, 1991 # # Françoise Gauquelin, Problèmes de l'heure résolus en astrologie, # Guy Trédaniel, Paris 1987 # # Shank & Pottenger seem to use '24:00' ambiguously; resolve it with Whitman. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule France 1916 only - Jun 14 23:00s 1:00 S Rule France 1916 1919 - Oct Sun>=1 23:00s 0 - Rule France 1917 only - Mar 24 23:00s 1:00 S Rule France 1918 only - Mar 9 23:00s 1:00 S Rule France 1919 only - Mar 1 23:00s 1:00 S Rule France 1920 only - Feb 14 23:00s 1:00 S Rule France 1920 only - Oct 23 23:00s 0 - Rule France 1921 only - Mar 14 23:00s 1:00 S Rule France 1921 only - Oct 25 23:00s 0 - Rule France 1922 only - Mar 25 23:00s 1:00 S # DSH writes that a law of 1923-05-24 specified 3rd Sat in Apr at 23:00 to 1st # Sat in Oct at 24:00; and that in 1930, because of Easter, the transitions # were Apr 12 and Oct 5. Go with Shanks & Pottenger. Rule France 1922 1938 - Oct Sat>=1 23:00s 0 - Rule France 1923 only - May 26 23:00s 1:00 S Rule France 1924 only - Mar 29 23:00s 1:00 S Rule France 1925 only - Apr 4 23:00s 1:00 S Rule France 1926 only - Apr 17 23:00s 1:00 S Rule France 1927 only - Apr 9 23:00s 1:00 S Rule France 1928 only - Apr 14 23:00s 1:00 S Rule France 1929 only - Apr 20 23:00s 1:00 S Rule France 1930 only - Apr 12 23:00s 1:00 S Rule France 1931 only - Apr 18 23:00s 1:00 S Rule France 1932 only - Apr 2 23:00s 1:00 S Rule France 1933 only - Mar 25 23:00s 1:00 S Rule France 1934 only - Apr 7 23:00s 1:00 S Rule France 1935 only - Mar 30 23:00s 1:00 S Rule France 1936 only - Apr 18 23:00s 1:00 S Rule France 1937 only - Apr 3 23:00s 1:00 S Rule France 1938 only - Mar 26 23:00s 1:00 S Rule France 1939 only - Apr 15 23:00s 1:00 S Rule France 1939 only - Nov 18 23:00s 0 - Rule France 1940 only - Feb 25 2:00 1:00 S # The French rules for 1941-1944 were not used in Paris, but Shanks & Pottenger # write that they were used in Monaco and in many French locations. # Le Corre writes that the upper limit of the free zone was Arnéguy, Orthez, # Mont-de-Marsan, Bazas, Langon, Lamothe-Montravel, Marœuil, La # Rochefoucauld, Champagne-Mouton, La Roche-Posay, La Haye-Descartes, # Loches, Montrichard, Vierzon, Bourges, Moulins, Digoin, # Paray-le-Monial, Montceau-les-Mines, Chalon-sur-Saône, Arbois, # Dole, Morez, St-Claude, and Collonges (Haute-Savoie). Rule France 1941 only - May 5 0:00 2:00 M # Midsummer # Shanks & Pottenger say this transition occurred at Oct 6 1:00, # but go with Denis Excoffier (1997-12-12), # who quotes the Ephémérides astronomiques for 1998 from Bureau des Longitudes # as saying 5/10/41 22hUT. Rule France 1941 only - Oct 6 0:00 1:00 S Rule France 1942 only - Mar 9 0:00 2:00 M Rule France 1942 only - Nov 2 3:00 1:00 S Rule France 1943 only - Mar 29 2:00 2:00 M Rule France 1943 only - Oct 4 3:00 1:00 S Rule France 1944 only - Apr 3 2:00 2:00 M Rule France 1944 only - Oct 8 1:00 1:00 S Rule France 1945 only - Apr 2 2:00 2:00 M Rule France 1945 only - Sep 16 3:00 0 - # Shanks & Pottenger give Mar 28 2:00 and Sep 26 3:00; # go with Excoffier's 28/3/76 0hUT and 25/9/76 23hUT. Rule France 1976 only - Mar 28 1:00 1:00 S Rule France 1976 only - Sep 26 1:00 0 - # Shanks & Pottenger give 0:09:20 for Paris Mean Time, and Whitman 0:09:05, # but Howse quotes the actual French legislation as saying 0:09:21. # Go with Howse. Howse writes that the time in France was officially based # on PMT-0:09:21 until 1978-08-09, when the time base finally switched to UTC. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Paris 0:09:21 - LMT 1891 Mar 15 0:01 0:09:21 - PMT 1911 Mar 11 0:01 # Paris MT # Shanks & Pottenger give 1940 Jun 14 0:00; go with Excoffier and Le Corre. 0:00 France WE%sT 1940 Jun 14 23:00 # Le Corre says Paris stuck with occupied-France time after the liberation; # go with Shanks & Pottenger. 1:00 C-Eur CE%sT 1944 Aug 25 0:00 France WE%sT 1945 Sep 16 3:00 1:00 France CE%sT 1977 1:00 EU CE%sT # Germany # From Markus Kuhn (1998-09-29): # The German time zone web site by the Physikalisch-Technische # Bundesanstalt contains DST information back to 1916. # [See tz-link.htm for the URL.] # From Jörg Schilling (2002-10-23): # In 1945, Berlin was switched to Moscow Summer time (GMT+4) by # http://www.dhm.de/lemo/html/biografien/BersarinNikolai/ # General [Nikolai] Bersarin. # From Paul Eggert (2003-03-08): # http://www.parlament-berlin.de/pds-fraktion.nsf/727459127c8b66ee8525662300459099/defc77cb784f180ac1256c2b0030274b/$FILE/bersarint.pdf # says that Bersarin issued an order to use Moscow time on May 20. # However, Moscow did not observe daylight saving in 1945, so # this was equivalent to CEMT (GMT+3), not GMT+4. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Germany 1946 only - Apr 14 2:00s 1:00 S Rule Germany 1946 only - Oct 7 2:00s 0 - Rule Germany 1947 1949 - Oct Sun>=1 2:00s 0 - # http://www.ptb.de/de/org/4/44/441/salt.htm says the following transition # occurred at 3:00 MEZ, not the 2:00 MEZ given in Shanks & Pottenger. # Go with the PTB. Rule Germany 1947 only - Apr 6 3:00s 1:00 S Rule Germany 1947 only - May 11 2:00s 2:00 M Rule Germany 1947 only - Jun 29 3:00 1:00 S Rule Germany 1948 only - Apr 18 2:00s 1:00 S Rule Germany 1949 only - Apr 10 2:00s 1:00 S Rule SovietZone 1945 only - May 24 2:00 2:00 M # Midsummer Rule SovietZone 1945 only - Sep 24 3:00 1:00 S Rule SovietZone 1945 only - Nov 18 2:00s 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Berlin 0:53:28 - LMT 1893 Apr 1:00 C-Eur CE%sT 1945 May 24 2:00 1:00 SovietZone CE%sT 1946 1:00 Germany CE%sT 1980 1:00 EU CE%sT # From Tobias Conradi (2011-09-12): # Büsingen , surrounded by the Swiss canton # Schaffhausen, did not start observing DST in 1980 as the rest of DE # (West Germany at that time) and DD (East Germany at that time) did. # DD merged into DE, the area is currently covered by code DE in ISO 3166-1, # which in turn is covered by the zone Europe/Berlin. # # Source for the time in Büsingen 1980: # http://www.srf.ch/player/video?id=c012c029-03b7-4c2b-9164-aa5902cd58d3 # From Arthur David Olson (2012-03-03): # Büsingen and Zurich have shared clocks since 1970. Link Europe/Zurich Europe/Busingen # Georgia # Please see the "asia" file for Asia/Tbilisi. # Herodotus (Histories, IV.45) says Georgia north of the Phasis (now Rioni) # is in Europe. Our reference location Tbilisi is in the Asian part. # Gibraltar # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Gibraltar -0:21:24 - LMT 1880 Aug 2 0:00s 0:00 GB-Eire %s 1957 Apr 14 2:00 1:00 - CET 1982 1:00 EU CE%sT # Greece # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S # Whitman gives 1932 Jul 5 - Nov 1; go with Shanks & Pottenger. Rule Greece 1932 only - Jul 7 0:00 1:00 S Rule Greece 1932 only - Sep 1 0:00 0 - # Whitman gives 1941 Apr 25 - ?; go with Shanks & Pottenger. Rule Greece 1941 only - Apr 7 0:00 1:00 S # Whitman gives 1942 Feb 2 - ?; go with Shanks & Pottenger. Rule Greece 1942 only - Nov 2 3:00 0 - Rule Greece 1943 only - Mar 30 0:00 1:00 S Rule Greece 1943 only - Oct 4 0:00 0 - # Whitman gives 1944 Oct 3 - Oct 31; go with Shanks & Pottenger. Rule Greece 1952 only - Jul 1 0:00 1:00 S Rule Greece 1952 only - Nov 2 0:00 0 - Rule Greece 1975 only - Apr 12 0:00s 1:00 S Rule Greece 1975 only - Nov 26 0:00s 0 - Rule Greece 1976 only - Apr 11 2:00s 1:00 S Rule Greece 1976 only - Oct 10 2:00s 0 - Rule Greece 1977 1978 - Apr Sun>=1 2:00s 1:00 S Rule Greece 1977 only - Sep 26 2:00s 0 - Rule Greece 1978 only - Sep 24 4:00 0 - Rule Greece 1979 only - Apr 1 9:00 1:00 S Rule Greece 1979 only - Sep 29 2:00 0 - Rule Greece 1980 only - Apr 1 0:00 1:00 S Rule Greece 1980 only - Sep 28 0:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Athens 1:34:52 - LMT 1895 Sep 14 1:34:52 - AMT 1916 Jul 28 0:01 # Athens MT 2:00 Greece EE%sT 1941 Apr 30 1:00 Greece CE%sT 1944 Apr 4 2:00 Greece EE%sT 1981 # Shanks & Pottenger say it switched to C-Eur in 1981; # go with EU instead, since Greece joined it on Jan 1. 2:00 EU EE%sT # Hungary # From Paul Eggert (2014-07-15): # Dates for 1916-1945 are taken from: # Oross A. Jelen a múlt jövője: a nyári időszámítás Magyarországon 1916-1945. # National Archives of Hungary (2012-10-29). # http://mnl.gov.hu/a_het_dokumentuma/a_nyari_idoszamitas_magyarorszagon_19161945.html # This source does not always give times, which are taken from Shanks # & Pottenger (which disagree about the dates). # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Hungary 1918 only - Apr 1 3:00 1:00 S Rule Hungary 1918 only - Sep 16 3:00 0 - Rule Hungary 1919 only - Apr 15 3:00 1:00 S Rule Hungary 1919 only - Nov 24 3:00 0 - Rule Hungary 1945 only - May 1 23:00 1:00 S Rule Hungary 1945 only - Nov 1 0:00 0 - Rule Hungary 1946 only - Mar 31 2:00s 1:00 S Rule Hungary 1946 1949 - Oct Sun>=1 2:00s 0 - Rule Hungary 1947 1949 - Apr Sun>=4 2:00s 1:00 S Rule Hungary 1950 only - Apr 17 2:00s 1:00 S Rule Hungary 1950 only - Oct 23 2:00s 0 - Rule Hungary 1954 1955 - May 23 0:00 1:00 S Rule Hungary 1954 1955 - Oct 3 0:00 0 - Rule Hungary 1956 only - Jun Sun>=1 0:00 1:00 S Rule Hungary 1956 only - Sep lastSun 0:00 0 - Rule Hungary 1957 only - Jun Sun>=1 1:00 1:00 S Rule Hungary 1957 only - Sep lastSun 3:00 0 - Rule Hungary 1980 only - Apr 6 1:00 1:00 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Budapest 1:16:20 - LMT 1890 Oct 1:00 C-Eur CE%sT 1918 1:00 Hungary CE%sT 1941 Apr 8 1:00 C-Eur CE%sT 1945 1:00 Hungary CE%sT 1980 Sep 28 2:00s 1:00 EU CE%sT # Iceland # # From Adam David (1993-11-06): # The name of the timezone in Iceland for system / mail / news purposes is GMT. # # (1993-12-05): # This material is paraphrased from the 1988 edition of the University of # Iceland Almanak. # # From January 1st, 1908 the whole of Iceland was standardised at 1 hour # behind GMT. Previously, local mean solar time was used in different parts # of Iceland, the almanak had been based on Reykjavik mean solar time which # was 1 hour and 28 minutes behind GMT. # # "first day of winter" referred to [below] means the first day of the 26 weeks # of winter, according to the old icelandic calendar that dates back to the # time the norsemen first settled Iceland. The first day of winter is always # Saturday, but is not dependent on the Julian or Gregorian calendars. # # (1993-12-10): # I have a reference from the Oxford Icelandic-English dictionary for the # beginning of winter, which ties it to the ecclesiastical calendar (and thus # to the julian/gregorian calendar) over the period in question. # the winter begins on the Saturday next before St. Luke's day # (old style), or on St. Luke's day, if a Saturday. # St. Luke's day ought to be traceable from ecclesiastical sources. "old style" # might be a reference to the Julian calendar as opposed to Gregorian, or it # might mean something else (???). # -# From Paul Eggert (2006-03-22): -# The Iceland Almanak, Shanks & Pottenger, and Whitman disagree on many points. -# We go with the Almanak, except for one claim from Shanks & Pottenger, namely -# that Reykavik was 21W57 from 1837 to 1908, local mean time before that. +# From Paul Eggert (2014-11-22): +# The information below is taken from the 1988 Almanak; see +# http://www.almanak.hi.is/klukkan.html # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S -Rule Iceland 1917 1918 - Feb 19 23:00 1:00 S +Rule Iceland 1917 1919 - Feb 19 23:00 1:00 S Rule Iceland 1917 only - Oct 21 1:00 0 - -Rule Iceland 1918 only - Nov 16 1:00 0 - +Rule Iceland 1918 1919 - Nov 16 1:00 0 - +Rule Iceland 1921 only - Mar 19 23:00 1:00 S +Rule Iceland 1921 only - Jun 23 1:00 0 - Rule Iceland 1939 only - Apr 29 23:00 1:00 S -Rule Iceland 1939 only - Nov 29 2:00 0 - +Rule Iceland 1939 only - Oct 29 2:00 0 - Rule Iceland 1940 only - Feb 25 2:00 1:00 S -Rule Iceland 1940 only - Nov 3 2:00 0 - -Rule Iceland 1941 only - Mar 2 1:00s 1:00 S -Rule Iceland 1941 only - Nov 2 1:00s 0 - -Rule Iceland 1942 only - Mar 8 1:00s 1:00 S -Rule Iceland 1942 only - Oct 25 1:00s 0 - +Rule Iceland 1940 1941 - Nov Sun>=2 1:00s 0 - +Rule Iceland 1941 1942 - Mar Sun>=2 1:00s 1:00 S # 1943-1946 - first Sunday in March until first Sunday in winter Rule Iceland 1943 1946 - Mar Sun>=1 1:00s 1:00 S -Rule Iceland 1943 1948 - Oct Sun>=22 1:00s 0 - +Rule Iceland 1942 1948 - Oct Sun>=22 1:00s 0 - # 1947-1967 - first Sunday in April until first Sunday in winter Rule Iceland 1947 1967 - Apr Sun>=1 1:00s 1:00 S -# 1949 Oct transition delayed by 1 week +# 1949 and 1967 Oct transitions delayed by 1 week Rule Iceland 1949 only - Oct 30 1:00s 0 - Rule Iceland 1950 1966 - Oct Sun>=22 1:00s 0 - Rule Iceland 1967 only - Oct 29 1:00s 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] -Zone Atlantic/Reykjavik -1:27:24 - LMT 1837 - -1:27:48 - RMT 1908 # Reykjavik Mean Time? +Zone Atlantic/Reykjavik -1:28 - LMT 1908 -1:00 Iceland IS%sT 1968 Apr 7 1:00s 0:00 - GMT # Italy # # From Paul Eggert (2001-03-06): # Sicily and Sardinia each had their own time zones from 1866 to 1893, # called Palermo Time (+00:53:28) and Cagliari Time (+00:36:32). # During World War II, German-controlled Italy used German time. # But these events all occurred before the 1970 cutoff, # so record only the time in Rome. # # From Paul Eggert (2006-03-22): # For Italian DST we have three sources: Shanks & Pottenger, Whitman, and # F. Pollastri # Day-light Saving Time in Italy (2006-02-03) # http://toi.iriti.cnr.it/uk/ienitlt.html # ('FP' below), taken from an Italian National Electrotechnical Institute # publication. When the three sources disagree, guess who's right, as follows: # # year FP Shanks&P. (S) Whitman (W) Go with: # 1916 06-03 06-03 24:00 06-03 00:00 FP & W # 09-30 09-30 24:00 09-30 01:00 FP; guess 24:00s # 1917 04-01 03-31 24:00 03-31 00:00 FP & S # 09-30 09-29 24:00 09-30 01:00 FP & W # 1918 03-09 03-09 24:00 03-09 00:00 FP & S # 10-06 10-05 24:00 10-06 01:00 FP & W # 1919 03-01 03-01 24:00 03-01 00:00 FP & S # 10-04 10-04 24:00 10-04 01:00 FP; guess 24:00s # 1920 03-20 03-20 24:00 03-20 00:00 FP & S # 09-18 09-18 24:00 10-01 01:00 FP; guess 24:00s # 1944 04-02 04-03 02:00 S (see C-Eur) # 09-16 10-02 03:00 FP; guess 24:00s # 1945 09-14 09-16 24:00 FP; guess 24:00s # 1970 05-21 05-31 00:00 S # 09-20 09-27 00:00 S # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Italy 1916 only - Jun 3 0:00s 1:00 S Rule Italy 1916 only - Oct 1 0:00s 0 - Rule Italy 1917 only - Apr 1 0:00s 1:00 S Rule Italy 1917 only - Sep 30 0:00s 0 - Rule Italy 1918 only - Mar 10 0:00s 1:00 S Rule Italy 1918 1919 - Oct Sun>=1 0:00s 0 - Rule Italy 1919 only - Mar 2 0:00s 1:00 S Rule Italy 1920 only - Mar 21 0:00s 1:00 S Rule Italy 1920 only - Sep 19 0:00s 0 - Rule Italy 1940 only - Jun 15 0:00s 1:00 S Rule Italy 1944 only - Sep 17 0:00s 0 - Rule Italy 1945 only - Apr 2 2:00 1:00 S Rule Italy 1945 only - Sep 15 0:00s 0 - Rule Italy 1946 only - Mar 17 2:00s 1:00 S Rule Italy 1946 only - Oct 6 2:00s 0 - Rule Italy 1947 only - Mar 16 0:00s 1:00 S Rule Italy 1947 only - Oct 5 0:00s 0 - Rule Italy 1948 only - Feb 29 2:00s 1:00 S Rule Italy 1948 only - Oct 3 2:00s 0 - Rule Italy 1966 1968 - May Sun>=22 0:00 1:00 S Rule Italy 1966 1969 - Sep Sun>=22 0:00 0 - Rule Italy 1969 only - Jun 1 0:00 1:00 S Rule Italy 1970 only - May 31 0:00 1:00 S Rule Italy 1970 only - Sep lastSun 0:00 0 - Rule Italy 1971 1972 - May Sun>=22 0:00 1:00 S Rule Italy 1971 only - Sep lastSun 1:00 0 - Rule Italy 1972 only - Oct 1 0:00 0 - Rule Italy 1973 only - Jun 3 0:00 1:00 S Rule Italy 1973 1974 - Sep lastSun 0:00 0 - Rule Italy 1974 only - May 26 0:00 1:00 S Rule Italy 1975 only - Jun 1 0:00s 1:00 S Rule Italy 1975 1977 - Sep lastSun 0:00s 0 - Rule Italy 1976 only - May 30 0:00s 1:00 S Rule Italy 1977 1979 - May Sun>=22 0:00s 1:00 S Rule Italy 1978 only - Oct 1 0:00s 0 - Rule Italy 1979 only - Sep 30 0:00s 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Rome 0:49:56 - LMT 1866 Sep 22 0:49:56 - RMT 1893 Nov 1 0:00s # Rome Mean 1:00 Italy CE%sT 1942 Nov 2 2:00s 1:00 C-Eur CE%sT 1944 Jul 1:00 Italy CE%sT 1980 1:00 EU CE%sT Link Europe/Rome Europe/Vatican Link Europe/Rome Europe/San_Marino # Latvia # From Liene Kanepe (1998-09-17): # I asked about this matter Scientific Secretary of the Institute of Astronomy # of The University of Latvia Dr. paed Mr. Ilgonis Vilks. I also searched the # correct data in juridical acts and I found some juridical documents about # changes in the counting of time in Latvia from 1981.... # # Act No.35 of the Council of Ministers of Latvian SSR of 1981-01-22 ... # according to the Act No.925 of the Council of Ministers of USSR of 1980-10-24 # ...: all year round the time of 2nd time zone + 1 hour, in addition turning # the hands of the clock 1 hour forward on 1 April at 00:00 (GMT 31 March 21:00) # and 1 hour backward on the 1 October at 00:00 (GMT 30 September 20:00). # # Act No.592 of the Council of Ministers of Latvian SSR of 1984-09-24 ... # according to the Act No.967 of the Council of Ministers of USSR of 1984-09-13 # ...: all year round the time of 2nd time zone + 1 hour, in addition turning # the hands of the clock 1 hour forward on the last Sunday of March at 02:00 # (GMT 23:00 on the previous day) and 1 hour backward on the last Sunday of # September at 03:00 (GMT 23:00 on the previous day). # # Act No.81 of the Council of Ministers of Latvian SSR of 1989-03-22 ... # according to the Act No.227 of the Council of Ministers of USSR of 1989-03-14 # ...: since the last Sunday of March 1989 in Lithuanian SSR, Latvian SSR, # Estonian SSR and Kaliningrad region of Russian Federation all year round the # time of 2nd time zone (Moscow time minus one hour). On the territory of Latvia # transition to summer time is performed on the last Sunday of March at 02:00 # (GMT 00:00), turning the hands of the clock 1 hour forward. The end of # daylight saving time is performed on the last Sunday of September at 03:00 # (GMT 00:00), turning the hands of the clock 1 hour backward. Exception is # 1989-03-26, when we must not turn the hands of the clock.... # # The Regulations of the Cabinet of Ministers of the Republic of Latvia of # 1997-01-21 on transition to Summer time ... established the same order of # daylight savings time settings as in the States of the European Union. # From Andrei Ivanov (2000-03-06): # This year Latvia will not switch to Daylight Savings Time (as specified in # The Regulations of the Cabinet of Ministers of the Rep. of Latvia of # 29-Feb-2000 (#79) , # in Latvian for subscribers only). # From RFE/RL Newsline # http://www.rferl.org/newsline/2001/01/3-CEE/cee-030101.html # (2001-01-03), noted after a heads-up by Rives McDow: # The Latvian government on 2 January decided that the country will # institute daylight-saving time this spring, LETA reported. # Last February the three Baltic states decided not to turn back their # clocks one hour in the spring.... # Minister of Economy Aigars Kalvītis noted that Latvia had too few # daylight hours and thus decided to comply with a draft European # Commission directive that provides for instituting daylight-saving # time in EU countries between 2002 and 2006. The Latvian government # urged Lithuania and Estonia to adopt a similar time policy, but it # appears that they will not do so.... # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Latvia 1989 1996 - Mar lastSun 2:00s 1:00 S Rule Latvia 1989 1996 - Sep lastSun 2:00s 0 - # Milne 1899 says Riga was 1:36:28 (Polytechnique House time). # Byalokoz 1919 says Latvia was 1:36:34. # Go with Byalokoz. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Riga 1:36:34 - LMT 1880 1:36:34 - RMT 1918 Apr 15 2:00 # Riga MT 1:36:34 1:00 LST 1918 Sep 16 3:00 # Latvian ST 1:36:34 - RMT 1919 Apr 1 2:00 1:36:34 1:00 LST 1919 May 22 3:00 1:36:34 - RMT 1926 May 11 2:00 - EET 1940 Aug 5 3:00 - MSK 1941 Jul 1:00 C-Eur CE%sT 1944 Oct 13 3:00 Russia MSK/MSD 1989 Mar lastSun 2:00s 2:00 1:00 EEST 1989 Sep lastSun 2:00s 2:00 Latvia EE%sT 1997 Jan 21 2:00 EU EE%sT 2000 Feb 29 2:00 - EET 2001 Jan 2 2:00 EU EE%sT # Liechtenstein # From Paul Eggert (2013-09-09): # Shanks & Pottenger say Vaduz is like Zurich. # From Alois Treindl (2013-09-18): # http://www.eliechtensteinensia.li/LIJ/1978/1938-1978/1941.pdf # ... confirms on p. 6 that Liechtenstein followed Switzerland in 1941 and 1942. # I ... translate only the last two paragraphs: # ... during second world war, in the years 1941 and 1942, Liechtenstein # introduced daylight saving time, adapting to Switzerland. From 1943 on # central European time was in force throughout the year. # From a report of the duke's government to the high council, # regarding the introduction of a time law, of 31 May 1977. Link Europe/Zurich Europe/Vaduz # Lithuania # From Paul Eggert (1996-11-22): # IATA SSIM (1992/1996) says Lithuania uses W-Eur rules, but since it is # known to be wrong about Estonia and Latvia, assume it's wrong here too. # From Marius Gedminas (1998-08-07): # I would like to inform that in this year Lithuanian time zone # (Europe/Vilnius) was changed. # From ELTA No. 972 (2582) (1999-09-29) , # via Steffen Thorsen: # Lithuania has shifted back to the second time zone (GMT plus two hours) # to be valid here starting from October 31, # as decided by the national government on Wednesday.... # The Lithuanian government also announced plans to consider a # motion to give up shifting to summer time in spring, as it was # already done by Estonia. # From the Fact File, Lithuanian State Department of Tourism # (2000-03-27): # Local time is GMT+2 hours ..., no daylight saving. # From a user via Klaus Marten (2003-02-07): # As a candidate for membership of the European Union, Lithuania will # observe Summer Time in 2003, changing its clocks at the times laid # down in EU Directive 2000/84 of 19.I.01 (i.e. at the same times as its # neighbour Latvia). The text of the Lithuanian government Order of # 7.XI.02 to this effect can be found at # http://www.lrvk.lt/nut/11/n1749.htm # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Vilnius 1:41:16 - LMT 1880 1:24:00 - WMT 1917 # Warsaw Mean Time 1:35:36 - KMT 1919 Oct 10 # Kaunas Mean Time 1:00 - CET 1920 Jul 12 2:00 - EET 1920 Oct 9 1:00 - CET 1940 Aug 3 3:00 - MSK 1941 Jun 24 1:00 C-Eur CE%sT 1944 Aug 3:00 Russia MSK/MSD 1991 Mar 31 2:00s 2:00 1:00 EEST 1991 Sep 29 2:00s 2:00 C-Eur EE%sT 1998 2:00 - EET 1998 Mar 29 1:00u 1:00 EU CE%sT 1999 Oct 31 1:00u 2:00 - EET 2003 Jan 1 2:00 EU EE%sT # Luxembourg # Whitman disagrees with most of these dates in minor ways; # go with Shanks & Pottenger. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Lux 1916 only - May 14 23:00 1:00 S Rule Lux 1916 only - Oct 1 1:00 0 - Rule Lux 1917 only - Apr 28 23:00 1:00 S Rule Lux 1917 only - Sep 17 1:00 0 - Rule Lux 1918 only - Apr Mon>=15 2:00s 1:00 S Rule Lux 1918 only - Sep Mon>=15 2:00s 0 - Rule Lux 1919 only - Mar 1 23:00 1:00 S Rule Lux 1919 only - Oct 5 3:00 0 - Rule Lux 1920 only - Feb 14 23:00 1:00 S Rule Lux 1920 only - Oct 24 2:00 0 - Rule Lux 1921 only - Mar 14 23:00 1:00 S Rule Lux 1921 only - Oct 26 2:00 0 - Rule Lux 1922 only - Mar 25 23:00 1:00 S Rule Lux 1922 only - Oct Sun>=2 1:00 0 - Rule Lux 1923 only - Apr 21 23:00 1:00 S Rule Lux 1923 only - Oct Sun>=2 2:00 0 - Rule Lux 1924 only - Mar 29 23:00 1:00 S Rule Lux 1924 1928 - Oct Sun>=2 1:00 0 - Rule Lux 1925 only - Apr 5 23:00 1:00 S Rule Lux 1926 only - Apr 17 23:00 1:00 S Rule Lux 1927 only - Apr 9 23:00 1:00 S Rule Lux 1928 only - Apr 14 23:00 1:00 S Rule Lux 1929 only - Apr 20 23:00 1:00 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Luxembourg 0:24:36 - LMT 1904 Jun 1:00 Lux CE%sT 1918 Nov 25 0:00 Lux WE%sT 1929 Oct 6 2:00s 0:00 Belgium WE%sT 1940 May 14 3:00 1:00 C-Eur WE%sT 1944 Sep 18 3:00 1:00 Belgium CE%sT 1977 1:00 EU CE%sT # Macedonia # See Europe/Belgrade. # Malta # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Malta 1973 only - Mar 31 0:00s 1:00 S Rule Malta 1973 only - Sep 29 0:00s 0 - Rule Malta 1974 only - Apr 21 0:00s 1:00 S Rule Malta 1974 only - Sep 16 0:00s 0 - Rule Malta 1975 1979 - Apr Sun>=15 2:00 1:00 S Rule Malta 1975 1980 - Sep Sun>=15 2:00 0 - Rule Malta 1980 only - Mar 31 2:00 1:00 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Malta 0:58:04 - LMT 1893 Nov 2 0:00s # Valletta 1:00 Italy CE%sT 1942 Nov 2 2:00s 1:00 C-Eur CE%sT 1945 Apr 2 2:00s 1:00 Italy CE%sT 1973 Mar 31 1:00 Malta CE%sT 1981 1:00 EU CE%sT # Moldova # From Paul Eggert (2006-03-22): # A previous version of this database followed Shanks & Pottenger, who write # that Tiraspol switched to Moscow time on 1992-01-19 at 02:00. # However, this is most likely an error, as Moldova declared independence # on 1991-08-27 (the 1992-01-19 date is that of a Russian decree). # In early 1992 there was large-scale interethnic violence in the area # and it's possible that some Russophones continued to observe Moscow time. # But [two people] separately reported via # Jesper Nørgaard that as of 2001-01-24 Tiraspol was like Chisinau. # The Tiraspol entry has therefore been removed for now. # # From Alexander Krivenyshev (2011-10-17): # Pridnestrovian Moldavian Republic (PMR, also known as # "Pridnestrovie") has abolished seasonal clock change (no transition # to the Winter Time). # # News (in Russian): # http://www.kyivpost.ua/russia/news/pridnestrove-otkazalos-ot-perehoda-na-zimnee-vremya-30954.html # http://www.allmoldova.com/moldova-news/1249064116.html # # The substance of this change (reinstatement of the Tiraspol entry) # is from a patch from Petr Machata (2011-10-17) # # From Tim Parenti (2011-10-19) # In addition, being situated at +4651+2938 would give Tiraspol # a pre-1880 LMT offset of 1:58:32. # # (which agrees with the earlier entry that had been removed) # # From Alexander Krivenyshev (2011-10-26) # NO need to divide Moldova into two timezones at this point. # As of today, Transnistria (Pridnestrovie)- Tiraspol reversed its own # decision to abolish DST this winter. # Following Moldova and neighboring Ukraine- Transnistria (Pridnestrovie)- # Tiraspol will go back to winter time on October 30, 2011. # News from Moldova (in russian): # http://ru.publika.md/link_317061.html # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Chisinau 1:55:20 - LMT 1880 1:55 - CMT 1918 Feb 15 # Chisinau MT 1:44:24 - BMT 1931 Jul 24 # Bucharest MT 2:00 Romania EE%sT 1940 Aug 15 2:00 1:00 EEST 1941 Jul 17 1:00 C-Eur CE%sT 1944 Aug 24 3:00 Russia MSK/MSD 1990 3:00 - MSK 1990 May 6 2:00 - EET 1991 2:00 Russia EE%sT 1992 2:00 E-Eur EE%sT 1997 # See Romania commentary for the guessed 1997 transition to EU rules. 2:00 EU EE%sT # Monaco # Shanks & Pottenger give 0:09:20 for Paris Mean Time; go with Howse's # more precise 0:09:21. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Monaco 0:29:32 - LMT 1891 Mar 15 0:09:21 - PMT 1911 Mar 11 # Paris Mean Time 0:00 France WE%sT 1945 Sep 16 3:00 1:00 France CE%sT 1977 1:00 EU CE%sT # Montenegro # See Europe/Belgrade. # Netherlands # Howse writes that the Netherlands' railways used GMT between 1892 and 1940, # but for other purposes the Netherlands used Amsterdam mean time. # However, Robert H. van Gent writes (2001-04-01): # Howse's statement is only correct up to 1909. From 1909-05-01 (00:00:00 # Amsterdam mean time) onwards, the whole of the Netherlands (including # the Dutch railways) was required by law to observe Amsterdam mean time # (19 minutes 32.13 seconds ahead of GMT). This had already been the # common practice (except for the railways) for many decades but it was # not until 1909 when the Dutch government finally defined this by law. # On 1937-07-01 this was changed to 20 minutes (exactly) ahead of GMT and # was generally known as Dutch Time ("Nederlandse Tijd"). # # (2001-04-08): # 1892-05-01 was the date when the Dutch railways were by law required to # observe GMT while the remainder of the Netherlands adhered to the common # practice of following Amsterdam mean time. # # (2001-04-09): # In 1835 the authorities of the province of North Holland requested the # municipal authorities of the towns and cities in the province to observe # Amsterdam mean time but I do not know in how many cases this request was # actually followed. # # From 1852 onwards the Dutch telegraph offices were by law required to # observe Amsterdam mean time. As the time signals from the observatory of # Leiden were also distributed by the telegraph system, I assume that most # places linked up with the telegraph (and railway) system automatically # adopted Amsterdam mean time. # # Although the early Dutch railway companies initially observed a variety # of times, most of them had adopted Amsterdam mean time by 1858 but it # was not until 1866 when they were all required by law to observe # Amsterdam mean time. # The data entries before 1945 are taken from # http://www.phys.uu.nl/~vgent/wettijd/wettijd.htm # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Neth 1916 only - May 1 0:00 1:00 NST # Netherlands Summer Time Rule Neth 1916 only - Oct 1 0:00 0 AMT # Amsterdam Mean Time Rule Neth 1917 only - Apr 16 2:00s 1:00 NST Rule Neth 1917 only - Sep 17 2:00s 0 AMT Rule Neth 1918 1921 - Apr Mon>=1 2:00s 1:00 NST Rule Neth 1918 1921 - Sep lastMon 2:00s 0 AMT Rule Neth 1922 only - Mar lastSun 2:00s 1:00 NST Rule Neth 1922 1936 - Oct Sun>=2 2:00s 0 AMT Rule Neth 1923 only - Jun Fri>=1 2:00s 1:00 NST Rule Neth 1924 only - Mar lastSun 2:00s 1:00 NST Rule Neth 1925 only - Jun Fri>=1 2:00s 1:00 NST # From 1926 through 1939 DST began 05-15, except that it was delayed by a week # in years when 05-15 fell in the Pentecost weekend. Rule Neth 1926 1931 - May 15 2:00s 1:00 NST Rule Neth 1932 only - May 22 2:00s 1:00 NST Rule Neth 1933 1936 - May 15 2:00s 1:00 NST Rule Neth 1937 only - May 22 2:00s 1:00 NST Rule Neth 1937 only - Jul 1 0:00 1:00 S Rule Neth 1937 1939 - Oct Sun>=2 2:00s 0 - Rule Neth 1938 1939 - May 15 2:00s 1:00 S Rule Neth 1945 only - Apr 2 2:00s 1:00 S Rule Neth 1945 only - Sep 16 2:00s 0 - # # Amsterdam Mean Time was +00:19:32.13 exactly, but the .13 is omitted # below because the current format requires GMTOFF to be an integer. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Amsterdam 0:19:32 - LMT 1835 0:19:32 Neth %s 1937 Jul 1 0:20 Neth NE%sT 1940 May 16 0:00 # Dutch Time 1:00 C-Eur CE%sT 1945 Apr 2 2:00 1:00 Neth CE%sT 1977 1:00 EU CE%sT # Norway # http://met.no/met/met_lex/q_u/sommertid.html (2004-01) agrees with Shanks & # Pottenger. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Norway 1916 only - May 22 1:00 1:00 S Rule Norway 1916 only - Sep 30 0:00 0 - Rule Norway 1945 only - Apr 2 2:00s 1:00 S Rule Norway 1945 only - Oct 1 2:00s 0 - Rule Norway 1959 1964 - Mar Sun>=15 2:00s 1:00 S Rule Norway 1959 1965 - Sep Sun>=15 2:00s 0 - Rule Norway 1965 only - Apr 25 2:00s 1:00 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Oslo 0:43:00 - LMT 1895 Jan 1 1:00 Norway CE%sT 1940 Aug 10 23:00 1:00 C-Eur CE%sT 1945 Apr 2 2:00 1:00 Norway CE%sT 1980 1:00 EU CE%sT # Svalbard & Jan Mayen # From Steffen Thorsen (2001-05-01): # Although I could not find it explicitly, it seems that Jan Mayen and # Svalbard have been using the same time as Norway at least since the # time they were declared as parts of Norway. Svalbard was declared # as a part of Norway by law of 1925-07-17 no 11, section 4 and Jan # Mayen by law of 1930-02-27 no 2, section 2. (From # and # ). The law/regulation # for normal/standard time in Norway is from 1894-06-29 no 1 (came # into operation on 1895-01-01) and Svalbard/Jan Mayen seem to be a # part of this law since 1925/1930. (From # ) I have not been # able to find if Jan Mayen used a different time zone (e.g. -0100) # before 1930. Jan Mayen has only been "inhabited" since 1921 by # Norwegian meteorologists and maybe used the same time as Norway ever # since 1921. Svalbard (Arctic/Longyearbyen) has been inhabited since # before 1895, and therefore probably changed the local time somewhere # between 1895 and 1925 (inclusive). # From Paul Eggert (2013-09-04): # # Actually, Jan Mayen was never occupied by Germany during World War II, # so it must have diverged from Oslo time during the war, as Oslo was # keeping Berlin time. # # says that the meteorologists # burned down their station in 1940 and left the island, but returned in # 1941 with a small Norwegian garrison and continued operations despite # frequent air attacks from Germans. In 1943 the Americans established a # radiolocating station on the island, called "Atlantic City". Possibly # the UT offset changed during the war, but I think it unlikely that # Jan Mayen used German daylight-saving rules. # # Svalbard is more complicated, as it was raided in August 1941 by an # Allied party that evacuated the civilian population to England (says # ). The Svalbard FAQ # says that the Germans were # expelled on 1942-05-14. However, small parties of Germans did return, # and according to Wilhelm Dege's book "War North of 80" (1954) # http://www.ucalgary.ca/UofC/departments/UP/1-55238/1-55238-110-2.html # the German armed forces at the Svalbard weather station code-named # Haudegen did not surrender to the Allies until September 1945. # # All these events predate our cutoff date of 1970, so use Europe/Oslo # for these regions. Link Europe/Oslo Arctic/Longyearbyen # Poland # The 1919 dates and times can be found in Tygodnik Urzędowy nr 1 (1919-03-20), # pp 1-2. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Poland 1918 1919 - Sep 16 2:00s 0 - Rule Poland 1919 only - Apr 15 2:00s 1:00 S Rule Poland 1944 only - Apr 3 2:00s 1:00 S # Whitman gives 1944 Nov 30; go with Shanks & Pottenger. Rule Poland 1944 only - Oct 4 2:00 0 - # For 1944-1948 Whitman gives the previous day; go with Shanks & Pottenger. Rule Poland 1945 only - Apr 29 0:00 1:00 S Rule Poland 1945 only - Nov 1 0:00 0 - # For 1946 on the source is Kazimierz Borkowski, # Toruń Center for Astronomy, Dept. of Radio Astronomy, Nicolaus Copernicus U., # http://www.astro.uni.torun.pl/~kb/Artykuly/U-PA/Czas2.htm#tth_tAb1 # Thanks to Przemysław Augustyniak (2005-05-28) for this reference. # He also gives these further references: # Mon Pol nr 13, poz 162 (1995) # Druk nr 2180 (2003) Rule Poland 1946 only - Apr 14 0:00s 1:00 S Rule Poland 1946 only - Oct 7 2:00s 0 - Rule Poland 1947 only - May 4 2:00s 1:00 S Rule Poland 1947 1949 - Oct Sun>=1 2:00s 0 - Rule Poland 1948 only - Apr 18 2:00s 1:00 S Rule Poland 1949 only - Apr 10 2:00s 1:00 S Rule Poland 1957 only - Jun 2 1:00s 1:00 S Rule Poland 1957 1958 - Sep lastSun 1:00s 0 - Rule Poland 1958 only - Mar 30 1:00s 1:00 S Rule Poland 1959 only - May 31 1:00s 1:00 S Rule Poland 1959 1961 - Oct Sun>=1 1:00s 0 - Rule Poland 1960 only - Apr 3 1:00s 1:00 S Rule Poland 1961 1964 - May lastSun 1:00s 1:00 S Rule Poland 1962 1964 - Sep lastSun 1:00s 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Warsaw 1:24:00 - LMT 1880 1:24:00 - WMT 1915 Aug 5 # Warsaw Mean Time 1:00 C-Eur CE%sT 1918 Sep 16 3:00 2:00 Poland EE%sT 1922 Jun 1:00 Poland CE%sT 1940 Jun 23 2:00 1:00 C-Eur CE%sT 1944 Oct 1:00 Poland CE%sT 1977 1:00 W-Eur CE%sT 1988 1:00 EU CE%sT # Portugal # # From Paul Eggert (2014-08-11), after a heads-up from Stephen Colebourne: # According to a Portuguese decree (1911-05-26) # http://dre.pt/pdf1sdip/1911/05/12500/23132313.pdf # Lisbon was at -0:36:44.68, but switched to GMT on 1912-01-01 at 00:00. # Round the old offset to -0:36:45. This agrees with Willett but disagrees # with Shanks, who says the transition occurred on 1911-05-24 at 00:00 for # Europe/Lisbon, Atlantic/Azores, and Atlantic/Madeira. # # From Rui Pedro Salgueiro (1992-11-12): # Portugal has recently (September, 27) changed timezone # (from WET to MET or CET) to harmonize with EEC. # # Martin Bruckmann (1996-02-29) reports via Peter Ilieve # that Portugal is reverting to 0:00 by not moving its clocks this spring. # The new Prime Minister was fed up with getting up in the dark in the winter. # # From Paul Eggert (1996-11-12): # IATA SSIM (1991-09) reports several 1991-09 and 1992-09 transitions # at 02:00u, not 01:00u. Assume that these are typos. # IATA SSIM (1991/1992) reports that the Azores were at -1:00. # IATA SSIM (1993-02) says +0:00; later issues (through 1996-09) say -1:00. # Guess that the Azores changed to EU rules in 1992 (since that's when Portugal # harmonized with the EU), and that they stayed +0:00 that winter. # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S # DSH writes that despite Decree 1,469 (1915), the change to the clocks was not # done every year, depending on what Spain did, because of railroad schedules. # Go with Shanks & Pottenger. Rule Port 1916 only - Jun 17 23:00 1:00 S # Whitman gives 1916 Oct 31; go with Shanks & Pottenger. Rule Port 1916 only - Nov 1 1:00 0 - Rule Port 1917 only - Feb 28 23:00s 1:00 S Rule Port 1917 1921 - Oct 14 23:00s 0 - Rule Port 1918 only - Mar 1 23:00s 1:00 S Rule Port 1919 only - Feb 28 23:00s 1:00 S Rule Port 1920 only - Feb 29 23:00s 1:00 S Rule Port 1921 only - Feb 28 23:00s 1:00 S Rule Port 1924 only - Apr 16 23:00s 1:00 S Rule Port 1924 only - Oct 14 23:00s 0 - Rule Port 1926 only - Apr 17 23:00s 1:00 S Rule Port 1926 1929 - Oct Sat>=1 23:00s 0 - Rule Port 1927 only - Apr 9 23:00s 1:00 S Rule Port 1928 only - Apr 14 23:00s 1:00 S Rule Port 1929 only - Apr 20 23:00s 1:00 S Rule Port 1931 only - Apr 18 23:00s 1:00 S # Whitman gives 1931 Oct 8; go with Shanks & Pottenger. Rule Port 1931 1932 - Oct Sat>=1 23:00s 0 - Rule Port 1932 only - Apr 2 23:00s 1:00 S Rule Port 1934 only - Apr 7 23:00s 1:00 S # Whitman gives 1934 Oct 5; go with Shanks & Pottenger. Rule Port 1934 1938 - Oct Sat>=1 23:00s 0 - # Shanks & Pottenger give 1935 Apr 30; go with Whitman. Rule Port 1935 only - Mar 30 23:00s 1:00 S Rule Port 1936 only - Apr 18 23:00s 1:00 S # Whitman gives 1937 Apr 2; go with Shanks & Pottenger. Rule Port 1937 only - Apr 3 23:00s 1:00 S Rule Port 1938 only - Mar 26 23:00s 1:00 S Rule Port 1939 only - Apr 15 23:00s 1:00 S # Whitman gives 1939 Oct 7; go with Shanks & Pottenger. Rule Port 1939 only - Nov 18 23:00s 0 - Rule Port 1940 only - Feb 24 23:00s 1:00 S # Shanks & Pottenger give 1940 Oct 7; go with Whitman. Rule Port 1940 1941 - Oct 5 23:00s 0 - Rule Port 1941 only - Apr 5 23:00s 1:00 S Rule Port 1942 1945 - Mar Sat>=8 23:00s 1:00 S Rule Port 1942 only - Apr 25 22:00s 2:00 M # Midsummer Rule Port 1942 only - Aug 15 22:00s 1:00 S Rule Port 1942 1945 - Oct Sat>=24 23:00s 0 - Rule Port 1943 only - Apr 17 22:00s 2:00 M Rule Port 1943 1945 - Aug Sat>=25 22:00s 1:00 S Rule Port 1944 1945 - Apr Sat>=21 22:00s 2:00 M Rule Port 1946 only - Apr Sat>=1 23:00s 1:00 S Rule Port 1946 only - Oct Sat>=1 23:00s 0 - Rule Port 1947 1949 - Apr Sun>=1 2:00s 1:00 S Rule Port 1947 1949 - Oct Sun>=1 2:00s 0 - # Shanks & Pottenger say DST was observed in 1950; go with Whitman. # Whitman gives Oct lastSun for 1952 on; go with Shanks & Pottenger. Rule Port 1951 1965 - Apr Sun>=1 2:00s 1:00 S Rule Port 1951 1965 - Oct Sun>=1 2:00s 0 - Rule Port 1977 only - Mar 27 0:00s 1:00 S Rule Port 1977 only - Sep 25 0:00s 0 - Rule Port 1978 1979 - Apr Sun>=1 0:00s 1:00 S Rule Port 1978 only - Oct 1 0:00s 0 - Rule Port 1979 1982 - Sep lastSun 1:00s 0 - Rule Port 1980 only - Mar lastSun 0:00s 1:00 S Rule Port 1981 1982 - Mar lastSun 1:00s 1:00 S Rule Port 1983 only - Mar lastSun 2:00s 1:00 S # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Lisbon -0:36:45 - LMT 1884 -0:36:45 - LMT 1912 Jan 1 # Lisbon Mean Time 0:00 Port WE%sT 1966 Apr 3 2:00 1:00 - CET 1976 Sep 26 1:00 0:00 Port WE%sT 1983 Sep 25 1:00s 0:00 W-Eur WE%sT 1992 Sep 27 1:00s 1:00 EU CE%sT 1996 Mar 31 1:00u 0:00 EU WE%sT Zone Atlantic/Azores -1:42:40 - LMT 1884 # Ponta Delgada -1:54:32 - HMT 1912 Jan 1 # Horta Mean Time -2:00 Port AZO%sT 1966 Apr 3 2:00 # Azores Time -1:00 Port AZO%sT 1983 Sep 25 1:00s -1:00 W-Eur AZO%sT 1992 Sep 27 1:00s 0:00 EU WE%sT 1993 Mar 28 1:00u -1:00 EU AZO%sT Zone Atlantic/Madeira -1:07:36 - LMT 1884 # Funchal -1:07:36 - FMT 1912 Jan 1 # Funchal Mean Time -1:00 Port MAD%sT 1966 Apr 3 2:00 # Madeira Time 0:00 Port WE%sT 1983 Sep 25 1:00s 0:00 EU WE%sT # Romania # # From Paul Eggert (1999-10-07): # Nine O'clock # (1998-10-23) reports that the switch occurred at # 04:00 local time in fall 1998. For lack of better info, # assume that Romania and Moldova switched to EU rules in 1997, # the same year as Bulgaria. # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Romania 1932 only - May 21 0:00s 1:00 S Rule Romania 1932 1939 - Oct Sun>=1 0:00s 0 - Rule Romania 1933 1939 - Apr Sun>=2 0:00s 1:00 S Rule Romania 1979 only - May 27 0:00 1:00 S Rule Romania 1979 only - Sep lastSun 0:00 0 - Rule Romania 1980 only - Apr 5 23:00 1:00 S Rule Romania 1980 only - Sep lastSun 1:00 0 - Rule Romania 1991 1993 - Mar lastSun 0:00s 1:00 S Rule Romania 1991 1993 - Sep lastSun 0:00s 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Bucharest 1:44:24 - LMT 1891 Oct 1:44:24 - BMT 1931 Jul 24 # Bucharest MT 2:00 Romania EE%sT 1981 Mar 29 2:00s 2:00 C-Eur EE%sT 1991 2:00 Romania EE%sT 1994 2:00 E-Eur EE%sT 1997 2:00 EU EE%sT # Russia # From Alexander Krivenyshev (2011-09-15): # Based on last Russian Government Decree # 725 on August 31, 2011 # (Government document # http://www.government.ru/gov/results/16355/print/ # in Russian) # there are few corrections have to be made for some Russian time zones... # All updated Russian Time Zones were placed in table and translated to English # by WorldTimeZone.com at the link below: # http://www.worldtimezone.com/dst_news/dst_news_russia36.htm # From Sanjeev Gupta (2011-09-27): # Scans of [Decree #23 of January 8, 1992] are available at: # http://government.consultant.ru/page.aspx?1223966 # They are in Cyrillic letters (presumably Russian). # From Arthur David Olson (2012-05-09): # Regarding the instant when clocks in time-zone-shifting parts of Russia # changed in September 2011: # # One source is # http://government.ru/gov/results/16355/ # which, according to translate.google.com, begins "Decree of August 31, # 2011 No 725" and contains no other dates or "effective date" information. # # Another source is # http://www.rg.ru/2011/09/06/chas-zona-dok.html # which, according to translate.google.com, begins "Resolution of the # Government of the Russian Federation on August 31, 2011 N 725" and also # contains "Date first official publication: September 6, 2011 Posted on: # in the 'RG' - Federal Issue number 5573 September 6, 2011" but which # does not contain any "effective date" information. # # Another source is # http://en.wikipedia.org/wiki/Oymyakonsky_District#cite_note-RuTime-7 # which, in note 8, contains "Resolution #725 of August 31, 2011... # Effective as of after 7 days following the day of the official publication" # but which does not contain any reference to September 6, 2011. # # The Wikipedia article refers to # http://base.consultant.ru/cons/cgi/online.cgi?req=doc;base=LAW;n=118896 # which seems to copy the text of the government.ru page. # # Tobias Conradi combines Wikipedia's # "as of after 7 days following the day of the official publication" # with www.rg.ru's "Date of first official publication: September 6, 2011" to # get September 13, 2011 as the cutover date (unusually, a Tuesday, as Tobias # Conradi notes). # # None of the sources indicates a time of day for changing clocks. # # Go with 2011-09-13 0:00s. # From Alexander Krivenyshev (2014-07-01): # According to the Russian news (ITAR-TASS News Agency) # http://en.itar-tass.com/russia/738562 # the State Duma has approved ... the draft bill on returning to # winter time standard and return Russia 11 time zones. The new # regulations will come into effect on October 26, 2014 at 02:00 ... # http://asozd2.duma.gov.ru/main.nsf/%28Spravka%29?OpenAgent&RN=431985-6&02 # Here is a link where we put together table (based on approved Bill N # 431985-6) with proposed 11 Russian time zones and corresponding # areas/cities/administrative centers in the Russian Federation (in English): # http://www.worldtimezone.com/dst_news/dst_news_russia65.html # # From Alexander Krivenyshev (2014-07-22): # Putin signed the Federal Law 431985-6 ... (in Russian) # http://itar-tass.com/obschestvo/1333711 # http://www.pravo.gov.ru:8080/page.aspx?111660 # http://www.kremlin.ru/acts/46279 # From October 26, 2014 the new Russian time zone map will looks like this: # http://www.worldtimezone.com/dst_news/dst_news_russia-map-2014-07.html # From Paul Eggert (2006-03-22): # Except for Moscow after 1919-07-01, I invented the time zone abbreviations. # Moscow time zone abbreviations after 1919-07-01, and Moscow rules after 1991, # are from Andrey A. Chernov. The rest is from Shanks & Pottenger, # except we follow Chernov's report that 1992 DST transitions were Sat # 23:00, not Sun 02:00s. # # From Stanislaw A. Kuzikowski (1994-06-29): # But now it is some months since Novosibirsk is 3 hours ahead of Moscow! # I do not know why they have decided to make this change; # as far as I remember it was done exactly during winter->summer switching # so we (Novosibirsk) simply did not switch. # # From Andrey A. Chernov (1996-10-04): # 'MSK' and 'MSD' were born and used initially on Moscow computers with # UNIX-like OSes by several developer groups (e.g. Demos group, Kiae group).... # The next step was the UUCP network, the Relcom predecessor # (used mainly for mail), and MSK/MSD was actively used there. # # From Chris Carrier (1996-10-30): # According to a friend of mine who rode the Trans-Siberian Railroad from # Moscow to Irkutsk in 1995, public air and rail transport in Russia ... # still follows Moscow time, no matter where in Russia it is located. # # For Grozny, Chechnya, we have the following story from # John Daniszewski, "Scavengers in the Rubble", Los Angeles Times (2001-02-07): # News - often false - is spread by word of mouth. A rumor that it was # time to move the clocks back put this whole city out of sync with # the rest of Russia for two weeks - even soldiers stationed here began # enforcing curfew at the wrong time. # # From Gwillim Law (2001-06-05): # There's considerable evidence that Sakhalin Island used to be in # UTC+11, and has changed to UTC+10, in this decade. I start with the # SSIM, which listed Yuzhno-Sakhalinsk in zone RU10 along with Magadan # until February 1997, and then in RU9 with Khabarovsk and Vladivostok # since September 1997.... Although the Kuril Islands are # administratively part of Sakhalin oblast', they appear to have # remained on UTC+11 along with Magadan. # From Tim Parenti (2014-07-06): # The comments detailing the coverage of each Russian zone are meant to assist # with maintenance only and represent our best guesses as to which regions # are covered by each zone. They are not meant to be taken as an authoritative # listing. The region codes listed come from # http://en.wikipedia.org/w/?title=Federal_subjects_of_Russia&oldid=611810498 # and are used for convenience only; no guarantees are made regarding their # future stability. ISO 3166-2:RU codes are also listed for first-level # divisions where available. # Zone NAME GMTOFF RULES FORMAT [UNTIL] # From Tim Parenti (2014-07-03): # Europe/Kaliningrad covers... # 39 RU-KGD Kaliningrad Oblast Zone Europe/Kaliningrad 1:22:00 - LMT 1893 Apr 1:00 C-Eur CE%sT 1945 2:00 Poland CE%sT 1946 3:00 Russia MSK/MSD 1991 Mar 31 2:00s 2:00 Russia EE%sT 2011 Mar 27 2:00s 3:00 - FET 2014 Oct 26 2:00s 2:00 - EET # From Tim Parenti (2014-07-03), per Oscar van Vlijmen (2001-08-25): # Europe/Moscow covers... # 01 RU-AD Adygea, Republic of # 05 RU-DA Dagestan, Republic of # 06 RU-IN Ingushetia, Republic of # 07 RU-KB Kabardino-Balkar Republic # 08 RU-KL Kalmykia, Republic of # 09 RU-KC Karachay-Cherkess Republic # 10 RU-KR Karelia, Republic of # 11 RU-KO Komi Republic # 12 RU-ME Mari El Republic # 13 RU-MO Mordovia, Republic of # 15 RU-SE North Ossetia-Alania, Republic of # 16 RU-TA Tatarstan, Republic of # 20 RU-CE Chechen Republic # 21 RU-CU Chuvash Republic # 23 RU-KDA Krasnodar Krai # 26 RU-STA Stavropol Krai # 29 RU-ARK Arkhangelsk Oblast # 31 RU-BEL Belgorod Oblast # 32 RU-BRY Bryansk Oblast # 33 RU-VLA Vladimir Oblast # 35 RU-VLG Vologda Oblast # 36 RU-VOR Voronezh Oblast # 37 RU-IVA Ivanovo Oblast # 40 RU-KLU Kaluga Oblast # 44 RU-KOS Kostroma Oblast # 46 RU-KRS Kursk Oblast # 47 RU-LEN Leningrad Oblast # 48 RU-LIP Lipetsk Oblast # 50 RU-MOS Moscow Oblast # 51 RU-MUR Murmansk Oblast # 52 RU-NIZ Nizhny Novgorod Oblast # 53 RU-NGR Novgorod Oblast # 57 RU-ORL Oryol Oblast # 58 RU-PNZ Penza Oblast # 60 RU-PSK Pskov Oblast # 61 RU-ROS Rostov Oblast # 62 RU-RYA Ryazan Oblast # 67 RU-SMO Smolensk Oblast # 68 RU-TAM Tambov Oblast # 69 RU-TVE Tver Oblast # 71 RU-TUL Tula Oblast # 73 RU-ULY Ulyanovsk Oblast # 76 RU-YAR Yaroslavl Oblast # 77 RU-MOW Moscow # 78 RU-SPE Saint Petersburg # 83 RU-NEN Nenets Autonomous Okrug # From Vladimir Karpinsky (2014-07-08): # LMT in Moscow (before Jul 3, 1916) is 2:30:17, that was defined by Moscow # Observatory (coordinates: 55 deg. 45'29.70", 37 deg. 34'05.30").... # LMT in Moscow since Jul 3, 1916 is 2:31:01 as a result of new standard. # (The info is from the book by Byalokoz ... p. 18.) # The time in St. Petersburg as capital of Russia was defined by # Pulkov observatory, near St. Petersburg. In 1916 LMT Moscow # was synchronized with LMT St. Petersburg (+30 minutes), (Pulkov observatory # coordinates: 59 deg. 46'18.70", 30 deg. 19'40.70") so 30 deg. 19'40.70" > # 2h01m18.7s = 2:01:19. LMT Moscow = LMT St.Petersburg + 30m 2:01:19 + 0:30 = # 2:31:19 ... # # From Paul Eggert (2014-07-08): # Milne does not list Moscow, but suggests that its time might be listed in # Résumés mensuels et annuels des observations météorologiques (1895). # Presumably this is OCLC 85825704, a journal published with parallel text in # Russian and French. This source has not been located; go with Karpinsky. Zone Europe/Moscow 2:30:17 - LMT 1880 2:30:17 - MMT 1916 Jul 3 # Moscow Mean Time 2:31:19 Russia %s 1919 Jul 1 2:00 3:00 Russia %s 1921 Oct 3:00 Russia MSK/MSD 1922 Oct 2:00 - EET 1930 Jun 21 3:00 Russia MSK/MSD 1991 Mar 31 2:00s 2:00 Russia EE%sT 1992 Jan 19 2:00s 3:00 Russia MSK/MSD 2011 Mar 27 2:00s 4:00 - MSK 2014 Oct 26 2:00s 3:00 - MSK # From Tim Parenti (2014-07-03): # Europe/Simferopol covers... # ** **** Crimea, Republic of # ** **** Sevastopol Zone Europe/Simferopol 2:16:24 - LMT 1880 2:16 - SMT 1924 May 2 # Simferopol Mean T 2:00 - EET 1930 Jun 21 3:00 - MSK 1941 Nov 1:00 C-Eur CE%sT 1944 Apr 13 3:00 Russia MSK/MSD 1990 3:00 - MSK 1990 Jul 1 2:00 2:00 - EET 1992 # Central Crimea used Moscow time 1994/1997. # # From Paul Eggert (2006-03-22): # The _Economist_ (1994-05-28, p 45) reports that central Crimea switched # from Kiev to Moscow time sometime after the January 1994 elections. # Shanks (1999) says "date of change uncertain", but implies that it happened # sometime between the 1994 DST switches. Shanks & Pottenger simply say # 1994-09-25 03:00, but that can't be right. For now, guess it # changed in May. 2:00 E-Eur EE%sT 1994 May # From IATA SSIM (1994/1997), which also says that Kerch is still like Kiev. 3:00 E-Eur MSK/MSD 1996 Mar 31 3:00s 3:00 1:00 MSD 1996 Oct 27 3:00s # IATA SSIM (1997-09) says Crimea switched to EET/EEST. # Assume it happened in March by not changing the clocks. 3:00 Russia MSK/MSD 1997 3:00 - MSK 1997 Mar lastSun 1:00u # From Alexander Krivenyshev (2014-03-17): # time change at 2:00 (2am) on March 30, 2014 # http://vz.ru/news/2014/3/17/677464.html # From Paul Eggert (2014-03-30): # Simferopol and Sevastopol reportedly changed their central town clocks # late the previous day, but this appears to have been ceremonial # and the discrepancies are small enough to not worry about. 2:00 EU EE%sT 2014 Mar 30 2:00 4:00 - MSK 2014 Oct 26 2:00s 3:00 - MSK # From Tim Parenti (2014-07-03): # Europe/Volgograd covers... # 30 RU-AST Astrakhan Oblast # 34 RU-VGG Volgograd Oblast # 43 RU-KIR Kirov Oblast # 64 RU-SAR Saratov Oblast # From Paul Eggert (2006-05-09): # Shanks & Pottenger say Kirov is still at +0400 but Wikipedia says +0300. # Perhaps it switched after the others? But we have no data. Zone Europe/Volgograd 2:57:40 - LMT 1920 Jan 3 3:00 - TSAT 1925 Apr 6 # Tsaritsyn Time 3:00 - STAT 1930 Jun 21 # Stalingrad Time 4:00 - STAT 1961 Nov 11 4:00 Russia VOL%sT 1989 Mar 26 2:00s # Volgograd T 3:00 Russia VOL%sT 1991 Mar 31 2:00s 4:00 - VOLT 1992 Mar 29 2:00s 3:00 Russia MSK 2011 Mar 27 2:00s 4:00 - MSK 2014 Oct 26 2:00s 3:00 - MSK # From Tim Parenti (2014-07-03), per Oscar van Vlijmen (2001-08-25): # Europe/Samara covers... # 18 RU-UD Udmurt Republic # 63 RU-SAM Samara Oblast # Byalokoz 1919 says Samara was 3:20:20. Zone Europe/Samara 3:20:20 - LMT 1919 Jul 1 2:00 3:00 - SAMT 1930 Jun 21 4:00 - SAMT 1935 Jan 27 4:00 Russia KUY%sT 1989 Mar 26 2:00s # Kuybyshev 3:00 Russia MSK/MSD 1991 Mar 31 2:00s 2:00 Russia EE%sT 1991 Sep 29 2:00s 3:00 - KUYT 1991 Oct 20 3:00 4:00 Russia SAM%sT 2010 Mar 28 2:00s # Samara Time 3:00 Russia SAM%sT 2011 Mar 27 2:00s 4:00 - SAMT # From Tim Parenti (2014-07-03), per Oscar van Vlijmen (2001-08-25): # Asia/Yekaterinburg covers... # 02 RU-BA Bashkortostan, Republic of # 90 RU-PER Perm Krai # 45 RU-KGN Kurgan Oblast # 56 RU-ORE Orenburg Oblast # 66 RU-SVE Sverdlovsk Oblast # 72 RU-TYU Tyumen Oblast # 74 RU-CHE Chelyabinsk Oblast # 86 RU-KHM Khanty-Mansi Autonomous Okrug - Yugra # 89 RU-YAN Yamalo-Nenets Autonomous Okrug # # Note: Effective 2005-12-01, (59) Perm Oblast and (81) Komi-Permyak # Autonomous Okrug merged to form (90, RU-PER) Perm Krai. # Milne says Yekaterinburg was 4:02:32.9; round to nearest. # Byalokoz 1919 says its provincial time was based on Perm, at 3:45:05. # Assume it switched on 1916-07-03, the time of the new standard. # The 1919 and 1930 transitions are from Shanks. Zone Asia/Yekaterinburg 4:02:33 - LMT 1916 Jul 3 3:45:05 - PMT 1919 Jul 15 4:00 4:00 - SVET 1930 Jun 21 # Sverdlovsk Time 5:00 Russia SVE%sT 1991 Mar 31 2:00s 4:00 Russia SVE%sT 1992 Jan 19 2:00s 5:00 Russia YEK%sT 2011 Mar 27 2:00s 6:00 - YEKT 2014 Oct 26 2:00s 5:00 - YEKT # From Tim Parenti (2014-07-03), per Oscar van Vlijmen (2001-08-25): # Asia/Omsk covers... # 04 RU-AL Altai Republic # 22 RU-ALT Altai Krai # 55 RU-OMS Omsk Oblast # Byalokoz 1919 says Omsk was 4:53:30. Zone Asia/Omsk 4:53:30 - LMT 1919 Nov 14 5:00 - OMST 1930 Jun 21 # Omsk Time 6:00 Russia OMS%sT 1991 Mar 31 2:00s 5:00 Russia OMS%sT 1992 Jan 19 2:00s 6:00 Russia OMS%sT 2011 Mar 27 2:00s 7:00 - OMST 2014 Oct 26 2:00s 6:00 - OMST # From Tim Parenti (2014-07-03): # Asia/Novosibirsk covers... # 54 RU-NVS Novosibirsk Oblast # 70 RU-TOM Tomsk Oblast # From Paul Eggert (2006-08-19): I'm guessing about Tomsk here; it's # not clear when it switched from +7 to +6. Zone Asia/Novosibirsk 5:31:40 - LMT 1919 Dec 14 6:00 6:00 - NOVT 1930 Jun 21 # Novosibirsk Time 7:00 Russia NOV%sT 1991 Mar 31 2:00s 6:00 Russia NOV%sT 1992 Jan 19 2:00s 7:00 Russia NOV%sT 1993 May 23 # say Shanks & P. 6:00 Russia NOV%sT 2011 Mar 27 2:00s 7:00 - NOVT 2014 Oct 26 2:00s 6:00 - NOVT # From Tim Parenti (2014-07-03): # Asia/Novokuznetsk covers... # 42 RU-KEM Kemerovo Oblast # From Alexander Krivenyshev (2009-10-13): # Kemerovo oblast' (Kemerovo region) in Russia will change current time zone on # March 28, 2010: # from current Russia Zone 6 - Krasnoyarsk Time Zone (KRA) UTC +0700 # to Russia Zone 5 - Novosibirsk Time Zone (NOV) UTC +0600 # # This is according to Government of Russia decree # 740, on September # 14, 2009 "Application in the territory of the Kemerovo region the Fifth # time zone." ("Russia Zone 5" or old "USSR Zone 5" is GMT +0600) # # Russian Government web site (Russian language) # http://www.government.ru/content/governmentactivity/rfgovernmentdecisions/archive/2009/09/14/991633.htm # or Russian-English translation by WorldTimeZone.com with reference # map to local region and new Russia Time Zone map after March 28, 2010 # http://www.worldtimezone.com/dst_news/dst_news_russia03.html # # Thus, when Russia will switch to DST on the night of March 28, 2010 # Kemerovo region (Kemerovo oblast') will not change the clock. # # As a result, Kemerovo oblast' will be in the same time zone as # Novosibirsk, Omsk, Tomsk, Barnaul and Altai Republic. # From Tim Parenti (2014-07-02), per Alexander Krivenyshev (2014-07-02): # The Kemerovo region will remain at UTC+7 through the 2014-10-26 change, thus # realigning itself with KRAT. Zone Asia/Novokuznetsk 5:48:48 - LMT 1924 May 1 6:00 - KRAT 1930 Jun 21 # Krasnoyarsk Time 7:00 Russia KRA%sT 1991 Mar 31 2:00s 6:00 Russia KRA%sT 1992 Jan 19 2:00s 7:00 Russia KRA%sT 2010 Mar 28 2:00s 6:00 Russia NOV%sT 2011 Mar 27 2:00s # Novosibirsk 7:00 - NOVT 2014 Oct 26 2:00s 7:00 - KRAT # Krasnoyarsk Time # From Tim Parenti (2014-07-03), per Oscar van Vlijmen (2001-08-25): # Asia/Krasnoyarsk covers... # 17 RU-TY Tuva Republic # 19 RU-KK Khakassia, Republic of # 24 RU-KYA Krasnoyarsk Krai # # Note: Effective 2007-01-01, (88) Evenk Autonomous Okrug and (84) Taymyr # Autonomous Okrug were merged into (24, RU-KYA) Krasnoyarsk Krai. # Byalokoz 1919 says Krasnoyarsk was 6:11:26. Zone Asia/Krasnoyarsk 6:11:26 - LMT 1920 Jan 6 6:00 - KRAT 1930 Jun 21 # Krasnoyarsk Time 7:00 Russia KRA%sT 1991 Mar 31 2:00s 6:00 Russia KRA%sT 1992 Jan 19 2:00s 7:00 Russia KRA%sT 2011 Mar 27 2:00s 8:00 - KRAT 2014 Oct 26 2:00s 7:00 - KRAT # From Tim Parenti (2014-07-03), per Oscar van Vlijmen (2001-08-25): # Asia/Irkutsk covers... # 03 RU-BU Buryatia, Republic of # 38 RU-IRK Irkutsk Oblast # # Note: Effective 2008-01-01, (85) Ust-Orda Buryat Autonomous Okrug was # merged into (38, RU-IRK) Irkutsk Oblast. # Milne 1899 says Irkutsk was 6:57:15. # Byalokoz 1919 says Irkutsk was 6:57:05. # Go with Byalokoz. Zone Asia/Irkutsk 6:57:05 - LMT 1880 6:57:05 - IMT 1920 Jan 25 # Irkutsk Mean Time 7:00 - IRKT 1930 Jun 21 # Irkutsk Time 8:00 Russia IRK%sT 1991 Mar 31 2:00s 7:00 Russia IRK%sT 1992 Jan 19 2:00s 8:00 Russia IRK%sT 2011 Mar 27 2:00s 9:00 - IRKT 2014 Oct 26 2:00s 8:00 - IRKT # From Tim Parenti (2014-07-06): # Asia/Chita covers... # 92 RU-ZAB Zabaykalsky Krai # # Note: Effective 2008-03-01, (75) Chita Oblast and (80) Agin-Buryat # Autonomous Okrug merged to form (92, RU-ZAB) Zabaykalsky Krai. Zone Asia/Chita 7:33:52 - LMT 1919 Dec 15 8:00 - YAKT 1930 Jun 21 # Yakutsk Time 9:00 Russia YAK%sT 1991 Mar 31 2:00s 8:00 Russia YAK%sT 1992 Jan 19 2:00s 9:00 Russia YAK%sT 2011 Mar 27 2:00s 10:00 - YAKT 2014 Oct 26 2:00s 8:00 - IRKT # From Tim Parenti (2014-07-03), per Oscar van Vlijmen (2009-11-29): # Asia/Yakutsk covers... # 28 RU-AMU Amur Oblast # # ...and parts of (14, RU-SA) Sakha (Yakutia) Republic: # 14-02 **** Aldansky District # 14-04 **** Amginsky District # 14-05 **** Anabarsky District # 14-06 **** Bulunsky District # 14-07 **** Verkhnevilyuysky District # 14-10 **** Vilyuysky District # 14-11 **** Gorny District # 14-12 **** Zhigansky District # 14-13 **** Kobyaysky District # 14-14 **** Lensky District # 14-15 **** Megino-Kangalassky District # 14-16 **** Mirninsky District # 14-18 **** Namsky District # 14-19 **** Neryungrinsky District # 14-21 **** Nyurbinsky District # 14-23 **** Olenyoksky District # 14-24 **** Olyokminsky District # 14-26 **** Suntarsky District # 14-27 **** Tattinsky District # 14-29 **** Ust-Aldansky District # 14-32 **** Khangalassky District # 14-33 **** Churapchinsky District # 14-34 **** Eveno-Bytantaysky National District # From Tim Parenti (2014-07-03): # Our commentary seems to have lost mention of (14-19) Neryungrinsky District. # Since the surrounding districts of Sakha are all YAKT, assume this is, too. # Also assume its history has been the same as the rest of Asia/Yakutsk. # Byalokoz 1919 says Yakutsk was 8:38:58. Zone Asia/Yakutsk 8:38:58 - LMT 1919 Dec 15 8:00 - YAKT 1930 Jun 21 # Yakutsk Time 9:00 Russia YAK%sT 1991 Mar 31 2:00s 8:00 Russia YAK%sT 1992 Jan 19 2:00s 9:00 Russia YAK%sT 2011 Mar 27 2:00s 10:00 - YAKT 2014 Oct 26 2:00s 9:00 - YAKT # From Tim Parenti (2014-07-03), per Oscar van Vlijmen (2009-11-29): # Asia/Vladivostok covers... # 25 RU-PRI Primorsky Krai # 27 RU-KHA Khabarovsk Krai # 79 RU-YEV Jewish Autonomous Oblast # # ...and parts of (14, RU-SA) Sakha (Yakutia) Republic: # 14-09 **** Verkhoyansky District # 14-31 **** Ust-Yansky District # Milne 1899 says Vladivostok was 8:47:33.5. # Byalokoz 1919 says Vladivostok was 8:47:31. # Go with Byalokoz. Zone Asia/Vladivostok 8:47:31 - LMT 1922 Nov 15 9:00 - VLAT 1930 Jun 21 # Vladivostok Time 10:00 Russia VLA%sT 1991 Mar 31 2:00s 9:00 Russia VLA%sT 1992 Jan 19 2:00s 10:00 Russia VLA%sT 2011 Mar 27 2:00s 11:00 - VLAT 2014 Oct 26 2:00s 10:00 - VLAT # From Tim Parenti (2014-07-03): # Asia/Khandyga covers parts of (14, RU-SA) Sakha (Yakutia) Republic: # 14-28 **** Tomponsky District # 14-30 **** Ust-Maysky District # From Arthur David Olson (2012-05-09): # Tomponskij and Ust'-Majskij switched from Vladivostok time to Yakutsk time # in 2011. # From Paul Eggert (2012-11-25): # Shanks and Pottenger (2003) has Khandyga on Yakutsk time. # Make a wild guess that it switched to Vladivostok time in 2004. # This transition is no doubt wrong, but we have no better info. Zone Asia/Khandyga 9:02:13 - LMT 1919 Dec 15 8:00 - YAKT 1930 Jun 21 # Yakutsk Time 9:00 Russia YAK%sT 1991 Mar 31 2:00s 8:00 Russia YAK%sT 1992 Jan 19 2:00s 9:00 Russia YAK%sT 2004 10:00 Russia VLA%sT 2011 Mar 27 2:00s 11:00 - VLAT 2011 Sep 13 0:00s # Decree 725? 10:00 - YAKT 2014 Oct 26 2:00s 9:00 - YAKT # From Tim Parenti (2014-07-03): # Asia/Sakhalin covers... # 65 RU-SAK Sakhalin Oblast # ...with the exception of: # 65-11 **** Severo-Kurilsky District (North Kuril Islands) # The Zone name should be Asia/Yuzhno-Sakhalinsk, but that's too long. Zone Asia/Sakhalin 9:30:48 - LMT 1905 Aug 23 9:00 - JCST 1937 Oct 1 9:00 - JST 1945 Aug 25 11:00 Russia SAK%sT 1991 Mar 31 2:00s # Sakhalin T 10:00 Russia SAK%sT 1992 Jan 19 2:00s 11:00 Russia SAK%sT 1997 Mar lastSun 2:00s 10:00 Russia SAK%sT 2011 Mar 27 2:00s 11:00 - SAKT 2014 Oct 26 2:00s 10:00 - SAKT # From Tim Parenti (2014-07-03), per Oscar van Vlijmen (2009-11-29): # Asia/Magadan covers... # 49 RU-MAG Magadan Oblast # From Tim Parenti (2014-07-06), per Alexander Krivenyshev (2014-07-02): # Magadan Oblast is moving from UTC+12 to UTC+10 on 2014-10-26; however, # several districts of Sakha Republic as well as Severo-Kurilsky District of # the Sakhalin Oblast (also known as the North Kuril Islands), represented # until now by Asia/Magadan, will instead move to UTC+11. These regions will # need their own zone. Zone Asia/Magadan 10:03:12 - LMT 1924 May 2 10:00 - MAGT 1930 Jun 21 # Magadan Time 11:00 Russia MAG%sT 1991 Mar 31 2:00s 10:00 Russia MAG%sT 1992 Jan 19 2:00s 11:00 Russia MAG%sT 2011 Mar 27 2:00s 12:00 - MAGT 2014 Oct 26 2:00s 10:00 - MAGT # From Tim Parenti (2014-07-06): # Asia/Srednekolymsk covers parts of (14, RU-SA) Sakha (Yakutia) Republic: # 14-01 **** Abyysky District # 14-03 **** Allaikhovsky District # 14-08 **** Verkhnekolymsky District # 14-17 **** Momsky District # 14-20 **** Nizhnekolymsky District # 14-25 **** Srednekolymsky District # # ...and parts of (65, RU-SAK) Sakhalin Oblast: # 65-11 **** Severo-Kurilsky District (North Kuril Islands) # From Tim Parenti (2014-07-02): # Oymyakonsky District of Sakha Republic (represented by Ust-Nera), along with # most of Sakhalin Oblast (represented by Sakhalin) will be moving to UTC+10 on # 2014-10-26 to stay aligned with VLAT/SAKT; however, Severo-Kurilsky District # of the Sakhalin Oblast (also known as the North Kuril Islands, represented by # Severo-Kurilsk) will remain on UTC+11. # From Tim Parenti (2014-07-06): # Assume North Kuril Islands have history like Magadan before 2011-03-27. # There is a decent chance this is wrong, in which case a new zone # Asia/Severo-Kurilsk would become necessary. # # Srednekolymsk and Zyryanka are the most populous places amongst these # districts, but have very similar populations. In fact, Wikipedia currently # lists them both as having 3528 people, exactly 1668 males and 1860 females # each! (Yikes!) # http://en.wikipedia.org/w/?title=Srednekolymsky_District&oldid=603435276 # http://en.wikipedia.org/w/?title=Verkhnekolymsky_District&oldid=594378493 # Assume this is a mistake, albeit an amusing one. # # Looking at censuses, the populations of the two municipalities seem to have # fluctuated recently. Zyryanka was more populous than Srednekolymsk in the # 1989 and 2002 censuses, but Srednekolymsk was more populous in the most # recent (2010) census, 3525 to 3170. (See pages 195 and 197 of # http://www.gks.ru/free_doc/new_site/perepis2010/croc/Documents/Vol1/pub-01-05.pdf # in Russian.) In addition, Srednekolymsk appears to be a much older # settlement and the population of Zyryanka seems to be declining. # Go with Srednekolymsk. # # Since Magadan Oblast moves to UTC+10 on 2014-10-26, we cannot keep using MAGT # as the abbreviation. Use SRET instead. Zone Asia/Srednekolymsk 10:14:52 - LMT 1924 May 2 10:00 - MAGT 1930 Jun 21 # Magadan Time 11:00 Russia MAG%sT 1991 Mar 31 2:00s 10:00 Russia MAG%sT 1992 Jan 19 2:00s 11:00 Russia MAG%sT 2011 Mar 27 2:00s 12:00 - MAGT 2014 Oct 26 2:00s 11:00 - SRET # Srednekolymsk Time # From Tim Parenti (2014-07-03): # Asia/Ust-Nera covers parts of (14, RU-SA) Sakha (Yakutia) Republic: # 14-22 **** Oymyakonsky District # From Arthur David Olson (2012-05-09): # Ojmyakonskij [and the Kuril Islands] switched from # Magadan time to Vladivostok time in 2011. # # From Tim Parenti (2014-07-06), per Alexander Krivenyshev (2014-07-02): # It's unlikely that any of the Kuril Islands were involved in such a switch, # as the South and Middle Kurils have been on UTC+11 (SAKT) with the rest of # Sakhalin Oblast since at least 2011-09, and the North Kurils have been on # UTC+12 since at least then, too. Zone Asia/Ust-Nera 9:32:54 - LMT 1919 Dec 15 8:00 - YAKT 1930 Jun 21 # Yakutsk Time 9:00 Russia YAKT 1981 Apr 1 11:00 Russia MAG%sT 1991 Mar 31 2:00s 10:00 Russia MAG%sT 1992 Jan 19 2:00s 11:00 Russia MAG%sT 2011 Mar 27 2:00s 12:00 - MAGT 2011 Sep 13 0:00s # Decree 725? 11:00 - VLAT 2014 Oct 26 2:00s 10:00 - VLAT # From Tim Parenti (2014-07-03), per Oscar van Vlijmen (2001-08-25): # Asia/Kamchatka covers... # 91 RU-KAM Kamchatka Krai # # Note: Effective 2007-07-01, (41) Kamchatka Oblast and (82) Koryak # Autonomous Okrug merged to form (91, RU-KAM) Kamchatka Krai. # The Zone name should be Asia/Petropavlovsk-Kamchatski or perhaps # Asia/Petropavlovsk-Kamchatsky, but these are too long. Zone Asia/Kamchatka 10:34:36 - LMT 1922 Nov 10 11:00 - PETT 1930 Jun 21 # P-K Time 12:00 Russia PET%sT 1991 Mar 31 2:00s 11:00 Russia PET%sT 1992 Jan 19 2:00s 12:00 Russia PET%sT 2010 Mar 28 2:00s 11:00 Russia PET%sT 2011 Mar 27 2:00s 12:00 - PETT # From Tim Parenti (2014-07-03): # Asia/Anadyr covers... # 87 RU-CHU Chukotka Autonomous Okrug Zone Asia/Anadyr 11:49:56 - LMT 1924 May 2 12:00 - ANAT 1930 Jun 21 # Anadyr Time 13:00 Russia ANA%sT 1982 Apr 1 0:00s 12:00 Russia ANA%sT 1991 Mar 31 2:00s 11:00 Russia ANA%sT 1992 Jan 19 2:00s 12:00 Russia ANA%sT 2010 Mar 28 2:00s 11:00 Russia ANA%sT 2011 Mar 27 2:00s 12:00 - ANAT # San Marino # See Europe/Rome. # Serbia # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Belgrade 1:22:00 - LMT 1884 1:00 - CET 1941 Apr 18 23:00 1:00 C-Eur CE%sT 1945 1:00 - CET 1945 May 8 2:00s 1:00 1:00 CEST 1945 Sep 16 2:00s # Metod Koželj reports that the legal date of # transition to EU rules was 1982-11-27, for all of Yugoslavia at the time. # Shanks & Pottenger don't give as much detail, so go with Koželj. 1:00 - CET 1982 Nov 27 1:00 EU CE%sT Link Europe/Belgrade Europe/Ljubljana # Slovenia Link Europe/Belgrade Europe/Podgorica # Montenegro Link Europe/Belgrade Europe/Sarajevo # Bosnia and Herzegovina Link Europe/Belgrade Europe/Skopje # Macedonia Link Europe/Belgrade Europe/Zagreb # Croatia # Slovakia Link Europe/Prague Europe/Bratislava # Slovenia # See Europe/Belgrade. # Spain # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S # For 1917-1919 Whitman gives Apr Sat>=1 - Oct Sat>=1; # go with Shanks & Pottenger. Rule Spain 1917 only - May 5 23:00s 1:00 S Rule Spain 1917 1919 - Oct 6 23:00s 0 - Rule Spain 1918 only - Apr 15 23:00s 1:00 S Rule Spain 1919 only - Apr 5 23:00s 1:00 S # Whitman gives 1921 Feb 28 - Oct 14; go with Shanks & Pottenger. Rule Spain 1924 only - Apr 16 23:00s 1:00 S # Whitman gives 1924 Oct 14; go with Shanks & Pottenger. Rule Spain 1924 only - Oct 4 23:00s 0 - Rule Spain 1926 only - Apr 17 23:00s 1:00 S # Whitman says no DST in 1929; go with Shanks & Pottenger. Rule Spain 1926 1929 - Oct Sat>=1 23:00s 0 - Rule Spain 1927 only - Apr 9 23:00s 1:00 S Rule Spain 1928 only - Apr 14 23:00s 1:00 S Rule Spain 1929 only - Apr 20 23:00s 1:00 S # Whitman gives 1937 Jun 16, 1938 Apr 16, 1940 Apr 13; # go with Shanks & Pottenger. Rule Spain 1937 only - May 22 23:00s 1:00 S Rule Spain 1937 1939 - Oct Sat>=1 23:00s 0 - Rule Spain 1938 only - Mar 22 23:00s 1:00 S Rule Spain 1939 only - Apr 15 23:00s 1:00 S Rule Spain 1940 only - Mar 16 23:00s 1:00 S # Whitman says no DST 1942-1945; go with Shanks & Pottenger. Rule Spain 1942 only - May 2 22:00s 2:00 M # Midsummer Rule Spain 1942 only - Sep 1 22:00s 1:00 S Rule Spain 1943 1946 - Apr Sat>=13 22:00s 2:00 M Rule Spain 1943 only - Oct 3 22:00s 1:00 S Rule Spain 1944 only - Oct 10 22:00s 1:00 S Rule Spain 1945 only - Sep 30 1:00 1:00 S Rule Spain 1946 only - Sep 30 0:00 0 - Rule Spain 1949 only - Apr 30 23:00 1:00 S Rule Spain 1949 only - Sep 30 1:00 0 - Rule Spain 1974 1975 - Apr Sat>=13 23:00 1:00 S Rule Spain 1974 1975 - Oct Sun>=1 1:00 0 - Rule Spain 1976 only - Mar 27 23:00 1:00 S Rule Spain 1976 1977 - Sep lastSun 1:00 0 - Rule Spain 1977 1978 - Apr 2 23:00 1:00 S Rule Spain 1978 only - Oct 1 1:00 0 - # The following rules are copied from Morocco from 1967 through 1978. Rule SpainAfrica 1967 only - Jun 3 12:00 1:00 S Rule SpainAfrica 1967 only - Oct 1 0:00 0 - Rule SpainAfrica 1974 only - Jun 24 0:00 1:00 S Rule SpainAfrica 1974 only - Sep 1 0:00 0 - Rule SpainAfrica 1976 1977 - May 1 0:00 1:00 S Rule SpainAfrica 1976 only - Aug 1 0:00 0 - Rule SpainAfrica 1977 only - Sep 28 0:00 0 - Rule SpainAfrica 1978 only - Jun 1 0:00 1:00 S Rule SpainAfrica 1978 only - Aug 4 0:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Madrid -0:14:44 - LMT 1901 Jan 1 0:00s 0:00 Spain WE%sT 1946 Sep 30 1:00 Spain CE%sT 1979 1:00 EU CE%sT Zone Africa/Ceuta -0:21:16 - LMT 1901 0:00 - WET 1918 May 6 23:00 0:00 1:00 WEST 1918 Oct 7 23:00 0:00 - WET 1924 0:00 Spain WE%sT 1929 0:00 SpainAfrica WE%sT 1984 Mar 16 1:00 - CET 1986 1:00 EU CE%sT Zone Atlantic/Canary -1:01:36 - LMT 1922 Mar # Las Palmas de Gran C. -1:00 - CANT 1946 Sep 30 1:00 # Canaries T 0:00 - WET 1980 Apr 6 0:00s 0:00 1:00 WEST 1980 Sep 28 0:00s 0:00 EU WE%sT # IATA SSIM (1996-09) says the Canaries switch at 2:00u, not 1:00u. # Ignore this for now, as the Canaries are part of the EU. # Sweden # From Ivan Nilsson (2001-04-13), superseding Shanks & Pottenger: # # The law "Svensk författningssamling 1878, no 14" about standard time in 1879: # From the beginning of 1879 (that is 01-01 00:00) the time for all # places in the country is "the mean solar time for the meridian at # three degrees, or twelve minutes of time, to the west of the # meridian of the Observatory of Stockholm". The law is dated 1878-05-31. # # The observatory at that time had the meridian 18 degrees 03' 30" # eastern longitude = 01:12:14 in time. Less 12 minutes gives the # national standard time as 01:00:14 ahead of GMT.... # # About the beginning of CET in Sweden. The lawtext ("Svensk # författningssamling 1899, no 44") states, that "from the beginning # of 1900... ... the same as the mean solar time for the meridian at # the distance of one hour of time from the meridian of the English # observatory at Greenwich, or at 12 minutes 14 seconds to the west # from the meridian of the Observatory of Stockholm". The law is dated # 1899-06-16. In short: At 1900-01-01 00:00:00 the new standard time # in Sweden is 01:00:00 ahead of GMT. # # 1916: The lawtext ("Svensk författningssamling 1916, no 124") states # that "1916-05-15 is considered to begin one hour earlier". It is # pretty obvious that at 05-14 23:00 the clocks are set to 05-15 00:00.... # Further the law says, that "1916-09-30 is considered to end one hour later". # # The laws regulating [DST] are available on the site of the Swedish # Parliament beginning with 1985 - the laws regulating 1980/1984 are # not available on the site (to my knowledge they are only available # in Swedish): (type # "sommartid" without the quotes in the field "Fritext" and then click # the Sök-button). # # (2001-05-13): # # I have now found a newspaper stating that at 1916-10-01 01:00 # summertime the church-clocks etc were set back one hour to show # 1916-10-01 00:00 standard time. The article also reports that some # people thought the switch to standard time would take place already # at 1916-10-01 00:00 summer time, but they had to wait for another # hour before the event took place. # # Source: The newspaper "Dagens Nyheter", 1916-10-01, page 7 upper left. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Stockholm 1:12:12 - LMT 1879 Jan 1 1:00:14 - SET 1900 Jan 1 # Swedish Time 1:00 - CET 1916 May 14 23:00 1:00 1:00 CEST 1916 Oct 1 1:00 1:00 - CET 1980 1:00 EU CE%sT # Switzerland # From Howse: # By the end of the 18th century clocks and watches became commonplace # and their performance improved enormously. Communities began to keep # mean time in preference to apparent time - Geneva from 1780 .... # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S # From Whitman (who writes "Midnight?"): # Rule Swiss 1940 only - Nov 2 0:00 1:00 S # Rule Swiss 1940 only - Dec 31 0:00 0 - # From Shanks & Pottenger: # Rule Swiss 1941 1942 - May Sun>=1 2:00 1:00 S # Rule Swiss 1941 1942 - Oct Sun>=1 0:00 0 - # From Alois Treindl (2008-12-17): # I have researched the DST usage in Switzerland during the 1940ies. # # As I wrote in an earlier message, I suspected the current tzdata values # to be wrong. This is now verified. # # I have found copies of the original ruling by the Swiss Federal # government, in 'Eidgenössische Gesetzessammlung 1941 and 1942' (Swiss # federal law collection)... # # DST began on Monday 5 May 1941, 1:00 am by shifting the clocks to 2:00 am # DST ended on Monday 6 Oct 1941, 2:00 am by shifting the clocks to 1:00 am. # # DST began on Monday, 4 May 1942 at 01:00 am # DST ended on Monday, 5 Oct 1942 at 02:00 am # # There was no DST in 1940, I have checked the law collection carefully. # It is also indicated by the fact that the 1942 entry in the law # collection points back to 1941 as a reference, but no reference to any # other years are made. # # Newspaper articles I have read in the archives on 6 May 1941 reported # about the introduction of DST (Sommerzeit in German) during the previous # night as an absolute novelty, because this was the first time that such # a thing had happened in Switzerland. # # I have also checked 1916, because one book source (Gabriel, Traité de # l'heure dans le monde) claims that Switzerland had DST in 1916. This is # false, no official document could be found. Probably Gabriel got misled # by references to Germany, which introduced DST in 1916 for the first time. # # The tzdata rules for Switzerland must be changed to: # Rule Swiss 1941 1942 - May Mon>=1 1:00 1:00 S # Rule Swiss 1941 1942 - Oct Mon>=1 2:00 0 - # # The 1940 rules must be deleted. # # One further detail for Switzerland, which is probably out of scope for # most users of tzdata: The [Europe/Zurich zone] ... # describes all of Switzerland correctly, with the exception of # the Canton de Genève (Geneva, Genf). Between 1848 and 1894 Geneva did not # follow Bern Mean Time but kept its own local mean time. # To represent this, an extra zone would be needed. # # From Alois Treindl (2013-09-11): # The Federal regulations say # http://www.admin.ch/opc/de/classified-compilation/20071096/index.html # ... the meridian for Bern mean time ... is 7 degrees 26' 22.50". # Expressed in time, it is 0h29m45.5s. # From Pierre-Yves Berger (2013-09-11): # the "Circulaire du conseil fédéral" (December 11 1893) # http://www.amtsdruckschriften.bar.admin.ch/viewOrigDoc.do?id=10071353 # clearly states that the [1894-06-01] change should be done at midnight # but if no one is present after 11 at night, could be postponed until one # hour before the beginning of service. # From Paul Eggert (2013-09-11): # Round BMT to the nearest even second, 0:29:46. # # We can find no reliable source for Shanks's assertion that all of Switzerland # except Geneva switched to Bern Mean Time at 00:00 on 1848-09-12. This book: # # Jakob Messerli. Gleichmässig, pünktlich, schnell. Zeiteinteilung und # Zeitgebrauch in der Schweiz im 19. Jahrhundert. Chronos, Zurich 1995, # ISBN 3-905311-68-2, OCLC 717570797. # # suggests that the transition was more gradual, and that the Swiss did not # agree about civil time during the transition. The timekeeping it gives the # most detail for is postal and telegraph time: here, federal legislation (the # "Bundesgesetz über die Erstellung von elektrischen Telegraphen") passed on # 1851-11-23, and an official implementation notice was published 1853-07-16 # (Bundesblatt 1853, Bd. II, S. 859). On p 72 Messerli writes that in # practice since July 1853 Bernese time was used in "all postal and telegraph # offices in Switzerland from Geneva to St. Gallen and Basel to Chiasso" # (Google translation). For now, model this transition as occurring on # 1853-07-16, though it probably occurred at some other date in Zurich, and # legal civil time probably changed at still some other transition date. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Swiss 1941 1942 - May Mon>=1 1:00 1:00 S Rule Swiss 1941 1942 - Oct Mon>=1 2:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Zurich 0:34:08 - LMT 1853 Jul 16 # See above comment. 0:29:46 - BMT 1894 Jun # Bern Mean Time 1:00 Swiss CE%sT 1981 1:00 EU CE%sT # Turkey # From Amar Devegowda (2007-01-03): # The time zone rules for Istanbul, Turkey have not been changed for years now. # ... The latest rules are available at: # http://www.timeanddate.com/worldclock/timezone.html?n=107 # From Steffen Thorsen (2007-01-03): # I have been able to find press records back to 1996 which all say that # DST started 01:00 local time and end at 02:00 local time. I am not sure # what happened before that. One example for each year from 1996 to 2001: # http://newspot.byegm.gov.tr/arsiv/1996/21/N4.htm # http://www.byegm.gov.tr/YAYINLARIMIZ/CHR/ING97/03/97X03X25.TXT # http://www.byegm.gov.tr/YAYINLARIMIZ/CHR/ING98/03/98X03X02.HTM # http://www.byegm.gov.tr/YAYINLARIMIZ/CHR/ING99/10/99X10X26.HTM#%2016 # http://www.byegm.gov.tr/YAYINLARIMIZ/CHR/ING2000/03/00X03X06.HTM#%2021 # http://www.byegm.gov.tr/YAYINLARIMIZ/CHR/ING2001/03/23x03x01.HTM#%2027 # From Paul Eggert (2007-01-03): # Prefer the above source to Shanks & Pottenger for time stamps after 1990. # From Steffen Thorsen (2007-03-09): # Starting 2007 though, it seems that they are adopting EU's 1:00 UTC # start/end time, according to the following page (2007-03-07): # http://www.ntvmsnbc.com/news/402029.asp # The official document is located here - it is in Turkish...: # http://rega.basbakanlik.gov.tr/eskiler/2007/03/20070307-7.htm # I was able to locate the following seemingly official document # (on a non-government server though) describing dates between 2002 and 2006: # http://www.alomaliye.com/bkk_2002_3769.htm # From Gökdeniz Karadağ (2011-03-10): # According to the articles linked below, Turkey will change into summer # time zone (GMT+3) on March 28, 2011 at 3:00 a.m. instead of March 27. # This change is due to a nationwide exam on 27th. # http://www.worldbulletin.net/?aType=haber&ArticleID=70872 # Turkish: # http://www.hurriyet.com.tr/ekonomi/17230464.asp?gid=373 # From Faruk Pasin (2014-02-14): # The DST for Turkey has been changed for this year because of the # Turkish Local election.... # http://www.sabah.com.tr/Ekonomi/2014/02/12/yaz-saatinde-onemli-degisiklik # ... so Turkey will move clocks forward one hour on March 31 at 3:00 a.m. # From Randal L. Schwartz (2014-04-15): # Having landed on a flight from the states to Istanbul (via AMS) on March 31, # I can tell you that NOBODY (even the airlines) respected this timezone DST # change delay. Maybe the word just didn't get out in time. # From Paul Eggert (2014-06-15): # The press reported massive confusion, as election officials obeyed the rule # change but cell phones (and airline baggage systems) did not. See: # Kostidis M. Eventful elections in Turkey. Balkan News Agency # http://www.balkaneu.com/eventful-elections-turkey/ 2014-03-30. # I guess the best we can do is document the official time. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Turkey 1916 only - May 1 0:00 1:00 S Rule Turkey 1916 only - Oct 1 0:00 0 - Rule Turkey 1920 only - Mar 28 0:00 1:00 S Rule Turkey 1920 only - Oct 25 0:00 0 - Rule Turkey 1921 only - Apr 3 0:00 1:00 S Rule Turkey 1921 only - Oct 3 0:00 0 - Rule Turkey 1922 only - Mar 26 0:00 1:00 S Rule Turkey 1922 only - Oct 8 0:00 0 - # Whitman gives 1923 Apr 28 - Sep 16 and no DST in 1924-1925; # go with Shanks & Pottenger. Rule Turkey 1924 only - May 13 0:00 1:00 S Rule Turkey 1924 1925 - Oct 1 0:00 0 - Rule Turkey 1925 only - May 1 0:00 1:00 S Rule Turkey 1940 only - Jun 30 0:00 1:00 S Rule Turkey 1940 only - Oct 5 0:00 0 - Rule Turkey 1940 only - Dec 1 0:00 1:00 S Rule Turkey 1941 only - Sep 21 0:00 0 - Rule Turkey 1942 only - Apr 1 0:00 1:00 S # Whitman omits the next two transition and gives 1945 Oct 1; # go with Shanks & Pottenger. Rule Turkey 1942 only - Nov 1 0:00 0 - Rule Turkey 1945 only - Apr 2 0:00 1:00 S Rule Turkey 1945 only - Oct 8 0:00 0 - Rule Turkey 1946 only - Jun 1 0:00 1:00 S Rule Turkey 1946 only - Oct 1 0:00 0 - Rule Turkey 1947 1948 - Apr Sun>=16 0:00 1:00 S Rule Turkey 1947 1950 - Oct Sun>=2 0:00 0 - Rule Turkey 1949 only - Apr 10 0:00 1:00 S Rule Turkey 1950 only - Apr 19 0:00 1:00 S Rule Turkey 1951 only - Apr 22 0:00 1:00 S Rule Turkey 1951 only - Oct 8 0:00 0 - Rule Turkey 1962 only - Jul 15 0:00 1:00 S Rule Turkey 1962 only - Oct 8 0:00 0 - Rule Turkey 1964 only - May 15 0:00 1:00 S Rule Turkey 1964 only - Oct 1 0:00 0 - Rule Turkey 1970 1972 - May Sun>=2 0:00 1:00 S Rule Turkey 1970 1972 - Oct Sun>=2 0:00 0 - Rule Turkey 1973 only - Jun 3 1:00 1:00 S Rule Turkey 1973 only - Nov 4 3:00 0 - Rule Turkey 1974 only - Mar 31 2:00 1:00 S Rule Turkey 1974 only - Nov 3 5:00 0 - Rule Turkey 1975 only - Mar 30 0:00 1:00 S Rule Turkey 1975 1976 - Oct lastSun 0:00 0 - Rule Turkey 1976 only - Jun 1 0:00 1:00 S Rule Turkey 1977 1978 - Apr Sun>=1 0:00 1:00 S Rule Turkey 1977 only - Oct 16 0:00 0 - Rule Turkey 1979 1980 - Apr Sun>=1 3:00 1:00 S Rule Turkey 1979 1982 - Oct Mon>=11 0:00 0 - Rule Turkey 1981 1982 - Mar lastSun 3:00 1:00 S Rule Turkey 1983 only - Jul 31 0:00 1:00 S Rule Turkey 1983 only - Oct 2 0:00 0 - Rule Turkey 1985 only - Apr 20 0:00 1:00 S Rule Turkey 1985 only - Sep 28 0:00 0 - Rule Turkey 1986 1990 - Mar lastSun 2:00s 1:00 S Rule Turkey 1986 1990 - Sep lastSun 2:00s 0 - Rule Turkey 1991 2006 - Mar lastSun 1:00s 1:00 S Rule Turkey 1991 1995 - Sep lastSun 1:00s 0 - Rule Turkey 1996 2006 - Oct lastSun 1:00s 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Istanbul 1:55:52 - LMT 1880 1:56:56 - IMT 1910 Oct # Istanbul Mean Time? 2:00 Turkey EE%sT 1978 Oct 15 3:00 Turkey TR%sT 1985 Apr 20 # Turkey Time 2:00 Turkey EE%sT 2007 2:00 EU EE%sT 2011 Mar 27 1:00u 2:00 - EET 2011 Mar 28 1:00u 2:00 EU EE%sT 2014 Mar 30 1:00u 2:00 - EET 2014 Mar 31 1:00u 2:00 EU EE%sT Link Europe/Istanbul Asia/Istanbul # Istanbul is in both continents. # Ukraine # # From Igor Karpov, who works for the Ukrainian Ministry of Justice, # via Garrett Wollman (2003-01-27): # BTW, I've found the official document on this matter. It's government # regulations number 509, May 13, 1996. In my poor translation it says: # "Time in Ukraine is set to second timezone (Kiev time). Each last Sunday # of March at 3am the time is changing to 4am and each last Sunday of # October the time at 4am is changing to 3am" # From Alexander Krivenyshev (2011-09-20): # On September 20, 2011 the deputies of the Verkhovna Rada agreed to # abolish the transfer clock to winter time. # # Bill number 8330 of MP from the Party of Regions Oleg Nadoshi got # approval from 266 deputies. # # Ukraine abolishes transfer back to the winter time (in Russian) # http://news.mail.ru/politics/6861560/ # # The Ukrainians will no longer change the clock (in Russian) # http://www.segodnya.ua/news/14290482.html # # Deputies cancelled the winter time (in Russian) # http://www.pravda.com.ua/rus/news/2011/09/20/6600616/ # # From Philip Pizzey (2011-10-18): # Today my Ukrainian colleagues have informed me that the # Ukrainian parliament have decided that they will go to winter # time this year after all. # # From Udo Schwedt (2011-10-18): # As far as I understand, the recent change to the Ukrainian time zone # (Europe/Kiev) to introduce permanent daylight saving time (similar # to Russia) was reverted today: # http://portal.rada.gov.ua/rada/control/en/publish/article/info_left?art_id=287324&cat_id=105995 # # Also reported by Alexander Bokovoy (2011-10-18) who also noted: # The law documents themselves are at # http://w1.c1.rada.gov.ua/pls/zweb_n/webproc4_1?id=&pf3511=41484 # From Vladimir in Moscow via Alois Treindl re Kiev time 1991/2 (2014-02-28): # First in Ukraine they changed Time zone from UTC+3 to UTC+2 with DST: # 03 25 1990 02:00 -03.00 1 Time Zone 3 with DST # 07 01 1990 02:00 -02.00 1 Time Zone 2 with DST # * Ukrainian Government's Resolution of 18.06.1990, No. 134. # http://search.ligazakon.ua/l_doc2.nsf/link1/T001500.html # # They did not end DST in September, 1990 (according to the law, # "summer time" was still in action): # 09 30 1990 03:00 -02.00 1 Time Zone 2 with DST # * Ukrainian Government's Resolution of 21.09.1990, No. 272. # http://search.ligazakon.ua/l_doc2.nsf/link1/KP900272.html # # Again no change in March, 1991 ("summer time" in action): # 03 31 1991 02:00 -02.00 1 Time Zone 2 with DST # # DST ended in September 1991 ("summer time" ended): # 09 29 1991 03:00 -02.00 0 Time Zone 2, no DST # * Ukrainian Government's Resolution of 25.09.1991, No. 225. # http://www.uazakon.com/documents/date_21/pg_iwgdoc.htm # This is an answer. # # Since 1992 they had normal DST procedure: # 03 29 1992 02:00 -02.00 1 DST started # 09 27 1992 03:00 -02.00 0 DST ended # * Ukrainian Government's Resolution of 20.03.1992, No. 139. # http://www.uazakon.com/documents/date_8u/pg_grcasa.htm # Zone NAME GMTOFF RULES FORMAT [UNTIL] # Most of Ukraine since 1970 has been like Kiev. # "Kyiv" is the transliteration of the Ukrainian name, but # "Kiev" is more common in English. Zone Europe/Kiev 2:02:04 - LMT 1880 2:02:04 - KMT 1924 May 2 # Kiev Mean Time 2:00 - EET 1930 Jun 21 3:00 - MSK 1941 Sep 20 1:00 C-Eur CE%sT 1943 Nov 6 3:00 Russia MSK/MSD 1990 Jul 1 2:00 2:00 1:00 EEST 1991 Sep 29 3:00 2:00 E-Eur EE%sT 1995 2:00 EU EE%sT # Ruthenia used CET 1990/1991. # "Uzhhorod" is the transliteration of the Rusyn/Ukrainian pronunciation, but # "Uzhgorod" is more common in English. Zone Europe/Uzhgorod 1:29:12 - LMT 1890 Oct 1:00 - CET 1940 1:00 C-Eur CE%sT 1944 Oct 1:00 1:00 CEST 1944 Oct 26 1:00 - CET 1945 Jun 29 3:00 Russia MSK/MSD 1990 3:00 - MSK 1990 Jul 1 2:00 1:00 - CET 1991 Mar 31 3:00 2:00 - EET 1992 2:00 E-Eur EE%sT 1995 2:00 EU EE%sT # Zaporozh'ye and eastern Lugansk oblasts observed DST 1990/1991. # "Zaporizhia" is the transliteration of the Ukrainian name, but # "Zaporozh'ye" is more common in English. Use the common English # spelling, except omit the apostrophe as it is not allowed in # portable Posix file names. Zone Europe/Zaporozhye 2:20:40 - LMT 1880 2:20 - CUT 1924 May 2 # Central Ukraine T 2:00 - EET 1930 Jun 21 3:00 - MSK 1941 Aug 25 1:00 C-Eur CE%sT 1943 Oct 25 3:00 Russia MSK/MSD 1991 Mar 31 2:00 2:00 E-Eur EE%sT 1995 2:00 EU EE%sT # Vatican City # See Europe/Rome. ############################################################################### # One source shows that Bulgaria, Cyprus, Finland, and Greece observe DST from # the last Sunday in March to the last Sunday in September in 1986. # The source shows Romania changing a day later than everybody else. # # According to Bernard Sieloff's source, Poland is in the MET time zone but # uses the WE DST rules. The Western USSR uses EET+1 and ME DST rules. # Bernard Sieloff's source claims Romania switches on the same day, but at # 00:00 standard time (i.e., 01:00 DST). It also claims that Turkey # switches on the same day, but switches on at 01:00 standard time # and off at 00:00 standard time (i.e., 01:00 DST) # ... # Date: Wed, 28 Jan 87 16:56:27 -0100 # From: Tom Hofmann # ... # # ...the European time rules are...standardized since 1981, when # most European countries started DST. Before that year, only # a few countries (UK, France, Italy) had DST, each according # to own national rules. In 1981, however, DST started on # 'Apr firstSun', and not on 'Mar lastSun' as in the following # years... # But also since 1981 there are some more national exceptions # than listed in 'europe': Switzerland, for example, joined DST # one year later, Denmark ended DST on 'Oct 1' instead of 'Sep # lastSun' in 1981 - I don't know how they handle now. # # Finally, DST ist always from 'Apr 1' to 'Oct 1' in the # Soviet Union (as far as I know). # # Tom Hofmann, Scientific Computer Center, CIBA-GEIGY AG, # 4002 Basle, Switzerland # ... # ... # Date: Wed, 4 Feb 87 22:35:22 +0100 # From: Dik T. Winter # ... # # The information from Tom Hofmann is (as far as I know) not entirely correct. # After a request from chongo at amdahl I tried to retrieve all information # about DST in Europe. I was able to find all from about 1969. # # ...standardization on DST in Europe started in about 1977 with switches on # first Sunday in April and last Sunday in September... # In 1981 UK joined Europe insofar that # the starting day for both shifted to last Sunday in March. And from 1982 # the whole of Europe used DST, with switch dates April 1 and October 1 in # the Sov[i]et Union. In 1985 the SU reverted to standard Europe[a]n switch # dates... # # It should also be remembered that time-zones are not constants; e.g. # Portugal switched in 1976 from MET (or CET) to WET with DST... # Note also that though there were rules for switch dates not # all countries abided to these dates, and many individual deviations # occurred, though not since 1982 I believe. Another note: it is always # assumed that DST is 1 hour ahead of normal time, this need not be the # case; at least in the Netherlands there have been times when DST was 2 hours # in advance of normal time. # # ... # dik t. winter, cwi, amsterdam, nederland # ... # From Bob Devine (1988-01-28): # ... # Greece: Last Sunday in April to last Sunday in September (iffy on dates). # Since 1978. Change at midnight. # ... # Monaco: has same DST as France. # ... Index: projects/cxl_iscsi/contrib/tzdata/leap-seconds.list =================================================================== --- projects/cxl_iscsi/contrib/tzdata/leap-seconds.list (revision 279716) +++ projects/cxl_iscsi/contrib/tzdata/leap-seconds.list (revision 279717) @@ -1,239 +1,249 @@ # # In the following text, the symbol '#' introduces # a comment, which continues from that symbol until # the end of the line. A plain comment line has a # whitespace character following the comment indicator. # There are also special comment lines defined below. # A special comment will always have a non-whitespace # character in column 2. # # A blank line should be ignored. # # The following table shows the corrections that must # be applied to compute International Atomic Time (TAI) # from the Coordinated Universal Time (UTC) values that # are transmitted by almost all time services. # # The first column shows an epoch as a number of seconds # since 1 January 1900, 00:00:00 (1900.0 is also used to # indicate the same epoch.) Both of these time stamp formats # ignore the complexities of the time scales that were # used before the current definition of UTC at the start # of 1972. (See note 3 below.) # The second column shows the number of seconds that # must be added to UTC to compute TAI for any timestamp # at or after that epoch. The value on each line is # valid from the indicated initial instant until the # epoch given on the next one or indefinitely into the # future if there is no next line. # (The comment on each line shows the representation of # the corresponding initial epoch in the usual # day-month-year format. The epoch always begins at # 00:00:00 UTC on the indicated day. See Note 5 below.) # # Important notes: # # 1. Coordinated Universal Time (UTC) is often referred to # as Greenwich Mean Time (GMT). The GMT time scale is no # longer used, and the use of GMT to designate UTC is # discouraged. # # 2. The UTC time scale is realized by many national # laboratories and timing centers. Each laboratory # identifies its realization with its name: Thus # UTC(NIST), UTC(USNO), etc. The differences among # these different realizations are typically on the # order of a few nanoseconds (i.e., 0.000 000 00x s) # and can be ignored for many purposes. These differences # are tabulated in Circular T, which is published monthly # by the International Bureau of Weights and Measures -# (BIPM). See www.bipm.fr for more information. +# (BIPM). See www.bipm.org for more information. # # 3. The current definition of the relationship between UTC # and TAI dates from 1 January 1972. A number of different # time scales were in use before that epoch, and it can be # quite difficult to compute precise timestamps and time # intervals in those "prehistoric" days. For more information, # consult: # # The Explanatory Supplement to the Astronomical # Ephemeris. # or # Terry Quinn, "The BIPM and the Accurate Measurement # of Time," Proc. of the IEEE, Vol. 79, pp. 894-905, # July, 1991. # # 4. The decision to insert a leap second into UTC is currently # the responsibility of the International Earth Rotation and # Reference Systems Service. (The name was changed from the # International Earth Rotation Service, but the acronym IERS # is still used.) # # Leap seconds are announced by the IERS in its Bulletin C. # # See www.iers.org for more details. # # Every national laboratory and timing center uses the # data from the BIPM and the IERS to construct UTC(lab), # their local realization of UTC. # # Although the definition also includes the possibility # of dropping seconds ("negative" leap seconds), this has # never been done and is unlikely to be necessary in the # foreseeable future. # # 5. If your system keeps time as the number of seconds since # some epoch (e.g., NTP timestamps), then the algorithm for # assigning a UTC time stamp to an event that happens during a positive # leap second is not well defined. The official name of that leap # second is 23:59:60, but there is no way of representing that time # in these systems. # Many systems of this type effectively stop the system clock for # one second during the leap second and use a time that is equivalent # to 23:59:59 UTC twice. For these systems, the corresponding TAI # timestamp would be obtained by advancing to the next entry in the # following table when the time equivalent to 23:59:59 UTC # is used for the second time. Thus the leap second which # occurred on 30 June 1972 at 23:59:59 UTC would have TAI # timestamps computed as follows: # # ... # 30 June 1972 23:59:59 (2287785599, first time): TAI= UTC + 10 seconds # 30 June 1972 23:59:60 (2287785599,second time): TAI= UTC + 11 seconds # 1 July 1972 00:00:00 (2287785600) TAI= UTC + 11 seconds # ... # # If your system realizes the leap second by repeating 00:00:00 UTC twice # (this is possible but not usual), then the advance to the next entry # in the table must occur the second time that a time equivalent to # 00:00:00 UTC is used. Thus, using the same example as above: # # ... # 30 June 1972 23:59:59 (2287785599): TAI= UTC + 10 seconds # 30 June 1972 23:59:60 (2287785600, first time): TAI= UTC + 10 seconds # 1 July 1972 00:00:00 (2287785600,second time): TAI= UTC + 11 seconds # ... # # in both cases the use of timestamps based on TAI produces a smooth # time scale with no discontinuity in the time interval. However, # although the long-term behavior of the time scale is correct in both # methods, the second method is technically not correct because it adds # the extra second to the wrong day. # # This complexity would not be needed for negative leap seconds (if they # are ever used). The UTC time would skip 23:59:59 and advance from # 23:59:58 to 00:00:00 in that case. The TAI offset would decrease by # 1 second at the same instant. This is a much easier situation to deal # with, since the difficulty of unambiguously representing the epoch # during the leap second does not arise. # +# Some systems implement leap seconds by amortizing the leap second +# over the last few minutes of the day. The frequency of the local +# clock is decreased (or increased) to realize the positive (or +# negative) leap second. This method removes the time step described +# above. Although the long-term behavior of the time scale is correct +# in this case, this method introduces an error during the adjustment +# period both in time and in frequency with respect to the official +# defintion of UTC. +# # Questions or comments to: # Judah Levine # Time and Frequency Division # NIST # Boulder, Colorado # Judah.Levine@nist.gov # -# Last Update of leap second values: 11 January 2012 +# Last Update of leap second values: 5 January 2015 # # The following line shows this last update date in NTP timestamp # format. This is the date on which the most recent change to # the leap second data was added to the file. This line can # be identified by the unique pair of characters in the first two # columns as shown below. # -#$ 3535228800 +#$ 3629404800 # # The NTP timestamps are in units of seconds since the NTP epoch, # which is 1 January 1900, 00:00:00. The Modified Julian Day number # corresponding to the NTP time stamp, X, can be computed as # # X/86400 + 15020 # # where the first term converts seconds to days and the second # term adds the MJD corresponding to the time origin defined above. # The integer portion of the result is the integer MJD for that # day, and any remainder is the time of day, expressed as the # fraction of the day since 0 hours UTC. The conversion from day # fraction to seconds or to hours, minutes, and seconds may involve # rounding or truncation, depending on the method used in the # computation. # # The data in this file will be updated periodically as new leap # seconds are announced. In addition to being entered on the line # above, the update time (in NTP format) will be added to the basic # file name leap-seconds to form the name leap-seconds.. # In addition, the generic name leap-seconds.list will always point to # the most recent version of the file. # # This update procedure will be performed only when a new leap second # is announced. # # The following entry specifies the expiration date of the data # in this file in units of seconds since the origin at the instant # 1 January 1900, 00:00:00. This expiration date will be changed # at least twice per year whether or not a new leap second is # announced. These semi-annual changes will be made no later # than 1 June and 1 December of each year to indicate what # action (if any) is to be taken on 30 June and 31 December, # respectively. (These are the customary effective dates for new # leap seconds.) This expiration date will be identified by a # unique pair of characters in columns 1 and 2 as shown below. # In the unlikely event that a leap second is announced with an # effective date other than 30 June or 31 December, then this # file will be edited to include that leap second as soon as it is # announced or at least one month before the effective date # (whichever is later). # If an announcement by the IERS specifies that no leap second is # scheduled, then only the expiration date of the file will # be advanced to show that the information in the file is still # current -- the update time stamp, the data and the name of the file # will not change. # -# Updated through IERS Bulletin C48 -# File expires on: 28 June 2015 +# Updated through IERS Bulletin C49 +# File expires on: 28 December 2015 # -#@ 3644438400 +#@ 3660249600 # 2272060800 10 # 1 Jan 1972 2287785600 11 # 1 Jul 1972 2303683200 12 # 1 Jan 1973 2335219200 13 # 1 Jan 1974 2366755200 14 # 1 Jan 1975 2398291200 15 # 1 Jan 1976 2429913600 16 # 1 Jan 1977 2461449600 17 # 1 Jan 1978 2492985600 18 # 1 Jan 1979 2524521600 19 # 1 Jan 1980 2571782400 20 # 1 Jul 1981 2603318400 21 # 1 Jul 1982 2634854400 22 # 1 Jul 1983 2698012800 23 # 1 Jul 1985 2776982400 24 # 1 Jan 1988 2840140800 25 # 1 Jan 1990 2871676800 26 # 1 Jan 1991 2918937600 27 # 1 Jul 1992 2950473600 28 # 1 Jul 1993 2982009600 29 # 1 Jul 1994 3029443200 30 # 1 Jan 1996 3076704000 31 # 1 Jul 1997 3124137600 32 # 1 Jan 1999 3345062400 33 # 1 Jan 2006 3439756800 34 # 1 Jan 2009 3550089600 35 # 1 Jul 2012 +3644697600 36 # 1 Jul 2015 # # the following special comment contains the # hash value of the data in this file computed # use the secure hash algorithm as specified # by FIPS 180-1. See the files in ~/pub/sha for # the details of how this hash value is # computed. Note that the hash computation # ignores comments and whitespace characters # in data lines. It includes the NTP values # of both the last modification time and the # expiration time of the file, but not the # white space on those lines. # the hash line is also ignored in the # computation. # -#h a4862ccd c6f43c6 964f3604 85944a26 b5cfad4e +#h 45e70fa7 a9df2033 f4a49ab0 ec648273 7b6c22c Index: projects/cxl_iscsi/contrib/tzdata/northamerica =================================================================== --- projects/cxl_iscsi/contrib/tzdata/northamerica (revision 279716) +++ projects/cxl_iscsi/contrib/tzdata/northamerica (revision 279717) @@ -1,3154 +1,3173 @@ # This file is in the public domain, so clarified as of # 2009-05-17 by Arthur David Olson. # also includes Central America and the Caribbean # This file is by no means authoritative; if you think you know better, # go ahead and edit the file (and please send any changes to # tz@iana.org for general use in the future). For more, please see # the file CONTRIBUTING in the tz distribution. # From Paul Eggert (1999-03-22): # A reliable and entertaining source about time zones is # Derek Howse, Greenwich time and longitude, Philip Wilson Publishers (1997). ############################################################################### # United States # From Paul Eggert (1999-03-31): # Howse writes (pp 121-125) that time zones were invented by # Professor Charles Ferdinand Dowd (1825-1904), # Principal of Temple Grove Ladies' Seminary (Saratoga Springs, NY). # His pamphlet "A System of National Time for Railroads" (1870) # was the result of his proposals at the Convention of Railroad Trunk Lines # in New York City (1869-10). His 1870 proposal was based on Washington, DC, # but in 1872-05 he moved the proposed origin to Greenwich. # His proposal was adopted by the railroads on 1883-11-18 at 12:00, # and the most of the country soon followed suit. # From Paul Eggert (2005-04-16): # That 1883 transition occurred at 12:00 new time, not at 12:00 old time. # See p 46 of David Prerau, Seize the daylight, Thunder's Mouth Press (2005). # From Paul Eggert (2006-03-22): # A good source for time zone historical data in the US is # Thomas G. Shanks, The American Atlas (5th edition), # San Diego: ACS Publications, Inc. (1991). # Make sure you have the errata sheet; the book is somewhat useless without it. # It is the source for most of the pre-1991 US entries below. # From Paul Eggert (2001-03-06): # Daylight Saving Time was first suggested as a joke by Benjamin Franklin # in his whimsical essay "An Economical Project for Diminishing the Cost # of Light" published in the Journal de Paris (1784-04-26). # Not everyone is happy with the results: # # I don't really care how time is reckoned so long as there is some # agreement about it, but I object to being told that I am saving # daylight when my reason tells me that I am doing nothing of the kind. # I even object to the implication that I am wasting something # valuable if I stay in bed after the sun has risen. As an admirer # of moonlight I resent the bossy insistence of those who want to # reduce my time for enjoying it. At the back of the Daylight Saving # scheme I detect the bony, blue-fingered hand of Puritanism, eager # to push people into bed earlier, and get them up earlier, to make # them healthy, wealthy and wise in spite of themselves. # # -- Robertson Davies, The diary of Samuel Marchbanks, # Clarke, Irwin (1947), XIX, Sunday # # For more about the first ten years of DST in the United States, see # Robert Garland, Ten years of daylight saving from the Pittsburgh standpoint # (Carnegie Library of Pittsburgh, 1927). # http://www.clpgh.org/exhibit/dst.html # # Shanks says that DST was called "War Time" in the US in 1918 and 1919. # However, DST was imposed by the Standard Time Act of 1918, which # was the first nationwide legal time standard, and apparently # time was just called "Standard Time" or "Daylight Saving Time". # From Arthur David Olson: # US Daylight Saving Time ended on the last Sunday of *October* in 1974. # See, for example, the front page of the Saturday, 1974-10-26 # and Sunday, 1974-10-27 editions of the Washington Post. # From Arthur David Olson: # Before the Uniform Time Act of 1966 took effect in 1967, observance of # Daylight Saving Time in the US was by local option, except during wartime. # From Arthur David Olson (2000-09-25): # Last night I heard part of a rebroadcast of a 1945 Arch Oboler radio drama. # In the introduction, Oboler spoke of "Eastern Peace Time." # An AltaVista search turned up: # http://rowayton.org/rhs/hstaug45.html # "When the time is announced over the radio now, it is 'Eastern Peace # Time' instead of the old familiar 'Eastern War Time.' Peace is wonderful." # (August 1945) by way of confirmation. # From Joseph Gallant citing # George H. Douglas, _The Early Days of Radio Broadcasting_ (1987): # At 7 P.M. (Eastern War Time) [on 1945-08-14], the networks were set # to switch to London for Attlee's address, but the American people # never got to hear his speech live. According to one press account, # CBS' Bob Trout was first to announce the word of Japan's surrender, # but a few seconds later, NBC, ABC and Mutual also flashed the word # of surrender, all of whom interrupting the bells of Big Ben in # London which were to precede Mr. Attlee's speech. # From Paul Eggert (2003-02-09): It was Robert St John, not Bob Trout. From # Myrna Oliver's obituary of St John on page B16 of today's Los Angeles Times: # # ... a war-weary U.S. clung to radios, awaiting word of Japan's surrender. # Any announcement from Asia would reach St. John's New York newsroom on a # wire service teletype machine, which had prescribed signals for major news. # Associated Press, for example, would ring five bells before spewing out # typed copy of an important story, and 10 bells for news "of transcendental # importance." # # On Aug. 14, stalling while talking steadily into the NBC networks' open # microphone, St. John heard five bells and waited only to hear a sixth bell, # before announcing confidently: "Ladies and gentlemen, World War II is over. # The Japanese have agreed to our surrender terms." # # He had scored a 20-second scoop on other broadcasters. # From Arthur David Olson (2005-08-22): # Paul has been careful to use the "US" rules only in those locations # that are part of the United States; this reflects the real scope of # U.S. government action. So even though the "US" rules have changed # in the latest release, other countries won't be affected. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule US 1918 1919 - Mar lastSun 2:00 1:00 D Rule US 1918 1919 - Oct lastSun 2:00 0 S Rule US 1942 only - Feb 9 2:00 1:00 W # War Rule US 1945 only - Aug 14 23:00u 1:00 P # Peace -Rule US 1945 only - Sep 30 2:00 0 S +Rule US 1945 only - Sep lastSun 2:00 0 S Rule US 1967 2006 - Oct lastSun 2:00 0 S Rule US 1967 1973 - Apr lastSun 2:00 1:00 D Rule US 1974 only - Jan 6 2:00 1:00 D Rule US 1975 only - Feb 23 2:00 1:00 D Rule US 1976 1986 - Apr lastSun 2:00 1:00 D Rule US 1987 2006 - Apr Sun>=1 2:00 1:00 D Rule US 2007 max - Mar Sun>=8 2:00 1:00 D Rule US 2007 max - Nov Sun>=1 2:00 0 S # From Arthur David Olson, 2005-12-19 # We generate the files specified below to guard against old files with # obsolete information being left in the time zone binary directory. # We limit the list to names that have appeared in previous versions of # this time zone package. # We do these as separate Zones rather than as Links to avoid problems if # a particular place changes whether it observes DST. # We put these specifications here in the northamerica file both to # increase the chances that they'll actually get compiled and to # avoid the need to duplicate the US rules in another file. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone EST -5:00 - EST Zone MST -7:00 - MST Zone HST -10:00 - HST Zone EST5EDT -5:00 US E%sT Zone CST6CDT -6:00 US C%sT Zone MST7MDT -7:00 US M%sT Zone PST8PDT -8:00 US P%sT # From Bob Devine (1988-01-28): # ...Alaska (and Hawaii) had the timezone names changed in 1967. # old new # Pacific Standard Time(PST) -same- # Yukon Standard Time(YST) -same- # Central Alaska S.T. (CAT) Alaska-Hawaii St[an]dard Time (AHST) # Nome Standard Time (NT) Bering Standard Time (BST) # # ...Alaska's timezone lines were redrawn in 1983 to give only 2 tz. # The YST zone now covers nearly all of the state, AHST just part # of the Aleutian islands. No DST. # From Paul Eggert (1995-12-19): # The tables below use 'NST', not 'NT', for Nome Standard Time. # I invented 'CAWT' for Central Alaska War Time. # From U. S. Naval Observatory (1989-01-19): # USA EASTERN 5 H BEHIND UTC NEW YORK, WASHINGTON # USA EASTERN 4 H BEHIND UTC APR 3 - OCT 30 # USA CENTRAL 6 H BEHIND UTC CHICAGO, HOUSTON # USA CENTRAL 5 H BEHIND UTC APR 3 - OCT 30 # USA MOUNTAIN 7 H BEHIND UTC DENVER # USA MOUNTAIN 6 H BEHIND UTC APR 3 - OCT 30 # USA PACIFIC 8 H BEHIND UTC L.A., SAN FRANCISCO # USA PACIFIC 7 H BEHIND UTC APR 3 - OCT 30 # USA ALASKA STD 9 H BEHIND UTC MOST OF ALASKA (AKST) # USA ALASKA STD 8 H BEHIND UTC APR 3 - OCT 30 (AKDT) # USA ALEUTIAN 10 H BEHIND UTC ISLANDS WEST OF 170W # USA " 9 H BEHIND UTC APR 3 - OCT 30 # USA HAWAII 10 H BEHIND UTC # USA BERING 11 H BEHIND UTC SAMOA, MIDWAY # From Arthur David Olson (1989-01-21): # The above dates are for 1988. # Note the "AKST" and "AKDT" abbreviations, the claim that there's # no DST in Samoa, and the claim that there is DST in Alaska and the # Aleutians. # From Arthur David Olson (1988-02-13): # Legal standard time zone names, from United States Code (1982 Edition and # Supplement III), Title 15, Chapter 6, Section 260 and forward. First, names # up to 1967-04-01 (when most provisions of the Uniform Time Act of 1966 # took effect), as explained in sections 263 and 261: # (none) # United States standard eastern time # United States standard mountain time # United States standard central time # United States standard Pacific time # (none) # United States standard Alaska time # (none) # Next, names from 1967-04-01 until 1983-11-30 (the date for # public law 98-181): # Atlantic standard time # eastern standard time # central standard time # mountain standard time # Pacific standard time # Yukon standard time # Alaska-Hawaii standard time # Bering standard time # And after 1983-11-30: # Atlantic standard time # eastern standard time # central standard time # mountain standard time # Pacific standard time # Alaska standard time # Hawaii-Aleutian standard time # Samoa standard time # The law doesn't give abbreviations. # # From Paul Eggert (2000-01-08), following a heads-up from Rives McDow: # Public law 106-564 (2000-12-23) introduced the abbreviation # "Chamorro Standard Time" for time in Guam and the Northern Marianas. # See the file "australasia". # From Arthur David Olson, 2005-08-09 # The following was signed into law on 2005-08-08. # # H.R. 6, Energy Policy Act of 2005, SEC. 110. DAYLIGHT SAVINGS. # (a) Amendment.--Section 3(a) of the Uniform Time Act of 1966 (15 # U.S.C. 260a(a)) is amended-- # (1) by striking "first Sunday of April" and inserting "second # Sunday of March"; and # (2) by striking "last Sunday of October" and inserting "first # Sunday of November'. # (b) Effective Date.--Subsection (a) shall take effect 1 year after the # date of enactment of this Act or March 1, 2007, whichever is later. # (c) Report to Congress.--Not later than 9 months after the effective # date stated in subsection (b), the Secretary shall report to Congress # on the impact of this section on energy consumption in the United # States. # (d) Right to Revert.--Congress retains the right to revert the # Daylight Saving Time back to the 2005 time schedules once the # Department study is complete. # US eastern time, represented by New York # Connecticut, Delaware, District of Columbia, most of Florida, # Georgia, southeast Indiana (Dearborn and Ohio counties), eastern Kentucky # (except America/Kentucky/Louisville below), Maine, Maryland, Massachusetts, # New Hampshire, New Jersey, New York, North Carolina, Ohio, # Pennsylvania, Rhode Island, South Carolina, eastern Tennessee, # Vermont, Virginia, West Virginia # From Dave Cantor (2004-11-02): # Early this summer I had the occasion to visit the Mount Washington # Observatory weather station atop (of course!) Mount Washington [, NH].... # One of the staff members said that the station was on Eastern Standard Time # and didn't change their clocks for Daylight Saving ... so that their # reports will always have times which are 5 hours behind UTC. # From Paul Eggert (2005-08-26): # According to today's Huntsville Times # http://www.al.com/news/huntsvilletimes/index.ssf?/base/news/1125047783228320.xml&coll=1 # a few towns on Alabama's "eastern border with Georgia, such as Phenix City # in Russell County, Lanett in Chambers County and some towns in Lee County, # set their watches and clocks on Eastern time." It quotes H.H. "Bubba" # Roberts, city administrator in Phenix City. as saying "We are in the Central # time zone, but we do go by the Eastern time zone because so many people work # in Columbus." # From Paul Eggert (2014-09-06): # Monthly Notices of the Royal Astronomical Society 44, 4 (1884-02-08), 208 # says that New York City Hall time was 3 minutes 58.4 seconds fast of # Eastern time (i.e., -4:56:01.6) just before the 1883 switch. Round to the # nearest second. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule NYC 1920 only - Mar lastSun 2:00 1:00 D Rule NYC 1920 only - Oct lastSun 2:00 0 S Rule NYC 1921 1966 - Apr lastSun 2:00 1:00 D Rule NYC 1921 1954 - Sep lastSun 2:00 0 S Rule NYC 1955 1966 - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/New_York -4:56:02 - LMT 1883 Nov 18 12:03:58 -5:00 US E%sT 1920 -5:00 NYC E%sT 1942 -5:00 US E%sT 1946 -5:00 NYC E%sT 1967 -5:00 US E%sT # US central time, represented by Chicago # Alabama, Arkansas, Florida panhandle (Bay, Calhoun, Escambia, # Gulf, Holmes, Jackson, Okaloosa, Santa Rosa, Walton, and # Washington counties), Illinois, western Indiana # (Gibson, Jasper, Lake, LaPorte, Newton, Porter, Posey, Spencer, # Vanderburgh, and Warrick counties), Iowa, most of Kansas, western # Kentucky, Louisiana, Minnesota, Mississippi, Missouri, eastern # Nebraska, eastern North Dakota, Oklahoma, eastern South Dakota, # western Tennessee, most of Texas, Wisconsin # From Larry M. Smith (2006-04-26) re Wisconsin: # http://www.legis.state.wi.us/statutes/Stat0175.pdf ... # is currently enforced at the 01:00 time of change. Because the local # "bar time" in the state corresponds to 02:00, a number of citations # are issued for the "sale of class 'B' alcohol after prohibited # hours" within the deviated hour of this change every year.... # # From Douglas R. Bomberg (2007-03-12): # Wisconsin has enacted (nearly eleventh-hour) legislation to get WI # Statue 175 closer in synch with the US Congress' intent.... # http://www.legis.state.wi.us/2007/data/acts/07Act3.pdf # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule Chicago 1920 only - Jun 13 2:00 1:00 D Rule Chicago 1920 1921 - Oct lastSun 2:00 0 S Rule Chicago 1921 only - Mar lastSun 2:00 1:00 D Rule Chicago 1922 1966 - Apr lastSun 2:00 1:00 D Rule Chicago 1922 1954 - Sep lastSun 2:00 0 S Rule Chicago 1955 1966 - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Chicago -5:50:36 - LMT 1883 Nov 18 12:09:24 -6:00 US C%sT 1920 -6:00 Chicago C%sT 1936 Mar 1 2:00 -5:00 - EST 1936 Nov 15 2:00 -6:00 Chicago C%sT 1942 -6:00 US C%sT 1946 -6:00 Chicago C%sT 1967 -6:00 US C%sT # Oliver County, ND switched from mountain to central time on 1992-10-25. Zone America/North_Dakota/Center -6:45:12 - LMT 1883 Nov 18 12:14:48 -7:00 US M%sT 1992 Oct 25 2:00 -6:00 US C%sT # Morton County, ND, switched from mountain to central time on # 2003-10-26, except for the area around Mandan which was already central time. # See . # Officially this switch also included part of Sioux County, and # Jones, Mellette, and Todd Counties in South Dakota; # but in practice these other counties were already observing central time. # See . Zone America/North_Dakota/New_Salem -6:45:39 - LMT 1883 Nov 18 12:14:21 -7:00 US M%sT 2003 Oct 26 2:00 -6:00 US C%sT # From Josh Findley (2011-01-21): # ...it appears that Mercer County, North Dakota, changed from the # mountain time zone to the central time zone at the last transition from # daylight-saving to standard time (on Nov. 7, 2010): # http://www.gpo.gov/fdsys/pkg/FR-2010-09-29/html/2010-24376.htm # http://www.bismarcktribune.com/news/local/article_1eb1b588-c758-11df-b472-001cc4c03286.html # From Andy Lipscomb (2011-01-24): # ...according to the Census Bureau, the largest city is Beulah (although # it's commonly referred to as Beulah-Hazen, with Hazen being the next # largest city in Mercer County). Google Maps places Beulah's city hall # at 47 degrees 15' 51" N, 101 degrees 46' 40" W, which yields an offset # of 6h47'07". Zone America/North_Dakota/Beulah -6:47:07 - LMT 1883 Nov 18 12:12:53 -7:00 US M%sT 2010 Nov 7 2:00 -6:00 US C%sT # US mountain time, represented by Denver # # Colorado, far western Kansas, Montana, western # Nebraska, Nevada border (Jackpot, Owyhee, and Mountain City), # New Mexico, southwestern North Dakota, # western South Dakota, far western Texas (El Paso County, Hudspeth County, # and Pine Springs and Nickel Creek in Culberson County), Utah, Wyoming # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule Denver 1920 1921 - Mar lastSun 2:00 1:00 D Rule Denver 1920 only - Oct lastSun 2:00 0 S Rule Denver 1921 only - May 22 2:00 0 S Rule Denver 1965 1966 - Apr lastSun 2:00 1:00 D Rule Denver 1965 1966 - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Denver -6:59:56 - LMT 1883 Nov 18 12:00:04 -7:00 US M%sT 1920 -7:00 Denver M%sT 1942 -7:00 US M%sT 1946 -7:00 Denver M%sT 1967 -7:00 US M%sT # US Pacific time, represented by Los Angeles # # California, northern Idaho (Benewah, Bonner, Boundary, Clearwater, # Kootenai, Latah, Lewis, Nez Perce, and Shoshone counties, Idaho county # north of the Salmon River, and the towns of Burgdorf and Warren), # Nevada (except West Wendover), Oregon (except the northern 3/4 of # Malheur county), and Washington # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule CA 1948 only - Mar 14 2:00 1:00 D Rule CA 1949 only - Jan 1 2:00 0 S Rule CA 1950 1966 - Apr lastSun 2:00 1:00 D Rule CA 1950 1961 - Sep lastSun 2:00 0 S Rule CA 1962 1966 - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Los_Angeles -7:52:58 - LMT 1883 Nov 18 12:07:02 -8:00 US P%sT 1946 -8:00 CA P%sT 1967 -8:00 US P%sT # Alaska # AK%sT is the modern abbreviation for -9:00 per USNO. # # From Paul Eggert (2001-05-30): # Howse writes that Alaska switched from the Julian to the Gregorian calendar, # and from east-of-GMT to west-of-GMT days, when the US bought it from Russia. # This was on 1867-10-18, a Friday; the previous day was 1867-10-06 Julian, # also a Friday. Include only the time zone part of this transition, # ignoring the switch from Julian to Gregorian, since we can't represent # the Julian calendar. # # As far as we know, none of the exact locations mentioned below were # permanently inhabited in 1867 by anyone using either calendar. # (Yakutat was colonized by the Russians in 1799, but the settlement # was destroyed in 1805 by a Yakutat-kon war party.) However, there # were nearby inhabitants in some cases and for our purposes perhaps # it's best to simply use the official transition. # From Paul Eggert (2014-07-18): # One opinion of the early-1980s turmoil in Alaska over time zones and # daylight saving time appeared as graffiti on a Juneau airport wall: # "Welcome to Juneau. Please turn your watch back to the 19th century." # See: Turner W. Alaska's four time zones now two. NY Times 1983-11-01. # http://www.nytimes.com/1983/11/01/us/alaska-s-four-time-zones-now-two.html # # Steve Ferguson (2011-01-31) referred to the following source: # Norris F. Keeping time in Alaska: national directives, local response. # Alaska History 2001;16(1-2). # http://alaskahistoricalsociety.org/discover-alaska/glimpses-of-the-past/keeping-time-in-alaska/ # From Arthur David Olson (2011-02-01): # Here's database-relevant material from the 2001 "Alaska History" article: # # On September 20 [1979]...DOT...officials decreed that on April 27, # 1980, Juneau and other nearby communities would move to Yukon Time. # Sitka, Petersburg, Wrangell, and Ketchikan, however, would remain on # Pacific Time. # # ...on September 22, 1980, DOT Secretary Neil E. Goldschmidt rescinded the # Department's September 1979 decision. Juneau and other communities in # northern Southeast reverted to Pacific Time on October 26. # # On October 28 [1983]...the Metlakatla Indian Community Council voted # unanimously to keep the reservation on Pacific Time. # # According to DOT official Joanne Petrie, Indian reservations are not # bound to follow time zones imposed by neighboring jurisdictions. # # (The last is consistent with how the database now handles the Navajo # Nation.) # From Arthur David Olson (2011-02-09): # I just spoke by phone with a staff member at the Metlakatla Indian # Community office (using contact information available at # http://www.commerce.state.ak.us/dca/commdb/CIS.cfm?Comm_Boro_name=Metlakatla # It's shortly after 1:00 here on the east coast of the United States; # the staffer said it was shortly after 10:00 there. When I asked whether # that meant they were on Pacific time, they said no - they were on their # own time. I asked about daylight saving; they said it wasn't used. I # did not inquire about practices in the past. # From Arthur David Olson (2011-08-17): # For lack of better information, assume that Metlakatla's # abandonment of use of daylight saving resulted from the 1983 vote. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Juneau 15:02:19 - LMT 1867 Oct 18 -8:57:41 - LMT 1900 Aug 20 12:00 -8:00 - PST 1942 -8:00 US P%sT 1946 -8:00 - PST 1969 -8:00 US P%sT 1980 Apr 27 2:00 -9:00 US Y%sT 1980 Oct 26 2:00 -8:00 US P%sT 1983 Oct 30 2:00 -9:00 US Y%sT 1983 Nov 30 -9:00 US AK%sT Zone America/Sitka 14:58:47 - LMT 1867 Oct 18 -9:01:13 - LMT 1900 Aug 20 12:00 -8:00 - PST 1942 -8:00 US P%sT 1946 -8:00 - PST 1969 -8:00 US P%sT 1983 Oct 30 2:00 -9:00 US Y%sT 1983 Nov 30 -9:00 US AK%sT Zone America/Metlakatla 15:13:42 - LMT 1867 Oct 18 -8:46:18 - LMT 1900 Aug 20 12:00 -8:00 - PST 1942 -8:00 US P%sT 1946 -8:00 - PST 1969 -8:00 US P%sT 1983 Oct 30 2:00 -8:00 - PST Zone America/Yakutat 14:41:05 - LMT 1867 Oct 18 -9:18:55 - LMT 1900 Aug 20 12:00 -9:00 - YST 1942 -9:00 US Y%sT 1946 -9:00 - YST 1969 -9:00 US Y%sT 1983 Nov 30 -9:00 US AK%sT Zone America/Anchorage 14:00:24 - LMT 1867 Oct 18 -9:59:36 - LMT 1900 Aug 20 12:00 -10:00 - CAT 1942 -10:00 US CAT/CAWT 1945 Aug 14 23:00u -10:00 US CAT/CAPT 1946 # Peace -10:00 - CAT 1967 Apr -10:00 - AHST 1969 -10:00 US AH%sT 1983 Oct 30 2:00 -9:00 US Y%sT 1983 Nov 30 -9:00 US AK%sT Zone America/Nome 12:58:21 - LMT 1867 Oct 18 -11:01:38 - LMT 1900 Aug 20 12:00 -11:00 - NST 1942 -11:00 US N%sT 1946 -11:00 - NST 1967 Apr -11:00 - BST 1969 -11:00 US B%sT 1983 Oct 30 2:00 -9:00 US Y%sT 1983 Nov 30 -9:00 US AK%sT Zone America/Adak 12:13:21 - LMT 1867 Oct 18 -11:46:38 - LMT 1900 Aug 20 12:00 -11:00 - NST 1942 -11:00 US N%sT 1946 -11:00 - NST 1967 Apr -11:00 - BST 1969 -11:00 US B%sT 1983 Oct 30 2:00 -10:00 US AH%sT 1983 Nov 30 -10:00 US HA%sT # The following switches don't quite make our 1970 cutoff. # # Shanks writes that part of southwest Alaska (e.g. Aniak) # switched from -11:00 to -10:00 on 1968-09-22 at 02:00, # and another part (e.g. Akiak) made the same switch five weeks later. # # From David Flater (2004-11-09): # In e-mail, 2004-11-02, Ray Hudson, historian/liaison to the Unalaska # Historic Preservation Commission, provided this information, which # suggests that Unalaska deviated from statutory time from early 1967 # possibly until 1983: # # Minutes of the Unalaska City Council Meeting, January 10, 1967: # "Except for St. Paul and Akutan, Unalaska is the only important # location not on Alaska Standard Time. The following resolution was # made by William Robinson and seconded by Henry Swanson: Be it # resolved that the City of Unalaska hereby goes to Alaska Standard # Time as of midnight Friday, January 13, 1967 (1 A.M. Saturday, # January 14, Alaska Standard Time.) This resolution was passed with # three votes for and one against." # Hawaii # From Arthur David Olson (2010-12-09): # "Hawaiian Time" by Robert C. Schmitt and Doak C. Cox appears on pages 207-225 # of volume 26 of The Hawaiian Journal of History (1992). As of 2010-12-09, # the article is available at # http://evols.library.manoa.hawaii.edu/bitstream/10524/239/2/JL26215.pdf # and indicates that standard time was adopted effective noon, January # 13, 1896 (page 218), that in "1933, the Legislature decreed daylight # saving for the period between the last Sunday of each April and the # last Sunday of each September, but less than a month later repealed the # act," (page 220), that year-round daylight saving time was in effect # from 1942-02-09 to 1945-09-30 (page 221, with no time of day given for # when clocks changed) and that clocks were changed by 30 minutes # effective the second Sunday of June, 1947 (page 219, with no time of # day given for when clocks changed). A footnote for the 1933 changes # cites Session Laws of Hawaii 1933, "Act. 90 (approved 26 Apr. 1933) # and Act 163 (approved 21 May 1933)." # From Arthur David Olson (2011-01-19): # The following is from "Laws of the Territory of Hawaii Passed by the # Seventeenth Legislature: Regular Session 1933," available (as of # 2011-01-19) at American University's Pence Law Library. Page 85: "Act # 90...At 2 o'clock ante meridian of the last Sunday in April of each # year, the standard time of this Territory shall be advanced one # hour...This Act shall take effect upon its approval. Approved this 26th # day of April, A. D. 1933. LAWRENCE M JUDD, Governor of the Territory of # Hawaii." Page 172: "Act 163...Act 90 of the Session Laws of 1933 is # hereby repealed...This Act shall take effect upon its approval, upon # which date the standard time of this Territory shall be restored to # that existing immediately prior to the taking effect of said Act 90. # Approved this 21st day of May, A. D. 1933. LAWRENCE M. JUDD, Governor # of the Territory of Hawaii." # # Note that 1933-05-21 was a Sunday. # We're left to guess the time of day when Act 163 was approved; guess noon. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Pacific/Honolulu -10:31:26 - LMT 1896 Jan 13 12:00 -10:30 - HST 1933 Apr 30 2:00 -10:30 1:00 HDT 1933 May 21 12:00 -10:30 - HST 1942 Feb 9 2:00 -10:30 1:00 HDT 1945 Sep 30 2:00 -10:30 - HST 1947 Jun 8 2:00 -10:00 - HST Link Pacific/Honolulu Pacific/Johnston # Now we turn to US areas that have diverged from the consensus since 1970. # Arizona mostly uses MST. # From Paul Eggert (2002-10-20): # # The information in the rest of this paragraph is derived from the # Daylight Saving Time web page # (2002-01-23) # maintained by the Arizona State Library, Archives and Public Records. # Between 1944-01-01 and 1944-04-01 the State of Arizona used standard # time, but by federal law railroads, airlines, bus lines, military # personnel, and some engaged in interstate commerce continued to # observe war (i.e., daylight saving) time. The 1944-03-17 Phoenix # Gazette says that was the date the law changed, and that 04-01 was # the date the state's clocks would change. In 1945 the State of # Arizona used standard time all year, again with exceptions only as # mandated by federal law. Arizona observed DST in 1967, but Arizona # Laws 1968, ch. 183 (effective 1968-03-21) repealed DST. # # Shanks says the 1944 experiment came to an end on 1944-03-17. # Go with the Arizona State Library instead. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Phoenix -7:28:18 - LMT 1883 Nov 18 11:31:42 -7:00 US M%sT 1944 Jan 1 0:01 -7:00 - MST 1944 Apr 1 0:01 -7:00 US M%sT 1944 Oct 1 0:01 -7:00 - MST 1967 -7:00 US M%sT 1968 Mar 21 -7:00 - MST # From Arthur David Olson (1988-02-13): # A writer from the Inter Tribal Council of Arizona, Inc., # notes in private correspondence dated 1987-12-28 that "Presently, only the # Navajo Nation participates in the Daylight Saving Time policy, due to its # large size and location in three states." (The "only" means that other # tribal nations don't use DST.) # # From Paul Eggert (2013-08-26): # See America/Denver for a zone appropriate for the Navajo Nation. # Southern Idaho (Ada, Adams, Bannock, Bear Lake, Bingham, Blaine, # Boise, Bonneville, Butte, Camas, Canyon, Caribou, Cassia, Clark, # Custer, Elmore, Franklin, Fremont, Gem, Gooding, Jefferson, Jerome, # Lemhi, Lincoln, Madison, Minidoka, Oneida, Owyhee, Payette, Power, # Teton, Twin Falls, Valley, Washington counties, and the southern # quarter of Idaho county) and eastern Oregon (most of Malheur County) # switched four weeks late in 1974. # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Boise -7:44:49 - LMT 1883 Nov 18 12:15:11 -8:00 US P%sT 1923 May 13 2:00 -7:00 US M%sT 1974 -7:00 - MST 1974 Feb 3 2:00 -7:00 US M%sT # Indiana # # For a map of Indiana's time zone regions, see: # http://en.wikipedia.org/wiki/Time_in_Indiana # # From Paul Eggert (2007-08-17): # Since 1970, most of Indiana has been like America/Indiana/Indianapolis, # with the following exceptions: # # - Gibson, Jasper, Lake, LaPorte, Newton, Porter, Posey, Spencer, # Vanderburgh, and Warrick counties have been like America/Chicago. # # - Dearborn and Ohio counties have been like America/New_York. # # - Clark, Floyd, and Harrison counties have been like # America/Kentucky/Louisville. # # - Crawford, Daviess, Dubois, Knox, Martin, Perry, Pike, Pulaski, Starke, # and Switzerland counties have their own time zone histories as noted below. # # Shanks partitioned Indiana into 345 regions, each with its own time history, # and wrote "Even newspaper reports present contradictory information." # Those Hoosiers! Such a flighty and changeable people! # Fortunately, most of the complexity occurred before our cutoff date of 1970. # # Other than Indianapolis, the Indiana place names are so nondescript # that they would be ambiguous if we left them at the 'America' level. # So we reluctantly put them all in a subdirectory 'America/Indiana'. # From Paul Eggert (2014-06-26): # https://www.federalregister.gov/articles/2006/01/20/06-563/standard-time-zone-boundary-in-the-state-of-indiana # says "DOT is relocating the time zone boundary in Indiana to move Starke, # Pulaski, Knox, Daviess, Martin, Pike, Dubois, and Perry Counties from the # Eastern Time Zone to the Central Time Zone.... The effective date of # this rule is 2 a.m. EST Sunday, April 2, 2006, which is the # changeover date from standard time to Daylight Saving Time." # Strictly speaking, this meant the affected counties changed their # clocks twice that night, but this obviously was in error. The intent # was that 01:59:59 EST be followed by 02:00:00 CDT. # From Gwillim Law (2007-02-10): # The Associated Press has been reporting that Pulaski County, Indiana is # going to switch from Central to Eastern Time on March 11, 2007.... # http://www.indystar.com/apps/pbcs.dll/article?AID=/20070207/LOCAL190108/702070524/0/LOCAL # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule Indianapolis 1941 only - Jun 22 2:00 1:00 D Rule Indianapolis 1941 1954 - Sep lastSun 2:00 0 S Rule Indianapolis 1946 1954 - Apr lastSun 2:00 1:00 D # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Indiana/Indianapolis -5:44:38 - LMT 1883 Nov 18 12:15:22 -6:00 US C%sT 1920 -6:00 Indianapolis C%sT 1942 -6:00 US C%sT 1946 -6:00 Indianapolis C%sT 1955 Apr 24 2:00 -5:00 - EST 1957 Sep 29 2:00 -6:00 - CST 1958 Apr 27 2:00 -5:00 - EST 1969 -5:00 US E%sT 1971 -5:00 - EST 2006 -5:00 US E%sT # # Eastern Crawford County, Indiana, left its clocks alone in 1974, # as well as from 1976 through 2005. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule Marengo 1951 only - Apr lastSun 2:00 1:00 D Rule Marengo 1951 only - Sep lastSun 2:00 0 S Rule Marengo 1954 1960 - Apr lastSun 2:00 1:00 D Rule Marengo 1954 1960 - Sep lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Indiana/Marengo -5:45:23 - LMT 1883 Nov 18 12:14:37 -6:00 US C%sT 1951 -6:00 Marengo C%sT 1961 Apr 30 2:00 -5:00 - EST 1969 -5:00 US E%sT 1974 Jan 6 2:00 -6:00 1:00 CDT 1974 Oct 27 2:00 -5:00 US E%sT 1976 -5:00 - EST 2006 -5:00 US E%sT # # Daviess, Dubois, Knox, and Martin Counties, Indiana, # switched from eastern to central time in April 2006, then switched back # in November 2007. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule Vincennes 1946 only - Apr lastSun 2:00 1:00 D Rule Vincennes 1946 only - Sep lastSun 2:00 0 S Rule Vincennes 1953 1954 - Apr lastSun 2:00 1:00 D Rule Vincennes 1953 1959 - Sep lastSun 2:00 0 S Rule Vincennes 1955 only - May 1 0:00 1:00 D Rule Vincennes 1956 1963 - Apr lastSun 2:00 1:00 D Rule Vincennes 1960 only - Oct lastSun 2:00 0 S Rule Vincennes 1961 only - Sep lastSun 2:00 0 S Rule Vincennes 1962 1963 - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Indiana/Vincennes -5:50:07 - LMT 1883 Nov 18 12:09:53 -6:00 US C%sT 1946 -6:00 Vincennes C%sT 1964 Apr 26 2:00 -5:00 - EST 1969 -5:00 US E%sT 1971 -5:00 - EST 2006 Apr 2 2:00 -6:00 US C%sT 2007 Nov 4 2:00 -5:00 US E%sT # # Perry County, Indiana, switched from eastern to central time in April 2006. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule Perry 1946 only - Apr lastSun 2:00 1:00 D Rule Perry 1946 only - Sep lastSun 2:00 0 S Rule Perry 1953 1954 - Apr lastSun 2:00 1:00 D Rule Perry 1953 1959 - Sep lastSun 2:00 0 S Rule Perry 1955 only - May 1 0:00 1:00 D Rule Perry 1956 1963 - Apr lastSun 2:00 1:00 D Rule Perry 1960 only - Oct lastSun 2:00 0 S Rule Perry 1961 only - Sep lastSun 2:00 0 S Rule Perry 1962 1963 - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Indiana/Tell_City -5:47:03 - LMT 1883 Nov 18 12:12:57 -6:00 US C%sT 1946 -6:00 Perry C%sT 1964 Apr 26 2:00 -5:00 - EST 1969 -5:00 US E%sT 1971 -5:00 - EST 2006 Apr 2 2:00 -6:00 US C%sT # # Pike County, Indiana moved from central to eastern time in 1977, # then switched back in 2006, then switched back again in 2007. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule Pike 1955 only - May 1 0:00 1:00 D Rule Pike 1955 1960 - Sep lastSun 2:00 0 S Rule Pike 1956 1964 - Apr lastSun 2:00 1:00 D Rule Pike 1961 1964 - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Indiana/Petersburg -5:49:07 - LMT 1883 Nov 18 12:10:53 -6:00 US C%sT 1955 -6:00 Pike C%sT 1965 Apr 25 2:00 -5:00 - EST 1966 Oct 30 2:00 -6:00 US C%sT 1977 Oct 30 2:00 -5:00 - EST 2006 Apr 2 2:00 -6:00 US C%sT 2007 Nov 4 2:00 -5:00 US E%sT # # Starke County, Indiana moved from central to eastern time in 1991, # then switched back in 2006. # From Arthur David Olson (1991-10-28): # An article on page A3 of the Sunday, 1991-10-27 Washington Post # notes that Starke County switched from Central time to Eastern time as of # 1991-10-27. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule Starke 1947 1961 - Apr lastSun 2:00 1:00 D Rule Starke 1947 1954 - Sep lastSun 2:00 0 S Rule Starke 1955 1956 - Oct lastSun 2:00 0 S Rule Starke 1957 1958 - Sep lastSun 2:00 0 S Rule Starke 1959 1961 - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Indiana/Knox -5:46:30 - LMT 1883 Nov 18 12:13:30 -6:00 US C%sT 1947 -6:00 Starke C%sT 1962 Apr 29 2:00 -5:00 - EST 1963 Oct 27 2:00 -6:00 US C%sT 1991 Oct 27 2:00 -5:00 - EST 2006 Apr 2 2:00 -6:00 US C%sT # # Pulaski County, Indiana, switched from eastern to central time in # April 2006 and then switched back in March 2007. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule Pulaski 1946 1960 - Apr lastSun 2:00 1:00 D Rule Pulaski 1946 1954 - Sep lastSun 2:00 0 S Rule Pulaski 1955 1956 - Oct lastSun 2:00 0 S Rule Pulaski 1957 1960 - Sep lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Indiana/Winamac -5:46:25 - LMT 1883 Nov 18 12:13:35 -6:00 US C%sT 1946 -6:00 Pulaski C%sT 1961 Apr 30 2:00 -5:00 - EST 1969 -5:00 US E%sT 1971 -5:00 - EST 2006 Apr 2 2:00 -6:00 US C%sT 2007 Mar 11 2:00 -5:00 US E%sT # # Switzerland County, Indiana, did not observe DST from 1973 through 2005. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Indiana/Vevay -5:40:16 - LMT 1883 Nov 18 12:19:44 -6:00 US C%sT 1954 Apr 25 2:00 -5:00 - EST 1969 -5:00 US E%sT 1973 -5:00 - EST 2006 -5:00 US E%sT # Part of Kentucky left its clocks alone in 1974. # This also includes Clark, Floyd, and Harrison counties in Indiana. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule Louisville 1921 only - May 1 2:00 1:00 D Rule Louisville 1921 only - Sep 1 2:00 0 S Rule Louisville 1941 1961 - Apr lastSun 2:00 1:00 D Rule Louisville 1941 only - Sep lastSun 2:00 0 S Rule Louisville 1946 only - Jun 2 2:00 0 S Rule Louisville 1950 1955 - Sep lastSun 2:00 0 S Rule Louisville 1956 1960 - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Kentucky/Louisville -5:43:02 - LMT 1883 Nov 18 12:16:58 -6:00 US C%sT 1921 -6:00 Louisville C%sT 1942 -6:00 US C%sT 1946 -6:00 Louisville C%sT 1961 Jul 23 2:00 -5:00 - EST 1968 -5:00 US E%sT 1974 Jan 6 2:00 -6:00 1:00 CDT 1974 Oct 27 2:00 -5:00 US E%sT # # Wayne County, Kentucky # # From Lake Cumberland LIFE # http://www.lake-cumberland.com/life/archive/news990129time.shtml # (1999-01-29) via WKYM-101.7: # Clinton County has joined Wayne County in asking the DoT to change from # the Central to the Eastern time zone.... The Wayne County government made # the same request in December. And while Russell County officials have not # taken action, the majority of respondents to a poll conducted there in # August indicated they would like to change to "fast time" also. # The three Lake Cumberland counties are the farthest east of any U.S. # location in the Central time zone. # # From Rich Wales (2000-08-29): # After prolonged debate, and despite continuing deep differences of opinion, # Wayne County (central Kentucky) is switching from Central (-0600) to Eastern # (-0500) time. They won't "fall back" this year. See Sara Shipley, # The difference an hour makes, Nando Times (2000-08-29 15:33 -0400). # # From Paul Eggert (2001-07-16): # The final rule was published in the # Federal Register 65, 160 (2000-08-17), pp 50154-50158. # http://frwebgate.access.gpo.gov/cgi-bin/getdoc.cgi?dbname=2000_register&docid=fr17au00-22 # Zone America/Kentucky/Monticello -5:39:24 - LMT 1883 Nov 18 12:20:36 -6:00 US C%sT 1946 -6:00 - CST 1968 -6:00 US C%sT 2000 Oct 29 2:00 -5:00 US E%sT # From Rives McDow (2000-08-30): # Here ... are all the changes in the US since 1985. # Kearny County, KS (put all of county on central; # previously split between MST and CST) ... 1990-10 # Starke County, IN (from CST to EST) ... 1991-10 # Oliver County, ND (from MST to CST) ... 1992-10 # West Wendover, NV (from PST TO MST) ... 1999-10 # Wayne County, KY (from CST to EST) ... 2000-10 # # From Paul Eggert (2001-07-17): # We don't know where the line used to be within Kearny County, KS, # so omit that change for now. # See America/Indiana/Knox for the Starke County, IN change. # See America/North_Dakota/Center for the Oliver County, ND change. # West Wendover, NV officially switched from Pacific to mountain time on # 1999-10-31. See the # Federal Register 64, 203 (1999-10-21), pp 56705-56707. # http://frwebgate.access.gpo.gov/cgi-bin/getdoc.cgi?dbname=1999_register&docid=fr21oc99-15 # However, the Federal Register says that West Wendover already operated # on mountain time, and the rule merely made this official; # hence a separate tz entry is not needed. # Michigan # # From Bob Devine (1988-01-28): # Michigan didn't observe DST from 1968 to 1973. # # From Paul Eggert (1999-03-31): # Shanks writes that Michigan started using standard time on 1885-09-18, # but Howse writes (pp 124-125, referring to Popular Astronomy, 1901-01) # that Detroit kept # # local time until 1900 when the City Council decreed that clocks should # be put back twenty-eight minutes to Central Standard Time. Half the # city obeyed, half refused. After considerable debate, the decision # was rescinded and the city reverted to Sun time. A derisive offer to # erect a sundial in front of the city hall was referred to the # Committee on Sewers. Then, in 1905, Central time was adopted # by city vote. # # This story is too entertaining to be false, so go with Howse over Shanks. # # From Paul Eggert (2001-03-06): # Garland (1927) writes "Cleveland and Detroit advanced their clocks # one hour in 1914." This change is not in Shanks. We have no more # info, so omit this for now. # # Most of Michigan observed DST from 1973 on, but was a bit late in 1975. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule Detroit 1948 only - Apr lastSun 2:00 1:00 D Rule Detroit 1948 only - Sep lastSun 2:00 0 S Rule Detroit 1967 only - Jun 14 2:00 1:00 D Rule Detroit 1967 only - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Detroit -5:32:11 - LMT 1905 -6:00 - CST 1915 May 15 2:00 -5:00 - EST 1942 -5:00 US E%sT 1946 -5:00 Detroit E%sT 1973 -5:00 US E%sT 1975 -5:00 - EST 1975 Apr 27 2:00 -5:00 US E%sT # # Dickinson, Gogebic, Iron, and Menominee Counties, Michigan, # switched from EST to CST/CDT in 1973. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER Rule Menominee 1946 only - Apr lastSun 2:00 1:00 D Rule Menominee 1946 only - Sep lastSun 2:00 0 S Rule Menominee 1966 only - Apr lastSun 2:00 1:00 D Rule Menominee 1966 only - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Menominee -5:50:27 - LMT 1885 Sep 18 12:00 -6:00 US C%sT 1946 -6:00 Menominee C%sT 1969 Apr 27 2:00 -5:00 - EST 1973 Apr 29 2:00 -6:00 US C%sT # Navassa # administered by the US Fish and Wildlife Service # claimed by US under the provisions of the 1856 Guano Islands Act # also claimed by Haiti # occupied 1857/1900 by the Navassa Phosphate Co # US lighthouse 1917/1996-09 # currently uninhabited # see Mark Fineman, "An Isle Rich in Guano and Discord", # _Los Angeles Times_ (1998-11-10), A1, A10; it cites # Jimmy Skaggs, _The Great Guano Rush_ (1994). ################################################################################ # From Paul Eggert (2014-10-31): # # Unless otherwise specified, the source for data through 1990 is: # Thomas G. Shanks and Rique Pottenger, The International Atlas (6th edition), # San Diego: ACS Publications, Inc. (2003). # Unfortunately this book contains many errors and cites no sources. # # Gwillim Law writes that a good source # for recent time zone data is the International Air Transport # Association's Standard Schedules Information Manual (IATA SSIM), # published semiannually. Law sent in several helpful summaries # of the IATA's data after 1990. Except where otherwise noted, # IATA SSIM is the source for entries after 1990. # # Other sources occasionally used include: # # Edward W. Whitman, World Time Differences, # Whitman Publishing Co, 2 Niagara Av, Ealing, London (undated), # which I found in the UCLA library. # # William Willett, The Waste of Daylight, 19th edition # # [PDF] (1914-03) # # Milne J. Civil time. Geogr J. 1899 Feb;13(2):173-94 # . # # See the 'europe' file for Greenland. # Canada # From Alain LaBonté (1994-11-14): # I post here the time zone abbreviations standardized in Canada # for both English and French in the CAN/CSA-Z234.4-89 standard.... # # UTC Standard time Daylight saving time # offset French English French English # -2:30 - - HAT NDT # -3 - - HAA ADT # -3:30 HNT NST - - # -4 HNA AST HAE EDT # -5 HNE EST HAC CDT # -6 HNC CST HAR MDT # -7 HNR MST HAP PDT # -8 HNP PST HAY YDT # -9 HNY YST - - # # HN: Heure Normale ST: Standard Time # HA: Heure Avancée DT: Daylight saving Time # # A: de l'Atlantique Atlantic # C: du Centre Central # E: de l'Est Eastern # M: Mountain # N: Newfoundland # P: du Pacifique Pacific # R: des Rocheuses # T: de Terre-Neuve # Y: du Yukon Yukon # # From Paul Eggert (1994-11-22): # Alas, this sort of thing must be handled by localization software. # Unless otherwise specified, the data entries for Canada are all from Shanks # & Pottenger. # From Chris Walton (2006-04-01, 2006-04-25, 2006-06-26, 2007-01-31, # 2007-03-01): # The British Columbia government announced yesterday that it will # adjust daylight savings next year to align with changes in the # U.S. and the rest of Canada.... # http://www2.news.gov.bc.ca/news_releases_2005-2009/2006AG0014-000330.htm # ... # Nova Scotia # Daylight saving time will be extended by four weeks starting in 2007.... # http://www.gov.ns.ca/just/regulations/rg2/2006/ma1206.pdf # # [For New Brunswick] the new legislation dictates that the time change is to # be done at 02:00 instead of 00:01. # http://www.gnb.ca/0062/acts/BBA-2006/Chap-19.pdf # ... # Manitoba has traditionally changed the clock every fall at 03:00. # As of 2006, the transition is to take place one hour earlier at 02:00. # http://web2.gov.mb.ca/laws/statutes/ccsm/o030e.php # ... # [Alberta, Ontario, Quebec] will follow US rules. # http://www.qp.gov.ab.ca/documents/spring/CH03_06.CFM # http://www.e-laws.gov.on.ca/DBLaws/Source/Regs/English/2006/R06111_e.htm # http://www2.publicationsduquebec.gouv.qc.ca/dynamicSearch/telecharge.php?type=5&file=2006C39A.PDF # ... # P.E.I. will follow US rules.... # http://www.assembly.pe.ca/bills/pdf_chapter/62/3/chapter-41.pdf # ... # Province of Newfoundland and Labrador.... # http://www.hoa.gov.nl.ca/hoa/bills/Bill0634.htm # ... # Yukon # http://www.gov.yk.ca/legislation/regs/oic2006_127.pdf # ... # N.W.T. will follow US rules. Whoever maintains the government web site # does not seem to believe in bookmarks. To see the news release, click the # following link and search for "Daylight Savings Time Change". Press the # "Daylight Savings Time Change" link; it will fire off a popup using # JavaScript. # http://www.exec.gov.nt.ca/currentnews/currentPR.asp?mode=archive # ... # Nunavut # An amendment to the Interpretation Act was registered on February 19/2007.... # http://action.attavik.ca/home/justice-gn/attach/2007/gaz02part2.pdf # From Paul Eggert (2014-10-18): # H. David Matthews and Mary Vincent's map # "It's about TIME", _Canadian Geographic_ (September-October 1998) # http://www.canadiangeographic.ca/Magazine/SO98/alacarte.asp # contains detailed boundaries for regions observing nonstandard # time and daylight saving time arrangements in Canada circa 1998. # # National Research Council Canada maintains info about time zones and DST. # http://www.nrc-cnrc.gc.ca/eng/services/time/time_zones.html # http://www.nrc-cnrc.gc.ca/eng/services/time/faq/index.html#Q5 # Its unofficial information is often taken from Matthews and Vincent. # From Paul Eggert (2006-06-27): # For now, assume all of DST-observing Canada will fall into line with the # new US DST rules, # From Chris Walton (2011-12-01) # In the first of Tammy Hardwick's articles # http://www.ilovecreston.com/?p=articles&t=spec&ar=260 # she quotes the Friday November 1/1918 edition of the Creston Review. # The quote includes these two statements: # 'Sunday the CPR went back to the old system of time...' # '... The daylight saving scheme was dropped all over Canada at the same time,' # These statements refer to a transition from daylight time to standard time # that occurred nationally on Sunday October 27/1918. This transition was # also documented in the Saturday October 26/1918 edition of the Toronto Star. # In light of that evidence, we alter the date from the earlier believed # Oct 31, to Oct 27, 1918 (and Sunday is a more likely transition day # than Thursday) in all Canadian rulesets. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Canada 1918 only - Apr 14 2:00 1:00 D Rule Canada 1918 only - Oct 27 2:00 0 S Rule Canada 1942 only - Feb 9 2:00 1:00 W # War Rule Canada 1945 only - Aug 14 23:00u 1:00 P # Peace Rule Canada 1945 only - Sep 30 2:00 0 S Rule Canada 1974 1986 - Apr lastSun 2:00 1:00 D Rule Canada 1974 2006 - Oct lastSun 2:00 0 S Rule Canada 1987 2006 - Apr Sun>=1 2:00 1:00 D Rule Canada 2007 max - Mar Sun>=8 2:00 1:00 D Rule Canada 2007 max - Nov Sun>=1 2:00 0 S # Newfoundland and Labrador # From Paul Eggert (2000-10-02): # Matthews and Vincent (1998) write that Labrador should use NST/NDT, # but the only part of Labrador that follows the rules is the # southeast corner, including Port Hope Simpson and Mary's Harbour, # but excluding, say, Black Tickle. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule StJohns 1917 only - Apr 8 2:00 1:00 D Rule StJohns 1917 only - Sep 17 2:00 0 S # Whitman gives 1919 Apr 5 and 1920 Apr 5; go with Shanks & Pottenger. Rule StJohns 1919 only - May 5 23:00 1:00 D Rule StJohns 1919 only - Aug 12 23:00 0 S # For 1931-1935 Whitman gives Apr same date; go with Shanks & Pottenger. Rule StJohns 1920 1935 - May Sun>=1 23:00 1:00 D Rule StJohns 1920 1935 - Oct lastSun 23:00 0 S # For 1936-1941 Whitman gives May Sun>=8 and Oct Sun>=1; go with Shanks & # Pottenger. Rule StJohns 1936 1941 - May Mon>=9 0:00 1:00 D Rule StJohns 1936 1941 - Oct Mon>=2 0:00 0 S # Whitman gives the following transitions: # 1942 03-01/12-31, 1943 05-30/09-05, 1944 07-10/09-02, 1945 01-01/10-07 # but go with Shanks & Pottenger and assume they used Canadian rules. # For 1946-9 Whitman gives May 5,4,9,1 - Oct 1,5,3,2, and for 1950 he gives # Apr 30 - Sep 24; go with Shanks & Pottenger. Rule StJohns 1946 1950 - May Sun>=8 2:00 1:00 D Rule StJohns 1946 1950 - Oct Sun>=2 2:00 0 S Rule StJohns 1951 1986 - Apr lastSun 2:00 1:00 D Rule StJohns 1951 1959 - Sep lastSun 2:00 0 S Rule StJohns 1960 1986 - Oct lastSun 2:00 0 S # From Paul Eggert (2000-10-02): # INMS (2000-09-12) says that, since 1988 at least, Newfoundland switches # at 00:01 local time. For now, assume it started in 1987. # From Michael Pelley (2011-09-12): # We received today, Monday, September 12, 2011, notification that the # changes to the Newfoundland Standard Time Act have been proclaimed. # The change in the Act stipulates that the change from Daylight Savings # Time to Standard Time and from Standard Time to Daylight Savings Time # now occurs at 2:00AM. # ... # http://www.assembly.nl.ca/legislation/sr/annualstatutes/2011/1106.chp.htm # ... # MICHAEL PELLEY | Manager of Enterprise Architecture - Solution Delivery # Office of the Chief Information Officer # Executive Council # Government of Newfoundland & Labrador Rule StJohns 1987 only - Apr Sun>=1 0:01 1:00 D Rule StJohns 1987 2006 - Oct lastSun 0:01 0 S Rule StJohns 1988 only - Apr Sun>=1 0:01 2:00 DD Rule StJohns 1989 2006 - Apr Sun>=1 0:01 1:00 D Rule StJohns 2007 2011 - Mar Sun>=8 0:01 1:00 D Rule StJohns 2007 2010 - Nov Sun>=1 0:01 0 S # # St John's has an apostrophe, but Posix file names can't have apostrophes. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/St_Johns -3:30:52 - LMT 1884 -3:30:52 StJohns N%sT 1918 -3:30:52 Canada N%sT 1919 -3:30:52 StJohns N%sT 1935 Mar 30 -3:30 StJohns N%sT 1942 May 11 -3:30 Canada N%sT 1946 -3:30 StJohns N%sT 2011 Nov -3:30 Canada N%sT # most of east Labrador # The name 'Happy Valley-Goose Bay' is too long; use 'Goose Bay'. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Goose_Bay -4:01:40 - LMT 1884 # Happy Valley-Goose Bay -3:30:52 - NST 1918 -3:30:52 Canada N%sT 1919 -3:30:52 - NST 1935 Mar 30 -3:30 - NST 1936 -3:30 StJohns N%sT 1942 May 11 -3:30 Canada N%sT 1946 -3:30 StJohns N%sT 1966 Mar 15 2:00 -4:00 StJohns A%sT 2011 Nov -4:00 Canada A%sT # west Labrador, Nova Scotia, Prince Edward I # From Paul Eggert (2006-03-22): # Shanks & Pottenger write that since 1970 most of this region has been like # Halifax. Many locales did not observe peacetime DST until 1972; # Glace Bay, NS is the largest that we know of. # Shanks & Pottenger also write that Liverpool, NS was the only town # in Canada to observe DST in 1971 but not 1970; for now we'll assume # this is a typo. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Halifax 1916 only - Apr 1 0:00 1:00 D Rule Halifax 1916 only - Oct 1 0:00 0 S Rule Halifax 1920 only - May 9 0:00 1:00 D Rule Halifax 1920 only - Aug 29 0:00 0 S Rule Halifax 1921 only - May 6 0:00 1:00 D Rule Halifax 1921 1922 - Sep 5 0:00 0 S Rule Halifax 1922 only - Apr 30 0:00 1:00 D Rule Halifax 1923 1925 - May Sun>=1 0:00 1:00 D Rule Halifax 1923 only - Sep 4 0:00 0 S Rule Halifax 1924 only - Sep 15 0:00 0 S Rule Halifax 1925 only - Sep 28 0:00 0 S Rule Halifax 1926 only - May 16 0:00 1:00 D Rule Halifax 1926 only - Sep 13 0:00 0 S Rule Halifax 1927 only - May 1 0:00 1:00 D Rule Halifax 1927 only - Sep 26 0:00 0 S Rule Halifax 1928 1931 - May Sun>=8 0:00 1:00 D Rule Halifax 1928 only - Sep 9 0:00 0 S Rule Halifax 1929 only - Sep 3 0:00 0 S Rule Halifax 1930 only - Sep 15 0:00 0 S Rule Halifax 1931 1932 - Sep Mon>=24 0:00 0 S Rule Halifax 1932 only - May 1 0:00 1:00 D Rule Halifax 1933 only - Apr 30 0:00 1:00 D Rule Halifax 1933 only - Oct 2 0:00 0 S Rule Halifax 1934 only - May 20 0:00 1:00 D Rule Halifax 1934 only - Sep 16 0:00 0 S Rule Halifax 1935 only - Jun 2 0:00 1:00 D Rule Halifax 1935 only - Sep 30 0:00 0 S Rule Halifax 1936 only - Jun 1 0:00 1:00 D Rule Halifax 1936 only - Sep 14 0:00 0 S Rule Halifax 1937 1938 - May Sun>=1 0:00 1:00 D Rule Halifax 1937 1941 - Sep Mon>=24 0:00 0 S Rule Halifax 1939 only - May 28 0:00 1:00 D Rule Halifax 1940 1941 - May Sun>=1 0:00 1:00 D Rule Halifax 1946 1949 - Apr lastSun 2:00 1:00 D Rule Halifax 1946 1949 - Sep lastSun 2:00 0 S Rule Halifax 1951 1954 - Apr lastSun 2:00 1:00 D Rule Halifax 1951 1954 - Sep lastSun 2:00 0 S Rule Halifax 1956 1959 - Apr lastSun 2:00 1:00 D Rule Halifax 1956 1959 - Sep lastSun 2:00 0 S Rule Halifax 1962 1973 - Apr lastSun 2:00 1:00 D Rule Halifax 1962 1973 - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Halifax -4:14:24 - LMT 1902 Jun 15 -4:00 Halifax A%sT 1918 -4:00 Canada A%sT 1919 -4:00 Halifax A%sT 1942 Feb 9 2:00s -4:00 Canada A%sT 1946 -4:00 Halifax A%sT 1974 -4:00 Canada A%sT Zone America/Glace_Bay -3:59:48 - LMT 1902 Jun 15 -4:00 Canada A%sT 1953 -4:00 Halifax A%sT 1954 -4:00 - AST 1972 -4:00 Halifax A%sT 1974 -4:00 Canada A%sT # New Brunswick # From Paul Eggert (2007-01-31): # The Time Definition Act # says they changed at 00:01 through 2006, and # makes it # clear that this was the case since at least 1993. # For now, assume it started in 1993. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Moncton 1933 1935 - Jun Sun>=8 1:00 1:00 D Rule Moncton 1933 1935 - Sep Sun>=8 1:00 0 S Rule Moncton 1936 1938 - Jun Sun>=1 1:00 1:00 D Rule Moncton 1936 1938 - Sep Sun>=1 1:00 0 S Rule Moncton 1939 only - May 27 1:00 1:00 D Rule Moncton 1939 1941 - Sep Sat>=21 1:00 0 S Rule Moncton 1940 only - May 19 1:00 1:00 D Rule Moncton 1941 only - May 4 1:00 1:00 D Rule Moncton 1946 1972 - Apr lastSun 2:00 1:00 D Rule Moncton 1946 1956 - Sep lastSun 2:00 0 S Rule Moncton 1957 1972 - Oct lastSun 2:00 0 S Rule Moncton 1993 2006 - Apr Sun>=1 0:01 1:00 D Rule Moncton 1993 2006 - Oct lastSun 0:01 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Moncton -4:19:08 - LMT 1883 Dec 9 -5:00 - EST 1902 Jun 15 -4:00 Canada A%sT 1933 -4:00 Moncton A%sT 1942 -4:00 Canada A%sT 1946 -4:00 Moncton A%sT 1973 -4:00 Canada A%sT 1993 -4:00 Moncton A%sT 2007 -4:00 Canada A%sT # Quebec # From Paul Eggert (2013-08-30): # Since 1970 most of Quebec has been like Toronto. # However, because earlier versions of the tz database mistakenly relied on data # from Shanks & Pottenger saying that Quebec differed from Ontario after 1970, # a separate entry was created for most of Quebec. We're loath to lose # its pre-1970 info, even though the tz database is normally limited to # zones that differ after 1970, so keep this otherwise out-of-scope entry. # Matthews and Vincent (1998) also write that Quebec east of the -63 # meridian is supposed to observe AST, but residents as far east as # Natashquan use EST/EDT, and residents east of Natashquan use AST. # The Quebec department of justice writes in # "The situation in Minganie and Basse-Côte-Nord" # http://www.justice.gouv.qc.ca/english/publications/generale/temps-minganie-a.htm # that the coastal strip from just east of Natashquan to Blanc-Sablon # observes Atlantic standard time all year round. # http://www.assnat.qc.ca/Media/Process.aspx?MediaId=ANQ.Vigie.Bll.DocumentGenerique_8845en # says this common practice was codified into law as of 2007. # For lack of better info, guess this practice began around 1970, contra to # Shanks & Pottenger who have this region observing AST/ADT. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Mont 1917 only - Mar 25 2:00 1:00 D Rule Mont 1917 only - Apr 24 0:00 0 S Rule Mont 1919 only - Mar 31 2:30 1:00 D Rule Mont 1919 only - Oct 25 2:30 0 S Rule Mont 1920 only - May 2 2:30 1:00 D Rule Mont 1920 1922 - Oct Sun>=1 2:30 0 S Rule Mont 1921 only - May 1 2:00 1:00 D Rule Mont 1922 only - Apr 30 2:00 1:00 D Rule Mont 1924 only - May 17 2:00 1:00 D Rule Mont 1924 1926 - Sep lastSun 2:30 0 S Rule Mont 1925 1926 - May Sun>=1 2:00 1:00 D Rule Mont 1927 1937 - Apr lastSat 24:00 1:00 D Rule Mont 1927 1937 - Sep lastSat 24:00 0 S Rule Mont 1938 1940 - Apr lastSun 0:00 1:00 D Rule Mont 1938 1939 - Sep lastSun 0:00 0 S Rule Mont 1946 1973 - Apr lastSun 2:00 1:00 D Rule Mont 1945 1948 - Sep lastSun 2:00 0 S Rule Mont 1949 1950 - Oct lastSun 2:00 0 S Rule Mont 1951 1956 - Sep lastSun 2:00 0 S Rule Mont 1957 1973 - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Blanc-Sablon -3:48:28 - LMT 1884 -4:00 Canada A%sT 1970 -4:00 - AST Zone America/Montreal -4:54:16 - LMT 1884 -5:00 Mont E%sT 1918 -5:00 Canada E%sT 1919 -5:00 Mont E%sT 1942 Feb 9 2:00s -5:00 Canada E%sT 1946 -5:00 Mont E%sT 1974 -5:00 Canada E%sT # Ontario # From Paul Eggert (2006-07-09): # Shanks & Pottenger write that since 1970 most of Ontario has been like # Toronto. # Thunder Bay skipped DST in 1973. # Many smaller locales did not observe peacetime DST until 1974; # Nipigon (EST) and Rainy River (CST) are the largest that we know of. # Far west Ontario is like Winnipeg; far east Quebec is like Halifax. # From Mark Brader (2003-07-26): # [According to the Toronto Star] Orillia, Ontario, adopted DST # effective Saturday, 1912-06-22, 22:00; the article mentions that # Port Arthur (now part of Thunder Bay, Ontario) as well as Moose Jaw # have already done so. In Orillia DST was to run until Saturday, # 1912-08-31 (no time mentioned), but it was met with considerable # hostility from certain segments of the public, and was revoked after # only two weeks - I copied it as Saturday, 1912-07-07, 22:00, but # presumably that should be -07-06. (1912-06-19, -07-12; also letters # earlier in June). # # Kenora, Ontario, was to abandon DST on 1914-06-01 (-05-21). # From Paul Eggert (1997-10-17): # Mark Brader writes that an article in the 1997-10-14 Toronto Star # says that Atikokan, Ontario currently does not observe DST, # but will vote on 11-10 whether to use EST/EDT. # He also writes that the Ontario Time Act (1990, Chapter T.9) # http://www.gov.on.ca/MBS/english/publications/statregs/conttext.html # says that Ontario east of 90W uses EST/EDT, and west of 90W uses CST/CDT. # Officially Atikokan is therefore on CST/CDT, and most likely this report # concerns a non-official time observed as a matter of local practice. # # From Paul Eggert (2000-10-02): # Matthews and Vincent (1998) write that Atikokan, Pickle Lake, and # New Osnaburgh observe CST all year, that Big Trout Lake observes # CST/CDT, and that Upsala and Shebandowan observe EST/EDT, all in # violation of the official Ontario rules. # # From Paul Eggert (2006-07-09): # Chris Walton (2006-07-06) mentioned an article by Stephanie MacLellan in the # 2005-07-21 Chronicle-Journal, which said: # # The clocks in Atikokan stay set on standard time year-round. # This means they spend about half the time on central time and # the other half on eastern time. # # For the most part, the system works, Mayor Dennis Brown said. # # "The majority of businesses in Atikokan deal more with Eastern # Canada, but there are some that deal with Western Canada," he # said. "I don't see any changes happening here." # # Walton also writes "Supposedly Pickle Lake and Mishkeegogamang # [New Osnaburgh] follow the same practice." # From Garry McKinnon (2006-07-14) via Chris Walton: # I chatted with a member of my board who has an outstanding memory # and a long history in Atikokan (and in the telecom industry) and he # can say for certain that Atikokan has been practicing the current # time keeping since 1952, at least. # From Paul Eggert (2006-07-17): # Shanks & Pottenger say that Atikokan has agreed with Rainy River # ever since standard time was introduced, but the information from # McKinnon sounds more authoritative. For now, assume that Atikokan # switched to EST immediately after WWII era daylight saving time # ended. This matches the old (less-populous) America/Coral_Harbour # entry since our cutoff date of 1970, so we can move # America/Coral_Harbour to the 'backward' file. # From Mark Brader (2010-03-06): # # Currently the database has: # # # Ontario # # # From Paul Eggert (2006-07-09): # # Shanks & Pottenger write that since 1970 most of Ontario has been like # # Toronto. # # Thunder Bay skipped DST in 1973. # # Many smaller locales did not observe peacetime DST until 1974; # # Nipigon (EST) and Rainy River (CST) are the largest that we know of. # # In the (Toronto) Globe and Mail for Saturday, 1955-09-24, in the bottom # right corner of page 1, it says that Toronto will return to standard # time at 2 am Sunday morning (which agrees with the database), and that: # # The one-hour setback will go into effect throughout most of Ontario, # except in areas like Windsor which remains on standard time all year. # # Windsor is, of course, a lot larger than Nipigon. # # I only came across this incidentally. I don't know if Windsor began # observing DST when Detroit did, or in 1974, or on some other date. # # By the way, the article continues by noting that: # # Some cities in the United States have pushed the deadline back # three weeks and will change over from daylight saving in October. # From Arthur David Olson (2010-07-17): # # "Standard Time and Time Zones in Canada" appeared in # The Journal of The Royal Astronomical Society of Canada, # volume 26, number 2 (February 1932) and, as of 2010-07-17, # was available at # http://adsabs.harvard.edu/full/1932JRASC..26...49S # # It includes the text below (starting on page 57): # # A list of the places in Canada using daylight saving time would # require yearly revision. From information kindly furnished by # the provincial governments and by the postmasters in many cities # and towns, it is found that the following places used daylight sav- # ing in 1930. The information for the province of Quebec is definite, # for the other provinces only approximate: # # Province Daylight saving time used # Prince Edward Island Not used. # Nova Scotia In Halifax only. # New Brunswick In St. John only. # Quebec In the following places: # Montreal Lachine # Quebec Mont-Royal # Lévis Iberville # St. Lambert Cap de la Madelèine # Verdun Loretteville # Westmount Richmond # Outremont St. Jérôme # Longueuil Greenfield Park # Arvida Waterloo # Chambly-Canton Beaulieu # Melbourne La Tuque # St. Théophile Buckingham # Ontario Used generally in the cities and towns along # the southerly part of the province. Not # used in the northwesterly part. # Manitoba Not used. # Saskatchewan In Regina only. # Alberta Not used. # British Columbia Not used. # # With some exceptions, the use of daylight saving may be said to be limited # to those cities and towns lying between Quebec city and Windsor, Ont. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Toronto 1919 only - Mar 30 23:30 1:00 D Rule Toronto 1919 only - Oct 26 0:00 0 S Rule Toronto 1920 only - May 2 2:00 1:00 D Rule Toronto 1920 only - Sep 26 0:00 0 S Rule Toronto 1921 only - May 15 2:00 1:00 D Rule Toronto 1921 only - Sep 15 2:00 0 S Rule Toronto 1922 1923 - May Sun>=8 2:00 1:00 D # Shanks & Pottenger say 1923-09-19; assume it's a typo and that "-16" # was meant. Rule Toronto 1922 1926 - Sep Sun>=15 2:00 0 S Rule Toronto 1924 1927 - May Sun>=1 2:00 1:00 D # The 1927-to-1939 rules can be expressed more simply as # Rule Toronto 1927 1937 - Sep Sun>=25 2:00 0 S # Rule Toronto 1928 1937 - Apr Sun>=25 2:00 1:00 D # Rule Toronto 1938 1940 - Apr lastSun 2:00 1:00 D # Rule Toronto 1938 1939 - Sep lastSun 2:00 0 S # The rules below avoid use of Sun>=25 # (which pre-2004 versions of zic cannot handle). Rule Toronto 1927 1932 - Sep lastSun 2:00 0 S Rule Toronto 1928 1931 - Apr lastSun 2:00 1:00 D Rule Toronto 1932 only - May 1 2:00 1:00 D Rule Toronto 1933 1940 - Apr lastSun 2:00 1:00 D Rule Toronto 1933 only - Oct 1 2:00 0 S Rule Toronto 1934 1939 - Sep lastSun 2:00 0 S Rule Toronto 1945 1946 - Sep lastSun 2:00 0 S Rule Toronto 1946 only - Apr lastSun 2:00 1:00 D Rule Toronto 1947 1949 - Apr lastSun 0:00 1:00 D Rule Toronto 1947 1948 - Sep lastSun 0:00 0 S Rule Toronto 1949 only - Nov lastSun 0:00 0 S Rule Toronto 1950 1973 - Apr lastSun 2:00 1:00 D Rule Toronto 1950 only - Nov lastSun 2:00 0 S Rule Toronto 1951 1956 - Sep lastSun 2:00 0 S # Shanks & Pottenger say Toronto ended DST a week early in 1971, # namely on 1971-10-24, but Mark Brader wrote (2003-05-31) that this # is wrong, and that he had confirmed it by checking the 1971-10-30 # Toronto Star, which said that DST was ending 1971-10-31 as usual. Rule Toronto 1957 1973 - Oct lastSun 2:00 0 S # From Paul Eggert (2003-07-27): # Willett (1914-03) writes (p. 17) "In the Cities of Fort William, and # Port Arthur, Ontario, the principle of the Bill has been in # operation for the past three years, and in the City of Moose Jaw, # Saskatchewan, for one year." # From David Bryan via Tory Tronrud, Director/Curator, # Thunder Bay Museum (2003-11-12): # There is some suggestion, however, that, by-law or not, daylight # savings time was being practiced in Fort William and Port Arthur # before 1909.... [I]n 1910, the line between the Eastern and Central # Time Zones was permanently moved about two hundred miles west to # include the Thunder Bay area.... When Canada adopted daylight # savings time in 1916, Fort William and Port Arthur, having done so # already, did not change their clocks.... During the Second World # War,... [t]he cities agreed to implement DST during the summer # months for the remainder of the war years. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Toronto -5:17:32 - LMT 1895 -5:00 Canada E%sT 1919 -5:00 Toronto E%sT 1942 Feb 9 2:00s -5:00 Canada E%sT 1946 -5:00 Toronto E%sT 1974 -5:00 Canada E%sT Zone America/Thunder_Bay -5:57:00 - LMT 1895 -6:00 - CST 1910 -5:00 - EST 1942 -5:00 Canada E%sT 1970 -5:00 Toronto E%sT 1973 -5:00 - EST 1974 -5:00 Canada E%sT Zone America/Nipigon -5:53:04 - LMT 1895 -5:00 Canada E%sT 1940 Sep 29 -5:00 1:00 EDT 1942 Feb 9 2:00s -5:00 Canada E%sT Zone America/Rainy_River -6:18:16 - LMT 1895 -6:00 Canada C%sT 1940 Sep 29 -6:00 1:00 CDT 1942 Feb 9 2:00s -6:00 Canada C%sT Zone America/Atikokan -6:06:28 - LMT 1895 -6:00 Canada C%sT 1940 Sep 29 -6:00 1:00 CDT 1942 Feb 9 2:00s -6:00 Canada C%sT 1945 Sep 30 2:00 -5:00 - EST # Manitoba # From Rob Douglas (2006-04-06): # the old Manitoba Time Act - as amended by Bill 2, assented to # March 27, 1987 ... said ... # "between two o'clock Central Standard Time in the morning of # the first Sunday of April of each year and two o'clock Central # Standard Time in the morning of the last Sunday of October next # following, one hour in advance of Central Standard Time."... # I believe that the English legislation [of the old time act] had # been assented to (March 22, 1967).... # Also, as far as I can tell, there was no order-in-council varying # the time of Daylight Saving Time for 2005 and so the provisions of # the 1987 version would apply - the changeover was at 2:00 Central # Standard Time (i.e. not until 3:00 Central Daylight Time). # From Paul Eggert (2006-04-10): # Shanks & Pottenger say Manitoba switched at 02:00 (not 02:00s) # starting 1966. Since 02:00s is clearly correct for 1967 on, assume # it was also 02:00s in 1966. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Winn 1916 only - Apr 23 0:00 1:00 D Rule Winn 1916 only - Sep 17 0:00 0 S Rule Winn 1918 only - Apr 14 2:00 1:00 D Rule Winn 1918 only - Oct 27 2:00 0 S Rule Winn 1937 only - May 16 2:00 1:00 D Rule Winn 1937 only - Sep 26 2:00 0 S Rule Winn 1942 only - Feb 9 2:00 1:00 W # War Rule Winn 1945 only - Aug 14 23:00u 1:00 P # Peace Rule Winn 1945 only - Sep lastSun 2:00 0 S Rule Winn 1946 only - May 12 2:00 1:00 D Rule Winn 1946 only - Oct 13 2:00 0 S Rule Winn 1947 1949 - Apr lastSun 2:00 1:00 D Rule Winn 1947 1949 - Sep lastSun 2:00 0 S Rule Winn 1950 only - May 1 2:00 1:00 D Rule Winn 1950 only - Sep 30 2:00 0 S Rule Winn 1951 1960 - Apr lastSun 2:00 1:00 D Rule Winn 1951 1958 - Sep lastSun 2:00 0 S Rule Winn 1959 only - Oct lastSun 2:00 0 S Rule Winn 1960 only - Sep lastSun 2:00 0 S Rule Winn 1963 only - Apr lastSun 2:00 1:00 D Rule Winn 1963 only - Sep 22 2:00 0 S Rule Winn 1966 1986 - Apr lastSun 2:00s 1:00 D Rule Winn 1966 2005 - Oct lastSun 2:00s 0 S Rule Winn 1987 2005 - Apr Sun>=1 2:00s 1:00 D # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Winnipeg -6:28:36 - LMT 1887 Jul 16 -6:00 Winn C%sT 2006 -6:00 Canada C%sT # Saskatchewan # From Mark Brader (2003-07-26): # The first actual adoption of DST in Canada was at the municipal # level. As the [Toronto] Star put it (1912-06-07), "While people # elsewhere have long been talking of legislation to save daylight, # the city of Moose Jaw [Saskatchewan] has acted on its own hook." # DST in Moose Jaw began on Saturday, 1912-06-01 (no time mentioned: # presumably late evening, as below), and would run until "the end of # the summer". The discrepancy between municipal time and railroad # time was noted. # From Paul Eggert (2003-07-27): # Willett (1914-03) notes that DST "has been in operation ... in the # City of Moose Jaw, Saskatchewan, for one year." # From Paul Eggert (2006-03-22): # Shanks & Pottenger say that since 1970 this region has mostly been as Regina. # Some western towns (e.g. Swift Current) switched from MST/MDT to CST in 1972. # Other western towns (e.g. Lloydminster) are like Edmonton. # Matthews and Vincent (1998) write that Denare Beach and Creighton # are like Winnipeg, in violation of Saskatchewan law. # From W. Jones (1992-11-06): # The. . .below is based on information I got from our law library, the # provincial archives, and the provincial Community Services department. # A precise history would require digging through newspaper archives, and # since you didn't say what you wanted, I didn't bother. # # Saskatchewan is split by a time zone meridian (105W) and over the years # the boundary became pretty ragged as communities near it reevaluated # their affiliations in one direction or the other. In 1965 a provincial # referendum favoured legislating common time practices. # # On 15 April 1966 the Time Act (c. T-14, Revised Statutes of # Saskatchewan 1978) was proclaimed, and established that the eastern # part of Saskatchewan would use CST year round, that districts in # northwest Saskatchewan would by default follow CST but could opt to # follow Mountain Time rules (thus 1 hour difference in the winter and # zero in the summer), and that districts in southwest Saskatchewan would # by default follow MT but could opt to follow CST. # # It took a few years for the dust to settle (I know one story of a town # on one time zone having its school in another, such that a mom had to # serve her family lunch in two shifts), but presently it seems that only # a few towns on the border with Alberta (e.g. Lloydminster) follow MT # rules any more; all other districts appear to have used CST year round # since sometime in the 1960s. # From Chris Walton (2006-06-26): # The Saskatchewan time act which was last updated in 1996 is about 30 pages # long and rather painful to read. # http://www.qp.gov.sk.ca/documents/English/Statutes/Statutes/T14.pdf # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Regina 1918 only - Apr 14 2:00 1:00 D Rule Regina 1918 only - Oct 27 2:00 0 S Rule Regina 1930 1934 - May Sun>=1 0:00 1:00 D Rule Regina 1930 1934 - Oct Sun>=1 0:00 0 S Rule Regina 1937 1941 - Apr Sun>=8 0:00 1:00 D Rule Regina 1937 only - Oct Sun>=8 0:00 0 S Rule Regina 1938 only - Oct Sun>=1 0:00 0 S Rule Regina 1939 1941 - Oct Sun>=8 0:00 0 S Rule Regina 1942 only - Feb 9 2:00 1:00 W # War Rule Regina 1945 only - Aug 14 23:00u 1:00 P # Peace Rule Regina 1945 only - Sep lastSun 2:00 0 S Rule Regina 1946 only - Apr Sun>=8 2:00 1:00 D Rule Regina 1946 only - Oct Sun>=8 2:00 0 S Rule Regina 1947 1957 - Apr lastSun 2:00 1:00 D Rule Regina 1947 1957 - Sep lastSun 2:00 0 S Rule Regina 1959 only - Apr lastSun 2:00 1:00 D Rule Regina 1959 only - Oct lastSun 2:00 0 S # Rule Swift 1957 only - Apr lastSun 2:00 1:00 D Rule Swift 1957 only - Oct lastSun 2:00 0 S Rule Swift 1959 1961 - Apr lastSun 2:00 1:00 D Rule Swift 1959 only - Oct lastSun 2:00 0 S Rule Swift 1960 1961 - Sep lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Regina -6:58:36 - LMT 1905 Sep -7:00 Regina M%sT 1960 Apr lastSun 2:00 -6:00 - CST Zone America/Swift_Current -7:11:20 - LMT 1905 Sep -7:00 Canada M%sT 1946 Apr lastSun 2:00 -7:00 Regina M%sT 1950 -7:00 Swift M%sT 1972 Apr lastSun 2:00 -6:00 - CST # Alberta # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Edm 1918 1919 - Apr Sun>=8 2:00 1:00 D Rule Edm 1918 only - Oct 27 2:00 0 S Rule Edm 1919 only - May 27 2:00 0 S Rule Edm 1920 1923 - Apr lastSun 2:00 1:00 D Rule Edm 1920 only - Oct lastSun 2:00 0 S Rule Edm 1921 1923 - Sep lastSun 2:00 0 S Rule Edm 1942 only - Feb 9 2:00 1:00 W # War Rule Edm 1945 only - Aug 14 23:00u 1:00 P # Peace Rule Edm 1945 only - Sep lastSun 2:00 0 S Rule Edm 1947 only - Apr lastSun 2:00 1:00 D Rule Edm 1947 only - Sep lastSun 2:00 0 S Rule Edm 1967 only - Apr lastSun 2:00 1:00 D Rule Edm 1967 only - Oct lastSun 2:00 0 S Rule Edm 1969 only - Apr lastSun 2:00 1:00 D Rule Edm 1969 only - Oct lastSun 2:00 0 S Rule Edm 1972 1986 - Apr lastSun 2:00 1:00 D Rule Edm 1972 2006 - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Edmonton -7:33:52 - LMT 1906 Sep -7:00 Edm M%sT 1987 -7:00 Canada M%sT # British Columbia # From Paul Eggert (2006-03-22): # Shanks & Pottenger write that since 1970 most of this region has # been like Vancouver. # Dawson Creek uses MST. Much of east BC is like Edmonton. # Matthews and Vincent (1998) write that Creston is like Dawson Creek. # It seems though that (re: Creston) is not entirely correct: # From Chris Walton (2011-12-01): # There are two areas within the Canadian province of British Columbia # that do not currently observe daylight saving: # a) The Creston Valley (includes the town of Creston and surrounding area) # b) The eastern half of the Peace River Regional District # (includes the cities of Dawson Creek and Fort St. John) # Earlier this year I stumbled across a detailed article about the time # keeping history of Creston; it was written by Tammy Hardwick who is the # manager of the Creston & District Museum. The article was written in May 2009. # http://www.ilovecreston.com/?p=articles&t=spec&ar=260 # According to the article, Creston has not changed its clocks since June 1918. # i.e. Creston has been stuck on UTC-7 for 93 years. # Dawson Creek, on the other hand, changed its clocks as recently as April 1972. # Unfortunately the exact date for the time change in June 1918 remains # unknown and will be difficult to ascertain. I e-mailed Tammy a few months # ago to ask if Sunday June 2 was a reasonable guess. She said it was just # as plausible as any other date (in June). She also said that after writing # the article she had discovered another time change in 1916; this is the # subject of another article which she wrote in October 2010. # http://www.creston.museum.bc.ca/index.php?module=comments&uop=view_comment&cm+id=56 # Here is a summary of the three clock change events in Creston's history: # 1. 1884 or 1885: adoption of Mountain Standard Time (GMT-7) # Exact date unknown # 2. Oct 1916: switch to Pacific Standard Time (GMT-8) # Exact date in October unknown; Sunday October 1 is a reasonable guess. # 3. June 1918: switch to Pacific Daylight Time (GMT-7) # Exact date in June unknown; Sunday June 2 is a reasonable guess. # note#1: # On Oct 27/1918 when daylight saving ended in the rest of Canada, # Creston did not change its clocks. # note#2: # During WWII when the Federal Government legislated a mandatory clock change, # Creston did not oblige. # note#3: # There is no guarantee that Creston will remain on Mountain Standard Time # (UTC-7) forever. # The subject was debated at least once this year by the town Council. # http://www.bclocalnews.com/kootenay_rockies/crestonvalleyadvance/news/116760809.html # During a period WWII, summer time (Daylight saying) was mandatory in Canada. # In Creston, that was handled by shifting the area to PST (-8:00) then applying # summer time to cause the offset to be -7:00, the same as it had been before # the change. It can be argued that the timezone abbreviation during this # period should be PDT rather than MST, but that doesn't seem important enough # (to anyone) to further complicate the rules. # The transition dates (and times) are guesses. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Vanc 1918 only - Apr 14 2:00 1:00 D Rule Vanc 1918 only - Oct 27 2:00 0 S Rule Vanc 1942 only - Feb 9 2:00 1:00 W # War Rule Vanc 1945 only - Aug 14 23:00u 1:00 P # Peace Rule Vanc 1945 only - Sep 30 2:00 0 S Rule Vanc 1946 1986 - Apr lastSun 2:00 1:00 D Rule Vanc 1946 only - Oct 13 2:00 0 S Rule Vanc 1947 1961 - Sep lastSun 2:00 0 S Rule Vanc 1962 2006 - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Vancouver -8:12:28 - LMT 1884 -8:00 Vanc P%sT 1987 -8:00 Canada P%sT Zone America/Dawson_Creek -8:00:56 - LMT 1884 -8:00 Canada P%sT 1947 -8:00 Vanc P%sT 1972 Aug 30 2:00 -7:00 - MST Zone America/Creston -7:46:04 - LMT 1884 -7:00 - MST 1916 Oct 1 -8:00 - PST 1918 Jun 2 -7:00 - MST # Northwest Territories, Nunavut, Yukon # From Paul Eggert (2006-03-22): # Dawson switched to PST in 1973. Inuvik switched to MST in 1979. # Mathew Englander (1996-10-07) gives the following refs: # * 1967. Paragraph 28(34)(g) of the Interpretation Act, S.C. 1967-68, # c. 7 defines Yukon standard time as UTC-9. This is still valid; # see Interpretation Act, R.S.C. 1985, c. I-21, s. 35(1). # * C.O. 1973/214 switched Yukon to PST on 1973-10-28 00:00. # * O.I.C. 1980/02 established DST. # * O.I.C. 1987/056 changed DST to Apr firstSun 2:00 to Oct lastSun 2:00. # Shanks & Pottenger say Yukon's 1973-10-28 switch was at 2:00; go # with Englander. # From Chris Walton (2006-06-26): # Here is a link to the old daylight saving portion of the interpretation # act which was last updated in 1987: # http://www.gov.yk.ca/legislation/regs/oic1987_056.pdf # From Rives McDow (1999-09-04): # Nunavut ... moved ... to incorporate the whole territory into one time zone. # Nunavut moves to single time zone Oct. 31 # http://www.nunatsiaq.com/nunavut/nvt90903_13.html # # From Antoine Leca (1999-09-06): # We then need to create a new timezone for the Kitikmeot region of Nunavut # to differentiate it from the Yellowknife region. # From Paul Eggert (1999-09-20): # Basic Facts: The New Territory # http://www.nunavut.com/basicfacts/english/basicfacts_1territory.html # (1999) reports that Pangnirtung operates on eastern time, # and that Coral Harbour does not observe DST. We don't know when # Pangnirtung switched to eastern time; we'll guess 1995. # From Rives McDow (1999-11-08): # On October 31, when the rest of Nunavut went to Central time, # Pangnirtung wobbled. Here is the result of their wobble: # # The following businesses and organizations in Pangnirtung use Central Time: # # First Air, Power Corp, Nunavut Construction, Health Center, RCMP, # Eastern Arctic National Parks, A & D Specialist # # The following businesses and organizations in Pangnirtung use Eastern Time: # # Hamlet office, All other businesses, Both schools, Airport operator # # This has made for an interesting situation there, which warranted the news. # No one there that I spoke with seems concerned, or has plans to # change the local methods of keeping time, as it evidently does not # really interfere with any activities or make things difficult locally. # They plan to celebrate New Year's turn-over twice, one hour apart, # so it appears that the situation will last at least that long. # The Nunavut Intergovernmental Affairs hopes that they will "come to # their senses", but the locals evidently don't see any problem with # the current state of affairs. # From Michaela Rodrigue, writing in the # Nunatsiaq News (1999-11-19): # http://www.nunatsiaq.com/archives/nunavut991130/nvt91119_17.html # Clyde River, Pangnirtung and Sanikiluaq now operate with two time zones, # central - or Nunavut time - for government offices, and eastern time # for municipal offices and schools.... Igloolik [was similar but then] # made the switch to central time on Saturday, Nov. 6. # From Paul Eggert (2000-10-02): # Matthews and Vincent (1998) say the following, but we lack histories # for these potential new Zones. # # The Canadian Forces station at Alert uses Eastern Time while the # handful of residents at the Eureka weather station [in the Central # zone] skip daylight savings. Baffin Island, which is crossed by the # Central, Eastern and Atlantic Time zones only uses Eastern Time. # Gjoa Haven, Taloyoak and Pelly Bay all use Mountain instead of # Central Time and Southampton Island [in the Central zone] is not # required to use daylight savings. # From # Nunavut now has two time zones (2000-11-10): # The Nunavut government would allow its employees in Kugluktuk and # Cambridge Bay to operate on central time year-round, putting them # one hour behind the rest of Nunavut for six months during the winter. # At the end of October the two communities had rebelled against # Nunavut's unified time zone, refusing to shift to eastern time with # the rest of the territory for the winter. Cambridge Bay remained on # central time, while Kugluktuk, even farther west, reverted to # mountain time, which they had used before the advent of Nunavut's # unified time zone in 1999. # # From Rives McDow (2001-01-20), quoting the Nunavut government: # The preceding decision came into effect at midnight, Saturday Nov 4, 2000. # From Paul Eggert (2000-12-04): # Let's just keep track of the official times for now. # From Rives McDow (2001-03-07): # The premier of Nunavut has issued a ministerial statement advising # that effective 2001-04-01, the territory of Nunavut will revert # back to three time zones (mountain, central, and eastern). Of the # cities in Nunavut, Coral Harbor is the only one that I know of that # has said it will not observe dst, staying on EST year round. I'm # checking for more info, and will get back to you if I come up with # more. # [Also see (2001-03-09).] # From Gwillim Law (2005-05-21): # According to ... # http://www.canadiangeographic.ca/Magazine/SO98/geomap.asp # (from a 1998 Canadian Geographic article), the de facto and de jure time # for Southampton Island (at the north end of Hudson Bay) is UTC-5 all year # round. Using Google, it's easy to find other websites that confirm this. # I wasn't able to find how far back this time regimen goes, but since it # predates the creation of Nunavut, it probably goes back many years.... # The Inuktitut name of Coral Harbour is Sallit, but it's rarely used. # # From Paul Eggert (2014-10-17): # For lack of better information, assume that Southampton Island observed # daylight saving only during wartime. Gwillim Law's email also # mentioned maps now maintained by National Research Council Canada; # see above for an up-to-date link. # From Chris Walton (2007-03-01): # ... the community of Resolute (located on Cornwallis Island in # Nunavut) moved from Central Time to Eastern Time last November. # Basically the community did not change its clocks at the end of # daylight saving.... # http://www.nnsl.com/frames/newspapers/2006-11/nov13_06none.html # From Chris Walton (2011-03-21): # Back in 2007 I initiated the creation of a new "zone file" for Resolute # Bay. Resolute Bay is a small community located about 900km north of # the Arctic Circle. The zone file was required because Resolute Bay had # decided to use UTC-5 instead of UTC-6 for the winter of 2006-2007. # # According to new information which I received last week, Resolute Bay # went back to using UTC-6 in the winter of 2007-2008... # # On March 11/2007 most of Canada went onto daylight saving. On March # 14/2007 I phoned the Resolute Bay hamlet office to do a "time check." I # talked to somebody that was both knowledgeable and helpful. I was able # to confirm that Resolute Bay was still operating on UTC-5. It was # explained to me that Resolute Bay had been on the Eastern Time zone # (EST) in the winter, and was now back on the Central Time zone (CDT). # i.e. the time zone had changed twice in the last year but the clocks # had not moved. The residents had to know which time zone they were in # so they could follow the correct TV schedule... # # On Nov 02/2008 most of Canada went onto standard time. On Nov 03/2008 I # phoned the Resolute Bay hamlet office...[D]ue to the challenging nature # of the phone call, I decided to seek out an alternate source of # information. I found an e-mail address for somebody by the name of # Stephanie Adams whose job was listed as "Inns North Support Officer for # Arctic Co-operatives." I was under the impression that Stephanie lived # and worked in Resolute Bay... # # On March 14/2011 I phoned the hamlet office again. I was told that # Resolute Bay had been using Central Standard Time over the winter of # 2010-2011 and that the clocks had therefore been moved one hour ahead # on March 13/2011. The person I talked to was aware that Resolute Bay # had previously experimented with Eastern Standard Time but he could not # tell me when the practice had stopped. # # On March 17/2011 I searched the Web to find an e-mail address of # somebody that might be able to tell me exactly when Resolute Bay went # off Eastern Standard Time. I stumbled on the name "Aziz Kheraj." Aziz # used to be the mayor of Resolute Bay and he apparently owns half the # businesses including "South Camp Inn." This website has some info on # Aziz: # http://www.uphere.ca/node/493 # # I sent Aziz an e-mail asking when Resolute Bay had stopped using # Eastern Standard Time. # # Aziz responded quickly with this: "hi, The time was not changed for the # 1 year only, the following year, the community went back to the old way # of "spring ahead-fall behind" currently we are zulu plus 5 hrs and in # the winter Zulu plus 6 hrs" # # This of course conflicted with everything I had ascertained in November 2008. # # I sent Aziz a copy of my 2008 e-mail exchange with Stephanie. Aziz # responded with this: "Hi, Stephanie lives in Winnipeg. I live here, You # may want to check with the weather office in Resolute Bay or do a # search on the weather through Env. Canada. web site" # # If I had realized the Stephanie did not live in Resolute Bay I would # never have contacted her. I now believe that all the information I # obtained in November 2008 should be ignored... # I apologize for reporting incorrect information in 2008. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule NT_YK 1918 only - Apr 14 2:00 1:00 D Rule NT_YK 1918 only - Oct 27 2:00 0 S Rule NT_YK 1919 only - May 25 2:00 1:00 D Rule NT_YK 1919 only - Nov 1 0:00 0 S Rule NT_YK 1942 only - Feb 9 2:00 1:00 W # War Rule NT_YK 1945 only - Aug 14 23:00u 1:00 P # Peace Rule NT_YK 1945 only - Sep 30 2:00 0 S Rule NT_YK 1965 only - Apr lastSun 0:00 2:00 DD Rule NT_YK 1965 only - Oct lastSun 2:00 0 S Rule NT_YK 1980 1986 - Apr lastSun 2:00 1:00 D Rule NT_YK 1980 2006 - Oct lastSun 2:00 0 S Rule NT_YK 1987 2006 - Apr Sun>=1 2:00 1:00 D # Zone NAME GMTOFF RULES FORMAT [UNTIL] # aka Panniqtuuq Zone America/Pangnirtung 0 - zzz 1921 # trading post est. -4:00 NT_YK A%sT 1995 Apr Sun>=1 2:00 -5:00 Canada E%sT 1999 Oct 31 2:00 -6:00 Canada C%sT 2000 Oct 29 2:00 -5:00 Canada E%sT # formerly Frobisher Bay Zone America/Iqaluit 0 - zzz 1942 Aug # Frobisher Bay est. -5:00 NT_YK E%sT 1999 Oct 31 2:00 -6:00 Canada C%sT 2000 Oct 29 2:00 -5:00 Canada E%sT # aka Qausuittuq Zone America/Resolute 0 - zzz 1947 Aug 31 # Resolute founded -6:00 NT_YK C%sT 2000 Oct 29 2:00 -5:00 - EST 2001 Apr 1 3:00 -6:00 Canada C%sT 2006 Oct 29 2:00 -5:00 - EST 2007 Mar 11 3:00 -6:00 Canada C%sT # aka Kangiqiniq Zone America/Rankin_Inlet 0 - zzz 1957 # Rankin Inlet founded -6:00 NT_YK C%sT 2000 Oct 29 2:00 -5:00 - EST 2001 Apr 1 3:00 -6:00 Canada C%sT # aka Iqaluktuuttiaq Zone America/Cambridge_Bay 0 - zzz 1920 # trading post est.? -7:00 NT_YK M%sT 1999 Oct 31 2:00 -6:00 Canada C%sT 2000 Oct 29 2:00 -5:00 - EST 2000 Nov 5 0:00 -6:00 - CST 2001 Apr 1 3:00 -7:00 Canada M%sT Zone America/Yellowknife 0 - zzz 1935 # Yellowknife founded? -7:00 NT_YK M%sT 1980 -7:00 Canada M%sT Zone America/Inuvik 0 - zzz 1953 # Inuvik founded -8:00 NT_YK P%sT 1979 Apr lastSun 2:00 -7:00 NT_YK M%sT 1980 -7:00 Canada M%sT Zone America/Whitehorse -9:00:12 - LMT 1900 Aug 20 -9:00 NT_YK Y%sT 1966 Jul 1 2:00 -8:00 NT_YK P%sT 1980 -8:00 Canada P%sT Zone America/Dawson -9:17:40 - LMT 1900 Aug 20 -9:00 NT_YK Y%sT 1973 Oct 28 0:00 -8:00 NT_YK P%sT 1980 -8:00 Canada P%sT ############################################################################### # Mexico -# From Paul Eggert (2001-03-05): +# From Paul Eggert (2014-12-07): # The Investigation and Analysis Service of the # Mexican Library of Congress (MLoC) has published a # history of Mexican local time (in Spanish) -# http://www.cddhcu.gob.mx/bibliot/publica/inveyana/polisoc/horver/ +# http://www.diputados.gob.mx/bibliot/publica/inveyana/polisoc/horver/index.htm # # Here are the discrepancies between Shanks & Pottenger (S&P) and the MLoC. # (In all cases we go with the MLoC.) # S&P report that Baja was at -8:00 in 1922/1923. # S&P say the 1930 transition in Baja was 1930-11-16. # S&P report no DST during summer 1931. # S&P report a transition at 1932-03-30 23:00, not 1932-04-01. # From Gwillim Law (2001-02-20): # There are some other discrepancies between the Decrees page and the # tz database. I think they can best be explained by supposing that # the researchers who prepared the Decrees page failed to find some of # the relevant documents. # From Alan Perry (1996-02-15): # A guy from our Mexico subsidiary finally found the Presidential Decree # outlining the timezone changes in Mexico. # # ------------- Begin Forwarded Message ------------- # # I finally got my hands on the Official Presidential Decree that sets up the # rules for the DST changes. The rules are: # # 1. The country is divided in 3 timezones: # - Baja California Norte (the Mexico/BajaNorte TZ) # - Baja California Sur, Nayarit, Sinaloa and Sonora (the Mexico/BajaSur TZ) # - The rest of the country (the Mexico/General TZ) # # 2. From the first Sunday in April at 2:00 AM to the last Sunday in October # at 2:00 AM, the times in each zone are as follows: # BajaNorte: GMT+7 # BajaSur: GMT+6 # General: GMT+5 # # 3. The rest of the year, the times are as follows: # BajaNorte: GMT+8 # BajaSur: GMT+7 # General: GMT+6 # # The Decree was published in Mexico's Official Newspaper on January 4th. # # -------------- End Forwarded Message -------------- # From Paul Eggert (1996-06-12): # For an English translation of the decree, see # "Diario Oficial: Time Zone Changeover" (1996-01-04). # http://mexico-travel.com/extra/timezone_eng.html # From Rives McDow (1998-10-08): # The State of Quintana Roo has reverted back to central STD and DST times # (i.e. UTC -0600 and -0500 as of 1998-08-02). # From Rives McDow (2000-01-10): # Effective April 4, 1999 at 2:00 AM local time, Sonora changed to the time # zone 5 hours from the International Date Line, and will not observe daylight # savings time so as to stay on the same time zone as the southern part of # Arizona year round. # From Jesper Nørgaard, translating # (2001-01-17): # In Oaxaca, the 55.000 teachers from the Section 22 of the National # Syndicate of Education Workers, refuse to apply daylight saving each # year, so that the more than 10,000 schools work at normal hour the # whole year. # From Gwillim Law (2001-01-19): # ... says # (translated):... # January 17, 2000 - The Energy Secretary, Ernesto Martens, announced # that Summer Time will be reduced from seven to five months, starting # this year.... # http://www.publico.com.mx/scripts/texto3.asp?action=pagina&pag=21&pos=p&secc=naci&date=01/17/2001 # [translated], says "summer time will ... take effect on the first Sunday # in May, and end on the last Sunday of September. # From Arthur David Olson (2001-01-25): # The 2001-01-24 traditional Washington Post contained the page one # story "Timely Issue Divides Mexicans."... # http://www.washingtonpost.com/wp-dyn/articles/A37383-2001Jan23.html # ... Mexico City Mayor López Obrador "...is threatening to keep # Mexico City and its 20 million residents on a different time than # the rest of the country..." In particular, López Obrador would abolish # observation of Daylight Saving Time. # Official statute published by the Energy Department # http://www.conae.gob.mx/ahorro/decretohorver2001.html#decre # (2001-02-01) shows Baja and Chihauhua as still using US DST rules, # and Sonora with no DST. This was reported by Jesper Nørgaard (2001-02-03). # From Paul Eggert (2001-03-03): # # http://www.latimes.com/news/nation/20010303/t000018766.html # James F. Smith writes in today's LA Times # * Sonora will continue to observe standard time. # * Last week Mexico City's mayor Andrés Manuel López Obrador decreed that # the Federal District will not adopt DST. # * 4 of 16 district leaders announced they'll ignore the decree. # * The decree does not affect federal-controlled facilities including # the airport, banks, hospitals, and schools. # # For now we'll assume that the Federal District will bow to federal rules. # From Jesper Nørgaard (2001-04-01): # I found some references to the Mexican application of daylight # saving, which modifies what I had already sent you, stating earlier # that a number of northern Mexican states would go on daylight # saving. The modification reverts this to only cover Baja California # (Norte), while all other states (except Sonora, who has no daylight # saving all year) will follow the original decree of president # Vicente Fox, starting daylight saving May 6, 2001 and ending # September 30, 2001. # References: "Diario de Monterrey" # Palabra (2001-03-31) # From Reuters (2001-09-04): # Mexico's Supreme Court on Tuesday declared that daylight savings was # unconstitutional in Mexico City, creating the possibility the # capital will be in a different time zone from the rest of the nation # next year.... The Supreme Court's ruling takes effect at 2:00 # a.m. (0800 GMT) on Sept. 30, when Mexico is scheduled to revert to # standard time. "This is so residents of the Federal District are not # subject to unexpected time changes," a statement from the court said. # From Jesper Nørgaard Welen (2002-03-12): # ... consulting my local grocery store(!) and my coworkers, they all insisted # that a new decision had been made to reinstate US style DST in Mexico.... # http://www.conae.gob.mx/ahorro/horaver2001_m1_2002.html (2002-02-20) # confirms this. Sonora as usual is the only state where DST is not applied. # From Steffen Thorsen (2009-12-28): # # Steffen Thorsen wrote: # > Mexico's House of Representatives has approved a proposal for northern # > Mexico's border cities to share the same daylight saving schedule as # > the United States. # Now this has passed both the Congress and the Senate, so starting from # 2010, some border regions will be the same: # http://www.signonsandiego.com/news/2009/dec/28/clocks-will-match-both-sides-border/ # http://www.elmananarey.com/diario/noticia/nacional/noticias/empatan_horario_de_frontera_con_eu/621939 # (Spanish) # # Could not find the new law text, but the proposed law text changes are here: # http://gaceta.diputados.gob.mx/Gaceta/61/2009/dic/20091210-V.pdf # (Gaceta Parlamentaria) # # There is also a list of the votes here: # http://gaceta.diputados.gob.mx/Gaceta/61/2009/dic/V2-101209.html # # Our page: # http://www.timeanddate.com/news/time/north-mexico-dst-change.html # From Arthur David Olson (2010-01-20): # The page # http://dof.gob.mx/nota_detalle.php?codigo=5127480&fecha=06/01/2010 # includes this text: # En los municipios fronterizos de Tijuana y Mexicali en Baja California; # Juárez y Ojinaga en Chihuahua; Acuña y Piedras Negras en Coahuila; # Anáhuac en Nuevo León; y Nuevo Laredo, Reynosa y Matamoros en # Tamaulipas, la aplicación de este horario estacional surtirá efecto # desde las dos horas del segundo domingo de marzo y concluirá a las dos # horas del primer domingo de noviembre. # En los municipios fronterizos que se encuentren ubicados en la franja # fronteriza norte en el territorio comprendido entre la línea # internacional y la línea paralela ubicada a una distancia de veinte # kilómetros, así como la Ciudad de Ensenada, Baja California, hacia el # interior del país, la aplicación de este horario estacional surtirá # efecto desde las dos horas del segundo domingo de marzo y concluirá a # las dos horas del primer domingo de noviembre. +# From Steffen Thorsen (2014-12-08), translated by Gwillim Law: +# The Mexican state of Quintana Roo will likely change to EST in 2015. +# +# http://www.unioncancun.mx/articulo/2014/12/04/medio-ambiente/congreso-aprueba-una-hora-mas-de-sol-en-qroo +# "With this change, the time conflict that has existed between the municipios +# of Quintana Roo and the municipio of Felipe Carrillo Puerto may come to an +# end. The latter declared itself in rebellion 15 years ago when a time change +# was initiated in Mexico, and since then it has refused to change its time +# zone along with the rest of the country." +# +# From Steffen Thorsen (2015-01-14), translated by Gwillim Law: +# http://sipse.com/novedades/confirman-aplicacion-de-nueva-zona-horaria-para-quintana-roo-132331.html +# "...the new time zone will come into effect at two o'clock on the first Sunday +# of February, when we will have to advance the clock one hour from its current +# time..." +# +# Also, the new zone will not use DST. + # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Mexico 1939 only - Feb 5 0:00 1:00 D Rule Mexico 1939 only - Jun 25 0:00 0 S Rule Mexico 1940 only - Dec 9 0:00 1:00 D Rule Mexico 1941 only - Apr 1 0:00 0 S Rule Mexico 1943 only - Dec 16 0:00 1:00 W # War Rule Mexico 1944 only - May 1 0:00 0 S Rule Mexico 1950 only - Feb 12 0:00 1:00 D Rule Mexico 1950 only - Jul 30 0:00 0 S Rule Mexico 1996 2000 - Apr Sun>=1 2:00 1:00 D Rule Mexico 1996 2000 - Oct lastSun 2:00 0 S Rule Mexico 2001 only - May Sun>=1 2:00 1:00 D Rule Mexico 2001 only - Sep lastSun 2:00 0 S Rule Mexico 2002 max - Apr Sun>=1 2:00 1:00 D Rule Mexico 2002 max - Oct lastSun 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] # Quintana Roo; represented by Cancún Zone America/Cancun -5:47:04 - LMT 1922 Jan 1 0:12:56 -6:00 - CST 1981 Dec 23 -5:00 Mexico E%sT 1998 Aug 2 2:00 - -6:00 Mexico C%sT + -6:00 Mexico C%sT 2015 Feb 1 2:00 + -5:00 - EST # Campeche, Yucatán; represented by Mérida Zone America/Merida -5:58:28 - LMT 1922 Jan 1 0:01:32 -6:00 - CST 1981 Dec 23 -5:00 - EST 1982 Dec 2 -6:00 Mexico C%sT # Coahuila, Durango, Nuevo León, Tamaulipas (near US border) Zone America/Matamoros -6:40:00 - LMT 1921 Dec 31 23:20:00 -6:00 - CST 1988 -6:00 US C%sT 1989 -6:00 Mexico C%sT 2010 -6:00 US C%sT # Coahuila, Durango, Nuevo León, Tamaulipas (away from US border) Zone America/Monterrey -6:41:16 - LMT 1921 Dec 31 23:18:44 -6:00 - CST 1988 -6:00 US C%sT 1989 -6:00 Mexico C%sT # Central Mexico Zone America/Mexico_City -6:36:36 - LMT 1922 Jan 1 0:23:24 -7:00 - MST 1927 Jun 10 23:00 -6:00 - CST 1930 Nov 15 -7:00 - MST 1931 May 1 23:00 -6:00 - CST 1931 Oct -7:00 - MST 1932 Apr 1 -6:00 Mexico C%sT 2001 Sep 30 2:00 -6:00 - CST 2002 Feb 20 -6:00 Mexico C%sT # Chihuahua (near US border) Zone America/Ojinaga -6:57:40 - LMT 1922 Jan 1 0:02:20 -7:00 - MST 1927 Jun 10 23:00 -6:00 - CST 1930 Nov 15 -7:00 - MST 1931 May 1 23:00 -6:00 - CST 1931 Oct -7:00 - MST 1932 Apr 1 -6:00 - CST 1996 -6:00 Mexico C%sT 1998 -6:00 - CST 1998 Apr Sun>=1 3:00 -7:00 Mexico M%sT 2010 -7:00 US M%sT # Chihuahua (away from US border) Zone America/Chihuahua -7:04:20 - LMT 1921 Dec 31 23:55:40 -7:00 - MST 1927 Jun 10 23:00 -6:00 - CST 1930 Nov 15 -7:00 - MST 1931 May 1 23:00 -6:00 - CST 1931 Oct -7:00 - MST 1932 Apr 1 -6:00 - CST 1996 -6:00 Mexico C%sT 1998 -6:00 - CST 1998 Apr Sun>=1 3:00 -7:00 Mexico M%sT # Sonora Zone America/Hermosillo -7:23:52 - LMT 1921 Dec 31 23:36:08 -7:00 - MST 1927 Jun 10 23:00 -6:00 - CST 1930 Nov 15 -7:00 - MST 1931 May 1 23:00 -6:00 - CST 1931 Oct -7:00 - MST 1932 Apr 1 -6:00 - CST 1942 Apr 24 -7:00 - MST 1949 Jan 14 -8:00 - PST 1970 -7:00 Mexico M%sT 1999 -7:00 - MST # From Alexander Krivenyshev (2010-04-21): # According to news, Bahía de Banderas (Mexican state of Nayarit) # changed time zone UTC-7 to new time zone UTC-6 on April 4, 2010 (to # share the same time zone as nearby city Puerto Vallarta, Jalisco). # # (Spanish) # Bahía de Banderas homologa su horario al del centro del # país, a partir de este domingo # http://www.nayarit.gob.mx/notes.asp?id=20748 # # Bahía de Banderas homologa su horario con el del Centro del # País # http://www.bahiadebanderas.gob.mx/principal/index.php?option=com_content&view=article&id=261:bahia-de-banderas-homologa-su-horario-con-el-del-centro-del-pais&catid=42:comunicacion-social&Itemid=50 # # (English) # Puerto Vallarta and Bahía de Banderas: One Time Zone # http://virtualvallarta.com/puertovallarta/puertovallarta/localnews/2009-12-03-Puerto-Vallarta-and-Bahia-de-Banderas-One-Time-Zone.shtml # http://www.worldtimezone.com/dst_news/dst_news_mexico08.html # # "Mexico's Senate approved the amendments to the Mexican Schedule System that # will allow Bahía de Banderas and Puerto Vallarta to share the same time # zone ..." # Baja California Sur, Nayarit, Sinaloa # From Arthur David Olson (2010-05-01): # Use "Bahia_Banderas" to keep the name to fourteen characters. # Mazatlán Zone America/Mazatlan -7:05:40 - LMT 1921 Dec 31 23:54:20 -7:00 - MST 1927 Jun 10 23:00 -6:00 - CST 1930 Nov 15 -7:00 - MST 1931 May 1 23:00 -6:00 - CST 1931 Oct -7:00 - MST 1932 Apr 1 -6:00 - CST 1942 Apr 24 -7:00 - MST 1949 Jan 14 -8:00 - PST 1970 -7:00 Mexico M%sT # Bahía de Banderas Zone America/Bahia_Banderas -7:01:00 - LMT 1921 Dec 31 23:59:00 -7:00 - MST 1927 Jun 10 23:00 -6:00 - CST 1930 Nov 15 -7:00 - MST 1931 May 1 23:00 -6:00 - CST 1931 Oct -7:00 - MST 1932 Apr 1 -6:00 - CST 1942 Apr 24 -7:00 - MST 1949 Jan 14 -8:00 - PST 1970 -7:00 Mexico M%sT 2010 Apr 4 2:00 -6:00 Mexico C%sT # Baja California (near US border) Zone America/Tijuana -7:48:04 - LMT 1922 Jan 1 0:11:56 -7:00 - MST 1924 -8:00 - PST 1927 Jun 10 23:00 -7:00 - MST 1930 Nov 15 -8:00 - PST 1931 Apr 1 -8:00 1:00 PDT 1931 Sep 30 -8:00 - PST 1942 Apr 24 -8:00 1:00 PWT 1945 Aug 14 23:00u -8:00 1:00 PPT 1945 Nov 12 # Peace -8:00 - PST 1948 Apr 5 -8:00 1:00 PDT 1949 Jan 14 -8:00 - PST 1954 -8:00 CA P%sT 1961 -8:00 - PST 1976 -8:00 US P%sT 1996 -8:00 Mexico P%sT 2001 -8:00 US P%sT 2002 Feb 20 -8:00 Mexico P%sT 2010 -8:00 US P%sT # Baja California (away from US border) Zone America/Santa_Isabel -7:39:28 - LMT 1922 Jan 1 0:20:32 -7:00 - MST 1924 -8:00 - PST 1927 Jun 10 23:00 -7:00 - MST 1930 Nov 15 -8:00 - PST 1931 Apr 1 -8:00 1:00 PDT 1931 Sep 30 -8:00 - PST 1942 Apr 24 -8:00 1:00 PWT 1945 Aug 14 23:00u -8:00 1:00 PPT 1945 Nov 12 # Peace -8:00 - PST 1948 Apr 5 -8:00 1:00 PDT 1949 Jan 14 -8:00 - PST 1954 -8:00 CA P%sT 1961 -8:00 - PST 1976 -8:00 US P%sT 1996 -8:00 Mexico P%sT 2001 -8:00 US P%sT 2002 Feb 20 -8:00 Mexico P%sT # From Paul Eggert (2006-03-22): # Formerly there was an America/Ensenada zone, which differed from # America/Tijuana only in that it did not observe DST from 1976 # through 1995. This was as per Shanks (1999). But Shanks & Pottenger say # Ensenada did not observe DST from 1948 through 1975. Guy Harris reports # that the 1987 OAG says "Only Ensenada, Mexicali, San Felipe and # Tijuana observe DST," which agrees with Shanks & Pottenger but implies that # DST-observance was a town-by-town matter back then. This concerns # data after 1970 so most likely there should be at least one Zone # other than America/Tijuana for Baja, but it's not clear yet what its # name or contents should be. # # Revillagigedo Is # no information ############################################################################### # Anguilla # See America/Port_of_Spain. # Antigua and Barbuda # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Antigua -4:07:12 - LMT 1912 Mar 2 -5:00 - EST 1951 -4:00 - AST # Bahamas # # For 1899 Milne gives -5:09:29.5; round that. # # From Sue Williams (2006-12-07): # The Bahamas announced about a month ago that they plan to change their DST # rules to sync with the U.S. starting in 2007.... # http://www.jonesbahamas.com/?c=45&a=10412 # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Bahamas 1964 1975 - Oct lastSun 2:00 0 S Rule Bahamas 1964 1975 - Apr lastSun 2:00 1:00 D # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Nassau -5:09:30 - LMT 1912 Mar 2 -5:00 Bahamas E%sT 1976 -5:00 US E%sT # Barbados # For 1899 Milne gives -3:58:29.2; round that. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Barb 1977 only - Jun 12 2:00 1:00 D Rule Barb 1977 1978 - Oct Sun>=1 2:00 0 S Rule Barb 1978 1980 - Apr Sun>=15 2:00 1:00 D Rule Barb 1979 only - Sep 30 2:00 0 S Rule Barb 1980 only - Sep 25 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Barbados -3:58:29 - LMT 1924 # Bridgetown -3:58:29 - BMT 1932 # Bridgetown Mean Time -4:00 Barb A%sT # Belize # Whitman entirely disagrees with Shanks; go with Shanks & Pottenger. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Belize 1918 1942 - Oct Sun>=2 0:00 0:30 HD Rule Belize 1919 1943 - Feb Sun>=9 0:00 0 S Rule Belize 1973 only - Dec 5 0:00 1:00 D Rule Belize 1974 only - Feb 9 0:00 0 S Rule Belize 1982 only - Dec 18 0:00 1:00 D Rule Belize 1983 only - Feb 12 0:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Belize -5:52:48 - LMT 1912 Apr -6:00 Belize C%sT # Bermuda # For 1899 Milne gives -4:19:18.3 as the meridian of the clock tower, # Bermuda dockyard, Ireland I; round that. # From Dan Jones, reporting in The Royal Gazette (2006-06-26): # Next year, however, clocks in the US will go forward on the second Sunday # in March, until the first Sunday in November. And, after the Time Zone # (Seasonal Variation) Bill 2006 was passed in the House of Assembly on # Friday, the same thing will happen in Bermuda. # http://www.theroyalgazette.com/apps/pbcs.dll/article?AID=/20060529/NEWS/105290135 # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Atlantic/Bermuda -4:19:18 - LMT 1930 Jan 1 2:00 # Hamilton -4:00 - AST 1974 Apr 28 2:00 -4:00 Canada A%sT 1976 -4:00 US A%sT # Cayman Is # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Cayman -5:25:32 - LMT 1890 # Georgetown -5:07:11 - KMT 1912 Feb # Kingston Mean Time -5:00 - EST # Costa Rica # Milne gives -5:36:13.3 as San José mean time; round to nearest. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule CR 1979 1980 - Feb lastSun 0:00 1:00 D Rule CR 1979 1980 - Jun Sun>=1 0:00 0 S Rule CR 1991 1992 - Jan Sat>=15 0:00 1:00 D # IATA SSIM (1991-09) says the following was at 1:00; # go with Shanks & Pottenger. Rule CR 1991 only - Jul 1 0:00 0 S Rule CR 1992 only - Mar 15 0:00 0 S # There are too many San Josés elsewhere, so we'll use 'Costa Rica'. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Costa_Rica -5:36:13 - LMT 1890 # San José -5:36:13 - SJMT 1921 Jan 15 # San José Mean Time -6:00 CR C%sT # Coco # no information; probably like America/Costa_Rica # Cuba # From Paul Eggert (2013-02-21): # Milne gives -5:28:50.45 for the observatory at Havana, -5:29:23.57 # for the port, and -5:30 for meteorological observations. # For now, stick with Shanks & Pottenger. # From Arthur David Olson (1999-03-29): # The 1999-03-28 exhibition baseball game held in Havana, Cuba, between # the Cuban National Team and the Baltimore Orioles was carried live on # the Orioles Radio Network, including affiliate WTOP in Washington, DC. # During the game, play-by-play announcer Jim Hunter noted that # "We'll be losing two hours of sleep...Cuba switched to Daylight Saving # Time today." (The "two hour" remark referred to losing one hour of # sleep on 1999-03-28 - when the announcers were in Cuba as it switched # to DST - and one more hour on 1999-04-04 - when the announcers will have # returned to Baltimore, which switches on that date.) # From Steffen Thorsen (2013-11-11): # DST start in Cuba in 2004 ... does not follow the same rules as the # years before. The correct date should be Sunday 2004-03-28 00:00 ... # https://web.archive.org/web/20040402060750/http://www.granma.cu/espanol/2004/marzo/sab27/reloj.html # From Evert van der Veer via Steffen Thorsen (2004-10-28): # Cuba is not going back to standard time this year. # From Paul Eggert (2006-03-22): # http://www.granma.cu/ingles/2004/septiembre/juev30/41medid-i.html # says that it's due to a problem at the Antonio Guiteras # thermoelectric plant, and says "This October there will be no return # to normal hours (after daylight saving time)". # For now, let's assume that it's a temporary measure. # From Carlos A. Carnero Delgado (2005-11-12): # This year (just like in 2004-2005) there's no change in time zone # adjustment in Cuba. We will stay in daylight saving time: # http://www.granma.cu/espanol/2005/noviembre/mier9/horario.html # From Jesper Nørgaard Welen (2006-10-21): # An article in GRANMA INTERNACIONAL claims that Cuba will end # the 3 years of permanent DST next weekend, see # http://www.granma.cu/ingles/2006/octubre/lun16/43horario.html # "On Saturday night, October 28 going into Sunday, October 29, at 01:00, # watches should be set back one hour - going back to 00:00 hours - returning # to the normal schedule.... # From Paul Eggert (2007-03-02): # , dated yesterday, # says Cuban clocks will advance at midnight on March 10. # For lack of better information, assume Cuba will use US rules, # except that it switches at midnight standard time as usual. # # From Steffen Thorsen (2007-10-25): # Carlos Alberto Fonseca Arauz informed me that Cuba will end DST one week # earlier - on the last Sunday of October, just like in 2006. # # He supplied these references: # # http://www.prensalatina.com.mx/article.asp?ID=%7B4CC32C1B-A9F7-42FB-8A07-8631AFC923AF%7D&language=ES # http://actualidad.terra.es/sociedad/articulo/cuba_llama_ahorrar_energia_cambio_1957044.htm # # From Alex Krivenyshev (2007-10-25): # Here is also article from Granma (Cuba): # # Regirá el Horario Normal desde el próximo domingo 28 de octubre # http://www.granma.cubaweb.cu/2007/10/24/nacional/artic07.html # # http://www.worldtimezone.com/dst_news/dst_news_cuba03.html # From Arthur David Olson (2008-03-09): # I'm in Maryland which is now observing United States Eastern Daylight # Time. At 9:44 local time I used RealPlayer to listen to # http://media.enet.cu/radioreloj # a Cuban information station, and heard # the time announced as "ocho cuarenta y cuatro" ("eight forty-four"), # indicating that Cuba is still on standard time. # From Steffen Thorsen (2008-03-12): # It seems that Cuba will start DST on Sunday, 2007-03-16... # It was announced yesterday, according to this source (in Spanish): # http://www.nnc.cubaweb.cu/marzo-2008/cien-1-11-3-08.htm # # Some more background information is posted here: # http://www.timeanddate.com/news/time/cuba-starts-dst-march-16.html # # The article also says that Cuba has been observing DST since 1963, # while Shanks (and tzdata) has 1965 as the first date (except in the # 1940's). Many other web pages in Cuba also claim that it has been # observed since 1963, but with the exception of 1970 - an exception # which is not present in tzdata/Shanks. So there is a chance we need to # change some historic records as well. # # One example: # http://www.radiohc.cu/espanol/noticias/mar07/11mar/hor.htm # From Jesper Nørgaard Welen (2008-03-13): # The Cuban time change has just been confirmed on the most authoritative # web site, the Granma. Please check out # http://www.granma.cubaweb.cu/2008/03/13/nacional/artic10.html # # Basically as expected after Steffen Thorsen's information, the change # will take place midnight between Saturday and Sunday. # From Arthur David Olson (2008-03-12): # Assume Sun>=15 (third Sunday) going forward. # From Alexander Krivenyshev (2009-03-04) # According to the Radio Reloj - Cuba will start Daylight Saving Time on # midnight between Saturday, March 07, 2009 and Sunday, March 08, 2009- # not on midnight March 14 / March 15 as previously thought. # # http://www.worldtimezone.com/dst_news/dst_news_cuba05.html # (in Spanish) # From Arthur David Olson (2009-03-09) # I listened over the Internet to # http://media.enet.cu/readioreloj # this morning; when it was 10:05 a. m. here in Bethesda, Maryland the # the time was announced as "diez cinco" - the same time as here, indicating # that has indeed switched to DST. Assume second Sunday from 2009 forward. # From Steffen Thorsen (2011-03-08): # Granma announced that Cuba is going to start DST on 2011-03-20 00:00:00 # this year. Nothing about the end date known so far (if that has # changed at all). # # Source: # http://granma.co.cu/2011/03/08/nacional/artic01.html # # Our info: # http://www.timeanddate.com/news/time/cuba-starts-dst-2011.html # # From Steffen Thorsen (2011-10-30) # Cuba will end DST two weeks later this year. Instead of going back # tonight, it has been delayed to 2011-11-13 at 01:00. # # One source (Spanish) # http://www.radioangulo.cu/noticias/cuba/17105-cuba-restablecera-el-horario-del-meridiano-de-greenwich.html # # Our page: # http://www.timeanddate.com/news/time/cuba-time-changes-2011.html # # From Steffen Thorsen (2012-03-01) # According to Radio Reloj, Cuba will start DST on Midnight between March # 31 and April 1. # # Radio Reloj has the following info (Spanish): # http://www.radioreloj.cu/index.php/noticias-radio-reloj/71-miscelaneas/7529-cuba-aplicara-el-horario-de-verano-desde-el-1-de-abril # # Our info on it: # http://www.timeanddate.com/news/time/cuba-starts-dst-2012.html # From Steffen Thorsen (2012-11-03): # Radio Reloj and many other sources report that Cuba is changing back # to standard time on 2012-11-04: # http://www.radioreloj.cu/index.php/noticias-radio-reloj/36-nacionales/9961-regira-horario-normal-en-cuba-desde-el-domingo-cuatro-de-noviembre # From Paul Eggert (2012-11-03): # For now, assume the future rule is first Sunday in November. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Cuba 1928 only - Jun 10 0:00 1:00 D Rule Cuba 1928 only - Oct 10 0:00 0 S Rule Cuba 1940 1942 - Jun Sun>=1 0:00 1:00 D Rule Cuba 1940 1942 - Sep Sun>=1 0:00 0 S Rule Cuba 1945 1946 - Jun Sun>=1 0:00 1:00 D Rule Cuba 1945 1946 - Sep Sun>=1 0:00 0 S Rule Cuba 1965 only - Jun 1 0:00 1:00 D Rule Cuba 1965 only - Sep 30 0:00 0 S Rule Cuba 1966 only - May 29 0:00 1:00 D Rule Cuba 1966 only - Oct 2 0:00 0 S Rule Cuba 1967 only - Apr 8 0:00 1:00 D Rule Cuba 1967 1968 - Sep Sun>=8 0:00 0 S Rule Cuba 1968 only - Apr 14 0:00 1:00 D Rule Cuba 1969 1977 - Apr lastSun 0:00 1:00 D Rule Cuba 1969 1971 - Oct lastSun 0:00 0 S Rule Cuba 1972 1974 - Oct 8 0:00 0 S Rule Cuba 1975 1977 - Oct lastSun 0:00 0 S Rule Cuba 1978 only - May 7 0:00 1:00 D Rule Cuba 1978 1990 - Oct Sun>=8 0:00 0 S Rule Cuba 1979 1980 - Mar Sun>=15 0:00 1:00 D Rule Cuba 1981 1985 - May Sun>=5 0:00 1:00 D Rule Cuba 1986 1989 - Mar Sun>=14 0:00 1:00 D Rule Cuba 1990 1997 - Apr Sun>=1 0:00 1:00 D Rule Cuba 1991 1995 - Oct Sun>=8 0:00s 0 S Rule Cuba 1996 only - Oct 6 0:00s 0 S Rule Cuba 1997 only - Oct 12 0:00s 0 S Rule Cuba 1998 1999 - Mar lastSun 0:00s 1:00 D Rule Cuba 1998 2003 - Oct lastSun 0:00s 0 S Rule Cuba 2000 2003 - Apr Sun>=1 0:00s 1:00 D Rule Cuba 2004 only - Mar lastSun 0:00s 1:00 D Rule Cuba 2006 2010 - Oct lastSun 0:00s 0 S Rule Cuba 2007 only - Mar Sun>=8 0:00s 1:00 D Rule Cuba 2008 only - Mar Sun>=15 0:00s 1:00 D Rule Cuba 2009 2010 - Mar Sun>=8 0:00s 1:00 D Rule Cuba 2011 only - Mar Sun>=15 0:00s 1:00 D Rule Cuba 2011 only - Nov 13 0:00s 0 S Rule Cuba 2012 only - Apr 1 0:00s 1:00 D Rule Cuba 2012 max - Nov Sun>=1 0:00s 0 S Rule Cuba 2013 max - Mar Sun>=8 0:00s 1:00 D # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Havana -5:29:28 - LMT 1890 -5:29:36 - HMT 1925 Jul 19 12:00 # Havana MT -5:00 Cuba C%sT # Dominica # See America/Port_of_Spain. # Dominican Republic # From Steffen Thorsen (2000-10-30): # Enrique Morales reported to me that the Dominican Republic has changed the # time zone to Eastern Standard Time as of Sunday 29 at 2 am.... # http://www.listin.com.do/antes/261000/republica/princi.html # From Paul Eggert (2000-12-04): # That URL (2000-10-26, in Spanish) says they planned to use US-style DST. # From Rives McDow (2000-12-01): # Dominican Republic changed its mind and presidential decree on Tuesday, # November 28, 2000, with a new decree. On Sunday, December 3 at 1:00 AM the # Dominican Republic will be reverting to 8 hours from the International Date # Line, and will not be using DST in the foreseeable future. The reason they # decided to use DST was to be in synch with Puerto Rico, who was also going # to implement DST. When Puerto Rico didn't implement DST, the president # decided to revert. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule DR 1966 only - Oct 30 0:00 1:00 D Rule DR 1967 only - Feb 28 0:00 0 S Rule DR 1969 1973 - Oct lastSun 0:00 0:30 HD Rule DR 1970 only - Feb 21 0:00 0 S Rule DR 1971 only - Jan 20 0:00 0 S Rule DR 1972 1974 - Jan 21 0:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Santo_Domingo -4:39:36 - LMT 1890 -4:40 - SDMT 1933 Apr 1 12:00 # S. Dom. MT -5:00 DR E%sT 1974 Oct 27 -4:00 - AST 2000 Oct 29 2:00 -5:00 US E%sT 2000 Dec 3 1:00 -4:00 - AST # El Salvador # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Salv 1987 1988 - May Sun>=1 0:00 1:00 D Rule Salv 1987 1988 - Sep lastSun 0:00 0 S # There are too many San Salvadors elsewhere, so use America/El_Salvador # instead of America/San_Salvador. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/El_Salvador -5:56:48 - LMT 1921 # San Salvador -6:00 Salv C%sT # Grenada # Guadeloupe # St Barthélemy # St Martin (French part) # See America/Port_of_Spain. # Guatemala # # From Gwillim Law (2006-04-22), after a heads-up from Oscar van Vlijmen: # Diario Co Latino, at # , # says in an article dated 2006-04-19 that the Guatemalan government had # decided on that date to advance official time by 60 minutes, to lessen the # impact of the elevated cost of oil.... Daylight saving time will last from # 2006-04-29 24:00 (Guatemalan standard time) to 2006-09-30 (time unspecified). # From Paul Eggert (2006-06-22): # The Ministry of Energy and Mines, press release CP-15/2006 # (2006-04-19), says DST ends at 24:00. See # http://www.sieca.org.gt/Sitio_publico/Energeticos/Doc/Medidas/Cambio_Horario_Nac_190406.pdf # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Guat 1973 only - Nov 25 0:00 1:00 D Rule Guat 1974 only - Feb 24 0:00 0 S Rule Guat 1983 only - May 21 0:00 1:00 D Rule Guat 1983 only - Sep 22 0:00 0 S Rule Guat 1991 only - Mar 23 0:00 1:00 D Rule Guat 1991 only - Sep 7 0:00 0 S Rule Guat 2006 only - Apr 30 0:00 1:00 D Rule Guat 2006 only - Oct 1 0:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Guatemala -6:02:04 - LMT 1918 Oct 5 -6:00 Guat C%sT # Haiti # From Gwillim Law (2005-04-15): # Risto O. Nykänen wrote me that Haiti is now on DST. # I searched for confirmation, and I found a press release # on the Web page of the Haitian Consulate in Chicago (2005-03-31), # . Translated from French, it says: # # "The Prime Minister's Communication Office notifies the public in general # and the press in particular that, following a decision of the Interior # Ministry and the Territorial Collectivities [I suppose that means the # provinces], Haiti will move to Eastern Daylight Time in the night from next # Saturday the 2nd to Sunday the 3rd. # # "Consequently, the Prime Minister's Communication Office wishes to inform # the population that the country's clocks will be set forward one hour # starting at midnight. This provision will hold until the last Saturday in # October 2005. # # "Port-au-Prince, March 31, 2005" # # From Steffen Thorsen (2006-04-04): # I have been informed by users that Haiti observes DST this year like # last year, so the current "only" rule for 2005 might be changed to a # "max" rule or to last until 2006. (Who knows if they will observe DST # next year or if they will extend their DST like US/Canada next year). # # I have found this article about it (in French): # http://www.haitipressnetwork.com/news.cfm?articleID=7612 # # The reason seems to be an energy crisis. # From Stephen Colebourne (2007-02-22): # Some IATA info: Haiti won't be having DST in 2007. # From Steffen Thorsen (2012-03-11): # According to several news sources, Haiti will observe DST this year, # apparently using the same start and end date as USA/Canada. # So this means they have already changed their time. # # http://www.alterpresse.org/spip.php?article12510 # http://radiovision2000haiti.net/home/?p=13253 # # From Arthur David Olson (2012-03-11): # The alterpresse.org source seems to show a US-style leap from 2:00 a.m. to # 3:00 a.m. rather than the traditional Haitian jump at midnight. # Assume a US-style fall back as well. # From Steffen Thorsen (2013-03-10): # It appears that Haiti is observing DST this year as well, same rules # as US/Canada. They did it last year as well, and it looks like they # are going to observe DST every year now... # # http://radiovision2000haiti.net/public/haiti-avis-changement-dheure-dimanche/ # http://www.canalplushaiti.net/?p=6714 # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Haiti 1983 only - May 8 0:00 1:00 D Rule Haiti 1984 1987 - Apr lastSun 0:00 1:00 D Rule Haiti 1983 1987 - Oct lastSun 0:00 0 S # Shanks & Pottenger say AT is 2:00, but IATA SSIM (1991/1997) says 1:00s. # Go with IATA. Rule Haiti 1988 1997 - Apr Sun>=1 1:00s 1:00 D Rule Haiti 1988 1997 - Oct lastSun 1:00s 0 S Rule Haiti 2005 2006 - Apr Sun>=1 0:00 1:00 D Rule Haiti 2005 2006 - Oct lastSun 0:00 0 S Rule Haiti 2012 max - Mar Sun>=8 2:00 1:00 D Rule Haiti 2012 max - Nov Sun>=1 2:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Port-au-Prince -4:49:20 - LMT 1890 -4:49 - PPMT 1917 Jan 24 12:00 # P-a-P MT -5:00 Haiti E%sT # Honduras # Shanks & Pottenger say 1921 Jan 1; go with Whitman's more precise Apr 1. # From Paul Eggert (2006-05-05): # worldtimezone.com reports a 2006-05-02 Spanish-language AP article # saying Honduras will start using DST midnight Saturday, effective 4 # months until September. La Tribuna reported today # that Manuel Zelaya, the president # of Honduras, refused to back down on this. # From Jesper Nørgaard Welen (2006-08-08): # It seems that Honduras has returned from DST to standard time this Monday at # 00:00 hours (prolonging Sunday to 25 hours duration). # http://www.worldtimezone.com/dst_news/dst_news_honduras04.html # From Paul Eggert (2006-08-08): # Also see Diario El Heraldo, The country returns to standard time (2006-08-08). # http://www.elheraldo.hn/nota.php?nid=54941&sec=12 # It mentions executive decree 18-2006. # From Steffen Thorsen (2006-08-17): # Honduras will observe DST from 2007 to 2009, exact dates are not # published, I have located this authoritative source: # http://www.presidencia.gob.hn/noticia.aspx?nId=47 # From Steffen Thorsen (2007-03-30): # http://www.laprensahn.com/pais_nota.php?id04962=7386 # So it seems that Honduras will not enter DST this year.... # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Hond 1987 1988 - May Sun>=1 0:00 1:00 D Rule Hond 1987 1988 - Sep lastSun 0:00 0 S Rule Hond 2006 only - May Sun>=1 0:00 1:00 D Rule Hond 2006 only - Aug Mon>=1 0:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Tegucigalpa -5:48:52 - LMT 1921 Apr -6:00 Hond C%sT # # Great Swan I ceded by US to Honduras in 1972 # Jamaica # Shanks & Pottenger give -5:07:12, but Milne records -5:07:10.41 from an # unspecified official document, and says "This time is used throughout the # island". Go with Milne. Round to the nearest second as required by zic. # # Shanks & Pottenger give April 28 for the 1974 spring-forward transition, but # Lance Neita writes that Prime Minister Michael Manley decreed it January 5. # Assume Neita meant Jan 6 02:00, the same as the US. Neita also writes that # Manley's supporters associated this act with Manley's nickname "Joshua" # (recall that in the Bible the sun stood still at Joshua's request), # and with the Rod of Correction which Manley said he had received from # Haile Selassie, Emperor of Ethiopia. See: # Neita L. The politician in all of us. Jamaica Observer 2014-09-20 # http://www.jamaicaobserver.com/columns/The-politician-in-all-of-us_17573647 # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Jamaica -5:07:11 - LMT 1890 # Kingston -5:07:11 - KMT 1912 Feb # Kingston Mean Time -5:00 - EST 1974 -5:00 US E%sT 1984 -5:00 - EST # Martinique # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Martinique -4:04:20 - LMT 1890 # Fort-de-France -4:04:20 - FFMT 1911 May # Fort-de-France MT -4:00 - AST 1980 Apr 6 -4:00 1:00 ADT 1980 Sep 28 -4:00 - AST # Montserrat # See America/Port_of_Spain. # Nicaragua # # This uses Shanks & Pottenger for times before 2005. # # From Steffen Thorsen (2005-04-12): # I've got reports from 8 different people that Nicaragua just started # DST on Sunday 2005-04-10, in order to save energy because of # expensive petroleum. The exact end date for DST is not yet # announced, only "September" but some sites also say "mid-September". # Some background information is available on the President's official site: # http://www.presidencia.gob.ni/Presidencia/Files_index/Secretaria/Notas%20de%20Prensa/Presidente/2005/ABRIL/Gobierno-de-nicaragua-adelanta-hora-oficial-06abril.htm # The Decree, no 23-2005 is available here: # http://www.presidencia.gob.ni/buscador_gaceta/BD/DECRETOS/2005/Decreto%2023-2005%20Se%20adelanta%20en%20una%20hora%20en%20todo%20el%20territorio%20nacional%20apartir%20de%20las%2024horas%20del%2009%20de%20Abril.pdf # # From Paul Eggert (2005-05-01): # The decree doesn't say anything about daylight saving, but for now let's # assume that it is daylight saving.... # # From Gwillim Law (2005-04-21): # The Associated Press story on the time change, which can be found at # http://www.lapalmainteractivo.com/guias/content/gen/ap/America_Latina/AMC_GEN_NICARAGUA_HORA.html # and elsewhere, says (fifth paragraph, translated from Spanish): "The last # time that a change of clocks was applied to save energy was in the year 2000 # during the Arnoldo Alemán administration."... # The northamerica file says that Nicaragua has been on UTC-6 continuously # since December 1998. I wasn't able to find any details of Nicaraguan time # changes in 2000. Perhaps a note could be added to the northamerica file, to # the effect that we have indirect evidence that DST was observed in 2000. # # From Jesper Nørgaard Welen (2005-11-02): # Nicaragua left DST the 2005-10-02 at 00:00 (local time). # http://www.presidencia.gob.ni/presidencia/files_index/secretaria/comunicados/2005/septiembre/26septiembre-cambio-hora.htm # (2005-09-26) # # From Jesper Nørgaard Welen (2006-05-05): # http://www.elnuevodiario.com.ni/2006/05/01/nacionales/18410 # (my informal translation) # By order of the president of the republic, Enrique Bolaños, Nicaragua # advanced by sixty minutes their official time, yesterday at 2 in the # morning, and will stay that way until 30th of September. # # From Jesper Nørgaard Welen (2006-09-30): # http://www.presidencia.gob.ni/buscador_gaceta/BD/DECRETOS/2006/D-063-2006P-PRN-Cambio-Hora.pdf # My informal translation runs: # The natural sun time is restored in all the national territory, in that the # time is returned one hour at 01:00 am of October 1 of 2006. # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Nic 1979 1980 - Mar Sun>=16 0:00 1:00 D Rule Nic 1979 1980 - Jun Mon>=23 0:00 0 S Rule Nic 2005 only - Apr 10 0:00 1:00 D Rule Nic 2005 only - Oct Sun>=1 0:00 0 S Rule Nic 2006 only - Apr 30 2:00 1:00 D Rule Nic 2006 only - Oct Sun>=1 1:00 0 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Managua -5:45:08 - LMT 1890 -5:45:12 - MMT 1934 Jun 23 # Managua Mean Time? -6:00 - CST 1973 May -5:00 - EST 1975 Feb 16 -6:00 Nic C%sT 1992 Jan 1 4:00 -5:00 - EST 1992 Sep 24 -6:00 - CST 1993 -5:00 - EST 1997 -6:00 Nic C%sT # Panama # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Panama -5:18:08 - LMT 1890 -5:19:36 - CMT 1908 Apr 22 # Colón Mean Time -5:00 - EST # Puerto Rico # There are too many San Juans elsewhere, so we'll use 'Puerto_Rico'. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Puerto_Rico -4:24:25 - LMT 1899 Mar 28 12:00 # San Juan -4:00 - AST 1942 May 3 -4:00 US A%sT 1946 -4:00 - AST # St Kitts-Nevis # St Lucia # See America/Port_of_Spain. # St Pierre and Miquelon # There are too many St Pierres elsewhere, so we'll use 'Miquelon'. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Miquelon -3:44:40 - LMT 1911 May 15 # St Pierre -4:00 - AST 1980 May -3:00 - PMST 1987 # Pierre & Miquelon Time -3:00 Canada PM%sT # St Vincent and the Grenadines # See America/Port_of_Spain. # Turks and Caicos # # From Chris Dunn in # http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=415007 # (2007-03-15): In the Turks & Caicos Islands (America/Grand_Turk) the # daylight saving dates for time changes have been adjusted to match # the recent U.S. change of dates. # # From Brian Inglis (2007-04-28): # http://www.turksandcaicos.tc/calendar/index.htm [2007-04-26] # there is an entry for Nov 4 "Daylight Savings Time Ends 2007" and three # rows before that there is an out of date entry for Oct: # "Eastern Standard Times Begins 2007 # Clocks are set back one hour at 2:00 a.m. local Daylight Saving Time" # indicating that the normal ET rules are followed. # # From Paul Eggert (2014-08-19): # The 2014-08-13 Cabinet meeting decided to stay on UTC-4 year-round. See: # http://tcweeklynews.com/daylight-savings-time-to-be-maintained-p5353-127.htm # Model this as a switch from EST/EDT to AST ... # From Chris Walton (2014-11-04): # ... the TCI government appears to have delayed the switch to # "permanent daylight saving time" by one year.... # http://tcweeklynews.com/time-change-to-go-ahead-this-november-p5437-127.htm # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Grand_Turk -4:44:32 - LMT 1890 -5:07:11 - KMT 1912 Feb # Kingston Mean Time -5:00 - EST 1979 -5:00 US E%sT 2015 Nov Sun>=1 2:00 -4:00 - AST # British Virgin Is # Virgin Is # See America/Port_of_Spain. # Local Variables: # coding: utf-8 # End: Index: projects/cxl_iscsi/contrib/tzdata/southamerica =================================================================== --- projects/cxl_iscsi/contrib/tzdata/southamerica (revision 279716) +++ projects/cxl_iscsi/contrib/tzdata/southamerica (revision 279717) @@ -1,1702 +1,1709 @@ # This file is in the public domain, so clarified as of # 2009-05-17 by Arthur David Olson. # This file is by no means authoritative; if you think you know better, # go ahead and edit the file (and please send any changes to # tz@iana.org for general use in the future). For more, please see # the file CONTRIBUTING in the tz distribution. # From Paul Eggert (2014-10-31): # # Unless otherwise specified, the source for data through 1990 is: # Thomas G. Shanks and Rique Pottenger, The International Atlas (6th edition), # San Diego: ACS Publications, Inc. (2003). # Unfortunately this book contains many errors and cites no sources. # # Gwillim Law writes that a good source # for recent time zone data is the International Air Transport # Association's Standard Schedules Information Manual (IATA SSIM), # published semiannually. Law sent in several helpful summaries # of the IATA's data after 1990. Except where otherwise noted, # IATA SSIM is the source for entries after 1990. # # For data circa 1899, a common source is: # Milne J. Civil time. Geogr J. 1899 Feb;13(2):173-94. # http://www.jstor.org/stable/1774359 # # Earlier editions of these tables used the North American style (e.g. ARST and # ARDT for Argentine Standard and Daylight Time), but the following quote # suggests that it's better to use European style (e.g. ART and ARST). # I suggest the use of _Summer time_ instead of the more cumbersome # _daylight-saving time_. _Summer time_ seems to be in general use # in Europe and South America. # -- E O Cutler, _New York Times_ (1937-02-14), quoted in # H L Mencken, _The American Language: Supplement I_ (1960), p 466 # # Earlier editions of these tables also used the North American style # for time zones in Brazil, but this was incorrect, as Brazilians say # "summer time". Reinaldo Goulart, a São Paulo businessman active in # the railroad sector, writes (1999-07-06): # The subject of time zones is currently a matter of discussion/debate in # Brazil. Let's say that "the Brasília time" is considered the # "official time" because Brasília is the capital city. # The other three time zones are called "Brasília time "minus one" or # "plus one" or "plus two". As far as I know there is no such # name/designation as "Eastern Time" or "Central Time". # So I invented the following (English-language) abbreviations for now. # Corrections are welcome! # std dst # -2:00 FNT FNST Fernando de Noronha # -3:00 BRT BRST Brasília # -4:00 AMT AMST Amazon # -5:00 ACT ACST Acre ############################################################################### ############################################################################### # Argentina # From Bob Devine (1988-01-28): # Argentina: first Sunday in October to first Sunday in April since 1976. # Double Summer time from 1969 to 1974. Switches at midnight. # From U. S. Naval Observatory (1988-01-19): # ARGENTINA 3 H BEHIND UTC # From Hernan G. Otero (1995-06-26): # I am sending modifications to the Argentine time zone table... # AR was chosen because they are the ISO letters that represent Argentina. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Arg 1930 only - Dec 1 0:00 1:00 S Rule Arg 1931 only - Apr 1 0:00 0 - Rule Arg 1931 only - Oct 15 0:00 1:00 S Rule Arg 1932 1940 - Mar 1 0:00 0 - Rule Arg 1932 1939 - Nov 1 0:00 1:00 S Rule Arg 1940 only - Jul 1 0:00 1:00 S Rule Arg 1941 only - Jun 15 0:00 0 - Rule Arg 1941 only - Oct 15 0:00 1:00 S Rule Arg 1943 only - Aug 1 0:00 0 - Rule Arg 1943 only - Oct 15 0:00 1:00 S Rule Arg 1946 only - Mar 1 0:00 0 - Rule Arg 1946 only - Oct 1 0:00 1:00 S Rule Arg 1963 only - Oct 1 0:00 0 - Rule Arg 1963 only - Dec 15 0:00 1:00 S Rule Arg 1964 1966 - Mar 1 0:00 0 - Rule Arg 1964 1966 - Oct 15 0:00 1:00 S Rule Arg 1967 only - Apr 2 0:00 0 - Rule Arg 1967 1968 - Oct Sun>=1 0:00 1:00 S Rule Arg 1968 1969 - Apr Sun>=1 0:00 0 - Rule Arg 1974 only - Jan 23 0:00 1:00 S Rule Arg 1974 only - May 1 0:00 0 - Rule Arg 1988 only - Dec 1 0:00 1:00 S # # From Hernan G. Otero (1995-06-26): # These corrections were contributed by InterSoft Argentina S.A., # obtaining the data from the: # Talleres de Hidrografía Naval Argentina # (Argentine Naval Hydrography Institute) Rule Arg 1989 1993 - Mar Sun>=1 0:00 0 - Rule Arg 1989 1992 - Oct Sun>=15 0:00 1:00 S # # From Hernan G. Otero (1995-06-26): # From this moment on, the law that mandated the daylight saving # time corrections was derogated and no more modifications # to the time zones (for daylight saving) are now made. # # From Rives McDow (2000-01-10): # On October 3, 1999, 0:00 local, Argentina implemented daylight savings time, # which did not result in the switch of a time zone, as they stayed 9 hours # from the International Date Line. Rule Arg 1999 only - Oct Sun>=1 0:00 1:00 S # From Paul Eggert (2007-12-28): # DST was set to expire on March 5, not March 3, but since it was converted # to standard time on March 3 it's more convenient for us to pretend that # it ended on March 3. Rule Arg 2000 only - Mar 3 0:00 0 - # # From Peter Gradelski via Steffen Thorsen (2000-03-01): # We just checked with our São Paulo office and they say the government of # Argentina decided not to become one of the countries that go on or off DST. # So Buenos Aires should be -3 hours from GMT at all times. # # From Fabián L. Arce Jofré (2000-04-04): # The law that claimed DST for Argentina was derogated by President Fernando # de la Rúa on March 2, 2000, because it would make people spend more energy # in the winter time, rather than less. The change took effect on March 3. # # From Mariano Absatz (2001-06-06): # one of the major newspapers here in Argentina said that the 1999 # Timezone Law (which never was effectively applied) will (would?) be # in effect.... The article is at # http://ar.clarin.com/diario/2001-06-06/e-01701.htm # ... The Law itself is "Ley No 25155", sanctioned on 1999-08-25, enacted # 1999-09-17, and published 1999-09-21. The official publication is at: # http://www.boletin.jus.gov.ar/BON/Primera/1999/09-Septiembre/21/PDF/BO21-09-99LEG.PDF # Regretfully, you have to subscribe (and pay) for the on-line version.... # # (2001-06-12): # the timezone for Argentina will not change next Sunday. # Apparently it will do so on Sunday 24th.... # http://ar.clarin.com/diario/2001-06-12/s-03501.htm # # (2001-06-25): # Last Friday (yes, the last working day before the date of the change), the # Senate annulled the 1999 law that introduced the changes later postponed. # http://www.clarin.com.ar/diario/2001-06-22/s-03601.htm # It remains the vote of the Deputies..., but it will be the same.... # This kind of things had always been done this way in Argentina. # We are still -03:00 all year round in all of the country. # # From Steffen Thorsen (2007-12-21): # A user (Leonardo Chaim) reported that Argentina will adopt DST.... # all of the country (all Zone-entries) are affected. News reports like # http://www.lanacion.com.ar/opinion/nota.asp?nota_id=973037 indicate # that Argentina will use DST next year as well, from October to # March, although exact rules are not given. # # From Jesper Nørgaard Welen (2007-12-26) # The last hurdle of Argentina DST is over, the proposal was approved in # the lower chamber too (Diputados) with a vote 192 for and 2 against. # By the way thanks to Mariano Absatz and Daniel Mario Vega for the link to # the original scanned proposal, where the dates and the zero hours are # clear and unambiguous...This is the article about final approval: # http://www.lanacion.com.ar/politica/nota.asp?nota_id=973996 # # From Paul Eggert (2007-12-22): # For dates after mid-2008, the following rules are my guesses and # are quite possibly wrong, but are more likely than no DST at all. # From Alexander Krivenyshev (2008-09-05): # As per message from Carlos Alberto Fonseca Arauz (Nicaragua), # Argentina will start DST on Sunday October 19, 2008. # # http://www.worldtimezone.com/dst_news/dst_news_argentina03.html # http://www.impulsobaires.com.ar/nota.php?id=57832 (in spanish) # From Rodrigo Severo (2008-10-06): # Here is some info available at a Gentoo bug related to TZ on Argentina's DST: # ... # ------- Comment #1 from [jmdocile] 2008-10-06 16:28 0000 ------- # Hi, there is a problem with timezone-data-2008e and maybe with # timezone-data-2008f # Argentinian law [Number] 25.155 is no longer valid. # http://www.infoleg.gov.ar/infolegInternet/anexos/60000-64999/60036/norma.htm # The new one is law [Number] 26.350 # http://www.infoleg.gov.ar/infolegInternet/anexos/135000-139999/136191/norma.htm # So there is no summer time in Argentina for now. # From Mariano Absatz (2008-10-20): # Decree 1693/2008 applies Law 26.350 for the summer 2008/2009 establishing DST # in Argentina from 2008-10-19 until 2009-03-15. # http://www.boletinoficial.gov.ar/Bora.Portal/CustomControls/PdfContent.aspx?fp=16102008&pi=3&pf=4&s=0&sec=01 # # Decree 1705/2008 excepting 12 Provinces from applying DST in the summer # 2008/2009: Catamarca, La Rioja, Mendoza, Salta, San Juan, San Luis, La # Pampa, Neuquén, Rio Negro, Chubut, Santa Cruz and Tierra del Fuego # http://www.boletinoficial.gov.ar/Bora.Portal/CustomControls/PdfContent.aspx?fp=17102008&pi=1&pf=1&s=0&sec=01 # # Press release 235 dated Saturday October 18th, from the Government of the # Province of Jujuy saying it will not apply DST either (even when it was not # included in Decree 1705/2008). # http://www.jujuy.gov.ar/index2/partes_prensa/18_10_08/235-181008.doc # From fullinet (2009-10-18): # As announced in # http://www.argentina.gob.ar/argentina/portal/paginas.dhtml?pagina=356 # (an official .gob.ar) under title: "Sin Cambio de Hora" # (English: "No hour change"). # # "Por el momento, el Gobierno Nacional resolvió no modificar la hora # oficial, decisión que estaba en estudio para su implementación el # domingo 18 de octubre. Desde el Ministerio de Planificación se anunció # que la Argentina hoy, en estas condiciones meteorológicas, no necesita # la modificación del huso horario, ya que 2009 nos encuentra con # crecimiento en la producción y distribución energética." Rule Arg 2007 only - Dec 30 0:00 1:00 S Rule Arg 2008 2009 - Mar Sun>=15 0:00 0 - Rule Arg 2008 only - Oct Sun>=15 0:00 1:00 S # From Mariano Absatz (2004-05-21): # Today it was officially published that the Province of Mendoza is changing # its timezone this winter... starting tomorrow night.... # http://www.gobernac.mendoza.gov.ar/boletin/pdf/20040521-27158-normas.pdf # From Paul Eggert (2004-05-24): # It's Law No. 7,210. This change is due to a public power emergency, so for # now we'll assume it's for this year only. # # From Paul Eggert (2014-08-09): # Hora de verano para la República Argentina # http://buenasiembra.com.ar/esoterismo/astrologia/hora-de-verano-de-la-republica-argentina-27.html # says that standard time in Argentina from 1894-10-31 # to 1920-05-01 was -4:16:48.25. Go with this more-precise value # over Shanks & Pottenger. # # From Mariano Absatz (2004-06-05): # These media articles from a major newspaper mostly cover the current state: # http://www.lanacion.com.ar/04/05/27/de_604825.asp # http://www.lanacion.com.ar/04/05/28/de_605203.asp # # The following eight (8) provinces pulled clocks back to UTC-04:00 at # midnight Monday May 31st. (that is, the night between 05/31 and 06/01). # Apparently, all nine provinces would go back to UTC-03:00 at the same # time in October 17th. # # Catamarca, Chubut, La Rioja, San Juan, San Luis, Santa Cruz, # Tierra del Fuego, Tucumán. # # From Mariano Absatz (2004-06-14): # ... this weekend, the Province of Tucumán decided it'd go back to UTC-03:00 # yesterday midnight (that is, at 24:00 Saturday 12th), since the people's # annoyance with the change is much higher than the power savings obtained.... # # From Gwillim Law (2004-06-14): # http://www.lanacion.com.ar/04/06/10/de_609078.asp ... # "The time change in Tierra del Fuego was a conflicted decision from # the start. The government had decreed that the measure would take # effect on June 1, but a normative error forced the new time to begin # three days earlier, from a Saturday to a Sunday.... # Our understanding was that the change was originally scheduled to take place # on June 1 at 00:00 in Chubut, Santa Cruz, Tierra del Fuego (and some other # provinces). Sunday was May 30, only two days earlier. So the article # contains a contradiction. I would give more credence to the Saturday/Sunday # date than the "three days earlier" phrase, and conclude that Tierra del # Fuego set its clocks back at 2004-05-30 00:00. # # From Steffen Thorsen (2004-10-05): # The previous law 7210 which changed the province of Mendoza's time zone # back in May have been modified slightly in a new law 7277, which set the # new end date to 2004-09-26 (original date was 2004-10-17). # http://www.gobernac.mendoza.gov.ar/boletin/pdf/20040924-27244-normas.pdf # # From Mariano Absatz (2004-10-05): # San Juan changed from UTC-03:00 to UTC-04:00 at midnight between # Sunday, May 30th and Monday, May 31st. It changed back to UTC-03:00 # at midnight between Saturday, July 24th and Sunday, July 25th.... # http://www.sanjuan.gov.ar/prensa/archivo/000329.html # http://www.sanjuan.gov.ar/prensa/archivo/000426.html # http://www.sanjuan.gov.ar/prensa/archivo/000441.html # From Alex Krivenyshev (2008-01-17): # Here are articles that Argentina Province San Luis is planning to end DST # as earlier as upcoming Monday January 21, 2008 or February 2008: # # Provincia argentina retrasa reloj y marca diferencia con resto del país # (Argentine Province delayed clock and mark difference with the rest of the # country) # http://cl.invertia.com/noticias/noticia.aspx?idNoticia=200801171849_EFE_ET4373&idtel # # Es inminente que en San Luis atrasen una hora los relojes # (It is imminent in San Luis clocks one hour delay) # http://www.lagaceta.com.ar/nota/253414/Economia/Es-inminente-que-en-San-Luis-atrasen-una-hora-los-relojes.html # http://www.worldtimezone.net/dst_news/dst_news_argentina02.html # From Jesper Nørgaard Welen (2008-01-18): # The page of the San Luis provincial government # http://www.sanluis.gov.ar/notas.asp?idCanal=0&id=22812 # confirms what Alex Krivenyshev has earlier sent to the tz # emailing list about that San Luis plans to return to standard # time much earlier than the rest of the country. It also # confirms that upon request the provinces San Juan and Mendoza # refused to follow San Luis in this change. # # The change is supposed to take place Monday the 21st at 0:00 # hours. As far as I understand it if this goes ahead, we need # a new timezone for San Luis (although there are also documented # independent changes in the southamerica file of San Luis in # 1990 and 1991 which has not been confirmed). # From Jesper Nørgaard Welen (2008-01-25): # Unfortunately the below page has become defunct, about the San Luis # time change. Perhaps because it now is part of a group of pages "Most # important pages of 2008." # # You can use # http://www.sanluis.gov.ar/notas.asp?idCanal=8141&id=22834 # instead it seems. Or use "Buscador" from the main page of the San Luis # government, and fill in "huso" and click OK, and you will get 3 pages # from which the first one is identical to the above. # From Mariano Absatz (2008-01-28): # I can confirm that the Province of San Luis (and so far only that # province) decided to go back to UTC-3 effective midnight Jan 20th 2008 # (that is, Monday 21st at 0:00 is the time the clocks were delayed back # 1 hour), and they intend to keep UTC-3 as their timezone all year round # (that is, unless they change their mind any minute now). # # So we'll have to add yet another city to 'southamerica' (I think San # Luis city is the mos populated city in the Province, so it'd be # America/Argentina/San_Luis... of course I can't remember if San Luis's # history of particular changes goes along with Mendoza or San Juan :-( # (I only remember not being able to collect hard facts about San Luis # back in 2004, when these provinces changed to UTC-4 for a few days, I # mailed them personally and never got an answer). # From Paul Eggert (2014-08-12): # Unless otherwise specified, data entries are from Shanks & Pottenger through # 1992, from the IATA otherwise. As noted below, Shanks & Pottenger say that # America/Cordoba split into 6 subregions during 1991/1992, one of which # was America/San_Luis, but we haven't verified this yet so for now we'll # keep America/Cordoba a single region rather than splitting it into the # other 5 subregions. # From Mariano Absatz (2009-03-13): # Yesterday (with our usual 2-day notice) the Province of San Luis # decided that next Sunday instead of "staying" @utc-03:00 they will go # to utc-04:00 until the second Saturday in October... # # The press release is at # http://www.sanluis.gov.ar/SL/Paginas/NoticiaDetalle.asp?TemaId=1&InfoPrensaId=3102 # (I couldn't find the decree, but www.sanluis.gov.ar # is the official page for the Province Government.) # # There's also a note in only one of the major national papers ... # http://www.lanacion.com.ar/nota.asp?nota_id=1107912 # # The press release says [quick and dirty translation]: # ... announced that next Sunday, at 00:00, Puntanos (the San Luis # inhabitants) will have to turn back one hour their clocks # # Since then, San Luis will establish its own Province timezone. Thus, # during 2009, this timezone change will run from 00:00 the third Sunday # in March until 24:00 of the second Saturday in October. # From Mariano Absatz (2009-10-16): # ...the Province of San Luis is a case in itself. # # The Law at # http://www.diputadossanluis.gov.ar/diputadosasp/paginas/verNorma.asp?NormaID=276 # is ambiguous because establishes a calendar from the 2nd Sunday in # October at 0:00 thru the 2nd Saturday in March at 24:00 and the # complement of that starting on the 2nd Sunday of March at 0:00 and # ending on the 2nd Saturday of March at 24:00. # # This clearly breaks every time the 1st of March or October is a Sunday. # # IMHO, the "spirit of the Law" is to make the changes at 0:00 on the 2nd # Sunday of October and March. # # The problem is that the changes in the rest of the Provinces that did # change in 2007/2008, were made according to the Federal Law and Decrees # that did so on the 3rd Sunday of October and March. # # In fact, San Luis actually switched from UTC-4 to UTC-3 last Sunday # (October 11th) at 0:00. # # So I guess a new set of rules, besides "Arg", must be made and the last # America/Argentina/San_Luis entries should change to use these... # # I'm enclosing a patch that does what I say... regretfully, the San Luis # timezone must be called "WART/WARST" even when most of the time (like, # right now) WARST == ART... that is, since last Sunday, all the country # is using UTC-3, but in my patch, San Luis calls it "WARST" and the rest # of the country calls it "ART". # ... # From Alexander Krivenyshev (2010-04-09): # According to news reports from El Diario de la República Province San # Luis, Argentina (standard time UTC-04) will keep Daylight Saving Time # after April 11, 2010 - will continue to have same time as rest of # Argentina (UTC-3) (no DST). # # Confirmaron la prórroga del huso horario de verano (Spanish) # http://www.eldiariodelarepublica.com/index.php?option=com_content&task=view&id=29383&Itemid=9 # or (some English translation): # http://www.worldtimezone.com/dst_news/dst_news_argentina08.html # From Mariano Absatz (2010-04-12): # yes...I can confirm this...and given that San Luis keeps calling # UTC-03:00 "summer time", we should't just let San Luis go back to "Arg" # rules...San Luis is still using "Western ARgentina Time" and it got # stuck on Summer daylight savings time even though the summer is over. # From Paul Eggert (2013-09-05): # Perhaps San Luis operates on the legal fiction that it is at UTC-4 # with perpetual summer time, but ordinary usage typically seems to # just say it's at UTC-3; see, for example, # http://es.wikipedia.org/wiki/Hora_oficial_argentina # We've documented similar situations as being plain changes to # standard time, so let's do that here too. This does not change UTC # offsets, only tm_isdst and the time zone abbreviations. One minor # plus is that this silences a zic complaint that there's no POSIX TZ # setting for time stamps past 2038. # From Paul Eggert (2013-02-21): # Milne says Córdoba time was -4:16:48.2. Round to the nearest second. # Zone NAME GMTOFF RULES FORMAT [UNTIL] # # Buenos Aires (BA), Capital Federal (CF), Zone America/Argentina/Buenos_Aires -3:53:48 - LMT 1894 Oct 31 -4:16:48 - CMT 1920 May # Córdoba Mean Time -4:00 - ART 1930 Dec -4:00 Arg AR%sT 1969 Oct 5 -3:00 Arg AR%sT 1999 Oct 3 -4:00 Arg AR%sT 2000 Mar 3 -3:00 Arg AR%sT # # Córdoba (CB), Santa Fe (SF), Entre Ríos (ER), Corrientes (CN), Misiones (MN), # Chaco (CC), Formosa (FM), Santiago del Estero (SE) # # Shanks & Pottenger also make the following claims, which we haven't verified: # - Formosa switched to -3:00 on 1991-01-07. # - Misiones switched to -3:00 on 1990-12-29. # - Chaco switched to -3:00 on 1991-01-04. # - Santiago del Estero switched to -4:00 on 1991-04-01, # then to -3:00 on 1991-04-26. # Zone America/Argentina/Cordoba -4:16:48 - LMT 1894 Oct 31 -4:16:48 - CMT 1920 May -4:00 - ART 1930 Dec -4:00 Arg AR%sT 1969 Oct 5 -3:00 Arg AR%sT 1991 Mar 3 -4:00 - WART 1991 Oct 20 -3:00 Arg AR%sT 1999 Oct 3 -4:00 Arg AR%sT 2000 Mar 3 -3:00 Arg AR%sT # # Salta (SA), La Pampa (LP), Neuquén (NQ), Rio Negro (RN) Zone America/Argentina/Salta -4:21:40 - LMT 1894 Oct 31 -4:16:48 - CMT 1920 May -4:00 - ART 1930 Dec -4:00 Arg AR%sT 1969 Oct 5 -3:00 Arg AR%sT 1991 Mar 3 -4:00 - WART 1991 Oct 20 -3:00 Arg AR%sT 1999 Oct 3 -4:00 Arg AR%sT 2000 Mar 3 -3:00 Arg AR%sT 2008 Oct 18 -3:00 - ART # # Tucumán (TM) Zone America/Argentina/Tucuman -4:20:52 - LMT 1894 Oct 31 -4:16:48 - CMT 1920 May -4:00 - ART 1930 Dec -4:00 Arg AR%sT 1969 Oct 5 -3:00 Arg AR%sT 1991 Mar 3 -4:00 - WART 1991 Oct 20 -3:00 Arg AR%sT 1999 Oct 3 -4:00 Arg AR%sT 2000 Mar 3 -3:00 - ART 2004 Jun 1 -4:00 - WART 2004 Jun 13 -3:00 Arg AR%sT # # La Rioja (LR) Zone America/Argentina/La_Rioja -4:27:24 - LMT 1894 Oct 31 -4:16:48 - CMT 1920 May -4:00 - ART 1930 Dec -4:00 Arg AR%sT 1969 Oct 5 -3:00 Arg AR%sT 1991 Mar 1 -4:00 - WART 1991 May 7 -3:00 Arg AR%sT 1999 Oct 3 -4:00 Arg AR%sT 2000 Mar 3 -3:00 - ART 2004 Jun 1 -4:00 - WART 2004 Jun 20 -3:00 Arg AR%sT 2008 Oct 18 -3:00 - ART # # San Juan (SJ) Zone America/Argentina/San_Juan -4:34:04 - LMT 1894 Oct 31 -4:16:48 - CMT 1920 May -4:00 - ART 1930 Dec -4:00 Arg AR%sT 1969 Oct 5 -3:00 Arg AR%sT 1991 Mar 1 -4:00 - WART 1991 May 7 -3:00 Arg AR%sT 1999 Oct 3 -4:00 Arg AR%sT 2000 Mar 3 -3:00 - ART 2004 May 31 -4:00 - WART 2004 Jul 25 -3:00 Arg AR%sT 2008 Oct 18 -3:00 - ART # # Jujuy (JY) Zone America/Argentina/Jujuy -4:21:12 - LMT 1894 Oct 31 -4:16:48 - CMT 1920 May -4:00 - ART 1930 Dec -4:00 Arg AR%sT 1969 Oct 5 -3:00 Arg AR%sT 1990 Mar 4 -4:00 - WART 1990 Oct 28 -4:00 1:00 WARST 1991 Mar 17 -4:00 - WART 1991 Oct 6 -3:00 1:00 ARST 1992 -3:00 Arg AR%sT 1999 Oct 3 -4:00 Arg AR%sT 2000 Mar 3 -3:00 Arg AR%sT 2008 Oct 18 -3:00 - ART # # Catamarca (CT), Chubut (CH) Zone America/Argentina/Catamarca -4:23:08 - LMT 1894 Oct 31 -4:16:48 - CMT 1920 May -4:00 - ART 1930 Dec -4:00 Arg AR%sT 1969 Oct 5 -3:00 Arg AR%sT 1991 Mar 3 -4:00 - WART 1991 Oct 20 -3:00 Arg AR%sT 1999 Oct 3 -4:00 Arg AR%sT 2000 Mar 3 -3:00 - ART 2004 Jun 1 -4:00 - WART 2004 Jun 20 -3:00 Arg AR%sT 2008 Oct 18 -3:00 - ART # # Mendoza (MZ) Zone America/Argentina/Mendoza -4:35:16 - LMT 1894 Oct 31 -4:16:48 - CMT 1920 May -4:00 - ART 1930 Dec -4:00 Arg AR%sT 1969 Oct 5 -3:00 Arg AR%sT 1990 Mar 4 -4:00 - WART 1990 Oct 15 -4:00 1:00 WARST 1991 Mar 1 -4:00 - WART 1991 Oct 15 -4:00 1:00 WARST 1992 Mar 1 -4:00 - WART 1992 Oct 18 -3:00 Arg AR%sT 1999 Oct 3 -4:00 Arg AR%sT 2000 Mar 3 -3:00 - ART 2004 May 23 -4:00 - WART 2004 Sep 26 -3:00 Arg AR%sT 2008 Oct 18 -3:00 - ART # # San Luis (SL) Rule SanLuis 2008 2009 - Mar Sun>=8 0:00 0 - Rule SanLuis 2007 2008 - Oct Sun>=8 0:00 1:00 S Zone America/Argentina/San_Luis -4:25:24 - LMT 1894 Oct 31 -4:16:48 - CMT 1920 May -4:00 - ART 1930 Dec -4:00 Arg AR%sT 1969 Oct 5 -3:00 Arg AR%sT 1990 -3:00 1:00 ARST 1990 Mar 14 -4:00 - WART 1990 Oct 15 -4:00 1:00 WARST 1991 Mar 1 -4:00 - WART 1991 Jun 1 -3:00 - ART 1999 Oct 3 -4:00 1:00 WARST 2000 Mar 3 -3:00 - ART 2004 May 31 -4:00 - WART 2004 Jul 25 -3:00 Arg AR%sT 2008 Jan 21 -4:00 SanLuis WAR%sT 2009 Oct 11 -3:00 - ART # # Santa Cruz (SC) Zone America/Argentina/Rio_Gallegos -4:36:52 - LMT 1894 Oct 31 -4:16:48 - CMT 1920 May # Córdoba Mean Time -4:00 - ART 1930 Dec -4:00 Arg AR%sT 1969 Oct 5 -3:00 Arg AR%sT 1999 Oct 3 -4:00 Arg AR%sT 2000 Mar 3 -3:00 - ART 2004 Jun 1 -4:00 - WART 2004 Jun 20 -3:00 Arg AR%sT 2008 Oct 18 -3:00 - ART # # Tierra del Fuego, Antártida e Islas del Atlántico Sur (TF) Zone America/Argentina/Ushuaia -4:33:12 - LMT 1894 Oct 31 -4:16:48 - CMT 1920 May # Córdoba Mean Time -4:00 - ART 1930 Dec -4:00 Arg AR%sT 1969 Oct 5 -3:00 Arg AR%sT 1999 Oct 3 -4:00 Arg AR%sT 2000 Mar 3 -3:00 - ART 2004 May 30 -4:00 - WART 2004 Jun 20 -3:00 Arg AR%sT 2008 Oct 18 -3:00 - ART # Aruba Link America/Curacao America/Aruba # Bolivia # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/La_Paz -4:32:36 - LMT 1890 -4:32:36 - CMT 1931 Oct 15 # Calamarca MT -4:32:36 1:00 BOST 1932 Mar 21 # Bolivia ST -4:00 - BOT # Bolivia Time # Brazil # From Paul Eggert (1993-11-18): # The mayor of Rio recently attempted to change the time zone rules # just in his city, in order to leave more summer time for the tourist trade. # The rule change lasted only part of the day; # the federal government refused to follow the city's rules, and business # was in a chaos, so the mayor backed down that afternoon. # From IATA SSIM (1996-02): # _Only_ the following states in BR1 observe DST: Rio Grande do Sul (RS), # Santa Catarina (SC), Paraná (PR), São Paulo (SP), Rio de Janeiro (RJ), # Espírito Santo (ES), Minas Gerais (MG), Bahia (BA), Goiás (GO), # Distrito Federal (DF), Tocantins (TO), Sergipe [SE] and Alagoas [AL]. # [The last three states are new to this issue of the IATA SSIM.] # From Gwillim Law (1996-10-07): # Geography, history (Tocantins was part of Goiás until 1989), and other # sources of time zone information lead me to believe that AL, SE, and TO were # always in BR1, and so the only change was whether or not they observed DST.... # The earliest issue of the SSIM I have is 2/91. Each issue from then until # 9/95 says that DST is observed only in the ten states I quoted from 9/95, # along with Mato Grosso (MT) and Mato Grosso do Sul (MS), which are in BR2 # (UTC-4).... The other two time zones given for Brazil are BR3, which is # UTC-5, no DST, and applies only in the state of Acre (AC); and BR4, which is # UTC-2, and applies to Fernando de Noronha (formerly FN, but I believe it's # become part of the state of Pernambuco). The boundary between BR1 and BR2 # has never been clearly stated. They've simply been called East and West. # However, some conclusions can be drawn from another IATA manual: the Airline # Coding Directory, which lists close to 400 airports in Brazil. For each # airport it gives a time zone which is coded to the SSIM. From that # information, I'm led to conclude that the states of Amapá (AP), Ceará (CE), # Maranhão (MA), Paraíba (PR), Pernambuco (PE), Piauí (PI), and Rio Grande do # Norte (RN), and the eastern part of Pará (PA) are all in BR1 without DST. # From Marcos Tadeu (1998-09-27): # Brazilian official page # From Jesper Nørgaard (2000-11-03): # [For an official list of which regions in Brazil use which time zones, see:] # http://pcdsh01.on.br/Fusbr.htm # http://pcdsh01.on.br/Fusbrhv.htm # From Celso Doria via David Madeo (2002-10-09): # The reason for the delay this year has to do with elections in Brazil. # # Unlike in the United States, elections in Brazil are 100% computerized and # the results are known almost immediately. Yesterday, it was the first # round of the elections when 115 million Brazilians voted for President, # Governor, Senators, Federal Deputies, and State Deputies. Nobody is # counting (or re-counting) votes anymore and we know there will be a second # round for the Presidency and also for some Governors. The 2nd round will # take place on October 27th. # # The reason why the DST will only begin November 3rd is that the thousands # of electoral machines used cannot have their time changed, and since the # Constitution says the elections must begin at 8:00 AM and end at 5:00 PM, # the Government decided to postpone DST, instead of changing the Constitution # (maybe, for the next elections, it will be possible to change the clock)... # From Rodrigo Severo (2004-10-04): # It's just the biannual change made necessary by the much hyped, supposedly # modern Brazilian eletronic voting machines which, apparently, can't deal # with a time change between the first and the second rounds of the elections. # From Steffen Thorsen (2007-09-20): # Brazil will start DST on 2007-10-14 00:00 and end on 2008-02-17 00:00: # http://www.mme.gov.br/site/news/detail.do;jsessionid=BBA06811AFCAAC28F0285210913513DA?newsId=13975 # From Paul Schulze (2008-06-24): # ...by law number 11.662 of April 24, 2008 (published in the "Diario # Oficial da União"...) in Brazil there are changes in the timezones, # effective today (00:00am at June 24, 2008) as follows: # # a) The timezone UTC+5 is extinguished, with all the Acre state and the # part of the Amazonas state that had this timezone now being put to the # timezone UTC+4 # b) The whole Pará state now is put at timezone UTC+3, instead of just # part of it, as was before. # # This change follows a proposal of senator Tiao Viana of Acre state, that # proposed it due to concerns about open television channels displaying # programs inappropriate to youths in the states that had the timezone # UTC+5 too early in the night. In the occasion, some more corrections # were proposed, trying to unify the timezones of any given state. This # change modifies timezone rules defined in decree 2.784 of 18 June, # 1913. # From Rodrigo Severo (2008-06-24): # Just correcting the URL: # https://www.in.gov.br/imprensa/visualiza/index.jsp?jornal=do&secao=1&pagina=1&data=25/04/2008 # # As a result of the above Decree I believe the America/Rio_Branco # timezone shall be modified from UTC-5 to UTC-4 and a new timezone shall # be created to represent the...west side of the Pará State. I # suggest this new timezone be called Santarem as the most # important/populated city in the affected area. # # This new timezone would be the same as the Rio_Branco timezone up to # the 2008/06/24 change which would be to UTC-3 instead of UTC-4. # From Alex Krivenyshev (2008-06-24): # This is a quick reference page for New and Old Brazil Time Zones map. # http://www.worldtimezone.com/brazil-time-new-old.php # # - 4 time zones replaced by 3 time zones - eliminating time zone UTC-05 # (state Acre and the part of the Amazonas will be UTC/GMT-04) - western # part of Par state is moving to one timezone UTC-03 (from UTC-04). # From Paul Eggert (2002-10-10): # The official decrees referenced below are mostly taken from # Decretos sobre o Horário de Verão no Brasil. # http://pcdsh01.on.br/DecHV.html # From Steffen Thorsen (2008-08-29): # As announced by the government and many newspapers in Brazil late # yesterday, Brazil will start DST on 2008-10-19 (need to change rule) and # it will end on 2009-02-15 (current rule for Brazil is fine). Based on # past years experience with the elections, there was a good chance that # the start was postponed to November, but it did not happen this year. # # It has not yet been posted to http://pcdsh01.on.br/DecHV.html # # An official page about it: # http://www.mme.gov.br/site/news/detail.do?newsId=16722 # Note that this link does not always work directly, but must be accessed # by going to # http://www.mme.gov.br/first # # One example link that works directly: # http://jornale.com.br/index.php?option=com_content&task=view&id=13530&Itemid=54 # (Portuguese) # # We have a written a short article about it as well: # http://www.timeanddate.com/news/time/brazil-dst-2008-2009.html # # From Alexander Krivenyshev (2011-10-04): # State Bahia will return to Daylight savings time this year after 8 years off. # The announcement was made by Governor Jaques Wagner in an interview to a # television station in Salvador. # In Portuguese: # http://g1.globo.com/bahia/noticia/2011/10/governador-jaques-wagner-confirma-horario-de-verao-na-bahia.html # http://noticias.terra.com.br/brasil/noticias/0,,OI5390887-EI8139,00-Bahia+volta+a+ter+horario+de+verao+apos+oito+anos.html # From Guilherme Bernardes Rodrigues (2011-10-07): # There is news in the media, however there is still no decree about it. # I just send a e-mail to Zulmira Brandao at http://pcdsh01.on.br/ the # official agency about time in Brazil, and she confirmed that the old rule is # still in force. # From Guilherme Bernardes Rodrigues (2011-10-14) # It's official, the President signed a decree that includes Bahia in summer # time. # [ and in a second message (same day): ] # I found the decree. # # DECRETO No- 7.584, DE 13 DE OUTUBRO DE 2011 # Link : # http://www.in.gov.br/visualiza/index.jsp?data=13/10/2011&jornal=1000&pagina=6&totalArquivos=6 # From Kelley Cook (2012-10-16): # The governor of state of Bahia in Brazil announced on Thursday that # due to public pressure, he is reversing the DST policy they implemented # last year and will not be going to Summer Time on October 21st.... # http://www.correio24horas.com.br/r/artigo/apos-pressoes-wagner-suspende-horario-de-verao-na-bahia # From Rodrigo Severo (2012-10-16): # Tocantins state will have DST. # http://noticias.terra.com.br/brasil/noticias/0,,OI6232536-EI306.html # From Steffen Thorsen (2013-09-20): # Tocantins in Brazil is very likely not to observe DST from October.... # http://conexaoto.com.br/2013/09/18/ministerio-confirma-que-tocantins-esta-fora-do-horario-de-verao-em-2013-mas-falta-publicacao-de-decreto # We will keep this article updated when this is confirmed: # http://www.timeanddate.com/news/time/brazil-starts-dst-2013.html # From Steffen Thorsen (2013-10-17): # http://www.timeanddate.com/news/time/acre-amazonas-change-time-zone.html # Senator Jorge Viana announced that Acre will change time zone on November 10. # He did not specify the time of the change, nor if western parts of Amazonas # will change as well. # # From Paul Eggert (2013-10-17): # For now, assume western Amazonas will change as well. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S # Decree 20,466 (1931-10-01) # Decree 21,896 (1932-01-10) Rule Brazil 1931 only - Oct 3 11:00 1:00 S Rule Brazil 1932 1933 - Apr 1 0:00 0 - Rule Brazil 1932 only - Oct 3 0:00 1:00 S # Decree 23,195 (1933-10-10) # revoked DST. # Decree 27,496 (1949-11-24) # Decree 27,998 (1950-04-13) Rule Brazil 1949 1952 - Dec 1 0:00 1:00 S Rule Brazil 1950 only - Apr 16 1:00 0 - Rule Brazil 1951 1952 - Apr 1 0:00 0 - # Decree 32,308 (1953-02-24) Rule Brazil 1953 only - Mar 1 0:00 0 - # Decree 34,724 (1953-11-30) # revoked DST. # Decree 52,700 (1963-10-18) # established DST from 1963-10-23 00:00 to 1964-02-29 00:00 # in SP, RJ, GB, MG, ES, due to the prolongation of the drought. # Decree 53,071 (1963-12-03) # extended the above decree to all of the national territory on 12-09. Rule Brazil 1963 only - Dec 9 0:00 1:00 S # Decree 53,604 (1964-02-25) # extended summer time by one day to 1964-03-01 00:00 (start of school). Rule Brazil 1964 only - Mar 1 0:00 0 - # Decree 55,639 (1965-01-27) Rule Brazil 1965 only - Jan 31 0:00 1:00 S Rule Brazil 1965 only - Mar 31 0:00 0 - # Decree 57,303 (1965-11-22) Rule Brazil 1965 only - Dec 1 0:00 1:00 S # Decree 57,843 (1966-02-18) Rule Brazil 1966 1968 - Mar 1 0:00 0 - Rule Brazil 1966 1967 - Nov 1 0:00 1:00 S # Decree 63,429 (1968-10-15) # revoked DST. # Decree 91,698 (1985-09-27) Rule Brazil 1985 only - Nov 2 0:00 1:00 S # Decree 92,310 (1986-01-21) # Decree 92,463 (1986-03-13) Rule Brazil 1986 only - Mar 15 0:00 0 - # Decree 93,316 (1986-10-01) Rule Brazil 1986 only - Oct 25 0:00 1:00 S Rule Brazil 1987 only - Feb 14 0:00 0 - # Decree 94,922 (1987-09-22) Rule Brazil 1987 only - Oct 25 0:00 1:00 S Rule Brazil 1988 only - Feb 7 0:00 0 - # Decree 96,676 (1988-09-12) # except for the states of AC, AM, PA, RR, RO, and AP (then a territory) Rule Brazil 1988 only - Oct 16 0:00 1:00 S Rule Brazil 1989 only - Jan 29 0:00 0 - # Decree 98,077 (1989-08-21) # with the same exceptions Rule Brazil 1989 only - Oct 15 0:00 1:00 S Rule Brazil 1990 only - Feb 11 0:00 0 - # Decree 99,530 (1990-09-17) # adopted by RS, SC, PR, SP, RJ, ES, MG, GO, MS, DF. # Decree 99,629 (1990-10-19) adds BA, MT. Rule Brazil 1990 only - Oct 21 0:00 1:00 S Rule Brazil 1991 only - Feb 17 0:00 0 - # Unnumbered decree (1991-09-25) # adopted by RS, SC, PR, SP, RJ, ES, MG, BA, GO, MT, MS, DF. Rule Brazil 1991 only - Oct 20 0:00 1:00 S Rule Brazil 1992 only - Feb 9 0:00 0 - # Unnumbered decree (1992-10-16) # adopted by same states. Rule Brazil 1992 only - Oct 25 0:00 1:00 S Rule Brazil 1993 only - Jan 31 0:00 0 - # Decree 942 (1993-09-28) # adopted by same states, plus AM. # Decree 1,252 (1994-09-22; # web page corrected 2004-01-07) adopted by same states, minus AM. # Decree 1,636 (1995-09-14) # adopted by same states, plus MT and TO. # Decree 1,674 (1995-10-13) # adds AL, SE. Rule Brazil 1993 1995 - Oct Sun>=11 0:00 1:00 S Rule Brazil 1994 1995 - Feb Sun>=15 0:00 0 - Rule Brazil 1996 only - Feb 11 0:00 0 - # Decree 2,000 (1996-09-04) # adopted by same states, minus AL, SE. Rule Brazil 1996 only - Oct 6 0:00 1:00 S Rule Brazil 1997 only - Feb 16 0:00 0 - # From Daniel C. Sobral (1998-02-12): # In 1997, the DS began on October 6. The stated reason was that # because international television networks ignored Brazil's policy on DS, # they bought the wrong times on satellite for coverage of Pope's visit. # This year, the ending date of DS was postponed to March 1 # to help dealing with the shortages of electric power. # # Decree 2,317 (1997-09-04), adopted by same states. Rule Brazil 1997 only - Oct 6 0:00 1:00 S # Decree 2,495 # (1998-02-10) Rule Brazil 1998 only - Mar 1 0:00 0 - # Decree 2,780 (1998-09-11) # adopted by the same states as before. Rule Brazil 1998 only - Oct 11 0:00 1:00 S Rule Brazil 1999 only - Feb 21 0:00 0 - # Decree 3,150 # (1999-08-23) adopted by same states. # Decree 3,188 (1999-09-30) # adds SE, AL, PB, PE, RN, CE, PI, MA and RR. Rule Brazil 1999 only - Oct 3 0:00 1:00 S Rule Brazil 2000 only - Feb 27 0:00 0 - # Decree 3,592 (2000-09-06) # adopted by the same states as before. # Decree 3,630 (2000-10-13) # repeals DST in PE and RR, effective 2000-10-15 00:00. # Decree 3,632 (2000-10-17) # repeals DST in SE, AL, PB, RN, CE, PI and MA, effective 2000-10-22 00:00. # Decree 3,916 # (2001-09-13) reestablishes DST in AL, CE, MA, PB, PE, PI, RN, SE. Rule Brazil 2000 2001 - Oct Sun>=8 0:00 1:00 S Rule Brazil 2001 2006 - Feb Sun>=15 0:00 0 - # Decree 4,399 (2002-10-01) repeals DST in AL, CE, MA, PB, PE, PI, RN, SE. # 4,399 Rule Brazil 2002 only - Nov 3 0:00 1:00 S # Decree 4,844 (2003-09-24; corrected 2003-09-26) repeals DST in BA, MT, TO. # 4,844 Rule Brazil 2003 only - Oct 19 0:00 1:00 S # Decree 5,223 (2004-10-01) reestablishes DST in MT. # 5,223 Rule Brazil 2004 only - Nov 2 0:00 1:00 S # Decree 5,539 (2005-09-19), # adopted by the same states as before. Rule Brazil 2005 only - Oct 16 0:00 1:00 S # Decree 5,920 (2006-10-03), # adopted by the same states as before. Rule Brazil 2006 only - Nov 5 0:00 1:00 S Rule Brazil 2007 only - Feb 25 0:00 0 - # Decree 6,212 (2007-09-26), # adopted by the same states as before. Rule Brazil 2007 only - Oct Sun>=8 0:00 1:00 S # From Frederico A. C. Neves (2008-09-10): # According to this decree # http://www.planalto.gov.br/ccivil_03/_Ato2007-2010/2008/Decreto/D6558.htm # [t]he DST period in Brazil now on will be from the 3rd Oct Sunday to the # 3rd Feb Sunday. There is an exception on the return date when this is # the Carnival Sunday then the return date will be the next Sunday... Rule Brazil 2008 max - Oct Sun>=15 0:00 1:00 S Rule Brazil 2008 2011 - Feb Sun>=15 0:00 0 - Rule Brazil 2012 only - Feb Sun>=22 0:00 0 - Rule Brazil 2013 2014 - Feb Sun>=15 0:00 0 - Rule Brazil 2015 only - Feb Sun>=22 0:00 0 - Rule Brazil 2016 2022 - Feb Sun>=15 0:00 0 - Rule Brazil 2023 only - Feb Sun>=22 0:00 0 - Rule Brazil 2024 2025 - Feb Sun>=15 0:00 0 - Rule Brazil 2026 only - Feb Sun>=22 0:00 0 - Rule Brazil 2027 2033 - Feb Sun>=15 0:00 0 - Rule Brazil 2034 only - Feb Sun>=22 0:00 0 - Rule Brazil 2035 2036 - Feb Sun>=15 0:00 0 - Rule Brazil 2037 only - Feb Sun>=22 0:00 0 - # From Arthur David Olson (2008-09-29): # The next is wrong in some years but is better than nothing. Rule Brazil 2038 max - Feb Sun>=15 0:00 0 - # The latest ruleset listed above says that the following states observe DST: # DF, ES, GO, MG, MS, MT, PR, RJ, RS, SC, SP. # Zone NAME GMTOFF RULES FORMAT [UNTIL] # # Fernando de Noronha (administratively part of PE) Zone America/Noronha -2:09:40 - LMT 1914 -2:00 Brazil FN%sT 1990 Sep 17 -2:00 - FNT 1999 Sep 30 -2:00 Brazil FN%sT 2000 Oct 15 -2:00 - FNT 2001 Sep 13 -2:00 Brazil FN%sT 2002 Oct 1 -2:00 - FNT # Other Atlantic islands have no permanent settlement. # These include Trindade and Martim Vaz (administratively part of ES), # Rocas Atoll (RN), and the St Peter and St Paul Archipelago (PE). # Fernando de Noronha was a separate territory from 1942-09-02 to 1989-01-01; # it also included the Penedos. # # Amapá (AP), east Pará (PA) # East Pará includes Belém, Marabá, Serra Norte, and São Félix do Xingu. # The division between east and west Pará is the river Xingu. # In the north a very small part from the river Javary (now Jari I guess, # the border with Amapá) to the Amazon, then to the Xingu. Zone America/Belem -3:13:56 - LMT 1914 -3:00 Brazil BR%sT 1988 Sep 12 -3:00 - BRT # # west Pará (PA) # West Pará includes Altamira, Óbidos, Prainha, Oriximiná, and Santarém. Zone America/Santarem -3:38:48 - LMT 1914 -4:00 Brazil AM%sT 1988 Sep 12 -4:00 - AMT 2008 Jun 24 0:00 -3:00 - BRT # # Maranhão (MA), Piauí (PI), Ceará (CE), Rio Grande do Norte (RN), # Paraíba (PB) Zone America/Fortaleza -2:34:00 - LMT 1914 -3:00 Brazil BR%sT 1990 Sep 17 -3:00 - BRT 1999 Sep 30 -3:00 Brazil BR%sT 2000 Oct 22 -3:00 - BRT 2001 Sep 13 -3:00 Brazil BR%sT 2002 Oct 1 -3:00 - BRT # # Pernambuco (PE) (except Atlantic islands) Zone America/Recife -2:19:36 - LMT 1914 -3:00 Brazil BR%sT 1990 Sep 17 -3:00 - BRT 1999 Sep 30 -3:00 Brazil BR%sT 2000 Oct 15 -3:00 - BRT 2001 Sep 13 -3:00 Brazil BR%sT 2002 Oct 1 -3:00 - BRT # # Tocantins (TO) Zone America/Araguaina -3:12:48 - LMT 1914 -3:00 Brazil BR%sT 1990 Sep 17 -3:00 - BRT 1995 Sep 14 -3:00 Brazil BR%sT 2003 Sep 24 -3:00 - BRT 2012 Oct 21 -3:00 Brazil BR%sT 2013 Sep -3:00 - BRT # # Alagoas (AL), Sergipe (SE) Zone America/Maceio -2:22:52 - LMT 1914 -3:00 Brazil BR%sT 1990 Sep 17 -3:00 - BRT 1995 Oct 13 -3:00 Brazil BR%sT 1996 Sep 4 -3:00 - BRT 1999 Sep 30 -3:00 Brazil BR%sT 2000 Oct 22 -3:00 - BRT 2001 Sep 13 -3:00 Brazil BR%sT 2002 Oct 1 -3:00 - BRT # # Bahia (BA) # There are too many Salvadors elsewhere, so use America/Bahia instead # of America/Salvador. Zone America/Bahia -2:34:04 - LMT 1914 -3:00 Brazil BR%sT 2003 Sep 24 -3:00 - BRT 2011 Oct 16 -3:00 Brazil BR%sT 2012 Oct 21 -3:00 - BRT # # Goiás (GO), Distrito Federal (DF), Minas Gerais (MG), # Espírito Santo (ES), Rio de Janeiro (RJ), São Paulo (SP), Paraná (PR), # Santa Catarina (SC), Rio Grande do Sul (RS) Zone America/Sao_Paulo -3:06:28 - LMT 1914 -3:00 Brazil BR%sT 1963 Oct 23 0:00 -3:00 1:00 BRST 1964 -3:00 Brazil BR%sT # # Mato Grosso do Sul (MS) Zone America/Campo_Grande -3:38:28 - LMT 1914 -4:00 Brazil AM%sT # # Mato Grosso (MT) Zone America/Cuiaba -3:44:20 - LMT 1914 -4:00 Brazil AM%sT 2003 Sep 24 -4:00 - AMT 2004 Oct 1 -4:00 Brazil AM%sT # # Rondônia (RO) Zone America/Porto_Velho -4:15:36 - LMT 1914 -4:00 Brazil AM%sT 1988 Sep 12 -4:00 - AMT # # Roraima (RR) Zone America/Boa_Vista -4:02:40 - LMT 1914 -4:00 Brazil AM%sT 1988 Sep 12 -4:00 - AMT 1999 Sep 30 -4:00 Brazil AM%sT 2000 Oct 15 -4:00 - AMT # # east Amazonas (AM): Boca do Acre, Jutaí, Manaus, Floriano Peixoto # The great circle line from Tabatinga to Porto Acre divides # east from west Amazonas. Zone America/Manaus -4:00:04 - LMT 1914 -4:00 Brazil AM%sT 1988 Sep 12 -4:00 - AMT 1993 Sep 28 -4:00 Brazil AM%sT 1994 Sep 22 -4:00 - AMT # # west Amazonas (AM): Atalaia do Norte, Boca do Maoco, Benjamin Constant, # Eirunepé, Envira, Ipixuna Zone America/Eirunepe -4:39:28 - LMT 1914 -5:00 Brazil AC%sT 1988 Sep 12 -5:00 - ACT 1993 Sep 28 -5:00 Brazil AC%sT 1994 Sep 22 -5:00 - ACT 2008 Jun 24 0:00 -4:00 - AMT 2013 Nov 10 -5:00 - ACT # # Acre (AC) Zone America/Rio_Branco -4:31:12 - LMT 1914 -5:00 Brazil AC%sT 1988 Sep 12 -5:00 - ACT 2008 Jun 24 0:00 -4:00 - AMT 2013 Nov 10 -5:00 - ACT # Chile # From Eduardo Krell (1995-10-19): # The law says to switch to DST at midnight [24:00] on the second SATURDAY # of October.... The law is the same for March and October. # (1998-09-29): # Because of the drought this year, the government decided to go into # DST earlier (saturday 9/26 at 24:00). This is a one-time change only ... # (unless there's another dry season next year, I guess). # From Julio I. Pacheco Troncoso (1999-03-18): # Because of the same drought, the government decided to end DST later, # on April 3, (one-time change). # From Oscar van Vlijmen (2006-10-08): # http://www.horaoficial.cl/cambio.htm # From Jesper Nørgaard Welen (2006-10-08): # I think that there are some obvious mistakes in the suggested link # from Oscar van Vlijmen,... for instance entry 66 says that GMT-4 # ended 1990-09-12 while entry 67 only begins GMT-3 at 1990-09-15 # (they should have been 1990-09-15 and 1990-09-16 respectively), but # anyhow it clears up some doubts too. # From Paul Eggert (2014-08-12): # The following data entries for Chile and America/Santiago are from # (2006-09-20), transcribed by # Jesper Nørgaard Welen. The data entries for Pacific/Easter are from Shanks # & Pottenger, except with DST transitions after 1932 cloned from # America/Santiago. The pre-1980 Pacific/Easter data entries are dubious, # but we have no other source. # From Germán Poo-Caamaño (2008-03-03): # Due to drought, Chile extends Daylight Time in three weeks. This # is one-time change (Saturday 3/29 at 24:00 for America/Santiago # and Saturday 3/29 at 22:00 for Pacific/Easter) # The Supreme Decree is located at # http://www.shoa.cl/servicios/supremo316.pdf # and the instructions for 2008 are located in: # http://www.horaoficial.cl/cambio.htm # From José Miguel Garrido (2008-03-05): # ... # You could see the announces of the change on # http://www.shoa.cl/noticias/2008/04hora/hora.htm # From Angel Chiang (2010-03-04): # Subject: DST in Chile exceptionally extended to 3 April due to earthquake # http://www.gobiernodechile.cl/viewNoticia.aspx?idArticulo=30098 # (in Spanish, last paragraph). # # This is breaking news. There should be more information available later. # From Arthur David Olson (2010-03-06): # Angel Chiang's message confirmed by Julio Pacheco; Julio provided a patch. # From Glenn Eychaner (2011-03-02): # It appears that the Chilean government has decided to postpone the # change from summer time to winter time again, by three weeks to April # 2nd: # http://www.emol.com/noticias/nacional/detalle/detallenoticias.asp?idnoticia=467651 # # This is not yet reflected in the official "cambio de hora" site, but # probably will be soon: # http://www.horaoficial.cl/cambio.htm # From Arthur David Olson (2011-03-02): # The emol.com article mentions a water shortage as the cause of the # postponement, which may mean that it's not a permanent change. # From Glenn Eychaner (2011-03-28): # The article: # http://diario.elmercurio.com/2011/03/28/_portada/_portada/noticias/7565897A-CA86-49E6-9E03-660B21A4883E.htm?id=3D{7565897A-CA86-49E6-9E03-660B21A4883E} # # In English: # Chile's clocks will go back an hour this year on the 7th of May instead # of this Saturday. They will go forward again the 3rd Saturday in # August, not in October as they have since 1968. This is a pilot plan # which will be reevaluated in 2012. # From Mauricio Parada (2012-02-22), translated by Glenn Eychaner (2012-02-23): # As stated in the website of the Chilean Energy Ministry # http://www.minenergia.cl/ministerio/noticias/generales/gobierno-anuncia-fechas-de-cambio-de.html # The Chilean Government has decided to postpone the entrance into winter time # (to leave DST) from March 11 2012 to April 28th 2012. The decision has not # been yet formalized but it will within the next days. # Quote from the website communication: # # 6. For the year 2012, the dates of entry into winter time will be as follows: # a. Saturday April 28, 2012, clocks should go back 60 minutes; that is, at # 23:59:59, instead of passing to 0:00, the time should be adjusted to be 23:00 # of the same day. # b. Saturday, September 1, 2012, clocks should go forward 60 minutes; that is, # at 23:59:59, instead of passing to 0:00, the time should be adjusted to be # 01:00 on September 2. # From Steffen Thorsen (2013-02-15): # According to several news sources, Chile has extended DST this year, # they will end DST later and start DST earlier than planned. They # hope to save energy. The new end date is 2013-04-28 00:00 and new # start date is 2013-09-08 00:00.... # http://www.gob.cl/informa/2013/02/15/gobierno-anuncia-fechas-de-cambio-de-hora-para-el-ano-2013.htm # From José Miguel Garrido (2014-02-19): # Today appeared in the Diario Oficial a decree amending the time change # dates to 2014. # DST End: last Saturday of April 2014 (Sun 27 Apr 2014 03:00 UTC) # DST Start: first Saturday of September 2014 (Sun 07 Sep 2014 04:00 UTC) # http://www.diariooficial.interior.gob.cl//media/2014/02/19/do-20140219.pdf +# From Juan Correa (2015-01-28): +# ... today the Ministry of Energy announced that Chile will drop DST, will keep +# "summer time" (UTC -3 / UTC -5) all year round.... +# http://www.minenergia.cl/ministerio/noticias/generales/ministerio-de-energia-anuncia.html + # NOTE: ChileAQ rules for Antarctic bases are stored separately in the # 'antarctica' file. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Chile 1927 1932 - Sep 1 0:00 1:00 S Rule Chile 1928 1932 - Apr 1 0:00 0 - Rule Chile 1942 only - Jun 1 4:00u 0 - Rule Chile 1942 only - Aug 1 5:00u 1:00 S Rule Chile 1946 only - Jul 15 4:00u 1:00 S Rule Chile 1946 only - Sep 1 3:00u 0:00 - Rule Chile 1947 only - Apr 1 4:00u 0 - Rule Chile 1968 only - Nov 3 4:00u 1:00 S Rule Chile 1969 only - Mar 30 3:00u 0 - Rule Chile 1969 only - Nov 23 4:00u 1:00 S Rule Chile 1970 only - Mar 29 3:00u 0 - Rule Chile 1971 only - Mar 14 3:00u 0 - Rule Chile 1970 1972 - Oct Sun>=9 4:00u 1:00 S Rule Chile 1972 1986 - Mar Sun>=9 3:00u 0 - Rule Chile 1973 only - Sep 30 4:00u 1:00 S Rule Chile 1974 1987 - Oct Sun>=9 4:00u 1:00 S Rule Chile 1987 only - Apr 12 3:00u 0 - Rule Chile 1988 1989 - Mar Sun>=9 3:00u 0 - Rule Chile 1988 only - Oct Sun>=1 4:00u 1:00 S Rule Chile 1989 only - Oct Sun>=9 4:00u 1:00 S Rule Chile 1990 only - Mar 18 3:00u 0 - Rule Chile 1990 only - Sep 16 4:00u 1:00 S Rule Chile 1991 1996 - Mar Sun>=9 3:00u 0 - Rule Chile 1991 1997 - Oct Sun>=9 4:00u 1:00 S Rule Chile 1997 only - Mar 30 3:00u 0 - Rule Chile 1998 only - Mar Sun>=9 3:00u 0 - Rule Chile 1998 only - Sep 27 4:00u 1:00 S Rule Chile 1999 only - Apr 4 3:00u 0 - Rule Chile 1999 2010 - Oct Sun>=9 4:00u 1:00 S Rule Chile 2000 2007 - Mar Sun>=9 3:00u 0 - # N.B.: the end of March 29 in Chile is March 30 in Universal time, # which is used below in specifying the transition. Rule Chile 2008 only - Mar 30 3:00u 0 - Rule Chile 2009 only - Mar Sun>=9 3:00u 0 - Rule Chile 2010 only - Apr Sun>=1 3:00u 0 - Rule Chile 2011 only - May Sun>=2 3:00u 0 - Rule Chile 2011 only - Aug Sun>=16 4:00u 1:00 S -Rule Chile 2012 max - Apr Sun>=23 3:00u 0 - -Rule Chile 2012 max - Sep Sun>=2 4:00u 1:00 S +Rule Chile 2012 2015 - Apr Sun>=23 3:00u 0 - +Rule Chile 2012 2014 - Sep Sun>=2 4:00u 1:00 S # IATA SSIM anomalies: (1992-02) says 1992-03-14; # (1996-09) says 1998-03-08. Ignore these. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Santiago -4:42:46 - LMT 1890 -4:42:46 - SMT 1910 # Santiago Mean Time -5:00 - CLT 1916 Jul 1 # Chile Time -4:42:46 - SMT 1918 Sep 1 # Santiago Mean Time -4:00 - CLT 1919 Jul 1 # Chile Time -4:42:46 - SMT 1927 Sep 1 # Santiago Mean Time -5:00 Chile CL%sT 1947 May 22 # Chile Time - -4:00 Chile CL%sT + -4:00 Chile CL%sT 2015 Apr 26 3:00u + -3:00 - CLT Zone Pacific/Easter -7:17:44 - LMT 1890 -7:17:28 - EMT 1932 Sep # Easter Mean Time - -7:00 Chile EAS%sT 1982 Mar 13 21:00 # Easter Time - -6:00 Chile EAS%sT + -7:00 Chile EAS%sT 1982 Mar 13 3:00u # Easter Time + -6:00 Chile EAS%sT 2015 Apr 26 3:00u + -5:00 - EAST # # Salas y Gómez Island is uninhabited. # Other Chilean locations, including Juan Fernández Is, Desventuradas Is, # and Antarctic bases, are like America/Santiago. # Colombia # Milne gives 4:56:16.4 for Bogotá time in 1899; round to nearest. He writes, # "A variation of fifteen minutes in the public clocks of Bogota is not rare." # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule CO 1992 only - May 3 0:00 1:00 S Rule CO 1993 only - Apr 4 0:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Bogota -4:56:16 - LMT 1884 Mar 13 -4:56:16 - BMT 1914 Nov 23 # Bogotá Mean Time -5:00 CO CO%sT # Colombia Time # Malpelo, Providencia, San Andres # no information; probably like America/Bogota # Curaçao # Milne gives 4:35:46.9 for Curaçao mean time; round to nearest. # # From Paul Eggert (2006-03-22): # Shanks & Pottenger say that The Bottom and Philipsburg have been at # -4:00 since standard time was introduced on 1912-03-02; and that # Kralendijk and Rincon used Kralendijk Mean Time (-4:33:08) from # 1912-02-02 to 1965-01-01. The former is dubious, since S&P also say # Saba Island has been like Curaçao. # This all predates our 1970 cutoff, though. # # By July 2007 Curaçao and St Maarten are planned to become # associated states within the Netherlands, much like Aruba; # Bonaire, Saba and St Eustatius would become directly part of the # Netherlands as Kingdom Islands. This won't affect their time zones # though, as far as we know. # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Curacao -4:35:47 - LMT 1912 Feb 12 # Willemstad -4:30 - ANT 1965 # Netherlands Antilles Time -4:00 - AST # From Arthur David Olson (2011-06-15): # use links for places with new iso3166 codes. # The name "Lower Prince's Quarter" is both longer than fourteen characters # and contains an apostrophe; use "Lower_Princes" below. Link America/Curacao America/Lower_Princes # Sint Maarten Link America/Curacao America/Kralendijk # Caribbean Netherlands # Ecuador # # Milne says the Central and South American Telegraph Company used -5:24:15. # # From Paul Eggert (2007-03-04): # Apparently Ecuador had a failed experiment with DST in 1992. # (2007-02-27) and # (2006-11-06) both # talk about "hora Sixto". Leave this alone for now, as we have no data. # # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Guayaquil -5:19:20 - LMT 1890 -5:14:00 - QMT 1931 # Quito Mean Time -5:00 - ECT # Ecuador Time Zone Pacific/Galapagos -5:58:24 - LMT 1931 # Puerto Baquerizo Moreno -5:00 - ECT 1986 -6:00 - GALT # Galápagos Time # Falklands # From Paul Eggert (2006-03-22): # Between 1990 and 2000 inclusive, Shanks & Pottenger and the IATA agree except # the IATA gives 1996-09-08. Go with Shanks & Pottenger. # From Falkland Islands Government Office, London (2001-01-22) # via Jesper Nørgaard: # ... the clocks revert back to Local Mean Time at 2 am on Sunday 15 # April 2001 and advance one hour to summer time at 2 am on Sunday 2 # September. It is anticipated that the clocks will revert back at 2 # am on Sunday 21 April 2002 and advance to summer time at 2 am on # Sunday 1 September. # From Rives McDow (2001-02-13): # # I have communicated several times with people there, and the last # time I had communications that was helpful was in 1998. Here is # what was said then: # # "The general rule was that Stanley used daylight saving and the Camp # did not. However for various reasons many people in the Camp have # started to use daylight saving (known locally as 'Stanley Time') # There is no rule as to who uses daylight saving - it is a matter of # personal choice and so it is impossible to draw a map showing who # uses it and who does not. Any list would be out of date as soon as # it was produced. This year daylight saving ended on April 18/19th # and started again on September 12/13th. I do not know what the rule # is, but can find out if you like. We do not change at the same time # as UK or Chile." # # I did have in my notes that the rule was "Second Saturday in Sep at # 0:00 until third Saturday in Apr at 0:00". I think that this does # not agree in some cases with Shanks; is this true? # # Also, there is no mention in the list that some areas in the # Falklands do not use DST. I have found in my communications there # that these areas are on the western half of East Falkland and all of # West Falkland. Stanley is the only place that consistently observes # DST. Again, as in other places in the world, the farmers don't like # it. West Falkland is almost entirely sheep farmers. # # I know one lady there that keeps a list of which farm keeps DST and # which doesn't each year. She runs a shop in Stanley, and says that # the list changes each year. She uses it to communicate to her # customers, catching them when they are home for lunch or dinner. # From Paul Eggert (2001-03-05): # For now, we'll just record the time in Stanley, since we have no # better info. # From Steffen Thorsen (2011-04-01): # The Falkland Islands will not turn back clocks this winter, but stay on # daylight saving time. # # One source: # http://www.falklandnews.com/public/story.cfm?get=5914&source=3 # # We have gotten this confirmed by a clerk of the legislative assembly: # Normally the clocks revert to Local Mean Time (UTC/GMT -4 hours) on the # third Sunday of April at 0200hrs and advance to Summer Time (UTC/GMT -3 # hours) on the first Sunday of September at 0200hrs. # # IMPORTANT NOTE: During 2011, on a trial basis, the Falkland Islands # will not revert to local mean time, but clocks will remain on Summer # time (UTC/GMT - 3 hours) throughout the whole of 2011. Any long term # change to local time following the trial period will be notified. # # From Andrew Newman (2012-02-24) # A letter from Justin McPhee, Chief Executive, # Cable & Wireless Falkland Islands (dated 2012-02-22) # states... # The current Atlantic/Stanley entry under South America expects the # clocks to go back to standard Falklands Time (FKT) on the 15th April. # The database entry states that in 2011 Stanley was staying on fixed # summer time on a trial basis only. FIG need to contact IANA and/or # the maintainers of the database to inform them we're adopting # the same policy this year and suggest recommendations for future years. # # For now we will assume permanent summer time for the Falklands # until advised differently (to apply for 2012 and beyond, after the 2011 # experiment was apparently successful.) # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Falk 1937 1938 - Sep lastSun 0:00 1:00 S Rule Falk 1938 1942 - Mar Sun>=19 0:00 0 - Rule Falk 1939 only - Oct 1 0:00 1:00 S Rule Falk 1940 1942 - Sep lastSun 0:00 1:00 S Rule Falk 1943 only - Jan 1 0:00 0 - Rule Falk 1983 only - Sep lastSun 0:00 1:00 S Rule Falk 1984 1985 - Apr lastSun 0:00 0 - Rule Falk 1984 only - Sep 16 0:00 1:00 S Rule Falk 1985 2000 - Sep Sun>=9 0:00 1:00 S Rule Falk 1986 2000 - Apr Sun>=16 0:00 0 - Rule Falk 2001 2010 - Apr Sun>=15 2:00 0 - Rule Falk 2001 2010 - Sep Sun>=1 2:00 1:00 S # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Atlantic/Stanley -3:51:24 - LMT 1890 -3:51:24 - SMT 1912 Mar 12 # Stanley Mean Time -4:00 Falk FK%sT 1983 May # Falkland Is Time -3:00 Falk FK%sT 1985 Sep 15 -4:00 Falk FK%sT 2010 Sep 5 2:00 -3:00 - FKST # French Guiana # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Cayenne -3:29:20 - LMT 1911 Jul -4:00 - GFT 1967 Oct # French Guiana Time -3:00 - GFT # Guyana # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Guyana -3:52:40 - LMT 1915 Mar # Georgetown -3:45 - GBGT 1966 May 26 # Br Guiana Time -3:45 - GYT 1975 Jul 31 # Guyana Time -3:00 - GYT 1991 # IATA SSIM (1996-06) says -4:00. Assume a 1991 switch. -4:00 - GYT # Paraguay # # From Paul Eggert (2006-03-22): # Shanks & Pottenger say that spring transitions are 01:00 -> 02:00, # and autumn transitions are 00:00 -> 23:00. Go with pre-1999 # editions of Shanks, and with the IATA, who say transitions occur at 00:00. # # From Waldemar Villamayor-Venialbo (2013-09-20): # No time of the day is established for the adjustment, so people normally # adjust their clocks at 0 hour of the given dates. # # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Para 1975 1988 - Oct 1 0:00 1:00 S Rule Para 1975 1978 - Mar 1 0:00 0 - Rule Para 1979 1991 - Apr 1 0:00 0 - Rule Para 1989 only - Oct 22 0:00 1:00 S Rule Para 1990 only - Oct 1 0:00 1:00 S Rule Para 1991 only - Oct 6 0:00 1:00 S Rule Para 1992 only - Mar 1 0:00 0 - Rule Para 1992 only - Oct 5 0:00 1:00 S Rule Para 1993 only - Mar 31 0:00 0 - Rule Para 1993 1995 - Oct 1 0:00 1:00 S Rule Para 1994 1995 - Feb lastSun 0:00 0 - Rule Para 1996 only - Mar 1 0:00 0 - # IATA SSIM (2000-02) says 1999-10-10; ignore this for now. # From Steffen Thorsen (2000-10-02): # I have three independent reports that Paraguay changed to DST this Sunday # (10-01). # # Translated by Gwillim Law (2001-02-27) from # Noticias, a daily paper in Asunción, Paraguay (2000-10-01): # http://www.diarionoticias.com.py/011000/nacional/naciona1.htm # Starting at 0:00 today, the clock will be set forward 60 minutes, in # fulfillment of Decree No. 7,273 of the Executive Power.... The time change # system has been operating for several years. Formerly there was a separate # decree each year; the new law has the same effect, but permanently. Every # year, the time will change on the first Sunday of October; likewise, the # clock will be set back on the first Sunday of March. # Rule Para 1996 2001 - Oct Sun>=1 0:00 1:00 S # IATA SSIM (1997-09) says Mar 1; go with Shanks & Pottenger. Rule Para 1997 only - Feb lastSun 0:00 0 - # Shanks & Pottenger say 1999-02-28; IATA SSIM (1999-02) says 1999-02-27, but # (1999-09) reports no date; go with above sources and Gerd Knops (2001-02-27). Rule Para 1998 2001 - Mar Sun>=1 0:00 0 - # From Rives McDow (2002-02-28): # A decree was issued in Paraguay (no. 16350) on 2002-02-26 that changed the # dst method to be from the first Sunday in September to the first Sunday in # April. Rule Para 2002 2004 - Apr Sun>=1 0:00 0 - Rule Para 2002 2003 - Sep Sun>=1 0:00 1:00 S # # From Jesper Nørgaard Welen (2005-01-02): # There are several sources that claim that Paraguay made # a timezone rule change in autumn 2004. # From Steffen Thorsen (2005-01-05): # Decree 1,867 (2004-03-05) # From Carlos Raúl Perasso via Jesper Nørgaard Welen (2006-10-13) # http://www.presidencia.gov.py/decretos/D1867.pdf Rule Para 2004 2009 - Oct Sun>=15 0:00 1:00 S Rule Para 2005 2009 - Mar Sun>=8 0:00 0 - # From Carlos Raúl Perasso (2010-02-18): # By decree number 3958 issued yesterday # http://www.presidencia.gov.py/v1/wp-content/uploads/2010/02/decreto3958.pdf # Paraguay changes its DST schedule, postponing the March rule to April and # modifying the October date. The decree reads: # ... # Art. 1. It is hereby established that from the second Sunday of the month of # April of this year (2010), the official time is to be set back 60 minutes, # and that on the first Sunday of the month of October, it is to be set # forward 60 minutes, in all the territory of the Paraguayan Republic. # ... Rule Para 2010 max - Oct Sun>=1 0:00 1:00 S Rule Para 2010 2012 - Apr Sun>=8 0:00 0 - # # From Steffen Thorsen (2013-03-07): # Paraguay will end DST on 2013-03-24 00:00.... # http://www.ande.gov.py/interna.php?id=1075 # # From Carlos Raúl Perasso (2013-03-15): # The change in Paraguay is now final. Decree number 10780 # http://www.presidencia.gov.py/uploads/pdf/presidencia-3b86ff4b691c79d4f5927ca964922ec74772ce857c02ca054a52a37b49afc7fb.pdf # From Carlos Raúl Perasso (2014-02-28): # Decree 1264 can be found at: # http://www.presidencia.gov.py/archivos/documentos/DECRETO1264_ey9r8zai.pdf Rule Para 2013 max - Mar Sun>=22 0:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Asuncion -3:50:40 - LMT 1890 -3:50:40 - AMT 1931 Oct 10 # Asunción Mean Time -4:00 - PYT 1972 Oct # Paraguay Time -3:00 - PYT 1974 Apr -4:00 Para PY%sT # Peru # # From Evelyn C. Leeper via Mark Brader (2003-10-26) # : # When we were in Peru in 1985-1986, they apparently switched over # sometime between December 29 and January 3 while we were on the Amazon. # # From Paul Eggert (2006-03-22): # Shanks & Pottenger don't have this transition. Assume 1986 was like 1987. # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S Rule Peru 1938 only - Jan 1 0:00 1:00 S Rule Peru 1938 only - Apr 1 0:00 0 - Rule Peru 1938 1939 - Sep lastSun 0:00 1:00 S Rule Peru 1939 1940 - Mar Sun>=24 0:00 0 - Rule Peru 1986 1987 - Jan 1 0:00 1:00 S Rule Peru 1986 1987 - Apr 1 0:00 0 - Rule Peru 1990 only - Jan 1 0:00 1:00 S Rule Peru 1990 only - Apr 1 0:00 0 - # IATA is ambiguous for 1993/1995; go with Shanks & Pottenger. Rule Peru 1994 only - Jan 1 0:00 1:00 S Rule Peru 1994 only - Apr 1 0:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Lima -5:08:12 - LMT 1890 -5:08:36 - LMT 1908 Jul 28 # Lima Mean Time? -5:00 Peru PE%sT # Peru Time # South Georgia # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Atlantic/South_Georgia -2:26:08 - LMT 1890 # Grytviken -2:00 - GST # South Georgia Time # South Sandwich Is # uninhabited; scientific personnel have wintered # Suriname # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Paramaribo -3:40:40 - LMT 1911 -3:40:52 - PMT 1935 # Paramaribo Mean Time -3:40:36 - PMT 1945 Oct # The capital moved? -3:30 - NEGT 1975 Nov 20 # Dutch Guiana Time -3:30 - SRT 1984 Oct # Suriname Time -3:00 - SRT # Trinidad and Tobago # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Port_of_Spain -4:06:04 - LMT 1912 Mar 2 -4:00 - AST # These all agree with Trinidad and Tobago since 1970. Link America/Port_of_Spain America/Anguilla Link America/Port_of_Spain America/Dominica Link America/Port_of_Spain America/Grenada Link America/Port_of_Spain America/Guadeloupe Link America/Port_of_Spain America/Marigot # St Martin (French part) Link America/Port_of_Spain America/Montserrat Link America/Port_of_Spain America/St_Barthelemy # St Barthélemy Link America/Port_of_Spain America/St_Kitts # St Kitts & Nevis Link America/Port_of_Spain America/St_Lucia Link America/Port_of_Spain America/St_Thomas # Virgin Islands (US) Link America/Port_of_Spain America/St_Vincent Link America/Port_of_Spain America/Tortola # Virgin Islands (UK) # Uruguay # From Paul Eggert (1993-11-18): # Uruguay wins the prize for the strangest peacetime manipulation of the rules. # From Shanks & Pottenger: # Rule NAME FROM TO TYPE IN ON AT SAVE LETTER/S # Whitman gives 1923 Oct 1; go with Shanks & Pottenger. Rule Uruguay 1923 only - Oct 2 0:00 0:30 HS Rule Uruguay 1924 1926 - Apr 1 0:00 0 - Rule Uruguay 1924 1925 - Oct 1 0:00 0:30 HS Rule Uruguay 1933 1935 - Oct lastSun 0:00 0:30 HS # Shanks & Pottenger give 1935 Apr 1 0:00 & 1936 Mar 30 0:00; go with Whitman. Rule Uruguay 1934 1936 - Mar Sat>=25 23:30s 0 - Rule Uruguay 1936 only - Nov 1 0:00 0:30 HS Rule Uruguay 1937 1941 - Mar lastSun 0:00 0 - # Whitman gives 1937 Oct 3; go with Shanks & Pottenger. Rule Uruguay 1937 1940 - Oct lastSun 0:00 0:30 HS # Whitman gives 1941 Oct 24 - 1942 Mar 27, 1942 Dec 14 - 1943 Apr 13, # and 1943 Apr 13 "to present time"; go with Shanks & Pottenger. Rule Uruguay 1941 only - Aug 1 0:00 0:30 HS Rule Uruguay 1942 only - Jan 1 0:00 0 - Rule Uruguay 1942 only - Dec 14 0:00 1:00 S Rule Uruguay 1943 only - Mar 14 0:00 0 - Rule Uruguay 1959 only - May 24 0:00 1:00 S Rule Uruguay 1959 only - Nov 15 0:00 0 - Rule Uruguay 1960 only - Jan 17 0:00 1:00 S Rule Uruguay 1960 only - Mar 6 0:00 0 - Rule Uruguay 1965 1967 - Apr Sun>=1 0:00 1:00 S Rule Uruguay 1965 only - Sep 26 0:00 0 - Rule Uruguay 1966 1967 - Oct 31 0:00 0 - Rule Uruguay 1968 1970 - May 27 0:00 0:30 HS Rule Uruguay 1968 1970 - Dec 2 0:00 0 - Rule Uruguay 1972 only - Apr 24 0:00 1:00 S Rule Uruguay 1972 only - Aug 15 0:00 0 - Rule Uruguay 1974 only - Mar 10 0:00 0:30 HS Rule Uruguay 1974 only - Dec 22 0:00 1:00 S Rule Uruguay 1976 only - Oct 1 0:00 0 - Rule Uruguay 1977 only - Dec 4 0:00 1:00 S Rule Uruguay 1978 only - Apr 1 0:00 0 - Rule Uruguay 1979 only - Oct 1 0:00 1:00 S Rule Uruguay 1980 only - May 1 0:00 0 - Rule Uruguay 1987 only - Dec 14 0:00 1:00 S Rule Uruguay 1988 only - Mar 14 0:00 0 - Rule Uruguay 1988 only - Dec 11 0:00 1:00 S Rule Uruguay 1989 only - Mar 12 0:00 0 - Rule Uruguay 1989 only - Oct 29 0:00 1:00 S # Shanks & Pottenger say no DST was observed in 1990/1 and 1991/2, # and that 1992/3's DST was from 10-25 to 03-01. Go with IATA. Rule Uruguay 1990 1992 - Mar Sun>=1 0:00 0 - Rule Uruguay 1990 1991 - Oct Sun>=21 0:00 1:00 S Rule Uruguay 1992 only - Oct 18 0:00 1:00 S Rule Uruguay 1993 only - Feb 28 0:00 0 - # From Eduardo Cota (2004-09-20): # The Uruguayan government has decreed a change in the local time.... # http://www.presidencia.gub.uy/decretos/2004091502.htm Rule Uruguay 2004 only - Sep 19 0:00 1:00 S # From Steffen Thorsen (2005-03-11): # Uruguay's DST was scheduled to end on Sunday, 2005-03-13, but in order to # save energy ... it was postponed two weeks.... # http://www.presidencia.gub.uy/_Web/noticias/2005/03/2005031005.htm Rule Uruguay 2005 only - Mar 27 2:00 0 - # From Eduardo Cota (2005-09-27): # http://www.presidencia.gub.uy/_Web/decretos/2005/09/CM%20119_09%2009%202005_00001.PDF # This means that from 2005-10-09 at 02:00 local time, until 2006-03-12 at # 02:00 local time, official time in Uruguay will be at GMT -2. Rule Uruguay 2005 only - Oct 9 2:00 1:00 S Rule Uruguay 2006 only - Mar 12 2:00 0 - # From Jesper Nørgaard Welen (2006-09-06): # http://www.presidencia.gub.uy/_web/decretos/2006/09/CM%20210_08%2006%202006_00001.PDF Rule Uruguay 2006 max - Oct Sun>=1 2:00 1:00 S Rule Uruguay 2007 max - Mar Sun>=8 2:00 0 - # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Montevideo -3:44:44 - LMT 1898 Jun 28 -3:44:44 - MMT 1920 May 1 # Montevideo MT -3:30 Uruguay UY%sT 1942 Dec 14 # Uruguay Time -3:00 Uruguay UY%sT # Venezuela # # From John Stainforth (2007-11-28): # ... the change for Venezuela originally expected for 2007-12-31 has # been brought forward to 2007-12-09. The official announcement was # published today in the "Gaceta Oficial de la República Bolivariana # de Venezuela, número 38.819" (official document for all laws or # resolution publication) # http://www.globovision.com/news.php?nid=72208 # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone America/Caracas -4:27:44 - LMT 1890 -4:27:40 - CMT 1912 Feb 12 # Caracas Mean Time? -4:30 - VET 1965 # Venezuela Time -4:00 - VET 2007 Dec 9 3:00 -4:30 - VET Index: projects/cxl_iscsi/contrib/tzdata/zone.tab =================================================================== --- projects/cxl_iscsi/contrib/tzdata/zone.tab (revision 279716) +++ projects/cxl_iscsi/contrib/tzdata/zone.tab (revision 279717) @@ -1,440 +1,440 @@ # tz zone descriptions (deprecated version) # # This file is in the public domain, so clarified as of # 2009-05-17 by Arthur David Olson. # # From Paul Eggert (2014-07-31): # This file is intended as a backward-compatibility aid for older programs. # New programs should use zone1970.tab. This file is like zone1970.tab (see # zone1970.tab's comments), but with the following additional restrictions: # # 1. This file contains only ASCII characters. # 2. The first data column contains exactly one country code. # # Because of (2), each row stands for an area that is the intersection # of a region identified by a country code and of a zone where civil # clocks have agreed since 1970; this is a narrower definition than # that of zone1970.tab. # # This table is intended as an aid for users, to help them select time # zone data entries appropriate for their practical needs. It is not # intended to take or endorse any position on legal or territorial claims. # #country- #code coordinates TZ comments AD +4230+00131 Europe/Andorra AE +2518+05518 Asia/Dubai AF +3431+06912 Asia/Kabul AG +1703-06148 America/Antigua AI +1812-06304 America/Anguilla AL +4120+01950 Europe/Tirane AM +4011+04430 Asia/Yerevan AO -0848+01314 Africa/Luanda AQ -7750+16636 Antarctica/McMurdo McMurdo, South Pole, Scott (New Zealand time) AQ -6734-06808 Antarctica/Rothera Rothera Station, Adelaide Island AQ -6448-06406 Antarctica/Palmer Palmer Station, Anvers Island AQ -6736+06253 Antarctica/Mawson Mawson Station, Holme Bay AQ -6835+07758 Antarctica/Davis Davis Station, Vestfold Hills AQ -6617+11031 Antarctica/Casey Casey Station, Bailey Peninsula AQ -7824+10654 Antarctica/Vostok Vostok Station, Lake Vostok AQ -6640+14001 Antarctica/DumontDUrville Dumont-d'Urville Station, Adelie Land AQ -690022+0393524 Antarctica/Syowa Syowa Station, E Ongul I AQ -720041+0023206 Antarctica/Troll Troll Station, Queen Maud Land AR -3436-05827 America/Argentina/Buenos_Aires Buenos Aires (BA, CF) AR -3124-06411 America/Argentina/Cordoba most locations (CB, CC, CN, ER, FM, MN, SE, SF) AR -2447-06525 America/Argentina/Salta (SA, LP, NQ, RN) AR -2411-06518 America/Argentina/Jujuy Jujuy (JY) AR -2649-06513 America/Argentina/Tucuman Tucuman (TM) AR -2828-06547 America/Argentina/Catamarca Catamarca (CT), Chubut (CH) AR -2926-06651 America/Argentina/La_Rioja La Rioja (LR) AR -3132-06831 America/Argentina/San_Juan San Juan (SJ) AR -3253-06849 America/Argentina/Mendoza Mendoza (MZ) AR -3319-06621 America/Argentina/San_Luis San Luis (SL) AR -5138-06913 America/Argentina/Rio_Gallegos Santa Cruz (SC) AR -5448-06818 America/Argentina/Ushuaia Tierra del Fuego (TF) AS -1416-17042 Pacific/Pago_Pago AT +4813+01620 Europe/Vienna AU -3133+15905 Australia/Lord_Howe Lord Howe Island AU -5430+15857 Antarctica/Macquarie Macquarie Island AU -4253+14719 Australia/Hobart Tasmania - most locations AU -3956+14352 Australia/Currie Tasmania - King Island AU -3749+14458 Australia/Melbourne Victoria AU -3352+15113 Australia/Sydney New South Wales - most locations AU -3157+14127 Australia/Broken_Hill New South Wales - Yancowinna AU -2728+15302 Australia/Brisbane Queensland - most locations AU -2016+14900 Australia/Lindeman Queensland - Holiday Islands AU -3455+13835 Australia/Adelaide South Australia AU -1228+13050 Australia/Darwin Northern Territory AU -3157+11551 Australia/Perth Western Australia - most locations AU -3143+12852 Australia/Eucla Western Australia - Eucla area AW +1230-06958 America/Aruba AX +6006+01957 Europe/Mariehamn AZ +4023+04951 Asia/Baku BA +4352+01825 Europe/Sarajevo BB +1306-05937 America/Barbados BD +2343+09025 Asia/Dhaka BE +5050+00420 Europe/Brussels BF +1222-00131 Africa/Ouagadougou BG +4241+02319 Europe/Sofia BH +2623+05035 Asia/Bahrain BI -0323+02922 Africa/Bujumbura BJ +0629+00237 Africa/Porto-Novo BL +1753-06251 America/St_Barthelemy BM +3217-06446 Atlantic/Bermuda BN +0456+11455 Asia/Brunei BO -1630-06809 America/La_Paz BQ +120903-0681636 America/Kralendijk BR -0351-03225 America/Noronha Atlantic islands BR -0127-04829 America/Belem Amapa, E Para BR -0343-03830 America/Fortaleza NE Brazil (MA, PI, CE, RN, PB) BR -0803-03454 America/Recife Pernambuco BR -0712-04812 America/Araguaina Tocantins BR -0940-03543 America/Maceio Alagoas, Sergipe BR -1259-03831 America/Bahia Bahia BR -2332-04637 America/Sao_Paulo S & SE Brazil (GO, DF, MG, ES, RJ, SP, PR, SC, RS) BR -2027-05437 America/Campo_Grande Mato Grosso do Sul BR -1535-05605 America/Cuiaba Mato Grosso BR -0226-05452 America/Santarem W Para BR -0846-06354 America/Porto_Velho Rondonia BR +0249-06040 America/Boa_Vista Roraima BR -0308-06001 America/Manaus E Amazonas BR -0640-06952 America/Eirunepe W Amazonas BR -0958-06748 America/Rio_Branco Acre BS +2505-07721 America/Nassau BT +2728+08939 Asia/Thimphu BW -2439+02555 Africa/Gaborone BY +5354+02734 Europe/Minsk BZ +1730-08812 America/Belize CA +4734-05243 America/St_Johns Newfoundland Time, including SE Labrador CA +4439-06336 America/Halifax Atlantic Time - Nova Scotia (most places), PEI CA +4612-05957 America/Glace_Bay Atlantic Time - Nova Scotia - places that did not observe DST 1966-1971 CA +4606-06447 America/Moncton Atlantic Time - New Brunswick CA +5320-06025 America/Goose_Bay Atlantic Time - Labrador - most locations CA +5125-05707 America/Blanc-Sablon Atlantic Standard Time - Quebec - Lower North Shore CA +4339-07923 America/Toronto Eastern Time - Ontario & Quebec - most locations CA +4901-08816 America/Nipigon Eastern Time - Ontario & Quebec - places that did not observe DST 1967-1973 CA +4823-08915 America/Thunder_Bay Eastern Time - Thunder Bay, Ontario CA +6344-06828 America/Iqaluit Eastern Time - east Nunavut - most locations CA +6608-06544 America/Pangnirtung Eastern Time - Pangnirtung, Nunavut CA +744144-0944945 America/Resolute Central Time - Resolute, Nunavut CA +484531-0913718 America/Atikokan Eastern Standard Time - Atikokan, Ontario and Southampton I, Nunavut CA +624900-0920459 America/Rankin_Inlet Central Time - central Nunavut CA +4953-09709 America/Winnipeg Central Time - Manitoba & west Ontario CA +4843-09434 America/Rainy_River Central Time - Rainy River & Fort Frances, Ontario CA +5024-10439 America/Regina Central Standard Time - Saskatchewan - most locations CA +5017-10750 America/Swift_Current Central Standard Time - Saskatchewan - midwest CA +5333-11328 America/Edmonton Mountain Time - Alberta, east British Columbia & west Saskatchewan CA +690650-1050310 America/Cambridge_Bay Mountain Time - west Nunavut CA +6227-11421 America/Yellowknife Mountain Time - central Northwest Territories CA +682059-1334300 America/Inuvik Mountain Time - west Northwest Territories CA +4906-11631 America/Creston Mountain Standard Time - Creston, British Columbia CA +5946-12014 America/Dawson_Creek Mountain Standard Time - Dawson Creek & Fort Saint John, British Columbia CA +4916-12307 America/Vancouver Pacific Time - west British Columbia CA +6043-13503 America/Whitehorse Pacific Time - south Yukon CA +6404-13925 America/Dawson Pacific Time - north Yukon CC -1210+09655 Indian/Cocos CD -0418+01518 Africa/Kinshasa west Dem. Rep. of Congo CD -1140+02728 Africa/Lubumbashi east Dem. Rep. of Congo CF +0422+01835 Africa/Bangui CG -0416+01517 Africa/Brazzaville CH +4723+00832 Europe/Zurich CI +0519-00402 Africa/Abidjan CK -2114-15946 Pacific/Rarotonga CL -3327-07040 America/Santiago most locations CL -2709-10926 Pacific/Easter Easter Island CM +0403+00942 Africa/Douala CN +3114+12128 Asia/Shanghai Beijing Time CN +4348+08735 Asia/Urumqi Xinjiang Time CO +0436-07405 America/Bogota CR +0956-08405 America/Costa_Rica CU +2308-08222 America/Havana CV +1455-02331 Atlantic/Cape_Verde CW +1211-06900 America/Curacao CX -1025+10543 Indian/Christmas CY +3510+03322 Asia/Nicosia CZ +5005+01426 Europe/Prague DE +5230+01322 Europe/Berlin most locations DE +4742+00841 Europe/Busingen Busingen DJ +1136+04309 Africa/Djibouti DK +5540+01235 Europe/Copenhagen DM +1518-06124 America/Dominica DO +1828-06954 America/Santo_Domingo DZ +3647+00303 Africa/Algiers EC -0210-07950 America/Guayaquil mainland EC -0054-08936 Pacific/Galapagos Galapagos Islands EE +5925+02445 Europe/Tallinn EG +3003+03115 Africa/Cairo EH +2709-01312 Africa/El_Aaiun ER +1520+03853 Africa/Asmara ES +4024-00341 Europe/Madrid mainland ES +3553-00519 Africa/Ceuta Ceuta & Melilla ES +2806-01524 Atlantic/Canary Canary Islands ET +0902+03842 Africa/Addis_Ababa FI +6010+02458 Europe/Helsinki FJ -1808+17825 Pacific/Fiji FK -5142-05751 Atlantic/Stanley FM +0725+15147 Pacific/Chuuk Chuuk (Truk) and Yap FM +0658+15813 Pacific/Pohnpei Pohnpei (Ponape) FM +0519+16259 Pacific/Kosrae Kosrae FO +6201-00646 Atlantic/Faroe FR +4852+00220 Europe/Paris GA +0023+00927 Africa/Libreville GB +513030-0000731 Europe/London GD +1203-06145 America/Grenada GE +4143+04449 Asia/Tbilisi GF +0456-05220 America/Cayenne GG +4927-00232 Europe/Guernsey GH +0533-00013 Africa/Accra GI +3608-00521 Europe/Gibraltar GL +6411-05144 America/Godthab most locations GL +7646-01840 America/Danmarkshavn east coast, north of Scoresbysund GL +7029-02158 America/Scoresbysund Scoresbysund / Ittoqqortoormiit GL +7634-06847 America/Thule Thule / Pituffik GM +1328-01639 Africa/Banjul GN +0931-01343 Africa/Conakry GP +1614-06132 America/Guadeloupe GQ +0345+00847 Africa/Malabo GR +3758+02343 Europe/Athens GS -5416-03632 Atlantic/South_Georgia GT +1438-09031 America/Guatemala GU +1328+14445 Pacific/Guam GW +1151-01535 Africa/Bissau GY +0648-05810 America/Guyana HK +2217+11409 Asia/Hong_Kong HN +1406-08713 America/Tegucigalpa HR +4548+01558 Europe/Zagreb HT +1832-07220 America/Port-au-Prince HU +4730+01905 Europe/Budapest ID -0610+10648 Asia/Jakarta Java & Sumatra ID -0002+10920 Asia/Pontianak west & central Borneo ID -0507+11924 Asia/Makassar east & south Borneo, Sulawesi (Celebes), Bali, Nusa Tengarra, west Timor ID -0232+14042 Asia/Jayapura west New Guinea (Irian Jaya) & Malukus (Moluccas) IE +5320-00615 Europe/Dublin IL +314650+0351326 Asia/Jerusalem IM +5409-00428 Europe/Isle_of_Man IN +2232+08822 Asia/Kolkata IO -0720+07225 Indian/Chagos IQ +3321+04425 Asia/Baghdad IR +3540+05126 Asia/Tehran IS +6409-02151 Atlantic/Reykjavik IT +4154+01229 Europe/Rome JE +4912-00207 Europe/Jersey JM +175805-0764736 America/Jamaica JO +3157+03556 Asia/Amman JP +353916+1394441 Asia/Tokyo KE -0117+03649 Africa/Nairobi KG +4254+07436 Asia/Bishkek KH +1133+10455 Asia/Phnom_Penh KI +0125+17300 Pacific/Tarawa Gilbert Islands KI -0308-17105 Pacific/Enderbury Phoenix Islands KI +0152-15720 Pacific/Kiritimati Line Islands KM -1141+04316 Indian/Comoro KN +1718-06243 America/St_Kitts KP +3901+12545 Asia/Pyongyang KR +3733+12658 Asia/Seoul KW +2920+04759 Asia/Kuwait KY +1918-08123 America/Cayman KZ +4315+07657 Asia/Almaty most locations KZ +4448+06528 Asia/Qyzylorda Qyzylorda (Kyzylorda, Kzyl-Orda) KZ +5017+05710 Asia/Aqtobe Aqtobe (Aktobe) KZ +4431+05016 Asia/Aqtau Atyrau (Atirau, Gur'yev), Mangghystau (Mankistau) KZ +5113+05121 Asia/Oral West Kazakhstan LA +1758+10236 Asia/Vientiane LB +3353+03530 Asia/Beirut LC +1401-06100 America/St_Lucia LI +4709+00931 Europe/Vaduz LK +0656+07951 Asia/Colombo LR +0618-01047 Africa/Monrovia LS -2928+02730 Africa/Maseru LT +5441+02519 Europe/Vilnius LU +4936+00609 Europe/Luxembourg LV +5657+02406 Europe/Riga LY +3254+01311 Africa/Tripoli MA +3339-00735 Africa/Casablanca MC +4342+00723 Europe/Monaco MD +4700+02850 Europe/Chisinau ME +4226+01916 Europe/Podgorica MF +1804-06305 America/Marigot MG -1855+04731 Indian/Antananarivo MH +0709+17112 Pacific/Majuro most locations MH +0905+16720 Pacific/Kwajalein Kwajalein MK +4159+02126 Europe/Skopje ML +1239-00800 Africa/Bamako MM +1647+09610 Asia/Rangoon MN +4755+10653 Asia/Ulaanbaatar most locations MN +4801+09139 Asia/Hovd Bayan-Olgiy, Govi-Altai, Hovd, Uvs, Zavkhan MN +4804+11430 Asia/Choibalsan Dornod, Sukhbaatar MO +2214+11335 Asia/Macau MP +1512+14545 Pacific/Saipan MQ +1436-06105 America/Martinique MR +1806-01557 Africa/Nouakchott MS +1643-06213 America/Montserrat MT +3554+01431 Europe/Malta MU -2010+05730 Indian/Mauritius MV +0410+07330 Indian/Maldives MW -1547+03500 Africa/Blantyre MX +1924-09909 America/Mexico_City Central Time - most locations -MX +2105-08646 America/Cancun Central Time - Quintana Roo +MX +2105-08646 America/Cancun Eastern Standard Time - Quintana Roo MX +2058-08937 America/Merida Central Time - Campeche, Yucatan MX +2540-10019 America/Monterrey Mexican Central Time - Coahuila, Durango, Nuevo Leon, Tamaulipas away from US border MX +2550-09730 America/Matamoros US Central Time - Coahuila, Durango, Nuevo Leon, Tamaulipas near US border MX +2313-10625 America/Mazatlan Mountain Time - S Baja, Nayarit, Sinaloa MX +2838-10605 America/Chihuahua Mexican Mountain Time - Chihuahua away from US border MX +2934-10425 America/Ojinaga US Mountain Time - Chihuahua near US border MX +2904-11058 America/Hermosillo Mountain Standard Time - Sonora MX +3232-11701 America/Tijuana US Pacific Time - Baja California near US border MX +3018-11452 America/Santa_Isabel Mexican Pacific Time - Baja California away from US border MX +2048-10515 America/Bahia_Banderas Mexican Central Time - Bahia de Banderas MY +0310+10142 Asia/Kuala_Lumpur peninsular Malaysia MY +0133+11020 Asia/Kuching Sabah & Sarawak MZ -2558+03235 Africa/Maputo NA -2234+01706 Africa/Windhoek NC -2216+16627 Pacific/Noumea NE +1331+00207 Africa/Niamey NF -2903+16758 Pacific/Norfolk NG +0627+00324 Africa/Lagos NI +1209-08617 America/Managua NL +5222+00454 Europe/Amsterdam NO +5955+01045 Europe/Oslo NP +2743+08519 Asia/Kathmandu NR -0031+16655 Pacific/Nauru NU -1901-16955 Pacific/Niue NZ -3652+17446 Pacific/Auckland most locations NZ -4357-17633 Pacific/Chatham Chatham Islands OM +2336+05835 Asia/Muscat PA +0858-07932 America/Panama PE -1203-07703 America/Lima PF -1732-14934 Pacific/Tahiti Society Islands PF -0900-13930 Pacific/Marquesas Marquesas Islands PF -2308-13457 Pacific/Gambier Gambier Islands PG -0930+14710 Pacific/Port_Moresby most locations PG -0613+15534 Pacific/Bougainville Bougainville PH +1435+12100 Asia/Manila PK +2452+06703 Asia/Karachi PL +5215+02100 Europe/Warsaw PM +4703-05620 America/Miquelon PN -2504-13005 Pacific/Pitcairn PR +182806-0660622 America/Puerto_Rico PS +3130+03428 Asia/Gaza Gaza Strip PS +313200+0350542 Asia/Hebron West Bank PT +3843-00908 Europe/Lisbon mainland PT +3238-01654 Atlantic/Madeira Madeira Islands PT +3744-02540 Atlantic/Azores Azores PW +0720+13429 Pacific/Palau PY -2516-05740 America/Asuncion QA +2517+05132 Asia/Qatar RE -2052+05528 Indian/Reunion RO +4426+02606 Europe/Bucharest RS +4450+02030 Europe/Belgrade RU +5443+02030 Europe/Kaliningrad Moscow-01 - Kaliningrad RU +554521+0373704 Europe/Moscow Moscow+00 - west Russia RU +4457+03406 Europe/Simferopol Moscow+00 - Crimea RU +4844+04425 Europe/Volgograd Moscow+00 - Caspian Sea RU +5312+05009 Europe/Samara Moscow+00 (Moscow+01 after 2014-10-26) - Samara, Udmurtia RU +5651+06036 Asia/Yekaterinburg Moscow+02 - Urals RU +5500+07324 Asia/Omsk Moscow+03 - west Siberia RU +5502+08255 Asia/Novosibirsk Moscow+03 - Novosibirsk RU +5345+08707 Asia/Novokuznetsk Moscow+03 (Moscow+04 after 2014-10-26) - Kemerovo RU +5601+09250 Asia/Krasnoyarsk Moscow+04 - Yenisei River RU +5216+10420 Asia/Irkutsk Moscow+05 - Lake Baikal RU +5203+11328 Asia/Chita Moscow+06 (Moscow+05 after 2014-10-26) - Zabaykalsky RU +6200+12940 Asia/Yakutsk Moscow+06 - Lena River RU +623923+1353314 Asia/Khandyga Moscow+06 - Tomponsky, Ust-Maysky RU +4310+13156 Asia/Vladivostok Moscow+07 - Amur River RU +4658+14242 Asia/Sakhalin Moscow+07 - Sakhalin Island RU +643337+1431336 Asia/Ust-Nera Moscow+07 - Oymyakonsky RU +5934+15048 Asia/Magadan Moscow+08 (Moscow+07 after 2014-10-26) - Magadan RU +6728+15343 Asia/Srednekolymsk Moscow+08 - E Sakha, N Kuril Is RU +5301+15839 Asia/Kamchatka Moscow+08 (Moscow+09 after 2014-10-26) - Kamchatka RU +6445+17729 Asia/Anadyr Moscow+08 (Moscow+09 after 2014-10-26) - Bering Sea RW -0157+03004 Africa/Kigali SA +2438+04643 Asia/Riyadh SB -0932+16012 Pacific/Guadalcanal SC -0440+05528 Indian/Mahe SD +1536+03232 Africa/Khartoum SE +5920+01803 Europe/Stockholm SG +0117+10351 Asia/Singapore SH -1555-00542 Atlantic/St_Helena SI +4603+01431 Europe/Ljubljana SJ +7800+01600 Arctic/Longyearbyen SK +4809+01707 Europe/Bratislava SL +0830-01315 Africa/Freetown SM +4355+01228 Europe/San_Marino SN +1440-01726 Africa/Dakar SO +0204+04522 Africa/Mogadishu SR +0550-05510 America/Paramaribo SS +0451+03136 Africa/Juba ST +0020+00644 Africa/Sao_Tome SV +1342-08912 America/El_Salvador SX +180305-0630250 America/Lower_Princes SY +3330+03618 Asia/Damascus SZ -2618+03106 Africa/Mbabane TC +2128-07108 America/Grand_Turk TD +1207+01503 Africa/Ndjamena TF -492110+0701303 Indian/Kerguelen TG +0608+00113 Africa/Lome TH +1345+10031 Asia/Bangkok TJ +3835+06848 Asia/Dushanbe TK -0922-17114 Pacific/Fakaofo TL -0833+12535 Asia/Dili TM +3757+05823 Asia/Ashgabat TN +3648+01011 Africa/Tunis TO -2110-17510 Pacific/Tongatapu TR +4101+02858 Europe/Istanbul TT +1039-06131 America/Port_of_Spain TV -0831+17913 Pacific/Funafuti TW +2503+12130 Asia/Taipei TZ -0648+03917 Africa/Dar_es_Salaam UA +5026+03031 Europe/Kiev most locations UA +4837+02218 Europe/Uzhgorod Ruthenia UA +4750+03510 Europe/Zaporozhye Zaporozh'ye, E Lugansk / Zaporizhia, E Luhansk UG +0019+03225 Africa/Kampala UM +1645-16931 Pacific/Johnston Johnston Atoll UM +2813-17722 Pacific/Midway Midway Islands UM +1917+16637 Pacific/Wake Wake Island US +404251-0740023 America/New_York Eastern Time US +421953-0830245 America/Detroit Eastern Time - Michigan - most locations US +381515-0854534 America/Kentucky/Louisville Eastern Time - Kentucky - Louisville area US +364947-0845057 America/Kentucky/Monticello Eastern Time - Kentucky - Wayne County US +394606-0860929 America/Indiana/Indianapolis Eastern Time - Indiana - most locations US +384038-0873143 America/Indiana/Vincennes Eastern Time - Indiana - Daviess, Dubois, Knox & Martin Counties US +410305-0863611 America/Indiana/Winamac Eastern Time - Indiana - Pulaski County US +382232-0862041 America/Indiana/Marengo Eastern Time - Indiana - Crawford County US +382931-0871643 America/Indiana/Petersburg Eastern Time - Indiana - Pike County US +384452-0850402 America/Indiana/Vevay Eastern Time - Indiana - Switzerland County US +415100-0873900 America/Chicago Central Time US +375711-0864541 America/Indiana/Tell_City Central Time - Indiana - Perry County US +411745-0863730 America/Indiana/Knox Central Time - Indiana - Starke County US +450628-0873651 America/Menominee Central Time - Michigan - Dickinson, Gogebic, Iron & Menominee Counties US +470659-1011757 America/North_Dakota/Center Central Time - North Dakota - Oliver County US +465042-1012439 America/North_Dakota/New_Salem Central Time - North Dakota - Morton County (except Mandan area) US +471551-1014640 America/North_Dakota/Beulah Central Time - North Dakota - Mercer County US +394421-1045903 America/Denver Mountain Time US +433649-1161209 America/Boise Mountain Time - south Idaho & east Oregon US +332654-1120424 America/Phoenix Mountain Standard Time - Arizona (except Navajo) US +340308-1181434 America/Los_Angeles Pacific Time US +550737-1313435 America/Metlakatla Pacific Standard Time - Annette Island, Alaska US +611305-1495401 America/Anchorage Alaska Time US +581807-1342511 America/Juneau Alaska Time - Alaska panhandle US +571035-1351807 America/Sitka Alaska Time - southeast Alaska panhandle US +593249-1394338 America/Yakutat Alaska Time - Alaska panhandle neck US +643004-1652423 America/Nome Alaska Time - west Alaska US +515248-1763929 America/Adak Aleutian Islands US +211825-1575130 Pacific/Honolulu Hawaii UY -3453-05611 America/Montevideo UZ +3940+06648 Asia/Samarkand west Uzbekistan UZ +4120+06918 Asia/Tashkent east Uzbekistan VA +415408+0122711 Europe/Vatican VC +1309-06114 America/St_Vincent VE +1030-06656 America/Caracas VG +1827-06437 America/Tortola VI +1821-06456 America/St_Thomas VN +1045+10640 Asia/Ho_Chi_Minh VU -1740+16825 Pacific/Efate WF -1318-17610 Pacific/Wallis WS -1350-17144 Pacific/Apia YE +1245+04512 Asia/Aden YT -1247+04514 Indian/Mayotte ZA -2615+02800 Africa/Johannesburg ZM -1525+02817 Africa/Lusaka ZW -1750+03103 Africa/Harare Index: projects/cxl_iscsi/contrib/tzdata/zone1970.tab =================================================================== --- projects/cxl_iscsi/contrib/tzdata/zone1970.tab (revision 279716) +++ projects/cxl_iscsi/contrib/tzdata/zone1970.tab (revision 279717) @@ -1,372 +1,372 @@ # tz zone descriptions # # This file is in the public domain. # # From Paul Eggert (2014-07-31): # This file contains a table where each row stands for a zone where # civil time stamps have agreed since 1970. Columns are separated by # a single tab. Lines beginning with '#' are comments. All text uses # UTF-8 encoding. The columns of the table are as follows: # # 1. The countries that overlap the zone, as a comma-separated list # of ISO 3166 2-character country codes. # See the file '/usr/share/misc/iso3166'. # 2. Latitude and longitude of the zone's principal location # in ISO 6709 sign-degrees-minutes-seconds format, # either +-DDMM+-DDDMM or +-DDMMSS+-DDDMMSS, # first latitude (+ is north), then longitude (+ is east). # 3. Zone name used in value of TZ environment variable. # Please see the 'Theory' file for how zone names are chosen. # If multiple zones overlap a country, each has a row in the # table, with each column 1 containing the country code. # 4. Comments; present if and only if a country has multiple zones. # # If a zone covers multiple countries, the most-populous city is used, # and that country is listed first in column 1; any other countries # are listed alphabetically by country code. The table is sorted # first by country code, then (if possible) by an order within the # country that (1) makes some geographical sense, and (2) puts the # most populous zones first, where that does not contradict (1). # # This table is intended as an aid for users, to help them select time # zone data entries appropriate for their practical needs. It is not # intended to take or endorse any position on legal or territorial claims. # #country- #codes coordinates TZ comments AD +4230+00131 Europe/Andorra AE,OM +2518+05518 Asia/Dubai AF +3431+06912 Asia/Kabul AL +4120+01950 Europe/Tirane AM +4011+04430 Asia/Yerevan AQ -6734-06808 Antarctica/Rothera Rothera Station, Adelaide Island AQ -6448-06406 Antarctica/Palmer Palmer Station, Anvers Island AQ -6736+06253 Antarctica/Mawson Mawson Station, Holme Bay AQ -6835+07758 Antarctica/Davis Davis Station, Vestfold Hills AQ -6617+11031 Antarctica/Casey Casey Station, Bailey Peninsula AQ -7824+10654 Antarctica/Vostok Vostok Station, Lake Vostok AQ -6640+14001 Antarctica/DumontDUrville Dumont-d'Urville Station, Adélie Land AQ -690022+0393524 Antarctica/Syowa Syowa Station, E Ongul I AQ -720041+0023206 Antarctica/Troll Troll Station, Queen Maud Land AR -3436-05827 America/Argentina/Buenos_Aires Buenos Aires (BA, CF) AR -3124-06411 America/Argentina/Cordoba most locations (CB, CC, CN, ER, FM, MN, SE, SF) AR -2447-06525 America/Argentina/Salta (SA, LP, NQ, RN) AR -2411-06518 America/Argentina/Jujuy Jujuy (JY) AR -2649-06513 America/Argentina/Tucuman Tucumán (TM) AR -2828-06547 America/Argentina/Catamarca Catamarca (CT), Chubut (CH) AR -2926-06651 America/Argentina/La_Rioja La Rioja (LR) AR -3132-06831 America/Argentina/San_Juan San Juan (SJ) AR -3253-06849 America/Argentina/Mendoza Mendoza (MZ) AR -3319-06621 America/Argentina/San_Luis San Luis (SL) AR -5138-06913 America/Argentina/Rio_Gallegos Santa Cruz (SC) AR -5448-06818 America/Argentina/Ushuaia Tierra del Fuego (TF) AS,UM -1416-17042 Pacific/Pago_Pago Samoa, Midway AT +4813+01620 Europe/Vienna AU -3133+15905 Australia/Lord_Howe Lord Howe Island AU -5430+15857 Antarctica/Macquarie Macquarie Island AU -4253+14719 Australia/Hobart Tasmania - most locations AU -3956+14352 Australia/Currie Tasmania - King Island AU -3749+14458 Australia/Melbourne Victoria AU -3352+15113 Australia/Sydney New South Wales - most locations AU -3157+14127 Australia/Broken_Hill New South Wales - Yancowinna AU -2728+15302 Australia/Brisbane Queensland - most locations AU -2016+14900 Australia/Lindeman Queensland - Holiday Islands AU -3455+13835 Australia/Adelaide South Australia AU -1228+13050 Australia/Darwin Northern Territory AU -3157+11551 Australia/Perth Western Australia - most locations AU -3143+12852 Australia/Eucla Western Australia - Eucla area AZ +4023+04951 Asia/Baku BB +1306-05937 America/Barbados BD +2343+09025 Asia/Dhaka BE +5050+00420 Europe/Brussels BG +4241+02319 Europe/Sofia BM +3217-06446 Atlantic/Bermuda BN +0456+11455 Asia/Brunei BO -1630-06809 America/La_Paz BR -0351-03225 America/Noronha Atlantic islands BR -0127-04829 America/Belem Amapá, E Pará BR -0343-03830 America/Fortaleza NE Brazil (MA, PI, CE, RN, PB) BR -0803-03454 America/Recife Pernambuco BR -0712-04812 America/Araguaina Tocantins BR -0940-03543 America/Maceio Alagoas, Sergipe BR -1259-03831 America/Bahia Bahia BR -2332-04637 America/Sao_Paulo S & SE Brazil (GO, DF, MG, ES, RJ, SP, PR, SC, RS) BR -2027-05437 America/Campo_Grande Mato Grosso do Sul BR -1535-05605 America/Cuiaba Mato Grosso BR -0226-05452 America/Santarem W Pará BR -0846-06354 America/Porto_Velho Rondônia BR +0249-06040 America/Boa_Vista Roraima BR -0308-06001 America/Manaus E Amazonas BR -0640-06952 America/Eirunepe W Amazonas BR -0958-06748 America/Rio_Branco Acre BS +2505-07721 America/Nassau BT +2728+08939 Asia/Thimphu BY +5354+02734 Europe/Minsk BZ +1730-08812 America/Belize CA +4734-05243 America/St_Johns Newfoundland Time, including SE Labrador CA +4439-06336 America/Halifax Atlantic Time - Nova Scotia (most places), PEI CA +4612-05957 America/Glace_Bay Atlantic Time - Nova Scotia - places that did not observe DST 1966-1971 CA +4606-06447 America/Moncton Atlantic Time - New Brunswick CA +5320-06025 America/Goose_Bay Atlantic Time - Labrador - most locations CA +5125-05707 America/Blanc-Sablon Atlantic Standard Time - Quebec - Lower North Shore CA +4339-07923 America/Toronto Eastern Time - Ontario & Quebec - most locations CA +4901-08816 America/Nipigon Eastern Time - Ontario & Quebec - places that did not observe DST 1967-1973 CA +4823-08915 America/Thunder_Bay Eastern Time - Thunder Bay, Ontario CA +6344-06828 America/Iqaluit Eastern Time - east Nunavut - most locations CA +6608-06544 America/Pangnirtung Eastern Time - Pangnirtung, Nunavut CA +744144-0944945 America/Resolute Central Time - Resolute, Nunavut CA +484531-0913718 America/Atikokan Eastern Standard Time - Atikokan, Ontario and Southampton I, Nunavut CA +624900-0920459 America/Rankin_Inlet Central Time - central Nunavut CA +4953-09709 America/Winnipeg Central Time - Manitoba & west Ontario CA +4843-09434 America/Rainy_River Central Time - Rainy River & Fort Frances, Ontario CA +5024-10439 America/Regina Central Standard Time - Saskatchewan - most locations CA +5017-10750 America/Swift_Current Central Standard Time - Saskatchewan - midwest CA +5333-11328 America/Edmonton Mountain Time - Alberta, east British Columbia & west Saskatchewan CA +690650-1050310 America/Cambridge_Bay Mountain Time - west Nunavut CA +6227-11421 America/Yellowknife Mountain Time - central Northwest Territories CA +682059-1334300 America/Inuvik Mountain Time - west Northwest Territories CA +4906-11631 America/Creston Mountain Standard Time - Creston, British Columbia CA +5946-12014 America/Dawson_Creek Mountain Standard Time - Dawson Creek & Fort Saint John, British Columbia CA +4916-12307 America/Vancouver Pacific Time - west British Columbia CA +6043-13503 America/Whitehorse Pacific Time - south Yukon CA +6404-13925 America/Dawson Pacific Time - north Yukon CC -1210+09655 Indian/Cocos CH,DE,LI +4723+00832 Europe/Zurich Swiss time CI,BF,GM,GN,ML,MR,SH,SL,SN,ST,TG +0519-00402 Africa/Abidjan CK -2114-15946 Pacific/Rarotonga CL -3327-07040 America/Santiago most locations CL -2709-10926 Pacific/Easter Easter Island CN +3114+12128 Asia/Shanghai Beijing Time CN +4348+08735 Asia/Urumqi Xinjiang Time CO +0436-07405 America/Bogota CR +0956-08405 America/Costa_Rica CU +2308-08222 America/Havana CV +1455-02331 Atlantic/Cape_Verde CW,AW,BQ,SX +1211-06900 America/Curacao CX -1025+10543 Indian/Christmas CY +3510+03322 Asia/Nicosia CZ,SK +5005+01426 Europe/Prague DE +5230+01322 Europe/Berlin Berlin time DK +5540+01235 Europe/Copenhagen DO +1828-06954 America/Santo_Domingo DZ +3647+00303 Africa/Algiers EC -0210-07950 America/Guayaquil mainland EC -0054-08936 Pacific/Galapagos Galápagos Islands EE +5925+02445 Europe/Tallinn EG +3003+03115 Africa/Cairo EH +2709-01312 Africa/El_Aaiun ES +4024-00341 Europe/Madrid mainland ES +3553-00519 Africa/Ceuta Ceuta & Melilla ES +2806-01524 Atlantic/Canary Canary Islands FI,AX +6010+02458 Europe/Helsinki FJ -1808+17825 Pacific/Fiji FK -5142-05751 Atlantic/Stanley FM +0725+15147 Pacific/Chuuk Chuuk (Truk) and Yap FM +0658+15813 Pacific/Pohnpei Pohnpei (Ponape) FM +0519+16259 Pacific/Kosrae Kosrae FO +6201-00646 Atlantic/Faroe FR +4852+00220 Europe/Paris GB,GG,IM,JE +513030-0000731 Europe/London GE +4143+04449 Asia/Tbilisi GF +0456-05220 America/Cayenne GH +0533-00013 Africa/Accra GI +3608-00521 Europe/Gibraltar GL +6411-05144 America/Godthab most locations GL +7646-01840 America/Danmarkshavn east coast, north of Scoresbysund GL +7029-02158 America/Scoresbysund Scoresbysund / Ittoqqortoormiit GL +7634-06847 America/Thule Thule / Pituffik GR +3758+02343 Europe/Athens GS -5416-03632 Atlantic/South_Georgia GT +1438-09031 America/Guatemala GU,MP +1328+14445 Pacific/Guam GW +1151-01535 Africa/Bissau GY +0648-05810 America/Guyana HK +2217+11409 Asia/Hong_Kong HN +1406-08713 America/Tegucigalpa HT +1832-07220 America/Port-au-Prince HU +4730+01905 Europe/Budapest ID -0610+10648 Asia/Jakarta Java & Sumatra ID -0002+10920 Asia/Pontianak west & central Borneo ID -0507+11924 Asia/Makassar east & south Borneo, Sulawesi (Celebes), Bali, Nusa Tengarra, west Timor ID -0232+14042 Asia/Jayapura west New Guinea (Irian Jaya) & Malukus (Moluccas) IE +5320-00615 Europe/Dublin IL +314650+0351326 Asia/Jerusalem IN +2232+08822 Asia/Kolkata IO -0720+07225 Indian/Chagos IQ +3321+04425 Asia/Baghdad IR +3540+05126 Asia/Tehran IS +6409-02151 Atlantic/Reykjavik IT,SM,VA +4154+01229 Europe/Rome JM +175805-0764736 America/Jamaica JO +3157+03556 Asia/Amman JP +353916+1394441 Asia/Tokyo KE,DJ,ER,ET,KM,MG,SO,TZ,UG,YT -0117+03649 Africa/Nairobi KG +4254+07436 Asia/Bishkek KI +0125+17300 Pacific/Tarawa Gilbert Islands KI -0308-17105 Pacific/Enderbury Phoenix Islands KI +0152-15720 Pacific/Kiritimati Line Islands KP +3901+12545 Asia/Pyongyang KR +3733+12658 Asia/Seoul KZ +4315+07657 Asia/Almaty most locations KZ +4448+06528 Asia/Qyzylorda Qyzylorda (Kyzylorda, Kzyl-Orda) KZ +5017+05710 Asia/Aqtobe Aqtobe (Aktobe) KZ +4431+05016 Asia/Aqtau Atyrau (Atirau, Gur'yev), Mangghystau (Mankistau) KZ +5113+05121 Asia/Oral West Kazakhstan LB +3353+03530 Asia/Beirut LK +0656+07951 Asia/Colombo LR +0618-01047 Africa/Monrovia LT +5441+02519 Europe/Vilnius LU +4936+00609 Europe/Luxembourg LV +5657+02406 Europe/Riga LY +3254+01311 Africa/Tripoli MA +3339-00735 Africa/Casablanca MC +4342+00723 Europe/Monaco MD +4700+02850 Europe/Chisinau MH +0709+17112 Pacific/Majuro most locations MH +0905+16720 Pacific/Kwajalein Kwajalein MM +1647+09610 Asia/Rangoon MN +4755+10653 Asia/Ulaanbaatar most locations MN +4801+09139 Asia/Hovd Bayan-Ölgii, Govi-Altai, Hovd, Uvs, Zavkhan MN +4804+11430 Asia/Choibalsan Dornod, Sükhbaatar MO +2214+11335 Asia/Macau MQ +1436-06105 America/Martinique MT +3554+01431 Europe/Malta MU -2010+05730 Indian/Mauritius MV +0410+07330 Indian/Maldives MX +1924-09909 America/Mexico_City Central Time - most locations -MX +2105-08646 America/Cancun Central Time - Quintana Roo +MX +2105-08646 America/Cancun Eastern Standard Time - Quintana Roo MX +2058-08937 America/Merida Central Time - Campeche, Yucatán MX +2540-10019 America/Monterrey Mexican Central Time - Coahuila, Durango, Nuevo León, Tamaulipas away from US border MX +2550-09730 America/Matamoros US Central Time - Coahuila, Durango, Nuevo León, Tamaulipas near US border MX +2313-10625 America/Mazatlan Mountain Time - S Baja, Nayarit, Sinaloa MX +2838-10605 America/Chihuahua Mexican Mountain Time - Chihuahua away from US border MX +2934-10425 America/Ojinaga US Mountain Time - Chihuahua near US border MX +2904-11058 America/Hermosillo Mountain Standard Time - Sonora MX +3232-11701 America/Tijuana US Pacific Time - Baja California near US border MX +3018-11452 America/Santa_Isabel Mexican Pacific Time - Baja California away from US border MX +2048-10515 America/Bahia_Banderas Mexican Central Time - Bahía de Banderas MY +0310+10142 Asia/Kuala_Lumpur peninsular Malaysia MY +0133+11020 Asia/Kuching Sabah & Sarawak MZ,BI,BW,CD,MW,RW,ZM,ZW -2558+03235 Africa/Maputo Central Africa Time (UTC+2) NA -2234+01706 Africa/Windhoek NC -2216+16627 Pacific/Noumea NF -2903+16758 Pacific/Norfolk NG,AO,BJ,CD,CF,CG,CM,GA,GQ,NE +0627+00324 Africa/Lagos West Africa Time (UTC+1) NI +1209-08617 America/Managua NL +5222+00454 Europe/Amsterdam NO,SJ +5955+01045 Europe/Oslo NP +2743+08519 Asia/Kathmandu NR -0031+16655 Pacific/Nauru NU -1901-16955 Pacific/Niue NZ,AQ -3652+17446 Pacific/Auckland New Zealand time NZ -4357-17633 Pacific/Chatham Chatham Islands PA,KY +0858-07932 America/Panama PE -1203-07703 America/Lima PF -1732-14934 Pacific/Tahiti Society Islands PF -0900-13930 Pacific/Marquesas Marquesas Islands PF -2308-13457 Pacific/Gambier Gambier Islands PG -0930+14710 Pacific/Port_Moresby most locations PG -0613+15534 Pacific/Bougainville Bougainville PH +1435+12100 Asia/Manila PK +2452+06703 Asia/Karachi PL +5215+02100 Europe/Warsaw PM +4703-05620 America/Miquelon PN -2504-13005 Pacific/Pitcairn PR +182806-0660622 America/Puerto_Rico PS +3130+03428 Asia/Gaza Gaza Strip PS +313200+0350542 Asia/Hebron West Bank PT +3843-00908 Europe/Lisbon mainland PT +3238-01654 Atlantic/Madeira Madeira Islands PT +3744-02540 Atlantic/Azores Azores PW +0720+13429 Pacific/Palau PY -2516-05740 America/Asuncion QA,BH +2517+05132 Asia/Qatar RE,TF -2052+05528 Indian/Reunion Réunion, Crozet Is, Scattered Is RO +4426+02606 Europe/Bucharest RS,BA,HR,ME,MK,SI +4450+02030 Europe/Belgrade RU +5443+02030 Europe/Kaliningrad Moscow-01 - Kaliningrad RU +554521+0373704 Europe/Moscow Moscow+00 - west Russia RU +4457+03406 Europe/Simferopol Moscow+00 - Crimea RU +4844+04425 Europe/Volgograd Moscow+00 - Caspian Sea RU +5312+05009 Europe/Samara Moscow+00 (Moscow+01 after 2014-10-26) - Samara, Udmurtia RU +5651+06036 Asia/Yekaterinburg Moscow+02 - Urals RU +5500+07324 Asia/Omsk Moscow+03 - west Siberia RU +5502+08255 Asia/Novosibirsk Moscow+03 - Novosibirsk RU +5345+08707 Asia/Novokuznetsk Moscow+03 (Moscow+04 after 2014-10-26) - Kemerovo RU +5601+09250 Asia/Krasnoyarsk Moscow+04 - Yenisei River RU +5216+10420 Asia/Irkutsk Moscow+05 - Lake Baikal RU +5203+11328 Asia/Chita Moscow+06 (Moscow+05 after 2014-10-26) - Zabaykalsky RU +6200+12940 Asia/Yakutsk Moscow+06 - Lena River RU +623923+1353314 Asia/Khandyga Moscow+06 - Tomponsky, Ust-Maysky RU +4310+13156 Asia/Vladivostok Moscow+07 - Amur River RU +4658+14242 Asia/Sakhalin Moscow+07 - Sakhalin Island RU +643337+1431336 Asia/Ust-Nera Moscow+07 - Oymyakonsky RU +5934+15048 Asia/Magadan Moscow+08 (Moscow+07 after 2014-10-26) - Magadan RU +6728+15343 Asia/Srednekolymsk Moscow+08 - E Sakha, N Kuril Is RU +5301+15839 Asia/Kamchatka Moscow+08 (Moscow+09 after 2014-10-26) - Kamchatka RU +6445+17729 Asia/Anadyr Moscow+08 (Moscow+09 after 2014-10-26) - Bering Sea SA,KW,YE +2438+04643 Asia/Riyadh SB -0932+16012 Pacific/Guadalcanal SC -0440+05528 Indian/Mahe SD,SS +1536+03232 Africa/Khartoum SE +5920+01803 Europe/Stockholm SG +0117+10351 Asia/Singapore SR +0550-05510 America/Paramaribo SV +1342-08912 America/El_Salvador SY +3330+03618 Asia/Damascus TC +2128-07108 America/Grand_Turk TD +1207+01503 Africa/Ndjamena TF -492110+0701303 Indian/Kerguelen Kerguelen, St Paul I, Amsterdam I TH,KH,LA,VN +1345+10031 Asia/Bangkok most of Indochina TJ +3835+06848 Asia/Dushanbe TK -0922-17114 Pacific/Fakaofo TL -0833+12535 Asia/Dili TM +3757+05823 Asia/Ashgabat TN +3648+01011 Africa/Tunis TO -2110-17510 Pacific/Tongatapu TR +4101+02858 Europe/Istanbul TT,AG,AI,BL,DM,GD,GP,KN,LC,MF,MS,VC,VG,VI +1039-06131 America/Port_of_Spain TV -0831+17913 Pacific/Funafuti TW +2503+12130 Asia/Taipei UA +5026+03031 Europe/Kiev most locations UA +4837+02218 Europe/Uzhgorod Ruthenia UA +4750+03510 Europe/Zaporozhye Zaporozh'ye, E Lugansk / Zaporizhia, E Luhansk UM +1917+16637 Pacific/Wake Wake Island US +404251-0740023 America/New_York Eastern Time US +421953-0830245 America/Detroit Eastern Time - Michigan - most locations US +381515-0854534 America/Kentucky/Louisville Eastern Time - Kentucky - Louisville area US +364947-0845057 America/Kentucky/Monticello Eastern Time - Kentucky - Wayne County US +394606-0860929 America/Indiana/Indianapolis Eastern Time - Indiana - most locations US +384038-0873143 America/Indiana/Vincennes Eastern Time - Indiana - Daviess, Dubois, Knox & Martin Counties US +410305-0863611 America/Indiana/Winamac Eastern Time - Indiana - Pulaski County US +382232-0862041 America/Indiana/Marengo Eastern Time - Indiana - Crawford County US +382931-0871643 America/Indiana/Petersburg Eastern Time - Indiana - Pike County US +384452-0850402 America/Indiana/Vevay Eastern Time - Indiana - Switzerland County US +415100-0873900 America/Chicago Central Time US +375711-0864541 America/Indiana/Tell_City Central Time - Indiana - Perry County US +411745-0863730 America/Indiana/Knox Central Time - Indiana - Starke County US +450628-0873651 America/Menominee Central Time - Michigan - Dickinson, Gogebic, Iron & Menominee Counties US +470659-1011757 America/North_Dakota/Center Central Time - North Dakota - Oliver County US +465042-1012439 America/North_Dakota/New_Salem Central Time - North Dakota - Morton County (except Mandan area) US +471551-1014640 America/North_Dakota/Beulah Central Time - North Dakota - Mercer County US +394421-1045903 America/Denver Mountain Time US +433649-1161209 America/Boise Mountain Time - south Idaho & east Oregon US +332654-1120424 America/Phoenix Mountain Standard Time - Arizona (except Navajo) US +340308-1181434 America/Los_Angeles Pacific Time US +550737-1313435 America/Metlakatla Pacific Standard Time - Annette Island, Alaska US +611305-1495401 America/Anchorage Alaska Time US +581807-1342511 America/Juneau Alaska Time - Alaska panhandle US +571035-1351807 America/Sitka Alaska Time - southeast Alaska panhandle US +593249-1394338 America/Yakutat Alaska Time - Alaska panhandle neck US +643004-1652423 America/Nome Alaska Time - west Alaska US +515248-1763929 America/Adak Aleutian Islands US,UM +211825-1575130 Pacific/Honolulu Hawaii time UY -3453-05611 America/Montevideo UZ +3940+06648 Asia/Samarkand west Uzbekistan UZ +4120+06918 Asia/Tashkent east Uzbekistan VE +1030-06656 America/Caracas VN +1045+10640 Asia/Ho_Chi_Minh south Vietnam VU -1740+16825 Pacific/Efate WF -1318-17610 Pacific/Wallis WS -1350-17144 Pacific/Apia ZA,LS,SZ -2615+02800 Africa/Johannesburg Index: projects/cxl_iscsi/contrib/tzdata =================================================================== --- projects/cxl_iscsi/contrib/tzdata (revision 279716) +++ projects/cxl_iscsi/contrib/tzdata (revision 279717) Property changes on: projects/cxl_iscsi/contrib/tzdata ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,2 ## Merged /head/contrib/tzdata:r278538-279716 Merged /vendor/tzdata/dist:r279704 Index: projects/cxl_iscsi/lib/libc/gen/getpwent.c =================================================================== --- projects/cxl_iscsi/lib/libc/gen/getpwent.c (revision 279716) +++ projects/cxl_iscsi/lib/libc/gen/getpwent.c (revision 279717) @@ -1,2011 +1,2011 @@ /*- * Copyright (c) 2003 Networks Associates Technology, Inc. * All rights reserved. * * This software was developed for the FreeBSD Project by * Jacques A. Vidrine, Safeport Network Services, and Network * Associates Laboratories, the Security Research Division of Network * Associates, Inc. under DARPA/SPAWAR contract N66001-01-C-8035 * ("CBOSS"), as part of the DARPA CHATS research program. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * */ #include __FBSDID("$FreeBSD$"); #include "namespace.h" #include #ifdef YP #include #include #include #endif #include #include #include #ifdef HESIOD #include #endif #include #include #include #include #include #include #include #include #include #include #include "un-namespace.h" #include #include "libc_private.h" #include "pw_scan.h" #include "nss_tls.h" #ifdef NS_CACHING #include "nscache.h" #endif #ifndef CTASSERT #define CTASSERT(x) _CTASSERT(x, __LINE__) #define _CTASSERT(x, y) __CTASSERT(x, y) #define __CTASSERT(x, y) typedef char __assert_ ## y [(x) ? 1 : -1] #endif /* Counter as stored in /etc/pwd.db */ typedef int pwkeynum; CTASSERT(MAXLOGNAME > sizeof(uid_t)); CTASSERT(MAXLOGNAME > sizeof(pwkeynum)); enum constants { PWD_STORAGE_INITIAL = 1 << 10, /* 1 KByte */ PWD_STORAGE_MAX = 1 << 20, /* 1 MByte */ SETPWENT = 1, ENDPWENT = 2, HESIOD_NAME_MAX = 256 }; static const ns_src defaultsrc[] = { { NSSRC_COMPAT, NS_SUCCESS }, { NULL, 0 } }; int __pw_match_entry(const char *, size_t, enum nss_lookup_type, const char *, uid_t); int __pw_parse_entry(char *, size_t, struct passwd *, int, int *errnop); static void pwd_init(struct passwd *); union key { const char *name; uid_t uid; }; static struct passwd *getpw(int (*fn)(union key, struct passwd *, char *, size_t, struct passwd **), union key); static int wrap_getpwnam_r(union key, struct passwd *, char *, size_t, struct passwd **); static int wrap_getpwuid_r(union key, struct passwd *, char *, size_t, struct passwd **); static int wrap_getpwent_r(union key, struct passwd *, char *, size_t, struct passwd **); static int pwdb_match_entry_v3(char *, size_t, enum nss_lookup_type, const char *, uid_t); static int pwdb_parse_entry_v3(char *, size_t, struct passwd *, int *); static int pwdb_match_entry_v4(char *, size_t, enum nss_lookup_type, const char *, uid_t); static int pwdb_parse_entry_v4(char *, size_t, struct passwd *, int *); struct { int (*match)(char *, size_t, enum nss_lookup_type, const char *, uid_t); int (*parse)(char *, size_t, struct passwd *, int *); } pwdb_versions[] = { { NULL, NULL }, /* version 0 */ { NULL, NULL }, /* version 1 */ { NULL, NULL }, /* version 2 */ { pwdb_match_entry_v3, pwdb_parse_entry_v3 }, /* version 3 */ { pwdb_match_entry_v4, pwdb_parse_entry_v4 }, /* version 4 */ }; struct files_state { DB *db; pwkeynum keynum; int stayopen; int version; }; static void files_endstate(void *); NSS_TLS_HANDLING(files); static DB *pwdbopen(int *); static void files_endstate(void *); static int files_setpwent(void *, void *, va_list); static int files_passwd(void *, void *, va_list); #ifdef HESIOD struct dns_state { long counter; }; static void dns_endstate(void *); NSS_TLS_HANDLING(dns); static int dns_setpwent(void *, void *, va_list); static int dns_passwd(void *, void *, va_list); #endif #ifdef YP struct nis_state { char domain[MAXHOSTNAMELEN]; int done; char *key; int keylen; }; static void nis_endstate(void *); NSS_TLS_HANDLING(nis); static int nis_setpwent(void *, void *, va_list); static int nis_passwd(void *, void *, va_list); static int nis_map(char *, enum nss_lookup_type, char *, size_t, int *); static int nis_adjunct(char *, const char *, char *, size_t); #endif struct compat_state { DB *db; pwkeynum keynum; int stayopen; int version; DB *exclude; struct passwd template; char *name; enum _compat { COMPAT_MODE_OFF = 0, COMPAT_MODE_ALL, COMPAT_MODE_NAME, COMPAT_MODE_NETGROUP } compat; }; static void compat_endstate(void *); NSS_TLS_HANDLING(compat); static int compat_setpwent(void *, void *, va_list); static int compat_passwd(void *, void *, va_list); static void compat_clear_template(struct passwd *); static int compat_set_template(struct passwd *, struct passwd *); static int compat_use_template(struct passwd *, struct passwd *, char *, size_t); static int compat_redispatch(struct compat_state *, enum nss_lookup_type, enum nss_lookup_type, const char *, const char *, uid_t, struct passwd *, char *, size_t, int *); #ifdef NS_CACHING static int pwd_id_func(char *, size_t *, va_list ap, void *); static int pwd_marshal_func(char *, size_t *, void *, va_list, void *); static int pwd_unmarshal_func(char *, size_t, void *, va_list, void *); static int pwd_id_func(char *buffer, size_t *buffer_size, va_list ap, void *cache_mdata) { char *name; uid_t uid; size_t size, desired_size; int res = NS_UNAVAIL; enum nss_lookup_type lookup_type; lookup_type = (enum nss_lookup_type)cache_mdata; switch (lookup_type) { case nss_lt_name: name = va_arg(ap, char *); size = strlen(name); desired_size = sizeof(enum nss_lookup_type) + size + 1; if (desired_size > *buffer_size) { res = NS_RETURN; goto fin; } memcpy(buffer, &lookup_type, sizeof(enum nss_lookup_type)); memcpy(buffer + sizeof(enum nss_lookup_type), name, size + 1); res = NS_SUCCESS; break; case nss_lt_id: uid = va_arg(ap, uid_t); desired_size = sizeof(enum nss_lookup_type) + sizeof(uid_t); if (desired_size > *buffer_size) { res = NS_RETURN; goto fin; } memcpy(buffer, &lookup_type, sizeof(enum nss_lookup_type)); memcpy(buffer + sizeof(enum nss_lookup_type), &uid, sizeof(uid_t)); res = NS_SUCCESS; break; default: /* should be unreachable */ return (NS_UNAVAIL); } fin: *buffer_size = desired_size; return (res); } static int pwd_marshal_func(char *buffer, size_t *buffer_size, void *retval, va_list ap, void *cache_mdata) { char *name; uid_t uid; struct passwd *pwd; char *orig_buf; size_t orig_buf_size; struct passwd new_pwd; size_t desired_size, size; char *p; switch ((enum nss_lookup_type)cache_mdata) { case nss_lt_name: name = va_arg(ap, char *); break; case nss_lt_id: uid = va_arg(ap, uid_t); break; case nss_lt_all: break; default: /* should be unreachable */ return (NS_UNAVAIL); } pwd = va_arg(ap, struct passwd *); orig_buf = va_arg(ap, char *); orig_buf_size = va_arg(ap, size_t); desired_size = sizeof(struct passwd) + sizeof(char *) + strlen(pwd->pw_name) + 1; if (pwd->pw_passwd != NULL) desired_size += strlen(pwd->pw_passwd) + 1; if (pwd->pw_class != NULL) desired_size += strlen(pwd->pw_class) + 1; if (pwd->pw_gecos != NULL) desired_size += strlen(pwd->pw_gecos) + 1; if (pwd->pw_dir != NULL) desired_size += strlen(pwd->pw_dir) + 1; if (pwd->pw_shell != NULL) desired_size += strlen(pwd->pw_shell) + 1; if (*buffer_size < desired_size) { /* this assignment is here for future use */ *buffer_size = desired_size; return (NS_RETURN); } memcpy(&new_pwd, pwd, sizeof(struct passwd)); memset(buffer, 0, desired_size); *buffer_size = desired_size; p = buffer + sizeof(struct passwd) + sizeof(char *); memcpy(buffer + sizeof(struct passwd), &p, sizeof(char *)); if (new_pwd.pw_name != NULL) { size = strlen(new_pwd.pw_name); memcpy(p, new_pwd.pw_name, size); new_pwd.pw_name = p; p += size + 1; } if (new_pwd.pw_passwd != NULL) { size = strlen(new_pwd.pw_passwd); memcpy(p, new_pwd.pw_passwd, size); new_pwd.pw_passwd = p; p += size + 1; } if (new_pwd.pw_class != NULL) { size = strlen(new_pwd.pw_class); memcpy(p, new_pwd.pw_class, size); new_pwd.pw_class = p; p += size + 1; } if (new_pwd.pw_gecos != NULL) { size = strlen(new_pwd.pw_gecos); memcpy(p, new_pwd.pw_gecos, size); new_pwd.pw_gecos = p; p += size + 1; } if (new_pwd.pw_dir != NULL) { size = strlen(new_pwd.pw_dir); memcpy(p, new_pwd.pw_dir, size); new_pwd.pw_dir = p; p += size + 1; } if (new_pwd.pw_shell != NULL) { size = strlen(new_pwd.pw_shell); memcpy(p, new_pwd.pw_shell, size); new_pwd.pw_shell = p; p += size + 1; } memcpy(buffer, &new_pwd, sizeof(struct passwd)); return (NS_SUCCESS); } static int pwd_unmarshal_func(char *buffer, size_t buffer_size, void *retval, va_list ap, void *cache_mdata) { char *name; uid_t uid; struct passwd *pwd; char *orig_buf; size_t orig_buf_size; int *ret_errno; char *p; switch ((enum nss_lookup_type)cache_mdata) { case nss_lt_name: name = va_arg(ap, char *); break; case nss_lt_id: uid = va_arg(ap, uid_t); break; case nss_lt_all: break; default: /* should be unreachable */ return (NS_UNAVAIL); } pwd = va_arg(ap, struct passwd *); orig_buf = va_arg(ap, char *); orig_buf_size = va_arg(ap, size_t); ret_errno = va_arg(ap, int *); if (orig_buf_size < buffer_size - sizeof(struct passwd) - sizeof(char *)) { *ret_errno = ERANGE; return (NS_RETURN); } memcpy(pwd, buffer, sizeof(struct passwd)); memcpy(&p, buffer + sizeof(struct passwd), sizeof(char *)); memcpy(orig_buf, buffer + sizeof(struct passwd) + sizeof(char *), buffer_size - sizeof(struct passwd) - sizeof(char *)); NS_APPLY_OFFSET(pwd->pw_name, orig_buf, p, char *); NS_APPLY_OFFSET(pwd->pw_passwd, orig_buf, p, char *); NS_APPLY_OFFSET(pwd->pw_class, orig_buf, p, char *); NS_APPLY_OFFSET(pwd->pw_gecos, orig_buf, p, char *); NS_APPLY_OFFSET(pwd->pw_dir, orig_buf, p, char *); NS_APPLY_OFFSET(pwd->pw_shell, orig_buf, p, char *); if (retval != NULL) *((struct passwd **)retval) = pwd; return (NS_SUCCESS); } NSS_MP_CACHE_HANDLING(passwd); #endif /* NS_CACHING */ void setpwent(void) { #ifdef NS_CACHING static const nss_cache_info cache_info = NS_MP_CACHE_INFO_INITIALIZER( passwd, (void *)nss_lt_all, NULL, NULL); #endif static const ns_dtab dtab[] = { { NSSRC_FILES, files_setpwent, (void *)SETPWENT }, #ifdef HESIOD { NSSRC_DNS, dns_setpwent, (void *)SETPWENT }, #endif #ifdef YP { NSSRC_NIS, nis_setpwent, (void *)SETPWENT }, #endif { NSSRC_COMPAT, compat_setpwent, (void *)SETPWENT }, #ifdef NS_CACHING NS_CACHE_CB(&cache_info) #endif { NULL, NULL, NULL } }; (void)_nsdispatch(NULL, dtab, NSDB_PASSWD, "setpwent", defaultsrc, 0); } int setpassent(int stayopen) { #ifdef NS_CACHING static const nss_cache_info cache_info = NS_MP_CACHE_INFO_INITIALIZER( passwd, (void *)nss_lt_all, NULL, NULL); #endif static const ns_dtab dtab[] = { { NSSRC_FILES, files_setpwent, (void *)SETPWENT }, #ifdef HESIOD { NSSRC_DNS, dns_setpwent, (void *)SETPWENT }, #endif #ifdef YP { NSSRC_NIS, nis_setpwent, (void *)SETPWENT }, #endif { NSSRC_COMPAT, compat_setpwent, (void *)SETPWENT }, #ifdef NS_CACHING NS_CACHE_CB(&cache_info) #endif { NULL, NULL, NULL } }; (void)_nsdispatch(NULL, dtab, NSDB_PASSWD, "setpwent", defaultsrc, stayopen); return (1); } void endpwent(void) { #ifdef NS_CACHING static const nss_cache_info cache_info = NS_MP_CACHE_INFO_INITIALIZER( passwd, (void *)nss_lt_all, NULL, NULL); #endif static const ns_dtab dtab[] = { { NSSRC_FILES, files_setpwent, (void *)ENDPWENT }, #ifdef HESIOD { NSSRC_DNS, dns_setpwent, (void *)ENDPWENT }, #endif #ifdef YP { NSSRC_NIS, nis_setpwent, (void *)ENDPWENT }, #endif { NSSRC_COMPAT, compat_setpwent, (void *)ENDPWENT }, #ifdef NS_CACHING NS_CACHE_CB(&cache_info) #endif { NULL, NULL, NULL } }; (void)_nsdispatch(NULL, dtab, NSDB_PASSWD, "endpwent", defaultsrc); } int getpwent_r(struct passwd *pwd, char *buffer, size_t bufsize, struct passwd **result) { #ifdef NS_CACHING static const nss_cache_info cache_info = NS_MP_CACHE_INFO_INITIALIZER( passwd, (void *)nss_lt_all, pwd_marshal_func, pwd_unmarshal_func); #endif static const ns_dtab dtab[] = { { NSSRC_FILES, files_passwd, (void *)nss_lt_all }, #ifdef HESIOD { NSSRC_DNS, dns_passwd, (void *)nss_lt_all }, #endif #ifdef YP { NSSRC_NIS, nis_passwd, (void *)nss_lt_all }, #endif { NSSRC_COMPAT, compat_passwd, (void *)nss_lt_all }, #ifdef NS_CACHING NS_CACHE_CB(&cache_info) #endif { NULL, NULL, NULL } }; int rv, ret_errno; pwd_init(pwd); ret_errno = 0; *result = NULL; rv = _nsdispatch(result, dtab, NSDB_PASSWD, "getpwent_r", defaultsrc, pwd, buffer, bufsize, &ret_errno); if (rv == NS_SUCCESS) return (0); else return (ret_errno); } int getpwnam_r(const char *name, struct passwd *pwd, char *buffer, size_t bufsize, struct passwd **result) { #ifdef NS_CACHING static const nss_cache_info cache_info = NS_COMMON_CACHE_INFO_INITIALIZER( passwd, (void *)nss_lt_name, pwd_id_func, pwd_marshal_func, pwd_unmarshal_func); #endif static const ns_dtab dtab[] = { { NSSRC_FILES, files_passwd, (void *)nss_lt_name }, #ifdef HESIOD { NSSRC_DNS, dns_passwd, (void *)nss_lt_name }, #endif #ifdef YP { NSSRC_NIS, nis_passwd, (void *)nss_lt_name }, #endif { NSSRC_COMPAT, compat_passwd, (void *)nss_lt_name }, #ifdef NS_CACHING NS_CACHE_CB(&cache_info) #endif { NULL, NULL, NULL } }; int rv, ret_errno; pwd_init(pwd); ret_errno = 0; *result = NULL; rv = _nsdispatch(result, dtab, NSDB_PASSWD, "getpwnam_r", defaultsrc, name, pwd, buffer, bufsize, &ret_errno); if (rv == NS_SUCCESS) return (0); else return (ret_errno); } int getpwuid_r(uid_t uid, struct passwd *pwd, char *buffer, size_t bufsize, struct passwd **result) { #ifdef NS_CACHING static const nss_cache_info cache_info = NS_COMMON_CACHE_INFO_INITIALIZER( passwd, (void *)nss_lt_id, pwd_id_func, pwd_marshal_func, pwd_unmarshal_func); #endif static const ns_dtab dtab[] = { { NSSRC_FILES, files_passwd, (void *)nss_lt_id }, #ifdef HESIOD { NSSRC_DNS, dns_passwd, (void *)nss_lt_id }, #endif #ifdef YP { NSSRC_NIS, nis_passwd, (void *)nss_lt_id }, #endif { NSSRC_COMPAT, compat_passwd, (void *)nss_lt_id }, #ifdef NS_CACHING NS_CACHE_CB(&cache_info) #endif { NULL, NULL, NULL } }; int rv, ret_errno; pwd_init(pwd); ret_errno = 0; *result = NULL; rv = _nsdispatch(result, dtab, NSDB_PASSWD, "getpwuid_r", defaultsrc, uid, pwd, buffer, bufsize, &ret_errno); if (rv == NS_SUCCESS) return (0); else return (ret_errno); } static void pwd_init(struct passwd *pwd) { static char nul[] = ""; memset(pwd, 0, sizeof(*pwd)); pwd->pw_uid = (uid_t)-1; /* Considered least likely to lead to */ pwd->pw_gid = (gid_t)-1; /* a security issue. */ pwd->pw_name = nul; pwd->pw_passwd = nul; pwd->pw_class = nul; pwd->pw_gecos = nul; pwd->pw_dir = nul; pwd->pw_shell = nul; } static struct passwd pwd; static char *pwd_storage; static size_t pwd_storage_size; static struct passwd * getpw(int (*fn)(union key, struct passwd *, char *, size_t, struct passwd **), union key key) { int rv; struct passwd *res; if (pwd_storage == NULL) { pwd_storage = malloc(PWD_STORAGE_INITIAL); if (pwd_storage == NULL) return (NULL); pwd_storage_size = PWD_STORAGE_INITIAL; } do { rv = fn(key, &pwd, pwd_storage, pwd_storage_size, &res); if (res == NULL && rv == ERANGE) { free(pwd_storage); if ((pwd_storage_size << 1) > PWD_STORAGE_MAX) { pwd_storage = NULL; errno = ERANGE; return (NULL); } pwd_storage_size <<= 1; pwd_storage = malloc(pwd_storage_size); if (pwd_storage == NULL) return (NULL); } } while (res == NULL && rv == ERANGE); if (rv != 0) errno = rv; return (res); } static int wrap_getpwnam_r(union key key, struct passwd *pwd, char *buffer, size_t bufsize, struct passwd **res) { return (getpwnam_r(key.name, pwd, buffer, bufsize, res)); } static int wrap_getpwuid_r(union key key, struct passwd *pwd, char *buffer, size_t bufsize, struct passwd **res) { return (getpwuid_r(key.uid, pwd, buffer, bufsize, res)); } static int wrap_getpwent_r(union key key __unused, struct passwd *pwd, char *buffer, size_t bufsize, struct passwd **res) { return (getpwent_r(pwd, buffer, bufsize, res)); } struct passwd * getpwnam(const char *name) { union key key; key.name = name; return (getpw(wrap_getpwnam_r, key)); } struct passwd * getpwuid(uid_t uid) { union key key; key.uid = uid; return (getpw(wrap_getpwuid_r, key)); } struct passwd * getpwent(void) { union key key; key.uid = 0; /* not used */ return (getpw(wrap_getpwent_r, key)); } /* * files backend */ static DB * pwdbopen(int *version) { DB *res; DBT key, entry; int rv; if (geteuid() != 0 || (res = dbopen(_PATH_SMP_DB, O_RDONLY, 0, DB_HASH, NULL)) == NULL) res = dbopen(_PATH_MP_DB, O_RDONLY, 0, DB_HASH, NULL); if (res == NULL) return (NULL); key.data = _PWD_VERSION_KEY; key.size = strlen(_PWD_VERSION_KEY); rv = res->get(res, &key, &entry, 0); if (rv == 0) *version = *(unsigned char *)entry.data; else *version = 3; if (*version < 3 || *version >= sizeof(pwdb_versions)/sizeof(pwdb_versions[0])) { syslog(LOG_CRIT, "Unsupported password database version %d", *version); res->close(res); res = NULL; } return (res); } static void files_endstate(void *p) { DB *db; if (p == NULL) return; db = ((struct files_state *)p)->db; if (db != NULL) db->close(db); free(p); } static int files_setpwent(void *retval, void *mdata, va_list ap) { struct files_state *st; int rv, stayopen; rv = files_getstate(&st); if (rv != 0) return (NS_UNAVAIL); switch ((enum constants)mdata) { case SETPWENT: stayopen = va_arg(ap, int); st->keynum = 0; if (stayopen) st->db = pwdbopen(&st->version); st->stayopen = stayopen; break; case ENDPWENT: if (st->db != NULL) { (void)st->db->close(st->db); st->db = NULL; } break; default: break; } return (NS_UNAVAIL); } static int files_passwd(void *retval, void *mdata, va_list ap) { char keybuf[MAXLOGNAME + 1]; DBT key, entry; struct files_state *st; enum nss_lookup_type how; const char *name; struct passwd *pwd; char *buffer; size_t bufsize, namesize; uid_t uid; uint32_t store; int rv, stayopen = 0, *errnop; name = NULL; uid = (uid_t)-1; how = (enum nss_lookup_type)mdata; switch (how) { case nss_lt_name: name = va_arg(ap, const char *); keybuf[0] = _PW_KEYBYNAME; break; case nss_lt_id: uid = va_arg(ap, uid_t); keybuf[0] = _PW_KEYBYUID; break; case nss_lt_all: keybuf[0] = _PW_KEYBYNUM; break; default: rv = NS_NOTFOUND; goto fin; } pwd = va_arg(ap, struct passwd *); buffer = va_arg(ap, char *); bufsize = va_arg(ap, size_t); errnop = va_arg(ap, int *); *errnop = files_getstate(&st); if (*errnop != 0) return (NS_UNAVAIL); if (how == nss_lt_all && st->keynum < 0) { rv = NS_NOTFOUND; goto fin; } if (st->db == NULL && (st->db = pwdbopen(&st->version)) == NULL) { *errnop = errno; rv = NS_UNAVAIL; goto fin; } if (how == nss_lt_all) stayopen = 1; else stayopen = st->stayopen; key.data = keybuf; do { switch (how) { case nss_lt_name: /* MAXLOGNAME includes NUL byte, but we do not * include the NUL byte in the key. */ namesize = strlcpy(&keybuf[1], name, sizeof(keybuf)-1); if (namesize >= sizeof(keybuf)-1) { *errnop = EINVAL; rv = NS_NOTFOUND; goto fin; } key.size = namesize + 1; break; case nss_lt_id: if (st->version < _PWD_CURRENT_VERSION) { memcpy(&keybuf[1], &uid, sizeof(uid)); key.size = sizeof(uid) + 1; } else { store = htonl(uid); memcpy(&keybuf[1], &store, sizeof(store)); key.size = sizeof(store) + 1; } break; case nss_lt_all: st->keynum++; if (st->version < _PWD_CURRENT_VERSION) { memcpy(&keybuf[1], &st->keynum, sizeof(st->keynum)); key.size = sizeof(st->keynum) + 1; } else { store = htonl(st->keynum); memcpy(&keybuf[1], &store, sizeof(store)); key.size = sizeof(store) + 1; } break; } keybuf[0] = _PW_VERSIONED(keybuf[0], st->version); rv = st->db->get(st->db, &key, &entry, 0); if (rv < 0 || rv > 1) { /* should never return > 1 */ *errnop = errno; rv = NS_UNAVAIL; goto fin; } else if (rv == 1) { if (how == nss_lt_all) st->keynum = -1; rv = NS_NOTFOUND; goto fin; } rv = pwdb_versions[st->version].match(entry.data, entry.size, how, name, uid); if (rv != NS_SUCCESS) continue; if (entry.size > bufsize) { *errnop = ERANGE; rv = NS_RETURN; break; } memcpy(buffer, entry.data, entry.size); rv = pwdb_versions[st->version].parse(buffer, entry.size, pwd, errnop); } while (how == nss_lt_all && !(rv & NS_TERMINATE)); fin: if (st->db != NULL && !stayopen) { (void)st->db->close(st->db); st->db = NULL; } if (rv == NS_SUCCESS) { pwd->pw_fields &= ~_PWF_SOURCE; pwd->pw_fields |= _PWF_FILES; if (retval != NULL) *(struct passwd **)retval = pwd; } return (rv); } static int pwdb_match_entry_v3(char *entry, size_t entrysize, enum nss_lookup_type how, const char *name, uid_t uid) { const char *p, *eom; uid_t uid2; eom = &entry[entrysize]; for (p = entry; p < eom; p++) if (*p == '\0') break; if (*p != '\0') return (NS_NOTFOUND); if (how == nss_lt_all) return (NS_SUCCESS); if (how == nss_lt_name) return (strcmp(name, entry) == 0 ? NS_SUCCESS : NS_NOTFOUND); for (p++; p < eom; p++) if (*p == '\0') break; if (*p != '\0' || (++p) + sizeof(uid) >= eom) return (NS_NOTFOUND); memcpy(&uid2, p, sizeof(uid2)); return (uid == uid2 ? NS_SUCCESS : NS_NOTFOUND); } static int pwdb_parse_entry_v3(char *buffer, size_t bufsize, struct passwd *pwd, int *errnop) { char *p, *eom; int32_t pw_change, pw_expire; /* THIS CODE MUST MATCH THAT IN pwd_mkdb. */ p = buffer; eom = &buffer[bufsize]; #define STRING(field) do { \ (field) = p; \ while (p < eom && *p != '\0') \ p++; \ if (p >= eom) \ return (NS_NOTFOUND); \ p++; \ } while (0) #define SCALAR(field) do { \ if (p + sizeof(field) > eom) \ return (NS_NOTFOUND); \ memcpy(&(field), p, sizeof(field)); \ p += sizeof(field); \ } while (0) STRING(pwd->pw_name); STRING(pwd->pw_passwd); SCALAR(pwd->pw_uid); SCALAR(pwd->pw_gid); SCALAR(pw_change); STRING(pwd->pw_class); STRING(pwd->pw_gecos); STRING(pwd->pw_dir); STRING(pwd->pw_shell); SCALAR(pw_expire); SCALAR(pwd->pw_fields); #undef STRING #undef SCALAR pwd->pw_change = pw_change; pwd->pw_expire = pw_expire; return (NS_SUCCESS); } static int pwdb_match_entry_v4(char *entry, size_t entrysize, enum nss_lookup_type how, const char *name, uid_t uid) { const char *p, *eom; uint32_t uid2; eom = &entry[entrysize]; for (p = entry; p < eom; p++) if (*p == '\0') break; if (*p != '\0') return (NS_NOTFOUND); if (how == nss_lt_all) return (NS_SUCCESS); if (how == nss_lt_name) return (strcmp(name, entry) == 0 ? NS_SUCCESS : NS_NOTFOUND); for (p++; p < eom; p++) if (*p == '\0') break; if (*p != '\0' || (++p) + sizeof(uid) >= eom) return (NS_NOTFOUND); memcpy(&uid2, p, sizeof(uid2)); uid2 = ntohl(uid2); return (uid == (uid_t)uid2 ? NS_SUCCESS : NS_NOTFOUND); } static int pwdb_parse_entry_v4(char *buffer, size_t bufsize, struct passwd *pwd, int *errnop) { char *p, *eom; uint32_t n; /* THIS CODE MUST MATCH THAT IN pwd_mkdb. */ p = buffer; eom = &buffer[bufsize]; #define STRING(field) do { \ (field) = p; \ while (p < eom && *p != '\0') \ p++; \ if (p >= eom) \ return (NS_NOTFOUND); \ p++; \ } while (0) #define SCALAR(field) do { \ if (p + sizeof(n) > eom) \ return (NS_NOTFOUND); \ memcpy(&n, p, sizeof(n)); \ (field) = ntohl(n); \ p += sizeof(n); \ } while (0) STRING(pwd->pw_name); STRING(pwd->pw_passwd); SCALAR(pwd->pw_uid); SCALAR(pwd->pw_gid); SCALAR(pwd->pw_change); STRING(pwd->pw_class); STRING(pwd->pw_gecos); STRING(pwd->pw_dir); STRING(pwd->pw_shell); SCALAR(pwd->pw_expire); SCALAR(pwd->pw_fields); #undef STRING #undef SCALAR return (NS_SUCCESS); } #ifdef HESIOD /* * dns backend */ static void dns_endstate(void *p) { free(p); } static int dns_setpwent(void *retval, void *mdata, va_list ap) { struct dns_state *st; int rv; rv = dns_getstate(&st); if (rv != 0) return (NS_UNAVAIL); st->counter = 0; return (NS_UNAVAIL); } static int dns_passwd(void *retval, void *mdata, va_list ap) { char buf[HESIOD_NAME_MAX]; struct dns_state *st; struct passwd *pwd; const char *name, *label; void *ctx; char *buffer, **hes; size_t bufsize, linesize; uid_t uid; enum nss_lookup_type how; int rv, *errnop; ctx = NULL; hes = NULL; name = NULL; uid = (uid_t)-1; how = (enum nss_lookup_type)mdata; switch (how) { case nss_lt_name: name = va_arg(ap, const char *); break; case nss_lt_id: uid = va_arg(ap, uid_t); break; case nss_lt_all: break; } pwd = va_arg(ap, struct passwd *); buffer = va_arg(ap, char *); bufsize = va_arg(ap, size_t); errnop = va_arg(ap, int *); *errnop = dns_getstate(&st); if (*errnop != 0) return (NS_UNAVAIL); if (hesiod_init(&ctx) != 0) { *errnop = errno; rv = NS_UNAVAIL; goto fin; } do { rv = NS_NOTFOUND; switch (how) { case nss_lt_name: label = name; break; case nss_lt_id: if (snprintf(buf, sizeof(buf), "%lu", (unsigned long)uid) >= sizeof(buf)) goto fin; label = buf; break; case nss_lt_all: if (st->counter < 0) goto fin; if (snprintf(buf, sizeof(buf), "passwd-%ld", st->counter++) >= sizeof(buf)) goto fin; label = buf; break; } hes = hesiod_resolve(ctx, label, how == nss_lt_id ? "uid" : "passwd"); if (hes == NULL) { if (how == nss_lt_all) st->counter = -1; if (errno != ENOENT) *errnop = errno; goto fin; } rv = __pw_match_entry(hes[0], strlen(hes[0]), how, name, uid); if (rv != NS_SUCCESS) { hesiod_free_list(ctx, hes); hes = NULL; continue; } linesize = strlcpy(buffer, hes[0], bufsize); if (linesize >= bufsize) { *errnop = ERANGE; rv = NS_RETURN; continue; } hesiod_free_list(ctx, hes); hes = NULL; rv = __pw_parse_entry(buffer, bufsize, pwd, 0, errnop); } while (how == nss_lt_all && !(rv & NS_TERMINATE)); fin: if (hes != NULL) hesiod_free_list(ctx, hes); if (ctx != NULL) hesiod_end(ctx); if (rv == NS_SUCCESS) { pwd->pw_fields &= ~_PWF_SOURCE; pwd->pw_fields |= _PWF_HESIOD; if (retval != NULL) *(struct passwd **)retval = pwd; } return (rv); } #endif /* HESIOD */ #ifdef YP /* * nis backend */ static void nis_endstate(void *p) { free(((struct nis_state *)p)->key); free(p); } /* * Test for the presence of special FreeBSD-specific master.passwd.by* * maps. We do this using yp_order(). If it fails, then either the server * doesn't have the map, or the YPPROC_ORDER procedure isn't supported by * the server (Sun NIS+ servers in YP compat mode behave this way). If * the master.passwd.by* maps don't exist, then let the lookup routine try * the regular passwd.by* maps instead. If the lookup routine fails, it * can return an error as needed. */ static int nis_map(char *domain, enum nss_lookup_type how, char *buffer, size_t bufsize, int *master) { int rv, order; *master = 0; if (geteuid() == 0) { if (snprintf(buffer, bufsize, "master.passwd.by%s", (how == nss_lt_id) ? "uid" : "name") >= bufsize) return (NS_UNAVAIL); rv = yp_order(domain, buffer, &order); if (rv == 0) { *master = 1; return (NS_SUCCESS); } } if (snprintf(buffer, bufsize, "passwd.by%s", (how == nss_lt_id) ? "uid" : "name") >= bufsize) return (NS_UNAVAIL); return (NS_SUCCESS); } static int nis_adjunct(char *domain, const char *name, char *buffer, size_t bufsize) { int rv; char *result, *p, *q, *eor; int resultlen; result = NULL; rv = yp_match(domain, "passwd.adjunct.byname", name, strlen(name), &result, &resultlen); if (rv != 0) rv = 1; else { eor = &result[resultlen]; p = memchr(result, ':', eor - result); if (p != NULL && ++p < eor && (q = memchr(p, ':', eor - p)) != NULL) { if (q - p >= bufsize) rv = -1; else { memcpy(buffer, p, q - p); buffer[q - p] ='\0'; } } else rv = 1; } free(result); return (rv); } static int nis_setpwent(void *retval, void *mdata, va_list ap) { struct nis_state *st; int rv; rv = nis_getstate(&st); if (rv != 0) return (NS_UNAVAIL); st->done = 0; free(st->key); st->key = NULL; return (NS_UNAVAIL); } static int nis_passwd(void *retval, void *mdata, va_list ap) { char map[YPMAXMAP]; struct nis_state *st; struct passwd *pwd; const char *name; char *buffer, *key, *result; size_t bufsize; uid_t uid; enum nss_lookup_type how; int *errnop, keylen, resultlen, rv, master; name = NULL; uid = (uid_t)-1; how = (enum nss_lookup_type)mdata; switch (how) { case nss_lt_name: name = va_arg(ap, const char *); break; case nss_lt_id: uid = va_arg(ap, uid_t); break; case nss_lt_all: break; } pwd = va_arg(ap, struct passwd *); buffer = va_arg(ap, char *); bufsize = va_arg(ap, size_t); errnop = va_arg(ap, int *); *errnop = nis_getstate(&st); if (*errnop != 0) return (NS_UNAVAIL); if (st->domain[0] == '\0') { if (getdomainname(st->domain, sizeof(st->domain)) != 0) { *errnop = errno; return (NS_UNAVAIL); } } rv = nis_map(st->domain, how, map, sizeof(map), &master); if (rv != NS_SUCCESS) return (rv); result = NULL; do { rv = NS_NOTFOUND; switch (how) { case nss_lt_name: if (strlcpy(buffer, name, bufsize) >= bufsize) goto erange; break; case nss_lt_id: if (snprintf(buffer, bufsize, "%lu", (unsigned long)uid) >= bufsize) goto erange; break; case nss_lt_all: if (st->done) goto fin; break; } result = NULL; if (how == nss_lt_all) { if (st->key == NULL) rv = yp_first(st->domain, map, &st->key, &st->keylen, &result, &resultlen); else { key = st->key; keylen = st->keylen; st->key = NULL; rv = yp_next(st->domain, map, key, keylen, &st->key, &st->keylen, &result, &resultlen); free(key); } if (rv != 0) { free(result); free(st->key); st->key = NULL; if (rv == YPERR_NOMORE) st->done = 1; else rv = NS_UNAVAIL; goto fin; } } else { rv = yp_match(st->domain, map, buffer, strlen(buffer), &result, &resultlen); if (rv == YPERR_KEY) { rv = NS_NOTFOUND; continue; } else if (rv != 0) { free(result); rv = NS_UNAVAIL; continue; } } if (resultlen >= bufsize) { free(result); goto erange; } memcpy(buffer, result, resultlen); buffer[resultlen] = '\0'; free(result); rv = __pw_match_entry(buffer, resultlen, how, name, uid); if (rv == NS_SUCCESS) rv = __pw_parse_entry(buffer, resultlen, pwd, master, errnop); } while (how == nss_lt_all && !(rv & NS_TERMINATE)); fin: if (rv == NS_SUCCESS) { if (strstr(pwd->pw_passwd, "##") != NULL) { rv = nis_adjunct(st->domain, pwd->pw_name, &buffer[resultlen+1], bufsize-resultlen-1); if (rv < 0) goto erange; else if (rv == 0) pwd->pw_passwd = &buffer[resultlen+1]; } pwd->pw_fields &= ~_PWF_SOURCE; pwd->pw_fields |= _PWF_NIS; if (retval != NULL) *(struct passwd **)retval = pwd; rv = NS_SUCCESS; } return (rv); erange: *errnop = ERANGE; return (NS_RETURN); } #endif /* YP */ /* * compat backend */ static void compat_clear_template(struct passwd *template) { free(template->pw_passwd); free(template->pw_gecos); free(template->pw_dir); free(template->pw_shell); memset(template, 0, sizeof(*template)); } static int compat_set_template(struct passwd *src, struct passwd *template) { compat_clear_template(template); #ifdef PW_OVERRIDE_PASSWD if ((src->pw_fields & _PWF_PASSWD) && (template->pw_passwd = strdup(src->pw_passwd)) == NULL) goto enomem; #endif if (src->pw_fields & _PWF_UID) template->pw_uid = src->pw_uid; if (src->pw_fields & _PWF_GID) template->pw_gid = src->pw_gid; if ((src->pw_fields & _PWF_GECOS) && (template->pw_gecos = strdup(src->pw_gecos)) == NULL) goto enomem; if ((src->pw_fields & _PWF_DIR) && (template->pw_dir = strdup(src->pw_dir)) == NULL) goto enomem; if ((src->pw_fields & _PWF_SHELL) && (template->pw_shell = strdup(src->pw_shell)) == NULL) goto enomem; template->pw_fields = src->pw_fields; return (0); enomem: syslog(LOG_ERR, "getpwent memory allocation failure"); return (-1); } static int compat_use_template(struct passwd *pwd, struct passwd *template, char *buffer, size_t bufsize) { struct passwd hold; char *copy, *p, *q, *eob; size_t n; /* We cannot know the layout of the password fields in `buffer', * so we have to copy everything. */ if (template->pw_fields == 0) /* nothing to fill-in */ return (0); n = 0; n += pwd->pw_name != NULL ? strlen(pwd->pw_name) + 1 : 0; n += pwd->pw_passwd != NULL ? strlen(pwd->pw_passwd) + 1 : 0; n += pwd->pw_class != NULL ? strlen(pwd->pw_class) + 1 : 0; n += pwd->pw_gecos != NULL ? strlen(pwd->pw_gecos) + 1 : 0; n += pwd->pw_dir != NULL ? strlen(pwd->pw_dir) + 1 : 0; n += pwd->pw_shell != NULL ? strlen(pwd->pw_shell) + 1 : 0; copy = malloc(n); if (copy == NULL) { syslog(LOG_ERR, "getpwent memory allocation failure"); return (ENOMEM); } p = copy; eob = ©[n]; #define COPY(field) do { \ if (pwd->field == NULL) \ hold.field = NULL; \ else { \ hold.field = p; \ p += strlcpy(p, pwd->field, eob-p) + 1; \ } \ } while (0) COPY(pw_name); COPY(pw_passwd); COPY(pw_class); COPY(pw_gecos); COPY(pw_dir); COPY(pw_shell); #undef COPY p = buffer; eob = &buffer[bufsize]; #define COPY(field, flag) do { \ q = (template->pw_fields & flag) ? template->field : hold.field; \ if (q == NULL) \ pwd->field = NULL; \ else { \ pwd->field = p; \ if ((n = strlcpy(p, q, eob-p)) >= eob-p) { \ free(copy); \ return (ERANGE); \ } \ p += n + 1; \ } \ } while (0) COPY(pw_name, 0); #ifdef PW_OVERRIDE_PASSWD COPY(pw_passwd, _PWF_PASSWD); #else COPY(pw_passwd, 0); #endif COPY(pw_class, 0); COPY(pw_gecos, _PWF_GECOS); COPY(pw_dir, _PWF_DIR); COPY(pw_shell, _PWF_SHELL); #undef COPY #define COPY(field, flag) do { \ if (template->pw_fields & flag) \ pwd->field = template->field; \ } while (0) COPY(pw_uid, _PWF_UID); COPY(pw_gid, _PWF_GID); #undef COPY free(copy); return (0); } static int compat_exclude(const char *name, DB **db) { DBT key, data; if (*db == NULL && (*db = dbopen(NULL, O_RDWR, 600, DB_HASH, 0)) == NULL) return (errno); key.size = strlen(name); key.data = (char *)name; data.size = 0; data.data = NULL; if ((*db)->put(*db, &key, &data, 0) == -1) return (errno); return (0); } static int compat_is_excluded(const char *name, DB *db) { DBT key, data; if (db == NULL) return (0); key.size = strlen(name); key.data = (char *)name; return (db->get(db, &key, &data, 0) == 0); } static int compat_redispatch(struct compat_state *st, enum nss_lookup_type how, enum nss_lookup_type lookup_how, const char *name, const char *lookup_name, uid_t uid, struct passwd *pwd, char *buffer, size_t bufsize, int *errnop) { static const ns_src compatsrc[] = { #ifdef YP { NSSRC_NIS, NS_SUCCESS }, #endif { NULL, 0 } }; ns_dtab dtab[] = { #ifdef YP { NSSRC_NIS, nis_passwd, NULL }, #endif #ifdef HESIOD { NSSRC_DNS, dns_passwd, NULL }, #endif { NULL, NULL, NULL } }; void *discard; int rv, e, i; for (i = 0; i < sizeof(dtab)/sizeof(dtab[0]) - 1; i++) dtab[i].mdata = (void *)lookup_how; more: pwd_init(pwd); switch (lookup_how) { case nss_lt_all: rv = _nsdispatch(&discard, dtab, NSDB_PASSWD_COMPAT, "getpwent_r", compatsrc, pwd, buffer, bufsize, errnop); break; case nss_lt_id: rv = _nsdispatch(&discard, dtab, NSDB_PASSWD_COMPAT, "getpwuid_r", compatsrc, uid, pwd, buffer, bufsize, errnop); break; case nss_lt_name: rv = _nsdispatch(&discard, dtab, NSDB_PASSWD_COMPAT, "getpwnam_r", compatsrc, lookup_name, pwd, buffer, bufsize, errnop); break; default: return (NS_UNAVAIL); } if (rv != NS_SUCCESS) return (rv); if (compat_is_excluded(pwd->pw_name, st->exclude)) { if (how == nss_lt_all) goto more; return (NS_NOTFOUND); } e = compat_use_template(pwd, &st->template, buffer, bufsize); if (e != 0) { *errnop = e; if (e == ERANGE) return (NS_RETURN); else return (NS_UNAVAIL); } switch (how) { case nss_lt_name: if (strcmp(name, pwd->pw_name) != 0) return (NS_NOTFOUND); break; case nss_lt_id: if (uid != pwd->pw_uid) return (NS_NOTFOUND); break; default: break; } return (NS_SUCCESS); } static void compat_endstate(void *p) { struct compat_state *st; if (p == NULL) return; st = (struct compat_state *)p; if (st->db != NULL) st->db->close(st->db); if (st->exclude != NULL) st->exclude->close(st->exclude); compat_clear_template(&st->template); free(p); } static int compat_setpwent(void *retval, void *mdata, va_list ap) { static const ns_src compatsrc[] = { #ifdef YP { NSSRC_NIS, NS_SUCCESS }, #endif { NULL, 0 } }; ns_dtab dtab[] = { #ifdef YP { NSSRC_NIS, nis_setpwent, NULL }, #endif #ifdef HESIOD { NSSRC_DNS, dns_setpwent, NULL }, #endif { NULL, NULL, NULL } }; struct compat_state *st; int rv, stayopen; #define set_setent(x, y) do { \ int i; \ \ for (i = 0; i < (sizeof(x)/sizeof(x[0])) - 1; i++) \ x[i].mdata = (void *)y; \ } while (0) rv = compat_getstate(&st); if (rv != 0) return (NS_UNAVAIL); switch ((enum constants)mdata) { case SETPWENT: stayopen = va_arg(ap, int); st->keynum = 0; if (stayopen) st->db = pwdbopen(&st->version); st->stayopen = stayopen; set_setent(dtab, mdata); (void)_nsdispatch(NULL, dtab, NSDB_PASSWD_COMPAT, "setpwent", compatsrc, 0); break; case ENDPWENT: if (st->db != NULL) { (void)st->db->close(st->db); st->db = NULL; } set_setent(dtab, mdata); (void)_nsdispatch(NULL, dtab, NSDB_PASSWD_COMPAT, "endpwent", compatsrc, 0); break; default: break; } return (NS_UNAVAIL); #undef set_setent } static int compat_passwd(void *retval, void *mdata, va_list ap) { char keybuf[MAXLOGNAME + 1]; DBT key, entry; struct compat_state *st; enum nss_lookup_type how; const char *name; struct passwd *pwd; char *buffer, *pw_name; char *host, *user, *domain; size_t bufsize; uid_t uid; uint32_t store; int rv, from_compat, stayopen, *errnop; from_compat = 0; name = NULL; uid = (uid_t)-1; how = (enum nss_lookup_type)mdata; switch (how) { case nss_lt_name: name = va_arg(ap, const char *); break; case nss_lt_id: uid = va_arg(ap, uid_t); break; case nss_lt_all: break; default: rv = NS_NOTFOUND; goto fin; } pwd = va_arg(ap, struct passwd *); buffer = va_arg(ap, char *); bufsize = va_arg(ap, size_t); errnop = va_arg(ap, int *); *errnop = compat_getstate(&st); if (*errnop != 0) return (NS_UNAVAIL); if (how == nss_lt_all && st->keynum < 0) { rv = NS_NOTFOUND; goto fin; } if (st->db == NULL && (st->db = pwdbopen(&st->version)) == NULL) { *errnop = errno; rv = NS_UNAVAIL; goto fin; } if (how == nss_lt_all) { if (st->keynum < 0) { rv = NS_NOTFOUND; goto fin; } stayopen = 1; } else { st->keynum = 0; stayopen = st->stayopen; } docompat: rv = NS_NOTFOUND; switch (st->compat) { case COMPAT_MODE_ALL: rv = compat_redispatch(st, how, how, name, name, uid, pwd, buffer, bufsize, errnop); if (rv != NS_SUCCESS) st->compat = COMPAT_MODE_OFF; break; case COMPAT_MODE_NETGROUP: /* XXX getnetgrent is not thread-safe. */ do { rv = getnetgrent(&host, &user, &domain); if (rv == 0) { endnetgrent(); st->compat = COMPAT_MODE_OFF; rv = NS_NOTFOUND; continue; } else if (user == NULL || user[0] == '\0') continue; rv = compat_redispatch(st, how, nss_lt_name, name, user, uid, pwd, buffer, bufsize, errnop); } while (st->compat == COMPAT_MODE_NETGROUP && !(rv & NS_TERMINATE)); break; case COMPAT_MODE_NAME: rv = compat_redispatch(st, how, nss_lt_name, name, st->name, uid, pwd, buffer, bufsize, errnop); free(st->name); st->name = NULL; st->compat = COMPAT_MODE_OFF; break; default: break; } if (rv & NS_TERMINATE) { from_compat = 1; goto fin; } key.data = keybuf; rv = NS_NOTFOUND; while (st->keynum >= 0) { st->keynum++; if (st->version < _PWD_CURRENT_VERSION) { memcpy(&keybuf[1], &st->keynum, sizeof(st->keynum)); key.size = sizeof(st->keynum) + 1; } else { store = htonl(st->keynum); memcpy(&keybuf[1], &store, sizeof(store)); key.size = sizeof(store) + 1; } keybuf[0] = _PW_VERSIONED(_PW_KEYBYNUM, st->version); rv = st->db->get(st->db, &key, &entry, 0); if (rv < 0 || rv > 1) { /* should never return > 1 */ *errnop = errno; rv = NS_UNAVAIL; goto fin; } else if (rv == 1) { st->keynum = -1; rv = NS_NOTFOUND; goto fin; } pw_name = (char *)entry.data; switch (pw_name[0]) { case '+': switch (pw_name[1]) { case '\0': st->compat = COMPAT_MODE_ALL; break; case '@': setnetgrent(&pw_name[2]); st->compat = COMPAT_MODE_NETGROUP; break; default: st->name = strdup(&pw_name[1]); if (st->name == NULL) { syslog(LOG_ERR, "getpwent memory allocation failure"); *errnop = ENOMEM; rv = NS_UNAVAIL; break; } st->compat = COMPAT_MODE_NAME; } if (entry.size > bufsize) { *errnop = ERANGE; rv = NS_RETURN; goto fin; } memcpy(buffer, entry.data, entry.size); rv = pwdb_versions[st->version].parse(buffer, entry.size, pwd, errnop); if (rv != NS_SUCCESS) ; else if (compat_set_template(pwd, &st->template) < 0) { *errnop = ENOMEM; rv = NS_UNAVAIL; goto fin; } goto docompat; case '-': switch (pw_name[1]) { case '\0': /* XXX Maybe syslog warning */ continue; case '@': setnetgrent(&pw_name[2]); while (getnetgrent(&host, &user, &domain) != 0) { if (user != NULL && user[0] != '\0') compat_exclude(user, &st->exclude); } endnetgrent(); continue; default: compat_exclude(&pw_name[1], &st->exclude); continue; } break; default: break; } if (compat_is_excluded((char *)entry.data, st->exclude)) continue; rv = pwdb_versions[st->version].match(entry.data, entry.size, how, name, uid); if (rv == NS_RETURN) break; else if (rv != NS_SUCCESS) continue; if (entry.size > bufsize) { *errnop = ERANGE; rv = NS_RETURN; break; } memcpy(buffer, entry.data, entry.size); rv = pwdb_versions[st->version].parse(buffer, entry.size, pwd, errnop); if (rv & NS_TERMINATE) break; } fin: - if (!stayopen && st->db != NULL) { + if (st->db != NULL && !stayopen) { (void)st->db->close(st->db); st->db = NULL; } if (rv == NS_SUCCESS) { if (!from_compat) { pwd->pw_fields &= ~_PWF_SOURCE; pwd->pw_fields |= _PWF_FILES; } if (retval != NULL) *(struct passwd **)retval = pwd; } return (rv); } /* * common passwd line matching and parsing */ int __pw_match_entry(const char *entry, size_t entrysize, enum nss_lookup_type how, const char *name, uid_t uid) { const char *p, *eom; char *q; size_t len; unsigned long m; eom = entry + entrysize; for (p = entry; p < eom; p++) if (*p == ':') break; if (*p != ':') return (NS_NOTFOUND); if (how == nss_lt_all) return (NS_SUCCESS); if (how == nss_lt_name) { len = strlen(name); if (len == (p - entry) && memcmp(name, entry, len) == 0) return (NS_SUCCESS); else return (NS_NOTFOUND); } for (p++; p < eom; p++) if (*p == ':') break; if (*p != ':') return (NS_NOTFOUND); m = strtoul(++p, &q, 10); if (q[0] != ':' || (uid_t)m != uid) return (NS_NOTFOUND); else return (NS_SUCCESS); } /* XXX buffer must be NUL-terminated. errnop is not set correctly. */ int __pw_parse_entry(char *buffer, size_t bufsize __unused, struct passwd *pwd, int master, int *errnop __unused) { if (__pw_scan(buffer, pwd, master ? _PWSCAN_MASTER : 0) == 0) return (NS_NOTFOUND); else return (NS_SUCCESS); } Index: projects/cxl_iscsi/lib/libc/stdlib/qsort.c =================================================================== --- projects/cxl_iscsi/lib/libc/stdlib/qsort.c (revision 279716) +++ projects/cxl_iscsi/lib/libc/stdlib/qsort.c (revision 279717) @@ -1,195 +1,204 @@ /*- * Copyright (c) 1992, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #if defined(LIBC_SCCS) && !defined(lint) static char sccsid[] = "@(#)qsort.c 8.1 (Berkeley) 6/4/93"; #endif /* LIBC_SCCS and not lint */ #include __FBSDID("$FreeBSD$"); #include #ifdef I_AM_QSORT_R typedef int cmp_t(void *, const void *, const void *); #else typedef int cmp_t(const void *, const void *); #endif static inline char *med3(char *, char *, char *, cmp_t *, void *); -static inline void swapfunc(char *, char *, int, int); +static inline void swapfunc(char *, char *, int, int, int); -#define min(a, b) (a) < (b) ? a : b +#define MIN(a, b) ((a) < (b) ? a : b) /* * Qsort routine from Bentley & McIlroy's "Engineering a Sort Function". */ -#define swapcode(TYPE, parmi, parmj, n) { \ - long i = (n) / sizeof (TYPE); \ - TYPE *pi = (TYPE *) (parmi); \ - TYPE *pj = (TYPE *) (parmj); \ +#define swapcode(TYPE, parmi, parmj, n) { \ + long i = (n) / sizeof (TYPE); \ + TYPE *pi = (TYPE *) (parmi); \ + TYPE *pj = (TYPE *) (parmj); \ do { \ TYPE t = *pi; \ *pi++ = *pj; \ *pj++ = t; \ - } while (--i > 0); \ + } while (--i > 0); \ } -#define SWAPINIT(a, es) swaptype = ((char *)a - (char *)0) % sizeof(long) || \ - es % sizeof(long) ? 2 : es == sizeof(long)? 0 : 1; +#define SWAPINIT(TYPE, a, es) swaptype_ ## TYPE = \ + ((char *)a - (char *)0) % sizeof(TYPE) || \ + es % sizeof(TYPE) ? 2 : es == sizeof(TYPE) ? 0 : 1; static inline void -swapfunc(a, b, n, swaptype) +swapfunc(a, b, n, swaptype_long, swaptype_int) char *a, *b; - int n, swaptype; + int n, swaptype_long, swaptype_int; { - if(swaptype <= 1) + if (swaptype_long <= 1) swapcode(long, a, b, n) + else if (swaptype_int <= 1) + swapcode(int, a, b, n) else swapcode(char, a, b, n) } -#define swap(a, b) \ - if (swaptype == 0) { \ +#define swap(a, b) \ + if (swaptype_long == 0) { \ long t = *(long *)(a); \ *(long *)(a) = *(long *)(b); \ *(long *)(b) = t; \ + } else if (swaptype_int == 0) { \ + int t = *(int *)(a); \ + *(int *)(a) = *(int *)(b); \ + *(int *)(b) = t; \ } else \ - swapfunc(a, b, es, swaptype) + swapfunc(a, b, es, swaptype_long, swaptype_int) -#define vecswap(a, b, n) if ((n) > 0) swapfunc(a, b, n, swaptype) +#define vecswap(a, b, n) \ + if ((n) > 0) swapfunc(a, b, n, swaptype_long, swaptype_int) #ifdef I_AM_QSORT_R #define CMP(t, x, y) (cmp((t), (x), (y))) #else #define CMP(t, x, y) (cmp((x), (y))) #endif static inline char * med3(char *a, char *b, char *c, cmp_t *cmp, void *thunk #ifndef I_AM_QSORT_R __unused #endif ) { return CMP(thunk, a, b) < 0 ? (CMP(thunk, b, c) < 0 ? b : (CMP(thunk, a, c) < 0 ? c : a )) - :(CMP(thunk, b, c) > 0 ? b : (CMP(thunk, a, c) < 0 ? a : c )); + :(CMP(thunk, b, c) > 0 ? b : (CMP(thunk, a, c) < 0 ? a : c )); } #ifdef I_AM_QSORT_R void qsort_r(void *a, size_t n, size_t es, void *thunk, cmp_t *cmp) #else -#define thunk NULL +#define thunk NULL void qsort(void *a, size_t n, size_t es, cmp_t *cmp) #endif { char *pa, *pb, *pc, *pd, *pl, *pm, *pn; size_t d, r; int cmp_result; - int swaptype, swap_cnt; + int swaptype_long, swaptype_int, swap_cnt; -loop: SWAPINIT(a, es); +loop: SWAPINIT(long, a, es); + SWAPINIT(int, a, es); swap_cnt = 0; if (n < 7) { for (pm = (char *)a + es; pm < (char *)a + n * es; pm += es) for (pl = pm; pl > (char *)a && CMP(thunk, pl - es, pl) > 0; pl -= es) swap(pl, pl - es); return; } pm = (char *)a + (n / 2) * es; if (n > 7) { pl = a; pn = (char *)a + (n - 1) * es; if (n > 40) { d = (n / 8) * es; pl = med3(pl, pl + d, pl + 2 * d, cmp, thunk); pm = med3(pm - d, pm, pm + d, cmp, thunk); pn = med3(pn - 2 * d, pn - d, pn, cmp, thunk); } pm = med3(pl, pm, pn, cmp, thunk); } swap(a, pm); pa = pb = (char *)a + es; pc = pd = (char *)a + (n - 1) * es; for (;;) { while (pb <= pc && (cmp_result = CMP(thunk, pb, a)) <= 0) { if (cmp_result == 0) { swap_cnt = 1; swap(pa, pb); pa += es; } pb += es; } while (pb <= pc && (cmp_result = CMP(thunk, pc, a)) >= 0) { if (cmp_result == 0) { swap_cnt = 1; swap(pc, pd); pd -= es; } pc -= es; } if (pb > pc) break; swap(pb, pc); swap_cnt = 1; pb += es; pc -= es; } if (swap_cnt == 0) { /* Switch to insertion sort */ for (pm = (char *)a + es; pm < (char *)a + n * es; pm += es) for (pl = pm; pl > (char *)a && CMP(thunk, pl - es, pl) > 0; pl -= es) swap(pl, pl - es); return; } pn = (char *)a + n * es; - r = min(pa - (char *)a, pb - pa); + r = MIN(pa - (char *)a, pb - pa); vecswap(a, pb - r, r); - r = min(pd - pc, pn - pd - es); + r = MIN(pd - pc, pn - pd - es); vecswap(pb, pn - r, r); if ((r = pb - pa) > es) #ifdef I_AM_QSORT_R qsort_r(a, r / es, es, thunk, cmp); #else qsort(a, r / es, es, cmp); #endif if ((r = pd - pc) > es) { /* Iterate rather than recurse to save stack space */ a = pn - r; n = r / es; goto loop; } /* qsort(pn - r, r / es, es, cmp);*/ } Index: projects/cxl_iscsi/lib/libc/sys/cap_ioctls_limit.2 =================================================================== --- projects/cxl_iscsi/lib/libc/sys/cap_ioctls_limit.2 (revision 279716) +++ projects/cxl_iscsi/lib/libc/sys/cap_ioctls_limit.2 (revision 279717) @@ -1,157 +1,159 @@ .\" .\" Copyright (c) 2012 The FreeBSD Foundation .\" All rights reserved. .\" .\" This documentation was written by Pawel Jakub Dawidek under sponsorship .\" the FreeBSD Foundation. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" $FreeBSD$ .\" .Dd March 27, 2014 .Dt CAP_IOCTLS_LIMIT 2 .Os .Sh NAME .Nm cap_ioctls_limit , .Nm cap_ioctls_get .Nd manage allowed ioctl commands .Sh LIBRARY .Lb libc .Sh SYNOPSIS .In sys/capsicum.h .Ft int .Fn cap_ioctls_limit "int fd" "const unsigned long *cmds" "size_t ncmds" .Ft ssize_t .Fn cap_ioctls_get "int fd" "unsigned long *cmds" "size_t maxcmds" .Sh DESCRIPTION If a file descriptor is granted the .Dv CAP_IOCTL capability right, the list of allowed .Xr ioctl 2 commands can be selectively reduced (but never expanded) with the .Fn cap_ioctls_limit system call. The .Fa cmds argument is an array of .Xr ioctl 2 commands and the .Fa ncmds argument specifies the number of elements in the array. There can be up to .Va 256 elements in the array. +Including an element that has been previously revoked will generate an error. +After a successful call only those listed in the array may be used. .Pp The list of allowed ioctl commands for a given file descriptor can be obtained with the .Fn cap_ioctls_get system call. The .Fa cmds argument points at memory that can hold up to .Fa maxcmds values. The function populates the provided buffer with up to .Fa maxcmds elements, but always returns the total number of ioctl commands allowed for the given file descriptor. The total number of ioctls commands for the given file descriptor can be obtained by passing .Dv NULL as the .Fa cmds argument and .Va 0 as the .Fa maxcmds argument. If all ioctl commands are allowed .Dv ( CAP_IOCTL capability right is assigned to the file descriptor and the .Fn cap_ioctls_limit system call was never called for this file descriptor), the .Fn cap_ioctls_get system call will return .Dv CAP_IOCTLS_ALL and won't modify the buffer pointed to by the .Fa cmds argument. .Sh RETURN VALUES .Rv -std cap_ioctls_limit .Pp The .Fn cap_ioctls_get function, if successful, returns the total number of allowed ioctl commands or the value .Dv CAP_IOCTLS_ALL if all ioctls commands are allowed. On failure the value .Va -1 is returned and the global variable errno is set to indicate the error. .Sh ERRORS .Fn cap_ioctls_limit succeeds unless: .Bl -tag -width Er .It Bq Er EBADF The .Fa fd argument is not a valid descriptor. .It Bq Er EFAULT The .Fa cmds argument points at an invalid address. .It Bq Er EINVAL The .Fa ncmds argument is greater than .Va 256 . .It Bq Er ENOTCAPABLE .Fa cmds would expand the list of allowed .Xr ioctl 2 commands. .El .Pp .Fn cap_ioctls_get succeeds unless: .Bl -tag -width Er .It Bq Er EBADF The .Fa fd argument is not a valid descriptor. .It Bq Er EFAULT The .Fa cmds argument points at invalid address. .El .Sh SEE ALSO .Xr cap_fcntls_limit 2 , .Xr cap_rights_limit 2 , .Xr ioctl 2 .Sh HISTORY Support for capabilities and capabilities mode was developed as part of the .Tn TrustedBSD Project. .Sh AUTHORS This function was created by .An Pawel Jakub Dawidek Aq Mt pawel@dawidek.net under sponsorship of the FreeBSD Foundation. Index: projects/cxl_iscsi/lib/libc =================================================================== --- projects/cxl_iscsi/lib/libc (revision 279716) +++ projects/cxl_iscsi/lib/libc (revision 279717) Property changes on: projects/cxl_iscsi/lib/libc ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/lib/libc:r279515-279716 Index: projects/cxl_iscsi/lib/libmt/mtlib.c =================================================================== --- projects/cxl_iscsi/lib/libmt/mtlib.c (revision 279716) +++ projects/cxl_iscsi/lib/libmt/mtlib.c (revision 279717) @@ -1,756 +1,773 @@ /*- * Copyright (c) 2013, 2014, 2015 Spectra Logic Corporation * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions, and the following disclaimer, * without modification. * 2. Redistributions in binary form must reproduce at minimum a disclaimer * substantially similar to the "NO WARRANTY" disclaimer below * ("Disclaimer") and any redistribution must be conditioned upon * including a substantially similar Disclaimer requirement for further * binary redistribution. * * NO WARRANTY * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT * HOLDERS OR CONTRIBUTORS BE LIABLE FOR SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGES. * * Authors: Ken Merry (Spectra Logic Corporation) */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * Called at the start of each XML element, and includes the list of * attributes for the element. */ void mt_start_element(void *user_data, const char *name, const char **attr) { int i; struct mt_status_data *mtinfo; struct mt_status_entry *entry; mtinfo = (struct mt_status_data *)user_data; if (mtinfo->error != 0) return; mtinfo->level++; if ((u_int)mtinfo->level >= (sizeof(mtinfo->cur_sb) / sizeof(mtinfo->cur_sb[0]))) { mtinfo->error = 1; snprintf(mtinfo->error_str, sizeof(mtinfo->error_str), "%s: too many nesting levels, %zd max", __func__, sizeof(mtinfo->cur_sb) / sizeof(mtinfo->cur_sb[0])); return; } mtinfo->cur_sb[mtinfo->level] = sbuf_new_auto(); if (mtinfo->cur_sb[mtinfo->level] == NULL) { mtinfo->error = 1; snprintf(mtinfo->error_str, sizeof(mtinfo->error_str), "%s: Unable to allocate sbuf", __func__); return; } entry = malloc(sizeof(*entry)); if (entry == NULL) { mtinfo->error = 1; snprintf(mtinfo->error_str, sizeof(mtinfo->error_str), "%s: unable to allocate %zd bytes", __func__, sizeof(*entry)); return; } bzero(entry, sizeof(*entry)); STAILQ_INIT(&entry->nv_list); STAILQ_INIT(&entry->child_entries); entry->entry_name = strdup(name); mtinfo->cur_entry[mtinfo->level] = entry; if (mtinfo->cur_entry[mtinfo->level - 1] == NULL) { STAILQ_INSERT_TAIL(&mtinfo->entries, entry, links); } else { STAILQ_INSERT_TAIL( &mtinfo->cur_entry[mtinfo->level - 1]->child_entries, entry, links); entry->parent = mtinfo->cur_entry[mtinfo->level - 1]; } for (i = 0; attr[i] != NULL; i+=2) { struct mt_status_nv *nv; int need_nv; need_nv = 0; if (strcmp(attr[i], "size") == 0) { entry->size = strtoull(attr[i+1], NULL, 0); } else if (strcmp(attr[i], "type") == 0) { if (strcmp(attr[i+1], "int") == 0) { entry->var_type = MT_TYPE_INT; } else if (strcmp(attr[i+1], "uint") == 0) { entry->var_type = MT_TYPE_UINT; } else if (strcmp(attr[i+1], "str") == 0) { entry->var_type = MT_TYPE_STRING; } else if (strcmp(attr[i+1], "node") == 0) { entry->var_type = MT_TYPE_NODE; } else { need_nv = 1; } } else if (strcmp(attr[i], "fmt") == 0) { entry->fmt = strdup(attr[i+1]); } else if (strcmp(attr[i], "desc") == 0) { entry->desc = strdup(attr[i+1]); } else { need_nv = 1; } if (need_nv != 0) { nv = malloc(sizeof(*nv)); if (nv == NULL) { mtinfo->error = 1; snprintf(mtinfo->error_str, sizeof(mtinfo->error_str), "%s: error allocating %zd bytes", __func__, sizeof(*nv)); } bzero(nv, sizeof(*nv)); nv->name = strdup(attr[i]); nv->value = strdup(attr[i+1]); STAILQ_INSERT_TAIL(&entry->nv_list, nv, links); } } } /* * Called on XML element close. */ void mt_end_element(void *user_data, const char *name) { struct mt_status_data *mtinfo; char *str; mtinfo = (struct mt_status_data *)user_data; if (mtinfo->error != 0) return; if (mtinfo->cur_sb[mtinfo->level] == NULL) { mtinfo->error = 1; snprintf(mtinfo->error_str, sizeof(mtinfo->error_str), "%s: no valid sbuf at level %d (name %s)", __func__, mtinfo->level, name); return; } sbuf_finish(mtinfo->cur_sb[mtinfo->level]); str = strdup(sbuf_data(mtinfo->cur_sb[mtinfo->level])); if (str == NULL) { mtinfo->error = 1; snprintf(mtinfo->error_str, sizeof(mtinfo->error_str), "%s can't allocate %zd bytes for string", __func__, sbuf_len(mtinfo->cur_sb[mtinfo->level])); return; } if (strlen(str) == 0) { free(str); str = NULL; } if (str != NULL) { struct mt_status_entry *entry; entry = mtinfo->cur_entry[mtinfo->level]; switch(entry->var_type) { case MT_TYPE_INT: entry->value_signed = strtoll(str, NULL, 0); break; case MT_TYPE_UINT: entry->value_unsigned = strtoull(str, NULL, 0); break; default: break; } } mtinfo->cur_entry[mtinfo->level]->value = str; sbuf_delete(mtinfo->cur_sb[mtinfo->level]); mtinfo->cur_sb[mtinfo->level] = NULL; mtinfo->cur_entry[mtinfo->level] = NULL; mtinfo->level--; } /* * Called to handle character strings in the current element. */ void mt_char_handler(void *user_data, const XML_Char *str, int len) { struct mt_status_data *mtinfo; mtinfo = (struct mt_status_data *)user_data; if (mtinfo->error != 0) return; sbuf_bcat(mtinfo->cur_sb[mtinfo->level], str, len); } void mt_status_tree_sbuf(struct sbuf *sb, struct mt_status_entry *entry, int indent, void (*sbuf_func)(struct sbuf *sb, struct mt_status_entry *entry, void *arg), void *arg) { struct mt_status_nv *nv; struct mt_status_entry *entry2; if (sbuf_func != NULL) { sbuf_func(sb, entry, arg); } else { sbuf_printf(sb, "%*sname: %s, value: %s, fmt: %s, size: %zd, " "type: %d, desc: %s\n", indent, "", entry->entry_name, entry->value, entry->fmt, entry->size, entry->var_type, entry->desc); STAILQ_FOREACH(nv, &entry->nv_list, links) { sbuf_printf(sb, "%*snv: name: %s, value: %s\n", indent + 1, "", nv->name, nv->value); } } STAILQ_FOREACH(entry2, &entry->child_entries, links) mt_status_tree_sbuf(sb, entry2, indent + 2, sbuf_func, arg); } void mt_status_tree_print(struct mt_status_entry *entry, int indent, void (*print_func)(struct mt_status_entry *entry, void *arg), void *arg) { if (print_func != NULL) { struct mt_status_entry *entry2; print_func(entry, arg); STAILQ_FOREACH(entry2, &entry->child_entries, links) mt_status_tree_print(entry2, indent + 2, print_func, arg); } else { struct sbuf *sb; sb = sbuf_new_auto(); if (sb == NULL) return; mt_status_tree_sbuf(sb, entry, indent, NULL, NULL); sbuf_finish(sb); printf("%s", sbuf_data(sb)); sbuf_delete(sb); } } /* * Given a parameter name in the form "foo" or "foo.bar.baz", traverse the * tree looking for the parameter (the first case) or series of parameters * (second case). */ struct mt_status_entry * mt_entry_find(struct mt_status_entry *entry, char *name) { struct mt_status_entry *entry2; char *tmpname = NULL, *tmpname2 = NULL, *tmpstr = NULL; tmpname = strdup(name); if (tmpname == NULL) goto bailout; /* Save a pointer so we can free this later */ tmpname2 = tmpname; tmpstr = strsep(&tmpname, "."); /* * Is this the entry we're looking for? Or do we have further * child entries that we need to grab? */ if (strcmp(entry->entry_name, tmpstr) == 0) { if (tmpname == NULL) { /* * There are no further child entries to find. We * have a complete match. */ free(tmpname2); return (entry); } else { /* * There are more child entries that we need to find. * Fall through to the recursive search off of this * entry, below. Use tmpname, which will contain * everything after the first period. */ name = tmpname; } } /* * Recursively look for further entries. */ STAILQ_FOREACH(entry2, &entry->child_entries, links) { struct mt_status_entry *entry3; entry3 = mt_entry_find(entry2, name); if (entry3 != NULL) { free(tmpname2); return (entry3); } } bailout: free(tmpname2); return (NULL); } struct mt_status_entry * mt_status_entry_find(struct mt_status_data *status_data, char *name) { struct mt_status_entry *entry, *entry2; STAILQ_FOREACH(entry, &status_data->entries, links) { entry2 = mt_entry_find(entry, name); if (entry2 != NULL) return (entry2); } return (NULL); } void mt_status_entry_free(struct mt_status_entry *entry) { struct mt_status_entry *entry2, *entry3; struct mt_status_nv *nv, *nv2; STAILQ_FOREACH_SAFE(entry2, &entry->child_entries, links, entry3) { STAILQ_REMOVE(&entry->child_entries, entry2, mt_status_entry, links); mt_status_entry_free(entry2); } free(entry->entry_name); free(entry->value); free(entry->fmt); free(entry->desc); STAILQ_FOREACH_SAFE(nv, &entry->nv_list, links, nv2) { STAILQ_REMOVE(&entry->nv_list, nv, mt_status_nv, links); free(nv->name); free(nv->value); free(nv); } free(entry); } void mt_status_free(struct mt_status_data *status_data) { struct mt_status_entry *entry, *entry2; STAILQ_FOREACH_SAFE(entry, &status_data->entries, links, entry2) { STAILQ_REMOVE(&status_data->entries, entry, mt_status_entry, links); mt_status_entry_free(entry); } } void mt_entry_sbuf(struct sbuf *sb, struct mt_status_entry *entry, char *fmt) { switch(entry->var_type) { case MT_TYPE_INT: if (fmt != NULL) sbuf_printf(sb, fmt, (intmax_t)entry->value_signed); else sbuf_printf(sb, "%jd", (intmax_t)entry->value_signed); break; case MT_TYPE_UINT: if (fmt != NULL) sbuf_printf(sb, fmt, (uintmax_t)entry->value_unsigned); else sbuf_printf(sb, "%ju", (uintmax_t)entry->value_unsigned); break; default: if (fmt != NULL) sbuf_printf(sb, fmt, entry->value); else sbuf_printf(sb, "%s", entry->value); break; } } void mt_param_parent_print(struct mt_status_entry *entry, struct mt_print_params *print_params) { if (entry->parent != NULL) mt_param_parent_print(entry->parent, print_params); if (((print_params->flags & MT_PF_INCLUDE_ROOT) == 0) && (strcmp(entry->entry_name, print_params->root_name) == 0)) return; printf("%s.", entry->entry_name); } void mt_param_parent_sbuf(struct sbuf *sb, struct mt_status_entry *entry, struct mt_print_params *print_params) { if (entry->parent != NULL) mt_param_parent_sbuf(sb, entry->parent, print_params); if (((print_params->flags & MT_PF_INCLUDE_ROOT) == 0) && (strcmp(entry->entry_name, print_params->root_name) == 0)) return; sbuf_printf(sb, "%s.", entry->entry_name); } void mt_param_entry_sbuf(struct sbuf *sb, struct mt_status_entry *entry, void *arg) { struct mt_print_params *print_params; print_params = (struct mt_print_params *)arg; /* * We don't want to print nodes. */ if (entry->var_type == MT_TYPE_NODE) return; if ((print_params->flags & MT_PF_FULL_PATH) && (entry->parent != NULL)) mt_param_parent_sbuf(sb, entry->parent, print_params); sbuf_printf(sb, "%s: %s", entry->entry_name, entry->value); if ((print_params->flags & MT_PF_VERBOSE) && (entry->desc != NULL) && (strlen(entry->desc) > 0)) sbuf_printf(sb, " (%s)", entry->desc); sbuf_printf(sb, "\n"); } void mt_param_entry_print(struct mt_status_entry *entry, void *arg) { struct mt_print_params *print_params; print_params = (struct mt_print_params *)arg; /* * We don't want to print nodes. */ if (entry->var_type == MT_TYPE_NODE) return; if ((print_params->flags & MT_PF_FULL_PATH) && (entry->parent != NULL)) mt_param_parent_print(entry->parent, print_params); printf("%s: %s", entry->entry_name, entry->value); if ((print_params->flags & MT_PF_VERBOSE) && (entry->desc != NULL) && (strlen(entry->desc) > 0)) printf(" (%s)", entry->desc); printf("\n"); } int mt_protect_print(struct mt_status_data *status_data, int verbose) { struct mt_status_entry *entry; const char *prot_name = MT_PROTECTION_NAME; struct mt_print_params print_params; snprintf(print_params.root_name, sizeof(print_params.root_name), MT_PARAM_ROOT_NAME); print_params.flags = MT_PF_FULL_PATH; if (verbose != 0) print_params.flags |= MT_PF_VERBOSE; entry = mt_status_entry_find(status_data, __DECONST(char *,prot_name)); if (entry == NULL) return (1); mt_status_tree_print(entry, 0, mt_param_entry_print, &print_params); return (0); } int mt_param_list(struct mt_status_data *status_data, char *param_name, int quiet) { struct mt_status_entry *entry; struct mt_print_params print_params; char root_name[20]; snprintf(root_name, sizeof(root_name), "mtparamget"); strlcpy(print_params.root_name, root_name, sizeof(print_params.root_name)); print_params.flags = MT_PF_FULL_PATH; if (quiet == 0) print_params.flags |= MT_PF_VERBOSE; if (param_name != NULL) { entry = mt_status_entry_find(status_data, param_name); if (entry == NULL) return (1); mt_param_entry_print(entry, &print_params); return (0); } else { entry = mt_status_entry_find(status_data, root_name); STAILQ_FOREACH(entry, &status_data->entries, links) mt_status_tree_print(entry, 0, mt_param_entry_print, &print_params); } return (0); } static struct densities { int dens; int bpmm; int bpi; const char *name; } dens[] = { /* * Taken from T10 Project 997D * SCSI-3 Stream Device Commands (SSC) * Revision 11, 4-Nov-97 * * LTO 1-6 definitions obtained from the eighth edition of the * IBM TotalStorage LTO Ultrium Tape Drive SCSI Reference * (July 2007) and the second edition of the IBM System Storage LTO * Tape Drive SCSI Reference (February 13, 2013). * * IBM 3592 definitions obtained from second edition of the IBM * System Storage Tape Drive 3592 SCSI Reference (May 25, 2012). + * + * DAT-72 and DAT-160 bpi values taken from "HP StorageWorks DAT160 + * tape drive white paper", dated June 2007. + * + * DAT-160 / SDLT220 density code (0x48) conflict information + * found here: + * + * http://h20564.www2.hp.com/hpsc/doc/public/display?docId=emr_na-c01065117&sp4ts.oid=429311 + * (Document ID c01065117) */ /*Num. bpmm bpi Reference */ { 0x1, 32, 800, "X3.22-1983" }, { 0x2, 63, 1600, "X3.39-1986" }, { 0x3, 246, 6250, "X3.54-1986" }, { 0x5, 315, 8000, "X3.136-1986" }, { 0x6, 126, 3200, "X3.157-1987" }, { 0x7, 252, 6400, "X3.116-1986" }, { 0x8, 315, 8000, "X3.158-1987" }, { 0x9, 491, 37871, "X3.180" }, { 0xA, 262, 6667, "X3B5/86-199" }, { 0xB, 63, 1600, "X3.56-1986" }, { 0xC, 500, 12690, "HI-TC1" }, { 0xD, 999, 25380, "HI-TC2" }, { 0xF, 394, 10000, "QIC-120" }, { 0x10, 394, 10000, "QIC-150" }, { 0x11, 630, 16000, "QIC-320" }, { 0x12, 2034, 51667, "QIC-1350" }, { 0x13, 2400, 61000, "X3B5/88-185A" }, { 0x14, 1703, 43245, "X3.202-1991" }, { 0x15, 1789, 45434, "ECMA TC17" }, { 0x16, 394, 10000, "X3.193-1990" }, { 0x17, 1673, 42500, "X3B5/91-174" }, { 0x18, 1673, 42500, "X3B5/92-50" }, { 0x19, 2460, 62500, "DLTapeIII" }, { 0x1A, 3214, 81633, "DLTapeIV(20GB)" }, { 0x1B, 3383, 85937, "DLTapeIV(35GB)" }, { 0x1C, 1654, 42000, "QIC-385M" }, { 0x1D, 1512, 38400, "QIC-410M" }, { 0x1E, 1385, 36000, "QIC-1000C" }, { 0x1F, 2666, 67733, "QIC-2100C" }, { 0x20, 2666, 67733, "QIC-6GB(M)" }, { 0x21, 2666, 67733, "QIC-20GB(C)" }, { 0x22, 1600, 40640, "QIC-2GB(C)" }, { 0x23, 2666, 67733, "QIC-875M" }, { 0x24, 2400, 61000, "DDS-2" }, { 0x25, 3816, 97000, "DDS-3" }, { 0x26, 3816, 97000, "DDS-4" }, { 0x27, 3056, 77611, "Mammoth" }, { 0x28, 1491, 37871, "X3.224" }, { 0x40, 4880, 123952, "LTO-1" }, { 0x41, 3868, 98250, "DLTapeIV(40GB)" }, { 0x42, 7398, 187909, "LTO-2" }, { 0x44, 9638, 244805, "LTO-3" }, { 0x46, 12725, 323215, "LTO-4" }, + { 0x47, 6417, 163000, "DAT-72" }, + /* + * XXX KDM note that 0x48 is also the density code for DAT-160. + * For some reason they used overlapping density codes. + */ +#if 0 + { 0x48, 6870, 174500, "DAT-160" }, +#endif { 0x48, 5236, 133000, "SDLTapeI(110)" }, { 0x49, 7598, 193000, "SDLTapeI(160)" }, { 0x4a, 0, 0, "T10000A" }, { 0x4b, 0, 0, "T10000B" }, { 0x4c, 0, 0, "T10000C" }, { 0x4d, 0, 0, "T10000D" }, { 0x51, 11800, 299720, "3592A1 (unencrypted)" }, { 0x52, 11800, 299720, "3592A2 (unencrypted)" }, { 0x53, 13452, 341681, "3592A3 (unencrypted)" }, { 0x54, 19686, 500024, "3592A4 (unencrypted)" }, { 0x55, 20670, 525018, "3592A5 (unencrypted)" }, { 0x58, 15142, 384607, "LTO-5" }, { 0x5A, 15142, 384607, "LTO-6" }, { 0x71, 11800, 299720, "3592A1 (encrypted)" }, { 0x72, 11800, 299720, "3592A2 (encrypted)" }, { 0x73, 13452, 341681, "3592A3 (encrypted)" }, { 0x74, 19686, 500024, "3592A4 (encrypted)" }, { 0x75, 20670, 525018, "3592A5 (encrypted)" }, { 0x8c, 1789, 45434, "EXB-8500c" }, { 0x90, 1703, 43245, "EXB-8200c" }, { 0, 0, 0, NULL } }; const char * mt_density_name(int density_num) { struct densities *sd; /* densities 0 and 0x7f are handled as special cases */ if (density_num == 0) return ("default"); if (density_num == 0x7f) return ("same"); for (sd = dens; sd->dens != 0; sd++) if (sd->dens == density_num) break; if (sd->dens == 0) return ("UNKNOWN"); return (sd->name); } /* * Given a specific density number, return either the bits per inch or bits * per millimeter for the given density. */ int mt_density_bp(int density_num, int bpi) { struct densities *sd; for (sd = dens; sd->dens; sd++) if (sd->dens == density_num) break; if (sd->dens == 0) return (0); if (bpi) return (sd->bpi); else return (sd->bpmm); } int mt_density_num(const char *density_name) { struct densities *sd; size_t l = strlen(density_name); for (sd = dens; sd->dens; sd++) if (strncasecmp(sd->name, density_name, l) == 0) break; return (sd->dens); } /* * Get the current status XML string. * Returns 0 on success, -1 on failure (with errno set, and *xml_str == NULL). */ int mt_get_xml_str(int mtfd, unsigned long cmd, char **xml_str) { size_t alloc_len = 32768; struct mtextget extget; int error; *xml_str = NULL; for (;;) { bzero(&extget, sizeof(extget)); *xml_str = malloc(alloc_len); if (*xml_str == NULL) return (-1); extget.status_xml = *xml_str; extget.alloc_len = alloc_len; error = ioctl(mtfd, cmd, (caddr_t)&extget); if (error == 0 && extget.status == MT_EXT_GET_OK) break; free(*xml_str); *xml_str = NULL; if (error != 0 || extget.status != MT_EXT_GET_NEED_MORE_SPACE) return (-1); /* The driver needs more space, so double and try again. */ alloc_len *= 2; } return (0); } /* * Populate a struct mt_status_data from the XML string via mt_get_xml_str(). * * Returns XML_STATUS_OK on success. * If XML_STATUS_ERROR is returned, errno may be set to indicate the reason. * The caller must check status_data->error. */ int mt_get_status(char *xml_str, struct mt_status_data *status_data) { XML_Parser parser; int retval; bzero(status_data, sizeof(*status_data)); STAILQ_INIT(&status_data->entries); parser = XML_ParserCreate(NULL); if (parser == NULL) { errno = ENOMEM; return (XML_STATUS_ERROR); } XML_SetUserData(parser, status_data); XML_SetElementHandler(parser, mt_start_element, mt_end_element); XML_SetCharacterDataHandler(parser, mt_char_handler); retval = XML_Parse(parser, xml_str, strlen(xml_str), 1); XML_ParserFree(parser); return (retval); } Index: projects/cxl_iscsi/lib/libucl/Makefile =================================================================== --- projects/cxl_iscsi/lib/libucl/Makefile (revision 279716) +++ projects/cxl_iscsi/lib/libucl/Makefile (revision 279717) @@ -1,27 +1,27 @@ # $FreeBSD$ LIBUCL= ${.CURDIR}/../../contrib/libucl LIB= ucl PRIVATELIB= true SHLIB_MAJOR= 1 SRCS= ucl_emitter_streamline.c \ ucl_emitter_utils.c \ ucl_emitter.c \ ucl_hash.c \ ucl_parser.c \ ucl_schema.c \ ucl_util.c \ xxhash.c .PATH: ${LIBUCL}/src LIBADD= m -WARNS= 2 +WARNS= 1 CFLAGS+= -I${LIBUCL}/include \ -I${LIBUCL}/src \ -I${LIBUCL}/uthash \ -I${LIBUCL}/klib .include Index: projects/cxl_iscsi/sbin/ifconfig/af_inet6.c =================================================================== --- projects/cxl_iscsi/sbin/ifconfig/af_inet6.c (revision 279716) +++ projects/cxl_iscsi/sbin/ifconfig/af_inet6.c (revision 279717) @@ -1,536 +1,540 @@ /* * Copyright (c) 1983, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef lint static const char rcsid[] = "$FreeBSD$"; #endif /* not lint */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* for struct ifaddr */ #include #include #include #include /* Define ND6_INFINITE_LIFETIME */ #include "ifconfig.h" static struct in6_ifreq in6_ridreq; static struct in6_aliasreq in6_addreq = { .ifra_flags = 0, .ifra_lifetime = { 0, 0, ND6_INFINITE_LIFETIME, ND6_INFINITE_LIFETIME } }; static int ip6lifetime; static int prefix(void *, int); static char *sec2str(time_t); static int explicit_prefix = 0; extern void setnd6flags(const char *, int, int, const struct afswtch *); extern void setnd6defif(const char *, int, int, const struct afswtch *); extern void nd6_status(int); static char addr_buf[MAXHOSTNAMELEN *2 + 1]; /*for getnameinfo()*/ static void setifprefixlen(const char *addr, int dummy __unused, int s, const struct afswtch *afp) { if (afp->af_getprefix != NULL) afp->af_getprefix(addr, MASK); explicit_prefix = 1; } static void setip6flags(const char *dummyaddr __unused, int flag, int dummysoc __unused, const struct afswtch *afp) { if (afp->af_af != AF_INET6) err(1, "address flags can be set only for inet6 addresses"); if (flag < 0) in6_addreq.ifra_flags &= ~(-flag); else in6_addreq.ifra_flags |= flag; } static void setip6lifetime(const char *cmd, const char *val, int s, const struct afswtch *afp) { struct timespec now; time_t newval; char *ep; clock_gettime(CLOCK_MONOTONIC_FAST, &now); newval = (time_t)strtoul(val, &ep, 0); if (val == ep) errx(1, "invalid %s", cmd); if (afp->af_af != AF_INET6) errx(1, "%s not allowed for the AF", cmd); if (strcmp(cmd, "vltime") == 0) { in6_addreq.ifra_lifetime.ia6t_expire = now.tv_sec + newval; in6_addreq.ifra_lifetime.ia6t_vltime = newval; } else if (strcmp(cmd, "pltime") == 0) { in6_addreq.ifra_lifetime.ia6t_preferred = now.tv_sec + newval; in6_addreq.ifra_lifetime.ia6t_pltime = newval; } } static void setip6pltime(const char *seconds, int dummy __unused, int s, const struct afswtch *afp) { setip6lifetime("pltime", seconds, s, afp); } static void setip6vltime(const char *seconds, int dummy __unused, int s, const struct afswtch *afp) { setip6lifetime("vltime", seconds, s, afp); } static void setip6eui64(const char *cmd, int dummy __unused, int s, const struct afswtch *afp) { struct ifaddrs *ifap, *ifa; const struct sockaddr_in6 *sin6 = NULL; const struct in6_addr *lladdr = NULL; struct in6_addr *in6; if (afp->af_af != AF_INET6) errx(EXIT_FAILURE, "%s not allowed for the AF", cmd); in6 = (struct in6_addr *)&in6_addreq.ifra_addr.sin6_addr; if (memcmp(&in6addr_any.s6_addr[8], &in6->s6_addr[8], 8) != 0) errx(EXIT_FAILURE, "interface index is already filled"); if (getifaddrs(&ifap) != 0) err(EXIT_FAILURE, "getifaddrs"); for (ifa = ifap; ifa; ifa = ifa->ifa_next) { if (ifa->ifa_addr->sa_family == AF_INET6 && strcmp(ifa->ifa_name, name) == 0) { sin6 = (const struct sockaddr_in6 *)ifa->ifa_addr; if (IN6_IS_ADDR_LINKLOCAL(&sin6->sin6_addr)) { lladdr = &sin6->sin6_addr; break; } } } if (!lladdr) errx(EXIT_FAILURE, "could not determine link local address"); memcpy(&in6->s6_addr[8], &lladdr->s6_addr[8], 8); freeifaddrs(ifap); } static void in6_status(int s __unused, const struct ifaddrs *ifa) { struct sockaddr_in6 *sin, null_sin; struct in6_ifreq ifr6; int s6; u_int32_t flags6; struct in6_addrlifetime lifetime; struct timespec now; int error; clock_gettime(CLOCK_MONOTONIC_FAST, &now); memset(&null_sin, 0, sizeof(null_sin)); sin = (struct sockaddr_in6 *)ifa->ifa_addr; if (sin == NULL) return; strncpy(ifr6.ifr_name, ifr.ifr_name, sizeof(ifr.ifr_name)); if ((s6 = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) { warn("socket(AF_INET6,SOCK_DGRAM)"); return; } ifr6.ifr_addr = *sin; if (ioctl(s6, SIOCGIFAFLAG_IN6, &ifr6) < 0) { warn("ioctl(SIOCGIFAFLAG_IN6)"); close(s6); return; } flags6 = ifr6.ifr_ifru.ifru_flags6; memset(&lifetime, 0, sizeof(lifetime)); ifr6.ifr_addr = *sin; if (ioctl(s6, SIOCGIFALIFETIME_IN6, &ifr6) < 0) { warn("ioctl(SIOCGIFALIFETIME_IN6)"); close(s6); return; } lifetime = ifr6.ifr_ifru.ifru_lifetime; close(s6); error = getnameinfo((struct sockaddr *)sin, sin->sin6_len, addr_buf, sizeof(addr_buf), NULL, 0, NI_NUMERICHOST); if (error != 0) inet_ntop(AF_INET6, &sin->sin6_addr, addr_buf, sizeof(addr_buf)); printf("\tinet6 %s ", addr_buf); if (ifa->ifa_flags & IFF_POINTOPOINT) { sin = (struct sockaddr_in6 *)ifa->ifa_dstaddr; /* * some of the interfaces do not have valid destination * address. */ if (sin != NULL && sin->sin6_family == AF_INET6) { int error; error = getnameinfo((struct sockaddr *)sin, sin->sin6_len, addr_buf, sizeof(addr_buf), NULL, 0, NI_NUMERICHOST); if (error != 0) inet_ntop(AF_INET6, &sin->sin6_addr, addr_buf, sizeof(addr_buf)); printf("--> %s ", addr_buf); } } sin = (struct sockaddr_in6 *)ifa->ifa_netmask; if (sin == NULL) sin = &null_sin; printf("prefixlen %d ", prefix(&sin->sin6_addr, sizeof(struct in6_addr))); if ((flags6 & IN6_IFF_ANYCAST) != 0) printf("anycast "); if ((flags6 & IN6_IFF_TENTATIVE) != 0) printf("tentative "); if ((flags6 & IN6_IFF_DUPLICATED) != 0) printf("duplicated "); if ((flags6 & IN6_IFF_DETACHED) != 0) printf("detached "); if ((flags6 & IN6_IFF_DEPRECATED) != 0) printf("deprecated "); if ((flags6 & IN6_IFF_AUTOCONF) != 0) printf("autoconf "); if ((flags6 & IN6_IFF_TEMPORARY) != 0) printf("temporary "); if ((flags6 & IN6_IFF_PREFER_SOURCE) != 0) printf("prefer_source "); if (((struct sockaddr_in6 *)(ifa->ifa_addr))->sin6_scope_id) printf("scopeid 0x%x ", ((struct sockaddr_in6 *)(ifa->ifa_addr))->sin6_scope_id); if (ip6lifetime && (lifetime.ia6t_preferred || lifetime.ia6t_expire)) { printf("pltime "); if (lifetime.ia6t_preferred) { printf("%s ", lifetime.ia6t_preferred < now.tv_sec ? "0" : sec2str(lifetime.ia6t_preferred - now.tv_sec)); } else printf("infty "); printf("vltime "); if (lifetime.ia6t_expire) { printf("%s ", lifetime.ia6t_expire < now.tv_sec ? "0" : sec2str(lifetime.ia6t_expire - now.tv_sec)); } else printf("infty "); } print_vhid(ifa, " "); putchar('\n'); } #define SIN6(x) ((struct sockaddr_in6 *) &(x)) static struct sockaddr_in6 *sin6tab[] = { SIN6(in6_ridreq.ifr_addr), SIN6(in6_addreq.ifra_addr), SIN6(in6_addreq.ifra_prefixmask), SIN6(in6_addreq.ifra_dstaddr) }; static void in6_getprefix(const char *plen, int which) { struct sockaddr_in6 *sin = sin6tab[which]; u_char *cp; int len = atoi(plen); if ((len < 0) || (len > 128)) errx(1, "%s: bad value", plen); sin->sin6_len = sizeof(*sin); if (which != MASK) sin->sin6_family = AF_INET6; if ((len == 0) || (len == 128)) { memset(&sin->sin6_addr, 0xff, sizeof(struct in6_addr)); return; } memset((void *)&sin->sin6_addr, 0x00, sizeof(sin->sin6_addr)); for (cp = (u_char *)&sin->sin6_addr; len > 7; len -= 8) *cp++ = 0xff; *cp = 0xff << (8 - len); } static void in6_getaddr(const char *s, int which) { struct sockaddr_in6 *sin = sin6tab[which]; struct addrinfo hints, *res; int error = -1; newaddr &= 1; sin->sin6_len = sizeof(*sin); if (which != MASK) sin->sin6_family = AF_INET6; if (which == ADDR) { char *p = NULL; if((p = strrchr(s, '/')) != NULL) { *p = '\0'; in6_getprefix(p + 1, MASK); explicit_prefix = 1; } } if (sin->sin6_family == AF_INET6) { bzero(&hints, sizeof(struct addrinfo)); hints.ai_family = AF_INET6; error = getaddrinfo(s, NULL, &hints, &res); } if (error != 0) { if (inet_pton(AF_INET6, s, &sin->sin6_addr) != 1) errx(1, "%s: bad value", s); } else bcopy(res->ai_addr, sin, res->ai_addrlen); } static int prefix(void *val, int size) { u_char *name = (u_char *)val; int byte, bit, plen = 0; for (byte = 0; byte < size; byte++, plen += 8) if (name[byte] != 0xff) break; if (byte == size) return (plen); for (bit = 7; bit != 0; bit--, plen++) if (!(name[byte] & (1 << bit))) break; for (; bit != 0; bit--) if (name[byte] & (1 << bit)) return(0); byte++; for (; byte < size; byte++) if (name[byte]) return(0); return (plen); } static char * sec2str(time_t total) { static char result[256]; int days, hours, mins, secs; int first = 1; char *p = result; if (0) { days = total / 3600 / 24; hours = (total / 3600) % 24; mins = (total / 60) % 60; secs = total % 60; if (days) { first = 0; p += sprintf(p, "%dd", days); } if (!first || hours) { first = 0; p += sprintf(p, "%dh", hours); } if (!first || mins) { first = 0; p += sprintf(p, "%dm", mins); } sprintf(p, "%ds", secs); } else sprintf(result, "%lu", (unsigned long)total); return(result); } static void in6_postproc(int s, const struct afswtch *afp) { if (explicit_prefix == 0) { /* Aggregatable address architecture defines all prefixes are 64. So, it is convenient to set prefixlen to 64 if it is not specified. */ setifprefixlen("64", 0, s, afp); /* in6_getprefix("64", MASK) if MASK is available here... */ } } static void in6_status_tunnel(int s) { char src[NI_MAXHOST]; char dst[NI_MAXHOST]; struct in6_ifreq in6_ifr; const struct sockaddr *sa = (const struct sockaddr *) &in6_ifr.ifr_addr; memset(&in6_ifr, 0, sizeof(in6_ifr)); strncpy(in6_ifr.ifr_name, name, IFNAMSIZ); if (ioctl(s, SIOCGIFPSRCADDR_IN6, (caddr_t)&in6_ifr) < 0) return; if (sa->sa_family != AF_INET6) return; if (getnameinfo(sa, sa->sa_len, src, sizeof(src), 0, 0, NI_NUMERICHOST) != 0) src[0] = '\0'; if (ioctl(s, SIOCGIFPDSTADDR_IN6, (caddr_t)&in6_ifr) < 0) return; if (sa->sa_family != AF_INET6) return; if (getnameinfo(sa, sa->sa_len, dst, sizeof(dst), 0, 0, NI_NUMERICHOST) != 0) dst[0] = '\0'; printf("\ttunnel inet6 %s --> %s\n", src, dst); } static void in6_set_tunnel(int s, struct addrinfo *srcres, struct addrinfo *dstres) { struct in6_aliasreq in6_addreq; memset(&in6_addreq, 0, sizeof(in6_addreq)); strncpy(in6_addreq.ifra_name, name, IFNAMSIZ); memcpy(&in6_addreq.ifra_addr, srcres->ai_addr, srcres->ai_addr->sa_len); memcpy(&in6_addreq.ifra_dstaddr, dstres->ai_addr, dstres->ai_addr->sa_len); if (ioctl(s, SIOCSIFPHYADDR_IN6, &in6_addreq) < 0) warn("SIOCSIFPHYADDR_IN6"); } static struct cmd inet6_cmds[] = { DEF_CMD_ARG("prefixlen", setifprefixlen), DEF_CMD("anycast", IN6_IFF_ANYCAST, setip6flags), DEF_CMD("tentative", IN6_IFF_TENTATIVE, setip6flags), DEF_CMD("-tentative", -IN6_IFF_TENTATIVE, setip6flags), DEF_CMD("deprecated", IN6_IFF_DEPRECATED, setip6flags), DEF_CMD("-deprecated", -IN6_IFF_DEPRECATED, setip6flags), DEF_CMD("autoconf", IN6_IFF_AUTOCONF, setip6flags), DEF_CMD("-autoconf", -IN6_IFF_AUTOCONF, setip6flags), DEF_CMD("prefer_source",IN6_IFF_PREFER_SOURCE, setip6flags), DEF_CMD("-prefer_source",-IN6_IFF_PREFER_SOURCE,setip6flags), DEF_CMD("accept_rtadv", ND6_IFF_ACCEPT_RTADV, setnd6flags), DEF_CMD("-accept_rtadv",-ND6_IFF_ACCEPT_RTADV, setnd6flags), DEF_CMD("no_radr", ND6_IFF_NO_RADR, setnd6flags), DEF_CMD("-no_radr", -ND6_IFF_NO_RADR, setnd6flags), DEF_CMD("defaultif", 1, setnd6defif), DEF_CMD("-defaultif", -1, setnd6defif), DEF_CMD("ifdisabled", ND6_IFF_IFDISABLED, setnd6flags), DEF_CMD("-ifdisabled", -ND6_IFF_IFDISABLED, setnd6flags), DEF_CMD("nud", ND6_IFF_PERFORMNUD, setnd6flags), DEF_CMD("-nud", -ND6_IFF_PERFORMNUD, setnd6flags), DEF_CMD("auto_linklocal",ND6_IFF_AUTO_LINKLOCAL,setnd6flags), DEF_CMD("-auto_linklocal",-ND6_IFF_AUTO_LINKLOCAL,setnd6flags), DEF_CMD("no_prefer_iface",ND6_IFF_NO_PREFER_IFACE,setnd6flags), DEF_CMD("-no_prefer_iface",-ND6_IFF_NO_PREFER_IFACE,setnd6flags), + DEF_CMD("no_dad", ND6_IFF_NO_DAD, setnd6flags), + DEF_CMD("-no_dad", -ND6_IFF_NO_DAD, setnd6flags), + DEF_CMD("ignoreloop", ND6_IFF_IGNORELOOP, setnd6flags), + DEF_CMD("-ignoreloop", -ND6_IFF_IGNORELOOP, setnd6flags), DEF_CMD_ARG("pltime", setip6pltime), DEF_CMD_ARG("vltime", setip6vltime), DEF_CMD("eui64", 0, setip6eui64), }; static struct afswtch af_inet6 = { .af_name = "inet6", .af_af = AF_INET6, .af_status = in6_status, .af_getaddr = in6_getaddr, .af_getprefix = in6_getprefix, .af_other_status = nd6_status, .af_postproc = in6_postproc, .af_status_tunnel = in6_status_tunnel, .af_settunnel = in6_set_tunnel, .af_difaddr = SIOCDIFADDR_IN6, .af_aifaddr = SIOCAIFADDR_IN6, .af_ridreq = &in6_addreq, .af_addreq = &in6_addreq, }; static void in6_Lopt_cb(const char *optarg __unused) { ip6lifetime++; /* print IPv6 address lifetime */ } static struct option in6_Lopt = { .opt = "L", .opt_usage = "[-L]", .cb = in6_Lopt_cb }; static __constructor void inet6_ctor(void) { #define N(a) (sizeof(a) / sizeof(a[0])) size_t i; #ifndef RESCUE if (!feature_present("inet6")) return; #endif for (i = 0; i < N(inet6_cmds); i++) cmd_register(&inet6_cmds[i]); af_register(&af_inet6); opt_register(&in6_Lopt); #undef N } Index: projects/cxl_iscsi/sbin/ifconfig/af_nd6.c =================================================================== --- projects/cxl_iscsi/sbin/ifconfig/af_nd6.c (revision 279716) +++ projects/cxl_iscsi/sbin/ifconfig/af_nd6.c (revision 279717) @@ -1,169 +1,170 @@ /* * Copyright (c) 2009 Hiroki Sato. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef lint static const char rcsid[] = "$FreeBSD$"; #endif /* not lint */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "ifconfig.h" #define MAX_SYSCTL_TRY 5 #define ND6BITS "\020\001PERFORMNUD\002ACCEPT_RTADV\003PREFER_SOURCE" \ "\004IFDISABLED\005DONT_SET_IFROUTE\006AUTO_LINKLOCAL" \ - "\007NO_RADR\010NO_PREFER_IFACE\020DEFAULTIF" + "\007NO_RADR\010NO_PREFER_IFACE\011IGNORELOOP\012NO_DAD" \ + "\020DEFAULTIF" static int isnd6defif(int); void setnd6flags(const char *, int, int, const struct afswtch *); void setnd6defif(const char *, int, int, const struct afswtch *); void nd6_status(int); void setnd6flags(const char *dummyaddr __unused, int d, int s, const struct afswtch *afp) { struct in6_ndireq nd; int error; memset(&nd, 0, sizeof(nd)); strncpy(nd.ifname, ifr.ifr_name, sizeof(nd.ifname)); error = ioctl(s, SIOCGIFINFO_IN6, &nd); if (error) { warn("ioctl(SIOCGIFINFO_IN6)"); return; } if (d < 0) nd.ndi.flags &= ~(-d); else nd.ndi.flags |= d; error = ioctl(s, SIOCSIFINFO_IN6, (caddr_t)&nd); if (error) warn("ioctl(SIOCSIFINFO_IN6)"); } void setnd6defif(const char *dummyaddr __unused, int d, int s, const struct afswtch *afp) { struct in6_ndifreq ndifreq; int ifindex; int error; memset(&ndifreq, 0, sizeof(ndifreq)); strncpy(ndifreq.ifname, ifr.ifr_name, sizeof(ndifreq.ifname)); if (d < 0) { if (isnd6defif(s)) { /* ifindex = 0 means to remove default if */ ifindex = 0; } else return; } else if ((ifindex = if_nametoindex(ndifreq.ifname)) == 0) { warn("if_nametoindex(%s)", ndifreq.ifname); return; } ndifreq.ifindex = ifindex; error = ioctl(s, SIOCSDEFIFACE_IN6, (caddr_t)&ndifreq); if (error) warn("ioctl(SIOCSDEFIFACE_IN6)"); } static int isnd6defif(int s) { struct in6_ndifreq ndifreq; unsigned int ifindex; int error; memset(&ndifreq, 0, sizeof(ndifreq)); strncpy(ndifreq.ifname, ifr.ifr_name, sizeof(ndifreq.ifname)); ifindex = if_nametoindex(ndifreq.ifname); error = ioctl(s, SIOCGDEFIFACE_IN6, (caddr_t)&ndifreq); if (error) { warn("ioctl(SIOCGDEFIFACE_IN6)"); return (error); } return (ndifreq.ifindex == ifindex); } void nd6_status(int s) { struct in6_ndireq nd; int s6; int error; int isdefif; memset(&nd, 0, sizeof(nd)); strncpy(nd.ifname, ifr.ifr_name, sizeof(nd.ifname)); if ((s6 = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) { if (errno != EAFNOSUPPORT && errno != EPROTONOSUPPORT) warn("socket(AF_INET6, SOCK_DGRAM)"); return; } error = ioctl(s6, SIOCGIFINFO_IN6, &nd); if (error) { if (errno != EPFNOSUPPORT) warn("ioctl(SIOCGIFINFO_IN6)"); close(s6); return; } isdefif = isnd6defif(s6); close(s6); if (nd.ndi.flags == 0 && !isdefif) return; printb("\tnd6 options", (unsigned int)(nd.ndi.flags | (isdefif << 15)), ND6BITS); putchar('\n'); } Index: projects/cxl_iscsi/sbin/ifconfig/ifconfig.8 =================================================================== --- projects/cxl_iscsi/sbin/ifconfig/ifconfig.8 (revision 279716) +++ projects/cxl_iscsi/sbin/ifconfig/ifconfig.8 (revision 279717) @@ -1,2844 +1,2858 @@ .\" Copyright (c) 1983, 1991, 1993 .\" The Regents of the University of California. All rights reserved. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" 4. Neither the name of the University nor the names of its contributors .\" may be used to endorse or promote products derived from this software .\" without specific prior written permission. .\" .\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" From: @(#)ifconfig.8 8.3 (Berkeley) 1/5/94 .\" $FreeBSD$ .\" -.Dd December 16, 2014 +.Dd March 6, 2015 .Dt IFCONFIG 8 .Os .Sh NAME .Nm ifconfig .Nd configure network interface parameters .Sh SYNOPSIS .Nm .Op Fl L .Op Fl k .Op Fl m .Op Fl n .Ar interface .Op Cm create .Ar address_family .Oo .Ar address .Op Ar dest_address .Oc .Op Ar parameters .Nm .Ar interface .Cm destroy .Nm .Fl a .Op Fl L .Op Fl d .Op Fl m .Op Fl u .Op Fl v .Op Ar address_family .Nm .Fl l .Op Fl d .Op Fl u .Op Ar address_family .Nm .Op Fl L .Op Fl d .Op Fl k .Op Fl m .Op Fl u .Op Fl v .Op Fl C .Nm .Op Fl g Ar groupname .Sh DESCRIPTION The .Nm utility is used to assign an address to a network interface and/or configure network interface parameters. The .Nm utility must be used at boot time to define the network address of each interface present on a machine; it may also be used at a later time to redefine an interface's address or other operating parameters. .Pp The following options are available: .Bl -tag -width indent .It Ar address For the .Tn DARPA Ns -Internet family, the address is either a host name present in the host name data base, .Xr hosts 5 , or a .Tn DARPA Internet address expressed in the Internet standard .Dq dot notation . .Pp It is also possible to use the CIDR notation (also known as the slash notation) to include the netmask. That is, one can specify an address like .Li 192.168.0.1/16 . .Pp For the .Dq inet6 family, it is also possible to specify the prefix length using the slash notation, like .Li ::1/128 . See the .Cm prefixlen parameter below for more information. .\" For the Xerox Network Systems(tm) family, .\" addresses are .\" .Ar net:a.b.c.d.e.f , .\" where .\" .Ar net .\" is the assigned network number (in decimal), .\" and each of the six bytes of the host number, .\" .Ar a .\" through .\" .Ar f , .\" are specified in hexadecimal. .\" The host number may be omitted on IEEE 802 protocol .\" (Ethernet, FDDI, and Token Ring) interfaces, .\" which use the hardware physical address, .\" and on interfaces other than the first. .\" For the .\" .Tn ISO .\" family, addresses are specified as a long hexadecimal string, .\" as in the Xerox family. .\" However, two consecutive dots imply a zero .\" byte, and the dots are optional, if the user wishes to (carefully) .\" count out long strings of digits in network byte order. .Pp The link-level .Pq Dq link address is specified as a series of colon-separated hex digits. This can be used to, for example, set a new MAC address on an Ethernet interface, though the mechanism used is not Ethernet specific. If the interface is already up when this option is used, it will be briefly brought down and then brought back up again in order to ensure that the receive filter in the underlying Ethernet hardware is properly reprogrammed. .It Ar address_family Specify the address family which affects interpretation of the remaining parameters. Since an interface can receive transmissions in differing protocols with different naming schemes, specifying the address family is recommended. The address or protocol families currently supported are .Dq inet , .Dq inet6 , and .Dq link . The default if available is .Dq inet or otherwise .Dq link . .Dq ether and .Dq lladdr are synonyms for .Dq link . When using the .Fl l flag, the .Dq ether address family has special meaning and is no longer synonymous with .Dq link or .Dq lladdr . Specifying .Fl l Dq ether will list only Ethernet interfaces, excluding all other interface types, including the loopback interface. .It Ar dest_address Specify the address of the correspondent on the other end of a point to point link. .It Ar interface This parameter is a string of the form .Dq name unit , for example, .Dq Li ed0 . .It Ar groupname List the interfaces in the given group. .El .Pp The following parameters may be set with .Nm : .Bl -tag -width indent .It Cm add Another name for the .Cm alias parameter. Introduced for compatibility with .Bsx . .It Cm alias Establish an additional network address for this interface. This is sometimes useful when changing network numbers, and one wishes to accept packets addressed to the old interface. If the address is on the same subnet as the first network address for this interface, a non-conflicting netmask must be given. Usually .Li 0xffffffff is most appropriate. .It Fl alias Remove the network address specified. This would be used if you incorrectly specified an alias, or it was no longer needed. If you have incorrectly set an NS address having the side effect of specifying the host portion, removing all NS addresses will allow you to respecify the host portion. .It Cm anycast (Inet6 only.) Specify that the address configured is an anycast address. Based on the current specification, only routers may configure anycast addresses. Anycast address will not be used as source address of any of outgoing IPv6 packets. .It Cm arp Enable the use of the Address Resolution Protocol .Pq Xr arp 4 in mapping between network level addresses and link level addresses (default). This is currently implemented for mapping between .Tn DARPA Internet addresses and .Tn IEEE 802 48-bit MAC addresses (Ethernet, FDDI, and Token Ring addresses). .It Fl arp Disable the use of the Address Resolution Protocol .Pq Xr arp 4 . .It Cm staticarp If the Address Resolution Protocol is enabled, the host will only reply to requests for its addresses, and will never send any requests. .It Fl staticarp If the Address Resolution Protocol is enabled, the host will perform normally, sending out requests and listening for replies. .It Cm broadcast (Inet only.) Specify the address to use to represent broadcasts to the network. The default broadcast address is the address with a host part of all 1's. .It Cm debug Enable driver dependent debugging code; usually, this turns on extra console error logging. .It Fl debug Disable driver dependent debugging code. .It Cm promisc Put interface into permanently promiscuous mode. .It Fl promisc Disable permanently promiscuous mode. .It Cm delete Another name for the .Fl alias parameter. .It Cm description Ar value , Cm descr Ar value Specify a description of the interface. This can be used to label interfaces in situations where they may otherwise be difficult to distinguish. .It Cm -description , Cm -descr Clear the interface description. .It Cm down Mark an interface .Dq down . When an interface is marked .Dq down , the system will not attempt to transmit messages through that interface. If possible, the interface will be reset to disable reception as well. This action does not automatically disable routes using the interface. .It Cm group Ar group-name Assign the interface to a .Dq group . Any interface can be in multiple groups. .Pp Cloned interfaces are members of their interface family group by default. For example, a PPP interface such as .Em ppp0 is a member of the PPP interface family group, .Em ppp . .\" The interface(s) the default route(s) point to are members of the .\" .Em egress .\" interface group. .It Cm -group Ar group-name Remove the interface from the given .Dq group . .It Cm eui64 (Inet6 only.) Fill interface index (lowermost 64bit of an IPv6 address) automatically. .It Cm fib Ar fib_number Specify interface FIB. A FIB .Ar fib_number is assigned to all frames or packets received on that interface. The FIB is not inherited, e.g., vlans or other sub-interfaces will use the default FIB (0) irrespective of the parent interface's FIB. The kernel needs to be tuned to support more than the default FIB using the .Va ROUTETABLES kernel configuration option, or the .Va net.fibs tunable. .It Cm maclabel Ar label If Mandatory Access Control support is enabled in the kernel, set the MAC label to .Ar label . .\" (see .\" .Xr maclabel 7 ) . .It Cm media Ar type If the driver supports the media selection system, set the media type of the interface to .Ar type . Some interfaces support the mutually exclusive use of one of several different physical media connectors. For example, a 10Mbit/s Ethernet interface might support the use of either .Tn AUI or twisted pair connectors. Setting the media type to .Cm 10base5/AUI would change the currently active connector to the AUI port. Setting it to .Cm 10baseT/UTP would activate twisted pair. Refer to the interfaces' driver specific documentation or man page for a complete list of the available types. .It Cm mediaopt Ar opts If the driver supports the media selection system, set the specified media options on the interface. The .Ar opts argument is a comma delimited list of options to apply to the interface. Refer to the interfaces' driver specific man page for a complete list of available options. .It Fl mediaopt Ar opts If the driver supports the media selection system, disable the specified media options on the interface. .It Cm mode Ar mode If the driver supports the media selection system, set the specified operating mode on the interface to .Ar mode . For IEEE 802.11 wireless interfaces that support multiple operating modes this directive is used to select between 802.11a .Pq Cm 11a , 802.11b .Pq Cm 11b , and 802.11g .Pq Cm 11g operating modes. .It Cm inst Ar minst , Cm instance Ar minst Set the media instance to .Ar minst . This is useful for devices which have multiple physical layer interfaces .Pq PHYs . .It Cm name Ar name Set the interface name to .Ar name . .It Cm rxcsum , txcsum , rxcsum6 , txcsum6 If the driver supports user-configurable checksum offloading, enable receive (or transmit) checksum offloading on the interface. The feature can be turned on selectively per protocol family. Use .Cm rxcsum6 , txcsum6 for .Xr ip6 4 or .Cm rxcsum , txcsum otherwise. Some drivers may not be able to enable these flags independently of each other, so setting one may also set the other. The driver will offload as much checksum work as it can reliably support, the exact level of offloading varies between drivers. .It Fl rxcsum , txcsum , rxcsum6 , txcsum6 If the driver supports user-configurable checksum offloading, disable receive (or transmit) checksum offloading on the interface. The feature can be turned off selectively per protocol family. Use .Fl rxcsum6 , txcsum6 for .Xr ip6 4 or .Fl rxcsum , txcsum otherwise. These settings may not always be independent of each other. .It Cm tso If the driver supports .Xr tcp 4 segmentation offloading, enable TSO on the interface. Some drivers may not be able to support TSO for .Xr ip 4 and .Xr ip6 4 packets, so they may enable only one of them. .It Fl tso If the driver supports .Xr tcp 4 segmentation offloading, disable TSO on the interface. It will always disable TSO for .Xr ip 4 and .Xr ip6 4 . .It Cm tso6 , tso4 If the driver supports .Xr tcp 4 segmentation offloading for .Xr ip6 4 or .Xr ip 4 use one of these to selectively enabled it only for one protocol family. .It Fl tso6 , tso4 If the driver supports .Xr tcp 4 segmentation offloading for .Xr ip6 4 or .Xr ip 4 use one of these to selectively disable it only for one protocol family. .It Cm lro If the driver supports .Xr tcp 4 large receive offloading, enable LRO on the interface. .It Fl lro If the driver supports .Xr tcp 4 large receive offloading, disable LRO on the interface. .It Cm wol , wol_ucast , wol_mcast , wol_magic Enable Wake On Lan (WOL) support, if available. WOL is a facility whereby a machine in a low power state may be woken in response to a received packet. There are three types of packets that may wake a system: ucast (directed solely to the machine's mac address), mcast (directed to a broadcast or multicast address), or magic (unicast or multicast frames with a ``magic contents''). Not all devices support WOL, those that do indicate the mechanisms they support in their capabilities. .Cm wol is a synonym for enabling all available WOL mechanisms. To disable WOL use .Fl wol . .It Cm vlanmtu , vlanhwtag, vlanhwfilter, vlanhwcsum, vlanhwtso If the driver offers user-configurable VLAN support, enable reception of extended frames, tag processing in hardware, frame filtering in hardware, checksum offloading, or TSO on VLAN, respectively. Note that this must be issued on a physical interface associated with .Xr vlan 4 , not on a .Xr vlan 4 interface itself. .It Fl vlanmtu , vlanhwtag, vlanhwfilter, vlanhwtso If the driver offers user-configurable VLAN support, disable reception of extended frames, tag processing in hardware, frame filtering in hardware, or TSO on VLAN, respectively. .It Cm vnet Ar jail Move the interface to the .Xr jail 8 , specified by name or JID. If the jail has a virtual network stack, the interface will disappear from the current environment and become visible to the jail. .It Fl vnet Ar jail Reclaim the interface from the .Xr jail 8 , specified by name or JID. If the jail has a virtual network stack, the interface will disappear from the jail, and become visible to the current network environment. .It Cm polling Turn on .Xr polling 4 feature and disable interrupts on the interface, if driver supports this mode. .It Fl polling Turn off .Xr polling 4 feature and enable interrupt mode on the interface. .It Cm create Create the specified network pseudo-device. If the interface is given without a unit number, try to create a new device with an arbitrary unit number. If creation of an arbitrary device is successful, the new device name is printed to standard output unless the interface is renamed or destroyed in the same .Nm invocation. .It Cm destroy Destroy the specified network pseudo-device. .It Cm plumb Another name for the .Cm create parameter. Included for .Tn Solaris compatibility. .It Cm unplumb Another name for the .Cm destroy parameter. Included for .Tn Solaris compatibility. .It Cm metric Ar n Set the routing metric of the interface to .Ar n , default 0. The routing metric is used by the routing protocol .Pq Xr routed 8 . Higher metrics have the effect of making a route less favorable; metrics are counted as additional hops to the destination network or host. .It Cm mtu Ar n Set the maximum transmission unit of the interface to .Ar n , default is interface specific. The MTU is used to limit the size of packets that are transmitted on an interface. Not all interfaces support setting the MTU, and some interfaces have range restrictions. .It Cm netmask Ar mask .\" (Inet and ISO.) (Inet only.) Specify how much of the address to reserve for subdividing networks into sub-networks. The mask includes the network part of the local address and the subnet part, which is taken from the host field of the address. The mask can be specified as a single hexadecimal number with a leading .Ql 0x , with a dot-notation Internet address, or with a pseudo-network name listed in the network table .Xr networks 5 . The mask contains 1's for the bit positions in the 32-bit address which are to be used for the network and subnet parts, and 0's for the host part. The mask should contain at least the standard network portion, and the subnet field should be contiguous with the network portion. .Pp The netmask can also be specified in CIDR notation after the address. See the .Ar address option above for more information. .It Cm prefixlen Ar len (Inet6 only.) Specify that .Ar len bits are reserved for subdividing networks into sub-networks. The .Ar len must be integer, and for syntactical reason it must be between 0 to 128. It is almost always 64 under the current IPv6 assignment rule. If the parameter is omitted, 64 is used. .Pp The prefix can also be specified using the slash notation after the address. See the .Ar address option above for more information. .It Cm remove Another name for the .Fl alias parameter. Introduced for compatibility with .Bsx . .Sm off .It Cm link Op Cm 0 No - Cm 2 .Sm on Enable special processing of the link level of the interface. These three options are interface specific in actual effect, however, they are in general used to select special modes of operation. An example of this is to enable SLIP compression, or to select the connector type for some Ethernet cards. Refer to the man page for the specific driver for more information. .Sm off .It Fl link Op Cm 0 No - Cm 2 .Sm on Disable special processing at the link level with the specified interface. .It Cm monitor Put the interface in monitor mode. No packets are transmitted, and received packets are discarded after .Xr bpf 4 processing. .It Fl monitor Take the interface out of monitor mode. .It Cm up Mark an interface .Dq up . This may be used to enable an interface after an .Dq Nm Cm down . It happens automatically when setting the first address on an interface. If the interface was reset when previously marked down, the hardware will be re-initialized. .El .Pp The following parameters are for ICMPv6 Neighbor Discovery Protocol. Note that the address family keyword .Dq Li inet6 is needed for them: .Bl -tag -width indent .It Cm accept_rtadv Set a flag to enable accepting ICMPv6 Router Advertisement messages. The .Xr sysctl 8 variable .Va net.inet6.ip6.accept_rtadv controls whether this flag is set by default or not. .It Cm -accept_rtadv Clear a flag .Cm accept_rtadv . .It Cm no_radr Set a flag to control whether routers from which the system accepts Router Advertisement messages will be added to the Default Router List or not. When the .Cm accept_rtadv flag is disabled, this flag has no effect. The .Xr sysctl 8 variable .Va net.inet6.ip6.no_radr controls whether this flag is set by default or not. .It Cm -no_radr Clear a flag .Cm no_radr . .It Cm auto_linklocal Set a flag to perform automatic link-local address configuration when the interface becomes available. The .Xr sysctl 8 variable .Va net.inet6.ip6.auto_linklocal controls whether this flag is set by default or not. .It Cm -auto_linklocal Clear a flag .Cm auto_linklocal . .It Cm defaultif Set the specified interface as the default route when there is no default router. .It Cm -defaultif Clear a flag .Cm defaultif . .It Cm ifdisabled Set a flag to disable all of IPv6 network communications on the specified interface. Note that if there are already configured IPv6 addresses on that interface, all of them are marked as .Dq tentative and DAD will be performed when this flag is cleared. .It Cm -ifdisabled Clear a flag .Cm ifdisabled . When this flag is cleared and .Cm auto_linklocal flag is enabled, automatic configuration of a link-local address is performed. .It Cm nud Set a flag to enable Neighbor Unreachability Detection. .It Cm -nud Clear a flag .Cm nud . .It Cm no_prefer_iface Set a flag to not honor rule 5 of source address selection in RFC 3484. In practice this means the address on the outgoing interface will not be preferred, effectively yielding the decision to the address selection policy table, configurable with .Xr ip6addrctl 8 . .It Cm -no_prefer_iface Clear a flag .Cm no_prefer_iface . +.It Cm no_dad +Set a flag to disable Duplicate Address Detection. +.It Cm -no_dad +Clear a flag +.Cm no_dad . +.It Cm ignoreloop +Set a flag to disable loopback detection in Enhanced Duplicate Address +Detection Algorithm. +When this flag is set, +Duplicate Address Detection will stop in a finite number of probings +even if a loopback configuration is detected. +.It Cm -ignoreloop +Clear a flag +.Cm ignoreloop . .El .Pp The following parameters are specific for IPv6 addresses. Note that the address family keyword .Dq Li inet6 is needed for them: .Bl -tag -width indent .It Cm prefer_source Set a flag to prefer address as a candidate of the source address for outgoing packets. .It Cm -prefer_source Clear a flag .Cm prefer_source . .El .Pp The following parameters are specific to cloning IEEE 802.11 wireless interfaces with the .Cm create request: .Bl -tag -width indent .It Cm wlandev Ar device Use .Ar device as the parent for the cloned device. .It Cm wlanmode Ar mode Specify the operating mode for this cloned device. .Ar mode is one of .Cm sta , .Cm ahdemo (or .Cm adhoc-demo ), .Cm ibss , (or .Cm adhoc ), .Cm ap , (or .Cm hostap ), .Cm wds , .Cm tdma , .Cm mesh , and .Cm monitor . The operating mode of a cloned interface cannot be changed. The .Cm tdma mode is actually implemented as an .Cm adhoc-demo interface with special properties. .It Cm wlanbssid Ar bssid The 802.11 mac address to use for the bssid. This must be specified at create time for a legacy .Cm wds device. .It Cm wlanaddr Ar address The local mac address. If this is not specified then a mac address will automatically be assigned to the cloned device. Typically this address is the same as the address of the parent device but if the .Cm bssid parameter is specified then the driver will craft a unique address for the device (if supported). .It Cm wdslegacy Mark a .Cm wds device as operating in ``legacy mode''. Legacy .Cm wds devices have a fixed peer relationship and do not, for example, roam if their peer stops communicating. For completeness a Dynamic WDS (DWDS) interface may marked as .Fl wdslegacy . .It Cm bssid Request a unique local mac address for the cloned device. This is only possible if the device supports multiple mac addresses. To force use of the parent's mac address use .Fl bssid . .It Cm beacons Mark the cloned interface as depending on hardware support to track received beacons. To have beacons tracked in software use .Fl beacons . For .Cm hostap mode .Fl beacons can also be used to indicate no beacons should be transmitted; this can be useful when creating a WDS configuration but .Cm wds interfaces can only be created as companions to an access point. .El .Pp The following parameters are specific to IEEE 802.11 wireless interfaces cloned with a .Cm create operation: .Bl -tag -width indent .It Cm ampdu Enable sending and receiving AMPDU frames when using 802.11n (default). The 802.11n specification states a compliant station must be capable of receiving AMPDU frames but transmission is optional. Use .Fl ampdu to disable all use of AMPDU with 802.11n. For testing and/or to work around interoperability problems one can use .Cm ampdutx and .Cm ampdurx to control use of AMPDU in one direction. .It Cm ampdudensity Ar density Set the AMPDU density parameter used when operating with 802.11n. This parameter controls the inter-packet gap for AMPDU frames. The sending device normally controls this setting but a receiving station may request wider gaps. Legal values for .Ar density are 0, .25, .5, 1, 2, 4, 8, and 16 (microseconds). A value of .Cm - is treated the same as 0. .It Cm ampdulimit Ar limit Set the limit on packet size for receiving AMPDU frames when operating with 802.11n. Legal values for .Ar limit are 8192, 16384, 32768, and 65536 but one can also specify just the unique prefix: 8, 16, 32, 64. Note the sender may limit the size of AMPDU frames to be less than the maximum specified by the receiving station. .It Cm amsdu Enable sending and receiving AMSDU frames when using 802.11n. By default AMSDU is received but not transmitted. Use .Fl amsdu to disable all use of AMSDU with 802.11n. For testing and/or to work around interoperability problems one can use .Cm amsdutx and .Cm amsdurx to control use of AMSDU in one direction. .It Cm amsdulimit Ar limit Set the limit on packet size for sending and receiving AMSDU frames when operating with 802.11n. Legal values for .Ar limit are 7935 and 3839 (bytes). Note the sender may limit the size of AMSDU frames to be less than the maximum specified by the receiving station. Note also that devices are not required to support the 7935 limit, only 3839 is required by the specification and the larger value may require more memory to be dedicated to support functionality that is rarely used. .It Cm apbridge When operating as an access point, pass packets between wireless clients directly (default). To instead let them pass up through the system and be forwarded using some other mechanism, use .Fl apbridge . Disabling the internal bridging is useful when traffic is to be processed with packet filtering. .It Cm authmode Ar mode Set the desired authentication mode in infrastructure mode. Not all adapters support all modes. The set of valid modes is .Cm none , open , shared (shared key), .Cm 8021x (IEEE 802.1x), and .Cm wpa (IEEE WPA/WPA2/802.11i). The .Cm 8021x and .Cm wpa modes are only useful when using an authentication service (a supplicant for client operation or an authenticator when operating as an access point). Modes are case insensitive. .It Cm bgscan Enable background scanning when operating as a station. Background scanning is a technique whereby a station associated to an access point will temporarily leave the channel to scan for neighboring stations. This allows a station to maintain a cache of nearby access points so that roaming between access points can be done without a lengthy scan operation. Background scanning is done only when a station is not busy and any outbound traffic will cancel a scan operation. Background scanning should never cause packets to be lost though there may be some small latency if outbound traffic interrupts a scan operation. By default background scanning is enabled if the device is capable. To disable background scanning, use .Fl bgscan . Background scanning is controlled by the .Cm bgscanidle and .Cm bgscanintvl parameters. Background scanning must be enabled for roaming; this is an artifact of the current implementation and may not be required in the future. .It Cm bgscanidle Ar idletime Set the minimum time a station must be idle (not transmitting or receiving frames) before a background scan is initiated. The .Ar idletime parameter is specified in milliseconds. By default a station must be idle at least 250 milliseconds before a background scan is initiated. The idle time may not be set to less than 100 milliseconds. .It Cm bgscanintvl Ar interval Set the interval at which background scanning is attempted. The .Ar interval parameter is specified in seconds. By default a background scan is considered every 300 seconds (5 minutes). The .Ar interval may not be set to less than 15 seconds. .It Cm bintval Ar interval Set the interval at which beacon frames are sent when operating in ad-hoc or ap mode. The .Ar interval parameter is specified in TU's (1024 usecs). By default beacon frames are transmitted every 100 TU's. .It Cm bmissthreshold Ar count Set the number of consecutive missed beacons at which the station will attempt to roam (i.e., search for a new access point). The .Ar count parameter must be in the range 1 to 255; though the upper bound may be reduced according to device capabilities. The default threshold is 7 consecutive missed beacons; but this may be overridden by the device driver. Another name for the .Cm bmissthreshold parameter is .Cm bmiss . .It Cm bssid Ar address Specify the MAC address of the access point to use when operating as a station in a BSS network. This overrides any automatic selection done by the system. To disable a previously selected access point, supply .Cm any , none , or .Cm - for the address. This option is useful when more than one access point uses the same SSID. Another name for the .Cm bssid parameter is .Cm ap . .It Cm burst Enable packet bursting. Packet bursting is a transmission technique whereby the wireless medium is acquired once to send multiple frames and the interframe spacing is reduced. This technique can significantly increase throughput by reducing transmission overhead. Packet bursting is supported by the 802.11e QoS specification and some devices that do not support QoS may still be capable. By default packet bursting is enabled if a device is capable of doing it. To disable packet bursting, use .Fl burst . .It Cm chanlist Ar channels Set the desired channels to use when scanning for access points, neighbors in an IBSS network, or looking for unoccupied channels when operating as an access point. The set of channels is specified as a comma-separated list with each element in the list representing either a single channel number or a range of the form .Dq Li a-b . Channel numbers must be in the range 1 to 255 and be permissible according to the operating characteristics of the device. .It Cm channel Ar number Set a single desired channel. Channels range from 1 to 255, but the exact selection available depends on the region your adaptor was manufactured for. Setting the channel to .Li any , or .Cm - will clear any desired channel and, if the device is marked up, force a scan for a channel to operate on. Alternatively the frequency, in megahertz, may be specified instead of the channel number. .Pp When there are several ways to use a channel the channel number/frequency may be appended with attributes to clarify. For example, if a device is capable of operating on channel 6 with 802.11n and 802.11g then one can specify that g-only use should be used by specifying ``6:g''. Similarly the channel width can be specified by appending it with ``/''; e.g., ``6/40'' specifies a 40MHz wide channel, These attributes can be combined as in: ``6:ht/40''. The full set of flags specified following a ``:'' are: .Cm a (802.11a), .Cm b (802.11b), .Cm d (Atheros Dynamic Turbo mode), .Cm g (802.11g), .Cm h or .Cm n (802.11n aka HT), .Cm s (Atheros Static Turbo mode), and .Cm t (Atheros Dynamic Turbo mode, or appended to ``st'' and ``dt''). The full set of channel widths following a '/' are: .Cm 5 (5MHz aka quarter-rate channel), .Cm 10 (10MHz aka half-rate channel), .Cm 20 (20MHz mostly for use in specifying ht20), and .Cm 40 (40MHz mostly for use in specifying ht40). In addition, a 40MHz HT channel specification may include the location of the extension channel by appending ``+'' or ``-'' for above and below, respectively; e.g., ``2437:ht/40+'' specifies 40MHz wide HT operation with the center channel at frequency 2437 and the extension channel above. .It Cm country Ar name Set the country code to use in calculating the regulatory constraints for operation. In particular the set of available channels, how the wireless device will operation on the channels, and the maximum transmit power that can be used on a channel are defined by this setting. Country/Region codes are specified as a 2-character abbreviation defined by ISO 3166 or using a longer, but possibly ambiguous, spelling; e.g., "ES" and "Spain". The set of country codes are taken from .Pa /etc/regdomain.xml and can also be viewed with the ``list countries'' request. Note that not all devices support changing the country code from a default setting; typically stored in EEPROM. See also .Cm regdomain , .Cm indoor , .Cm outdoor , and .Cm anywhere . .It Cm dfs Enable Dynamic Frequency Selection (DFS) as specified in 802.11h. DFS embodies several facilities including detection of overlapping radar signals, dynamic transmit power control, and channel selection according to a least-congested criteria. DFS support is mandatory for some 5GHz frequencies in certain locales (e.g., ETSI). By default DFS is enabled according to the regulatory definitions specified in .Pa /etc/regdomain.xml and the current country code, regdomain, and channel. Note the underlying device (and driver) must support radar detection for full DFS support to work. To be fully compliant with the local regulatory agency frequencies that require DFS should not be used unless it is fully supported. Use .Fl dfs to disable this functionality for testing. .It Cm dotd Enable support for the 802.11d specification (default). When this support is enabled in station mode, beacon frames that advertise a country code different than the currently configured country code will cause an event to be dispatched to user applications. This event can be used by the station to adopt that country code and operate according to the associated regulatory constraints. When operating as an access point with 802.11d enabled the beacon and probe response frames transmitted will advertise the current regulatory domain settings. To disable 802.11d use .Fl dotd . .It Cm doth Enable 802.11h support including spectrum management. When 802.11h is enabled beacon and probe response frames will have the SpectrumMgt bit set in the capabilities field and country and power constraint information elements will be present. 802.11h support also includes handling Channel Switch Announcements (CSA) which are a mechanism to coordinate channel changes by an access point. By default 802.11h is enabled if the device is capable. To disable 802.11h use .Fl doth . .It Cm deftxkey Ar index Set the default key to use for transmission. Typically this is only set when using WEP encryption. Note that you must set a default transmit key for the system to know which key to use in encrypting outbound traffic. The .Cm weptxkey is an alias for this request; it is provided for backwards compatibility. .It Cm dtimperiod Ar period Set the DTIM period for transmitting buffered multicast data frames when operating in ap mode. The .Ar period specifies the number of beacon intervals between DTIM and must be in the range 1 to 15. By default DTIM is 1 (i.e., DTIM occurs at each beacon). .It Cm quiet Enable the use of quiet IE. Hostap will use this to silence other stations to reduce interference for radar detection when operating on 5GHz frequency and doth support is enabled. Use .Fl quiet to disable this functionality. .It Cm quiet_period Ar period Set the QUIET .Ar period to the number of beacon intervals between the start of regularly scheduled quiet intervals defined by Quiet element. .It Cm quiet_count Ar count Set the QUIET .Ar count to the number of TBTTs until the beacon interval during which the next quiet interval shall start. A value of 1 indicates the quiet interval will start during the beacon interval starting at the next TBTT. A value 0 is reserved. .It Cm quiet_offset Ar offset Set the QUIET .Ar offset to the offset of the start of the quiet interval from the TBTT specified by the Quiet count, expressed in TUs. The value of the .Ar offset shall be less than one beacon interval. .It Cm quiet_duration Ar dur Set the QUIET .Ar dur to the duration of the Quiet interval, expressed in TUs. The value should be less than beacon interval. .It Cm dturbo Enable the use of Atheros Dynamic Turbo mode when communicating with another Dynamic Turbo-capable station. Dynamic Turbo mode is an Atheros-specific mechanism by which stations switch between normal 802.11 operation and a ``boosted'' mode in which a 40MHz wide channel is used for communication. Stations using Dynamic Turbo mode operate boosted only when the channel is free of non-dturbo stations; when a non-dturbo station is identified on the channel all stations will automatically drop back to normal operation. By default, Dynamic Turbo mode is not enabled, even if the device is capable. Note that turbo mode (dynamic or static) is only allowed on some channels depending on the regulatory constraints; use the .Cm list chan command to identify the channels where turbo mode may be used. To disable Dynamic Turbo mode use .Fl dturbo . .It Cm dwds Enable Dynamic WDS (DWDS) support. DWDS is a facility by which 4-address traffic can be carried between stations operating in infrastructure mode. A station first associates to an access point and authenticates using normal procedures (e.g., WPA). Then 4-address frames are passed to carry traffic for stations operating on either side of the wireless link. DWDS extends the normal WDS mechanism by leveraging existing security protocols and eliminating static binding. .Pp When DWDS is enabled on an access point 4-address frames received from an authorized station will generate a ``DWDS discovery'' event to user applications. This event should be used to create a WDS interface that is bound to the remote station (and usually plumbed into a bridge). Once the WDS interface is up and running 4-address traffic then logically flows through that interface. .Pp When DWDS is enabled on a station, traffic with a destination address different from the peer station are encapsulated in a 4-address frame and transmitted to the peer. All 4-address traffic uses the security information of the stations (e.g., cryptographic keys). A station is associated using 802.11n facilities may transport 4-address traffic using these same mechanisms; this depends on available resources and capabilities of the device. The DWDS implementation guards against layer 2 routing loops of multicast traffic. .It Cm ff Enable the use of Atheros Fast Frames when communicating with another Fast Frames-capable station. Fast Frames are an encapsulation technique by which two 802.3 frames are transmitted in a single 802.11 frame. This can noticeably improve throughput but requires that the receiving station understand how to decapsulate the frame. Fast frame use is negotiated using the Atheros 802.11 vendor-specific protocol extension so enabling use is safe when communicating with non-Atheros devices. By default, use of fast frames is enabled if the device is capable. To explicitly disable fast frames, use .Fl ff . .It Cm fragthreshold Ar length Set the threshold for which transmitted frames are broken into fragments. The .Ar length argument is the frame size in bytes and must be in the range 256 to 2346. Setting .Ar length to .Li 2346 , .Cm any , or .Cm - disables transmit fragmentation. Not all adapters honor the fragmentation threshold. .It Cm hidessid When operating as an access point, do not broadcast the SSID in beacon frames or respond to probe request frames unless they are directed to the ap (i.e., they include the ap's SSID). By default, the SSID is included in beacon frames and undirected probe request frames are answered. To re-enable the broadcast of the SSID etc., use .Fl hidessid . .It Cm ht Enable use of High Throughput (HT) when using 802.11n (default). The 802.11n specification includes mechanisms for operation on 20MHz and 40MHz wide channels using different signalling mechanisms than specified in 802.11b, 802.11g, and 802.11a. Stations negotiate use of these facilities, termed HT20 and HT40, when they associate. To disable all use of 802.11n use .Fl ht . To disable use of HT20 (e.g., to force only HT40 use) use .Fl ht20 . To disable use of HT40 use .Fl ht40 . .Pp HT configuration is used to ``auto promote'' operation when several choices are available. For example, if a station associates to an 11n-capable access point it controls whether the station uses legacy operation, HT20, or HT40. When an 11n-capable device is setup as an access point and Auto Channel Selection is used to locate a channel to operate on, HT configuration controls whether legacy, HT20, or HT40 operation is setup on the selected channel. If a fixed channel is specified for a station then HT configuration can be given as part of the channel specification; e.g., 6:ht/20 to setup HT20 operation on channel 6. .It Cm htcompat Enable use of compatibility support for pre-802.11n devices (default). The 802.11n protocol specification went through several incompatible iterations. Some vendors implemented 11n support to older specifications that will not interoperate with a purely 11n-compliant station. In particular the information elements included in management frames for old devices are different. When compatibility support is enabled both standard and compatible data will be provided. Stations that associate using the compatibility mechanisms are flagged in ``list sta''. To disable compatibility support use .Fl htcompat . .It Cm htprotmode Ar technique For interfaces operating in 802.11n, use the specified .Ar technique for protecting HT frames in a mixed legacy/HT network. The set of valid techniques is .Cm off , and .Cm rts (RTS/CTS, default). Technique names are case insensitive. .It Cm inact Enable inactivity processing for stations associated to an access point (default). When operating as an access point the 802.11 layer monitors the activity of each associated station. When a station is inactive for 5 minutes it will send several ``probe frames'' to see if the station is still present. If no response is received then the station is deauthenticated. Applications that prefer to handle this work can disable this facility by using .Fl inact . .It Cm indoor Set the location to use in calculating regulatory constraints. The location is also advertised in beacon and probe response frames when 802.11d is enabled with .Cm dotd . See also .Cm outdoor , .Cm anywhere , .Cm country , and .Cm regdomain . .It Cm list active Display the list of channels available for use taking into account any restrictions set with the .Cm chanlist directive. See the description of .Cm list chan for more information. .It Cm list caps Display the adaptor's capabilities, including the operating modes supported. .It Cm list chan Display the list of channels available for use. Channels are shown with their IEEE channel number, equivalent frequency, and usage modes. Channels identified as .Ql 11g are also usable in .Ql 11b mode. Channels identified as .Ql 11a Turbo may be used only for Atheros' Static Turbo mode (specified with . Cm mediaopt turbo ) . Channels marked with a .Ql * have a regulatory constraint that they be passively scanned. This means a station is not permitted to transmit on the channel until it identifies the channel is being used for 802.11 communication; typically by hearing a beacon frame from an access point operating on the channel. .Cm list freq is another way of requesting this information. By default a compacted list of channels is displayed; if the .Fl v option is specified then all channels are shown. .It Cm list countries Display the set of country codes and regulatory domains that can be used in regulatory configuration. .It Cm list mac Display the current MAC Access Control List state. Each address is prefixed with a character that indicates the current policy applied to it: .Ql + indicates the address is allowed access, .Ql - indicates the address is denied access, .Ql * indicates the address is present but the current policy open (so the ACL is not consulted). .It Cm list mesh Displays the mesh routing table, used for forwarding packets on a mesh network. .It Cm list regdomain Display the current regulatory settings including the available channels and transmit power caps. .It Cm list roam Display the parameters that govern roaming operation. .It Cm list txparam Display the parameters that govern transmit operation. .It Cm list txpower Display the transmit power caps for each channel. .It Cm list scan Display the access points and/or ad-hoc neighbors located in the vicinity. This information may be updated automatically by the adapter with a .Cm scan request or through background scanning. Depending on the capabilities of the stations the following flags can be included in the output: .Bl -tag -width 3n .It Li A Authorized. Indicates that the station is permitted to send/receive data frames. .It Li E Extended Rate Phy (ERP). Indicates that the station is operating in an 802.11g network using extended transmit rates. .It Li H High Throughput (HT). Indicates that the station is using HT transmit rates. If a `+' follows immediately after then the station associated using deprecated mechanisms supported only when .Cm htcompat is enabled. .It Li P Power Save. Indicates that the station is operating in power save mode. .It Li Q Quality of Service (QoS). Indicates that the station is using QoS encapsulation for data frame. QoS encapsulation is enabled only when WME mode is enabled. .It Li S Short Preamble. Indicates that the station is doing short preamble to optionally improve throughput performance with 802.11g and 802.11b. .It Li T Transitional Security Network (TSN). Indicates that the station associated using TSN; see also .Cm tsn below. .It Li W Wi-Fi Protected Setup (WPS). Indicates that the station associated using WPS. .El .Pp By default interesting information elements captured from the neighboring stations are displayed at the end of each row. Possible elements include: .Cm WME (station supports WME), .Cm WPA (station supports WPA), .Cm WPS (station supports WPS), .Cm RSN (station supports 802.11i/RSN), .Cm HTCAP (station supports 802.11n/HT communication), .Cm ATH (station supports Atheros protocol extensions), .Cm VEN (station supports unknown vendor-specific extensions). If the .Fl v flag is used all the information elements and their contents will be shown. Specifying the .Fl v flag also enables display of long SSIDs. The .Cm list ap command is another way of requesting this information. .It Cm list sta When operating as an access point display the stations that are currently associated. When operating in ad-hoc mode display stations identified as neighbors in the IBSS. When operating in mesh mode display stations identified as neighbors in the MBSS. When operating in station mode display the access point. Capabilities advertised by the stations are described under the .Cm scan request. Depending on the capabilities of the stations the following flags can be included in the output: .Bl -tag -width 3n .It Li A Authorized. Indicates that the station is permitted to send/receive data frames. .It Li E Extended Rate Phy (ERP). Indicates that the station is operating in an 802.11g network using extended transmit rates. .It Li H High Throughput (HT). Indicates that the station is using HT transmit rates. If a `+' follows immediately after then the station associated using deprecated mechanisms supported only when .Cm htcompat is enabled. .It Li P Power Save. Indicates that the station is operating in power save mode. .It Li Q Quality of Service (QoS). Indicates that the station is using QoS encapsulation for data frame. QoS encapsulation is enabled only when WME mode is enabled. .It Li S Short Preamble. Indicates that the station is doing short preamble to optionally improve throughput performance with 802.11g and 802.11b. .It Li T Transitional Security Network (TSN). Indicates that the station associated using TSN; see also .Cm tsn below. .It Li W Wi-Fi Protected Setup (WPS). Indicates that the station associated using WPS. .El .Pp By default information elements received from associated stations are displayed in a short form; the .Fl v flag causes this information to be displayed symbolically. .It Cm list wme Display the current channel parameters to use when operating in WME mode. If the .Fl v option is specified then both channel and BSS parameters are displayed for each AC (first channel, then BSS). When WME mode is enabled for an adaptor this information will be displayed with the regular status; this command is mostly useful for examining parameters when WME mode is disabled. See the description of the .Cm wme directive for information on the various parameters. .It Cm maxretry Ar count Set the maximum number of tries to use in sending unicast frames. The default setting is 6 but drivers may override this with a value they choose. .It Cm mcastrate Ar rate Set the rate for transmitting multicast/broadcast frames. Rates are specified as megabits/second in decimal; e.g.,\& 5.5 for 5.5 Mb/s. This rate should be valid for the current operating conditions; if an invalid rate is specified drivers are free to chose an appropriate rate. .It Cm mgtrate Ar rate Set the rate for transmitting management and/or control frames. Rates are specified as megabits/second in decimal; e.g.,\& 5.5 for 5.5 Mb/s. .It Cm outdoor Set the location to use in calculating regulatory constraints. The location is also advertised in beacon and probe response frames when 802.11d is enabled with .Cm dotd . See also .Cm anywhere , .Cm country , .Cm indoor , and .Cm regdomain . .It Cm powersave Enable powersave operation. When operating as a client, the station will conserve power by periodically turning off the radio and listening for messages from the access point telling it there are packets waiting. The station must then retrieve the packets. Not all devices support power save operation as a client. The 802.11 specification requires that all access points support power save but some drivers do not. Use .Fl powersave to disable powersave operation when operating as a client. .It Cm powersavesleep Ar sleep Set the desired max powersave sleep time in TU's (1024 usecs). By default the max powersave sleep time is 100 TU's. .It Cm protmode Ar technique For interfaces operating in 802.11g, use the specified .Ar technique for protecting OFDM frames in a mixed 11b/11g network. The set of valid techniques is .Cm off , cts (CTS to self), and .Cm rtscts (RTS/CTS). Technique names are case insensitive. Not all devices support .Cm cts as a protection technique. .It Cm pureg When operating as an access point in 802.11g mode allow only 11g-capable stations to associate (11b-only stations are not permitted to associate). To allow both 11g and 11b-only stations to associate, use .Fl pureg . .It Cm puren When operating as an access point in 802.11n mode allow only HT-capable stations to associate (legacy stations are not permitted to associate). To allow both HT and legacy stations to associate, use .Fl puren . .It Cm regdomain Ar sku Set the regulatory domain to use in calculating the regulatory constraints for operation. In particular the set of available channels, how the wireless device will operation on the channels, and the maximum transmit power that can be used on a channel are defined by this setting. Regdomain codes (SKU's) are taken from .Pa /etc/regdomain.xml and can also be viewed with the ``list countries'' request. Note that not all devices support changing the regdomain from a default setting; typically stored in EEPROM. See also .Cm country , .Cm indoor , .Cm outdoor , and .Cm anywhere . .It Cm rifs Enable use of Reduced InterFrame Spacing (RIFS) when operating in 802.11n on an HT channel. Note that RIFS must be supported by both the station and access point for it to be used. To disable RIFS use .Fl rifs . .It Cm roam:rate Ar rate Set the threshold for controlling roaming when operating in a BSS. The .Ar rate parameter specifies the transmit rate in megabits at which roaming should be considered. If the current transmit rate drops below this setting and background scanning is enabled, then the system will check if a more desirable access point is available and switch over to it. The current scan cache contents are used if they are considered valid according to the .Cm scanvalid parameter; otherwise a background scan operation is triggered before any selection occurs. Each channel type has a separate rate threshold; the default values are: 12 Mb/s (11a), 2 Mb/s (11b), 2 Mb/s (11g), MCS 1 (11na, 11ng). .It Cm roam:rssi Ar rssi Set the threshold for controlling roaming when operating in a BSS. The .Ar rssi parameter specifies the receive signal strength in dBm units at which roaming should be considered. If the current rssi drops below this setting and background scanning is enabled, then the system will check if a more desirable access point is available and switch over to it. The current scan cache contents are used if they are considered valid according to the .Cm scanvalid parameter; otherwise a background scan operation is triggered before any selection occurs. Each channel type has a separate rssi threshold; the default values are all 7 dBm. .It Cm roaming Ar mode When operating as a station, control how the system will behave when communication with the current access point is broken. The .Ar mode argument may be one of .Cm device (leave it to the hardware device to decide), .Cm auto (handle either in the device or the operating system\[em]as appropriate), .Cm manual (do nothing until explicitly instructed). By default, the device is left to handle this if it is capable; otherwise, the operating system will automatically attempt to reestablish communication. Manual mode is used by applications such as .Xr wpa_supplicant 8 that want to control the selection of an access point. .It Cm rtsthreshold Ar length Set the threshold for which transmitted frames are preceded by transmission of an RTS control frame. The .Ar length argument is the frame size in bytes and must be in the range 1 to 2346. Setting .Ar length to .Li 2346 , .Cm any , or .Cm - disables transmission of RTS frames. Not all adapters support setting the RTS threshold. .It Cm scan Initiate a scan of neighboring stations, wait for it to complete, and display all stations found. Only the super-user can initiate a scan. See .Cm list scan for information on the display. By default a background scan is done; otherwise a foreground scan is done and the station may roam to a different access point. The .Cm list scan request can be used to show recent scan results without initiating a new scan. .It Cm scanvalid Ar threshold Set the maximum time the scan cache contents are considered valid; i.e., will be used without first triggering a scan operation to refresh the data. The .Ar threshold parameter is specified in seconds and defaults to 60 seconds. The minimum setting for .Ar threshold is 10 seconds. One should take care setting this threshold; if it is set too low then attempts to roam to another access point may trigger unnecessary background scan operations. .It Cm shortgi Enable use of Short Guard Interval when operating in 802.11n on an HT channel. NB: this currently enables Short GI on both HT40 and HT20 channels. To disable Short GI use .Fl shortgi . .It Cm smps Enable use of Static Spatial Multiplexing Power Save (SMPS) when operating in 802.11n. A station operating with Static SMPS maintains only a single receive chain active (this can significantly reduce power consumption). To disable SMPS use .Fl smps . .It Cm smpsdyn Enable use of Dynamic Spatial Multiplexing Power Save (SMPS) when operating in 802.11n. A station operating with Dynamic SMPS maintains only a single receive chain active but switches to multiple receive chains when it receives an RTS frame (this can significantly reduce power consumption). Note that stations cannot distinguish between RTS/CTS intended to enable multiple receive chains and those used for other purposes. To disable SMPS use .Fl smps . .It Cm ssid Ar ssid Set the desired Service Set Identifier (aka network name). The SSID is a string up to 32 characters in length and may be specified as either a normal string or in hexadecimal when preceded by .Ql 0x . Additionally, the SSID may be cleared by setting it to .Ql - . .It Cm tdmaslot Ar slot When operating with TDMA, use the specified .Ar slot configuration. The .Ar slot is a number between 0 and the maximum number of slots in the BSS. Note that a station configured as slot 0 is a master and will broadcast beacon frames advertising the BSS; stations configured to use other slots will always scan to locate a master before they ever transmit. By default .Cm tdmaslot is set to 1. .It Cm tdmaslotcnt Ar cnt When operating with TDMA, setup a BSS with .Ar cnt slots. The slot count may be at most 8. The current implementation is only tested with two stations (i.e., point to point applications). This setting is only meaningful when a station is configured as slot 0; other stations adopt this setting from the BSS they join. By default .Cm tdmaslotcnt is set to 2. .It Cm tdmaslotlen Ar len When operating with TDMA, setup a BSS such that each station has a slot .Ar len microseconds long. The slot length must be at least 150 microseconds (1/8 TU) and no more than 65 milliseconds. Note that setting too small a slot length may result in poor channel bandwidth utilization due to factors such as timer granularity and guard time. This setting is only meaningful when a station is configured as slot 0; other stations adopt this setting from the BSS they join. By default .Cm tdmaslotlen is set to 10 milliseconds. .It Cm tdmabintval Ar intval When operating with TDMA, setup a BSS such that beacons are transmitted every .Ar intval superframes to synchronize the TDMA slot timing. A superframe is defined as the number of slots times the slot length; e.g., a BSS with two slots of 10 milliseconds has a 20 millisecond superframe. The beacon interval may not be zero. A lower setting of .Cm tdmabintval causes the timers to be resynchronized more often; this can be help if significant timer drift is observed. By default .Cm tdmabintval is set to 5. .It Cm tsn When operating as an access point with WPA/802.11i allow legacy stations to associate using static key WEP and open authentication. To disallow legacy station use of WEP, use .Fl tsn . .It Cm txpower Ar power Set the power used to transmit frames. The .Ar power argument is specified in .5 dBm units. Out of range values are truncated. Typically only a few discreet power settings are available and the driver will use the setting closest to the specified value. Not all adapters support changing the transmit power. .It Cm ucastrate Ar rate Set a fixed rate for transmitting unicast frames. Rates are specified as megabits/second in decimal; e.g.,\& 5.5 for 5.5 Mb/s. This rate should be valid for the current operating conditions; if an invalid rate is specified drivers are free to chose an appropriate rate. .It Cm wepmode Ar mode Set the desired WEP mode. Not all adapters support all modes. The set of valid modes is .Cm off , on , and .Cm mixed . The .Cm mixed mode explicitly tells the adaptor to allow association with access points which allow both encrypted and unencrypted traffic. On these adapters, .Cm on means that the access point must only allow encrypted connections. On other adapters, .Cm on is generally another name for .Cm mixed . Modes are case insensitive. .It Cm weptxkey Ar index Set the WEP key to be used for transmission. This is the same as setting the default transmission key with .Cm deftxkey . .It Cm wepkey Ar key Ns | Ns Ar index : Ns Ar key Set the selected WEP key. If an .Ar index is not given, key 1 is set. A WEP key will be either 5 or 13 characters (40 or 104 bits) depending on the local network and the capabilities of the adaptor. It may be specified either as a plain string or as a string of hexadecimal digits preceded by .Ql 0x . For maximum portability, hex keys are recommended; the mapping of text keys to WEP encryption is usually driver-specific. In particular, the .Tn Windows drivers do this mapping differently to .Fx . A key may be cleared by setting it to .Ql - . If WEP is supported then there are at least four keys. Some adapters support more than four keys. If that is the case, then the first four keys (1-4) will be the standard temporary keys and any others will be adaptor specific keys such as permanent keys stored in NVRAM. .Pp Note that you must set a default transmit key with .Cm deftxkey for the system to know which key to use in encrypting outbound traffic. .It Cm wme Enable Wireless Multimedia Extensions (WME) support, if available, for the specified interface. WME is a subset of the IEEE 802.11e standard to support the efficient communication of realtime and multimedia data. To disable WME support, use .Fl wme . Another name for this parameter is .Cm wmm . .Pp The following parameters are meaningful only when WME support is in use. Parameters are specified per-AC (Access Category) and split into those that are used by a station when acting as an access point and those for client stations in the BSS. The latter are received from the access point and may not be changed (at the station). The following Access Categories are recognized: .Pp .Bl -tag -width ".Cm AC_BK" -compact .It Cm AC_BE (or .Cm BE ) best effort delivery, .It Cm AC_BK (or .Cm BK ) background traffic, .It Cm AC_VI (or .Cm VI ) video traffic, .It Cm AC_VO (or .Cm VO ) voice traffic. .El .Pp AC parameters are case-insensitive. Traffic classification is done in the operating system using the vlan priority associated with data frames or the ToS (Type of Service) indication in IP-encapsulated frames. If neither information is present, traffic is assigned to the Best Effort (BE) category. .Bl -tag -width indent .It Cm ack Ar ac Set the ACK policy for QoS transmissions by the local station; this controls whether or not data frames transmitted by a station require an ACK response from the receiving station. To disable waiting for an ACK use .Fl ack . This parameter is applied only to the local station. .It Cm acm Ar ac Enable the Admission Control Mandatory (ACM) mechanism for transmissions by the local station. To disable the ACM use .Fl acm . On stations in a BSS this parameter is read-only and indicates the setting received from the access point. NB: ACM is not supported right now. .It Cm aifs Ar ac Ar count Set the Arbitration Inter Frame Spacing (AIFS) channel access parameter to use for transmissions by the local station. On stations in a BSS this parameter is read-only and indicates the setting received from the access point. .It Cm cwmin Ar ac Ar count Set the CWmin channel access parameter to use for transmissions by the local station. On stations in a BSS this parameter is read-only and indicates the setting received from the access point. .It Cm cwmax Ar ac Ar count Set the CWmax channel access parameter to use for transmissions by the local station. On stations in a BSS this parameter is read-only and indicates the setting received from the access point. .It Cm txoplimit Ar ac Ar limit Set the Transmission Opportunity Limit channel access parameter to use for transmissions by the local station. This parameter defines an interval of time when a WME station has the right to initiate transmissions onto the wireless medium. On stations in a BSS this parameter is read-only and indicates the setting received from the access point. .It Cm bss:aifs Ar ac Ar count Set the AIFS channel access parameter to send to stations in a BSS. This parameter is meaningful only when operating in ap mode. .It Cm bss:cwmin Ar ac Ar count Set the CWmin channel access parameter to send to stations in a BSS. This parameter is meaningful only when operating in ap mode. .It Cm bss:cwmax Ar ac Ar count Set the CWmax channel access parameter to send to stations in a BSS. This parameter is meaningful only when operating in ap mode. .It Cm bss:txoplimit Ar ac Ar limit Set the TxOpLimit channel access parameter to send to stations in a BSS. This parameter is meaningful only when operating in ap mode. .El .It Cm wps Enable Wireless Privacy Subscriber support. Note that WPS support requires a WPS-capable supplicant. To disable this function use .Fl wps . .El .Pp The following parameters support an optional access control list feature available with some adapters when operating in ap mode; see .Xr wlan_acl 4 . This facility allows an access point to accept/deny association requests based on the MAC address of the station. Note that this feature does not significantly enhance security as MAC address spoofing is easy to do. .Bl -tag -width indent .It Cm mac:add Ar address Add the specified MAC address to the database. Depending on the policy setting association requests from the specified station will be allowed or denied. .It Cm mac:allow Set the ACL policy to permit association only by stations registered in the database. .It Cm mac:del Ar address Delete the specified MAC address from the database. .It Cm mac:deny Set the ACL policy to deny association only by stations registered in the database. .It Cm mac:kick Ar address Force the specified station to be deauthenticated. This typically is done to block a station after updating the address database. .It Cm mac:open Set the ACL policy to allow all stations to associate. .It Cm mac:flush Delete all entries in the database. .It Cm mac:radius Set the ACL policy to permit association only by stations approved by a RADIUS server. Note that this feature requires the .Xr hostapd 8 program be configured to do the right thing as it handles the RADIUS processing (and marks stations as authorized). .El .Pp The following parameters are related to a wireless interface operating in mesh mode: .Bl -tag -width indent .It Cm meshid Ar meshid Set the desired Mesh Identifier. The Mesh ID is a string up to 32 characters in length. A mesh interface must have a Mesh Identifier specified to reach an operational state. .It Cm meshttl Ar ttl Set the desired ``time to live'' for mesh forwarded packets; this is the number of hops a packet may be forwarded before it is discarded. The default setting for .Cm meshttl is 31. .It Cm meshpeering Enable or disable peering with neighbor mesh stations. Stations must peer before any data packets can be exchanged. By default .Cm meshpeering is enabled. .It Cm meshforward Enable or disable forwarding packets by a mesh interface. By default .Cm meshforward is enabled. .It Cm meshgate This attribute specifies whether or not the mesh STA activates mesh gate announcements. By default .Cm meshgate is disabled. .It Cm meshmetric Ar protocol Set the specified .Ar protocol as the link metric protocol used on a mesh network. The default protocol is called .Ar AIRTIME . The mesh interface will restart after changing this setting. .It Cm meshpath Ar protocol Set the specified .Ar protocol as the path selection protocol used on a mesh network. The only available protocol at the moment is called .Ar HWMP (Hybrid Wireless Mesh Protocol). The mesh interface will restart after changing this setting. .It Cm hwmprootmode Ar mode Stations on a mesh network can operate as ``root nodes.'' Root nodes try to find paths to all mesh nodes and advertise themselves regularly. When there is a root mesh node on a network, other mesh nodes can setup paths between themselves faster because they can use the root node to find the destination. This path may not be the best, but on-demand routing will eventually find the best path. The following modes are recognized: .Pp .Bl -tag -width ".Cm PROACTIVE" -compact .It Cm DISABLED Disable root mode. .It Cm NORMAL Send broadcast path requests every two seconds. Nodes on the mesh without a path to this root mesh station with try to discover a path to us. .It Cm PROACTIVE Send broadcast path requests every two seconds and every node must reply with a path reply even if it already has a path to this root mesh station. .It Cm RANN Send broadcast root announcement (RANN) frames. Nodes on the mesh without a path to this root mesh station with try to discover a path to us. .El By default .Cm hwmprootmode is set to .Ar DISABLED . .It Cm hwmpmaxhops Ar cnt Set the maximum number of hops allowed in an HMWP path to .Ar cnt . The default setting for .Cm hwmpmaxhops is 31. .El .Pp The following parameters are for compatibility with other systems: .Bl -tag -width indent .It Cm nwid Ar ssid Another name for the .Cm ssid parameter. Included for .Nx compatibility. .It Cm stationname Ar name Set the name of this station. The station name is not part of the IEEE 802.11 protocol though some interfaces support it. As such it only seems to be meaningful to identical or virtually identical equipment. Setting the station name is identical in syntax to setting the SSID. One can also use .Cm station for .Bsx compatibility. .It Cm wep Another way of saying .Cm wepmode on . Included for .Bsx compatibility. .It Fl wep Another way of saying .Cm wepmode off . Included for .Bsx compatibility. .It Cm nwkey key Another way of saying: .Dq Li "wepmode on weptxkey 1 wepkey 1:key wepkey 2:- wepkey 3:- wepkey 4:-" . Included for .Nx compatibility. .It Cm nwkey Xo .Sm off .Ar n : k1 , k2 , k3 , k4 .Sm on .Xc Another way of saying .Dq Li "wepmode on weptxkey n wepkey 1:k1 wepkey 2:k2 wepkey 3:k3 wepkey 4:k4" . Included for .Nx compatibility. .It Fl nwkey Another way of saying .Cm wepmode off . Included for .Nx compatibility. .El .Pp The following parameters are specific to bridge interfaces: .Bl -tag -width indent .It Cm addm Ar interface Add the interface named by .Ar interface as a member of the bridge. The interface is put into promiscuous mode so that it can receive every packet sent on the network. .It Cm deletem Ar interface Remove the interface named by .Ar interface from the bridge. Promiscuous mode is disabled on the interface when it is removed from the bridge. .It Cm maxaddr Ar size Set the size of the bridge address cache to .Ar size . The default is 2000 entries. .It Cm timeout Ar seconds Set the timeout of address cache entries to .Ar seconds seconds. If .Ar seconds is zero, then address cache entries will not be expired. The default is 1200 seconds. .It Cm addr Display the addresses that have been learned by the bridge. .It Cm static Ar interface-name Ar address Add a static entry into the address cache pointing to .Ar interface-name . Static entries are never aged out of the cache or re-placed, even if the address is seen on a different interface. .It Cm deladdr Ar address Delete .Ar address from the address cache. .It Cm flush Delete all dynamically-learned addresses from the address cache. .It Cm flushall Delete all addresses, including static addresses, from the address cache. .It Cm discover Ar interface Mark an interface as a .Dq discovering interface. When the bridge has no address cache entry (either dynamic or static) for the destination address of a packet, the bridge will forward the packet to all member interfaces marked as .Dq discovering . This is the default for all interfaces added to a bridge. .It Cm -discover Ar interface Clear the .Dq discovering attribute on a member interface. For packets without the .Dq discovering attribute, the only packets forwarded on the interface are broadcast or multicast packets and packets for which the destination address is known to be on the interface's segment. .It Cm learn Ar interface Mark an interface as a .Dq learning interface. When a packet arrives on such an interface, the source address of the packet is entered into the address cache as being a destination address on the interface's segment. This is the default for all interfaces added to a bridge. .It Cm -learn Ar interface Clear the .Dq learning attribute on a member interface. .It Cm sticky Ar interface Mark an interface as a .Dq sticky interface. Dynamically learned address entries are treated at static once entered into the cache. Sticky entries are never aged out of the cache or replaced, even if the address is seen on a different interface. .It Cm -sticky Ar interface Clear the .Dq sticky attribute on a member interface. .It Cm private Ar interface Mark an interface as a .Dq private interface. A private interface does not forward any traffic to any other port that is also a private interface. .It Cm -private Ar interface Clear the .Dq private attribute on a member interface. .It Cm span Ar interface Add the interface named by .Ar interface as a span port on the bridge. Span ports transmit a copy of every frame received by the bridge. This is most useful for snooping a bridged network passively on another host connected to one of the span ports of the bridge. .It Cm -span Ar interface Delete the interface named by .Ar interface from the list of span ports of the bridge. .It Cm stp Ar interface Enable Spanning Tree protocol on .Ar interface . The .Xr if_bridge 4 driver has support for the IEEE 802.1D Spanning Tree protocol (STP). Spanning Tree is used to detect and remove loops in a network topology. .It Cm -stp Ar interface Disable Spanning Tree protocol on .Ar interface . This is the default for all interfaces added to a bridge. .It Cm edge Ar interface Set .Ar interface as an edge port. An edge port connects directly to end stations cannot create bridging loops in the network, this allows it to transition straight to forwarding. .It Cm -edge Ar interface Disable edge status on .Ar interface . .It Cm autoedge Ar interface Allow .Ar interface to automatically detect edge status. This is the default for all interfaces added to a bridge. .It Cm -autoedge Ar interface Disable automatic edge status on .Ar interface . .It Cm ptp Ar interface Set the .Ar interface as a point to point link. This is required for straight transitions to forwarding and should be enabled on a direct link to another RSTP capable switch. .It Cm -ptp Ar interface Disable point to point link status on .Ar interface . This should be disabled for a half duplex link and for an interface connected to a shared network segment, like a hub or a wireless network. .It Cm autoptp Ar interface Automatically detect the point to point status on .Ar interface by checking the full duplex link status. This is the default for interfaces added to the bridge. .It Cm -autoptp Ar interface Disable automatic point to point link detection on .Ar interface . .It Cm maxage Ar seconds Set the time that a Spanning Tree protocol configuration is valid. The default is 20 seconds. The minimum is 6 seconds and the maximum is 40 seconds. .It Cm fwddelay Ar seconds Set the time that must pass before an interface begins forwarding packets when Spanning Tree is enabled. The default is 15 seconds. The minimum is 4 seconds and the maximum is 30 seconds. .It Cm hellotime Ar seconds Set the time between broadcasting of Spanning Tree protocol configuration messages. The hello time may only be changed when operating in legacy stp mode. The default is 2 seconds. The minimum is 1 second and the maximum is 2 seconds. .It Cm priority Ar value Set the bridge priority for Spanning Tree. The default is 32768. The minimum is 0 and the maximum is 61440. .It Cm proto Ar value Set the Spanning Tree protocol. The default is rstp. The available options are stp and rstp. .It Cm holdcnt Ar value Set the transmit hold count for Spanning Tree. This is the number of packets transmitted before being rate limited. The default is 6. The minimum is 1 and the maximum is 10. .It Cm ifpriority Ar interface Ar value Set the Spanning Tree priority of .Ar interface to .Ar value . The default is 128. The minimum is 0 and the maximum is 240. .It Cm ifpathcost Ar interface Ar value Set the Spanning Tree path cost of .Ar interface to .Ar value . The default is calculated from the link speed. To change a previously selected path cost back to automatic, set the cost to 0. The minimum is 1 and the maximum is 200000000. .It Cm ifmaxaddr Ar interface Ar size Set the maximum number of hosts allowed from an interface, packets with unknown source addresses are dropped until an existing host cache entry expires or is removed. Set to 0 to disable. .El .Pp The following parameters are specific to lagg interfaces: .Bl -tag -width indent .It Cm laggport Ar interface Add the interface named by .Ar interface as a port of the aggregation interface. .It Cm -laggport Ar interface Remove the interface named by .Ar interface from the aggregation interface. .It Cm laggproto Ar proto Set the aggregation protocol. The default is .Li failover . The available options are .Li failover , .Li lacp , .Li loadbalance , .Li roundrobin , .Li broadcast and .Li none . .It Cm lagghash Ar option Ns Oo , Ns Ar option Oc Set the packet layers to hash for aggregation protocols which load balance. The default is .Dq l2,l3,l4 . The options can be combined using commas. .Pp .Bl -tag -width ".Cm l2" -compact .It Cm l2 src/dst mac address and optional vlan number. .It Cm l3 src/dst address for IPv4 or IPv6. .It Cm l4 src/dst port for TCP/UDP/SCTP. .El .It Cm use_flowid Enable local hash computation for RSS hash on the interface. The .Li loadbalance and .Li lacp modes will use the RSS hash from the network card if available to avoid computing one, this may give poor traffic distribution if the hash is invalid or uses less of the protocol header information. .Cm use_flowid disables use of RSS hash from the network card. The default value can be set via the .Va net.link.lagg.default_use_flowid .Xr sysctl 8 variable. .Li 0 means .Dq disabled and .Li 1 means .Dq enabled . .It Cm -use_flowid Disable local hash computation for RSS hash on the interface. .It Cm flowid_shift Ar number Set a shift parameter for RSS local hash computation. Hash is calculated by using flowid bits in a packet header mbuf which are shifted by the number of this parameter. .El .Pp The following parameters are specific to IP tunnel interfaces, .Xr gif 4 : .Bl -tag -width indent .It Cm tunnel Ar src_addr dest_addr Configure the physical source and destination address for IP tunnel interfaces. The arguments .Ar src_addr and .Ar dest_addr are interpreted as the outer source/destination for the encapsulating IPv4/IPv6 header. .It Fl tunnel Unconfigure the physical source and destination address for IP tunnel interfaces previously configured with .Cm tunnel . .It Cm deletetunnel Another name for the .Fl tunnel parameter. .It Cm accept_rev_ethip_ver Set a flag to accept both correct EtherIP packets and ones with reversed version field. Enabled by default. This is for backward compatibility with .Fx 6.1 , 6.2, 6.3, 7.0, and 7.1. .It Cm -accept_rev_ethip_ver Clear a flag .Cm accept_rev_ethip_ver . .It Cm send_rev_ethip_ver Set a flag to send EtherIP packets with reversed version field intentionally. Disabled by default. This is for backward compatibility with .Fx 6.1 , 6.2, 6.3, 7.0, and 7.1. .It Cm -send_rev_ethip_ver Clear a flag .Cm send_rev_ethip_ver . .El .Pp The following parameters are specific to GRE tunnel interfaces, .Xr gre 4 : .Bl -tag -width indent .It Cm grekey Ar key Configure the GRE key to be used for outgoing packets. Note that .Xr gre 4 will always accept GRE packets with invalid or absent keys. This command will result in a four byte MTU reduction on the interface. .El .Pp The following parameters are specific to .Xr pfsync 4 interfaces: .Bl -tag -width indent .It Cm syncdev Ar iface Use the specified interface to send and receive pfsync state synchronisation messages. .It Fl syncdev Stop sending pfsync state synchronisation messages over the network. .It Cm syncpeer Ar peer_address Make the pfsync link point-to-point rather than using multicast to broadcast the state synchronisation messages. The peer_address is the IP address of the other host taking part in the pfsync cluster. .It Fl syncpeer Broadcast the packets using multicast. .It Cm maxupd Ar n Set the maximum number of updates for a single state which can be collapsed into one. This is an 8-bit number; the default value is 128. .It Cm defer Defer transmission of the first packet in a state until a peer has acknowledged that the associated state has been inserted. .It Fl defer Do not defer the first packet in a state. This is the default. .El .Pp The following parameters are specific to .Xr vlan 4 interfaces: .Bl -tag -width indent .It Cm vlan Ar vlan_tag Set the VLAN tag value to .Ar vlan_tag . This value is a 12-bit VLAN Identifier (VID) which is used to create an 802.1Q VLAN header for packets sent from the .Xr vlan 4 interface. Note that .Cm vlan and .Cm vlandev must both be set at the same time. .It Cm vlandev Ar iface Associate the physical interface .Ar iface with a .Xr vlan 4 interface. Packets transmitted through the .Xr vlan 4 interface will be diverted to the specified physical interface .Ar iface with 802.1Q VLAN encapsulation. Packets with 802.1Q encapsulation received by the parent interface with the correct VLAN Identifier will be diverted to the associated .Xr vlan 4 pseudo-interface. The .Xr vlan 4 interface is assigned a copy of the parent interface's flags and the parent's Ethernet address. The .Cm vlandev and .Cm vlan must both be set at the same time. If the .Xr vlan 4 interface already has a physical interface associated with it, this command will fail. To change the association to another physical interface, the existing association must be cleared first. .Pp Note: if the hardware tagging capability is set on the parent interface, the .Xr vlan 4 pseudo interface's behavior changes: the .Xr vlan 4 interface recognizes that the parent interface supports insertion and extraction of VLAN tags on its own (usually in firmware) and that it should pass packets to and from the parent unaltered. .It Fl vlandev Op Ar iface If the driver is a .Xr vlan 4 pseudo device, disassociate the parent interface from it. This breaks the link between the .Xr vlan 4 interface and its parent, clears its VLAN Identifier, flags and its link address and shuts the interface down. The .Ar iface argument is useless and hence deprecated. .El .Pp The following parameters are used to configure .Xr vxlan 4 interfaces. .Bl -tag -width indent .It Cm vxlanid Ar identifier This value is a 24-bit VXLAN Network Identifier (VNI) that identifies the virtual network segment membership of the interface. .It Cm vxlanlocal Ar address The source address used in the encapsulating IPv4/IPv6 header. The address should already be assigned to an existing interface. When the interface is configured in unicast mode, the listening socket is bound to this address. .It Cm vxlanremote Ar address The interface can be configured in a unicast, or point-to-point, mode to create a tunnel between two hosts. This is the IP address of the remote end of the tunnel. .It Cm vxlangroup Ar address The interface can be configured in a multicast mode to create a virtual network of hosts. This is the IP multicast group address the interface will join. .It Cm vxlanlocalport Ar port The port number the interface will listen on. The default port number is 4789. .It Cm vxlanremoteport Ar port The destination port number used in the encapsulating IPv4/IPv6 header. The remote host should be listening on this port. The default port number is 4789. Note some other implementations, such as Linux, do not default to the IANA assigned port, but instead listen on port 8472. .It Cm vxlanportrange Ar low high The range of source ports used in the encapsulating IPv4/IPv6 header. The port selected within the range is based on a hash of the inner frame. A range is useful to provide entropy within the outer IP header for more effective load balancing. The default range is between the .Xr sysctl 8 variables .Va net.inet.ip.portrange.first and .Va net.inet.ip.portrange.last .It Cm vxlantimeout Ar timeout The maximum time, in seconds, before an entry in the forwarding table is pruned. The default is 1200 seconds (20 minutes). .It Cm vxlanmaxaddr Ar max The maximum number of entries in the forwarding table. The default is 2000. .It Cm vxlandev Ar dev When the interface is configured in multicast mode, the .Cm dev interface is used to transmit IP multicast packets. .It Cm vxlanttl Ar ttl The TTL used in the encapsulating IPv4/IPv6 header. The default is 64. .It Cm vxlanlearn The source IP address and inner source Ethernet MAC address of received packets are used to dynamically populate the forwarding table. When in multicast mode, an entry in the forwarding table allows the interface to send the frame directly to the remote host instead of broadcasting the frame to the multicast group. This is the default. .It Fl vxlanlearn The forwarding table is not populated by recevied packets. .It Cm vxlanflush Delete all dynamically-learned addresses from the forwarding table. .It Cm vxlanflushall Delete all addresses, including static addresses, from the forwarding table. .El .Pp The following parameters are used to configure .Xr carp 4 protocol on an interface: .Bl -tag -width indent .It Cm vhid Ar n Set the virtual host ID. This is a required setting to initiate .Xr carp 4 . If the virtual host ID does not exist yet, it is created and attached to the interface, otherwise configuration of an existing vhid is adjusted. If the .Cm vhid keyword is supplied along with an .Dq inet6 or .Dq inet address, then this address is configured to be run under control of the specified vhid. Whenever a last address that refers to a particular vhid is removed from an interface, the vhid is automatically removed from interface and destroyed. Any other configuration parameters for the .Xr carp 4 protocol should be supplied along with the .Cm vhid keyword. Acceptable values for vhid are 1 to 255. .It Cm advbase Ar seconds Specifies the base of the advertisement interval in seconds. The acceptable values are 1 to 255. The default value is 1. .It Cm advskew Ar interval Specifies the skew to add to the base advertisement interval to make one host advertise slower than another host. It is specified in 1/256 of seconds. The acceptable values are 1 to 254. The default value is 0. .It Cm pass Ar phrase Set the authentication key to .Ar phrase . .It Cm state Ar MASTER|BACKUP Forcibly change state of a given vhid. .El .Pp The .Nm utility displays the current configuration for a network interface when no optional parameters are supplied. If a protocol family is specified, .Nm will report only the details specific to that protocol family. .Pp If the .Fl m flag is passed before an interface name, .Nm will display the capability list and all of the supported media for the specified interface. If .Fl L flag is supplied, address lifetime is displayed for IPv6 addresses, as time offset string. .Pp Optionally, the .Fl a flag may be used instead of an interface name. This flag instructs .Nm to display information about all interfaces in the system. The .Fl d flag limits this to interfaces that are down, and .Fl u limits this to interfaces that are up. When no arguments are given, .Fl a is implied. .Pp The .Fl l flag may be used to list all available interfaces on the system, with no other additional information. If an .Ar address_family is specified, only interfaces of that type will be listed. .Fl l Dq ether will list only Ethernet adapters, excluding the loopback interface. Use of this flag is mutually exclusive with all other flags and commands, except for .Fl d (only list interfaces that are down) and .Fl u (only list interfaces that are up). .Pp The .Fl v flag may be used to get more verbose status for an interface. .Pp The .Fl C flag may be used to list all of the interface cloners available on the system, with no additional information. Use of this flag is mutually exclusive with all other flags and commands. .Pp The .Fl k flag causes keying information for the interface, if available, to be printed. For example, the values of 802.11 WEP keys and .Xr carp 4 passphrases will be printed, if accessible to the current user. This information is not printed by default, as it may be considered sensitive. .Pp If the network interface driver is not present in the kernel then .Nm will attempt to load it. The .Fl n flag disables this behavior. .Pp Only the super-user may modify the configuration of a network interface. .Sh EXAMPLES Assign the IPv4 address .Li 192.0.2.10 , with a network mask of .Li 255.255.255.0 , to the interface .Li fxp0 : .Dl # ifconfig fxp0 inet 192.0.2.10 netmask 255.255.255.0 .Pp Add the IPv4 address .Li 192.0.2.45 , with the CIDR network prefix .Li /28 , to the interface .Li ed0 , using .Cm add as a synonym for the canonical form of the option .Cm alias : .Dl # ifconfig ed0 inet 192.0.2.45/28 add .Pp Remove the IPv4 address .Li 192.0.2.45 from the interface .Li ed0 : .Dl # ifconfig ed0 inet 192.0.2.45 -alias .Pp Enable IPv6 functionality of the interface: .Dl # ifconfig em0 inet6 -ifdisabled .Pp Add the IPv6 address .Li 2001:DB8:DBDB::123/48 to the interface .Li em0 : .Dl # ifconfig em0 inet6 2001:db8:bdbd::123 prefixlen 48 alias Note that lower case hexadecimal IPv6 addresses are acceptable. .Pp Remove the IPv6 address added in the above example, using the .Li / character as shorthand for the network prefix, and using .Cm delete as a synonym for the canonical form of the option .Fl alias : .Dl # ifconfig em0 inet6 2001:db8:bdbd::123/48 delete .Pp Configure a single CARP redundant address on igb0, and then switch it to be master: .Dl # ifconfig igb0 vhid 1 10.0.0.1/24 pass foobar up .Dl # ifconfig igb0 vhid 1 state master .Pp Configure the interface .Li xl0 , to use 100baseTX, full duplex Ethernet media options: .Dl # ifconfig xl0 media 100baseTX mediaopt full-duplex .Pp Label the em0 interface as an uplink: .Dl # ifconfig em0 description \&"Uplink to Gigabit Switch 2\&" .Pp Create the software network interface .Li gif1 : .Dl # ifconfig gif1 create .Pp Destroy the software network interface .Li gif1 : .Dl # ifconfig gif1 destroy .Pp Display available wireless networks using .Li wlan0 : .Dl # ifconfig wlan0 list scan .Sh DIAGNOSTICS Messages indicating the specified interface does not exist, the requested address is unknown, or the user is not privileged and tried to alter an interface's configuration. .Sh SEE ALSO .Xr netstat 1 , .Xr carp 4 , .Xr gif 4 , .Xr netintro 4 , .Xr pfsync 4 , .Xr polling 4 , .Xr vlan 4 , .Xr vxlan 4 , .Xr devd.conf 5 , .\" .Xr eon 5 , .Xr devd 8 , .Xr jail 8 , .Xr rc 8 , .Xr routed 8 , .Xr sysctl 8 .Sh HISTORY The .Nm utility appeared in .Bx 4.2 . .Sh BUGS Basic IPv6 node operation requires a link-local address on each interface configured for IPv6. Normally, such an address is automatically configured by the kernel on each interface added to the system or enabled; this behavior may be disabled by setting per-interface flag .Cm -auto_linklocal . The default value of this flag is 1 and can be disabled by using the sysctl MIB variable .Va net.inet6.ip6.auto_linklocal . .Pp Do not configure IPv6 addresses with no link-local address by using .Nm . It can result in unexpected behaviors of the kernel. Index: projects/cxl_iscsi/sbin =================================================================== --- projects/cxl_iscsi/sbin (revision 279716) +++ projects/cxl_iscsi/sbin (revision 279717) Property changes on: projects/cxl_iscsi/sbin ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sbin:r279515-279716 Index: projects/cxl_iscsi/share/misc/committers-src.dot =================================================================== --- projects/cxl_iscsi/share/misc/committers-src.dot (revision 279716) +++ projects/cxl_iscsi/share/misc/committers-src.dot (revision 279717) @@ -1,717 +1,721 @@ # $FreeBSD$ # This file is meant to list all FreeBSD src committers and describe the # mentor-mentee relationships between them. # The graphical output can be generated from this file with the following # command: # $ dot -T png -o file.png committers-src.dot # # The dot binary is part of the graphics/graphviz port. digraph src { # Node definitions follow this example: # # foo [label="Foo Bar\nfoo@FreeBSD.org\n????/??/??"] # # ????/??/?? is the date when the commit bit was obtained, usually the one you # can find looking at svn logs for the svnadmin/access file. # Use YYYY/MM/DD format. # # For returned commit bits, the node definition will follow this example: # # foo [label="Foo Bar\nfoo@FreeBSD.org\n????/??/??\n????/??/??"] # # The first date is the same as for an active committer, the second date is # the date when the commit bit has been returned. Again, check svn logs. node [color=grey62, style=filled, bgcolor=black]; # Alumni go here.. Try to keep things sorted. alm [label="Andrew Moore\nalm@FreeBSD.org\n1993/06/12\n????/??/??"] anholt [label="Eric Anholt\nanholt@FreeBSD.org\n2002/04/22\n2008/08/07"] archie [label="Archie Cobbs\narchie@FreeBSD.org\n1998/11/06\n2006/06/09"] arr [label="Andrew R. Reiter\narr@FreeBSD.org\n2001/11/02\n2005/05/25"] arun [label="Arun Sharma\narun@FreeBSD.org\n2003/03/06\n2006/12/16"] asmodai [label="Jeroen Ruigrok\nasmodai@FreeBSD.org\n1999/12/16\n2001/11/16"] benjsc [label="Benjamin Close\nbenjsc@FreeBSD.org\n2007/02/09\n2010/09/15"] billf [label="Bill Fumerola\nbillf@FreeBSD.org\n1998/11/11\n2008/11/10"] bmah [label="Bruce A. Mah\nbmah@FreeBSD.org\n2002/01/29\n2009/09/13"] bmilekic [label="Bosko Milekic\nbmilekic@FreeBSD.org\n2000/09/21\n2008/11/10"] bushman [label="Michael Bushkov\nbushman@FreeBSD.org\n2007/03/10\n2010/04/29"] carl [label="Carl Delsey\ncarl@FreeBSD.org\n2013/01/14\n2014/03/06"] ceri [label="Ceri Davies\nceri@FreeBSD.org\n2006/11/07\n2012/03/07"] cjc [label="Crist J. Clark\ncjc@FreeBSD.org\n2001/06/01\n2006/12/29"] davidxu [label="David Xu\ndavidxu@FreeBSD.org\n2002/09/02\n2014/04/14"] dds [label="Diomidis Spinellis\ndds@FreeBSD.org\n2003/06/20\n2010/09/22"] dhartmei [label="Daniel Hartmeier\ndhartmei@FreeBSD.org\n2004/04/06\n2008/12/08"] dmlb [label="Duncan Barclay\ndmlb@FreeBSD.org\n2001/12/14\n2008/11/10"] dougb [label="Doug Barton\ndougb@FreeBSD.org\n2000/10/26\n2012/10/08"] eik [label="Oliver Eikemeier\neik@FreeBSD.org\n2004/05/20\n2008/11/10"] furuta [label="Atsushi Furuta\nfuruta@FreeBSD.org\n2000/06/21\n2003/03/08"] gj [label="Gary L. Jennejohn\ngj@FreeBSD.org\n1994/??/??\n2006/04/28"] groudier [label="Gerard Roudier\ngroudier@FreeBSD.org\n1999/12/30\n2006/04/06"] jake [label="Jake Burkholder\njake@FreeBSD.org\n2000/05/16\n2008/11/10"] jayanth [label="Jayanth Vijayaraghavan\njayanth@FreeBSD.org\n2000/05/08\n2008/11/10"] jb [label="John Birrell\njb@FreeBSD.org\n1997/03/27\n2009/12/15"] jdp [label="John Polstra\njdp@FreeBSD.org\n1995/12/07\n2008/02/26"] jedgar [label="Chris D. Faulhaber\njedgar@FreeBSD.org\n1999/12/15\n2006/04/07"] jkh [label="Jordan K. Hubbard\njkh@FreeBSD.org\n1993/06/12\n2008/06/13"] jlemon [label="Jonathan Lemon\njlemon@FreeBSD.org\n1997/08/14\n2008/11/10"] joe [label="Josef Karthauser\njoe@FreeBSD.org\n1999/10/22\n2008/08/10"] jtc [label="J.T. Conklin\njtc@FreeBSD.org\n1993/06/12\n????/??/??"] kbyanc [label="Kelly Yancey\nkbyanc@FreeBSD.org\n2000/07/11\n2006/07/25"] keichii [label="Michael Wu\nkeichii@FreeBSD.org\n2001/03/07\n2006/04/28"] linimon [label="Mark Linimon\nlinimon@FreeBSD.org\n2006/09/30\n2008/05/04"] lulf [label="Ulf Lilleengen\nlulf@FreeBSD.org\n2007/10/24\n2012/01/19"] mb [label="Maxim Bolotin\nmb@FreeBSD.org\n2000/04/06\n2003/03/08"] marks [label="Mark Santcroos\nmarks@FreeBSD.org\n2004/03/18\n2008/09/29"] mike [label="Mike Barcroft\nmike@FreeBSD.org\n2001/07/17\n2006/04/28"] msmith [label="Mike Smith\nmsmith@FreeBSD.org\n1996/10/22\n2003/12/15"] murray [label="Murray Stokely\nmurray@FreeBSD.org\n2000/04/05\n2010/07/25"] mux [label="Maxime Henrion\nmux@FreeBSD.org\n2002/03/03\n2011/06/22"] nate [label="Nate Willams\nnate@FreeBSD.org\n1993/06/12\n2003/12/15"] njl [label="Nate Lawson\nnjl@FreeBSD.org\n2002/08/07\n2008/02/16"] non [label="Noriaki Mitsnaga\nnon@FreeBSD.org\n2000/06/19\n2007/03/06"] onoe [label="Atsushi Onoe\nonoe@FreeBSD.org\n2000/07/21\n2008/11/10"] rafan [label="Rong-En Fan\nrafan@FreeBSD.org\n2007/01/31\n2012/07/23"] randi [label="Randi Harper\nrandi@FreeBSD.org\n2010/04/20\n2012/05/10"] rgrimes [label="Rod Grimes\nrgrimes@FreeBSD.org\n1993/06/12\n2003/03/08"] rink [label="Rink Springer\nrink@FreeBSD.org\n2006/01/16\n2010/11/04"] robert [label="Robert Drehmel\nrobert@FreeBSD.org\n2001/08/23\n2006/05/13"] sah [label="Sam Hopkins\nsah@FreeBSD.org\n2004/12/15\n2008/11/10"] shafeeq [label="Shafeeq Sinnamohideen\nshafeeq@FreeBSD.org\n2000/06/19\n2006/04/06"] sheldonh [label="Sheldon Hearn\nsheldonh@FreeBSD.org\n1999/06/14\n2006/05/13"] shiba [label="Takeshi Shibagaki\nshiba@FreeBSD.org\n2000/06/19\n2008/11/10"] shin [label="Yoshinobu Inoue\nshin@FreeBSD.org\n1999/07/29\n2003/03/08"] snb [label="Nick Barkas\nsnb@FreeBSD.org\n2009/05/05\n2010/11/04"] tmm [label="Thomas Moestl\ntmm@FreeBSD.org\n2001/03/07\n2006/07/12"] toshi [label="Toshihiko Arai\ntoshi@FreeBSD.org\n2000/07/06\n2003/03/08"] tshiozak [label="Takuya SHIOZAKI\ntshiozak@FreeBSD.org\n2001/04/25\n2003/03/08"] uch [label="UCHIYAMA Yasushi\nuch@FreeBSD.org\n2000/06/21\n2002/04/24"] wilko [label="Wilko Bulte\nwilko@FreeBSD.org\n2000/01/13\n2013/01/17"] yar [label="Yar Tikhiy\nyar@FreeBSD.org\n2001/03/25\n2012/05/23"] zack [label="Zack Kirsch\nzack@FreeBSD.org\n2010/11/05\n2012/09/08"] node [color=lightblue2, style=filled, bgcolor=black]; # Current src committers go here. Try to keep things sorted. ache [label="Andrey Chernov\nache@FreeBSD.org\n1993/10/31"] achim [label="Achim Leubner\nachim@FreeBSD.org\n2013/01/23"] adrian [label="Adrian Chadd\nadrian@FreeBSD.org\n2000/07/03"] ae [label="Andrey V. Elsukov\nae@FreeBSD.org\n2010/06/03"] akiyama [label="Shunsuke Akiyama\nakiyama@FreeBSD.org\n2000/06/19"] alc [label="Alan Cox\nalc@FreeBSD.org\n1999/02/23"] ambrisko [label="Doug Ambrisko\nambrisko@FreeBSD.org\n2001/12/19"] anchie [label="Ana Kukec\nanchie@FreeBSD.org\n2010/04/14"] andre [label="Andre Oppermann\nandre@FreeBSD.org\n2003/11/12"] andreast [label="Andreas Tobler\nandreast@FreeBSD.org\n2010/09/05"] andrew [label="Andrew Turner\nandrew@FreeBSD.org\n2010/07/19"] antoine [label="Antoine Brodin\nantoine@FreeBSD.org\n2008/02/03"] ariff [label="Ariff Abdullah\nariff@FreeBSD.org\n2005/11/14"] art [label="Artem Belevich\nart@FreeBSD.org\n2011/03/29"] arybchik [label="Andrew Rybchenko\narybchik@FreeBSD.org\n2014/10/12"] asomers [label="Alan Somers\nasomers@FreeBSD.org\n2013/04/24"] avg [label="Andriy Gapon\navg@FreeBSD.org\n2009/02/18"] bapt [label="Baptiste Daroussin\nbapt@FreeBSD.org\n2011/12/23"] bdrewery [label="Bryan Drewery\nbdrewery@FreeBSD.org\n2013/12/14"] benl [label="Ben Laurie\nbenl@FreeBSD.org\n2011/05/18"] benno [label="Benno Rice\nbenno@FreeBSD.org\n2000/11/02"] bms [label="Bruce M Simpson\nbms@FreeBSD.org\n2003/08/06"] br [label="Ruslan Bukin\nbr@FreeBSD.org\n2013/09/02"] brian [label="Brian Somers\nbrian@FreeBSD.org\n1996/12/16"] brooks [label="Brooks Davis\nbrooks@FreeBSD.org\n2001/06/21"] brucec [label="Bruce Cran\nbrucec@FreeBSD.org\n2010/01/29"] brueffer [label="Christian Brueffer\nbrueffer@FreeBSD.org\n2006/02/28"] bruno [label="Bruno Ducrot\nbruno@FreeBSD.org\n2005/07/18"] bryanv [label="Bryan Venteicher\nbryanv@FreeBSD.org\n2012/11/03"] bschmidt [label="Bernhard Schmidt\nbschmidt@FreeBSD.org\n2010/02/06"] bz [label="Bjoern A. Zeeb\nbz@FreeBSD.org\n2004/07/27"] cognet [label="Olivier Houchard\ncognet@FreeBSD.org\n2002/10/09"] cokane [label="Coleman Kane\ncokane@FreeBSD.org\n2000/06/19"] cperciva [label="Colin Percival\ncperciva@FreeBSD.org\n2004/01/20"] csjp [label="Christian S.J. Peron\ncsjp@FreeBSD.org\n2004/05/04"] das [label="David Schultz\ndas@FreeBSD.org\n2003/02/21"] davide [label="Davide Italiano\ndavide@FreeBSD.org\n2012/01/27"] dchagin [label="Dmitry Chagin\ndchagin@FreeBSD.org\n2009/02/28"] delphij [label="Xin Li\ndelphij@FreeBSD.org\n2004/09/14"] des [label="Dag-Erling Smorgrav\ndes@FreeBSD.org\n1998/04/03"] dfr [label="Doug Rabson\ndfr@FreeBSD.org\n????/??/??"] dg [label="David Greenman\ndg@FreeBSD.org\n1993/06/14"] dim [label="Dimitry Andric\ndim@FreeBSD.org\n2010/08/30"] dteske [label="Devin Teske\ndteske@FreeBSD.org\n2012/04/10"] dumbbell [label="Jean-Sebastien Pedron\ndumbbell@FreeBSD.org\n2004/11/29"] dwmalone [label="David Malone\ndwmalone@FreeBSD.org\n2000/07/11"] eadler [label="Eitan Adler\neadler@FreeBSD.org\n2012/01/18"] ed [label="Ed Schouten\ned@FreeBSD.org\n2008/05/22"] edavis [label="Eric Davis\nedavis@FreeBSD.org\n2013/10/09"] edwin [label="Edwin Groothuis\nedwin@FreeBSD.org\n2007/06/25"] eivind [label="Eivind Eklund\neivind@FreeBSD.org\n1997/02/02"] emaste [label="Ed Maste\nemaste@FreeBSD.org\n2005/10/04"] emax [label="Maksim Yevmenkin\nemax@FreeBSD.org\n2003/10/12"] eri [label="Ermal Luci\neri@FreeBSD.org\n2008/06/11"] erj [label="Eric Joyner\nerj@FreeBSD.org\n2014/12/14"] fabient [label="Fabien Thomas\nfabient@FreeBSD.org\n2009/03/16"] fanf [label="Tony Finch\nfanf@FreeBSD.org\n2002/05/05"] fjoe [label="Max Khon\nfjoe@FreeBSD.org\n2001/08/06"] flz [label="Florent Thoumie\nflz@FreeBSD.org\n2006/03/30"] gabor [label="Gabor Kovesdan\ngabor@FreeBSD.org\n2010/02/02"] gad [label="Garance A. Drosehn\ngad@FreeBSD.org\n2000/10/27"] gallatin [label="Andrew Gallatin\ngallatin@FreeBSD.org\n1999/01/15"] gavin [label="Gavin Atkinson\ngavin@FreeBSD.org\n2009/12/07"] gibbs [label="Justin T. Gibbs\ngibbs@FreeBSD.org\n????/??/??"] gjb [label="Glen Barber\ngjb@FreeBSD.org\n2013/06/04"] gleb [label="Gleb Kurtsou\ngleb@FreeBSD.org\n2011/09/19"] glebius [label="Gleb Smirnoff\nglebius@FreeBSD.org\n2004/07/14"] gnn [label="George V. Neville-Neil\ngnn@FreeBSD.org\n2004/10/11"] gordon [label="Gordon Tetlow\ngordon@FreeBSD.org\n2002/05/17"] grehan [label="Peter Grehan\ngrehan@FreeBSD.org\n2002/08/08"] grog [label="Greg Lehey\ngrog@FreeBSD.org\n1998/08/30"] gshapiro [label="Gregory Shapiro\ngshapiro@FreeBSD.org\n2000/07/12"] harti [label="Hartmut Brandt\nharti@FreeBSD.org\n2003/01/29"] hiren [label="Hiren Panchasara\nhiren@FreeBSD.org\n2013/04/12"] hmp [label="Hiten Pandya\nhmp@FreeBSD.org\n2004/03/23"] ian [label="Ian Lepore\nian@FreeBSD.org\n2013/01/07"] iedowse [label="Ian Dowse\niedowse@FreeBSD.org\n2000/12/01"] imp [label="Warner Losh\nimp@FreeBSD.org\n1996/09/20"] ivoras [label="Ivan Voras\nivoras@FreeBSD.org\n2008/06/10"] jamie [label="Jamie Gritton\njamie@FreeBSD.org\n2009/01/28"] jasone [label="Jason Evans\njasone@FreeBSD.org\n1999/03/03"] jceel [label="Jakub Klama\njceel@FreeBSD.org\n2011/09/25"] jch [label="Julien Charbon\njch@FreeBSD.org\n2014/09/24"] jchandra [label="Jayachandran C.\njchandra@FreeBSD.org\n2010/05/19"] jeff [label="Jeff Roberson\njeff@FreeBSD.org\n2002/02/21"] jh [label="Jaakko Heinonen\njh@FreeBSD.org\n2009/10/02"] jhb [label="John Baldwin\njhb@FreeBSD.org\n1999/08/23"] jhibbits [label="Justin Hibbits\njhibbits@FreeBSD.org\n2011/11/30"] jilles [label="Jilles Tjoelker\njilles@FreeBSD.org\n2009/05/22"] jimharris [label="Jim Harris\njimharris@FreeBSD.org\n2011/12/09"] jinmei [label="JINMEI Tatuya\njinmei@FreeBSD.org\n2007/03/17"] jkim [label="Jung-uk Kim\njkim@FreeBSD.org\n2005/07/06"] jkoshy [label="A. Joseph Koshy\njkoshy@FreeBSD.org\n1998/05/13"] jlh [label="Jeremie Le Hen\njlh@FreeBSD.org\n2012/04/22"] jls [label="Jordan Sissel\njls@FreeBSD.org\n2006/12/06"] jmg [label="John-Mark Gurney\njmg@FreeBSD.org\n1997/02/13"] jmmv [label="Julio Merino\njmmv@FreeBSD.org\n2013/11/02"] joerg [label="Joerg Wunsch\njoerg@FreeBSD.org\n1993/11/14"] jon [label="Jonathan Chen\njon@FreeBSD.org\n2000/10/17"] jonathan [label="Jonathan Anderson\njonathan@FreeBSD.org\n2010/10/07"] jpaetzel [label="Josh Paetzel\njpaetzel@FreeBSD.org\n2011/01/21"] julian [label="Julian Elischer\njulian@FreeBSD.org\n1993/04/19"] jwd [label="John De Boskey\njwd@FreeBSD.org\n2000/05/19"] kaiw [label="Kai Wang\nkaiw@FreeBSD.org\n2007/09/26"] kan [label="Alexander Kabaev\nkan@FreeBSD.org\n2002/07/21"] kargl [label="Steven G. Kargl\nkargl@FreeBSD.org\n2011/01/17"] ken [label="Ken Merry\nken@FreeBSD.org\n1998/09/08"] kensmith [label="Ken Smith\nkensmith@FreeBSD.org\n2004/01/23"] kevlo [label="Kevin Lo\nkevlo@FreeBSD.org\n2006/07/23"] kib [label="Konstantin Belousov\nkib@FreeBSD.org\n2006/06/03"] kmacy [label="Kip Macy\nkmacy@FreeBSD.org\n2005/06/01"] le [label="Lukas Ertl\nle@FreeBSD.org\n2004/02/02"] loos [label="Luiz Otavio O Souza\nloos@FreeBSD.org\n2013/07/03"] lstewart [label="Lawrence Stewart\nlstewart@FreeBSD.org\n2008/10/06"] marcel [label="Marcel Moolenaar\nmarcel@FreeBSD.org\n1999/07/03"] marius [label="Marius Strobl\nmarius@FreeBSD.org\n2004/04/17"] markj [label="Mark Johnston\nmarkj@FreeBSD.org\n2012/12/18"] markm [label="Mark Murray\nmarkm@FreeBSD.org\n1995/04/24"] markus [label="Markus Brueffer\nmarkus@FreeBSD.org\n2006/06/01"] matteo [label="Matteo Riondato\nmatteo@FreeBSD.org\n2006/01/18"] mav [label="Alexander Motin\nmav@FreeBSD.org\n2007/04/12"] maxim [label="Maxim Konovalov\nmaxim@FreeBSD.org\n2002/02/07"] mdf [label="Matthew Fleming\nmdf@FreeBSD.org\n2010/06/04"] mdodd [label="Matthew N. Dodd\nmdodd@FreeBSD.org\n1999/07/27"] melifaro [label="Alexander V. Chernikov\nmelifaro@FreeBSD.org\n2011/10/04"] mjacob [label="Matt Jacob\nmjacob@FreeBSD.org\n1997/08/13"] mjg [label="Mateusz Guzik\nmjg@FreeBSD.org\n2012/06/04"] mlaier [label="Max Laier\nmlaier@FreeBSD.org\n2004/02/10"] monthadar [label="Monthadar Al Jaberi\nmonthadar@FreeBSD.org\n2012/04/02"] mp [label="Mark Peek\nmp@FreeBSD.org\n2001/07/27"] mr [label="Michael Reifenberger\nmr@FreeBSD.org\n2001/09/30"] neel [label="Neel Natu\nneel@FreeBSD.org\n2009/09/20"] netchild [label="Alexander Leidinger\nnetchild@FreeBSD.org\n2005/03/31"] ngie [label="Garrett Cooper\nngie@FreeBSD.org\n2014/07/27"] nork [label="Norikatsu Shigemura\nnork@FreeBSD.org\n2009/06/09"] np [label="Navdeep Parhar\nnp@FreeBSD.org\n2009/06/05"] nwhitehorn [label="Nathan Whitehorn\nnwhitehorn@FreeBSD.org\n2008/07/03"] obrien [label="David E. O'Brien\nobrien@FreeBSD.org\n1996/10/29"] olli [label="Oliver Fromme\nolli@FreeBSD.org\n2008/02/14"] peadar [label="Peter Edwards\npeadar@FreeBSD.org\n2004/03/08"] peter [label="Peter Wemm\npeter@FreeBSD.org\n1995/07/04"] peterj [label="Peter Jeremy\npeterj@FreeBSD.org\n2012/09/14"] pfg [label="Pedro Giffuni\npfg@FreeBSD.org\n2011/12/01"] philip [label="Philip Paeps\nphilip@FreeBSD.org\n2004/01/21"] phk [label="Poul-Henning Kamp\nphk@FreeBSD.org\n1994/02/21"] pho [label="Peter Holm\npho@FreeBSD.org\n2008/11/16"] pjd [label="Pawel Jakub Dawidek\npjd@FreeBSD.org\n2004/02/02"] pkelsey [label="Patrick Kelsey\pkelsey@FreeBSD.org\n2014/05/29"] pluknet [label="Sergey Kandaurov\npluknet@FreeBSD.org\n2010/10/05"] ps [label="Paul Saab\nps@FreeBSD.org\n2000/02/23"] qingli [label="Qing Li\nqingli@FreeBSD.org\n2005/04/13"] ray [label="Aleksandr Rybalko\nray@FreeBSD.org\n2011/05/25"] rdivacky [label="Roman Divacky\nrdivacky@FreeBSD.org\n2008/03/13"] remko [label="Remko Lodder\nremko@FreeBSD.org\n2007/02/23"] rik [label="Roman Kurakin\nrik@FreeBSD.org\n2003/12/18"] rmacklem [label="Rick Macklem\nrmacklem@FreeBSD.org\n2009/03/27"] rmh [label="Robert Millan\nrmh@FreeBSD.org\n2011/09/18"] rnoland [label="Robert Noland\nrnoland@FreeBSD.org\n2008/09/15"] roberto [label="Ollivier Robert\nroberto@FreeBSD.org\n1995/02/22"] royger [label="Roger Pau Monne\nroyger@FreeBSD.org\n2013/11/26"] rpaulo [label="Rui Paulo\nrpaulo@FreeBSD.org\n2007/09/25"] rrs [label="Randall R Stewart\nrrs@FreeBSD.org\n2007/02/08"] rse [label="Ralf S. Engelschall\nrse@FreeBSD.org\n1997/07/31"] rstone [label="Ryan Stone\nrstone@FreeBSD.org\n2010/04/19"] ru [label="Ruslan Ermilov\nru@FreeBSD.org\n1999/05/27"] rwatson [label="Robert N. M. Watson\nrwatson@FreeBSD.org\n1999/12/16"] sam [label="Sam Leffler\nsam@FreeBSD.org\n2002/07/02"] sanpei [label="MIHIRA Sanpei Yoshiro\nsanpei@FreeBSD.org\n2000/06/19"] sbruno [label="Sean Bruno\nsbruno@FreeBSD.org\n2008/08/02"] scf [label="Sean C. Farley\nscf@FreeBSD.org\n2007/06/24"] schweikh [label="Jens Schweikhardt\nschweikh@FreeBSD.org\n2001/04/06"] scottl [label="Scott Long\nscottl@FreeBSD.org\n2000/09/28"] se [label="Stefan Esser\nse@FreeBSD.org\n1994/08/26"] sephe [label="Sepherosa Ziehau\nsephe@FreeBSD.org\n2007/03/28"] sepotvin [label="Stephane E. Potvin\nsepotvin@FreeBSD.org\n2007/02/15"] simon [label="Simon L. Nielsen\nsimon@FreeBSD.org\n2006/03/07"] sjg [label="Simon J. Gerraty\nsjg@FreeBSD.org\n2012/10/23"] slm [label="Stephen McConnell\nslm@FreeBSD.org\n2014/05/07"] smh [label="Steven Hartland\nsmh@FreeBSD.org\n2012/11/12"] sobomax [label="Maxim Sobolev\nsobomax@FreeBSD.org\n2001/07/25"] sos [label="Soren Schmidt\nsos@FreeBSD.org\n????/??/??"] sson [label="Stacey Son\nsson@FreeBSD.org\n2008/07/08"] stas [label="Stanislav Sedov\nstas@FreeBSD.org\n2008/08/22"] suz [label="SUZUKI Shinsuke\nsuz@FreeBSD.org\n2002/03/26"] syrinx [label="Shteryana Shopova\nsyrinx@FreeBSD.org\n2006/10/07"] takawata [label="Takanori Watanabe\ntakawata@FreeBSD.org\n2000/07/06"] theraven [label="David Chisnall\ntheraven@FreeBSD.org\n2011/11/11"] thompsa [label="Andrew Thompson\nthompsa@FreeBSD.org\n2005/05/25"] ticso [label="Bernd Walter\nticso@FreeBSD.org\n2002/01/31"] tijl [label="Tijl Coosemans\ntijl@FreeBSD.org\n2010/07/16"] trasz [label="Edward Tomasz Napierala\ntrasz@FreeBSD.org\n2008/08/22"] trhodes [label="Tom Rhodes\ntrhodes@FreeBSD.org\n2002/05/28"] trociny [label="Mikolaj Golub\ntrociny@FreeBSD.org\n2011/03/10"] tuexen [label="Michael Tuexen\ntuexen@FreeBSD.org\n2009/06/06"] tychon [label="Tycho Nightingale\ntychon@FreeBSD.org\n2014/01/21"] ume [label="Hajimu UMEMOTO\nume@FreeBSD.org\n2000/02/26"] uqs [label="Ulrich Spoerlein\nuqs@FreeBSD.org\n2010/01/28"] vanhu [label="Yvan Vanhullebus\nvanhu@FreeBSD.org\n2008/07/21"] versus [label="Konrad Jankowski\nversus@FreeBSD.org\n2008/10/27"] weongyo [label="Weongyo Jeong\nweongyo@FreeBSD.org\n2007/12/21"] wes [label="Wes Peters\nwes@FreeBSD.org\n1998/11/25"] +whu [label="Wei Hu\nwhu@FreeBSD.org\n2015/02/11"] wkoszek [label="Wojciech A. Koszek\nwkoszek@FreeBSD.org\n2006/02/21"] wollman [label="Garrett Wollman\nwollman@FreeBSD.org\n????/??/??"] wsalamon [label="Wayne Salamon\nwsalamon@FreeBSD.org\n2005/06/25"] yongari [label="Pyun YongHyeon\nyongari@FreeBSD.org\n2004/08/01"] zbb [label="Zbigniew Bodek\nzbb@FreeBSD.org\n2013/09/02"] zec [label="Marko Zec\nzec@FreeBSD.org\n2008/06/22"] zml [label="Zachary Loafman\nzml@FreeBSD.org\n2009/05/27"] zont [label="Andrey Zonov\nzont@FreeBSD.org\n2012/08/21"] # Pseudo target representing rev 1.1 of commit.allow day1 [label="Birth of FreeBSD"] # Here are the mentor/mentee relationships. # Group together all the mentees for a particular mentor. # Keep the list sorted by mentor login. day1 -> jtc day1 -> jkh day1 -> nate day1 -> rgrimes day1 -> alm day1 -> dg adrian -> loos adrian -> monthadar adrian -> ray adrian -> rmh ae -> melifaro alc -> davide andre -> qingli anholt -> jkim avg -> art avg -> pluknet avg -> smh bapt -> bdrewery benno -> grehan billf -> dougb billf -> gad billf -> jedgar billf -> jhb billf -> shafeeq bmilekic -> csjp bms -> dhartmei bms -> mlaier bms -> thompsa brian -> joe brooks -> bushman brooks -> jamie brooks -> theraven bz -> anchie bz -> jamie bz -> syrinx cognet -> br cognet -> jceel cognet -> kevlo cognet -> ian cognet -> wkoszek cognet -> zbb cperciva -> eadler cperciva -> flz cperciva -> randi cperciva -> simon csjp -> bushman das -> kargl delphij -> gabor delphij -> rafan des -> anholt des -> hmp des -> mike des -> olli des -> ru des -> bapt dds -> versus dfr -> gallatin dfr -> zml dg -> peter dim -> theraven dwmalone -> fanf dwmalone -> peadar dwmalone -> snb ed -> dim ed -> gavin ed -> jilles ed -> rdivacky ed -> uqs eivind -> des eivind -> rwatson emaste -> achim emaste -> rstone emaste -> dteske emaste -> markj emax -> markus fjoe -> versus gallatin -> ticso gavin -> versus gibbs -> mjacob gibbs -> njl gibbs -> royger +gibbs -> whu glebius -> mav gnn -> jinmei gnn -> rrs gnn -> ivoras gnn -> vanhu gnn -> lstewart gnn -> np gnn -> davide gnn -> arybchik gnn -> erj grehan -> bryanv grog -> edwin grog -> le grog -> peterj imp -> akiyama imp -> ambrisko imp -> andrew imp -> bmah imp -> bruno imp -> dmlb imp -> emax imp -> furuta imp -> joe imp -> jon imp -> keichii imp -> mb imp -> mr imp -> neel imp -> non imp -> nork imp -> onoe imp -> remko imp -> rik imp -> rink imp -> sanpei imp -> shiba imp -> takawata imp -> toshi imp -> uch jake -> bms jake -> gordon jake -> harti jake -> jeff jake -> kmacy jake -> robert jake -> yongari jb -> sson jdp -> fjoe jfv -> erj jhb -> arr jhb -> avg jhb -> jch jhb -> jeff jhb -> kbyanc jhb -> peterj jhb -> pfg jhb -> rnoland jimharris -> carl jkh -> dfr jkh -> gj jkh -> grog jkh -> imp jkh -> jlemon jkh -> joerg jkh -> jwd jkh -> msmith jkh -> murray jkh -> phk jkh -> wes jkh -> yar jkoshy -> kaiw jkoshy -> fabient jkoshy -> rstone jlemon -> bmilekic jlemon -> brooks jmallett -> pkelsey jmmv -> ngie joerg -> brian joerg -> eik joerg -> jmg joerg -> le joerg -> netchild joerg -> schweikh julian -> glebius julian -> davidxu julian -> archie julian -> adrian julian -> zec julian -> mp kan -> kib ken -> asomers ken -> slm kib -> ae kib -> dchagin kib -> gjb kib -> jlh kib -> jpaetzel kib -> lulf kib -> melifaro kib -> pho kib -> pluknet kib -> rdivacky kib -> rmacklem kib -> rmh kib -> stas kib -> tijl kib -> trociny kib -> zont kmacy -> lstewart marcel -> art marcel -> arun marcel -> marius marcel -> nwhitehorn marcel -> sjg markm -> jasone markm -> sheldonh mav -> ae mdf -> gleb mdodd -> jake mike -> das mlaier -> benjsc mlaier -> dhartmei mlaier -> thompsa mlaier -> eri msmith -> cokane msmith -> jasone msmith -> scottl murray -> delphij mux -> cognet mux -> dumbbell netchild -> ariff njl -> marks njl -> philip njl -> rpaulo njl -> sepotvin nwhitehorn -> andreast nwhitehorn -> jhibbits obrien -> benno obrien -> groudier obrien -> gshapiro obrien -> kan obrien -> sam peter -> asmodai peter -> jayanth peter -> ps philip -> benl philip -> ed philip -> jls philip -> matteo philip -> uqs phk -> jkoshy phk -> mux pjd -> kib pjd -> lulf pjd -> smh pjd -> trociny rgrimes -> markm rmacklem -> jwd + +royger -> whu rpaulo -> avg rpaulo -> bschmidt rpaulo -> dim rpaulo -> jmmv rpaulo -> ngie rrs -> brucec rrs -> jchandra rrs -> tuexen rstone -> markj ru -> ceri ru -> cjc ru -> eik ru -> maxim ru -> sobomax rwatson -> adrian rwatson -> antoine rwatson -> bmah rwatson -> brueffer rwatson -> bz rwatson -> cperciva rwatson -> emaste rwatson -> gnn rwatson -> jh rwatson -> jonathan rwatson -> kensmith rwatson -> kmacy rwatson -> linimon rwatson -> rmacklem rwatson -> shafeeq rwatson -> tmm rwatson -> trasz rwatson -> trhodes rwatson -> wsalamon sam -> andre sam -> benjsc sam -> sephe sbruno -> hiren sbruno -> jimharris schweikh -> dds scottl -> achim scottl -> jimharris scottl -> pjd scottl -> sah scottl -> sbruno scottl -> slm scottl -> yongari sheldonh -> dwmalone sheldonh -> iedowse shin -> ume simon -> benl sos -> marcel thompsa -> weongyo thompsa -> eri trasz -> jh trasz -> mjg ume -> jinmei ume -> suz ume -> tshiozak wes -> scf wkoszek -> jceel wollman -> gad zml -> mdf zml -> zack } Index: projects/cxl_iscsi/share =================================================================== --- projects/cxl_iscsi/share (revision 279716) +++ projects/cxl_iscsi/share (revision 279717) Property changes on: projects/cxl_iscsi/share ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/share:r279515-279716 Index: projects/cxl_iscsi/sys/amd64/amd64/mp_machdep.c =================================================================== --- projects/cxl_iscsi/sys/amd64/amd64/mp_machdep.c (revision 279716) +++ projects/cxl_iscsi/sys/amd64/amd64/mp_machdep.c (revision 279717) @@ -1,1740 +1,1740 @@ /*- * Copyright (c) 1996, by Steve Passe * Copyright (c) 2003, by Peter Wemm * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. The name of the developer may NOT be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include "opt_cpu.h" #include "opt_ddb.h" #include "opt_kstack_pages.h" #include "opt_sched.h" #include "opt_smp.h" #include #include #include #include #ifdef GPROF #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define WARMBOOT_TARGET 0 #define WARMBOOT_OFF (KERNBASE + 0x0467) #define WARMBOOT_SEG (KERNBASE + 0x0469) #define CMOS_REG (0x70) #define CMOS_DATA (0x71) #define BIOS_RESET (0x0f) #define BIOS_WARM (0x0a) /* lock region used by kernel profiling */ int mcount_lock; int mp_naps; /* # of Applications processors */ int boot_cpu_id = -1; /* designated BSP */ extern struct pcpu __pcpu[]; /* AP uses this during bootstrap. Do not staticize. */ char *bootSTK; int bootAP; /* Free these after use */ void *bootstacks[MAXCPU]; /* Temporary variables for init_secondary() */ char *doublefault_stack; char *nmi_stack; void *dpcpu; struct pcb stoppcbs[MAXCPU]; struct susppcb **susppcbs; /* Variables needed for SMP tlb shootdown. */ vm_offset_t smp_tlb_addr2; struct invpcid_descr smp_tlb_invpcid; volatile int smp_tlb_wait; uint64_t pcid_cr3; pmap_t smp_tlb_pmap; extern int invpcid_works; #ifdef COUNT_IPIS /* Interrupt counts. */ static u_long *ipi_preempt_counts[MAXCPU]; static u_long *ipi_ast_counts[MAXCPU]; u_long *ipi_invltlb_counts[MAXCPU]; u_long *ipi_invlrng_counts[MAXCPU]; u_long *ipi_invlpg_counts[MAXCPU]; u_long *ipi_invlcache_counts[MAXCPU]; u_long *ipi_rendezvous_counts[MAXCPU]; static u_long *ipi_hardclock_counts[MAXCPU]; #endif /* Default cpu_ops implementation. */ struct cpu_ops cpu_ops; extern inthand_t IDTVEC(fast_syscall), IDTVEC(fast_syscall32); extern int pmap_pcid_enabled; /* * Local data and functions. */ static volatile cpuset_t ipi_nmi_pending; /* used to hold the AP's until we are ready to release them */ struct mtx ap_boot_mtx; /* Set to 1 once we're ready to let the APs out of the pen. */ static volatile int aps_ready = 0; /* * Store data from cpu_add() until later in the boot when we actually setup * the APs. */ struct cpu_info { int cpu_present:1; int cpu_bsp:1; int cpu_disabled:1; int cpu_hyperthread:1; } static cpu_info[MAX_APIC_ID + 1]; int cpu_apic_ids[MAXCPU]; int apic_cpuids[MAX_APIC_ID + 1]; /* Holds pending bitmap based IPIs per CPU */ volatile u_int cpu_ipi_pending[MAXCPU]; static u_int boot_address; static int cpu_logical; /* logical cpus per core */ static int cpu_cores; /* cores per package */ static void assign_cpu_ids(void); static void set_interrupt_apic_ids(void); static int start_ap(int apic_id); static void release_aps(void *dummy); static u_int hyperthreading_cpus; /* logical cpus sharing L1 cache */ static int hyperthreading_allowed = 1; static u_int bootMP_size; static void mem_range_AP_init(void) { if (mem_range_softc.mr_op && mem_range_softc.mr_op->initAP) mem_range_softc.mr_op->initAP(&mem_range_softc); } static void topo_probe_amd(void) { int core_id_bits; int id; /* AMD processors do not support HTT. */ cpu_logical = 1; if ((amd_feature2 & AMDID2_CMP) == 0) { cpu_cores = 1; return; } core_id_bits = (cpu_procinfo2 & AMDID_COREID_SIZE) >> AMDID_COREID_SIZE_SHIFT; if (core_id_bits == 0) { cpu_cores = (cpu_procinfo2 & AMDID_CMP_CORES) + 1; return; } /* Fam 10h and newer should get here. */ for (id = 0; id <= MAX_APIC_ID; id++) { /* Check logical CPU availability. */ if (!cpu_info[id].cpu_present || cpu_info[id].cpu_disabled) continue; /* Check if logical CPU has the same package ID. */ if ((id >> core_id_bits) != (boot_cpu_id >> core_id_bits)) continue; cpu_cores++; } } /* * Round up to the next power of two, if necessary, and then * take log2. * Returns -1 if argument is zero. */ static __inline int mask_width(u_int x) { return (fls(x << (1 - powerof2(x))) - 1); } static void topo_probe_0x4(void) { u_int p[4]; int pkg_id_bits; int core_id_bits; int max_cores; int max_logical; int id; /* Both zero and one here mean one logical processor per package. */ max_logical = (cpu_feature & CPUID_HTT) != 0 ? (cpu_procinfo & CPUID_HTT_CORES) >> 16 : 1; if (max_logical <= 1) return; /* * Because of uniformity assumption we examine only * those logical processors that belong to the same * package as BSP. Further, we count number of * logical processors that belong to the same core * as BSP thus deducing number of threads per core. */ if (cpu_high >= 0x4) { cpuid_count(0x04, 0, p); max_cores = ((p[0] >> 26) & 0x3f) + 1; } else max_cores = 1; core_id_bits = mask_width(max_logical/max_cores); if (core_id_bits < 0) return; pkg_id_bits = core_id_bits + mask_width(max_cores); for (id = 0; id <= MAX_APIC_ID; id++) { /* Check logical CPU availability. */ if (!cpu_info[id].cpu_present || cpu_info[id].cpu_disabled) continue; /* Check if logical CPU has the same package ID. */ if ((id >> pkg_id_bits) != (boot_cpu_id >> pkg_id_bits)) continue; cpu_cores++; /* Check if logical CPU has the same package and core IDs. */ if ((id >> core_id_bits) == (boot_cpu_id >> core_id_bits)) cpu_logical++; } KASSERT(cpu_cores >= 1 && cpu_logical >= 1, ("topo_probe_0x4 couldn't find BSP")); cpu_cores /= cpu_logical; hyperthreading_cpus = cpu_logical; } static void topo_probe_0xb(void) { u_int p[4]; int bits; int cnt; int i; int logical; int type; int x; /* We only support three levels for now. */ for (i = 0; i < 3; i++) { cpuid_count(0x0b, i, p); /* Fall back if CPU leaf 11 doesn't really exist. */ if (i == 0 && p[1] == 0) { topo_probe_0x4(); return; } bits = p[0] & 0x1f; logical = p[1] &= 0xffff; type = (p[2] >> 8) & 0xff; if (type == 0 || logical == 0) break; /* * Because of uniformity assumption we examine only * those logical processors that belong to the same * package as BSP. */ for (cnt = 0, x = 0; x <= MAX_APIC_ID; x++) { if (!cpu_info[x].cpu_present || cpu_info[x].cpu_disabled) continue; if (x >> bits == boot_cpu_id >> bits) cnt++; } if (type == CPUID_TYPE_SMT) cpu_logical = cnt; else if (type == CPUID_TYPE_CORE) cpu_cores = cnt; } if (cpu_logical == 0) cpu_logical = 1; cpu_cores /= cpu_logical; } /* * Both topology discovery code and code that consumes topology * information assume top-down uniformity of the topology. * That is, all physical packages must be identical and each * core in a package must have the same number of threads. * Topology information is queried only on BSP, on which this * code runs and for which it can query CPUID information. * Then topology is extrapolated on all packages using the * uniformity assumption. */ static void topo_probe(void) { static int cpu_topo_probed = 0; if (cpu_topo_probed) return; CPU_ZERO(&logical_cpus_mask); if (mp_ncpus <= 1) cpu_cores = cpu_logical = 1; else if (cpu_vendor_id == CPU_VENDOR_AMD) topo_probe_amd(); else if (cpu_vendor_id == CPU_VENDOR_INTEL) { /* * See Intel(R) 64 Architecture Processor * Topology Enumeration article for details. * * Note that 0x1 <= cpu_high < 4 case should be * compatible with topo_probe_0x4() logic when * CPUID.1:EBX[23:16] > 0 (cpu_cores will be 1) * or it should trigger the fallback otherwise. */ if (cpu_high >= 0xb) topo_probe_0xb(); else if (cpu_high >= 0x1) topo_probe_0x4(); } /* * Fallback: assume each logical CPU is in separate * physical package. That is, no multi-core, no SMT. */ if (cpu_cores == 0 || cpu_logical == 0) cpu_cores = cpu_logical = 1; cpu_topo_probed = 1; } struct cpu_group * cpu_topo(void) { int cg_flags; /* * Determine whether any threading flags are * necessry. */ topo_probe(); if (cpu_logical > 1 && hyperthreading_cpus) cg_flags = CG_FLAG_HTT; else if (cpu_logical > 1) cg_flags = CG_FLAG_SMT; else cg_flags = 0; if (mp_ncpus % (cpu_cores * cpu_logical) != 0) { printf("WARNING: Non-uniform processors.\n"); printf("WARNING: Using suboptimal topology.\n"); return (smp_topo_none()); } /* * No multi-core or hyper-threaded. */ if (cpu_logical * cpu_cores == 1) return (smp_topo_none()); /* * Only HTT no multi-core. */ if (cpu_logical > 1 && cpu_cores == 1) return (smp_topo_1level(CG_SHARE_L1, cpu_logical, cg_flags)); /* * Only multi-core no HTT. */ if (cpu_cores > 1 && cpu_logical == 1) return (smp_topo_1level(CG_SHARE_L2, cpu_cores, cg_flags)); /* * Both HTT and multi-core. */ return (smp_topo_2level(CG_SHARE_L2, cpu_cores, CG_SHARE_L1, cpu_logical, cg_flags)); } /* * Calculate usable address in base memory for AP trampoline code. */ u_int mp_bootaddress(u_int basemem) { bootMP_size = mptramp_end - mptramp_start; boot_address = trunc_page(basemem * 1024); /* round down to 4k boundary */ if (((basemem * 1024) - boot_address) < bootMP_size) boot_address -= PAGE_SIZE; /* not enough, lower by 4k */ /* 3 levels of page table pages */ mptramp_pagetables = boot_address - (PAGE_SIZE * 3); return mptramp_pagetables; } void cpu_add(u_int apic_id, char boot_cpu) { if (apic_id > MAX_APIC_ID) { panic("SMP: APIC ID %d too high", apic_id); return; } KASSERT(cpu_info[apic_id].cpu_present == 0, ("CPU %d added twice", apic_id)); cpu_info[apic_id].cpu_present = 1; if (boot_cpu) { KASSERT(boot_cpu_id == -1, ("CPU %d claims to be BSP, but CPU %d already is", apic_id, boot_cpu_id)); boot_cpu_id = apic_id; cpu_info[apic_id].cpu_bsp = 1; } if (mp_ncpus < MAXCPU) { mp_ncpus++; mp_maxid = mp_ncpus - 1; } if (bootverbose) printf("SMP: Added CPU %d (%s)\n", apic_id, boot_cpu ? "BSP" : "AP"); } void cpu_mp_setmaxid(void) { /* * mp_maxid should be already set by calls to cpu_add(). * Just sanity check its value here. */ if (mp_ncpus == 0) KASSERT(mp_maxid == 0, ("%s: mp_ncpus is zero, but mp_maxid is not", __func__)); else if (mp_ncpus == 1) mp_maxid = 0; else KASSERT(mp_maxid >= mp_ncpus - 1, ("%s: counters out of sync: max %d, count %d", __func__, mp_maxid, mp_ncpus)); } int cpu_mp_probe(void) { /* * Always record BSP in CPU map so that the mbuf init code works * correctly. */ CPU_SETOF(0, &all_cpus); if (mp_ncpus == 0) { /* * No CPUs were found, so this must be a UP system. Setup * the variables to represent a system with a single CPU * with an id of 0. */ mp_ncpus = 1; return (0); } /* At least one CPU was found. */ if (mp_ncpus == 1) { /* * One CPU was found, so this must be a UP system with * an I/O APIC. */ mp_maxid = 0; return (0); } /* At least two CPUs were found. */ return (1); } /* * Initialize the IPI handlers and start up the AP's. */ void cpu_mp_start(void) { int i; /* Initialize the logical ID to APIC ID table. */ for (i = 0; i < MAXCPU; i++) { cpu_apic_ids[i] = -1; cpu_ipi_pending[i] = 0; } /* Install an inter-CPU IPI for TLB invalidation */ if (pmap_pcid_enabled) { setidt(IPI_INVLTLB, IDTVEC(invltlb_pcid), SDT_SYSIGT, SEL_KPL, 0); setidt(IPI_INVLPG, IDTVEC(invlpg_pcid), SDT_SYSIGT, SEL_KPL, 0); } else { setidt(IPI_INVLTLB, IDTVEC(invltlb), SDT_SYSIGT, SEL_KPL, 0); setidt(IPI_INVLPG, IDTVEC(invlpg), SDT_SYSIGT, SEL_KPL, 0); } setidt(IPI_INVLRNG, IDTVEC(invlrng), SDT_SYSIGT, SEL_KPL, 0); /* Install an inter-CPU IPI for cache invalidation. */ setidt(IPI_INVLCACHE, IDTVEC(invlcache), SDT_SYSIGT, SEL_KPL, 0); /* Install an inter-CPU IPI for all-CPU rendezvous */ setidt(IPI_RENDEZVOUS, IDTVEC(rendezvous), SDT_SYSIGT, SEL_KPL, 0); /* Install generic inter-CPU IPI handler */ setidt(IPI_BITMAP_VECTOR, IDTVEC(ipi_intr_bitmap_handler), SDT_SYSIGT, SEL_KPL, 0); /* Install an inter-CPU IPI for CPU stop/restart */ setidt(IPI_STOP, IDTVEC(cpustop), SDT_SYSIGT, SEL_KPL, 0); /* Install an inter-CPU IPI for CPU suspend/resume */ setidt(IPI_SUSPEND, IDTVEC(cpususpend), SDT_SYSIGT, SEL_KPL, 0); /* Set boot_cpu_id if needed. */ if (boot_cpu_id == -1) { boot_cpu_id = PCPU_GET(apic_id); cpu_info[boot_cpu_id].cpu_bsp = 1; } else KASSERT(boot_cpu_id == PCPU_GET(apic_id), ("BSP's APIC ID doesn't match boot_cpu_id")); /* Probe logical/physical core configuration. */ topo_probe(); assign_cpu_ids(); /* Start each Application Processor */ init_ops.start_all_aps(); set_interrupt_apic_ids(); } /* * Print various information about the SMP system hardware and setup. */ void cpu_mp_announce(void) { const char *hyperthread; int i; printf("FreeBSD/SMP: %d package(s) x %d core(s)", mp_ncpus / (cpu_cores * cpu_logical), cpu_cores); if (hyperthreading_cpus > 1) printf(" x %d HTT threads", cpu_logical); else if (cpu_logical > 1) printf(" x %d SMT threads", cpu_logical); printf("\n"); /* List active CPUs first. */ printf(" cpu0 (BSP): APIC ID: %2d\n", boot_cpu_id); for (i = 1; i < mp_ncpus; i++) { if (cpu_info[cpu_apic_ids[i]].cpu_hyperthread) hyperthread = "/HT"; else hyperthread = ""; printf(" cpu%d (AP%s): APIC ID: %2d\n", i, hyperthread, cpu_apic_ids[i]); } /* List disabled CPUs last. */ for (i = 0; i <= MAX_APIC_ID; i++) { if (!cpu_info[i].cpu_present || !cpu_info[i].cpu_disabled) continue; if (cpu_info[i].cpu_hyperthread) hyperthread = "/HT"; else hyperthread = ""; printf(" cpu (AP%s): APIC ID: %2d (disabled)\n", hyperthread, i); } } /* * AP CPU's call this to initialize themselves. */ void init_secondary(void) { struct pcpu *pc; struct nmi_pcpu *np; u_int64_t msr, cr0; u_int cpuid; int cpu, gsel_tss, x; struct region_descriptor ap_gdt; /* Set by the startup code for us to use */ cpu = bootAP; /* Init tss */ common_tss[cpu] = common_tss[0]; common_tss[cpu].tss_rsp0 = 0; /* not used until after switch */ common_tss[cpu].tss_iobase = sizeof(struct amd64tss) + IOPERM_BITMAP_SIZE; common_tss[cpu].tss_ist1 = (long)&doublefault_stack[PAGE_SIZE]; /* The NMI stack runs on IST2. */ np = ((struct nmi_pcpu *) &nmi_stack[PAGE_SIZE]) - 1; common_tss[cpu].tss_ist2 = (long) np; /* Prepare private GDT */ gdt_segs[GPROC0_SEL].ssd_base = (long) &common_tss[cpu]; for (x = 0; x < NGDT; x++) { if (x != GPROC0_SEL && x != (GPROC0_SEL + 1) && x != GUSERLDT_SEL && x != (GUSERLDT_SEL + 1)) ssdtosd(&gdt_segs[x], &gdt[NGDT * cpu + x]); } ssdtosyssd(&gdt_segs[GPROC0_SEL], (struct system_segment_descriptor *)&gdt[NGDT * cpu + GPROC0_SEL]); ap_gdt.rd_limit = NGDT * sizeof(gdt[0]) - 1; ap_gdt.rd_base = (long) &gdt[NGDT * cpu]; lgdt(&ap_gdt); /* does magic intra-segment return */ /* Get per-cpu data */ pc = &__pcpu[cpu]; /* prime data page for it to use */ pcpu_init(pc, cpu, sizeof(struct pcpu)); dpcpu_init(dpcpu, cpu); pc->pc_apic_id = cpu_apic_ids[cpu]; pc->pc_prvspace = pc; pc->pc_curthread = 0; pc->pc_tssp = &common_tss[cpu]; pc->pc_commontssp = &common_tss[cpu]; pc->pc_rsp0 = 0; pc->pc_tss = (struct system_segment_descriptor *)&gdt[NGDT * cpu + GPROC0_SEL]; pc->pc_fs32p = &gdt[NGDT * cpu + GUFS32_SEL]; pc->pc_gs32p = &gdt[NGDT * cpu + GUGS32_SEL]; pc->pc_ldt = (struct system_segment_descriptor *)&gdt[NGDT * cpu + GUSERLDT_SEL]; /* Save the per-cpu pointer for use by the NMI handler. */ np->np_pcpu = (register_t) pc; wrmsr(MSR_FSBASE, 0); /* User value */ wrmsr(MSR_GSBASE, (u_int64_t)pc); wrmsr(MSR_KGSBASE, (u_int64_t)pc); /* XXX User value while we're in the kernel */ lidt(&r_idt); gsel_tss = GSEL(GPROC0_SEL, SEL_KPL); ltr(gsel_tss); /* * Set to a known state: * Set by mpboot.s: CR0_PG, CR0_PE * Set by cpu_setregs: CR0_NE, CR0_MP, CR0_TS, CR0_WP, CR0_AM */ cr0 = rcr0(); cr0 &= ~(CR0_CD | CR0_NW | CR0_EM); load_cr0(cr0); /* Set up the fast syscall stuff */ msr = rdmsr(MSR_EFER) | EFER_SCE; wrmsr(MSR_EFER, msr); wrmsr(MSR_LSTAR, (u_int64_t)IDTVEC(fast_syscall)); wrmsr(MSR_CSTAR, (u_int64_t)IDTVEC(fast_syscall32)); msr = ((u_int64_t)GSEL(GCODE_SEL, SEL_KPL) << 32) | ((u_int64_t)GSEL(GUCODE32_SEL, SEL_UPL) << 48); wrmsr(MSR_STAR, msr); wrmsr(MSR_SF_MASK, PSL_NT|PSL_T|PSL_I|PSL_C|PSL_D); /* signal our startup to the BSP. */ mp_naps++; /* Spin until the BSP releases the AP's. */ while (!aps_ready) ia32_pause(); /* * On real hardware, switch to x2apic mode if possible. Do it * after aps_ready was signalled, to avoid manipulating the * mode while BSP might still want to send some IPI to us * (second startup IPI is ignored on modern hardware etc). */ lapic_xapic_mode(); /* Initialize the PAT MSR. */ pmap_init_pat(); /* set up CPU registers and state */ cpu_setregs(); /* set up SSE/NX */ initializecpu(); /* set up FPU state on the AP */ fpuinit(); if (cpu_ops.cpu_init) cpu_ops.cpu_init(); /* A quick check from sanity claus */ cpuid = PCPU_GET(cpuid); if (PCPU_GET(apic_id) != lapic_id()) { printf("SMP: cpuid = %d\n", cpuid); printf("SMP: actual apic_id = %d\n", lapic_id()); printf("SMP: correct apic_id = %d\n", PCPU_GET(apic_id)); panic("cpuid mismatch! boom!!"); } /* Initialize curthread. */ KASSERT(PCPU_GET(idlethread) != NULL, ("no idle thread")); PCPU_SET(curthread, PCPU_GET(idlethread)); mca_init(); mtx_lock_spin(&ap_boot_mtx); /* Init local apic for irq's */ lapic_setup(1); /* Set memory range attributes for this CPU to match the BSP */ mem_range_AP_init(); smp_cpus++; CTR1(KTR_SMP, "SMP: AP CPU #%d Launched", cpuid); printf("SMP: AP CPU #%d Launched!\n", cpuid); /* Determine if we are a logical CPU. */ /* XXX Calculation depends on cpu_logical being a power of 2, e.g. 2 */ if (cpu_logical > 1 && PCPU_GET(apic_id) % cpu_logical != 0) CPU_SET(cpuid, &logical_cpus_mask); if (bootverbose) lapic_dump("AP"); if (smp_cpus == mp_ncpus) { /* enable IPI's, tlb shootdown, freezes etc */ atomic_store_rel_int(&smp_started, 1); } /* * Enable global pages TLB extension * This also implicitly flushes the TLB */ load_cr4(rcr4() | CR4_PGE); if (pmap_pcid_enabled) load_cr4(rcr4() | CR4_PCIDE); load_ds(_udatasel); load_es(_udatasel); load_fs(_ufssel); mtx_unlock_spin(&ap_boot_mtx); /* Wait until all the AP's are up. */ while (smp_started == 0) ia32_pause(); /* Start per-CPU event timers. */ cpu_initclocks_ap(); sched_throw(NULL); panic("scheduler returned us to %s", __func__); /* NOTREACHED */ } /******************************************************************* * local functions and data */ /* * We tell the I/O APIC code about all the CPUs we want to receive * interrupts. If we don't want certain CPUs to receive IRQs we * can simply not tell the I/O APIC code about them in this function. * We also do not tell it about the BSP since it tells itself about * the BSP internally to work with UP kernels and on UP machines. */ static void set_interrupt_apic_ids(void) { u_int i, apic_id; for (i = 0; i < MAXCPU; i++) { apic_id = cpu_apic_ids[i]; if (apic_id == -1) continue; if (cpu_info[apic_id].cpu_bsp) continue; if (cpu_info[apic_id].cpu_disabled) continue; /* Don't let hyperthreads service interrupts. */ - if (hyperthreading_cpus > 1 && - apic_id % hyperthreading_cpus != 0) + if (cpu_logical > 1 && + apic_id % cpu_logical != 0) continue; intr_add_cpu(i); } } /* * Assign logical CPU IDs to local APICs. */ static void assign_cpu_ids(void) { u_int i; TUNABLE_INT_FETCH("machdep.hyperthreading_allowed", &hyperthreading_allowed); /* Check for explicitly disabled CPUs. */ for (i = 0; i <= MAX_APIC_ID; i++) { if (!cpu_info[i].cpu_present || cpu_info[i].cpu_bsp) continue; if (hyperthreading_cpus > 1 && i % hyperthreading_cpus != 0) { cpu_info[i].cpu_hyperthread = 1; /* * Don't use HT CPU if it has been disabled by a * tunable. */ if (hyperthreading_allowed == 0) { cpu_info[i].cpu_disabled = 1; continue; } } /* Don't use this CPU if it has been disabled by a tunable. */ if (resource_disabled("lapic", i)) { cpu_info[i].cpu_disabled = 1; continue; } } if (hyperthreading_allowed == 0 && hyperthreading_cpus > 1) { hyperthreading_cpus = 0; cpu_logical = 1; } /* * Assign CPU IDs to local APIC IDs and disable any CPUs * beyond MAXCPU. CPU 0 is always assigned to the BSP. * * To minimize confusion for userland, we attempt to number * CPUs such that all threads and cores in a package are * grouped together. For now we assume that the BSP is always * the first thread in a package and just start adding APs * starting with the BSP's APIC ID. */ mp_ncpus = 1; cpu_apic_ids[0] = boot_cpu_id; apic_cpuids[boot_cpu_id] = 0; for (i = boot_cpu_id + 1; i != boot_cpu_id; i == MAX_APIC_ID ? i = 0 : i++) { if (!cpu_info[i].cpu_present || cpu_info[i].cpu_bsp || cpu_info[i].cpu_disabled) continue; if (mp_ncpus < MAXCPU) { cpu_apic_ids[mp_ncpus] = i; apic_cpuids[i] = mp_ncpus; mp_ncpus++; } else cpu_info[i].cpu_disabled = 1; } KASSERT(mp_maxid >= mp_ncpus - 1, ("%s: counters out of sync: max %d, count %d", __func__, mp_maxid, mp_ncpus)); } /* * start each AP in our list */ int native_start_all_aps(void) { vm_offset_t va = boot_address + KERNBASE; u_int64_t *pt4, *pt3, *pt2; u_int32_t mpbioswarmvec; int apic_id, cpu, i; u_char mpbiosreason; mtx_init(&ap_boot_mtx, "ap boot", NULL, MTX_SPIN); /* install the AP 1st level boot code */ pmap_kenter(va, boot_address); pmap_invalidate_page(kernel_pmap, va); bcopy(mptramp_start, (void *)va, bootMP_size); /* Locate the page tables, they'll be below the trampoline */ pt4 = (u_int64_t *)(uintptr_t)(mptramp_pagetables + KERNBASE); pt3 = pt4 + (PAGE_SIZE) / sizeof(u_int64_t); pt2 = pt3 + (PAGE_SIZE) / sizeof(u_int64_t); /* Create the initial 1GB replicated page tables */ for (i = 0; i < 512; i++) { /* Each slot of the level 4 pages points to the same level 3 page */ pt4[i] = (u_int64_t)(uintptr_t)(mptramp_pagetables + PAGE_SIZE); pt4[i] |= PG_V | PG_RW | PG_U; /* Each slot of the level 3 pages points to the same level 2 page */ pt3[i] = (u_int64_t)(uintptr_t)(mptramp_pagetables + (2 * PAGE_SIZE)); pt3[i] |= PG_V | PG_RW | PG_U; /* The level 2 page slots are mapped with 2MB pages for 1GB. */ pt2[i] = i * (2 * 1024 * 1024); pt2[i] |= PG_V | PG_RW | PG_PS | PG_U; } /* save the current value of the warm-start vector */ mpbioswarmvec = *((u_int32_t *) WARMBOOT_OFF); outb(CMOS_REG, BIOS_RESET); mpbiosreason = inb(CMOS_DATA); /* setup a vector to our boot code */ *((volatile u_short *) WARMBOOT_OFF) = WARMBOOT_TARGET; *((volatile u_short *) WARMBOOT_SEG) = (boot_address >> 4); outb(CMOS_REG, BIOS_RESET); outb(CMOS_DATA, BIOS_WARM); /* 'warm-start' */ /* start each AP */ for (cpu = 1; cpu < mp_ncpus; cpu++) { apic_id = cpu_apic_ids[cpu]; /* allocate and set up an idle stack data page */ bootstacks[cpu] = (void *)kmem_malloc(kernel_arena, KSTACK_PAGES * PAGE_SIZE, M_WAITOK | M_ZERO); doublefault_stack = (char *)kmem_malloc(kernel_arena, PAGE_SIZE, M_WAITOK | M_ZERO); nmi_stack = (char *)kmem_malloc(kernel_arena, PAGE_SIZE, M_WAITOK | M_ZERO); dpcpu = (void *)kmem_malloc(kernel_arena, DPCPU_SIZE, M_WAITOK | M_ZERO); bootSTK = (char *)bootstacks[cpu] + KSTACK_PAGES * PAGE_SIZE - 8; bootAP = cpu; /* attempt to start the Application Processor */ if (!start_ap(apic_id)) { /* restore the warmstart vector */ *(u_int32_t *) WARMBOOT_OFF = mpbioswarmvec; panic("AP #%d (PHY# %d) failed!", cpu, apic_id); } CPU_SET(cpu, &all_cpus); /* record AP in CPU map */ } /* restore the warmstart vector */ *(u_int32_t *) WARMBOOT_OFF = mpbioswarmvec; outb(CMOS_REG, BIOS_RESET); outb(CMOS_DATA, mpbiosreason); /* number of APs actually started */ return mp_naps; } /* * This function starts the AP (application processor) identified * by the APIC ID 'physicalCpu'. It does quite a "song and dance" * to accomplish this. This is necessary because of the nuances * of the different hardware we might encounter. It isn't pretty, * but it seems to work. */ static int start_ap(int apic_id) { int vector, ms; int cpus; /* calculate the vector */ vector = (boot_address >> 12) & 0xff; /* used as a watchpoint to signal AP startup */ cpus = mp_naps; ipi_startup(apic_id, vector); /* Wait up to 5 seconds for it to start. */ for (ms = 0; ms < 5000; ms++) { if (mp_naps > cpus) return 1; /* return SUCCESS */ DELAY(1000); } return 0; /* return FAILURE */ } #ifdef COUNT_XINVLTLB_HITS u_int xhits_gbl[MAXCPU]; u_int xhits_pg[MAXCPU]; u_int xhits_rng[MAXCPU]; static SYSCTL_NODE(_debug, OID_AUTO, xhits, CTLFLAG_RW, 0, ""); SYSCTL_OPAQUE(_debug_xhits, OID_AUTO, global, CTLFLAG_RW, &xhits_gbl, sizeof(xhits_gbl), "IU", ""); SYSCTL_OPAQUE(_debug_xhits, OID_AUTO, page, CTLFLAG_RW, &xhits_pg, sizeof(xhits_pg), "IU", ""); SYSCTL_OPAQUE(_debug_xhits, OID_AUTO, range, CTLFLAG_RW, &xhits_rng, sizeof(xhits_rng), "IU", ""); u_int ipi_global; u_int ipi_page; u_int ipi_range; u_int ipi_range_size; SYSCTL_UINT(_debug_xhits, OID_AUTO, ipi_global, CTLFLAG_RW, &ipi_global, 0, ""); SYSCTL_UINT(_debug_xhits, OID_AUTO, ipi_page, CTLFLAG_RW, &ipi_page, 0, ""); SYSCTL_UINT(_debug_xhits, OID_AUTO, ipi_range, CTLFLAG_RW, &ipi_range, 0, ""); SYSCTL_UINT(_debug_xhits, OID_AUTO, ipi_range_size, CTLFLAG_RW, &ipi_range_size, 0, ""); u_int ipi_masked_global; u_int ipi_masked_page; u_int ipi_masked_range; u_int ipi_masked_range_size; SYSCTL_UINT(_debug_xhits, OID_AUTO, ipi_masked_global, CTLFLAG_RW, &ipi_masked_global, 0, ""); SYSCTL_UINT(_debug_xhits, OID_AUTO, ipi_masked_page, CTLFLAG_RW, &ipi_masked_page, 0, ""); SYSCTL_UINT(_debug_xhits, OID_AUTO, ipi_masked_range, CTLFLAG_RW, &ipi_masked_range, 0, ""); SYSCTL_UINT(_debug_xhits, OID_AUTO, ipi_masked_range_size, CTLFLAG_RW, &ipi_masked_range_size, 0, ""); #endif /* COUNT_XINVLTLB_HITS */ /* * Init and startup IPI. */ void ipi_startup(int apic_id, int vector) { /* * This attempts to follow the algorithm described in the * Intel Multiprocessor Specification v1.4 in section B.4. * For each IPI, we allow the local APIC ~20us to deliver the * IPI. If that times out, we panic. */ /* * first we do an INIT IPI: this INIT IPI might be run, resetting * and running the target CPU. OR this INIT IPI might be latched (P5 * bug), CPU waiting for STARTUP IPI. OR this INIT IPI might be * ignored. */ lapic_ipi_raw(APIC_DEST_DESTFLD | APIC_TRIGMOD_LEVEL | APIC_LEVEL_ASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_INIT, apic_id); lapic_ipi_wait(20); /* Explicitly deassert the INIT IPI. */ lapic_ipi_raw(APIC_DEST_DESTFLD | APIC_TRIGMOD_LEVEL | APIC_LEVEL_DEASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_INIT, apic_id); DELAY(10000); /* wait ~10mS */ /* * next we do a STARTUP IPI: the previous INIT IPI might still be * latched, (P5 bug) this 1st STARTUP would then terminate * immediately, and the previously started INIT IPI would continue. OR * the previous INIT IPI has already run. and this STARTUP IPI will * run. OR the previous INIT IPI was ignored. and this STARTUP IPI * will run. */ lapic_ipi_raw(APIC_DEST_DESTFLD | APIC_TRIGMOD_EDGE | APIC_LEVEL_ASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_STARTUP | vector, apic_id); if (!lapic_ipi_wait(20)) panic("Failed to deliver first STARTUP IPI to APIC %d", apic_id); DELAY(200); /* wait ~200uS */ /* * finally we do a 2nd STARTUP IPI: this 2nd STARTUP IPI should run IF * the previous STARTUP IPI was cancelled by a latched INIT IPI. OR * this STARTUP IPI will be ignored, as only ONE STARTUP IPI is * recognized after hardware RESET or INIT IPI. */ lapic_ipi_raw(APIC_DEST_DESTFLD | APIC_TRIGMOD_EDGE | APIC_LEVEL_ASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_STARTUP | vector, apic_id); if (!lapic_ipi_wait(20)) panic("Failed to deliver second STARTUP IPI to APIC %d", apic_id); DELAY(200); /* wait ~200uS */ } /* * Send an IPI to specified CPU handling the bitmap logic. */ static void ipi_send_cpu(int cpu, u_int ipi) { u_int bitmap, old_pending, new_pending; KASSERT(cpu_apic_ids[cpu] != -1, ("IPI to non-existent CPU %d", cpu)); if (IPI_IS_BITMAPED(ipi)) { bitmap = 1 << ipi; ipi = IPI_BITMAP_VECTOR; do { old_pending = cpu_ipi_pending[cpu]; new_pending = old_pending | bitmap; } while (!atomic_cmpset_int(&cpu_ipi_pending[cpu], old_pending, new_pending)); if (old_pending) return; } lapic_ipi_vectored(ipi, cpu_apic_ids[cpu]); } /* * Flush the TLB on all other CPU's */ static void smp_tlb_shootdown(u_int vector, pmap_t pmap, vm_offset_t addr1, vm_offset_t addr2) { u_int ncpu; ncpu = mp_ncpus - 1; /* does not shootdown self */ if (ncpu < 1) return; /* no other cpus */ if (!(read_rflags() & PSL_I)) panic("%s: interrupts disabled", __func__); mtx_lock_spin(&smp_ipi_mtx); smp_tlb_invpcid.addr = addr1; if (pmap == NULL) { smp_tlb_invpcid.pcid = 0; } else { smp_tlb_invpcid.pcid = pmap->pm_pcid; pcid_cr3 = pmap->pm_cr3; } smp_tlb_addr2 = addr2; smp_tlb_pmap = pmap; atomic_store_rel_int(&smp_tlb_wait, 0); ipi_all_but_self(vector); while (smp_tlb_wait < ncpu) ia32_pause(); mtx_unlock_spin(&smp_ipi_mtx); } static void smp_targeted_tlb_shootdown(cpuset_t mask, u_int vector, pmap_t pmap, vm_offset_t addr1, vm_offset_t addr2) { int cpu, ncpu, othercpus; othercpus = mp_ncpus - 1; if (CPU_ISFULLSET(&mask)) { if (othercpus < 1) return; } else { CPU_CLR(PCPU_GET(cpuid), &mask); if (CPU_EMPTY(&mask)) return; } if (!(read_rflags() & PSL_I)) panic("%s: interrupts disabled", __func__); mtx_lock_spin(&smp_ipi_mtx); smp_tlb_invpcid.addr = addr1; if (pmap == NULL) { smp_tlb_invpcid.pcid = 0; } else { smp_tlb_invpcid.pcid = pmap->pm_pcid; pcid_cr3 = pmap->pm_cr3; } smp_tlb_addr2 = addr2; smp_tlb_pmap = pmap; atomic_store_rel_int(&smp_tlb_wait, 0); if (CPU_ISFULLSET(&mask)) { ncpu = othercpus; ipi_all_but_self(vector); } else { ncpu = 0; while ((cpu = CPU_FFS(&mask)) != 0) { cpu--; CPU_CLR(cpu, &mask); CTR3(KTR_SMP, "%s: cpu: %d ipi: %x", __func__, cpu, vector); ipi_send_cpu(cpu, vector); ncpu++; } } while (smp_tlb_wait < ncpu) ia32_pause(); mtx_unlock_spin(&smp_ipi_mtx); } void smp_cache_flush(void) { if (smp_started) smp_tlb_shootdown(IPI_INVLCACHE, NULL, 0, 0); } void smp_invltlb(pmap_t pmap) { if (smp_started) { smp_tlb_shootdown(IPI_INVLTLB, pmap, 0, 0); #ifdef COUNT_XINVLTLB_HITS ipi_global++; #endif } } void smp_invlpg(pmap_t pmap, vm_offset_t addr) { if (smp_started) { smp_tlb_shootdown(IPI_INVLPG, pmap, addr, 0); #ifdef COUNT_XINVLTLB_HITS ipi_page++; #endif } } void smp_invlpg_range(pmap_t pmap, vm_offset_t addr1, vm_offset_t addr2) { if (smp_started) { smp_tlb_shootdown(IPI_INVLRNG, pmap, addr1, addr2); #ifdef COUNT_XINVLTLB_HITS ipi_range++; ipi_range_size += (addr2 - addr1) / PAGE_SIZE; #endif } } void smp_masked_invltlb(cpuset_t mask, pmap_t pmap) { if (smp_started) { smp_targeted_tlb_shootdown(mask, IPI_INVLTLB, pmap, 0, 0); #ifdef COUNT_XINVLTLB_HITS ipi_masked_global++; #endif } } void smp_masked_invlpg(cpuset_t mask, pmap_t pmap, vm_offset_t addr) { if (smp_started) { smp_targeted_tlb_shootdown(mask, IPI_INVLPG, pmap, addr, 0); #ifdef COUNT_XINVLTLB_HITS ipi_masked_page++; #endif } } void smp_masked_invlpg_range(cpuset_t mask, pmap_t pmap, vm_offset_t addr1, vm_offset_t addr2) { if (smp_started) { smp_targeted_tlb_shootdown(mask, IPI_INVLRNG, pmap, addr1, addr2); #ifdef COUNT_XINVLTLB_HITS ipi_masked_range++; ipi_masked_range_size += (addr2 - addr1) / PAGE_SIZE; #endif } } void ipi_bitmap_handler(struct trapframe frame) { struct trapframe *oldframe; struct thread *td; int cpu = PCPU_GET(cpuid); u_int ipi_bitmap; critical_enter(); td = curthread; td->td_intr_nesting_level++; oldframe = td->td_intr_frame; td->td_intr_frame = &frame; ipi_bitmap = atomic_readandclear_int(&cpu_ipi_pending[cpu]); if (ipi_bitmap & (1 << IPI_PREEMPT)) { #ifdef COUNT_IPIS (*ipi_preempt_counts[cpu])++; #endif sched_preempt(td); } if (ipi_bitmap & (1 << IPI_AST)) { #ifdef COUNT_IPIS (*ipi_ast_counts[cpu])++; #endif /* Nothing to do for AST */ } if (ipi_bitmap & (1 << IPI_HARDCLOCK)) { #ifdef COUNT_IPIS (*ipi_hardclock_counts[cpu])++; #endif hardclockintr(); } td->td_intr_frame = oldframe; td->td_intr_nesting_level--; critical_exit(); } /* * send an IPI to a set of cpus. */ void ipi_selected(cpuset_t cpus, u_int ipi) { int cpu; /* * IPI_STOP_HARD maps to a NMI and the trap handler needs a bit * of help in order to understand what is the source. * Set the mask of receiving CPUs for this purpose. */ if (ipi == IPI_STOP_HARD) CPU_OR_ATOMIC(&ipi_nmi_pending, &cpus); while ((cpu = CPU_FFS(&cpus)) != 0) { cpu--; CPU_CLR(cpu, &cpus); CTR3(KTR_SMP, "%s: cpu: %d ipi: %x", __func__, cpu, ipi); ipi_send_cpu(cpu, ipi); } } /* * send an IPI to a specific CPU. */ void ipi_cpu(int cpu, u_int ipi) { /* * IPI_STOP_HARD maps to a NMI and the trap handler needs a bit * of help in order to understand what is the source. * Set the mask of receiving CPUs for this purpose. */ if (ipi == IPI_STOP_HARD) CPU_SET_ATOMIC(cpu, &ipi_nmi_pending); CTR3(KTR_SMP, "%s: cpu: %d ipi: %x", __func__, cpu, ipi); ipi_send_cpu(cpu, ipi); } /* * send an IPI to all CPUs EXCEPT myself */ void ipi_all_but_self(u_int ipi) { cpuset_t other_cpus; other_cpus = all_cpus; CPU_CLR(PCPU_GET(cpuid), &other_cpus); if (IPI_IS_BITMAPED(ipi)) { ipi_selected(other_cpus, ipi); return; } /* * IPI_STOP_HARD maps to a NMI and the trap handler needs a bit * of help in order to understand what is the source. * Set the mask of receiving CPUs for this purpose. */ if (ipi == IPI_STOP_HARD) CPU_OR_ATOMIC(&ipi_nmi_pending, &other_cpus); CTR2(KTR_SMP, "%s: ipi: %x", __func__, ipi); lapic_ipi_vectored(ipi, APIC_IPI_DEST_OTHERS); } int ipi_nmi_handler() { u_int cpuid; /* * As long as there is not a simple way to know about a NMI's * source, if the bitmask for the current CPU is present in * the global pending bitword an IPI_STOP_HARD has been issued * and should be handled. */ cpuid = PCPU_GET(cpuid); if (!CPU_ISSET(cpuid, &ipi_nmi_pending)) return (1); CPU_CLR_ATOMIC(cpuid, &ipi_nmi_pending); cpustop_handler(); return (0); } /* * Handle an IPI_STOP by saving our current context and spinning until we * are resumed. */ void cpustop_handler(void) { u_int cpu; cpu = PCPU_GET(cpuid); savectx(&stoppcbs[cpu]); /* Indicate that we are stopped */ CPU_SET_ATOMIC(cpu, &stopped_cpus); /* Wait for restart */ while (!CPU_ISSET(cpu, &started_cpus)) ia32_pause(); CPU_CLR_ATOMIC(cpu, &started_cpus); CPU_CLR_ATOMIC(cpu, &stopped_cpus); #ifdef DDB amd64_db_resume_dbreg(); #endif if (cpu == 0 && cpustop_restartfunc != NULL) { cpustop_restartfunc(); cpustop_restartfunc = NULL; } } /* * Handle an IPI_SUSPEND by saving our current context and spinning until we * are resumed. */ void cpususpend_handler(void) { u_int cpu; mtx_assert(&smp_ipi_mtx, MA_NOTOWNED); cpu = PCPU_GET(cpuid); if (savectx(&susppcbs[cpu]->sp_pcb)) { fpususpend(susppcbs[cpu]->sp_fpususpend); wbinvd(); CPU_SET_ATOMIC(cpu, &suspended_cpus); } else { fpuresume(susppcbs[cpu]->sp_fpususpend); pmap_init_pat(); initializecpu(); PCPU_SET(switchtime, 0); PCPU_SET(switchticks, ticks); /* Indicate that we are resumed */ CPU_CLR_ATOMIC(cpu, &suspended_cpus); } /* Wait for resume */ while (!CPU_ISSET(cpu, &started_cpus)) ia32_pause(); if (cpu_ops.cpu_resume) cpu_ops.cpu_resume(); if (vmm_resume_p) vmm_resume_p(); /* Resume MCA and local APIC */ lapic_xapic_mode(); mca_resume(); lapic_setup(0); CPU_CLR_ATOMIC(cpu, &started_cpus); /* Indicate that we are resumed */ CPU_CLR_ATOMIC(cpu, &suspended_cpus); } /* * Handlers for TLB related IPIs */ void invltlb_handler(void) { #ifdef COUNT_XINVLTLB_HITS xhits_gbl[PCPU_GET(cpuid)]++; #endif /* COUNT_XINVLTLB_HITS */ #ifdef COUNT_IPIS (*ipi_invltlb_counts[PCPU_GET(cpuid)])++; #endif /* COUNT_IPIS */ invltlb(); atomic_add_int(&smp_tlb_wait, 1); } void invltlb_pcid_handler(void) { uint64_t cr3; u_int cpuid; #ifdef COUNT_XINVLTLB_HITS xhits_gbl[PCPU_GET(cpuid)]++; #endif /* COUNT_XINVLTLB_HITS */ #ifdef COUNT_IPIS (*ipi_invltlb_counts[PCPU_GET(cpuid)])++; #endif /* COUNT_IPIS */ if (smp_tlb_invpcid.pcid != (uint64_t)-1 && smp_tlb_invpcid.pcid != 0) { if (invpcid_works) { invpcid(&smp_tlb_invpcid, INVPCID_CTX); } else { /* Otherwise reload %cr3 twice. */ cr3 = rcr3(); if (cr3 != pcid_cr3) { load_cr3(pcid_cr3); cr3 |= CR3_PCID_SAVE; } load_cr3(cr3); } } else { invltlb_globpcid(); } if (smp_tlb_pmap != NULL) { cpuid = PCPU_GET(cpuid); if (!CPU_ISSET(cpuid, &smp_tlb_pmap->pm_active)) CPU_CLR_ATOMIC(cpuid, &smp_tlb_pmap->pm_save); } atomic_add_int(&smp_tlb_wait, 1); } void invlpg_handler(void) { #ifdef COUNT_XINVLTLB_HITS xhits_pg[PCPU_GET(cpuid)]++; #endif /* COUNT_XINVLTLB_HITS */ #ifdef COUNT_IPIS (*ipi_invlpg_counts[PCPU_GET(cpuid)])++; #endif /* COUNT_IPIS */ invlpg(smp_tlb_invpcid.addr); atomic_add_int(&smp_tlb_wait, 1); } void invlpg_pcid_handler(void) { uint64_t cr3; #ifdef COUNT_XINVLTLB_HITS xhits_pg[PCPU_GET(cpuid)]++; #endif /* COUNT_XINVLTLB_HITS */ #ifdef COUNT_IPIS (*ipi_invlpg_counts[PCPU_GET(cpuid)])++; #endif /* COUNT_IPIS */ if (smp_tlb_invpcid.pcid == (uint64_t)-1) { invltlb_globpcid(); } else if (smp_tlb_invpcid.pcid == 0) { invlpg(smp_tlb_invpcid.addr); } else if (invpcid_works) { invpcid(&smp_tlb_invpcid, INVPCID_ADDR); } else { /* * PCID supported, but INVPCID is not. * Temporarily switch to the target address * space and do INVLPG. */ cr3 = rcr3(); if (cr3 != pcid_cr3) load_cr3(pcid_cr3 | CR3_PCID_SAVE); invlpg(smp_tlb_invpcid.addr); load_cr3(cr3 | CR3_PCID_SAVE); } atomic_add_int(&smp_tlb_wait, 1); } static inline void invlpg_range(vm_offset_t start, vm_offset_t end) { do { invlpg(start); start += PAGE_SIZE; } while (start < end); } void invlrng_handler(void) { struct invpcid_descr d; vm_offset_t addr; uint64_t cr3; u_int cpuid; #ifdef COUNT_XINVLTLB_HITS xhits_rng[PCPU_GET(cpuid)]++; #endif /* COUNT_XINVLTLB_HITS */ #ifdef COUNT_IPIS (*ipi_invlrng_counts[PCPU_GET(cpuid)])++; #endif /* COUNT_IPIS */ addr = smp_tlb_invpcid.addr; if (pmap_pcid_enabled) { if (smp_tlb_invpcid.pcid == 0) { /* * kernel pmap - use invlpg to invalidate * global mapping. */ invlpg_range(addr, smp_tlb_addr2); } else if (smp_tlb_invpcid.pcid == (uint64_t)-1) { invltlb_globpcid(); if (smp_tlb_pmap != NULL) { cpuid = PCPU_GET(cpuid); if (!CPU_ISSET(cpuid, &smp_tlb_pmap->pm_active)) CPU_CLR_ATOMIC(cpuid, &smp_tlb_pmap->pm_save); } } else if (invpcid_works) { d = smp_tlb_invpcid; do { invpcid(&d, INVPCID_ADDR); d.addr += PAGE_SIZE; } while (d.addr <= smp_tlb_addr2); } else { cr3 = rcr3(); if (cr3 != pcid_cr3) load_cr3(pcid_cr3 | CR3_PCID_SAVE); invlpg_range(addr, smp_tlb_addr2); load_cr3(cr3 | CR3_PCID_SAVE); } } else { invlpg_range(addr, smp_tlb_addr2); } atomic_add_int(&smp_tlb_wait, 1); } void invlcache_handler(void) { #ifdef COUNT_IPIS (*ipi_invlcache_counts[PCPU_GET(cpuid)])++; #endif /* COUNT_IPIS */ wbinvd(); atomic_add_int(&smp_tlb_wait, 1); } /* * This is called once the rest of the system is up and running and we're * ready to let the AP's out of the pen. */ static void release_aps(void *dummy __unused) { if (mp_ncpus == 1) return; atomic_store_rel_int(&aps_ready, 1); while (smp_started == 0) ia32_pause(); } SYSINIT(start_aps, SI_SUB_SMP, SI_ORDER_FIRST, release_aps, NULL); #ifdef COUNT_IPIS /* * Setup interrupt counters for IPI handlers. */ static void mp_ipi_intrcnt(void *dummy) { char buf[64]; int i; CPU_FOREACH(i) { snprintf(buf, sizeof(buf), "cpu%d:invltlb", i); intrcnt_add(buf, &ipi_invltlb_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:invlrng", i); intrcnt_add(buf, &ipi_invlrng_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:invlpg", i); intrcnt_add(buf, &ipi_invlpg_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:invlcache", i); intrcnt_add(buf, &ipi_invlcache_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:preempt", i); intrcnt_add(buf, &ipi_preempt_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:ast", i); intrcnt_add(buf, &ipi_ast_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:rendezvous", i); intrcnt_add(buf, &ipi_rendezvous_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:hardclock", i); intrcnt_add(buf, &ipi_hardclock_counts[i]); } } SYSINIT(mp_ipi_intrcnt, SI_SUB_INTR, SI_ORDER_MIDDLE, mp_ipi_intrcnt, NULL); #endif Index: projects/cxl_iscsi/sys/amd64/vmm/io/vatpic.c =================================================================== --- projects/cxl_iscsi/sys/amd64/vmm/io/vatpic.c (revision 279716) +++ projects/cxl_iscsi/sys/amd64/vmm/io/vatpic.c (revision 279717) @@ -1,808 +1,809 @@ /*- * Copyright (c) 2014 Tycho Nightingale * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include "vmm_ktr.h" #include "vmm_lapic.h" #include "vioapic.h" #include "vatpic.h" static MALLOC_DEFINE(M_VATPIC, "atpic", "bhyve virtual atpic (8259)"); #define VATPIC_LOCK(vatpic) mtx_lock_spin(&((vatpic)->mtx)) #define VATPIC_UNLOCK(vatpic) mtx_unlock_spin(&((vatpic)->mtx)) #define VATPIC_LOCKED(vatpic) mtx_owned(&((vatpic)->mtx)) enum irqstate { IRQSTATE_ASSERT, IRQSTATE_DEASSERT, IRQSTATE_PULSE }; struct atpic { bool ready; int icw_num; int rd_cmd_reg; bool aeoi; bool poll; bool rotate; bool sfn; /* special fully-nested mode */ int irq_base; uint8_t request; /* Interrupt Request Register (IIR) */ uint8_t service; /* Interrupt Service (ISR) */ uint8_t mask; /* Interrupt Mask Register (IMR) */ uint8_t smm; /* special mask mode */ int acnt[8]; /* sum of pin asserts and deasserts */ int lowprio; /* lowest priority irq */ bool intr_raised; }; struct vatpic { struct vm *vm; struct mtx mtx; struct atpic atpic[2]; uint8_t elc[2]; }; #define VATPIC_CTR0(vatpic, fmt) \ VM_CTR0((vatpic)->vm, fmt) #define VATPIC_CTR1(vatpic, fmt, a1) \ VM_CTR1((vatpic)->vm, fmt, a1) #define VATPIC_CTR2(vatpic, fmt, a1, a2) \ VM_CTR2((vatpic)->vm, fmt, a1, a2) #define VATPIC_CTR3(vatpic, fmt, a1, a2, a3) \ VM_CTR3((vatpic)->vm, fmt, a1, a2, a3) #define VATPIC_CTR4(vatpic, fmt, a1, a2, a3, a4) \ VM_CTR4((vatpic)->vm, fmt, a1, a2, a3, a4) /* * Loop over all the pins in priority order from highest to lowest. */ #define ATPIC_PIN_FOREACH(pinvar, atpic, tmpvar) \ for (tmpvar = 0, pinvar = (atpic->lowprio + 1) & 0x7; \ tmpvar < 8; \ tmpvar++, pinvar = (pinvar + 1) & 0x7) static void vatpic_set_pinstate(struct vatpic *vatpic, int pin, bool newstate); static __inline bool master_atpic(struct vatpic *vatpic, struct atpic *atpic) { if (atpic == &vatpic->atpic[0]) return (true); else return (false); } static __inline int vatpic_get_highest_isrpin(struct atpic *atpic) { int bit, pin; int i; ATPIC_PIN_FOREACH(pin, atpic, i) { bit = (1 << pin); if (atpic->service & bit) { /* * An IS bit that is masked by an IMR bit will not be * cleared by a non-specific EOI in Special Mask Mode. */ if (atpic->smm && (atpic->mask & bit) != 0) continue; else return (pin); } } return (-1); } static __inline int vatpic_get_highest_irrpin(struct atpic *atpic) { int serviced; int bit, pin, tmp; /* * In 'Special Fully-Nested Mode' when an interrupt request from * a slave is in service, the slave is not locked out from the * master's priority logic. */ serviced = atpic->service; if (atpic->sfn) serviced &= ~(1 << 2); /* * In 'Special Mask Mode', when a mask bit is set in OCW1 it inhibits * further interrupts at that level and enables interrupts from all * other levels that are not masked. In other words the ISR has no * bearing on the levels that can generate interrupts. */ if (atpic->smm) serviced = 0; ATPIC_PIN_FOREACH(pin, atpic, tmp) { bit = 1 << pin; /* * If there is already an interrupt in service at the same * or higher priority then bail. */ if ((serviced & bit) != 0) break; /* * If an interrupt is asserted and not masked then return * the corresponding 'pin' to the caller. */ if ((atpic->request & bit) != 0 && (atpic->mask & bit) == 0) return (pin); } return (-1); } static void vatpic_notify_intr(struct vatpic *vatpic) { struct atpic *atpic; int pin; KASSERT(VATPIC_LOCKED(vatpic), ("vatpic_notify_intr not locked")); /* * First check the slave. */ atpic = &vatpic->atpic[1]; if (!atpic->intr_raised && (pin = vatpic_get_highest_irrpin(atpic)) != -1) { VATPIC_CTR4(vatpic, "atpic slave notify pin = %d " "(imr 0x%x irr 0x%x isr 0x%x)", pin, atpic->mask, atpic->request, atpic->service); /* * Cascade the request from the slave to the master. */ atpic->intr_raised = true; vatpic_set_pinstate(vatpic, 2, true); vatpic_set_pinstate(vatpic, 2, false); } else { VATPIC_CTR3(vatpic, "atpic slave no eligible interrupts " "(imr 0x%x irr 0x%x isr 0x%x)", atpic->mask, atpic->request, atpic->service); } /* * Then check the master. */ atpic = &vatpic->atpic[0]; if (!atpic->intr_raised && (pin = vatpic_get_highest_irrpin(atpic)) != -1) { VATPIC_CTR4(vatpic, "atpic master notify pin = %d " "(imr 0x%x irr 0x%x isr 0x%x)", pin, atpic->mask, atpic->request, atpic->service); /* * From Section 3.6.2, "Interrupt Modes", in the * MPtable Specification, Version 1.4 * * PIC interrupts are routed to both the Local APIC * and the I/O APIC to support operation in 1 of 3 * modes. * * 1. Legacy PIC Mode: the PIC effectively bypasses * all APIC components. In this mode the local APIC is * disabled and LINT0 is reconfigured as INTR to * deliver the PIC interrupt directly to the CPU. * * 2. Virtual Wire Mode: the APIC is treated as a * virtual wire which delivers interrupts from the PIC * to the CPU. In this mode LINT0 is programmed as * ExtINT to indicate that the PIC is the source of * the interrupt. * * 3. Virtual Wire Mode via I/O APIC: PIC interrupts are * fielded by the I/O APIC and delivered to the appropriate * CPU. In this mode the I/O APIC input 0 is programmed * as ExtINT to indicate that the PIC is the source of the * interrupt. */ atpic->intr_raised = true; lapic_set_local_intr(vatpic->vm, -1, APIC_LVT_LINT0); vioapic_pulse_irq(vatpic->vm, 0); } else { VATPIC_CTR3(vatpic, "atpic master no eligible interrupts " "(imr 0x%x irr 0x%x isr 0x%x)", atpic->mask, atpic->request, atpic->service); } } static int vatpic_icw1(struct vatpic *vatpic, struct atpic *atpic, uint8_t val) { VATPIC_CTR1(vatpic, "atpic icw1 0x%x", val); atpic->ready = false; atpic->icw_num = 1; + atpic->request = 0; atpic->mask = 0; atpic->lowprio = 7; atpic->rd_cmd_reg = 0; atpic->poll = 0; atpic->smm = 0; if ((val & ICW1_SNGL) != 0) { VATPIC_CTR0(vatpic, "vatpic cascade mode required"); return (-1); } if ((val & ICW1_IC4) == 0) { VATPIC_CTR0(vatpic, "vatpic icw4 required"); return (-1); } atpic->icw_num++; return (0); } static int vatpic_icw2(struct vatpic *vatpic, struct atpic *atpic, uint8_t val) { VATPIC_CTR1(vatpic, "atpic icw2 0x%x", val); atpic->irq_base = val & 0xf8; atpic->icw_num++; return (0); } static int vatpic_icw3(struct vatpic *vatpic, struct atpic *atpic, uint8_t val) { VATPIC_CTR1(vatpic, "atpic icw3 0x%x", val); atpic->icw_num++; return (0); } static int vatpic_icw4(struct vatpic *vatpic, struct atpic *atpic, uint8_t val) { VATPIC_CTR1(vatpic, "atpic icw4 0x%x", val); if ((val & ICW4_8086) == 0) { VATPIC_CTR0(vatpic, "vatpic microprocessor mode required"); return (-1); } if ((val & ICW4_AEOI) != 0) atpic->aeoi = true; if ((val & ICW4_SFNM) != 0) { if (master_atpic(vatpic, atpic)) { atpic->sfn = true; } else { VATPIC_CTR1(vatpic, "Ignoring special fully nested " "mode on slave atpic: %#x", val); } } atpic->icw_num = 0; atpic->ready = true; return (0); } static int vatpic_ocw1(struct vatpic *vatpic, struct atpic *atpic, uint8_t val) { VATPIC_CTR1(vatpic, "atpic ocw1 0x%x", val); atpic->mask = val & 0xff; return (0); } static int vatpic_ocw2(struct vatpic *vatpic, struct atpic *atpic, uint8_t val) { VATPIC_CTR1(vatpic, "atpic ocw2 0x%x", val); atpic->rotate = ((val & OCW2_R) != 0); if ((val & OCW2_EOI) != 0) { int isr_bit; if ((val & OCW2_SL) != 0) { /* specific EOI */ isr_bit = val & 0x7; } else { /* non-specific EOI */ isr_bit = vatpic_get_highest_isrpin(atpic); } if (isr_bit != -1) { atpic->service &= ~(1 << isr_bit); if (atpic->rotate) atpic->lowprio = isr_bit; } } else if ((val & OCW2_SL) != 0 && atpic->rotate == true) { /* specific priority */ atpic->lowprio = val & 0x7; } return (0); } static int vatpic_ocw3(struct vatpic *vatpic, struct atpic *atpic, uint8_t val) { VATPIC_CTR1(vatpic, "atpic ocw3 0x%x", val); if (val & OCW3_ESMM) { atpic->smm = val & OCW3_SMM ? 1 : 0; VATPIC_CTR2(vatpic, "%s atpic special mask mode %s", master_atpic(vatpic, atpic) ? "master" : "slave", atpic->smm ? "enabled" : "disabled"); } if (val & OCW3_RR) { /* read register command */ atpic->rd_cmd_reg = val & OCW3_RIS; /* Polling mode */ atpic->poll = ((val & OCW3_P) != 0); } return (0); } static void vatpic_set_pinstate(struct vatpic *vatpic, int pin, bool newstate) { struct atpic *atpic; int oldcnt, newcnt; bool level; KASSERT(pin >= 0 && pin < 16, ("vatpic_set_pinstate: invalid pin number %d", pin)); KASSERT(VATPIC_LOCKED(vatpic), ("vatpic_set_pinstate: vatpic is not locked")); atpic = &vatpic->atpic[pin >> 3]; oldcnt = atpic->acnt[pin & 0x7]; if (newstate) atpic->acnt[pin & 0x7]++; else atpic->acnt[pin & 0x7]--; newcnt = atpic->acnt[pin & 0x7]; if (newcnt < 0) { VATPIC_CTR2(vatpic, "atpic pin%d: bad acnt %d", pin, newcnt); } level = ((vatpic->elc[pin >> 3] & (1 << (pin & 0x7))) != 0); if ((oldcnt == 0 && newcnt == 1) || (newcnt > 0 && level == true)) { /* rising edge or level */ VATPIC_CTR1(vatpic, "atpic pin%d: asserted", pin); atpic->request |= (1 << (pin & 0x7)); } else if (oldcnt == 1 && newcnt == 0) { /* falling edge */ VATPIC_CTR1(vatpic, "atpic pin%d: deasserted", pin); if (level) atpic->request &= ~(1 << (pin & 0x7)); } else { VATPIC_CTR3(vatpic, "atpic pin%d: %s, ignored, acnt %d", pin, newstate ? "asserted" : "deasserted", newcnt); } vatpic_notify_intr(vatpic); } static int vatpic_set_irqstate(struct vm *vm, int irq, enum irqstate irqstate) { struct vatpic *vatpic; struct atpic *atpic; if (irq < 0 || irq > 15) return (EINVAL); vatpic = vm_atpic(vm); atpic = &vatpic->atpic[irq >> 3]; if (atpic->ready == false) return (0); VATPIC_LOCK(vatpic); switch (irqstate) { case IRQSTATE_ASSERT: vatpic_set_pinstate(vatpic, irq, true); break; case IRQSTATE_DEASSERT: vatpic_set_pinstate(vatpic, irq, false); break; case IRQSTATE_PULSE: vatpic_set_pinstate(vatpic, irq, true); vatpic_set_pinstate(vatpic, irq, false); break; default: panic("vatpic_set_irqstate: invalid irqstate %d", irqstate); } VATPIC_UNLOCK(vatpic); return (0); } int vatpic_assert_irq(struct vm *vm, int irq) { return (vatpic_set_irqstate(vm, irq, IRQSTATE_ASSERT)); } int vatpic_deassert_irq(struct vm *vm, int irq) { return (vatpic_set_irqstate(vm, irq, IRQSTATE_DEASSERT)); } int vatpic_pulse_irq(struct vm *vm, int irq) { return (vatpic_set_irqstate(vm, irq, IRQSTATE_PULSE)); } int vatpic_set_irq_trigger(struct vm *vm, int irq, enum vm_intr_trigger trigger) { struct vatpic *vatpic; if (irq < 0 || irq > 15) return (EINVAL); /* * See comment in vatpic_elc_handler. These IRQs must be * edge triggered. */ if (trigger == LEVEL_TRIGGER) { switch (irq) { case 0: case 1: case 2: case 8: case 13: return (EINVAL); } } vatpic = vm_atpic(vm); VATPIC_LOCK(vatpic); if (trigger == LEVEL_TRIGGER) vatpic->elc[irq >> 3] |= 1 << (irq & 0x7); else vatpic->elc[irq >> 3] &= ~(1 << (irq & 0x7)); VATPIC_UNLOCK(vatpic); return (0); } void vatpic_pending_intr(struct vm *vm, int *vecptr) { struct vatpic *vatpic; struct atpic *atpic; int pin; vatpic = vm_atpic(vm); atpic = &vatpic->atpic[0]; VATPIC_LOCK(vatpic); pin = vatpic_get_highest_irrpin(atpic); if (pin == 2) { atpic = &vatpic->atpic[1]; pin = vatpic_get_highest_irrpin(atpic); } /* * If there are no pins active at this moment then return the spurious * interrupt vector instead. */ if (pin == -1) pin = 7; KASSERT(pin >= 0 && pin <= 7, ("%s: invalid pin %d", __func__, pin)); *vecptr = atpic->irq_base + pin; VATPIC_UNLOCK(vatpic); } static void vatpic_pin_accepted(struct atpic *atpic, int pin) { atpic->intr_raised = false; if (atpic->acnt[pin] == 0) atpic->request &= ~(1 << pin); if (atpic->aeoi == true) { if (atpic->rotate == true) atpic->lowprio = pin; } else { atpic->service |= (1 << pin); } } void vatpic_intr_accepted(struct vm *vm, int vector) { struct vatpic *vatpic; int pin; vatpic = vm_atpic(vm); VATPIC_LOCK(vatpic); pin = vector & 0x7; if ((vector & ~0x7) == vatpic->atpic[1].irq_base) { vatpic_pin_accepted(&vatpic->atpic[1], pin); /* * If this vector originated from the slave, * accept the cascaded interrupt too. */ vatpic_pin_accepted(&vatpic->atpic[0], 2); } else { vatpic_pin_accepted(&vatpic->atpic[0], pin); } vatpic_notify_intr(vatpic); VATPIC_UNLOCK(vatpic); } static int vatpic_read(struct vatpic *vatpic, struct atpic *atpic, bool in, int port, int bytes, uint32_t *eax) { int pin; VATPIC_LOCK(vatpic); if (atpic->poll) { atpic->poll = 0; pin = vatpic_get_highest_irrpin(atpic); if (pin >= 0) { vatpic_pin_accepted(atpic, pin); *eax = 0x80 | pin; } else { *eax = 0; } } else { if (port & ICU_IMR_OFFSET) { /* read interrrupt mask register */ *eax = atpic->mask; } else { if (atpic->rd_cmd_reg == OCW3_RIS) { /* read interrupt service register */ *eax = atpic->service; } else { /* read interrupt request register */ *eax = atpic->request; } } } VATPIC_UNLOCK(vatpic); return (0); } static int vatpic_write(struct vatpic *vatpic, struct atpic *atpic, bool in, int port, int bytes, uint32_t *eax) { int error; uint8_t val; error = 0; val = *eax; VATPIC_LOCK(vatpic); if (port & ICU_IMR_OFFSET) { switch (atpic->icw_num) { case 2: error = vatpic_icw2(vatpic, atpic, val); break; case 3: error = vatpic_icw3(vatpic, atpic, val); break; case 4: error = vatpic_icw4(vatpic, atpic, val); break; default: error = vatpic_ocw1(vatpic, atpic, val); break; } } else { if (val & (1 << 4)) error = vatpic_icw1(vatpic, atpic, val); if (atpic->ready) { if (val & (1 << 3)) error = vatpic_ocw3(vatpic, atpic, val); else error = vatpic_ocw2(vatpic, atpic, val); } } if (atpic->ready) vatpic_notify_intr(vatpic); VATPIC_UNLOCK(vatpic); return (error); } int vatpic_master_handler(struct vm *vm, int vcpuid, bool in, int port, int bytes, uint32_t *eax) { struct vatpic *vatpic; struct atpic *atpic; vatpic = vm_atpic(vm); atpic = &vatpic->atpic[0]; if (bytes != 1) return (-1); if (in) { return (vatpic_read(vatpic, atpic, in, port, bytes, eax)); } return (vatpic_write(vatpic, atpic, in, port, bytes, eax)); } int vatpic_slave_handler(struct vm *vm, int vcpuid, bool in, int port, int bytes, uint32_t *eax) { struct vatpic *vatpic; struct atpic *atpic; vatpic = vm_atpic(vm); atpic = &vatpic->atpic[1]; if (bytes != 1) return (-1); if (in) { return (vatpic_read(vatpic, atpic, in, port, bytes, eax)); } return (vatpic_write(vatpic, atpic, in, port, bytes, eax)); } int vatpic_elc_handler(struct vm *vm, int vcpuid, bool in, int port, int bytes, uint32_t *eax) { struct vatpic *vatpic; bool is_master; vatpic = vm_atpic(vm); is_master = (port == IO_ELCR1); if (bytes != 1) return (-1); VATPIC_LOCK(vatpic); if (in) { if (is_master) *eax = vatpic->elc[0]; else *eax = vatpic->elc[1]; } else { /* * For the master PIC the cascade channel (IRQ2), the * heart beat timer (IRQ0), and the keyboard * controller (IRQ1) cannot be programmed for level * mode. * * For the slave PIC the real time clock (IRQ8) and * the floating point error interrupt (IRQ13) cannot * be programmed for level mode. */ if (is_master) vatpic->elc[0] = (*eax & 0xf8); else vatpic->elc[1] = (*eax & 0xde); } VATPIC_UNLOCK(vatpic); return (0); } struct vatpic * vatpic_init(struct vm *vm) { struct vatpic *vatpic; vatpic = malloc(sizeof(struct vatpic), M_VATPIC, M_WAITOK | M_ZERO); vatpic->vm = vm; mtx_init(&vatpic->mtx, "vatpic lock", NULL, MTX_SPIN); return (vatpic); } void vatpic_cleanup(struct vatpic *vatpic) { free(vatpic, M_VATPIC); } Index: projects/cxl_iscsi/sys/amd64/vmm =================================================================== --- projects/cxl_iscsi/sys/amd64/vmm (revision 279716) +++ projects/cxl_iscsi/sys/amd64/vmm (revision 279717) Property changes on: projects/cxl_iscsi/sys/amd64/vmm ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/amd64/vmm:r279555-279716 Index: projects/cxl_iscsi/sys/arm/arm/db_trace.c =================================================================== --- projects/cxl_iscsi/sys/arm/arm/db_trace.c (revision 279716) +++ projects/cxl_iscsi/sys/arm/arm/db_trace.c (revision 279717) @@ -1,183 +1,183 @@ /* $NetBSD: db_trace.c,v 1.8 2003/01/17 22:28:48 thorpej Exp $ */ /*- * Copyright (c) 2000, 2001 Ben Harris * Copyright (c) 1996 Scott K. Stevens * * Mach Operating System * Copyright (c) 1991,1990 Carnegie Mellon University * All Rights Reserved. * * Permission to use, copy, modify and distribute this software and its * documentation is hereby granted, provided that both the copyright * notice and this permission notice appear in all copies of the * software, derivative works or modified versions, and any portions * thereof, and that both notices appear in supporting documentation. * * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND FOR * ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. * * Carnegie Mellon requests users of this software to return to * * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU * School of Computer Science * Carnegie Mellon University * Pittsburgh PA 15213-3890 * * any improvements or extensions that they make and grant Carnegie the * rights to redistribute these changes. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include static void db_stack_trace_cmd(struct unwind_state *state) { const char *name; db_expr_t value; db_expr_t offset; c_db_sym_t sym; u_int reg, i; char *sep; uint16_t upd_mask; bool finished; finished = false; while (!finished) { - finished = unwind_stack_one(state, 0); + finished = unwind_stack_one(state, 1); /* Print the frame details */ sym = db_search_symbol(state->start_pc, DB_STGY_ANY, &offset); if (sym == C_DB_SYM_NULL) { value = 0; name = "(null)"; } else db_symbol_values(sym, &name, &value); db_printf("%s() at ", name); db_printsym(state->start_pc, DB_STGY_PROC); db_printf("\n"); db_printf("\t pc = 0x%08x lr = 0x%08x (", state->start_pc, state->registers[LR]); db_printsym(state->registers[LR], DB_STGY_PROC); db_printf(")\n"); db_printf("\t sp = 0x%08x fp = 0x%08x", state->registers[SP], state->registers[FP]); /* Don't print the registers we have already printed */ upd_mask = state->update_mask & ~((1 << SP) | (1 << FP) | (1 << LR) | (1 << PC)); sep = "\n\t"; for (i = 0, reg = 0; upd_mask != 0; upd_mask >>= 1, reg++) { if ((upd_mask & 1) != 0) { db_printf("%s%sr%d = 0x%08x", sep, (reg < 10) ? " " : "", reg, state->registers[reg]); i++; if (i == 2) { sep = "\n\t"; i = 0; } else sep = " "; } } db_printf("\n"); if (finished) break; /* * Stop if directed to do so, or if we've unwound back to the * kernel entry point, or if the unwind function didn't change * anything (to avoid getting stuck in this loop forever). * If the latter happens, it's an indication that the unwind * information is incorrect somehow for the function named in * the last frame printed before you see the unwind failure * message (maybe it needs a STOP_UNWINDING). */ if (state->registers[PC] < VM_MIN_KERNEL_ADDRESS) { db_printf("Unable to unwind into user mode\n"); finished = true; } else if (state->update_mask == 0) { db_printf("Unwind failure (no registers changed)\n"); finished = true; } } } /* XXX stubs */ void db_md_list_watchpoints() { } int db_md_clr_watchpoint(db_expr_t addr, db_expr_t size) { return (0); } int db_md_set_watchpoint(db_expr_t addr, db_expr_t size) { return (0); } int db_trace_thread(struct thread *thr, int count) { struct unwind_state state; struct pcb *ctx; if (thr != curthread) { ctx = kdb_thr_ctx(thr); state.registers[FP] = ctx->pcb_regs.sf_r11; state.registers[SP] = ctx->pcb_regs.sf_sp; state.registers[LR] = ctx->pcb_regs.sf_lr; state.registers[PC] = ctx->pcb_regs.sf_pc; db_stack_trace_cmd(&state); } else db_trace_self(); return (0); } void db_trace_self(void) { struct unwind_state state; uint32_t sp; /* Read the stack pointer */ __asm __volatile("mov %0, sp" : "=&r" (sp)); state.registers[FP] = (uint32_t)__builtin_frame_address(0); state.registers[SP] = sp; state.registers[LR] = (uint32_t)__builtin_return_address(0); state.registers[PC] = (uint32_t)db_trace_self; db_stack_trace_cmd(&state); } Index: projects/cxl_iscsi/sys/arm/arm/exception.S =================================================================== --- projects/cxl_iscsi/sys/arm/arm/exception.S (revision 279716) +++ projects/cxl_iscsi/sys/arm/arm/exception.S (revision 279717) @@ -1,480 +1,477 @@ /* $NetBSD: exception.S,v 1.13 2003/10/31 16:30:15 scw Exp $ */ /*- * Copyright (c) 1994-1997 Mark Brinicombe. * Copyright (c) 1994 Brini. * All rights reserved. * * This code is derived from software written for Brini by Mark Brinicombe * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by Brini. * 4. The name of the company nor the name of the author may be used to * endorse or promote products derived from this software without specific * prior written permission. * * THIS SOFTWARE IS PROVIDED BY BRINI ``AS IS'' AND ANY EXPRESS OR IMPLIED * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL BRINI OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * RiscBSD kernel project * * exception.S * * Low level handlers for exception vectors * * Created : 24/09/94 * * Based on kate/display/abort.s * */ #include "assym.s" #include "opt_kdtrace.h" #include #include #include __FBSDID("$FreeBSD$"); #ifdef KDTRACE_HOOKS .bss .align 4 - .global _C_LABEL(dtrace_invop_jump_addr) -_C_LABEL(dtrace_invop_jump_addr): - .word 0 - .word 0 - .global _C_LABEL(dtrace_invop_calltrap_addr) _C_LABEL(dtrace_invop_calltrap_addr): .word 0 .word 0 #endif .text .align 2 /* * ASM macros for pushing and pulling trapframes from the stack * * These macros are used to handle the irqframe and trapframe structures * defined above. */ /* * PUSHFRAME - macro to push a trap frame on the stack in the current mode * Since the current mode is used, the SVC lr field is not defined. * * NOTE: r13 and r14 are stored separately as a work around for the * SA110 rev 2 STM^ bug */ #ifdef ARM_TP_ADDRESS #define PUSHFRAME \ sub sp, sp, #4; /* Align the stack */ \ str lr, [sp, #-4]!; /* Push the return address */ \ sub sp, sp, #(4*17); /* Adjust the stack pointer */ \ stmia sp, {r0-r12}; /* Push the user mode registers */ \ add r0, sp, #(4*13); /* Adjust the stack pointer */ \ stmia r0, {r13-r14}^; /* Push the user mode registers */ \ mov r0, r0; /* NOP for previous instruction */ \ mrs r0, spsr; /* Put the SPSR on the stack */ \ str r0, [sp, #-4]!; \ ldr r0, =ARM_RAS_START; \ mov r1, #0; \ str r1, [r0]; \ mov r1, #0xffffffff; \ str r1, [r0, #4]; #else #define PUSHFRAME \ sub sp, sp, #4; /* Align the stack */ \ str lr, [sp, #-4]!; /* Push the return address */ \ sub sp, sp, #(4*17); /* Adjust the stack pointer */ \ stmia sp, {r0-r12}; /* Push the user mode registers */ \ add r0, sp, #(4*13); /* Adjust the stack pointer */ \ stmia r0, {r13-r14}^; /* Push the user mode registers */ \ mov r0, r0; /* NOP for previous instruction */ \ mrs r0, spsr; /* Put the SPSR on the stack */ \ str r0, [sp, #-4]!; #endif /* * PULLFRAME - macro to pull a trap frame from the stack in the current mode * Since the current mode is used, the SVC lr field is ignored. */ #ifdef ARM_TP_ADDRESS #define PULLFRAME \ ldr r0, [sp], #4; /* Get the SPSR from stack */ \ msr spsr_fsxc, r0; \ ldmia sp, {r0-r14}^; /* Restore registers (usr mode) */ \ mov r0, r0; /* NOP for previous instruction */ \ add sp, sp, #(4*17); /* Adjust the stack pointer */ \ ldr lr, [sp], #4; /* Pull the return address */ \ add sp, sp, #4 /* Align the stack */ #else #define PULLFRAME \ ldr r0, [sp], #4 ; /* Get the SPSR from stack */ \ msr spsr_fsxc, r0; \ clrex; \ ldmia sp, {r0-r14}^; /* Restore registers (usr mode) */ \ mov r0, r0; /* NOP for previous instruction */ \ add sp, sp, #(4*17); /* Adjust the stack pointer */ \ ldr lr, [sp], #4; /* Pull the return address */ \ add sp, sp, #4 /* Align the stack */ #endif /* * PUSHFRAMEINSVC - macro to push a trap frame on the stack in SVC32 mode * This should only be used if the processor is not currently in SVC32 * mode. The processor mode is switched to SVC mode and the trap frame is * stored. The SVC lr field is used to store the previous value of * lr in SVC mode. * * NOTE: r13 and r14 are stored separately as a work around for the * SA110 rev 2 STM^ bug */ #ifdef ARM_TP_ADDRESS #define PUSHFRAMEINSVC \ stmdb sp, {r0-r3}; /* Save 4 registers */ \ mov r0, lr; /* Save xxx32 r14 */ \ mov r1, sp; /* Save xxx32 sp */ \ mrs r3, spsr; /* Save xxx32 spsr */ \ mrs r2, cpsr; /* Get the CPSR */ \ bic r2, r2, #(PSR_MODE); /* Fix for SVC mode */ \ orr r2, r2, #(PSR_SVC32_MODE); \ msr cpsr_c, r2; /* Punch into SVC mode */ \ mov r2, sp; /* Save SVC sp */ \ bic sp, sp, #7; /* Align sp to an 8-byte addrress */ \ - sub sp, sp, #4; /* Pad trapframe to keep alignment */ \ + sub sp, sp, #(4 * 17); /* Pad trapframe to keep alignment */ \ + /* and for dtrace to emulate push/pop */ \ str r0, [sp, #-4]!; /* Push return address */ \ str lr, [sp, #-4]!; /* Push SVC lr */ \ str r2, [sp, #-4]!; /* Push SVC sp */ \ msr spsr_fsxc, r3; /* Restore correct spsr */ \ ldmdb r1, {r0-r3}; /* Restore 4 regs from xxx mode */ \ sub sp, sp, #(4*15); /* Adjust the stack pointer */ \ stmia sp, {r0-r12}; /* Push the user mode registers */ \ add r0, sp, #(4*13); /* Adjust the stack pointer */ \ stmia r0, {r13-r14}^; /* Push the user mode registers */ \ mov r0, r0; /* NOP for previous instruction */ \ ldr r5, =ARM_RAS_START; /* Check if there's any RAS */ \ ldr r4, [r5, #4]; /* reset it to point at the */ \ cmp r4, #0xffffffff; /* end of memory if necessary; */ \ movne r1, #0xffffffff; /* leave value in r4 for later */ \ strne r1, [r5, #4]; /* comparision against PC. */ \ ldr r3, [r5]; /* Retrieve global RAS_START */ \ cmp r3, #0; /* and reset it if non-zero. */ \ movne r1, #0; /* If non-zero RAS_START and */ \ strne r1, [r5]; /* PC was lower than RAS_END, */ \ ldrne r1, [r0, #16]; /* adjust the saved PC so that */ \ cmpne r4, r1; /* execution later resumes at */ \ strhi r3, [r0, #16]; /* the RAS_START location. */ \ mrs r0, spsr; \ str r0, [sp, #-4]! #else #define PUSHFRAMEINSVC \ stmdb sp, {r0-r3}; /* Save 4 registers */ \ mov r0, lr; /* Save xxx32 r14 */ \ mov r1, sp; /* Save xxx32 sp */ \ mrs r3, spsr; /* Save xxx32 spsr */ \ mrs r2, cpsr; /* Get the CPSR */ \ bic r2, r2, #(PSR_MODE); /* Fix for SVC mode */ \ orr r2, r2, #(PSR_SVC32_MODE); \ msr cpsr_c, r2; /* Punch into SVC mode */ \ mov r2, sp; /* Save SVC sp */ \ bic sp, sp, #7; /* Align sp to an 8-byte addrress */ \ - sub sp, sp, #4; /* Pad trapframe to keep alignment */ \ + sub sp, sp, #(4 * 17); /* Pad trapframe to keep alignment */ \ + /* and for dtrace to emulate push/pop */ \ str r0, [sp, #-4]!; /* Push return address */ \ str lr, [sp, #-4]!; /* Push SVC lr */ \ str r2, [sp, #-4]!; /* Push SVC sp */ \ msr spsr_fsxc, r3; /* Restore correct spsr */ \ ldmdb r1, {r0-r3}; /* Restore 4 regs from xxx mode */ \ sub sp, sp, #(4*15); /* Adjust the stack pointer */ \ stmia sp, {r0-r12}; /* Push the user mode registers */ \ add r0, sp, #(4*13); /* Adjust the stack pointer */ \ stmia r0, {r13-r14}^; /* Push the user mode registers */ \ mov r0, r0; /* NOP for previous instruction */ \ mrs r0, spsr; /* Put the SPSR on the stack */ \ str r0, [sp, #-4]! #endif /* * PULLFRAMEFROMSVCANDEXIT - macro to pull a trap frame from the stack * in SVC32 mode and restore the saved processor mode and PC. * This should be used when the SVC lr register needs to be restored on * exit. */ #ifdef ARM_TP_ADDRESS #define PULLFRAMEFROMSVCANDEXIT \ ldr r0, [sp], #4; /* Get the SPSR from stack */ \ msr spsr_fsxc, r0; /* restore SPSR */ \ ldmia sp, {r0-r14}^; /* Restore registers (usr mode) */ \ mov r0, r0; /* NOP for previous instruction */ \ add sp, sp, #(4*15); /* Adjust the stack pointer */ \ ldmia sp, {sp, lr, pc}^ /* Restore lr and exit */ #else #define PULLFRAMEFROMSVCANDEXIT \ ldr r0, [sp], #4; /* Get the SPSR from stack */ \ msr spsr_fsxc, r0; /* restore SPSR */ \ clrex; \ ldmia sp, {r0-r14}^; /* Restore registers (usr mode) */ \ mov r0, r0; /* NOP for previous instruction */ \ add sp, sp, #(4*15); /* Adjust the stack pointer */ \ ldmia sp, {sp, lr, pc}^ /* Restore lr and exit */ #endif #if defined(__ARM_EABI__) /* * Unwind hints so we can unwind past functions that use * PULLFRAMEFROMSVCANDEXIT. They are run in reverse order. * As the last thing we do is restore the stack pointer * we can ignore the padding at the end of struct trapframe. */ #define UNWINDSVCFRAME \ .save {r13-r15}; /* Restore sp, lr, pc */ \ .pad #(2*4); /* Skip user sp and lr */ \ .save {r0-r12}; /* Restore r0-r12 */ \ .pad #(4) /* Skip spsr */ #else #define UNWINDSVCFRAME #endif #define DO_AST \ ldr r0, [sp]; /* Get the SPSR from stack */ \ mrs r4, cpsr; /* save CPSR */ \ orr r1, r4, #(PSR_I|PSR_F); \ msr cpsr_c, r1; /* Disable interrupts */ \ and r0, r0, #(PSR_MODE); /* Returning to USR mode? */ \ teq r0, #(PSR_USR32_MODE); \ bne 2f; /* Nope, get out now */ \ bic r4, r4, #(PSR_I|PSR_F); \ 1: GET_CURTHREAD_PTR(r5); \ ldr r1, [r5, #(TD_FLAGS)]; \ and r1, r1, #(TDF_ASTPENDING|TDF_NEEDRESCHED); \ teq r1, #0; \ beq 2f; /* Nope. Just bail */ \ msr cpsr_c, r4; /* Restore interrupts */ \ mov r0, sp; \ bl _C_LABEL(ast); /* ast(frame) */ \ orr r0, r4, #(PSR_I|PSR_F); \ msr cpsr_c, r0; \ b 1b; \ 2: /* * Entry point for a Software Interrupt (SWI). * * The hardware switches to svc32 mode on a swi, so we're already on the * right stack; just build a trapframe and call the handler. */ ASENTRY_NP(swi_entry) PUSHFRAME /* Build the trapframe on the */ mov r0, sp /* scv32 stack, pass it to the */ bl _C_LABEL(swi_handler) /* swi handler. */ /* * The fork_trampoline() code in swtch.S aranges for the MI fork_exit() * to return to swi_exit here, to return to userland. The net effect is * that a newly created thread appears to return from a SWI just like * the parent thread that created it. */ ASEENTRY_NP(swi_exit) DO_AST /* Handle pending signals. */ PULLFRAME /* Deallocate trapframe. */ movs pc, lr /* Return to userland. */ STOP_UNWINDING /* Don't unwind into user mode. */ EEND(swi_exit) END(swi_entry) /* * Standard exception exit handler. * * This is used to return from all exceptions except SWI. It uses DO_AST and * PULLFRAMEFROMSVCANDEXIT and can only be called if the exception entry code * used PUSHFRAMEINSVC. * * If the return is to user mode, this uses DO_AST to deliver any pending * signals and/or handle TDF_NEEDRESCHED first. */ ASENTRY_NP(exception_exit) DO_AST /* Handle pending signals. */ PULLFRAMEFROMSVCANDEXIT /* Return. */ UNWINDSVCFRAME /* Special unwinding for exceptions. */ END(exception_exit) /* * Entry point for a Prefetch Abort exception. * * The hardware switches to the abort mode stack; we switch to svc32 before * calling the handler, then return directly to the original mode/stack * on exit (without transitioning back through the abort mode stack). */ ASENTRY_NP(prefetch_abort_entry) #ifdef __XSCALE__ nop /* Make absolutely sure any pending */ nop /* imprecise aborts have occurred. */ #endif sub lr, lr, #4 /* Adjust the lr. Transition to scv32 */ PUSHFRAMEINSVC /* mode stack, build trapframe there. */ adr lr, exception_exit /* Return from handler via standard */ mov r0, sp /* exception exit routine. Pass the */ mov r1, #1 /* Type flag */ b _C_LABEL(abort_handler) END(prefetch_abort_entry) /* * Entry point for a Data Abort exception. * * The hardware switches to the abort mode stack; we switch to svc32 before * calling the handler, then return directly to the original mode/stack * on exit (without transitioning back through the abort mode stack). */ ASENTRY_NP(data_abort_entry) #ifdef __XSCALE__ nop /* Make absolutely sure any pending */ nop /* imprecise aborts have occurred. */ #endif sub lr, lr, #8 /* Adjust the lr. Transition to scv32 */ PUSHFRAMEINSVC /* mode stack, build trapframe there. */ adr lr, exception_exit /* Exception exit routine */ mov r0, sp /* Trapframe to the handler */ mov r1, #0 /* Type flag */ b _C_LABEL(abort_handler) END(data_abort_entry) /* * Entry point for an Undefined Instruction exception. * * The hardware switches to the undefined mode stack; we switch to svc32 before * calling the handler, then return directly to the original mode/stack * on exit (without transitioning back through the undefined mode stack). */ ASENTRY_NP(undefined_entry) sub lr, lr, #4 /* Adjust the lr. Transition to scv32 */ PUSHFRAMEINSVC /* mode stack, build trapframe there. */ adr lr, exception_exit /* Return from handler via standard */ mov r0, sp /* exception exit routine. Pass the */ b undefinedinstruction /* trapframe to the handler. */ END(undefined_entry) /* * Entry point for a normal IRQ. * * The hardware switches to the IRQ mode stack; we switch to svc32 before * calling the handler, then return directly to the original mode/stack * on exit (without transitioning back through the IRQ mode stack). */ ASENTRY_NP(irq_entry) sub lr, lr, #4 /* Adjust the lr. Transition to scv32 */ PUSHFRAMEINSVC /* mode stack, build trapframe there. */ adr lr, exception_exit /* Return from handler via standard */ mov r0, sp /* exception exit routine. Pass the */ b _C_LABEL(arm_irq_handler)/* trapframe to the handler. */ END(irq_entry) /* * Entry point for an FIQ interrupt. * * We don't currently support FIQ handlers very much. Something can * install itself in the FIQ vector using code (that may or may not work * these days) in fiq.c. If nobody does that and an FIQ happens, this * default handler just disables FIQs and otherwise ignores it. */ ASENTRY_NP(fiq_entry) mrs r8, cpsr /* FIQ handling isn't supported, */ bic r8, #(PSR_F) /* just disable FIQ and return. */ msr cpsr_c, r8 /* The r8 we trash here is the */ subs pc, lr, #4 /* banked FIQ-mode r8. */ END(fiq_entry) /* * Entry point for an Address Exception exception. * This is an arm26 exception that should never happen. */ ASENTRY_NP(addr_exception_entry) mov r3, lr mrs r2, spsr mrs r1, cpsr adr r0, Laddr_exception_msg b _C_LABEL(panic) Laddr_exception_msg: .asciz "Address Exception CPSR=0x%08x SPSR=0x%08x LR=0x%08x\n" .balign 4 END(addr_exception_entry) /* * Entry point for the system Reset vector. * This should never happen, so panic. */ ASENTRY_NP(reset_entry) mov r1, lr adr r0, Lreset_panicmsg b _C_LABEL(panic) /* NOTREACHED */ Lreset_panicmsg: .asciz "Reset vector called, LR = 0x%08x" .balign 4 END(reset_entry) /* * page0 and page0_data -- An image of the ARM vectors which is copied to * the ARM vectors page (high or low) as part of CPU initialization. The * code that does the copy assumes that page0_data holds one 32-bit word * of data for each of the predefined ARM vectors. It also assumes that * page0_data follows the vectors in page0, but other stuff can appear * between the two. We currently leave room between the two for some fiq * handler code to be copied in. */ .global _C_LABEL(page0), _C_LABEL(page0_data) _C_LABEL(page0): ldr pc, .Lreset_entry ldr pc, .Lundefined_entry ldr pc, .Lswi_entry ldr pc, .Lprefetch_abort_entry ldr pc, .Ldata_abort_entry ldr pc, .Laddr_exception_entry ldr pc, .Lirq_entry .fiqv: ldr pc, .Lfiq_entry .space 256 /* room for some fiq handler code */ _C_LABEL(page0_data): .Lreset_entry: .word reset_entry .Lundefined_entry: .word undefined_entry .Lswi_entry: .word swi_entry .Lprefetch_abort_entry: .word prefetch_abort_entry .Ldata_abort_entry: .word data_abort_entry .Laddr_exception_entry: .word addr_exception_entry .Lirq_entry: .word irq_entry .Lfiq_entry: .word fiq_entry /* * These items are used by the code in fiq.c to install what it calls the * "null" handler. It's actually our default vector entry that just jumps * to the default handler which just disables FIQs and returns. */ .global _C_LABEL(fiq_nullhandler_code), _C_LABEL(fiq_nullhandler_size) _C_LABEL(fiq_nullhandler_code): .word .fiqv _C_LABEL(fiq_nullhandler_size): .word 4 Index: projects/cxl_iscsi/sys/arm/arm/physmem.c =================================================================== --- projects/cxl_iscsi/sys/arm/arm/physmem.c (revision 279716) +++ projects/cxl_iscsi/sys/arm/arm/physmem.c (revision 279717) @@ -1,350 +1,350 @@ /*- * Copyright (c) 2014 Ian Lepore * All rights excluded. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include "opt_ddb.h" /* * Routines for describing and initializing anything related to physical memory. */ #include #include #include #include #include /* * These structures are used internally to keep track of regions of physical * ram, and regions within the physical ram that need to be excluded. An * exclusion region can be excluded from crash dumps, from the vm pool of pages * that can be allocated, or both, depending on the exclusion flags associated * with the region. */ #define MAX_HWCNT 10 #define MAX_EXCNT 10 struct region { vm_paddr_t addr; vm_size_t size; uint32_t flags; }; static struct region hwregions[MAX_HWCNT]; static struct region exregions[MAX_EXCNT]; static size_t hwcnt; static size_t excnt; /* * These "avail lists" are globals used to communicate physical memory layout to * other parts of the kernel. Within the arrays, each value is the starting * address of a contiguous area of physical address space. The values at even * indexes are areas that contain usable memory and the values at odd indexes * are areas that aren't usable. Each list is terminated by a pair of zero * entries. * * dump_avail tells the dump code what regions to include in a crash dump, and * phys_avail is the way we hand all the remaining physical ram we haven't used * in early kernel init over to the vm system for allocation management. * * We size these arrays to hold twice as many available regions as we allow for * hardware memory regions, to allow for the fact that exclusions can split a * hardware region into two or more available regions. In the real world there * will typically be one or two hardware regions and two or three exclusions. * * Each available region in this list occupies two array slots (the start of the * available region and the start of the unavailable region that follows it). */ #define MAX_AVAIL_REGIONS (MAX_HWCNT * 2) #define MAX_AVAIL_ENTRIES (MAX_AVAIL_REGIONS * 2) vm_paddr_t phys_avail[MAX_AVAIL_ENTRIES + 2]; /* +2 to allow for a pair */ vm_paddr_t dump_avail[MAX_AVAIL_ENTRIES + 2]; /* of zeroes to terminate. */ /* * realmem is the total number of hardware pages, excluded or not. * Maxmem is one greater than the last physical page number. */ long realmem; long Maxmem; /* The address at which the kernel was loaded. Set early in initarm(). */ vm_paddr_t arm_physmem_kernaddr; /* * Print the contents of the physical and excluded region tables using the * provided printf-like output function (which will be either printf or * db_printf). */ static void physmem_dump_tables(int (*prfunc)(const char *, ...)) { int flags, i; uintmax_t addr, size; const unsigned int mbyte = 1024 * 1024; prfunc("Physical memory chunk(s):\n"); for (i = 0; i < hwcnt; ++i) { addr = hwregions[i].addr; size = hwregions[i].size; prfunc(" 0x%08jx - 0x%08jx, %5ju MB (%7ju pages)\n", addr, addr + size - 1, size / mbyte, size / PAGE_SIZE); } prfunc("Excluded memory regions:\n"); for (i = 0; i < excnt; ++i) { addr = exregions[i].addr; size = exregions[i].size; flags = exregions[i].flags; prfunc(" 0x%08jx - 0x%08jx, %5ju MB (%7ju pages) %s %s\n", addr, addr + size - 1, size / mbyte, size / PAGE_SIZE, (flags & EXFLAG_NOALLOC) ? "NoAlloc" : "", (flags & EXFLAG_NODUMP) ? "NoDump" : ""); } #ifdef DEBUG prfunc("Avail lists:\n"); for (i = 0; phys_avail[i] != 0; ++i) { prfunc(" phys_avail[%d] 0x%08x\n", i, phys_avail[i]); } for (i = 0; dump_avail[i] != 0; ++i) { prfunc(" dump_avail[%d] 0x%08x\n", i, dump_avail[i]); } #endif } /* * Print the contents of the static mapping table. Used for bootverbose. */ void arm_physmem_print_tables() { physmem_dump_tables(printf); } /* * Walk the list of hardware regions, processing it against the list of * exclusions that contain the given exflags, and generating an "avail list". * - * Updates the kernel global 'realmem' with the sum of all pages in hw regions. + * Updates the value at *pavail with the sum of all pages in all hw regions. * * Returns the number of pages of non-excluded memory added to the avail list. */ static size_t regions_to_avail(vm_paddr_t *avail, uint32_t exflags, long *pavail) { size_t acnt, exi, hwi; vm_paddr_t end, start, xend, xstart; long availmem; const struct region *exp, *hwp; realmem = 0; availmem = 0; acnt = 0; for (hwi = 0, hwp = hwregions; hwi < hwcnt; ++hwi, ++hwp) { start = hwp->addr; end = hwp->size + start; realmem += arm32_btop(end - start); for (exi = 0, exp = exregions; exi < excnt; ++exi, ++exp) { /* * If the excluded region does not match given flags, * continue checking with the next excluded region. */ if ((exp->flags & exflags) == 0) continue; xstart = exp->addr; xend = exp->size + xstart; /* * If the excluded region ends before this hw region, * continue checking with the next excluded region. */ if (xend <= start) continue; /* * If the excluded region begins after this hw region * we're done because both lists are sorted. */ if (xstart >= end) break; /* * If the excluded region completely covers this hw * region, shrink this hw region to zero size. */ if ((start >= xstart) && (end <= xend)) { start = xend; end = xend; break; } /* * If the excluded region falls wholly within this hw * region without abutting or overlapping the beginning * or end, create an available entry from the leading * fragment, then adjust the start of this hw region to * the end of the excluded region, and continue checking * the next excluded region because another exclusion * could affect the remainder of this hw region. */ if ((xstart > start) && (xend < end)) { avail[acnt++] = start; avail[acnt++] = xstart; availmem += arm32_btop(xstart - start); start = xend; continue; } /* * We know the excluded region overlaps either the start * or end of this hardware region (but not both), trim * the excluded portion off the appropriate end. */ if (xstart <= start) start = xend; else end = xstart; } /* * If the trimming actions above left a non-zero size, create an * available entry for it. */ if (end > start) { avail[acnt++] = start; avail[acnt++] = end; availmem += arm32_btop(end - start); } if (acnt >= MAX_AVAIL_ENTRIES) panic("Not enough space in the dump/phys_avail arrays"); } if (pavail) *pavail = availmem; return (acnt); } /* * Insertion-sort a new entry into a regions list; sorted by start address. */ static void insert_region(struct region *regions, size_t rcnt, vm_paddr_t addr, vm_size_t size, uint32_t flags) { size_t i; struct region *ep, *rp; ep = regions + rcnt; for (i = 0, rp = regions; i < rcnt; ++i, ++rp) { if (addr < rp->addr) { bcopy(rp, rp + 1, (ep - rp) * sizeof(*rp)); break; } } rp->addr = addr; rp->size = size; rp->flags = flags; } /* * Add a hardware memory region. */ void arm_physmem_hardware_region(vm_paddr_t pa, vm_size_t sz) { vm_offset_t adj; /* * Filter out the page at PA 0x00000000. The VM can't handle it, as * pmap_extract() == 0 means failure. */ if (pa == 0) { pa = PAGE_SIZE; sz -= PAGE_SIZE; } /* * Round the starting address up to a page boundary, and truncate the * ending page down to a page boundary. */ adj = round_page(pa) - pa; pa = round_page(pa); sz = trunc_page(sz - adj); if (hwcnt < nitems(hwregions)) insert_region(hwregions, hwcnt++, pa, sz, 0); } /* * Add an exclusion region. */ void arm_physmem_exclude_region(vm_paddr_t pa, vm_size_t sz, uint32_t exflags) { vm_offset_t adj; /* * Truncate the starting address down to a page boundary, and round the * ending page up to a page boundary. */ adj = pa - trunc_page(pa); pa = trunc_page(pa); sz = round_page(sz + adj); if (excnt < nitems(exregions)) insert_region(exregions, excnt++, pa, sz, exflags); } /* * Process all the regions added earlier into the global avail lists. * * Updates the kernel global 'physmem' with the number of physical pages * available for use (all pages not in any exclusion region). * * Updates the kernel global 'Maxmem' with the page number one greater then the * last page of physical memory in the system. */ void arm_physmem_init_kernel_globals(void) { size_t nextidx; regions_to_avail(dump_avail, EXFLAG_NODUMP, NULL); nextidx = regions_to_avail(phys_avail, EXFLAG_NOALLOC, &physmem); if (nextidx == 0) panic("No memory entries in phys_avail"); Maxmem = atop(phys_avail[nextidx - 1]); } #ifdef DDB #include DB_SHOW_COMMAND(physmem, db_show_physmem) { physmem_dump_tables(db_printf); } #endif /* DDB */ Index: projects/cxl_iscsi/sys/arm/arm/undefined.c =================================================================== --- projects/cxl_iscsi/sys/arm/arm/undefined.c (revision 279716) +++ projects/cxl_iscsi/sys/arm/arm/undefined.c (revision 279717) @@ -1,294 +1,305 @@ /* $NetBSD: undefined.c,v 1.22 2003/11/29 22:21:29 bjh21 Exp $ */ /*- * Copyright (c) 2001 Ben Harris. * Copyright (c) 1995 Mark Brinicombe. * Copyright (c) 1995 Brini. * All rights reserved. * * This code is derived from software written for Brini by Mark Brinicombe * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by Brini. * 4. The name of the company nor the name of the author may be used to * endorse or promote products derived from this software without specific * prior written permission. * * THIS SOFTWARE IS PROVIDED BY BRINI ``AS IS'' AND ANY EXPRESS OR IMPLIED * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL BRINI OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * RiscBSD kernel project * * undefined.c * * Fault handler * * Created : 06/01/95 */ #include "opt_ddb.h" #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #ifdef KDB #include #endif #include #include #include #include #include #include #include #include #include #ifdef DDB #include #endif #ifdef KDB #include #endif +#ifdef KDTRACE_HOOKS +int (*dtrace_invop_jump_addr)(struct trapframe *); +#endif + static int gdb_trapper(u_int, u_int, struct trapframe *, int); LIST_HEAD(, undefined_handler) undefined_handlers[MAX_COPROCS]; void * install_coproc_handler(int coproc, undef_handler_t handler) { struct undefined_handler *uh; KASSERT(coproc >= 0 && coproc < MAX_COPROCS, ("bad coproc")); KASSERT(handler != NULL, ("handler is NULL")); /* Used to be legal. */ /* XXX: M_TEMP??? */ uh = malloc(sizeof(*uh), M_TEMP, M_WAITOK); uh->uh_handler = handler; install_coproc_handler_static(coproc, uh); return uh; } void install_coproc_handler_static(int coproc, struct undefined_handler *uh) { LIST_INSERT_HEAD(&undefined_handlers[coproc], uh, uh_link); } void remove_coproc_handler(void *cookie) { struct undefined_handler *uh = cookie; LIST_REMOVE(uh, uh_link); free(uh, M_TEMP); } static int gdb_trapper(u_int addr, u_int insn, struct trapframe *frame, int code) { struct thread *td; ksiginfo_t ksi; td = (curthread == NULL) ? &thread0 : curthread; if (insn == GDB_BREAKPOINT || insn == GDB5_BREAKPOINT) { if (code == FAULT_USER) { ksiginfo_init_trap(&ksi); ksi.ksi_signo = SIGTRAP; ksi.ksi_code = TRAP_BRKPT; ksi.ksi_addr = (u_int32_t *)addr; trapsignal(td, &ksi); return 0; } #if 0 #ifdef KGDB return !kgdb_trap(T_BREAKPOINT, frame); #endif #endif } return 1; } static struct undefined_handler gdb_uh; void undefined_init() { int loop; /* Not actually necessary -- the initialiser is just NULL */ for (loop = 0; loop < MAX_COPROCS; ++loop) LIST_INIT(&undefined_handlers[loop]); /* Install handler for GDB breakpoints */ gdb_uh.uh_handler = gdb_trapper; install_coproc_handler_static(0, &gdb_uh); } void undefinedinstruction(struct trapframe *frame) { struct thread *td; u_int fault_pc; int fault_instruction; int fault_code; int coprocessor; struct undefined_handler *uh; #ifdef VERBOSE_ARM32 int s; #endif ksiginfo_t ksi; /* Enable interrupts if they were enabled before the exception. */ if (__predict_true(frame->tf_spsr & PSR_I) == 0) enable_interrupts(PSR_I); if (__predict_true(frame->tf_spsr & PSR_F) == 0) enable_interrupts(PSR_F); PCPU_INC(cnt.v_trap); fault_pc = frame->tf_pc; /* * Get the current thread/proc structure or thread0/proc0 if there is * none. */ td = curthread == NULL ? &thread0 : curthread; /* * Make sure the program counter is correctly aligned so we * don't take an alignment fault trying to read the opcode. */ if (__predict_false((fault_pc & 3) != 0)) { ksiginfo_init_trap(&ksi); ksi.ksi_signo = SIGILL; ksi.ksi_code = ILL_ILLADR; ksi.ksi_addr = (u_int32_t *)(intptr_t) fault_pc; trapsignal(td, &ksi); userret(td, frame); return; } /* * Should use fuword() here .. but in the interests of squeezing every * bit of speed we will just use ReadWord(). We know the instruction * can be read as was just executed so this will never fail unless the * kernel is screwed up in which case it does not really matter does * it ? */ fault_instruction = *(u_int32_t *)fault_pc; /* Update vmmeter statistics */ #if 0 uvmexp.traps++; #endif /* Check for coprocessor instruction */ /* * According to the datasheets you only need to look at bit 27 of the * instruction to tell the difference between and undefined * instruction and a coprocessor instruction following an undefined * instruction trap. */ coprocessor = 0; if ((fault_instruction & (1 << 27)) != 0) coprocessor = (fault_instruction >> 8) & 0x0f; #ifdef VFP else { /* check for special instructions */ if (((fault_instruction & 0xfe000000) == 0xf2000000) || ((fault_instruction & 0xff100000) == 0xf4000000)) coprocessor = 10; /* vfp / simd */ } #endif /* VFP */ if ((frame->tf_spsr & PSR_MODE) == PSR_USR32_MODE) { /* * Modify the fault_code to reflect the USR/SVC state at * time of fault. */ fault_code = FAULT_USER; td->td_frame = frame; } else fault_code = 0; /* OK this is were we do something about the instruction. */ LIST_FOREACH(uh, &undefined_handlers[coprocessor], uh_link) if (uh->uh_handler(fault_pc, fault_instruction, frame, fault_code) == 0) break; if (fault_code & FAULT_USER && fault_instruction == PTRACE_BREAKPOINT) { PROC_LOCK(td->td_proc); _PHOLD(td->td_proc); ptrace_clear_single_step(td); _PRELE(td->td_proc); PROC_UNLOCK(td->td_proc); return; } if (uh == NULL && (fault_code & FAULT_USER)) { /* Fault has not been handled */ ksiginfo_init_trap(&ksi); ksi.ksi_signo = SIGILL; ksi.ksi_code = ILL_ILLOPC; ksi.ksi_addr = (u_int32_t *)(intptr_t) fault_pc; trapsignal(td, &ksi); } if ((fault_code & FAULT_USER) == 0) { if (fault_instruction == KERNEL_BREAKPOINT) { #ifdef KDB kdb_trap(T_BREAKPOINT, 0, frame); #else printf("No debugger in kernel.\n"); #endif return; - } else + } +#ifdef KDTRACE_HOOKS + else if (dtrace_invop_jump_addr != 0) { + dtrace_invop_jump_addr(frame); + return; + } +#endif + else panic("Undefined instruction in kernel.\n"); } userret(td, frame); } Index: projects/cxl_iscsi/sys/arm/conf/BEAGLEBONE =================================================================== --- projects/cxl_iscsi/sys/arm/conf/BEAGLEBONE (revision 279716) +++ projects/cxl_iscsi/sys/arm/conf/BEAGLEBONE (revision 279717) @@ -1,169 +1,169 @@ # # BEAGLEBONE -- Custom configuration for the BeagleBone ARM development # platforms, check out http://www.beagleboard.org/bone and # http://www.beagleboard.org/black. This kernel config file is used for the # original BeagleBone and the BeagleBone Black. # # For more information on this file, please read the config(5) manual page, # and/or the handbook section on Kernel Configuration Files: # # http://www.FreeBSD.org/doc/en_US.ISO8859-1/books/handbook/kernelconfig-config.html # # The handbook is also available locally in /usr/share/doc/handbook # if you've installed the doc distribution, otherwise always see the # FreeBSD World Wide Web server (http://www.FreeBSD.org/) for the # latest information. # # An exhaustive list of options and more detailed explanations of the # device lines is also present in the ../../conf/NOTES and NOTES files. # If you are in doubt as to the purpose or necessity of a line, check first # in NOTES. # # $FreeBSD$ ident BEAGLEBONE include "../ti/am335x/std.am335x" makeoptions WITHOUT_MODULES="ahc" # DTrace support -options KDTRACE_HOOKS # Kernel DTrace hooks -options DDB_CTF # all architectures - kernel ELF linker loads CTF data -makeoptions WITH_CTF=1 -makeoptions MODULES_OVERRIDE="opensolaris dtrace dtrace/lockstat dtrace/profile dtrace/fbt" +options KDTRACE_HOOKS # Kernel DTrace hooks +options DDB_CTF # all architectures - kernel ELF linker loads CTF data +makeoptions WITH_CTF=1 +makeoptions MODULES_OVERRIDE="opensolaris dtrace dtrace/lockstat dtrace/profile dtrace/fbt" options HZ=100 options SCHED_4BSD # 4BSD scheduler options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories options UFS_GJOURNAL # Enable gjournal-based UFS journaling options QUOTA # Enable disk quotas for UFS options NFSCL # Network Filesystem Client options NFSLOCKD # Network Lock Manager options NFS_ROOT # NFS usable as /, requires NFSCL options MSDOSFS # MSDOS Filesystem options CD9660 # ISO 9660 Filesystem options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework options TMPFS # Efficient memory filesystem options GEOM_PART_GPT # GUID Partition Tables options GEOM_PART_BSD # BSD partition scheme options GEOM_PART_MBR # MBR partition scheme options COMPAT_43 # Compatible with BSD 4.3 [KEEP THIS!] options SCSI_DELAY=5000 # Delay (in ms) before probing SCSI options KTRACE # ktrace(1) support options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions options KBD_INSTALL_CDEV # install a CDEV entry in /dev options PLATFORM options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) options VFP # Enable floating point hardware support # Debugging for use in -current makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options BREAK_TO_DEBUGGER #options VERBOSE_SYSINIT # Enable verbose sysinit messages options KDB # Enable kernel debugger support # For minimum debugger support (stable branch) use: #options KDB_TRACE # Print a stack trace for a panic # For full debugger support use this instead: options DDB # Enable the kernel debugger options INVARIANTS # Enable calls of extra sanity checking options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS options WITNESS # Enable checks to detect deadlocks and cycles options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed #options DIAGNOSTIC # NFS server support #options NFSD # NFS root from boopt/dhcp #options BOOTP #options BOOTP_NFSROOT #options BOOTP_COMPAT #options BOOTP_NFSV3 #options BOOTP_WIRED_TO=cpsw0 # Boot device is 2nd slice on MMC/SD card options ROOTDEVNAME=\"ufs:mmcsd0s2\" # MMC/SD/SDIO Card slot support device mmc # mmc/sd bus device mmcsd # mmc/sd flash cards device sdhci # mmc/sd host controller # I2C support device iicbus device iic device ti_i2c device am335x_pmic # AM335x Power Management IC (TPC65217) device am335x_rtc # RTC support (power management only) # Console and misc device uart device uart_ns8250 device pty device snp device md device random # Entropy device # GPIO device gpio device gpioled # ADC support device ti_adc # Watchdog support # If we don't enable the watchdog driver, the system could potentially # reboot automatically because the boot loader might have enabled the # watchdog. device ti_wdt # TI Programmable Realtime Unit support device ti_pruss # Mailbox support device ti_mbox # USB support device usb options USB_HOST_ALIGN=64 # Align usb buffers to cache line size. options USB_DEBUG #options USB_REQ_DEBUG #options USB_VERBOSE device musb device umass device scbus # SCSI bus (required for ATA/SCSI) device da # Direct Access (disks) # Ethernet device loop device ether device mii device smscphy device cpsw device bpf # USB Ethernet support, requires miibus device miibus device axe # ASIX Electronics USB Ethernet # Device mode support and USFS template device usb_template # Control of the gadget device usfs # Flattened Device Tree options FDT # Configure using FDT/DTB data options FDT_DTB_STATIC makeoptions FDT_DTS_FILE=beaglebone.dts Index: projects/cxl_iscsi/sys/arm/xilinx/zy7_devcfg.c =================================================================== --- projects/cxl_iscsi/sys/arm/xilinx/zy7_devcfg.c (revision 279716) +++ projects/cxl_iscsi/sys/arm/xilinx/zy7_devcfg.c (revision 279717) @@ -1,668 +1,847 @@ /*- * Copyright (c) 2013 Thomas Skibo * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ /* * Zynq-7000 Devcfg driver. This allows programming the PL (FPGA) section * of Zynq. * * Reference: Zynq-7000 All Programmable SoC Technical Reference Manual. * (v1.4) November 16, 2012. Xilinx doc UG585. PL Configuration is * covered in section 6.4.5. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include struct zy7_devcfg_softc { device_t dev; struct mtx sc_mtx; struct resource *mem_res; struct resource *irq_res; struct cdev *sc_ctl_dev; void *intrhandle; bus_dma_tag_t dma_tag; bus_dmamap_t dma_map; int is_open; + + struct sysctl_ctx_list sysctl_tree; + struct sysctl_oid *sysctl_tree_top; }; static struct zy7_devcfg_softc *zy7_devcfg_softc_p; +#define FCLK_NUM 4 + +struct zy7_fclk_config { + int source; + int frequency; + int actual_frequency; +}; + +static struct zy7_fclk_config fclk_configs[FCLK_NUM]; + #define DEVCFG_SC_LOCK(sc) mtx_lock(&(sc)->sc_mtx) #define DEVCFG_SC_UNLOCK(sc) mtx_unlock(&(sc)->sc_mtx) #define DEVCFG_SC_LOCK_INIT(sc) \ mtx_init(&(sc)->sc_mtx, device_get_nameunit((sc)->dev), \ "zy7_devcfg", MTX_DEF) #define DEVCFG_SC_LOCK_DESTROY(sc) mtx_destroy(&(sc)->sc_mtx); #define DEVCFG_SC_ASSERT_LOCKED(sc) mtx_assert(&(sc)->sc_mtx, MA_OWNED); #define RD4(sc, off) (bus_read_4((sc)->mem_res, (off))) #define WR4(sc, off, val) (bus_write_4((sc)->mem_res, (off), (val))) SYSCTL_NODE(_hw, OID_AUTO, fpga, CTLFLAG_RD, 0, \ "Xilinx Zynq-7000 PL (FPGA) section"); static int zy7_devcfg_sysctl_pl_done(SYSCTL_HANDLER_ARGS); SYSCTL_PROC(_hw_fpga, OID_AUTO, pl_done, CTLTYPE_INT | CTLFLAG_RD, NULL, 0, zy7_devcfg_sysctl_pl_done, "I", "PL section config DONE signal"); static int zy7_en_level_shifters = 1; SYSCTL_INT(_hw_fpga, OID_AUTO, en_level_shifters, CTLFLAG_RW, &zy7_en_level_shifters, 0, "Enable PS-PL level shifters after device config"); static int zy7_ps_vers = 0; SYSCTL_INT(_hw, OID_AUTO, ps_vers, CTLFLAG_RD, &zy7_ps_vers, 0, "Zynq-7000 PS version"); +static int zy7_devcfg_fclk_sysctl_level_shifters(SYSCTL_HANDLER_ARGS); +SYSCTL_PROC(_hw_fpga, OID_AUTO, level_shifters, + CTLFLAG_RW | CTLTYPE_INT, + NULL, 0, zy7_devcfg_fclk_sysctl_level_shifters, + "I", "Enable/disable level shifters"); /* cdev entry points. */ static int zy7_devcfg_open(struct cdev *, int, int, struct thread *); static int zy7_devcfg_write(struct cdev *, struct uio *, int); static int zy7_devcfg_close(struct cdev *, int, int, struct thread *); - struct cdevsw zy7_devcfg_cdevsw = { .d_version = D_VERSION, .d_open = zy7_devcfg_open, .d_write = zy7_devcfg_write, .d_close = zy7_devcfg_close, .d_name = "devcfg", }; /* Devcfg block registers. */ #define ZY7_DEVCFG_CTRL 0x0000 #define ZY7_DEVCFG_CTRL_FORCE_RST (1<<31) #define ZY7_DEVCFG_CTRL_PCFG_PROG_B (1<<30) #define ZY7_DEVCFG_CTRL_PCFG_POR_CNT_4K (1<<29) #define ZY7_DEVCFG_CTRL_PCAP_PR (1<<27) #define ZY7_DEVCFG_CTRL_PCAP_MODE (1<<26) #define ZY7_DEVCFG_CTRL_QTR_PCAP_RATE_EN (1<<25) #define ZY7_DEVCFG_CTRL_MULTIBOOT_EN (1<<24) #define ZY7_DEVCFG_CTRL_JTAG_CHAIN_DIS (1<<23) #define ZY7_DEVCFG_CTRL_USER_MODE (1<<15) #define ZY7_DEVCFG_CTRL_RESVD_WR11 (3<<13) /* always write 11 */ #define ZY7_DEVCFG_CTRL_PCFG_AES_FUSE (1<<12) #define ZY7_DEVCFG_CTRL_PCFG_AES_EN_MASK (7<<9) /* all 1's or 0's */ #define ZY7_DEVCFG_CTRL_SEU_EN (1<<8) #define ZY7_DEVCFG_CTRL_SEC_EN (1<<7) #define ZY7_DEVCFG_CTRL_SPNIDEN (1<<6) #define ZY7_DEVCFG_CTRL_SPIDEN (1<<5) #define ZY7_DEVCFG_CTRL_NIDEN (1<<4) #define ZY7_DEVCFG_CTRL_DBGEN (1<<3) #define ZY7_DEVCFG_CTRL_DAP_EN_MASK (7<<0) /* all 1's to enable */ #define ZY7_DEVCFG_LOCK 0x004 #define ZY7_DEVCFG_LOCK_AES_FUSE_LOCK (1<<4) #define ZY7_DEVCFG_LOCK_AES_EN (1<<3) #define ZY7_DEVCFG_LOCK_SEU_LOCK (1<<2) #define ZY7_DEVCFG_LOCK_SEC_LOCK (1<<1) #define ZY7_DEVCFG_LOCK_DBG_LOCK (1<<0) #define ZY7_DEVCFG_CFG 0x008 #define ZY7_DEVCFG_CFG_RFIFO_TH_MASK (3<<10) #define ZY7_DEVCFG_CFG_WFIFO_TH_MASK (3<<8) #define ZY7_DEVCFG_CFG_RCLK_EDGE (1<<7) #define ZY7_DEVCFG_CFG_WCLK_EDGE (1<<6) #define ZY7_DEVCFG_CFG_DIS_SRC_INC (1<<5) #define ZY7_DEVCFG_CFG_DIS_DST_INC (1<<4) #define ZY7_DEVCFG_INT_STATUS 0x00C #define ZY7_DEVCFG_INT_MASK 0x010 #define ZY7_DEVCFG_INT_PSS_GTS_USR_B (1<<31) #define ZY7_DEVCFG_INT_PSS_FST_CFG_B (1<<30) #define ZY7_DEVCFG_INT_PSS_GPWRDWN_B (1<<29) #define ZY7_DEVCFG_INT_PSS_GTS_CFG_B (1<<28) #define ZY7_DEVCFG_INT_CFG_RESET_B (1<<27) #define ZY7_DEVCFG_INT_AXI_WTO (1<<23) /* axi write timeout */ #define ZY7_DEVCFG_INT_AXI_WERR (1<<22) /* axi write err */ #define ZY7_DEVCFG_INT_AXI_RTO (1<<21) /* axi read timeout */ #define ZY7_DEVCFG_INT_AXI_RERR (1<<20) /* axi read err */ #define ZY7_DEVCFG_INT_RX_FIFO_OV (1<<18) /* rx fifo overflow */ #define ZY7_DEVCFG_INT_WR_FIFO_LVL (1<<17) /* wr fifo < level */ #define ZY7_DEVCFG_INT_RD_FIFO_LVL (1<<16) /* rd fifo >= level */ #define ZY7_DEVCFG_INT_DMA_CMD_ERR (1<<15) #define ZY7_DEVCFG_INT_DMA_Q_OV (1<<14) #define ZY7_DEVCFG_INT_DMA_DONE (1<<13) #define ZY7_DEVCFG_INT_DMA_PCAP_DONE (1<<12) #define ZY7_DEVCFG_INT_P2D_LEN_ERR (1<<11) #define ZY7_DEVCFG_INT_PCFG_HMAC_ERR (1<<6) #define ZY7_DEVCFG_INT_PCFG_SEU_ERR (1<<5) #define ZY7_DEVCFG_INT_PCFG_POR_B (1<<4) #define ZY7_DEVCFG_INT_PCFG_CFG_RST (1<<3) #define ZY7_DEVCFG_INT_PCFG_DONE (1<<2) #define ZY7_DEVCFG_INT_PCFG_INIT_PE (1<<1) #define ZY7_DEVCFG_INT_PCFG_INIT_NE (1<<0) #define ZY7_DEVCFG_INT_ERRORS 0x00f0f860 #define ZY7_DEVCFG_INT_ALL 0xf8f7f87f #define ZY7_DEVCFG_STATUS 0x014 #define ZY7_DEVCFG_STATUS_DMA_CMD_Q_F (1<<31) /* cmd queue full */ #define ZY7_DEVCFG_STATUS_DMA_CMD_Q_E (1<<30) /* cmd queue empty */ #define ZY7_DEVCFG_STATUS_DONE_COUNT_MASK (3<<28) #define ZY7_DEVCFG_STATUS_DONE_COUNT_SHIFT 28 #define ZY7_DEVCFG_STATUS_RX_FIFO_LVL_MASK (0x1f<<20) #define ZY7_DEVCFG_STATUS_RX_FIFO_LVL_SHIFT 20 #define ZY7_DEVCFG_STATUS_TX_FIFO_LVL_MASK (0x7f<<12) #define ZY7_DEVCFG_STATUS_TX_FIFO_LVL_SHIFT 12 #define ZY7_DEVCFG_STATUS_PSS_GTS_USR_B (1<<11) #define ZY7_DEVCFG_STATUS_PSS_FST_CFG_B (1<<10) #define ZY7_DEVCFG_STATUS_PSS_GPWRDWN_B (1<<9) #define ZY7_DEVCFG_STATUS_PSS_GTS_CFG_B (1<<8) #define ZY7_DEVCFG_STATUS_ILL_APB_ACCE (1<<6) #define ZY7_DEVCFG_STATUS_PSS_CFG_RESET_B (1<<5) #define ZY7_DEVCFG_STATUS_PCFG_INIT (1<<4) #define ZY7_DEVCFG_STATUS_EFUSE_BBRAM_KEY_DIS (1<<3) #define ZY7_DEVCFG_STATUS_EFUSE_SEC_EN (1<<2) #define ZY7_DEVCFG_STATUS_EFUSE_JTAG_DIS (1<<1) #define ZY7_DEVCFG_DMA_SRC_ADDR 0x018 #define ZY7_DEVCFG_DMA_DST_ADDR 0x01c #define ZY7_DEVCFG_DMA_ADDR_WAIT_PCAP 1 #define ZY7_DEVCFG_DMA_ADDR_ILLEGAL 0xffffffff #define ZY7_DEVCFG_DMA_SRC_LEN 0x020 /* in 4-byte words. */ #define ZY7_DEVCFG_DMA_SRC_LEN_MAX 0x7ffffff #define ZY7_DEVCFG_DMA_DST_LEN 0x024 #define ZY7_DEVCFG_ROM_SHADOW 0x028 #define ZY7_DEVCFG_MULTIBOOT_ADDR 0x02c #define ZY7_DEVCFG_SW_ID 0x030 #define ZY7_DEVCFG_UNLOCK 0x034 #define ZY7_DEVCFG_UNLOCK_MAGIC 0x757bdf0d #define ZY7_DEVCFG_MCTRL 0x080 #define ZY7_DEVCFG_MCTRL_PS_VERS_MASK (0xf<<28) #define ZY7_DEVCFG_MCTRL_PS_VERS_SHIFT 28 #define ZY7_DEVCFG_MCTRL_PCFG_POR_B (1<<8) #define ZY7_DEVCFG_MCTRL_INT_PCAP_LPBK (1<<4) #define ZY7_DEVCFG_XADCIF_CFG 0x100 #define ZY7_DEVCFG_XADCIF_INT_STAT 0x104 #define ZY7_DEVCFG_XADCIF_INT_MASK 0x108 #define ZY7_DEVCFG_XADCIF_MSTS 0x10c #define ZY7_DEVCFG_XADCIF_CMD_FIFO 0x110 #define ZY7_DEVCFG_XADCIF_RD_FIFO 0x114 #define ZY7_DEVCFG_XADCIF_MCTL 0x118 +static int +zy7_devcfg_fclk_sysctl_source(SYSCTL_HANDLER_ARGS) +{ + char buf[4]; + struct zy7_fclk_config *cfg; + int unit; + int error; + cfg = arg1; + unit = arg2; + + switch (cfg->source) { + case ZY7_PL_FCLK_SRC_IO: + case ZY7_PL_FCLK_SRC_IO_ALT: + strncpy(buf, "IO", sizeof(buf)); + break; + case ZY7_PL_FCLK_SRC_DDR: + strncpy(buf, "DDR", sizeof(buf)); + break; + case ZY7_PL_FCLK_SRC_ARM: + strncpy(buf, "ARM", sizeof(buf)); + break; + default: + strncpy(buf, "???", sizeof(buf)); + break; + } + + error = sysctl_handle_string(oidp, buf, sizeof(buf), req); + if (error != 0 || req->newptr == NULL) + return (error); + + if (strcasecmp(buf, "io") == 0) + cfg->source = ZY7_PL_FCLK_SRC_IO; + else if (strcasecmp(buf, "ddr") == 0) + cfg->source = ZY7_PL_FCLK_SRC_DDR; + else if (strcasecmp(buf, "arm") == 0) + cfg->source = ZY7_PL_FCLK_SRC_ARM; + else + return (EINVAL); + + zy7_pl_fclk_set_source(unit, cfg->source); + if (cfg->frequency > 0) + cfg->actual_frequency = zy7_pl_fclk_get_freq(unit); + + return (0); +} + +static int +zy7_devcfg_fclk_sysctl_freq(SYSCTL_HANDLER_ARGS) +{ + struct zy7_fclk_config *cfg; + int unit; + int error; + int freq; + int new_actual_freq; + + cfg = arg1; + unit = arg2; + + freq = cfg->frequency; + + error = sysctl_handle_int(oidp, &freq, 0, req); + if (error != 0 || req->newptr == NULL) + return (error); + + if (freq > 0) { + new_actual_freq = zy7_pl_fclk_set_freq(unit, freq); + if (new_actual_freq < 0) + return (EINVAL); + if (!zy7_pl_fclk_enabled(unit)) + zy7_pl_fclk_enable(unit); + } + else { + zy7_pl_fclk_disable(unit); + new_actual_freq = 0; + } + + cfg->frequency = freq; + cfg->actual_frequency = new_actual_freq; + + return (0); +} + +static int +zy7_devcfg_fclk_sysctl_level_shifters(SYSCTL_HANDLER_ARGS) +{ + int error, enabled; + + enabled = zy7_pl_level_shifters_enabled(); + + error = sysctl_handle_int(oidp, &enabled, 0, req); + if (error != 0 || req->newptr == NULL) + return (error); + + if (enabled) + zy7_pl_level_shifters_enable(); + else + zy7_pl_level_shifters_disable(); + + return (0); +} + +static int +zy7_devcfg_init_fclk_sysctl(struct zy7_devcfg_softc *sc) +{ + struct sysctl_oid *fclk_node; + char fclk_num[4]; + int i; + + sysctl_ctx_init(&sc->sysctl_tree); + sc->sysctl_tree_top = SYSCTL_ADD_NODE(&sc->sysctl_tree, + SYSCTL_STATIC_CHILDREN(_hw_fpga), OID_AUTO, "fclk", + CTLFLAG_RD, 0, ""); + if (sc->sysctl_tree_top == NULL) { + sysctl_ctx_free(&sc->sysctl_tree); + return (-1); + } + + for (i = 0; i < FCLK_NUM; i++) { + snprintf(fclk_num, sizeof(fclk_num), "%d", i); + fclk_node = SYSCTL_ADD_NODE(&sc->sysctl_tree, + SYSCTL_CHILDREN(sc->sysctl_tree_top), OID_AUTO, fclk_num, + CTLFLAG_RD, 0, ""); + + SYSCTL_ADD_INT(&sc->sysctl_tree, + SYSCTL_CHILDREN(fclk_node), OID_AUTO, + "actual_freq", CTLFLAG_RD, + &fclk_configs[i].actual_frequency, i, + "Actual frequency"); + SYSCTL_ADD_PROC(&sc->sysctl_tree, + SYSCTL_CHILDREN(fclk_node), OID_AUTO, + "freq", CTLFLAG_RW | CTLTYPE_INT, + &fclk_configs[i], i, + zy7_devcfg_fclk_sysctl_freq, + "I", "Configured frequency"); + SYSCTL_ADD_PROC(&sc->sysctl_tree, + SYSCTL_CHILDREN(fclk_node), OID_AUTO, + "source", CTLFLAG_RW | CTLTYPE_STRING, + &fclk_configs[i], i, + zy7_devcfg_fclk_sysctl_source, + "A", "Clock source"); + } + + return (0); +} + /* Enable programming the PL through PCAP. */ static void zy7_devcfg_init_hw(struct zy7_devcfg_softc *sc) { DEVCFG_SC_ASSERT_LOCKED(sc); /* Set devcfg control register. */ WR4(sc, ZY7_DEVCFG_CTRL, ZY7_DEVCFG_CTRL_PCFG_PROG_B | ZY7_DEVCFG_CTRL_PCAP_PR | ZY7_DEVCFG_CTRL_PCAP_MODE | ZY7_DEVCFG_CTRL_USER_MODE | ZY7_DEVCFG_CTRL_RESVD_WR11 | ZY7_DEVCFG_CTRL_SPNIDEN | ZY7_DEVCFG_CTRL_SPIDEN | ZY7_DEVCFG_CTRL_NIDEN | ZY7_DEVCFG_CTRL_DBGEN | ZY7_DEVCFG_CTRL_DAP_EN_MASK); /* Turn off internal PCAP loopback. */ WR4(sc, ZY7_DEVCFG_MCTRL, RD4(sc, ZY7_DEVCFG_MCTRL) & ~ZY7_DEVCFG_MCTRL_INT_PCAP_LPBK); } /* Clear previous configuration of the PL by asserting PROG_B. */ static int zy7_devcfg_reset_pl(struct zy7_devcfg_softc *sc) { uint32_t devcfg_ctl; int tries, err; DEVCFG_SC_ASSERT_LOCKED(sc); devcfg_ctl = RD4(sc, ZY7_DEVCFG_CTRL); /* Clear sticky bits and set up INIT signal positive edge interrupt. */ WR4(sc, ZY7_DEVCFG_INT_STATUS, ZY7_DEVCFG_INT_ALL); WR4(sc, ZY7_DEVCFG_INT_MASK, ~ZY7_DEVCFG_INT_PCFG_INIT_PE); /* Deassert PROG_B (active low). */ devcfg_ctl |= ZY7_DEVCFG_CTRL_PCFG_PROG_B; WR4(sc, ZY7_DEVCFG_CTRL, devcfg_ctl); /* * Wait for INIT to assert. If it is already asserted, we may not get * an edge interrupt so cancel it and continue. */ if ((RD4(sc, ZY7_DEVCFG_STATUS) & ZY7_DEVCFG_STATUS_PCFG_INIT) != 0) { /* Already asserted. Cancel interrupt. */ WR4(sc, ZY7_DEVCFG_INT_MASK, ~0); } else { /* Wait for positive edge interrupt. */ err = mtx_sleep(sc, &sc->sc_mtx, PCATCH, "zy7i1", hz); if (err != 0) return (err); } /* Reassert PROG_B (active low). */ devcfg_ctl &= ~ZY7_DEVCFG_CTRL_PCFG_PROG_B; WR4(sc, ZY7_DEVCFG_CTRL, devcfg_ctl); /* Wait for INIT deasserted. This happens almost instantly. */ tries = 0; while ((RD4(sc, ZY7_DEVCFG_STATUS) & ZY7_DEVCFG_STATUS_PCFG_INIT) != 0) { if (++tries >= 100) return (EIO); DELAY(5); } /* Clear sticky bits and set up INIT positive edge interrupt. */ WR4(sc, ZY7_DEVCFG_INT_STATUS, ZY7_DEVCFG_INT_ALL); WR4(sc, ZY7_DEVCFG_INT_MASK, ~ZY7_DEVCFG_INT_PCFG_INIT_PE); /* Deassert PROG_B again. */ devcfg_ctl |= ZY7_DEVCFG_CTRL_PCFG_PROG_B; WR4(sc, ZY7_DEVCFG_CTRL, devcfg_ctl); /* * Wait for INIT asserted indicating FPGA internal initialization * is complete. */ err = mtx_sleep(sc, &sc->sc_mtx, PCATCH, "zy7i2", hz); if (err != 0) return (err); /* Clear sticky DONE bit in interrupt status. */ WR4(sc, ZY7_DEVCFG_INT_STATUS, ZY7_DEVCFG_INT_ALL); return (0); } /* Callback function for bus_dmamap_load(). */ static void zy7_dma_cb2(void *arg, bus_dma_segment_t *seg, int nsegs, int error) { if (!error && nsegs == 1) *(bus_addr_t *)arg = seg[0].ds_addr; } - static int zy7_devcfg_open(struct cdev *dev, int oflags, int devtype, struct thread *td) { struct zy7_devcfg_softc *sc = dev->si_drv1; int err; DEVCFG_SC_LOCK(sc); if (sc->is_open) { DEVCFG_SC_UNLOCK(sc); return (EBUSY); } sc->dma_map = NULL; err = bus_dma_tag_create(bus_get_dma_tag(sc->dev), 4, 0, BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, NULL, PAGE_SIZE, 1, PAGE_SIZE, 0, busdma_lock_mutex, &sc->sc_mtx, &sc->dma_tag); if (err) { DEVCFG_SC_UNLOCK(sc); return (err); } sc->is_open = 1; DEVCFG_SC_UNLOCK(sc); return (0); } static int zy7_devcfg_write(struct cdev *dev, struct uio *uio, int ioflag) { struct zy7_devcfg_softc *sc = dev->si_drv1; void *dma_mem; bus_addr_t dma_physaddr; int segsz, err; DEVCFG_SC_LOCK(sc); /* First write? Reset PL. */ if (uio->uio_offset == 0 && uio->uio_resid > 0) { zy7_devcfg_init_hw(sc); zy7_slcr_preload_pl(); err = zy7_devcfg_reset_pl(sc); if (err != 0) { DEVCFG_SC_UNLOCK(sc); return (err); } } /* Allocate dma memory and load. */ err = bus_dmamem_alloc(sc->dma_tag, &dma_mem, BUS_DMA_NOWAIT, &sc->dma_map); if (err != 0) { DEVCFG_SC_UNLOCK(sc); return (err); } err = bus_dmamap_load(sc->dma_tag, sc->dma_map, dma_mem, PAGE_SIZE, zy7_dma_cb2, &dma_physaddr, 0); if (err != 0) { bus_dmamem_free(sc->dma_tag, dma_mem, sc->dma_map); DEVCFG_SC_UNLOCK(sc); return (err); } while (uio->uio_resid > 0) { /* If DONE signal has been set, we shouldn't write anymore. */ if ((RD4(sc, ZY7_DEVCFG_INT_STATUS) & ZY7_DEVCFG_INT_PCFG_DONE) != 0) { err = EIO; break; } /* uiomove the data from user buffer to our dma map. */ segsz = MIN(PAGE_SIZE, uio->uio_resid); DEVCFG_SC_UNLOCK(sc); err = uiomove(dma_mem, segsz, uio); DEVCFG_SC_LOCK(sc); if (err != 0) break; /* Flush the cache to memory. */ bus_dmamap_sync(sc->dma_tag, sc->dma_map, BUS_DMASYNC_PREWRITE); /* Program devcfg's DMA engine. The ordering of these * register writes is critical. */ if (uio->uio_resid > segsz) WR4(sc, ZY7_DEVCFG_DMA_SRC_ADDR, (uint32_t) dma_physaddr); else WR4(sc, ZY7_DEVCFG_DMA_SRC_ADDR, (uint32_t) dma_physaddr | ZY7_DEVCFG_DMA_ADDR_WAIT_PCAP); WR4(sc, ZY7_DEVCFG_DMA_DST_ADDR, ZY7_DEVCFG_DMA_ADDR_ILLEGAL); WR4(sc, ZY7_DEVCFG_DMA_SRC_LEN, (segsz+3)/4); WR4(sc, ZY7_DEVCFG_DMA_DST_LEN, 0); /* Now clear done bit and set up DMA done interrupt. */ WR4(sc, ZY7_DEVCFG_INT_STATUS, ZY7_DEVCFG_INT_ALL); WR4(sc, ZY7_DEVCFG_INT_MASK, ~ZY7_DEVCFG_INT_DMA_DONE); /* Wait for DMA done interrupt. */ err = mtx_sleep(sc->dma_map, &sc->sc_mtx, PCATCH, "zy7dma", hz); if (err != 0) break; bus_dmamap_sync(sc->dma_tag, sc->dma_map, BUS_DMASYNC_POSTWRITE); /* Check DONE signal. */ if ((RD4(sc, ZY7_DEVCFG_INT_STATUS) & ZY7_DEVCFG_INT_PCFG_DONE) != 0) zy7_slcr_postload_pl(zy7_en_level_shifters); } bus_dmamap_unload(sc->dma_tag, sc->dma_map); bus_dmamem_free(sc->dma_tag, dma_mem, sc->dma_map); DEVCFG_SC_UNLOCK(sc); return (err); } static int zy7_devcfg_close(struct cdev *dev, int fflag, int devtype, struct thread *td) { struct zy7_devcfg_softc *sc = dev->si_drv1; DEVCFG_SC_LOCK(sc); sc->is_open = 0; bus_dma_tag_destroy(sc->dma_tag); DEVCFG_SC_UNLOCK(sc); + zy7_slcr_postload_pl(zy7_en_level_shifters); + return (0); } - static void zy7_devcfg_intr(void *arg) { struct zy7_devcfg_softc *sc = (struct zy7_devcfg_softc *)arg; uint32_t istatus, imask; DEVCFG_SC_LOCK(sc); istatus = RD4(sc, ZY7_DEVCFG_INT_STATUS); imask = ~RD4(sc, ZY7_DEVCFG_INT_MASK); /* Turn interrupt off. */ WR4(sc, ZY7_DEVCFG_INT_MASK, ~0); if ((istatus & imask) == 0) { DEVCFG_SC_UNLOCK(sc); return; } /* DMA done? */ if ((istatus & ZY7_DEVCFG_INT_DMA_DONE) != 0) wakeup(sc->dma_map); /* INIT_B positive edge? */ if ((istatus & ZY7_DEVCFG_INT_PCFG_INIT_PE) != 0) wakeup(sc); DEVCFG_SC_UNLOCK(sc); } /* zy7_devcfg_sysctl_pl_done() returns status of the PL_DONE signal. */ static int zy7_devcfg_sysctl_pl_done(SYSCTL_HANDLER_ARGS) { struct zy7_devcfg_softc *sc = zy7_devcfg_softc_p; int pl_done = 0; if (sc) { DEVCFG_SC_LOCK(sc); /* PCFG_DONE bit is sticky. Clear it before checking it. */ WR4(sc, ZY7_DEVCFG_INT_STATUS, ZY7_DEVCFG_INT_PCFG_DONE); pl_done = ((RD4(sc, ZY7_DEVCFG_INT_STATUS) & ZY7_DEVCFG_INT_PCFG_DONE) != 0); DEVCFG_SC_UNLOCK(sc); } return (sysctl_handle_int(oidp, &pl_done, 0, req)); } static int zy7_devcfg_probe(device_t dev) { if (!ofw_bus_status_okay(dev)) return (ENXIO); if (!ofw_bus_is_compatible(dev, "xlnx,zy7_devcfg")) return (ENXIO); device_set_desc(dev, "Zynq devcfg block"); return (0); } static int zy7_devcfg_detach(device_t dev); static int zy7_devcfg_attach(device_t dev) { struct zy7_devcfg_softc *sc = device_get_softc(dev); + int i; int rid, err; /* Allow only one attach. */ if (zy7_devcfg_softc_p != NULL) return (ENXIO); sc->dev = dev; DEVCFG_SC_LOCK_INIT(sc); /* Get memory resource. */ rid = 0; sc->mem_res = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &rid, RF_ACTIVE); if (sc->mem_res == NULL) { device_printf(dev, "could not allocate memory resources.\n"); zy7_devcfg_detach(dev); return (ENOMEM); } /* Allocate IRQ. */ rid = 0; sc->irq_res = bus_alloc_resource_any(dev, SYS_RES_IRQ, &rid, RF_ACTIVE); if (sc->irq_res == NULL) { device_printf(dev, "cannot allocate IRQ\n"); zy7_devcfg_detach(dev); return (ENOMEM); } /* Activate the interrupt. */ err = bus_setup_intr(dev, sc->irq_res, INTR_TYPE_MISC | INTR_MPSAFE, NULL, zy7_devcfg_intr, sc, &sc->intrhandle); if (err) { device_printf(dev, "cannot setup IRQ\n"); zy7_devcfg_detach(dev); return (err); } /* Create /dev/devcfg */ sc->sc_ctl_dev = make_dev(&zy7_devcfg_cdevsw, 0, UID_ROOT, GID_WHEEL, 0600, "devcfg"); if (sc->sc_ctl_dev == NULL) { device_printf(dev, "failed to create /dev/devcfg"); zy7_devcfg_detach(dev); return (ENXIO); } sc->sc_ctl_dev->si_drv1 = sc; zy7_devcfg_softc_p = sc; /* Unlock devcfg registers. */ WR4(sc, ZY7_DEVCFG_UNLOCK, ZY7_DEVCFG_UNLOCK_MAGIC); /* Make sure interrupts are completely disabled. */ WR4(sc, ZY7_DEVCFG_INT_STATUS, ZY7_DEVCFG_INT_ALL); WR4(sc, ZY7_DEVCFG_INT_MASK, 0xffffffff); /* Get PS_VERS for SYSCTL. */ zy7_ps_vers = (RD4(sc, ZY7_DEVCFG_MCTRL) & ZY7_DEVCFG_MCTRL_PS_VERS_MASK) >> ZY7_DEVCFG_MCTRL_PS_VERS_SHIFT; + for (i = 0; i < FCLK_NUM; i++) { + fclk_configs[i].source = zy7_pl_fclk_get_source(i); + fclk_configs[i].actual_frequency = + zy7_pl_fclk_enabled(i) ? zy7_pl_fclk_get_freq(i) : 0; + /* Initially assume actual frequency is the configure one */ + fclk_configs[i].frequency = fclk_configs[i].actual_frequency; + } + + if (zy7_devcfg_init_fclk_sysctl(sc) < 0) + device_printf(dev, "failed to initialized sysctl tree\n"); + return (0); } static int zy7_devcfg_detach(device_t dev) { struct zy7_devcfg_softc *sc = device_get_softc(dev); + + if (sc->sysctl_tree_top != NULL) { + sysctl_ctx_free(&sc->sysctl_tree); + sc->sysctl_tree_top = NULL; + } if (device_is_attached(dev)) bus_generic_detach(dev); /* Get rid of /dev/devcfg0. */ if (sc->sc_ctl_dev != NULL) destroy_dev(sc->sc_ctl_dev); /* Teardown and release interrupt. */ if (sc->irq_res != NULL) { if (sc->intrhandle) bus_teardown_intr(dev, sc->irq_res, sc->intrhandle); bus_release_resource(dev, SYS_RES_IRQ, rman_get_rid(sc->irq_res), sc->irq_res); } /* Release memory resource. */ if (sc->mem_res != NULL) bus_release_resource(dev, SYS_RES_MEMORY, rman_get_rid(sc->mem_res), sc->mem_res); zy7_devcfg_softc_p = NULL; DEVCFG_SC_LOCK_DESTROY(sc); return (0); } static device_method_t zy7_devcfg_methods[] = { /* device_if */ DEVMETHOD(device_probe, zy7_devcfg_probe), DEVMETHOD(device_attach, zy7_devcfg_attach), DEVMETHOD(device_detach, zy7_devcfg_detach), DEVMETHOD_END }; static driver_t zy7_devcfg_driver = { "zy7_devcfg", zy7_devcfg_methods, sizeof(struct zy7_devcfg_softc), }; static devclass_t zy7_devcfg_devclass; DRIVER_MODULE(zy7_devcfg, simplebus, zy7_devcfg_driver, zy7_devcfg_devclass, \ 0, 0); MODULE_DEPEND(zy7_devcfg, zy7_slcr, 1, 1, 1); Index: projects/cxl_iscsi/sys/arm/xilinx/zy7_slcr.c =================================================================== --- projects/cxl_iscsi/sys/arm/xilinx/zy7_slcr.c (revision 279716) +++ projects/cxl_iscsi/sys/arm/xilinx/zy7_slcr.c (revision 279717) @@ -1,424 +1,712 @@ /*- * Copyright (c) 2013 Thomas Skibo * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ /* * Zynq-700 SLCR driver. Provides hooks for cpu_reset and PL control stuff. * In the future, maybe MIO control, clock control, etc. could go here. * * Reference: Zynq-7000 All Programmable SoC Technical Reference Manual. * (v1.4) November 16, 2012. Xilinx doc UG585. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include struct zy7_slcr_softc { device_t dev; struct mtx sc_mtx; struct resource *mem_res; }; static struct zy7_slcr_softc *zy7_slcr_softc_p; extern void (*zynq7_cpu_reset); #define ZSLCR_LOCK(sc) mtx_lock(&(sc)->sc_mtx) #define ZSLCR_UNLOCK(sc) mtx_unlock(&(sc)->sc_mtx) #define ZSLCR_LOCK_INIT(sc) \ mtx_init(&(sc)->sc_mtx, device_get_nameunit((sc)->dev), \ "zy7_slcr", MTX_DEF) #define ZSLCR_LOCK_DESTROY(_sc) mtx_destroy(&_sc->sc_mtx); #define RD4(sc, off) (bus_read_4((sc)->mem_res, (off))) #define WR4(sc, off, val) (bus_write_4((sc)->mem_res, (off), (val))) #define ZYNQ_DEFAULT_PS_CLK_FREQUENCY 33333333 /* 33.3 Mhz */ - SYSCTL_NODE(_hw, OID_AUTO, zynq, CTLFLAG_RD, 0, "Xilinx Zynq-7000"); static char zynq_bootmode[64]; SYSCTL_STRING(_hw_zynq, OID_AUTO, bootmode, CTLFLAG_RD, zynq_bootmode, 0, "Zynq boot mode"); static char zynq_pssid[100]; SYSCTL_STRING(_hw_zynq, OID_AUTO, pssid, CTLFLAG_RD, zynq_pssid, 0, "Zynq PSS IDCODE"); static uint32_t zynq_reboot_status; SYSCTL_INT(_hw_zynq, OID_AUTO, reboot_status, CTLFLAG_RD, &zynq_reboot_status, 0, "Zynq REBOOT_STATUS register"); static int ps_clk_frequency; SYSCTL_INT(_hw_zynq, OID_AUTO, ps_clk_frequency, CTLFLAG_RD, &ps_clk_frequency, 0, "Zynq PS_CLK Frequency"); static int io_pll_frequency; SYSCTL_INT(_hw_zynq, OID_AUTO, io_pll_frequency, CTLFLAG_RD, &io_pll_frequency, 0, "Zynq IO PLL Frequency"); static int arm_pll_frequency; SYSCTL_INT(_hw_zynq, OID_AUTO, arm_pll_frequency, CTLFLAG_RD, &arm_pll_frequency, 0, "Zynq ARM PLL Frequency"); static int ddr_pll_frequency; SYSCTL_INT(_hw_zynq, OID_AUTO, ddr_pll_frequency, CTLFLAG_RD, &ddr_pll_frequency, 0, "Zynq DDR PLL Frequency"); static void zy7_slcr_unlock(struct zy7_slcr_softc *sc) { /* Unlock SLCR with magic number. */ WR4(sc, ZY7_SLCR_UNLOCK, ZY7_SLCR_UNLOCK_MAGIC); } static void zy7_slcr_lock(struct zy7_slcr_softc *sc) { /* Lock SLCR with magic number. */ WR4(sc, ZY7_SLCR_LOCK, ZY7_SLCR_LOCK_MAGIC); } - static void zy7_slcr_cpu_reset(void) { struct zy7_slcr_softc *sc = zy7_slcr_softc_p; /* Unlock SLCR registers. */ zy7_slcr_unlock(sc); /* This has something to do with a work-around so the fsbl will load * the bitstream after soft-reboot. It's very important. */ WR4(sc, ZY7_SLCR_REBOOT_STAT, RD4(sc, ZY7_SLCR_REBOOT_STAT) & 0xf0ffffff); /* Soft reset */ WR4(sc, ZY7_SLCR_PSS_RST_CTRL, ZY7_SLCR_PSS_RST_CTRL_SOFT_RESET); for (;;) ; } /* Assert PL resets and disable level shifters in preparation of programming * the PL (FPGA) section. Called from zy7_devcfg.c. */ void zy7_slcr_preload_pl(void) { struct zy7_slcr_softc *sc = zy7_slcr_softc_p; if (!sc) return; ZSLCR_LOCK(sc); /* Unlock SLCR registers. */ zy7_slcr_unlock(sc); /* Assert top level output resets. */ WR4(sc, ZY7_SLCR_FPGA_RST_CTRL, ZY7_SLCR_FPGA_RST_CTRL_RST_ALL); /* Disable all level shifters. */ WR4(sc, ZY7_SLCR_LVL_SHFTR_EN, 0); /* Lock SLCR registers. */ zy7_slcr_lock(sc); ZSLCR_UNLOCK(sc); } /* After PL configuration, enable level shifters and deassert top-level * PL resets. Called from zy7_devcfg.c. Optionally, the level shifters * can be left disabled but that's rare of an FPGA application. That option * is controled by a sysctl in the devcfg driver. */ void zy7_slcr_postload_pl(int en_level_shifters) { struct zy7_slcr_softc *sc = zy7_slcr_softc_p; if (!sc) return; ZSLCR_LOCK(sc); /* Unlock SLCR registers. */ zy7_slcr_unlock(sc); if (en_level_shifters) /* Enable level shifters. */ WR4(sc, ZY7_SLCR_LVL_SHFTR_EN, ZY7_SLCR_LVL_SHFTR_EN_ALL); /* Deassert top level output resets. */ WR4(sc, ZY7_SLCR_FPGA_RST_CTRL, 0); /* Lock SLCR registers. */ zy7_slcr_lock(sc); ZSLCR_UNLOCK(sc); } /* Override cgem_set_refclk() in gigabit ethernet driver * (sys/dev/cadence/if_cgem.c). This function is called to * request a change in the gem's reference clock speed. */ int cgem_set_ref_clk(int unit, int frequency) { struct zy7_slcr_softc *sc = zy7_slcr_softc_p; int div0, div1; if (!sc) return (-1); /* Find suitable divisor pairs. Round result to nearest khz * to test for match. */ for (div1 = 1; div1 <= ZY7_SLCR_GEM_CLK_CTRL_DIVISOR1_MAX; div1++) { div0 = (io_pll_frequency + div1 * frequency / 2) / div1 / frequency; if (div0 > 0 && div0 <= ZY7_SLCR_GEM_CLK_CTRL_DIVISOR_MAX && ((io_pll_frequency / div0 / div1) + 500) / 1000 == (frequency + 500) / 1000) break; } if (div1 > ZY7_SLCR_GEM_CLK_CTRL_DIVISOR1_MAX) return (-1); ZSLCR_LOCK(sc); /* Unlock SLCR registers. */ zy7_slcr_unlock(sc); /* Modify GEM reference clock. */ WR4(sc, unit ? ZY7_SLCR_GEM1_CLK_CTRL : ZY7_SLCR_GEM0_CLK_CTRL, (div1 << ZY7_SLCR_GEM_CLK_CTRL_DIVISOR1_SHIFT) | (div0 << ZY7_SLCR_GEM_CLK_CTRL_DIVISOR_SHIFT) | ZY7_SLCR_GEM_CLK_CTRL_SRCSEL_IO_PLL | ZY7_SLCR_GEM_CLK_CTRL_CLKACT); /* Lock SLCR registers. */ zy7_slcr_lock(sc); ZSLCR_UNLOCK(sc); return (0); +} + +/* + * PL clocks management function + */ +int +zy7_pl_fclk_set_source(int unit, int source) +{ + struct zy7_slcr_softc *sc = zy7_slcr_softc_p; + uint32_t reg; + + if (!sc) + return (-1); + + ZSLCR_LOCK(sc); + + /* Unlock SLCR registers. */ + zy7_slcr_unlock(sc); + + /* Modify FPGAx source. */ + reg = RD4(sc, ZY7_SLCR_FPGA_CLK_CTRL(unit)); + reg &= ~(ZY7_SLCR_FPGA_CLK_CTRL_SRCSEL_MASK); + reg |= (source << ZY7_SLCR_FPGA_CLK_CTRL_SRCSEL_SHIFT); + WR4(sc, ZY7_SLCR_FPGA_CLK_CTRL(unit), reg); + + /* Lock SLCR registers. */ + zy7_slcr_lock(sc); + + ZSLCR_UNLOCK(sc); + + return (0); +} + +int +zy7_pl_fclk_get_source(int unit) +{ + struct zy7_slcr_softc *sc = zy7_slcr_softc_p; + uint32_t reg; + int source; + + if (!sc) + return (-1); + + ZSLCR_LOCK(sc); + + /* Modify GEM reference clock. */ + reg = RD4(sc, ZY7_SLCR_FPGA_CLK_CTRL(unit)); + source = (reg & ZY7_SLCR_FPGA_CLK_CTRL_SRCSEL_MASK) >> + ZY7_SLCR_FPGA_CLK_CTRL_SRCSEL_SHIFT; + + /* ZY7_PL_FCLK_SRC_IO is actually b0x */ + if ((source & 2) == 0) + source = ZY7_PL_FCLK_SRC_IO; + + ZSLCR_UNLOCK(sc); + + return (source); +} + +int +zy7_pl_fclk_set_freq(int unit, int frequency) +{ + struct zy7_slcr_softc *sc = zy7_slcr_softc_p; + int div0, div1; + int base_frequency; + uint32_t reg; + int source; + + if (!sc) + return (-1); + + source = zy7_pl_fclk_get_source(unit); + switch (source) { + case ZY7_PL_FCLK_SRC_IO: + base_frequency = io_pll_frequency; + break; + + case ZY7_PL_FCLK_SRC_ARM: + base_frequency = arm_pll_frequency; + break; + + case ZY7_PL_FCLK_SRC_DDR: + base_frequency = ddr_pll_frequency; + break; + + default: + return (-1); + } + + /* Find suitable divisor pairs. Round result to nearest khz + * to test for match. + */ + for (div1 = 1; div1 <= ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR_MAX; div1++) { + div0 = (base_frequency + div1 * frequency / 2) / + div1 / frequency; + if (div0 > 0 && div0 <= ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR_MAX && + ((base_frequency / div0 / div1) + 500) / 1000 == + (frequency + 500) / 1000) + break; + } + + if (div1 > ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR_MAX) + return (-1); + + ZSLCR_LOCK(sc); + + /* Unlock SLCR registers. */ + zy7_slcr_unlock(sc); + + /* Modify FPGAx reference clock. */ + reg = RD4(sc, ZY7_SLCR_FPGA_CLK_CTRL(unit)); + reg &= ~(ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR1_MASK | + ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR0_MASK); + reg |= (div1 << ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR1_SHIFT) | + (div0 << ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR0_SHIFT); + WR4(sc, ZY7_SLCR_FPGA_CLK_CTRL(unit), reg); + + /* Lock SLCR registers. */ + zy7_slcr_lock(sc); + + ZSLCR_UNLOCK(sc); + + return (base_frequency / div0 / div1); +} + +int +zy7_pl_fclk_get_freq(int unit) +{ + struct zy7_slcr_softc *sc = zy7_slcr_softc_p; + int div0, div1; + int base_frequency; + int frequency; + uint32_t reg; + int source; + + if (!sc) + return (-1); + + source = zy7_pl_fclk_get_source(unit); + switch (source) { + case ZY7_PL_FCLK_SRC_IO: + base_frequency = io_pll_frequency; + break; + + case ZY7_PL_FCLK_SRC_ARM: + base_frequency = arm_pll_frequency; + break; + + case ZY7_PL_FCLK_SRC_DDR: + base_frequency = ddr_pll_frequency; + break; + + default: + return (-1); + } + + ZSLCR_LOCK(sc); + + /* Modify FPGAx reference clock. */ + reg = RD4(sc, ZY7_SLCR_FPGA_CLK_CTRL(unit)); + div1 = (reg & ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR1_MASK) >> + ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR1_SHIFT; + div0 = (reg & ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR0_MASK) >> + ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR0_SHIFT; + + ZSLCR_UNLOCK(sc); + + if (div0 == 0) + div0 = 1; + + if (div1 == 0) + div1 = 1; + + frequency = (base_frequency / div0 / div1); + /* Round to KHz */ + frequency = (frequency + 500) / 1000; + frequency = frequency * 1000; + + return (frequency); +} + +int +zy7_pl_fclk_enable(int unit) +{ + struct zy7_slcr_softc *sc = zy7_slcr_softc_p; + + if (!sc) + return (-1); + + ZSLCR_LOCK(sc); + + /* Unlock SLCR registers. */ + zy7_slcr_unlock(sc); + + WR4(sc, ZY7_SLCR_FPGA_THR_CTRL(unit), 0); + WR4(sc, ZY7_SLCR_FPGA_THR_CNT(unit), 0); + + /* Lock SLCR registers. */ + zy7_slcr_lock(sc); + + ZSLCR_UNLOCK(sc); + + return (0); +} + +int +zy7_pl_fclk_disable(int unit) +{ + struct zy7_slcr_softc *sc = zy7_slcr_softc_p; + + if (!sc) + return (-1); + + ZSLCR_LOCK(sc); + + /* Unlock SLCR registers. */ + zy7_slcr_unlock(sc); + + WR4(sc, ZY7_SLCR_FPGA_THR_CTRL(unit), 0); + WR4(sc, ZY7_SLCR_FPGA_THR_CNT(unit), 1); + + /* Lock SLCR registers. */ + zy7_slcr_lock(sc); + + ZSLCR_UNLOCK(sc); + + return (0); +} + +int +zy7_pl_fclk_enabled(int unit) +{ + struct zy7_slcr_softc *sc = zy7_slcr_softc_p; + uint32_t reg; + + if (!sc) + return (-1); + + ZSLCR_LOCK(sc); + reg = RD4(sc, ZY7_SLCR_FPGA_THR_CNT(unit)); + ZSLCR_UNLOCK(sc); + + return !(reg & 1); +} + +int +zy7_pl_level_shifters_enabled() +{ + struct zy7_slcr_softc *sc = zy7_slcr_softc_p; + + uint32_t reg; + + if (!sc) + return (-1); + + ZSLCR_LOCK(sc); + reg = RD4(sc, ZY7_SLCR_LVL_SHFTR_EN); + ZSLCR_UNLOCK(sc); + + return (reg == ZY7_SLCR_LVL_SHFTR_EN_ALL); +} + +void +zy7_pl_level_shifters_enable() +{ + struct zy7_slcr_softc *sc = zy7_slcr_softc_p; + + if (!sc) + return; + + ZSLCR_LOCK(sc); + zy7_slcr_unlock(sc); + WR4(sc, ZY7_SLCR_LVL_SHFTR_EN, ZY7_SLCR_LVL_SHFTR_EN_ALL); + zy7_slcr_lock(sc); + ZSLCR_UNLOCK(sc); +} + +void +zy7_pl_level_shifters_disable() +{ + struct zy7_slcr_softc *sc = zy7_slcr_softc_p; + + if (!sc) + return; + + ZSLCR_LOCK(sc); + zy7_slcr_unlock(sc); + WR4(sc, ZY7_SLCR_LVL_SHFTR_EN, 0); + zy7_slcr_lock(sc); + ZSLCR_UNLOCK(sc); } static int zy7_slcr_probe(device_t dev) { if (!ofw_bus_status_okay(dev)) return (ENXIO); if (!ofw_bus_is_compatible(dev, "xlnx,zy7_slcr")) return (ENXIO); device_set_desc(dev, "Zynq-7000 slcr block"); return (0); } static int zy7_slcr_attach(device_t dev) { struct zy7_slcr_softc *sc = device_get_softc(dev); int rid; phandle_t node; pcell_t cell; uint32_t bootmode; uint32_t pss_idcode; uint32_t arm_pll_ctrl; uint32_t ddr_pll_ctrl; uint32_t io_pll_ctrl; static char *bootdev_names[] = { "JTAG", "Quad-SPI", "NOR", "(3?)", "NAND", "SD Card", "(6?)", "(7?)" }; /* Allow only one attach. */ if (zy7_slcr_softc_p != NULL) return (ENXIO); sc->dev = dev; ZSLCR_LOCK_INIT(sc); /* Get memory resource. */ rid = 0; sc->mem_res = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &rid, RF_ACTIVE); if (sc->mem_res == NULL) { device_printf(dev, "could not allocate memory resources.\n"); return (ENOMEM); } /* Hook up cpu_reset. */ zy7_slcr_softc_p = sc; zynq7_cpu_reset = zy7_slcr_cpu_reset; /* Read info and set sysctls. */ bootmode = RD4(sc, ZY7_SLCR_BOOT_MODE); snprintf(zynq_bootmode, sizeof(zynq_bootmode), "0x%x: boot device: %s", bootmode, bootdev_names[bootmode & ZY7_SLCR_BOOT_MODE_BOOTDEV_MASK]); pss_idcode = RD4(sc, ZY7_SLCR_PSS_IDCODE); snprintf(zynq_pssid, sizeof(zynq_pssid), "0x%x: manufacturer: 0x%x device: 0x%x " "family: 0x%x sub-family: 0x%x rev: 0x%x", pss_idcode, (pss_idcode & ZY7_SLCR_PSS_IDCODE_MNFR_ID_MASK) >> ZY7_SLCR_PSS_IDCODE_MNFR_ID_SHIFT, (pss_idcode & ZY7_SLCR_PSS_IDCODE_DEVICE_MASK) >> ZY7_SLCR_PSS_IDCODE_DEVICE_SHIFT, (pss_idcode & ZY7_SLCR_PSS_IDCODE_FAMILY_MASK) >> ZY7_SLCR_PSS_IDCODE_FAMILY_SHIFT, (pss_idcode & ZY7_SLCR_PSS_IDCODE_SUB_FAMILY_MASK) >> ZY7_SLCR_PSS_IDCODE_SUB_FAMILY_SHIFT, (pss_idcode & ZY7_SLCR_PSS_IDCODE_REVISION_MASK) >> ZY7_SLCR_PSS_IDCODE_REVISION_SHIFT); zynq_reboot_status = RD4(sc, ZY7_SLCR_REBOOT_STAT); /* Derive PLL frequencies from PS_CLK. */ node = ofw_bus_get_node(dev); if (OF_getprop(node, "clock-frequency", &cell, sizeof(cell)) > 0) ps_clk_frequency = fdt32_to_cpu(cell); else ps_clk_frequency = ZYNQ_DEFAULT_PS_CLK_FREQUENCY; arm_pll_ctrl = RD4(sc, ZY7_SLCR_ARM_PLL_CTRL); ddr_pll_ctrl = RD4(sc, ZY7_SLCR_DDR_PLL_CTRL); io_pll_ctrl = RD4(sc, ZY7_SLCR_IO_PLL_CTRL); /* Determine ARM PLL frequency. */ if (((arm_pll_ctrl & ZY7_SLCR_PLL_CTRL_BYPASS_QUAL) == 0 && (arm_pll_ctrl & ZY7_SLCR_PLL_CTRL_BYPASS_FORCE) != 0) || ((arm_pll_ctrl & ZY7_SLCR_PLL_CTRL_BYPASS_QUAL) != 0 && (bootmode & ZY7_SLCR_BOOT_MODE_PLL_BYPASS) != 0)) /* PLL is bypassed. */ arm_pll_frequency = ps_clk_frequency; else arm_pll_frequency = ps_clk_frequency * ((arm_pll_ctrl & ZY7_SLCR_PLL_CTRL_FDIV_MASK) >> ZY7_SLCR_PLL_CTRL_FDIV_SHIFT); /* Determine DDR PLL frequency. */ if (((ddr_pll_ctrl & ZY7_SLCR_PLL_CTRL_BYPASS_QUAL) == 0 && (ddr_pll_ctrl & ZY7_SLCR_PLL_CTRL_BYPASS_FORCE) != 0) || ((ddr_pll_ctrl & ZY7_SLCR_PLL_CTRL_BYPASS_QUAL) != 0 && (bootmode & ZY7_SLCR_BOOT_MODE_PLL_BYPASS) != 0)) /* PLL is bypassed. */ ddr_pll_frequency = ps_clk_frequency; else ddr_pll_frequency = ps_clk_frequency * ((ddr_pll_ctrl & ZY7_SLCR_PLL_CTRL_FDIV_MASK) >> ZY7_SLCR_PLL_CTRL_FDIV_SHIFT); /* Determine IO PLL frequency. */ if (((io_pll_ctrl & ZY7_SLCR_PLL_CTRL_BYPASS_QUAL) == 0 && (io_pll_ctrl & ZY7_SLCR_PLL_CTRL_BYPASS_FORCE) != 0) || ((io_pll_ctrl & ZY7_SLCR_PLL_CTRL_BYPASS_QUAL) != 0 && (bootmode & ZY7_SLCR_BOOT_MODE_PLL_BYPASS) != 0)) /* PLL is bypassed. */ io_pll_frequency = ps_clk_frequency; else io_pll_frequency = ps_clk_frequency * ((io_pll_ctrl & ZY7_SLCR_PLL_CTRL_FDIV_MASK) >> ZY7_SLCR_PLL_CTRL_FDIV_SHIFT); /* Lock SLCR registers. */ zy7_slcr_lock(sc); return (0); } static int zy7_slcr_detach(device_t dev) { struct zy7_slcr_softc *sc = device_get_softc(dev); bus_generic_detach(dev); /* Release memory resource. */ if (sc->mem_res != NULL) bus_release_resource(dev, SYS_RES_MEMORY, rman_get_rid(sc->mem_res), sc->mem_res); zy7_slcr_softc_p = NULL; zynq7_cpu_reset = NULL; ZSLCR_LOCK_DESTROY(sc); return (0); } static device_method_t zy7_slcr_methods[] = { /* device_if */ DEVMETHOD(device_probe, zy7_slcr_probe), DEVMETHOD(device_attach, zy7_slcr_attach), DEVMETHOD(device_detach, zy7_slcr_detach), DEVMETHOD_END }; static driver_t zy7_slcr_driver = { "zy7_slcr", zy7_slcr_methods, sizeof(struct zy7_slcr_softc), }; static devclass_t zy7_slcr_devclass; DRIVER_MODULE(zy7_slcr, simplebus, zy7_slcr_driver, zy7_slcr_devclass, 0, 0); MODULE_VERSION(zy7_slcr, 1); Index: projects/cxl_iscsi/sys/arm/xilinx/zy7_slcr.h =================================================================== --- projects/cxl_iscsi/sys/arm/xilinx/zy7_slcr.h (revision 279716) +++ projects/cxl_iscsi/sys/arm/xilinx/zy7_slcr.h (revision 279717) @@ -1,292 +1,318 @@ /*- * Copyright (c) 2013 Thomas Skibo * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ /* * Defines for Zynq-7000 SLCR registers. * * Most of these registers are initialized by the First Stage Boot * Loader and are not modified by the kernel. * * Reference: Zynq-7000 All Programmable SoC Technical Reference Manual. * (v1.4) November 16, 2012. Xilinx doc UG585. SLCR register definitions * are in appendix B.28. */ - #ifndef _ZY7_SLCR_H_ #define _ZY7_SLCR_H_ #define ZY7_SCLR_SCL 0x0000 #define ZY7_SLCR_LOCK 0x0004 #define ZY7_SLCR_LOCK_MAGIC 0x767b #define ZY7_SLCR_UNLOCK 0x0008 #define ZY7_SLCR_UNLOCK_MAGIC 0xdf0d #define ZY7_SLCR_LOCKSTA 0x000c /* PLL controls. */ #define ZY7_SLCR_ARM_PLL_CTRL 0x0100 #define ZY7_SLCR_DDR_PLL_CTRL 0x0104 #define ZY7_SLCR_IO_PLL_CTRL 0x0108 #define ZY7_SLCR_PLL_CTRL_RESET (1<<0) #define ZY7_SLCR_PLL_CTRL_PWRDWN (1<<1) #define ZY7_SLCR_PLL_CTRL_BYPASS_QUAL (1<<3) #define ZY7_SLCR_PLL_CTRL_BYPASS_FORCE (1<<4) #define ZY7_SLCR_PLL_CTRL_FDIV_SHIFT 12 #define ZY7_SLCR_PLL_CTRL_FDIV_MASK (0x7f<<12) #define ZY7_SLCR_PLL_STATUS 0x010c #define ZY7_SLCR_PLL_STAT_ARM_PLL_LOCK (1<<0) #define ZY7_SLCR_PLL_STAT_DDR_PLL_LOCK (1<<1) #define ZY7_SLCR_PLL_STAT_IO_PLL_LOCK (1<<2) #define ZY7_SLCR_PLL_STAT_ARM_PLL_STABLE (1<<3) #define ZY7_SLCR_PLL_STAT_DDR_PLL_STABLE (1<<4) #define ZY7_SLCR_PLL_STAT_IO_PLL_STABLE (1<<5) #define ZY7_SLCR_ARM_PLL_CFG 0x0110 #define ZY7_SLCR_DDR_PLL_CFG 0x0114 #define ZY7_SLCR_IO_PLL_CFG 0x0118 #define ZY7_SLCR_PLL_CFG_RES_SHIFT 4 #define ZY7_SLCR_PLL_CFG_RES_MASK (0xf<<4) #define ZY7_SLCR_PLL_CFG_PLL_CP_SHIFT 8 #define ZY7_SLCR_PLL_CFG_PLL_CP_MASK (0xf<<8) #define ZY7_SLCR_PLL_CFG_LOCK_CNT_SHIFT 12 #define ZY7_SLCR_PLL_CFG_LOCK_CNT_MASK (0x3ff<<12) /* Clock controls. */ #define ZY7_SLCR_ARM_CLK_CTRL 0x0120 #define ZY7_SLCR_ARM_CLK_CTRL_CPU_PERI_CLKACT (1<<28) #define ZY7_SLCR_ARM_CLK_CTRL_CPU_1XCLKACT (1<<27) #define ZY7_SLCR_ARM_CLK_CTRL_CPU_2XCLKACT (1<<26) #define ZY7_SLCR_ARM_CLK_CTRL_CPU_3OR2XCLKACT (1<<25) #define ZY7_SLCR_ARM_CLK_CTRL_CPU_6OR4XCLKACT (1<<24) #define ZY7_SLCR_ARM_CLK_CTRL_SRCSEL_MASK (3<<4) #define ZY7_SLCR_ARM_CLK_CTRL_SRCSEL_ARM_PLL (0<<4) #define ZY7_SLCR_ARM_CLK_CTRL_SRCSEL_DDR_PLL (2<<4) #define ZY7_SLCR_ARM_CLK_CTRL_SRCSEL_IO_PLL (3<<4) #define ZY7_SLCR_ARM_CLK_CTRL_DIVISOR_SHIFT 8 #define ZY7_SLCR_ARM_CLK_CTRL_DIVISOR_MASK (0x3f<<8) #define ZY7_SLCR_DDR_CLK_CTRL 0x0124 #define ZY7_SLCR_DDR_CLK_CTRL_2XCLK_DIV_SHIFT 26 #define ZY7_SLCR_DDR_CLK_CTRL_2XCLK_DIV_MASK (0x3f<<26) #define ZY7_SLCR_DDR_CLK_CTRL_3XCLK_DIV_SHIFT 20 #define ZY7_SLCR_DDR_CLK_CTRL_3XCLK_DIV_MASK (0x3f<<20) #define ZY7_SLCR_DDR_CLK_CTRL_2XCLKACT (1<<1) #define ZY7_SLCR_DDR_CLK_CTRL_3XCLKACT (1<<0) #define ZY7_SLCR_DCI_CLK_CTRL 0x0128 #define ZY7_SLCR_DCI_CLK_CTRL_DIVISOR1_SHIFT 20 #define ZY7_SLCR_DCI_CLK_CTRL_DIVISOR1_MASK (0x3f<<20) #define ZY7_SLCR_DCI_CLK_CTRL_DIVISOR0_SHIFT 8 #define ZY7_SLCR_DCI_CLK_CTRL_DIVISOR0_MASK (0x3f<<8) #define ZY7_SLCR_DCI_CLK_CTRL_CLKACT (1<<0) #define ZY7_SLCR_APER_CLK_CTRL 0x012c /* amba periph clk ctrl */ #define ZY7_SLCR_APER_CLK_CTRL_SMC_CPU_1XCLKACT (1<<24) #define ZY7_SLCR_APER_CLK_CTRL_LQSPI_CPU_1XCLKACT (1<<23) #define ZY7_SLCR_APER_CLK_CTRL_GPIO_CPU_1XCLKACT (1<<22) #define ZY7_SLCR_APER_CLK_CTRL_UART1_CPU_1XCLKACT (1<<21) #define ZY7_SLCR_APER_CLK_CTRL_UART0_CPU_1XCLKACT (1<<20) #define ZY7_SLCR_APER_CLK_CTRL_I2C1_CPU_1XCLKACT (1<<19) #define ZY7_SLCR_APER_CLK_CTRL_I2C0_CPU_1XCLKACT (1<<18) #define ZY7_SLCR_APER_CLK_CTRL_CAN1_CPU_1XCLKACT (1<<17) #define ZY7_SLCR_APER_CLK_CTRL_CAN0_CPU_1XCLKACT (1<<16) #define ZY7_SLCR_APER_CLK_CTRL_SPI1_CPU_1XCLKACT (1<<15) #define ZY7_SLCR_APER_CLK_CTRL_SPI0_CPU_1XCLKACT (1<<14) #define ZY7_SLCR_APER_CLK_CTRL_SDI1_CPU_1XCLKACT (1<<11) #define ZY7_SLCR_APER_CLK_CTRL_SDI0_CPU_1XCLKACT (1<<10) #define ZY7_SLCR_APER_CLK_CTRL_GEM1_CPU_1XCLKACT (1<<7) #define ZY7_SLCR_APER_CLK_CTRL_GEM0_CPU_1XCLKACT (1<<6) #define ZY7_SLCR_APER_CLK_CTRL_USB1_CPU_1XCLKACT (1<<3) #define ZY7_SLCR_APER_CLK_CTRL_USB0_CPU_1XCLKACT (1<<2) #define ZY7_SLCR_APER_CLK_CTRL_DMA_CPU_1XCLKACT (1<<0) #define ZY7_SLCR_USB0_CLK_CTRL 0x0130 #define ZY7_SLCR_USB1_CLK_CTRL 0x0134 #define ZY7_SLCR_GEM0_RCLK_CTRL 0x0138 #define ZY7_SLCR_GEM1_RCLK_CTRL 0x013c #define ZY7_SLCR_GEM0_CLK_CTRL 0x0140 #define ZY7_SLCR_GEM1_CLK_CTRL 0x0144 #define ZY7_SLCR_GEM_CLK_CTRL_DIVISOR1_MASK (0x3f<<20) #define ZY7_SLCR_GEM_CLK_CTRL_DIVISOR1_SHIFT 20 #define ZY7_SLCR_GEM_CLK_CTRL_DIVISOR1_MAX 0x3f #define ZY7_SLCR_GEM_CLK_CTRL_DIVISOR_MASK (0x3f<<8) #define ZY7_SLCR_GEM_CLK_CTRL_DIVISOR_SHIFT 8 #define ZY7_SLCR_GEM_CLK_CTRL_DIVISOR_MAX 0x3f #define ZY7_SLCR_GEM_CLK_CTRL_SRCSEL_MASK (7<<4) #define ZY7_SLCR_GEM_CLK_CTRL_SRCSEL_IO_PLL (0<<4) #define ZY7_SLCR_GEM_CLK_CTRL_SRCSEL_ARM_PLL (2<<4) #define ZY7_SLCR_GEM_CLK_CTRL_SRCSEL_DDR_PLL (3<<4) #define ZY7_SLCR_GEM_CLK_CTRL_SRCSEL_EMIO_CLK (4<<4) #define ZY7_SLCR_GEM_CLK_CTRL_CLKACT 1 #define ZY7_SLCR_SMC_CLK_CTRL 0x0148 #define ZY7_SLCR_LQSPI_CLK_CTRL 0x014c #define ZY7_SLCR_SDIO_CLK_CTRL 0x0150 #define ZY7_SLCR_UART_CLK_CTRL 0x0154 #define ZY7_SLCR_SPI_CLK_CTRL 0x0158 #define ZY7_SLCR_CAN_CLK_CTRL 0x015c #define ZY7_SLCR_CAN_MIOCLK_CTRL 0x0160 #define ZY7_SLCR_DBG_CLK_CTRL 0x0164 #define ZY7_SLCR_PCAP_CLK_CTRL 0x0168 #define ZY7_SLCR_TOPSW_CLK_CTRL 0x016c /* central intercnn clk ctrl */ -#define ZY7_SLCR_FPGA0_CLK_CTRL 0x0170 -#define ZY7_SLCR_FPGA1_CLK_CTRL 0x0180 -#define ZY7_SLCR_FPGA2_CLK_CTRL 0x0190 -#define ZY7_SLCR_FPGA3_CLK_CTRL 0x01a0 +#define ZY7_SLCR_FPGA_CLK_CTRL(unit) (0x0170 + 0x10*(unit)) +#define ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR1_SHIFT 20 +#define ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR1_MASK (0x3f << 20) +#define ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR0_SHIFT 8 +#define ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR0_MASK (0x3f << 8) +#define ZY7_SLCR_FPGA_CLK_CTRL_DIVISOR_MAX 0x3f +#define ZY7_SLCR_FPGA_CLK_CTRL_SRCSEL_SHIFT 4 +#define ZY7_SLCR_FPGA_CLK_CTRL_SRCSEL_MASK (3 << 4) +#define ZY7_SLCR_FPGA_THR_CTRL(unit) (0x0174 + 0x10*(unit)) +#define ZY7_SLCR_FPGA_THR_CTRL_CNT_RST (1 << 1) +#define ZY7_SLCR_FPGA_THR_CTRL_CPU_START (1 << 0) +#define ZY7_SLCR_FPGA_THR_CNT(unit) (0x0178 + 0x10*(unit)) +#define ZY7_SLCR_FPGA_THR_STA(unit) (0x017c + 0x10*(unit)) #define ZY7_SLCR_CLK_621_TRUE 0x01c4 /* cpu clock ratio mode */ /* Reset controls. */ #define ZY7_SLCR_PSS_RST_CTRL 0x0200 #define ZY7_SLCR_PSS_RST_CTRL_SOFT_RESET (1<<0) #define ZY7_SLCR_DDR_RST_CTRL 0x0204 #define ZY7_SLCR_TOPSW_RST_CTRL 0x0208 #define ZY7_SLCR_DMAC_RST_CTRL 0x020c #define ZY7_SLCR_USB_RST_CTRL 0x0210 #define ZY7_SLCR_GEM_RST_CTRL 0x0214 #define ZY7_SLCR_SDIO_RST_CTRL 0x0218 #define ZY7_SLCR_SPI_RST_CTRL 0x021c #define ZY7_SLCR_CAN_RST_CTRL 0x0220 #define ZY7_SLCR_I2C_RST_CTRL 0x0224 #define ZY7_SLCR_UART_RST_CTRL 0x0228 #define ZY7_SLCR_GPIO_RST_CTRL 0x022c #define ZY7_SLCR_LQSPI_RST_CTRL 0x0230 #define ZY7_SLCR_SMC_RST_CTRL 0x0234 #define ZY7_SLCR_OCM_RST_CTRL 0x0238 #define ZY7_SLCR_DEVCI_RST_CTRL 0x023c #define ZY7_SLCR_FPGA_RST_CTRL 0x0240 #define ZY7_SLCR_FPGA_RST_CTRL_FPGA3_OUT_RST (1<<3) #define ZY7_SLCR_FPGA_RST_CTRL_FPGA2_OUT_RST (1<<2) #define ZY7_SLCR_FPGA_RST_CTRL_FPGA1_OUT_RST (1<<1) #define ZY7_SLCR_FPGA_RST_CTRL_FPGA0_OUT_RST (1<<0) #define ZY7_SLCR_FPGA_RST_CTRL_RST_ALL 0xf #define ZY7_SLCR_A9_CPU_RST_CTRL 0x0244 #define ZY7_SLCR_RS_AWDT_CTRL 0x024c #define ZY7_SLCR_REBOOT_STAT 0x0258 #define ZY7_SLCR_REBOOT_STAT_STATE_MASK (0xff<<24) #define ZY7_SLCR_REBOOT_STAT_POR (1<<22) #define ZY7_SLCR_REBOOT_STAT_SRST_B (1<<21) #define ZY7_SLCR_REBOOT_STAT_DBG_RST (1<<20) #define ZY7_SLCR_REBOOT_STAT_SLC_RST (1<<19) #define ZY7_SLCR_REBOOT_STAT_AWDT1_RST (1<<18) #define ZY7_SLCR_REBOOT_STAT_AWDT0_RST (1<<17) #define ZY7_SLCR_REBOOT_STAT_SWDT_RST (1<<16) #define ZY7_SLCR_REBOOT_STAT_BOOTROM_ERR_CODE_MASK (0xffff) #define ZY7_SLCR_BOOT_MODE 0x025c #define ZY7_SLCR_BOOT_MODE_PLL_BYPASS (1<<4) #define ZY7_SLCR_BOOT_MODE_JTAG_INDEP (1<<3) #define ZY7_SLCR_BOOT_MODE_BOOTDEV_MASK 7 #define ZY7_SLCR_BOOT_MODE_BOOTDEV_JTAG 0 #define ZY7_SLCR_BOOT_MODE_BOOTDEV_QUAD_SPI 1 #define ZY7_SLCR_BOOT_MODE_BOOTDEV_NOR 2 #define ZY7_SLCR_BOOT_MODE_BOOTDEV_NAND 4 #define ZY7_SLCR_BOOT_MODE_BOOTDEV_SD_CARD 5 #define ZY7_SLCR_APU_CTRL 0x0300 #define ZY7_SLCR_WDT_CLK_SEL 0x0304 #define ZY7_SLCR_PSS_IDCODE 0x0530 #define ZY7_SLCR_PSS_IDCODE_REVISION_MASK (0xf<<28) #define ZY7_SLCR_PSS_IDCODE_REVISION_SHIFT 28 #define ZY7_SLCR_PSS_IDCODE_FAMILY_MASK (0x7f<<21) #define ZY7_SLCR_PSS_IDCODE_FAMILY_SHIFT 21 #define ZY7_SLCR_PSS_IDCODE_SUB_FAMILY_MASK (0xf<<17) #define ZY7_SLCR_PSS_IDCODE_SUB_FAMILY_SHIFT 17 #define ZY7_SLCR_PSS_IDCODE_DEVICE_MASK (0x1f<<12) #define ZY7_SLCR_PSS_IDCODE_DEVICE_SHIFT 12 #define ZY7_SLCR_PSS_IDCODE_MNFR_ID_MASK (0x7ff<<1) #define ZY7_SLCR_PSS_IDCODE_MNFR_ID_SHIFT 1 #define ZY7_SLCR_DDR_URGENT 0x0600 #define ZY7_SLCR_DDR_CAL_START 0x060c #define ZY7_SLCR_DDR_REF_START 0x0614 #define ZY7_SLCR_DDR_CMD_STA 0x0618 #define ZY7_SLCR_DDR_URGENT_SEL 0x061c #define ZY7_SLCR_DDR_DFI_STATUS 0x0620 /* MIO Pin controls */ #define ZY7_SLCR_MIO_PIN(n) (0x0700+(n)*4) /* 0-53 */ #define ZY7_SLCR_MIO_PIN_RCVR_DIS (1<<13) #define ZY7_SLCR_MIO_PIN_PULLUP_EN (1<<12) #define ZY7_SLCR_MIO_PIN_IO_TYPE_MASK (7<<9) #define ZY7_SLCR_MIO_PIN_IO_TYPE_LVTTL (0<<9) #define ZY7_SLCR_MIO_PIN_IO_TYPE_LVCMOS18 (1<<9) #define ZY7_SLCR_MIO_PIN_IO_TYPE_LVCMOS25 (2<<9) #define ZY7_SLCR_MIO_PIN_IO_TYPE_LVCMOS33 (3<<9) #define ZY7_SLCR_MIO_PIN_IO_TYPE_HSTL (4<<9) #define ZY7_SLCR_MIO_PIN_L2_SEL_MASK (3<<3) #define ZY7_SLCR_MIO_PIN_L2_SEL_L3_MUX (0<<3) #define ZY7_SLCR_MIO_PIN_L2_SEL_SRAM_NOR_CS0 (1<<3) #define ZY7_SLCR_MIO_PIN_L2_SEL_NAND_CS (2<<3) #define ZY7_SLCR_MIO_PIN_L2_SEL_SDIO0_PC (3<<3) #define ZY7_SLCR_MIO_PIN_L1_SEL (1<<2) #define ZY7_SLCR_MIO_PIN_L0_SEL (1<<1) #define ZY7_SLCR_MIO_PIN_TRI_EN (1<<0) #define ZY7_SLCR_MIO_LOOPBACK 0x0804 #define ZY7_SLCR_MIO_LOOPBACK_I2C0_I2C1 (1<<3) #define ZY7_SLCR_MIO_LOOPBACK_CAN0_CAN1 (1<<2) #define ZY7_SLCR_MIO_LOOPBACK_UA0_UA1 (1<<1) #define ZY7_SLCR_MIO_LOOPBACK_SPI0_SPI1 (1<<0) #define ZY7_SLCR_MIO_MST_TRI0 0x080c #define ZY7_SLCR_MIO_MST_TRI1 0x0810 #define ZY7_SLCR_SD0_WP_CD_SEL 0x0830 #define ZY7_SLCR_SD1_WP_CD_SEL 0x0834 /* PS-PL level shifter control. */ #define ZY7_SLCR_LVL_SHFTR_EN 0x900 #define ZY7_SLCR_LVL_SHFTR_EN_USER_LVL_IN_EN_0 (1<<3) /* PL to PS */ #define ZY7_SLCR_LVL_SHFTR_EN_USER_LVL_OUT_EN_0 (1<<2) /* PS to PL */ #define ZY7_SLCR_LVL_SHFTR_EN_USER_LVL_IN_EN_1 (1<<1) /* PL to PS */ #define ZY7_SLCR_LVL_SHFTR_EN_USER_LVL_OUT_EN_1 (1<<0) /* PS to PL */ #define ZY7_SLCR_LVL_SHFTR_EN_ALL 0xf #define ZY7_SLCR_OCM_CFG 0x0910 #define ZY7_SLCR_GPIOB_CTRL 0x0b00 #define ZY7_SLCR_GPIOB_CFG_CMOS18 0x0b04 #define ZY7_SLCR_GPIOB_CFG_CMOS25 0x0b08 #define ZY7_SLCR_GPIOB_CFG_CMOS33 0x0b0c #define ZY7_SLCR_GPIOB_CFG_LVTTL 0x0b10 #define ZY7_SLCR_GPIOB_CFG_HSTL 0x0b14 #define ZY7_SLCR_GPIOB_DRVR_BIAS_CTRL 0x0b18 #define ZY7_SLCR_DDRIOB_ADDR0 0x0b40 #define ZY7_SLCR_DDRIOB_ADDR1 0x0b44 #define ZY7_SLCR_DDRIOB_DATA0 0x0b48 #define ZY7_SLCR_DDRIOB_DATA1 0x0b4c #define ZY7_SLCR_DDRIOB_DIFF0 0x0b50 #define ZY7_SLCR_DDRIOB_DIFF1 0x0b54 #define ZY7_SLCR_DDRIOB_CLK 0x0b58 #define ZY7_SLCR_DDRIOB_DRIVE_SLEW_ADDR 0x0b5c #define ZY7_SLCR_DDRIOB_DRIVE_SLEW_DATA 0x0b60 #define ZY7_SLCR_DDRIOB_DRIVE_SLEW_DIFF 0x0b64 #define ZY7_SLCR_DDRIOB_DRIVE_SLEW_CLK 0x0b68 #define ZY7_SLCR_DDRIOB_DDR_CTRL 0x0b6c #define ZY7_SLCR_DDRIOB_DCI_CTRL 0x0b70 #define ZY7_SLCR_DDRIOB_DCI_STATUS 0x0b74 #ifdef _KERNEL extern void zy7_slcr_preload_pl(void); extern void zy7_slcr_postload_pl(int en_level_shifters); extern int cgem_set_ref_clk(int unit, int frequency); + +/* Should be consistent with SRCSEL field of FPGAx_CLK_CTRL */ +#define ZY7_PL_FCLK_SRC_IO 0 +#define ZY7_PL_FCLK_SRC_IO_ALT 1 /* ZY7_PL_FCLK_SRC_IO is b0x */ +#define ZY7_PL_FCLK_SRC_ARM 2 +#define ZY7_PL_FCLK_SRC_DDR 3 + +int zy7_pl_fclk_set_source(int unit, int source); +int zy7_pl_fclk_get_source(int unit); +int zy7_pl_fclk_set_freq(int unit, int freq); +int zy7_pl_fclk_get_freq(int unit); +int zy7_pl_fclk_enable(int unit); +int zy7_pl_fclk_disable(int unit); +int zy7_pl_fclk_enabled(int unit); +int zy7_pl_level_shifters_enabled(void); +void zy7_pl_level_shifters_enable(void); +void zy7_pl_level_shifters_disable(void); + #endif #endif /* _ZY7_SLCR_H_ */ Index: projects/cxl_iscsi/sys/boot/i386/boot2/Makefile =================================================================== --- projects/cxl_iscsi/sys/boot/i386/boot2/Makefile (revision 279716) +++ projects/cxl_iscsi/sys/boot/i386/boot2/Makefile (revision 279717) @@ -1,116 +1,118 @@ # $FreeBSD$ .include FILES= boot boot1 boot2 NM?= nm # A value of 0x80 enables LBA support. BOOT_BOOT1_FLAGS?= 0x80 BOOT_COMCONSOLE_PORT?= 0x3f8 BOOT_COMCONSOLE_SPEED?= 9600 B2SIOFMT?= 0x3 REL1= 0x700 ORG1= 0x7c00 ORG2= 0x2000 # Decide level of UFS support. BOOT2_UFS?= UFS1_AND_UFS2 #BOOT2_UFS?= UFS2_ONLY #BOOT2_UFS?= UFS1_ONLY CFLAGS= -fomit-frame-pointer \ -mrtd \ -mregparm=3 \ -DUSE_XREAD \ -D${BOOT2_UFS} \ -DFLAGS=${BOOT_BOOT1_FLAGS} \ -DSIOPRT=${BOOT_COMCONSOLE_PORT} \ -DSIOFMT=${B2SIOFMT} \ -DSIOSPD=${BOOT_COMCONSOLE_SPEED} \ -I${.CURDIR}/../../common \ -I${.CURDIR}/../btx/lib -I. \ -Wall -Waggregate-return -Wbad-function-cast -Wcast-align \ -Wmissing-declarations -Wmissing-prototypes -Wnested-externs \ -Wpointer-arith -Wshadow -Wstrict-prototypes -Wwrite-strings \ -Winline CFLAGS.gcc+= -Os \ -fno-guess-branch-probability \ -fno-unit-at-a-time \ - -mno-align-long-strings \ --param max-inline-insns-single=100 +.if ${COMPILER_TYPE} == "gcc" && ${COMPILER_VERSION} <= 40201 +CFLAGS.gcc+= -mno-align-long-strings +.endif CFLAGS.clang+= -Oz ${CLANG_OPT_SMALL} LD_FLAGS=-static -N --gc-sections # Pick up ../Makefile.inc early. .include CLEANFILES= boot boot: boot1 boot2 cat boot1 boot2 > boot CLEANFILES+= boot1 boot1.out boot1.o boot1: boot1.out objcopy -S -O binary boot1.out ${.TARGET} boot1.out: boot1.o ${LD} ${LD_FLAGS} -e start -Ttext ${ORG1} -o ${.TARGET} boot1.o CLEANFILES+= boot2 boot2.ld boot2.ldr boot2.bin boot2.out boot2.o \ boot2.s boot2.s.tmp boot2.h sio.o boot2: boot2.ld @set -- `ls -l boot2.ld`; x=$$((7680-$$5)); \ echo "$$x bytes available"; test $$x -ge 0 dd if=boot2.ld of=${.TARGET} obs=7680 conv=osync boot2.ld: boot2.ldr boot2.bin ${BTXKERN} btxld -v -E ${ORG2} -f bin -b ${BTXKERN} -l boot2.ldr \ -o ${.TARGET} -P 1 boot2.bin boot2.ldr: dd if=/dev/zero of=${.TARGET} bs=512 count=1 boot2.bin: boot2.out objcopy -S -O binary boot2.out ${.TARGET} boot2.out: ${BTXCRT} boot2.o sio.o ${LD} ${LD_FLAGS} -Ttext ${ORG2} -o ${.TARGET} ${.ALLSRC} boot2.o: boot2.s ${CC} ${ACFLAGS} -c boot2.s SRCS= boot2.c boot2.h boot2.s: boot2.c boot2.h ${.CURDIR}/../../common/ufsread.c ${CC} ${CFLAGS} -S -o boot2.s.tmp ${.CURDIR}/boot2.c sed -e '/align/d' -e '/nop/d' < boot2.s.tmp > boot2.s rm -f boot2.s.tmp boot2.h: boot1.out ${NM} -t d ${.ALLSRC} | awk '/([0-9])+ T xread/ \ { x = $$1 - ORG1; \ printf("#define XREADORG %#x\n", REL1 + x) }' \ ORG1=`printf "%d" ${ORG1}` \ REL1=`printf "%d" ${REL1}` > ${.TARGET} .if ${MACHINE_CPUARCH} == "amd64" beforedepend boot2.s: machine CLEANFILES+= machine machine: ln -sf ${.CURDIR}/../../../i386/include machine .endif .include # XXX: clang integrated-as doesn't grok .codeNN directives yet CFLAGS.boot1.S= ${CLANG_NO_IAS} CFLAGS+= ${CFLAGS.${.IMPSRC:T}} Index: projects/cxl_iscsi/sys/boot/pc98/boot2/Makefile =================================================================== --- projects/cxl_iscsi/sys/boot/pc98/boot2/Makefile (revision 279716) +++ projects/cxl_iscsi/sys/boot/pc98/boot2/Makefile (revision 279717) @@ -1,115 +1,117 @@ # $FreeBSD$ .include FILES= boot boot1 boot2 NM?= nm BOOT_COMCONSOLE_PORT?= 0x238 BOOT_COMCONSOLE_SPEED?= 9600 B2SIOFMT?= 0x3 REL1= 0x700 ORG1= 0 ORG2= 0x2000 # Decide level of UFS support. BOOT2_UFS?= UFS1_AND_UFS2 #BOOT2_UFS?= UFS2_ONLY #BOOT2_UFS?= UFS1_ONLY CFLAGS= -fomit-frame-pointer \ -mrtd \ -mregparm=3 \ -D${BOOT2_UFS} \ -DFLAGS=${BOOT_BOOT1_FLAGS} \ -DSIOPRT=${BOOT_COMCONSOLE_PORT} \ -DSIOFMT=${B2SIOFMT} \ -DSIOSPD=${BOOT_COMCONSOLE_SPEED} \ -I${.CURDIR}/../../.. \ -I${.CURDIR}/../../i386/boot2 \ -I${.CURDIR}/../../common \ -I${.CURDIR}/../btx/lib -I. \ -Wall -Waggregate-return -Wbad-function-cast -Wcast-align \ -Wmissing-declarations -Wmissing-prototypes -Wnested-externs \ -Wpointer-arith -Wshadow -Wstrict-prototypes -Wwrite-strings \ -Winline CFLAGS.gcc+= -Os \ -fno-guess-branch-probability \ -fno-unit-at-a-time \ - -mno-align-long-strings \ --param max-inline-insns-single=100 +.if ${COMPILER_TYPE} == "gcc" && ${COMPILER_VERSION} <= 40201 +CFLAGS.gcc+= -mno-align-long-strings +.endif # Set machine type to PC98_SYSTEM_PARAMETER #CFLAGS+= -DSET_MACHINE_TYPE # Initialize the bi_bios_geom using the BIOS geometry #CFLAGS+= -DGET_BIOSGEOM CFLAGS.clang+= -Oz ${CLANG_OPT_SMALL} LD_FLAGS=-static -N --gc-sections # Pick up ../Makefile.inc early. .include .PATH: ${.CURDIR}/../../i386/boot2 CLEANFILES= boot boot: boot1 boot2 cat boot1 boot2 > boot CLEANFILES+= boot1 boot1.out boot1.o boot1: boot1.out objcopy -S -O binary boot1.out ${.TARGET} boot1.out: boot1.o ${LD} ${LD_FLAGS} -e start -Ttext ${ORG1} -o ${.TARGET} boot1.o CLEANFILES+= boot2 boot2.ld boot2.ldr boot2.bin boot2.out boot2.o \ boot2.s boot2.s.tmp boot2.h sio.o boot2: boot2.ld @set -- `ls -l boot2.ld`; x=$$((7680-$$5)); \ echo "$$x bytes available"; test $$x -ge 0 dd if=boot2.ld of=${.TARGET} obs=7680 conv=osync boot2.ld: boot2.ldr boot2.bin ${BTXKERN} btxld -v -E ${ORG2} -f bin -b ${BTXKERN} -l boot2.ldr \ -o ${.TARGET} -P 1 boot2.bin boot2.ldr: dd if=/dev/zero of=${.TARGET} bs=276 count=1 boot2.bin: boot2.out objcopy -S -O binary boot2.out ${.TARGET} boot2.out: ${BTXCRT} boot2.o sio.o ${LD} ${LD_FLAGS} -Ttext ${ORG2} -o ${.TARGET} ${.ALLSRC} boot2.o: boot2.s ${CC} ${ACFLAGS} -c boot2.s SRCS= boot2.c boot2.h boot2.s: boot2.c boot2.h ${.CURDIR}/../../common/ufsread.c ${CC} ${CFLAGS} -S -o boot2.s.tmp ${.CURDIR}/boot2.c sed -e '/align/d' -e '/nop/d' < boot2.s.tmp > boot2.s rm -f boot2.s.tmp boot2.h: boot1.out ${NM} -t d ${.ALLSRC} | awk '/([0-9])+ T (read|putc)/ \ { x = $$1 - ORG1; \ printf("#define %sORG %#x\n", toupper($$3), REL1 + x) }' \ ORG1=`printf "%d" ${ORG1}` \ REL1=`printf "%d" ${REL1}` > ${.TARGET} .include # XXX: clang integrated-as doesn't grok .codeNN directives yet CFLAGS.boot1.S= ${CLANG_NO_IAS} CFLAGS+= ${CFLAGS.${.IMPSRC:T}} Index: projects/cxl_iscsi/sys/boot =================================================================== --- projects/cxl_iscsi/sys/boot (revision 279716) +++ projects/cxl_iscsi/sys/boot (revision 279717) Property changes on: projects/cxl_iscsi/sys/boot ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/boot:r279555-279716 Index: projects/cxl_iscsi/sys/cddl/contrib/opensolaris/uts/common/sys/dtrace.h =================================================================== --- projects/cxl_iscsi/sys/cddl/contrib/opensolaris/uts/common/sys/dtrace.h (revision 279716) +++ projects/cxl_iscsi/sys/cddl/contrib/opensolaris/uts/common/sys/dtrace.h (revision 279717) @@ -1,2450 +1,2454 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2007 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ /* * Copyright (c) 2013, Joyent, Inc. All rights reserved. * Copyright (c) 2013 by Delphix. All rights reserved. */ #ifndef _SYS_DTRACE_H #define _SYS_DTRACE_H #pragma ident "%Z%%M% %I% %E% SMI" #ifdef __cplusplus extern "C" { #endif /* * DTrace Dynamic Tracing Software: Kernel Interfaces * * Note: The contents of this file are private to the implementation of the * Solaris system and DTrace subsystem and are subject to change at any time * without notice. Applications and drivers using these interfaces will fail * to run on future releases. These interfaces should not be used for any * purpose except those expressly outlined in dtrace(7D) and libdtrace(3LIB). * Please refer to the "Solaris Dynamic Tracing Guide" for more information. */ #ifndef _ASM #include #include #include #ifdef illumos #include #else #include #include #include #include #include typedef int model_t; #endif #include #ifdef illumos #include #include #else #include #endif /* * DTrace Universal Constants and Typedefs */ #define DTRACE_CPUALL -1 /* all CPUs */ #define DTRACE_IDNONE 0 /* invalid probe identifier */ #define DTRACE_EPIDNONE 0 /* invalid enabled probe identifier */ #define DTRACE_AGGIDNONE 0 /* invalid aggregation identifier */ #define DTRACE_AGGVARIDNONE 0 /* invalid aggregation variable ID */ #define DTRACE_CACHEIDNONE 0 /* invalid predicate cache */ #define DTRACE_PROVNONE 0 /* invalid provider identifier */ #define DTRACE_METAPROVNONE 0 /* invalid meta-provider identifier */ #define DTRACE_ARGNONE -1 /* invalid argument index */ #define DTRACE_PROVNAMELEN 64 #define DTRACE_MODNAMELEN 64 #define DTRACE_FUNCNAMELEN 128 #define DTRACE_NAMELEN 64 #define DTRACE_FULLNAMELEN (DTRACE_PROVNAMELEN + DTRACE_MODNAMELEN + \ DTRACE_FUNCNAMELEN + DTRACE_NAMELEN + 4) #define DTRACE_ARGTYPELEN 128 typedef uint32_t dtrace_id_t; /* probe identifier */ typedef uint32_t dtrace_epid_t; /* enabled probe identifier */ typedef uint32_t dtrace_aggid_t; /* aggregation identifier */ typedef int64_t dtrace_aggvarid_t; /* aggregation variable identifier */ typedef uint16_t dtrace_actkind_t; /* action kind */ typedef int64_t dtrace_optval_t; /* option value */ typedef uint32_t dtrace_cacheid_t; /* predicate cache identifier */ typedef enum dtrace_probespec { DTRACE_PROBESPEC_NONE = -1, DTRACE_PROBESPEC_PROVIDER = 0, DTRACE_PROBESPEC_MOD, DTRACE_PROBESPEC_FUNC, DTRACE_PROBESPEC_NAME } dtrace_probespec_t; /* * DTrace Intermediate Format (DIF) * * The following definitions describe the DTrace Intermediate Format (DIF), a * a RISC-like instruction set and program encoding used to represent * predicates and actions that can be bound to DTrace probes. The constants * below defining the number of available registers are suggested minimums; the * compiler should use DTRACEIOC_CONF to dynamically obtain the number of * registers provided by the current DTrace implementation. */ #define DIF_VERSION_1 1 /* DIF version 1: Solaris 10 Beta */ #define DIF_VERSION_2 2 /* DIF version 2: Solaris 10 FCS */ #define DIF_VERSION DIF_VERSION_2 /* latest DIF instruction set version */ #define DIF_DIR_NREGS 8 /* number of DIF integer registers */ #define DIF_DTR_NREGS 8 /* number of DIF tuple registers */ #define DIF_OP_OR 1 /* or r1, r2, rd */ #define DIF_OP_XOR 2 /* xor r1, r2, rd */ #define DIF_OP_AND 3 /* and r1, r2, rd */ #define DIF_OP_SLL 4 /* sll r1, r2, rd */ #define DIF_OP_SRL 5 /* srl r1, r2, rd */ #define DIF_OP_SUB 6 /* sub r1, r2, rd */ #define DIF_OP_ADD 7 /* add r1, r2, rd */ #define DIF_OP_MUL 8 /* mul r1, r2, rd */ #define DIF_OP_SDIV 9 /* sdiv r1, r2, rd */ #define DIF_OP_UDIV 10 /* udiv r1, r2, rd */ #define DIF_OP_SREM 11 /* srem r1, r2, rd */ #define DIF_OP_UREM 12 /* urem r1, r2, rd */ #define DIF_OP_NOT 13 /* not r1, rd */ #define DIF_OP_MOV 14 /* mov r1, rd */ #define DIF_OP_CMP 15 /* cmp r1, r2 */ #define DIF_OP_TST 16 /* tst r1 */ #define DIF_OP_BA 17 /* ba label */ #define DIF_OP_BE 18 /* be label */ #define DIF_OP_BNE 19 /* bne label */ #define DIF_OP_BG 20 /* bg label */ #define DIF_OP_BGU 21 /* bgu label */ #define DIF_OP_BGE 22 /* bge label */ #define DIF_OP_BGEU 23 /* bgeu label */ #define DIF_OP_BL 24 /* bl label */ #define DIF_OP_BLU 25 /* blu label */ #define DIF_OP_BLE 26 /* ble label */ #define DIF_OP_BLEU 27 /* bleu label */ #define DIF_OP_LDSB 28 /* ldsb [r1], rd */ #define DIF_OP_LDSH 29 /* ldsh [r1], rd */ #define DIF_OP_LDSW 30 /* ldsw [r1], rd */ #define DIF_OP_LDUB 31 /* ldub [r1], rd */ #define DIF_OP_LDUH 32 /* lduh [r1], rd */ #define DIF_OP_LDUW 33 /* lduw [r1], rd */ #define DIF_OP_LDX 34 /* ldx [r1], rd */ #define DIF_OP_RET 35 /* ret rd */ #define DIF_OP_NOP 36 /* nop */ #define DIF_OP_SETX 37 /* setx intindex, rd */ #define DIF_OP_SETS 38 /* sets strindex, rd */ #define DIF_OP_SCMP 39 /* scmp r1, r2 */ #define DIF_OP_LDGA 40 /* ldga var, ri, rd */ #define DIF_OP_LDGS 41 /* ldgs var, rd */ #define DIF_OP_STGS 42 /* stgs var, rs */ #define DIF_OP_LDTA 43 /* ldta var, ri, rd */ #define DIF_OP_LDTS 44 /* ldts var, rd */ #define DIF_OP_STTS 45 /* stts var, rs */ #define DIF_OP_SRA 46 /* sra r1, r2, rd */ #define DIF_OP_CALL 47 /* call subr, rd */ #define DIF_OP_PUSHTR 48 /* pushtr type, rs, rr */ #define DIF_OP_PUSHTV 49 /* pushtv type, rs, rv */ #define DIF_OP_POPTS 50 /* popts */ #define DIF_OP_FLUSHTS 51 /* flushts */ #define DIF_OP_LDGAA 52 /* ldgaa var, rd */ #define DIF_OP_LDTAA 53 /* ldtaa var, rd */ #define DIF_OP_STGAA 54 /* stgaa var, rs */ #define DIF_OP_STTAA 55 /* sttaa var, rs */ #define DIF_OP_LDLS 56 /* ldls var, rd */ #define DIF_OP_STLS 57 /* stls var, rs */ #define DIF_OP_ALLOCS 58 /* allocs r1, rd */ #define DIF_OP_COPYS 59 /* copys r1, r2, rd */ #define DIF_OP_STB 60 /* stb r1, [rd] */ #define DIF_OP_STH 61 /* sth r1, [rd] */ #define DIF_OP_STW 62 /* stw r1, [rd] */ #define DIF_OP_STX 63 /* stx r1, [rd] */ #define DIF_OP_ULDSB 64 /* uldsb [r1], rd */ #define DIF_OP_ULDSH 65 /* uldsh [r1], rd */ #define DIF_OP_ULDSW 66 /* uldsw [r1], rd */ #define DIF_OP_ULDUB 67 /* uldub [r1], rd */ #define DIF_OP_ULDUH 68 /* ulduh [r1], rd */ #define DIF_OP_ULDUW 69 /* ulduw [r1], rd */ #define DIF_OP_ULDX 70 /* uldx [r1], rd */ #define DIF_OP_RLDSB 71 /* rldsb [r1], rd */ #define DIF_OP_RLDSH 72 /* rldsh [r1], rd */ #define DIF_OP_RLDSW 73 /* rldsw [r1], rd */ #define DIF_OP_RLDUB 74 /* rldub [r1], rd */ #define DIF_OP_RLDUH 75 /* rlduh [r1], rd */ #define DIF_OP_RLDUW 76 /* rlduw [r1], rd */ #define DIF_OP_RLDX 77 /* rldx [r1], rd */ #define DIF_OP_XLATE 78 /* xlate xlrindex, rd */ #define DIF_OP_XLARG 79 /* xlarg xlrindex, rd */ #define DIF_INTOFF_MAX 0xffff /* highest integer table offset */ #define DIF_STROFF_MAX 0xffff /* highest string table offset */ #define DIF_REGISTER_MAX 0xff /* highest register number */ #define DIF_VARIABLE_MAX 0xffff /* highest variable identifier */ #define DIF_SUBROUTINE_MAX 0xffff /* highest subroutine code */ #define DIF_VAR_ARRAY_MIN 0x0000 /* lowest numbered array variable */ #define DIF_VAR_ARRAY_UBASE 0x0080 /* lowest user-defined array */ #define DIF_VAR_ARRAY_MAX 0x00ff /* highest numbered array variable */ #define DIF_VAR_OTHER_MIN 0x0100 /* lowest numbered scalar or assc */ #define DIF_VAR_OTHER_UBASE 0x0500 /* lowest user-defined scalar or assc */ #define DIF_VAR_OTHER_MAX 0xffff /* highest numbered scalar or assc */ #define DIF_VAR_ARGS 0x0000 /* arguments array */ #define DIF_VAR_REGS 0x0001 /* registers array */ #define DIF_VAR_UREGS 0x0002 /* user registers array */ #define DIF_VAR_CURTHREAD 0x0100 /* thread pointer */ #define DIF_VAR_TIMESTAMP 0x0101 /* timestamp */ #define DIF_VAR_VTIMESTAMP 0x0102 /* virtual timestamp */ #define DIF_VAR_IPL 0x0103 /* interrupt priority level */ #define DIF_VAR_EPID 0x0104 /* enabled probe ID */ #define DIF_VAR_ID 0x0105 /* probe ID */ #define DIF_VAR_ARG0 0x0106 /* first argument */ #define DIF_VAR_ARG1 0x0107 /* second argument */ #define DIF_VAR_ARG2 0x0108 /* third argument */ #define DIF_VAR_ARG3 0x0109 /* fourth argument */ #define DIF_VAR_ARG4 0x010a /* fifth argument */ #define DIF_VAR_ARG5 0x010b /* sixth argument */ #define DIF_VAR_ARG6 0x010c /* seventh argument */ #define DIF_VAR_ARG7 0x010d /* eighth argument */ #define DIF_VAR_ARG8 0x010e /* ninth argument */ #define DIF_VAR_ARG9 0x010f /* tenth argument */ #define DIF_VAR_STACKDEPTH 0x0110 /* stack depth */ #define DIF_VAR_CALLER 0x0111 /* caller */ #define DIF_VAR_PROBEPROV 0x0112 /* probe provider */ #define DIF_VAR_PROBEMOD 0x0113 /* probe module */ #define DIF_VAR_PROBEFUNC 0x0114 /* probe function */ #define DIF_VAR_PROBENAME 0x0115 /* probe name */ #define DIF_VAR_PID 0x0116 /* process ID */ #define DIF_VAR_TID 0x0117 /* (per-process) thread ID */ #define DIF_VAR_EXECNAME 0x0118 /* name of executable */ #define DIF_VAR_ZONENAME 0x0119 /* zone name associated with process */ #define DIF_VAR_WALLTIMESTAMP 0x011a /* wall-clock timestamp */ #define DIF_VAR_USTACKDEPTH 0x011b /* user-land stack depth */ #define DIF_VAR_UCALLER 0x011c /* user-level caller */ #define DIF_VAR_PPID 0x011d /* parent process ID */ #define DIF_VAR_UID 0x011e /* process user ID */ #define DIF_VAR_GID 0x011f /* process group ID */ #define DIF_VAR_ERRNO 0x0120 /* thread errno */ #define DIF_VAR_EXECARGS 0x0121 /* process arguments */ #ifndef illumos #define DIF_VAR_CPU 0x0200 #endif #define DIF_SUBR_RAND 0 #define DIF_SUBR_MUTEX_OWNED 1 #define DIF_SUBR_MUTEX_OWNER 2 #define DIF_SUBR_MUTEX_TYPE_ADAPTIVE 3 #define DIF_SUBR_MUTEX_TYPE_SPIN 4 #define DIF_SUBR_RW_READ_HELD 5 #define DIF_SUBR_RW_WRITE_HELD 6 #define DIF_SUBR_RW_ISWRITER 7 #define DIF_SUBR_COPYIN 8 #define DIF_SUBR_COPYINSTR 9 #define DIF_SUBR_SPECULATION 10 #define DIF_SUBR_PROGENYOF 11 #define DIF_SUBR_STRLEN 12 #define DIF_SUBR_COPYOUT 13 #define DIF_SUBR_COPYOUTSTR 14 #define DIF_SUBR_ALLOCA 15 #define DIF_SUBR_BCOPY 16 #define DIF_SUBR_COPYINTO 17 #define DIF_SUBR_MSGDSIZE 18 #define DIF_SUBR_MSGSIZE 19 #define DIF_SUBR_GETMAJOR 20 #define DIF_SUBR_GETMINOR 21 #define DIF_SUBR_DDI_PATHNAME 22 #define DIF_SUBR_STRJOIN 23 #define DIF_SUBR_LLTOSTR 24 #define DIF_SUBR_BASENAME 25 #define DIF_SUBR_DIRNAME 26 #define DIF_SUBR_CLEANPATH 27 #define DIF_SUBR_STRCHR 28 #define DIF_SUBR_STRRCHR 29 #define DIF_SUBR_STRSTR 30 #define DIF_SUBR_STRTOK 31 #define DIF_SUBR_SUBSTR 32 #define DIF_SUBR_INDEX 33 #define DIF_SUBR_RINDEX 34 #define DIF_SUBR_HTONS 35 #define DIF_SUBR_HTONL 36 #define DIF_SUBR_HTONLL 37 #define DIF_SUBR_NTOHS 38 #define DIF_SUBR_NTOHL 39 #define DIF_SUBR_NTOHLL 40 #define DIF_SUBR_INET_NTOP 41 #define DIF_SUBR_INET_NTOA 42 #define DIF_SUBR_INET_NTOA6 43 #define DIF_SUBR_TOUPPER 44 #define DIF_SUBR_TOLOWER 45 #define DIF_SUBR_MEMREF 46 #define DIF_SUBR_TYPEREF 47 #define DIF_SUBR_SX_SHARED_HELD 48 #define DIF_SUBR_SX_EXCLUSIVE_HELD 49 #define DIF_SUBR_SX_ISEXCLUSIVE 50 #define DIF_SUBR_MEMSTR 51 #define DIF_SUBR_GETF 52 #define DIF_SUBR_JSON 53 #define DIF_SUBR_STRTOLL 54 #define DIF_SUBR_MAX 54 /* max subroutine value */ typedef uint32_t dif_instr_t; #define DIF_INSTR_OP(i) (((i) >> 24) & 0xff) #define DIF_INSTR_R1(i) (((i) >> 16) & 0xff) #define DIF_INSTR_R2(i) (((i) >> 8) & 0xff) #define DIF_INSTR_RD(i) ((i) & 0xff) #define DIF_INSTR_RS(i) ((i) & 0xff) #define DIF_INSTR_LABEL(i) ((i) & 0xffffff) #define DIF_INSTR_VAR(i) (((i) >> 8) & 0xffff) #define DIF_INSTR_INTEGER(i) (((i) >> 8) & 0xffff) #define DIF_INSTR_STRING(i) (((i) >> 8) & 0xffff) #define DIF_INSTR_SUBR(i) (((i) >> 8) & 0xffff) #define DIF_INSTR_TYPE(i) (((i) >> 16) & 0xff) #define DIF_INSTR_XLREF(i) (((i) >> 8) & 0xffff) #define DIF_INSTR_FMT(op, r1, r2, d) \ (((op) << 24) | ((r1) << 16) | ((r2) << 8) | (d)) #define DIF_INSTR_NOT(r1, d) (DIF_INSTR_FMT(DIF_OP_NOT, r1, 0, d)) #define DIF_INSTR_MOV(r1, d) (DIF_INSTR_FMT(DIF_OP_MOV, r1, 0, d)) #define DIF_INSTR_CMP(op, r1, r2) (DIF_INSTR_FMT(op, r1, r2, 0)) #define DIF_INSTR_TST(r1) (DIF_INSTR_FMT(DIF_OP_TST, r1, 0, 0)) #define DIF_INSTR_BRANCH(op, label) (((op) << 24) | (label)) #define DIF_INSTR_LOAD(op, r1, d) (DIF_INSTR_FMT(op, r1, 0, d)) #define DIF_INSTR_STORE(op, r1, d) (DIF_INSTR_FMT(op, r1, 0, d)) #define DIF_INSTR_SETX(i, d) ((DIF_OP_SETX << 24) | ((i) << 8) | (d)) #define DIF_INSTR_SETS(s, d) ((DIF_OP_SETS << 24) | ((s) << 8) | (d)) #define DIF_INSTR_RET(d) (DIF_INSTR_FMT(DIF_OP_RET, 0, 0, d)) #define DIF_INSTR_NOP (DIF_OP_NOP << 24) #define DIF_INSTR_LDA(op, v, r, d) (DIF_INSTR_FMT(op, v, r, d)) #define DIF_INSTR_LDV(op, v, d) (((op) << 24) | ((v) << 8) | (d)) #define DIF_INSTR_STV(op, v, rs) (((op) << 24) | ((v) << 8) | (rs)) #define DIF_INSTR_CALL(s, d) ((DIF_OP_CALL << 24) | ((s) << 8) | (d)) #define DIF_INSTR_PUSHTS(op, t, r2, rs) (DIF_INSTR_FMT(op, t, r2, rs)) #define DIF_INSTR_POPTS (DIF_OP_POPTS << 24) #define DIF_INSTR_FLUSHTS (DIF_OP_FLUSHTS << 24) #define DIF_INSTR_ALLOCS(r1, d) (DIF_INSTR_FMT(DIF_OP_ALLOCS, r1, 0, d)) #define DIF_INSTR_COPYS(r1, r2, d) (DIF_INSTR_FMT(DIF_OP_COPYS, r1, r2, d)) #define DIF_INSTR_XLATE(op, r, d) (((op) << 24) | ((r) << 8) | (d)) #define DIF_REG_R0 0 /* %r0 is always set to zero */ /* * A DTrace Intermediate Format Type (DIF Type) is used to represent the types * of variables, function and associative array arguments, and the return type * for each DIF object (shown below). It contains a description of the type, * its size in bytes, and a module identifier. */ typedef struct dtrace_diftype { uint8_t dtdt_kind; /* type kind (see below) */ uint8_t dtdt_ckind; /* type kind in CTF */ uint8_t dtdt_flags; /* type flags (see below) */ uint8_t dtdt_pad; /* reserved for future use */ uint32_t dtdt_size; /* type size in bytes (unless string) */ } dtrace_diftype_t; #define DIF_TYPE_CTF 0 /* type is a CTF type */ #define DIF_TYPE_STRING 1 /* type is a D string */ #define DIF_TF_BYREF 0x1 /* type is passed by reference */ #define DIF_TF_BYUREF 0x2 /* user type is passed by reference */ /* * A DTrace Intermediate Format variable record is used to describe each of the * variables referenced by a given DIF object. It contains an integer variable * identifier along with variable scope and properties, as shown below. The * size of this structure must be sizeof (int) aligned. */ typedef struct dtrace_difv { uint32_t dtdv_name; /* variable name index in dtdo_strtab */ uint32_t dtdv_id; /* variable reference identifier */ uint8_t dtdv_kind; /* variable kind (see below) */ uint8_t dtdv_scope; /* variable scope (see below) */ uint16_t dtdv_flags; /* variable flags (see below) */ dtrace_diftype_t dtdv_type; /* variable type (see above) */ } dtrace_difv_t; #define DIFV_KIND_ARRAY 0 /* variable is an array of quantities */ #define DIFV_KIND_SCALAR 1 /* variable is a scalar quantity */ #define DIFV_SCOPE_GLOBAL 0 /* variable has global scope */ #define DIFV_SCOPE_THREAD 1 /* variable has thread scope */ #define DIFV_SCOPE_LOCAL 2 /* variable has local scope */ #define DIFV_F_REF 0x1 /* variable is referenced by DIFO */ #define DIFV_F_MOD 0x2 /* variable is written by DIFO */ /* * DTrace Actions * * The upper byte determines the class of the action; the low bytes determines * the specific action within that class. The classes of actions are as * follows: * * [ no class ] <= May record process- or kernel-related data * DTRACEACT_PROC <= Only records process-related data * DTRACEACT_PROC_DESTRUCTIVE <= Potentially destructive to processes * DTRACEACT_KERNEL <= Only records kernel-related data * DTRACEACT_KERNEL_DESTRUCTIVE <= Potentially destructive to the kernel * DTRACEACT_SPECULATIVE <= Speculation-related action * DTRACEACT_AGGREGATION <= Aggregating action */ #define DTRACEACT_NONE 0 /* no action */ #define DTRACEACT_DIFEXPR 1 /* action is DIF expression */ #define DTRACEACT_EXIT 2 /* exit() action */ #define DTRACEACT_PRINTF 3 /* printf() action */ #define DTRACEACT_PRINTA 4 /* printa() action */ #define DTRACEACT_LIBACT 5 /* library-controlled action */ #define DTRACEACT_TRACEMEM 6 /* tracemem() action */ #define DTRACEACT_TRACEMEM_DYNSIZE 7 /* dynamic tracemem() size */ #define DTRACEACT_PRINTM 8 /* printm() action (BSD) */ #define DTRACEACT_PRINTT 9 /* printt() action (BSD) */ #define DTRACEACT_PROC 0x0100 #define DTRACEACT_USTACK (DTRACEACT_PROC + 1) #define DTRACEACT_JSTACK (DTRACEACT_PROC + 2) #define DTRACEACT_USYM (DTRACEACT_PROC + 3) #define DTRACEACT_UMOD (DTRACEACT_PROC + 4) #define DTRACEACT_UADDR (DTRACEACT_PROC + 5) #define DTRACEACT_PROC_DESTRUCTIVE 0x0200 #define DTRACEACT_STOP (DTRACEACT_PROC_DESTRUCTIVE + 1) #define DTRACEACT_RAISE (DTRACEACT_PROC_DESTRUCTIVE + 2) #define DTRACEACT_SYSTEM (DTRACEACT_PROC_DESTRUCTIVE + 3) #define DTRACEACT_FREOPEN (DTRACEACT_PROC_DESTRUCTIVE + 4) #define DTRACEACT_PROC_CONTROL 0x0300 #define DTRACEACT_KERNEL 0x0400 #define DTRACEACT_STACK (DTRACEACT_KERNEL + 1) #define DTRACEACT_SYM (DTRACEACT_KERNEL + 2) #define DTRACEACT_MOD (DTRACEACT_KERNEL + 3) #define DTRACEACT_KERNEL_DESTRUCTIVE 0x0500 #define DTRACEACT_BREAKPOINT (DTRACEACT_KERNEL_DESTRUCTIVE + 1) #define DTRACEACT_PANIC (DTRACEACT_KERNEL_DESTRUCTIVE + 2) #define DTRACEACT_CHILL (DTRACEACT_KERNEL_DESTRUCTIVE + 3) #define DTRACEACT_SPECULATIVE 0x0600 #define DTRACEACT_SPECULATE (DTRACEACT_SPECULATIVE + 1) #define DTRACEACT_COMMIT (DTRACEACT_SPECULATIVE + 2) #define DTRACEACT_DISCARD (DTRACEACT_SPECULATIVE + 3) #define DTRACEACT_CLASS(x) ((x) & 0xff00) #define DTRACEACT_ISDESTRUCTIVE(x) \ (DTRACEACT_CLASS(x) == DTRACEACT_PROC_DESTRUCTIVE || \ DTRACEACT_CLASS(x) == DTRACEACT_KERNEL_DESTRUCTIVE) #define DTRACEACT_ISSPECULATIVE(x) \ (DTRACEACT_CLASS(x) == DTRACEACT_SPECULATIVE) #define DTRACEACT_ISPRINTFLIKE(x) \ ((x) == DTRACEACT_PRINTF || (x) == DTRACEACT_PRINTA || \ (x) == DTRACEACT_SYSTEM || (x) == DTRACEACT_FREOPEN) /* * DTrace Aggregating Actions * * These are functions f(x) for which the following is true: * * f(f(x_0) U f(x_1) U ... U f(x_n)) = f(x_0 U x_1 U ... U x_n) * * where x_n is a set of arbitrary data. Aggregating actions are in their own * DTrace action class, DTTRACEACT_AGGREGATION. The macros provided here allow * for easier processing of the aggregation argument and data payload for a few * aggregating actions (notably: quantize(), lquantize(), and ustack()). */ #define DTRACEACT_AGGREGATION 0x0700 #define DTRACEAGG_COUNT (DTRACEACT_AGGREGATION + 1) #define DTRACEAGG_MIN (DTRACEACT_AGGREGATION + 2) #define DTRACEAGG_MAX (DTRACEACT_AGGREGATION + 3) #define DTRACEAGG_AVG (DTRACEACT_AGGREGATION + 4) #define DTRACEAGG_SUM (DTRACEACT_AGGREGATION + 5) #define DTRACEAGG_STDDEV (DTRACEACT_AGGREGATION + 6) #define DTRACEAGG_QUANTIZE (DTRACEACT_AGGREGATION + 7) #define DTRACEAGG_LQUANTIZE (DTRACEACT_AGGREGATION + 8) #define DTRACEAGG_LLQUANTIZE (DTRACEACT_AGGREGATION + 9) #define DTRACEACT_ISAGG(x) \ (DTRACEACT_CLASS(x) == DTRACEACT_AGGREGATION) #define DTRACE_QUANTIZE_NBUCKETS \ (((sizeof (uint64_t) * NBBY) - 1) * 2 + 1) #define DTRACE_QUANTIZE_ZEROBUCKET ((sizeof (uint64_t) * NBBY) - 1) #define DTRACE_QUANTIZE_BUCKETVAL(buck) \ (int64_t)((buck) < DTRACE_QUANTIZE_ZEROBUCKET ? \ -(1LL << (DTRACE_QUANTIZE_ZEROBUCKET - 1 - (buck))) : \ (buck) == DTRACE_QUANTIZE_ZEROBUCKET ? 0 : \ 1LL << ((buck) - DTRACE_QUANTIZE_ZEROBUCKET - 1)) #define DTRACE_LQUANTIZE_STEPSHIFT 48 #define DTRACE_LQUANTIZE_STEPMASK ((uint64_t)UINT16_MAX << 48) #define DTRACE_LQUANTIZE_LEVELSHIFT 32 #define DTRACE_LQUANTIZE_LEVELMASK ((uint64_t)UINT16_MAX << 32) #define DTRACE_LQUANTIZE_BASESHIFT 0 #define DTRACE_LQUANTIZE_BASEMASK UINT32_MAX #define DTRACE_LQUANTIZE_STEP(x) \ (uint16_t)(((x) & DTRACE_LQUANTIZE_STEPMASK) >> \ DTRACE_LQUANTIZE_STEPSHIFT) #define DTRACE_LQUANTIZE_LEVELS(x) \ (uint16_t)(((x) & DTRACE_LQUANTIZE_LEVELMASK) >> \ DTRACE_LQUANTIZE_LEVELSHIFT) #define DTRACE_LQUANTIZE_BASE(x) \ (int32_t)(((x) & DTRACE_LQUANTIZE_BASEMASK) >> \ DTRACE_LQUANTIZE_BASESHIFT) #define DTRACE_LLQUANTIZE_FACTORSHIFT 48 #define DTRACE_LLQUANTIZE_FACTORMASK ((uint64_t)UINT16_MAX << 48) #define DTRACE_LLQUANTIZE_LOWSHIFT 32 #define DTRACE_LLQUANTIZE_LOWMASK ((uint64_t)UINT16_MAX << 32) #define DTRACE_LLQUANTIZE_HIGHSHIFT 16 #define DTRACE_LLQUANTIZE_HIGHMASK ((uint64_t)UINT16_MAX << 16) #define DTRACE_LLQUANTIZE_NSTEPSHIFT 0 #define DTRACE_LLQUANTIZE_NSTEPMASK UINT16_MAX #define DTRACE_LLQUANTIZE_FACTOR(x) \ (uint16_t)(((x) & DTRACE_LLQUANTIZE_FACTORMASK) >> \ DTRACE_LLQUANTIZE_FACTORSHIFT) #define DTRACE_LLQUANTIZE_LOW(x) \ (uint16_t)(((x) & DTRACE_LLQUANTIZE_LOWMASK) >> \ DTRACE_LLQUANTIZE_LOWSHIFT) #define DTRACE_LLQUANTIZE_HIGH(x) \ (uint16_t)(((x) & DTRACE_LLQUANTIZE_HIGHMASK) >> \ DTRACE_LLQUANTIZE_HIGHSHIFT) #define DTRACE_LLQUANTIZE_NSTEP(x) \ (uint16_t)(((x) & DTRACE_LLQUANTIZE_NSTEPMASK) >> \ DTRACE_LLQUANTIZE_NSTEPSHIFT) #define DTRACE_USTACK_NFRAMES(x) (uint32_t)((x) & UINT32_MAX) #define DTRACE_USTACK_STRSIZE(x) (uint32_t)((x) >> 32) #define DTRACE_USTACK_ARG(x, y) \ ((((uint64_t)(y)) << 32) | ((x) & UINT32_MAX)) #ifndef _LP64 #if BYTE_ORDER == _BIG_ENDIAN #define DTRACE_PTR(type, name) uint32_t name##pad; type *name #else #define DTRACE_PTR(type, name) type *name; uint32_t name##pad #endif #else #define DTRACE_PTR(type, name) type *name #endif /* * DTrace Object Format (DOF) * * DTrace programs can be persistently encoded in the DOF format so that they * may be embedded in other programs (for example, in an ELF file) or in the * dtrace driver configuration file for use in anonymous tracing. The DOF * format is versioned and extensible so that it can be revised and so that * internal data structures can be modified or extended compatibly. All DOF * structures use fixed-size types, so the 32-bit and 64-bit representations * are identical and consumers can use either data model transparently. * * The file layout is structured as follows: * * +---------------+-------------------+----- ... ----+---- ... ------+ * | dof_hdr_t | dof_sec_t[ ... ] | loadable | non-loadable | * | (file header) | (section headers) | section data | section data | * +---------------+-------------------+----- ... ----+---- ... ------+ * |<------------ dof_hdr.dofh_loadsz --------------->| | * |<------------ dof_hdr.dofh_filesz ------------------------------->| * * The file header stores meta-data including a magic number, data model for * the instrumentation, data encoding, and properties of the DIF code within. * The header describes its own size and the size of the section headers. By * convention, an array of section headers follows the file header, and then * the data for all loadable sections and unloadable sections. This permits * consumer code to easily download the headers and all loadable data into the * DTrace driver in one contiguous chunk, omitting other extraneous sections. * * The section headers describe the size, offset, alignment, and section type * for each section. Sections are described using a set of #defines that tell * the consumer what kind of data is expected. Sections can contain links to * other sections by storing a dof_secidx_t, an index into the section header * array, inside of the section data structures. The section header includes * an entry size so that sections with data arrays can grow their structures. * * The DOF data itself can contain many snippets of DIF (i.e. >1 DIFOs), which * are represented themselves as a collection of related DOF sections. This * permits us to change the set of sections associated with a DIFO over time, * and also permits us to encode DIFOs that contain different sets of sections. * When a DOF section wants to refer to a DIFO, it stores the dof_secidx_t of a * section of type DOF_SECT_DIFOHDR. This section's data is then an array of * dof_secidx_t's which in turn denote the sections associated with this DIFO. * * This loose coupling of the file structure (header and sections) to the * structure of the DTrace program itself (ECB descriptions, action * descriptions, and DIFOs) permits activities such as relocation processing * to occur in a single pass without having to understand D program structure. * * Finally, strings are always stored in ELF-style string tables along with a * string table section index and string table offset. Therefore strings in * DOF are always arbitrary-length and not bound to the current implementation. */ #define DOF_ID_SIZE 16 /* total size of dofh_ident[] in bytes */ typedef struct dof_hdr { uint8_t dofh_ident[DOF_ID_SIZE]; /* identification bytes (see below) */ uint32_t dofh_flags; /* file attribute flags (if any) */ uint32_t dofh_hdrsize; /* size of file header in bytes */ uint32_t dofh_secsize; /* size of section header in bytes */ uint32_t dofh_secnum; /* number of section headers */ uint64_t dofh_secoff; /* file offset of section headers */ uint64_t dofh_loadsz; /* file size of loadable portion */ uint64_t dofh_filesz; /* file size of entire DOF file */ uint64_t dofh_pad; /* reserved for future use */ } dof_hdr_t; #define DOF_ID_MAG0 0 /* first byte of magic number */ #define DOF_ID_MAG1 1 /* second byte of magic number */ #define DOF_ID_MAG2 2 /* third byte of magic number */ #define DOF_ID_MAG3 3 /* fourth byte of magic number */ #define DOF_ID_MODEL 4 /* DOF data model (see below) */ #define DOF_ID_ENCODING 5 /* DOF data encoding (see below) */ #define DOF_ID_VERSION 6 /* DOF file format major version (see below) */ #define DOF_ID_DIFVERS 7 /* DIF instruction set version */ #define DOF_ID_DIFIREG 8 /* DIF integer registers used by compiler */ #define DOF_ID_DIFTREG 9 /* DIF tuple registers used by compiler */ #define DOF_ID_PAD 10 /* start of padding bytes (all zeroes) */ #define DOF_MAG_MAG0 0x7F /* DOF_ID_MAG[0-3] */ #define DOF_MAG_MAG1 'D' #define DOF_MAG_MAG2 'O' #define DOF_MAG_MAG3 'F' #define DOF_MAG_STRING "\177DOF" #define DOF_MAG_STRLEN 4 #define DOF_MODEL_NONE 0 /* DOF_ID_MODEL */ #define DOF_MODEL_ILP32 1 #define DOF_MODEL_LP64 2 #ifdef _LP64 #define DOF_MODEL_NATIVE DOF_MODEL_LP64 #else #define DOF_MODEL_NATIVE DOF_MODEL_ILP32 #endif #define DOF_ENCODE_NONE 0 /* DOF_ID_ENCODING */ #define DOF_ENCODE_LSB 1 #define DOF_ENCODE_MSB 2 #if BYTE_ORDER == _BIG_ENDIAN #define DOF_ENCODE_NATIVE DOF_ENCODE_MSB #else #define DOF_ENCODE_NATIVE DOF_ENCODE_LSB #endif #define DOF_VERSION_1 1 /* DOF version 1: Solaris 10 FCS */ #define DOF_VERSION_2 2 /* DOF version 2: Solaris Express 6/06 */ #define DOF_VERSION DOF_VERSION_2 /* Latest DOF version */ #define DOF_FL_VALID 0 /* mask of all valid dofh_flags bits */ typedef uint32_t dof_secidx_t; /* section header table index type */ typedef uint32_t dof_stridx_t; /* string table index type */ #define DOF_SECIDX_NONE (-1U) /* null value for section indices */ #define DOF_STRIDX_NONE (-1U) /* null value for string indices */ typedef struct dof_sec { uint32_t dofs_type; /* section type (see below) */ uint32_t dofs_align; /* section data memory alignment */ uint32_t dofs_flags; /* section flags (if any) */ uint32_t dofs_entsize; /* size of section entry (if table) */ uint64_t dofs_offset; /* offset of section data within file */ uint64_t dofs_size; /* size of section data in bytes */ } dof_sec_t; #define DOF_SECT_NONE 0 /* null section */ #define DOF_SECT_COMMENTS 1 /* compiler comments */ #define DOF_SECT_SOURCE 2 /* D program source code */ #define DOF_SECT_ECBDESC 3 /* dof_ecbdesc_t */ #define DOF_SECT_PROBEDESC 4 /* dof_probedesc_t */ #define DOF_SECT_ACTDESC 5 /* dof_actdesc_t array */ #define DOF_SECT_DIFOHDR 6 /* dof_difohdr_t (variable length) */ #define DOF_SECT_DIF 7 /* uint32_t array of byte code */ #define DOF_SECT_STRTAB 8 /* string table */ #define DOF_SECT_VARTAB 9 /* dtrace_difv_t array */ #define DOF_SECT_RELTAB 10 /* dof_relodesc_t array */ #define DOF_SECT_TYPTAB 11 /* dtrace_diftype_t array */ #define DOF_SECT_URELHDR 12 /* dof_relohdr_t (user relocations) */ #define DOF_SECT_KRELHDR 13 /* dof_relohdr_t (kernel relocations) */ #define DOF_SECT_OPTDESC 14 /* dof_optdesc_t array */ #define DOF_SECT_PROVIDER 15 /* dof_provider_t */ #define DOF_SECT_PROBES 16 /* dof_probe_t array */ #define DOF_SECT_PRARGS 17 /* uint8_t array (probe arg mappings) */ #define DOF_SECT_PROFFS 18 /* uint32_t array (probe arg offsets) */ #define DOF_SECT_INTTAB 19 /* uint64_t array */ #define DOF_SECT_UTSNAME 20 /* struct utsname */ #define DOF_SECT_XLTAB 21 /* dof_xlref_t array */ #define DOF_SECT_XLMEMBERS 22 /* dof_xlmember_t array */ #define DOF_SECT_XLIMPORT 23 /* dof_xlator_t */ #define DOF_SECT_XLEXPORT 24 /* dof_xlator_t */ #define DOF_SECT_PREXPORT 25 /* dof_secidx_t array (exported objs) */ #define DOF_SECT_PRENOFFS 26 /* uint32_t array (enabled offsets) */ #define DOF_SECF_LOAD 1 /* section should be loaded */ #define DOF_SEC_ISLOADABLE(x) \ (((x) == DOF_SECT_ECBDESC) || ((x) == DOF_SECT_PROBEDESC) || \ ((x) == DOF_SECT_ACTDESC) || ((x) == DOF_SECT_DIFOHDR) || \ ((x) == DOF_SECT_DIF) || ((x) == DOF_SECT_STRTAB) || \ ((x) == DOF_SECT_VARTAB) || ((x) == DOF_SECT_RELTAB) || \ ((x) == DOF_SECT_TYPTAB) || ((x) == DOF_SECT_URELHDR) || \ ((x) == DOF_SECT_KRELHDR) || ((x) == DOF_SECT_OPTDESC) || \ ((x) == DOF_SECT_PROVIDER) || ((x) == DOF_SECT_PROBES) || \ ((x) == DOF_SECT_PRARGS) || ((x) == DOF_SECT_PROFFS) || \ ((x) == DOF_SECT_INTTAB) || ((x) == DOF_SECT_XLTAB) || \ ((x) == DOF_SECT_XLMEMBERS) || ((x) == DOF_SECT_XLIMPORT) || \ ((x) == DOF_SECT_XLIMPORT) || ((x) == DOF_SECT_XLEXPORT) || \ ((x) == DOF_SECT_PREXPORT) || ((x) == DOF_SECT_PRENOFFS)) typedef struct dof_ecbdesc { dof_secidx_t dofe_probes; /* link to DOF_SECT_PROBEDESC */ dof_secidx_t dofe_pred; /* link to DOF_SECT_DIFOHDR */ dof_secidx_t dofe_actions; /* link to DOF_SECT_ACTDESC */ uint32_t dofe_pad; /* reserved for future use */ uint64_t dofe_uarg; /* user-supplied library argument */ } dof_ecbdesc_t; typedef struct dof_probedesc { dof_secidx_t dofp_strtab; /* link to DOF_SECT_STRTAB section */ dof_stridx_t dofp_provider; /* provider string */ dof_stridx_t dofp_mod; /* module string */ dof_stridx_t dofp_func; /* function string */ dof_stridx_t dofp_name; /* name string */ uint32_t dofp_id; /* probe identifier (or zero) */ } dof_probedesc_t; typedef struct dof_actdesc { dof_secidx_t dofa_difo; /* link to DOF_SECT_DIFOHDR */ dof_secidx_t dofa_strtab; /* link to DOF_SECT_STRTAB section */ uint32_t dofa_kind; /* action kind (DTRACEACT_* constant) */ uint32_t dofa_ntuple; /* number of subsequent tuple actions */ uint64_t dofa_arg; /* kind-specific argument */ uint64_t dofa_uarg; /* user-supplied argument */ } dof_actdesc_t; typedef struct dof_difohdr { dtrace_diftype_t dofd_rtype; /* return type for this fragment */ dof_secidx_t dofd_links[1]; /* variable length array of indices */ } dof_difohdr_t; typedef struct dof_relohdr { dof_secidx_t dofr_strtab; /* link to DOF_SECT_STRTAB for names */ dof_secidx_t dofr_relsec; /* link to DOF_SECT_RELTAB for relos */ dof_secidx_t dofr_tgtsec; /* link to section we are relocating */ } dof_relohdr_t; typedef struct dof_relodesc { dof_stridx_t dofr_name; /* string name of relocation symbol */ uint32_t dofr_type; /* relo type (DOF_RELO_* constant) */ uint64_t dofr_offset; /* byte offset for relocation */ uint64_t dofr_data; /* additional type-specific data */ } dof_relodesc_t; #define DOF_RELO_NONE 0 /* empty relocation entry */ #define DOF_RELO_SETX 1 /* relocate setx value */ typedef struct dof_optdesc { uint32_t dofo_option; /* option identifier */ dof_secidx_t dofo_strtab; /* string table, if string option */ uint64_t dofo_value; /* option value or string index */ } dof_optdesc_t; typedef uint32_t dof_attr_t; /* encoded stability attributes */ #define DOF_ATTR(n, d, c) (((n) << 24) | ((d) << 16) | ((c) << 8)) #define DOF_ATTR_NAME(a) (((a) >> 24) & 0xff) #define DOF_ATTR_DATA(a) (((a) >> 16) & 0xff) #define DOF_ATTR_CLASS(a) (((a) >> 8) & 0xff) typedef struct dof_provider { dof_secidx_t dofpv_strtab; /* link to DOF_SECT_STRTAB section */ dof_secidx_t dofpv_probes; /* link to DOF_SECT_PROBES section */ dof_secidx_t dofpv_prargs; /* link to DOF_SECT_PRARGS section */ dof_secidx_t dofpv_proffs; /* link to DOF_SECT_PROFFS section */ dof_stridx_t dofpv_name; /* provider name string */ dof_attr_t dofpv_provattr; /* provider attributes */ dof_attr_t dofpv_modattr; /* module attributes */ dof_attr_t dofpv_funcattr; /* function attributes */ dof_attr_t dofpv_nameattr; /* name attributes */ dof_attr_t dofpv_argsattr; /* args attributes */ dof_secidx_t dofpv_prenoffs; /* link to DOF_SECT_PRENOFFS section */ } dof_provider_t; typedef struct dof_probe { uint64_t dofpr_addr; /* probe base address or offset */ dof_stridx_t dofpr_func; /* probe function string */ dof_stridx_t dofpr_name; /* probe name string */ dof_stridx_t dofpr_nargv; /* native argument type strings */ dof_stridx_t dofpr_xargv; /* translated argument type strings */ uint32_t dofpr_argidx; /* index of first argument mapping */ uint32_t dofpr_offidx; /* index of first offset entry */ uint8_t dofpr_nargc; /* native argument count */ uint8_t dofpr_xargc; /* translated argument count */ uint16_t dofpr_noffs; /* number of offset entries for probe */ uint32_t dofpr_enoffidx; /* index of first is-enabled offset */ uint16_t dofpr_nenoffs; /* number of is-enabled offsets */ uint16_t dofpr_pad1; /* reserved for future use */ uint32_t dofpr_pad2; /* reserved for future use */ } dof_probe_t; typedef struct dof_xlator { dof_secidx_t dofxl_members; /* link to DOF_SECT_XLMEMBERS section */ dof_secidx_t dofxl_strtab; /* link to DOF_SECT_STRTAB section */ dof_stridx_t dofxl_argv; /* input parameter type strings */ uint32_t dofxl_argc; /* input parameter list length */ dof_stridx_t dofxl_type; /* output type string name */ dof_attr_t dofxl_attr; /* output stability attributes */ } dof_xlator_t; typedef struct dof_xlmember { dof_secidx_t dofxm_difo; /* member link to DOF_SECT_DIFOHDR */ dof_stridx_t dofxm_name; /* member name */ dtrace_diftype_t dofxm_type; /* member type */ } dof_xlmember_t; typedef struct dof_xlref { dof_secidx_t dofxr_xlator; /* link to DOF_SECT_XLATORS section */ uint32_t dofxr_member; /* index of referenced dof_xlmember */ uint32_t dofxr_argn; /* index of argument for DIF_OP_XLARG */ } dof_xlref_t; /* * DTrace Intermediate Format Object (DIFO) * * A DIFO is used to store the compiled DIF for a D expression, its return * type, and its string and variable tables. The string table is a single * buffer of character data into which sets instructions and variable * references can reference strings using a byte offset. The variable table * is an array of dtrace_difv_t structures that describe the name and type of * each variable and the id used in the DIF code. This structure is described * above in the DIF section of this header file. The DIFO is used at both * user-level (in the library) and in the kernel, but the structure is never * passed between the two: the DOF structures form the only interface. As a * result, the definition can change depending on the presence of _KERNEL. */ typedef struct dtrace_difo { dif_instr_t *dtdo_buf; /* instruction buffer */ uint64_t *dtdo_inttab; /* integer table (optional) */ char *dtdo_strtab; /* string table (optional) */ dtrace_difv_t *dtdo_vartab; /* variable table (optional) */ uint_t dtdo_len; /* length of instruction buffer */ uint_t dtdo_intlen; /* length of integer table */ uint_t dtdo_strlen; /* length of string table */ uint_t dtdo_varlen; /* length of variable table */ dtrace_diftype_t dtdo_rtype; /* return type */ uint_t dtdo_refcnt; /* owner reference count */ uint_t dtdo_destructive; /* invokes destructive subroutines */ #ifndef _KERNEL dof_relodesc_t *dtdo_kreltab; /* kernel relocations */ dof_relodesc_t *dtdo_ureltab; /* user relocations */ struct dt_node **dtdo_xlmtab; /* translator references */ uint_t dtdo_krelen; /* length of krelo table */ uint_t dtdo_urelen; /* length of urelo table */ uint_t dtdo_xlmlen; /* length of translator table */ #endif } dtrace_difo_t; /* * DTrace Enabling Description Structures * * When DTrace is tracking the description of a DTrace enabling entity (probe, * predicate, action, ECB, record, etc.), it does so in a description * structure. These structures all end in "desc", and are used at both * user-level and in the kernel -- but (with the exception of * dtrace_probedesc_t) they are never passed between them. Typically, * user-level will use the description structures when assembling an enabling. * It will then distill those description structures into a DOF object (see * above), and send it into the kernel. The kernel will again use the * description structures to create a description of the enabling as it reads * the DOF. When the description is complete, the enabling will be actually * created -- turning it into the structures that represent the enabling * instead of merely describing it. Not surprisingly, the description * structures bear a strong resemblance to the DOF structures that act as their * conduit. */ struct dtrace_predicate; typedef struct dtrace_probedesc { dtrace_id_t dtpd_id; /* probe identifier */ char dtpd_provider[DTRACE_PROVNAMELEN]; /* probe provider name */ char dtpd_mod[DTRACE_MODNAMELEN]; /* probe module name */ char dtpd_func[DTRACE_FUNCNAMELEN]; /* probe function name */ char dtpd_name[DTRACE_NAMELEN]; /* probe name */ } dtrace_probedesc_t; typedef struct dtrace_repldesc { dtrace_probedesc_t dtrpd_match; /* probe descr. to match */ dtrace_probedesc_t dtrpd_create; /* probe descr. to create */ } dtrace_repldesc_t; typedef struct dtrace_preddesc { dtrace_difo_t *dtpdd_difo; /* pointer to DIF object */ struct dtrace_predicate *dtpdd_predicate; /* pointer to predicate */ } dtrace_preddesc_t; typedef struct dtrace_actdesc { dtrace_difo_t *dtad_difo; /* pointer to DIF object */ struct dtrace_actdesc *dtad_next; /* next action */ dtrace_actkind_t dtad_kind; /* kind of action */ uint32_t dtad_ntuple; /* number in tuple */ uint64_t dtad_arg; /* action argument */ uint64_t dtad_uarg; /* user argument */ int dtad_refcnt; /* reference count */ } dtrace_actdesc_t; typedef struct dtrace_ecbdesc { dtrace_actdesc_t *dted_action; /* action description(s) */ dtrace_preddesc_t dted_pred; /* predicate description */ dtrace_probedesc_t dted_probe; /* probe description */ uint64_t dted_uarg; /* library argument */ int dted_refcnt; /* reference count */ } dtrace_ecbdesc_t; /* * DTrace Metadata Description Structures * * DTrace separates the trace data stream from the metadata stream. The only * metadata tokens placed in the data stream are the dtrace_rechdr_t (EPID + * timestamp) or (in the case of aggregations) aggregation identifiers. To * determine the structure of the data, DTrace consumers pass the token to the * kernel, and receive in return a corresponding description of the enabled * probe (via the dtrace_eprobedesc structure) or the aggregation (via the * dtrace_aggdesc structure). Both of these structures are expressed in terms * of record descriptions (via the dtrace_recdesc structure) that describe the * exact structure of the data. Some record descriptions may also contain a * format identifier; this additional bit of metadata can be retrieved from the * kernel, for which a format description is returned via the dtrace_fmtdesc * structure. Note that all four of these structures must be bitness-neutral * to allow for a 32-bit DTrace consumer on a 64-bit kernel. */ typedef struct dtrace_recdesc { dtrace_actkind_t dtrd_action; /* kind of action */ uint32_t dtrd_size; /* size of record */ uint32_t dtrd_offset; /* offset in ECB's data */ uint16_t dtrd_alignment; /* required alignment */ uint16_t dtrd_format; /* format, if any */ uint64_t dtrd_arg; /* action argument */ uint64_t dtrd_uarg; /* user argument */ } dtrace_recdesc_t; typedef struct dtrace_eprobedesc { dtrace_epid_t dtepd_epid; /* enabled probe ID */ dtrace_id_t dtepd_probeid; /* probe ID */ uint64_t dtepd_uarg; /* library argument */ uint32_t dtepd_size; /* total size */ int dtepd_nrecs; /* number of records */ dtrace_recdesc_t dtepd_rec[1]; /* records themselves */ } dtrace_eprobedesc_t; typedef struct dtrace_aggdesc { DTRACE_PTR(char, dtagd_name); /* not filled in by kernel */ dtrace_aggvarid_t dtagd_varid; /* not filled in by kernel */ int dtagd_flags; /* not filled in by kernel */ dtrace_aggid_t dtagd_id; /* aggregation ID */ dtrace_epid_t dtagd_epid; /* enabled probe ID */ uint32_t dtagd_size; /* size in bytes */ int dtagd_nrecs; /* number of records */ uint32_t dtagd_pad; /* explicit padding */ dtrace_recdesc_t dtagd_rec[1]; /* record descriptions */ } dtrace_aggdesc_t; typedef struct dtrace_fmtdesc { DTRACE_PTR(char, dtfd_string); /* format string */ int dtfd_length; /* length of format string */ uint16_t dtfd_format; /* format identifier */ } dtrace_fmtdesc_t; #define DTRACE_SIZEOF_EPROBEDESC(desc) \ (sizeof (dtrace_eprobedesc_t) + ((desc)->dtepd_nrecs ? \ (((desc)->dtepd_nrecs - 1) * sizeof (dtrace_recdesc_t)) : 0)) #define DTRACE_SIZEOF_AGGDESC(desc) \ (sizeof (dtrace_aggdesc_t) + ((desc)->dtagd_nrecs ? \ (((desc)->dtagd_nrecs - 1) * sizeof (dtrace_recdesc_t)) : 0)) /* * DTrace Option Interface * * Run-time DTrace options are set and retrieved via DOF_SECT_OPTDESC sections * in a DOF image. The dof_optdesc structure contains an option identifier and * an option value. The valid option identifiers are found below; the mapping * between option identifiers and option identifying strings is maintained at * user-level. Note that the value of DTRACEOPT_UNSET is such that all of the * following are potentially valid option values: all positive integers, zero * and negative one. Some options (notably "bufpolicy" and "bufresize") take * predefined tokens as their values; these are defined with * DTRACEOPT_{option}_{token}. */ #define DTRACEOPT_BUFSIZE 0 /* buffer size */ #define DTRACEOPT_BUFPOLICY 1 /* buffer policy */ #define DTRACEOPT_DYNVARSIZE 2 /* dynamic variable size */ #define DTRACEOPT_AGGSIZE 3 /* aggregation size */ #define DTRACEOPT_SPECSIZE 4 /* speculation size */ #define DTRACEOPT_NSPEC 5 /* number of speculations */ #define DTRACEOPT_STRSIZE 6 /* string size */ #define DTRACEOPT_CLEANRATE 7 /* dynvar cleaning rate */ #define DTRACEOPT_CPU 8 /* CPU to trace */ #define DTRACEOPT_BUFRESIZE 9 /* buffer resizing policy */ #define DTRACEOPT_GRABANON 10 /* grab anonymous state, if any */ #define DTRACEOPT_FLOWINDENT 11 /* indent function entry/return */ #define DTRACEOPT_QUIET 12 /* only output explicitly traced data */ #define DTRACEOPT_STACKFRAMES 13 /* number of stack frames */ #define DTRACEOPT_USTACKFRAMES 14 /* number of user stack frames */ #define DTRACEOPT_AGGRATE 15 /* aggregation snapshot rate */ #define DTRACEOPT_SWITCHRATE 16 /* buffer switching rate */ #define DTRACEOPT_STATUSRATE 17 /* status rate */ #define DTRACEOPT_DESTRUCTIVE 18 /* destructive actions allowed */ #define DTRACEOPT_STACKINDENT 19 /* output indent for stack traces */ #define DTRACEOPT_RAWBYTES 20 /* always print bytes in raw form */ #define DTRACEOPT_JSTACKFRAMES 21 /* number of jstack() frames */ #define DTRACEOPT_JSTACKSTRSIZE 22 /* size of jstack() string table */ #define DTRACEOPT_AGGSORTKEY 23 /* sort aggregations by key */ #define DTRACEOPT_AGGSORTREV 24 /* reverse-sort aggregations */ #define DTRACEOPT_AGGSORTPOS 25 /* agg. position to sort on */ #define DTRACEOPT_AGGSORTKEYPOS 26 /* agg. key position to sort on */ #define DTRACEOPT_TEMPORAL 27 /* temporally ordered output */ #define DTRACEOPT_AGGHIST 28 /* histogram aggregation output */ #define DTRACEOPT_AGGPACK 29 /* packed aggregation output */ #define DTRACEOPT_AGGZOOM 30 /* zoomed aggregation scaling */ #define DTRACEOPT_ZONE 31 /* zone in which to enable probes */ #define DTRACEOPT_MAX 32 /* number of options */ #define DTRACEOPT_UNSET (dtrace_optval_t)-2 /* unset option */ #define DTRACEOPT_BUFPOLICY_RING 0 /* ring buffer */ #define DTRACEOPT_BUFPOLICY_FILL 1 /* fill buffer, then stop */ #define DTRACEOPT_BUFPOLICY_SWITCH 2 /* switch buffers */ #define DTRACEOPT_BUFRESIZE_AUTO 0 /* automatic resizing */ #define DTRACEOPT_BUFRESIZE_MANUAL 1 /* manual resizing */ /* * DTrace Buffer Interface * * In order to get a snapshot of the principal or aggregation buffer, * user-level passes a buffer description to the kernel with the dtrace_bufdesc * structure. This describes which CPU user-level is interested in, and * where user-level wishes the kernel to snapshot the buffer to (the * dtbd_data field). The kernel uses the same structure to pass back some * information regarding the buffer: the size of data actually copied out, the * number of drops, the number of errors, the offset of the oldest record, * and the time of the snapshot. * * If the buffer policy is a "switch" policy, taking a snapshot of the * principal buffer has the additional effect of switching the active and * inactive buffers. Taking a snapshot of the aggregation buffer _always_ has * the additional effect of switching the active and inactive buffers. */ typedef struct dtrace_bufdesc { uint64_t dtbd_size; /* size of buffer */ uint32_t dtbd_cpu; /* CPU or DTRACE_CPUALL */ uint32_t dtbd_errors; /* number of errors */ uint64_t dtbd_drops; /* number of drops */ DTRACE_PTR(char, dtbd_data); /* data */ uint64_t dtbd_oldest; /* offset of oldest record */ uint64_t dtbd_timestamp; /* hrtime of snapshot */ } dtrace_bufdesc_t; /* * Each record in the buffer (dtbd_data) begins with a header that includes * the epid and a timestamp. The timestamp is split into two 4-byte parts * so that we do not require 8-byte alignment. */ typedef struct dtrace_rechdr { dtrace_epid_t dtrh_epid; /* enabled probe id */ uint32_t dtrh_timestamp_hi; /* high bits of hrtime_t */ uint32_t dtrh_timestamp_lo; /* low bits of hrtime_t */ } dtrace_rechdr_t; #define DTRACE_RECORD_LOAD_TIMESTAMP(dtrh) \ ((dtrh)->dtrh_timestamp_lo + \ ((uint64_t)(dtrh)->dtrh_timestamp_hi << 32)) #define DTRACE_RECORD_STORE_TIMESTAMP(dtrh, hrtime) { \ (dtrh)->dtrh_timestamp_lo = (uint32_t)hrtime; \ (dtrh)->dtrh_timestamp_hi = hrtime >> 32; \ } /* * DTrace Status * * The status of DTrace is relayed via the dtrace_status structure. This * structure contains members to count drops other than the capacity drops * available via the buffer interface (see above). This consists of dynamic * drops (including capacity dynamic drops, rinsing drops and dirty drops), and * speculative drops (including capacity speculative drops, drops due to busy * speculative buffers and drops due to unavailable speculative buffers). * Additionally, the status structure contains a field to indicate the number * of "fill"-policy buffers have been filled and a boolean field to indicate * that exit() has been called. If the dtst_exiting field is non-zero, no * further data will be generated until tracing is stopped (at which time any * enablings of the END action will be processed); if user-level sees that * this field is non-zero, tracing should be stopped as soon as possible. */ typedef struct dtrace_status { uint64_t dtst_dyndrops; /* dynamic drops */ uint64_t dtst_dyndrops_rinsing; /* dyn drops due to rinsing */ uint64_t dtst_dyndrops_dirty; /* dyn drops due to dirty */ uint64_t dtst_specdrops; /* speculative drops */ uint64_t dtst_specdrops_busy; /* spec drops due to busy */ uint64_t dtst_specdrops_unavail; /* spec drops due to unavail */ uint64_t dtst_errors; /* total errors */ uint64_t dtst_filled; /* number of filled bufs */ uint64_t dtst_stkstroverflows; /* stack string tab overflows */ uint64_t dtst_dblerrors; /* errors in ERROR probes */ char dtst_killed; /* non-zero if killed */ char dtst_exiting; /* non-zero if exit() called */ char dtst_pad[6]; /* pad out to 64-bit align */ } dtrace_status_t; /* * DTrace Configuration * * User-level may need to understand some elements of the kernel DTrace * configuration in order to generate correct DIF. This information is * conveyed via the dtrace_conf structure. */ typedef struct dtrace_conf { uint_t dtc_difversion; /* supported DIF version */ uint_t dtc_difintregs; /* # of DIF integer registers */ uint_t dtc_diftupregs; /* # of DIF tuple registers */ uint_t dtc_ctfmodel; /* CTF data model */ uint_t dtc_pad[8]; /* reserved for future use */ } dtrace_conf_t; /* * DTrace Faults * * The constants below DTRACEFLT_LIBRARY indicate probe processing faults; * constants at or above DTRACEFLT_LIBRARY indicate faults in probe * postprocessing at user-level. Probe processing faults induce an ERROR * probe and are replicated in unistd.d to allow users' ERROR probes to decode * the error condition using thse symbolic labels. */ #define DTRACEFLT_UNKNOWN 0 /* Unknown fault */ #define DTRACEFLT_BADADDR 1 /* Bad address */ #define DTRACEFLT_BADALIGN 2 /* Bad alignment */ #define DTRACEFLT_ILLOP 3 /* Illegal operation */ #define DTRACEFLT_DIVZERO 4 /* Divide-by-zero */ #define DTRACEFLT_NOSCRATCH 5 /* Out of scratch space */ #define DTRACEFLT_KPRIV 6 /* Illegal kernel access */ #define DTRACEFLT_UPRIV 7 /* Illegal user access */ #define DTRACEFLT_TUPOFLOW 8 /* Tuple stack overflow */ #define DTRACEFLT_BADSTACK 9 /* Bad stack */ #define DTRACEFLT_LIBRARY 1000 /* Library-level fault */ /* * DTrace Argument Types * * Because it would waste both space and time, argument types do not reside * with the probe. In order to determine argument types for args[X] * variables, the D compiler queries for argument types on a probe-by-probe * basis. (This optimizes for the common case that arguments are either not * used or used in an untyped fashion.) Typed arguments are specified with a * string of the type name in the dtragd_native member of the argument * description structure. Typed arguments may be further translated to types * of greater stability; the provider indicates such a translated argument by * filling in the dtargd_xlate member with the string of the translated type. * Finally, the provider may indicate which argument value a given argument * maps to by setting the dtargd_mapping member -- allowing a single argument * to map to multiple args[X] variables. */ typedef struct dtrace_argdesc { dtrace_id_t dtargd_id; /* probe identifier */ int dtargd_ndx; /* arg number (-1 iff none) */ int dtargd_mapping; /* value mapping */ char dtargd_native[DTRACE_ARGTYPELEN]; /* native type name */ char dtargd_xlate[DTRACE_ARGTYPELEN]; /* translated type name */ } dtrace_argdesc_t; /* * DTrace Stability Attributes * * Each DTrace provider advertises the name and data stability of each of its * probe description components, as well as its architectural dependencies. * The D compiler can query the provider attributes (dtrace_pattr_t below) in * order to compute the properties of an input program and report them. */ typedef uint8_t dtrace_stability_t; /* stability code (see attributes(5)) */ typedef uint8_t dtrace_class_t; /* architectural dependency class */ #define DTRACE_STABILITY_INTERNAL 0 /* private to DTrace itself */ #define DTRACE_STABILITY_PRIVATE 1 /* private to Sun (see docs) */ #define DTRACE_STABILITY_OBSOLETE 2 /* scheduled for removal */ #define DTRACE_STABILITY_EXTERNAL 3 /* not controlled by Sun */ #define DTRACE_STABILITY_UNSTABLE 4 /* new or rapidly changing */ #define DTRACE_STABILITY_EVOLVING 5 /* less rapidly changing */ #define DTRACE_STABILITY_STABLE 6 /* mature interface from Sun */ #define DTRACE_STABILITY_STANDARD 7 /* industry standard */ #define DTRACE_STABILITY_MAX 7 /* maximum valid stability */ #define DTRACE_CLASS_UNKNOWN 0 /* unknown architectural dependency */ #define DTRACE_CLASS_CPU 1 /* CPU-module-specific */ #define DTRACE_CLASS_PLATFORM 2 /* platform-specific (uname -i) */ #define DTRACE_CLASS_GROUP 3 /* hardware-group-specific (uname -m) */ #define DTRACE_CLASS_ISA 4 /* ISA-specific (uname -p) */ #define DTRACE_CLASS_COMMON 5 /* common to all systems */ #define DTRACE_CLASS_MAX 5 /* maximum valid class */ #define DTRACE_PRIV_NONE 0x0000 #define DTRACE_PRIV_KERNEL 0x0001 #define DTRACE_PRIV_USER 0x0002 #define DTRACE_PRIV_PROC 0x0004 #define DTRACE_PRIV_OWNER 0x0008 #define DTRACE_PRIV_ZONEOWNER 0x0010 #define DTRACE_PRIV_ALL \ (DTRACE_PRIV_KERNEL | DTRACE_PRIV_USER | \ DTRACE_PRIV_PROC | DTRACE_PRIV_OWNER | DTRACE_PRIV_ZONEOWNER) typedef struct dtrace_ppriv { uint32_t dtpp_flags; /* privilege flags */ uid_t dtpp_uid; /* user ID */ zoneid_t dtpp_zoneid; /* zone ID */ } dtrace_ppriv_t; typedef struct dtrace_attribute { dtrace_stability_t dtat_name; /* entity name stability */ dtrace_stability_t dtat_data; /* entity data stability */ dtrace_class_t dtat_class; /* entity data dependency */ } dtrace_attribute_t; typedef struct dtrace_pattr { dtrace_attribute_t dtpa_provider; /* provider attributes */ dtrace_attribute_t dtpa_mod; /* module attributes */ dtrace_attribute_t dtpa_func; /* function attributes */ dtrace_attribute_t dtpa_name; /* name attributes */ dtrace_attribute_t dtpa_args; /* args[] attributes */ } dtrace_pattr_t; typedef struct dtrace_providerdesc { char dtvd_name[DTRACE_PROVNAMELEN]; /* provider name */ dtrace_pattr_t dtvd_attr; /* stability attributes */ dtrace_ppriv_t dtvd_priv; /* privileges required */ } dtrace_providerdesc_t; /* * DTrace Pseudodevice Interface * * DTrace is controlled through ioctl(2)'s to the in-kernel dtrace:dtrace * pseudodevice driver. These ioctls comprise the user-kernel interface to * DTrace. */ #ifdef illumos #define DTRACEIOC (('d' << 24) | ('t' << 16) | ('r' << 8)) #define DTRACEIOC_PROVIDER (DTRACEIOC | 1) /* provider query */ #define DTRACEIOC_PROBES (DTRACEIOC | 2) /* probe query */ #define DTRACEIOC_BUFSNAP (DTRACEIOC | 4) /* snapshot buffer */ #define DTRACEIOC_PROBEMATCH (DTRACEIOC | 5) /* match probes */ #define DTRACEIOC_ENABLE (DTRACEIOC | 6) /* enable probes */ #define DTRACEIOC_AGGSNAP (DTRACEIOC | 7) /* snapshot agg. */ #define DTRACEIOC_EPROBE (DTRACEIOC | 8) /* get eprobe desc. */ #define DTRACEIOC_PROBEARG (DTRACEIOC | 9) /* get probe arg */ #define DTRACEIOC_CONF (DTRACEIOC | 10) /* get config. */ #define DTRACEIOC_STATUS (DTRACEIOC | 11) /* get status */ #define DTRACEIOC_GO (DTRACEIOC | 12) /* start tracing */ #define DTRACEIOC_STOP (DTRACEIOC | 13) /* stop tracing */ #define DTRACEIOC_AGGDESC (DTRACEIOC | 15) /* get agg. desc. */ #define DTRACEIOC_FORMAT (DTRACEIOC | 16) /* get format str */ #define DTRACEIOC_DOFGET (DTRACEIOC | 17) /* get DOF */ #define DTRACEIOC_REPLICATE (DTRACEIOC | 18) /* replicate enab */ #else #define DTRACEIOC_PROVIDER _IOWR('x',1,dtrace_providerdesc_t) /* provider query */ #define DTRACEIOC_PROBES _IOWR('x',2,dtrace_probedesc_t) /* probe query */ #define DTRACEIOC_BUFSNAP _IOW('x',4,dtrace_bufdesc_t *) /* snapshot buffer */ #define DTRACEIOC_PROBEMATCH _IOWR('x',5,dtrace_probedesc_t) /* match probes */ typedef struct { void *dof; /* DOF userland address written to driver. */ int n_matched; /* # matches returned by driver. */ } dtrace_enable_io_t; #define DTRACEIOC_ENABLE _IOWR('x',6,dtrace_enable_io_t) /* enable probes */ #define DTRACEIOC_AGGSNAP _IOW('x',7,dtrace_bufdesc_t *) /* snapshot agg. */ #define DTRACEIOC_EPROBE _IOW('x',8,dtrace_eprobedesc_t) /* get eprobe desc. */ #define DTRACEIOC_PROBEARG _IOWR('x',9,dtrace_argdesc_t) /* get probe arg */ #define DTRACEIOC_CONF _IOR('x',10,dtrace_conf_t) /* get config. */ #define DTRACEIOC_STATUS _IOR('x',11,dtrace_status_t) /* get status */ #define DTRACEIOC_GO _IOR('x',12,processorid_t) /* start tracing */ #define DTRACEIOC_STOP _IOWR('x',13,processorid_t) /* stop tracing */ #define DTRACEIOC_AGGDESC _IOW('x',15,dtrace_aggdesc_t *) /* get agg. desc. */ #define DTRACEIOC_FORMAT _IOWR('x',16,dtrace_fmtdesc_t) /* get format str */ #define DTRACEIOC_DOFGET _IOW('x',17,dof_hdr_t *) /* get DOF */ #define DTRACEIOC_REPLICATE _IOW('x',18,dtrace_repldesc_t) /* replicate enab */ #endif /* * DTrace Helpers * * In general, DTrace establishes probes in processes and takes actions on * processes without knowing their specific user-level structures. Instead of * existing in the framework, process-specific knowledge is contained by the * enabling D program -- which can apply process-specific knowledge by making * appropriate use of DTrace primitives like copyin() and copyinstr() to * operate on user-level data. However, there may exist some specific probes * of particular semantic relevance that the application developer may wish to * explicitly export. For example, an application may wish to export a probe * at the point that it begins and ends certain well-defined transactions. In * addition to providing probes, programs may wish to offer assistance for * certain actions. For example, in highly dynamic environments (e.g., Java), * it may be difficult to obtain a stack trace in terms of meaningful symbol * names (the translation from instruction addresses to corresponding symbol * names may only be possible in situ); these environments may wish to define * a series of actions to be applied in situ to obtain a meaningful stack * trace. * * These two mechanisms -- user-level statically defined tracing and assisting * DTrace actions -- are provided via DTrace _helpers_. Helpers are specified * via DOF, but unlike enabling DOF, helper DOF may contain definitions of * providers, probes and their arguments. If a helper wishes to provide * action assistance, probe descriptions and corresponding DIF actions may be * specified in the helper DOF. For such helper actions, however, the probe * description describes the specific helper: all DTrace helpers have the * provider name "dtrace" and the module name "helper", and the name of the * helper is contained in the function name (for example, the ustack() helper * is named "ustack"). Any helper-specific name may be contained in the name * (for example, if a helper were to have a constructor, it might be named * "dtrace:helper::init"). Helper actions are only called when the * action that they are helping is taken. Helper actions may only return DIF * expressions, and may only call the following subroutines: * * alloca() <= Allocates memory out of the consumer's scratch space * bcopy() <= Copies memory to scratch space * copyin() <= Copies memory from user-level into consumer's scratch * copyinto() <= Copies memory into a specific location in scratch * copyinstr() <= Copies a string into a specific location in scratch * * Helper actions may only access the following built-in variables: * * curthread <= Current kthread_t pointer * tid <= Current thread identifier * pid <= Current process identifier * ppid <= Parent process identifier * uid <= Current user ID * gid <= Current group ID * execname <= Current executable name * zonename <= Current zone name * * Helper actions may not manipulate or allocate dynamic variables, but they * may have clause-local and statically-allocated global variables. The * helper action variable state is specific to the helper action -- variables * used by the helper action may not be accessed outside of the helper * action, and the helper action may not access variables that like outside * of it. Helper actions may not load from kernel memory at-large; they are * restricting to loading current user state (via copyin() and variants) and * scratch space. As with probe enablings, helper actions are executed in * program order. The result of the helper action is the result of the last * executing helper expression. * * Helpers -- composed of either providers/probes or probes/actions (or both) * -- are added by opening the "helper" minor node, and issuing an ioctl(2) * (DTRACEHIOC_ADDDOF) that specifies the dof_helper_t structure. This * encapsulates the name and base address of the user-level library or * executable publishing the helpers and probes as well as the DOF that * contains the definitions of those helpers and probes. * * The DTRACEHIOC_ADD and DTRACEHIOC_REMOVE are left in place for legacy * helpers and should no longer be used. No other ioctls are valid on the * helper minor node. */ #ifdef illumos #define DTRACEHIOC (('d' << 24) | ('t' << 16) | ('h' << 8)) #define DTRACEHIOC_ADD (DTRACEHIOC | 1) /* add helper */ #define DTRACEHIOC_REMOVE (DTRACEHIOC | 2) /* remove helper */ #define DTRACEHIOC_ADDDOF (DTRACEHIOC | 3) /* add helper DOF */ #else #define DTRACEHIOC_ADD _IOWR('z', 1, dof_hdr_t)/* add helper */ #define DTRACEHIOC_REMOVE _IOW('z', 2, int) /* remove helper */ #define DTRACEHIOC_ADDDOF _IOWR('z', 3, dof_helper_t)/* add helper DOF */ #endif typedef struct dof_helper { char dofhp_mod[DTRACE_MODNAMELEN]; /* executable or library name */ uint64_t dofhp_addr; /* base address of object */ uint64_t dofhp_dof; /* address of helper DOF */ #ifndef illumos int gen; #endif } dof_helper_t; #define DTRACEMNR_DTRACE "dtrace" /* node for DTrace ops */ #define DTRACEMNR_HELPER "helper" /* node for helpers */ #define DTRACEMNRN_DTRACE 0 /* minor for DTrace ops */ #define DTRACEMNRN_HELPER 1 /* minor for helpers */ #define DTRACEMNRN_CLONE 2 /* first clone minor */ #ifdef _KERNEL /* * DTrace Provider API * * The following functions are implemented by the DTrace framework and are * used to implement separate in-kernel DTrace providers. Common functions * are provided in uts/common/os/dtrace.c. ISA-dependent subroutines are * defined in uts//dtrace/dtrace_asm.s or uts//dtrace/dtrace_isa.c. * * The provider API has two halves: the API that the providers consume from * DTrace, and the API that providers make available to DTrace. * * 1 Framework-to-Provider API * * 1.1 Overview * * The Framework-to-Provider API is represented by the dtrace_pops structure * that the provider passes to the framework when registering itself. This * structure consists of the following members: * * dtps_provide() <-- Provide all probes, all modules * dtps_provide_module() <-- Provide all probes in specified module * dtps_enable() <-- Enable specified probe * dtps_disable() <-- Disable specified probe * dtps_suspend() <-- Suspend specified probe * dtps_resume() <-- Resume specified probe * dtps_getargdesc() <-- Get the argument description for args[X] * dtps_getargval() <-- Get the value for an argX or args[X] variable * dtps_usermode() <-- Find out if the probe was fired in user mode * dtps_destroy() <-- Destroy all state associated with this probe * * 1.2 void dtps_provide(void *arg, const dtrace_probedesc_t *spec) * * 1.2.1 Overview * * Called to indicate that the provider should provide all probes. If the * specified description is non-NULL, dtps_provide() is being called because * no probe matched a specified probe -- if the provider has the ability to * create custom probes, it may wish to create a probe that matches the * specified description. * * 1.2.2 Arguments and notes * * The first argument is the cookie as passed to dtrace_register(). The * second argument is a pointer to a probe description that the provider may * wish to consider when creating custom probes. The provider is expected to * call back into the DTrace framework via dtrace_probe_create() to create * any necessary probes. dtps_provide() may be called even if the provider * has made available all probes; the provider should check the return value * of dtrace_probe_create() to handle this case. Note that the provider need * not implement both dtps_provide() and dtps_provide_module(); see * "Arguments and Notes" for dtrace_register(), below. * * 1.2.3 Return value * * None. * * 1.2.4 Caller's context * * dtps_provide() is typically called from open() or ioctl() context, but may * be called from other contexts as well. The DTrace framework is locked in * such a way that providers may not register or unregister. This means that * the provider may not call any DTrace API that affects its registration with * the framework, including dtrace_register(), dtrace_unregister(), * dtrace_invalidate(), and dtrace_condense(). However, the context is such * that the provider may (and indeed, is expected to) call probe-related * DTrace routines, including dtrace_probe_create(), dtrace_probe_lookup(), * and dtrace_probe_arg(). * * 1.3 void dtps_provide_module(void *arg, modctl_t *mp) * * 1.3.1 Overview * * Called to indicate that the provider should provide all probes in the * specified module. * * 1.3.2 Arguments and notes * * The first argument is the cookie as passed to dtrace_register(). The * second argument is a pointer to a modctl structure that indicates the * module for which probes should be created. * * 1.3.3 Return value * * None. * * 1.3.4 Caller's context * * dtps_provide_module() may be called from open() or ioctl() context, but * may also be called from a module loading context. mod_lock is held, and * the DTrace framework is locked in such a way that providers may not * register or unregister. This means that the provider may not call any * DTrace API that affects its registration with the framework, including * dtrace_register(), dtrace_unregister(), dtrace_invalidate(), and * dtrace_condense(). However, the context is such that the provider may (and * indeed, is expected to) call probe-related DTrace routines, including * dtrace_probe_create(), dtrace_probe_lookup(), and dtrace_probe_arg(). Note * that the provider need not implement both dtps_provide() and * dtps_provide_module(); see "Arguments and Notes" for dtrace_register(), * below. * * 1.4 void dtps_enable(void *arg, dtrace_id_t id, void *parg) * * 1.4.1 Overview * * Called to enable the specified probe. * * 1.4.2 Arguments and notes * * The first argument is the cookie as passed to dtrace_register(). The * second argument is the identifier of the probe to be enabled. The third * argument is the probe argument as passed to dtrace_probe_create(). * dtps_enable() will be called when a probe transitions from not being * enabled at all to having one or more ECB. The number of ECBs associated * with the probe may change without subsequent calls into the provider. * When the number of ECBs drops to zero, the provider will be explicitly * told to disable the probe via dtps_disable(). dtrace_probe() should never * be called for a probe identifier that hasn't been explicitly enabled via * dtps_enable(). * * 1.4.3 Return value * * None. * * 1.4.4 Caller's context * * The DTrace framework is locked in such a way that it may not be called * back into at all. cpu_lock is held. mod_lock is not held and may not * be acquired. * * 1.5 void dtps_disable(void *arg, dtrace_id_t id, void *parg) * * 1.5.1 Overview * * Called to disable the specified probe. * * 1.5.2 Arguments and notes * * The first argument is the cookie as passed to dtrace_register(). The * second argument is the identifier of the probe to be disabled. The third * argument is the probe argument as passed to dtrace_probe_create(). * dtps_disable() will be called when a probe transitions from being enabled * to having zero ECBs. dtrace_probe() should never be called for a probe * identifier that has been explicitly enabled via dtps_disable(). * * 1.5.3 Return value * * None. * * 1.5.4 Caller's context * * The DTrace framework is locked in such a way that it may not be called * back into at all. cpu_lock is held. mod_lock is not held and may not * be acquired. * * 1.6 void dtps_suspend(void *arg, dtrace_id_t id, void *parg) * * 1.6.1 Overview * * Called to suspend the specified enabled probe. This entry point is for * providers that may need to suspend some or all of their probes when CPUs * are being powered on or when the boot monitor is being entered for a * prolonged period of time. * * 1.6.2 Arguments and notes * * The first argument is the cookie as passed to dtrace_register(). The * second argument is the identifier of the probe to be suspended. The * third argument is the probe argument as passed to dtrace_probe_create(). * dtps_suspend will only be called on an enabled probe. Providers that * provide a dtps_suspend entry point will want to take roughly the action * that it takes for dtps_disable. * * 1.6.3 Return value * * None. * * 1.6.4 Caller's context * * Interrupts are disabled. The DTrace framework is in a state such that the * specified probe cannot be disabled or destroyed for the duration of * dtps_suspend(). As interrupts are disabled, the provider is afforded * little latitude; the provider is expected to do no more than a store to * memory. * * 1.7 void dtps_resume(void *arg, dtrace_id_t id, void *parg) * * 1.7.1 Overview * * Called to resume the specified enabled probe. This entry point is for * providers that may need to resume some or all of their probes after the * completion of an event that induced a call to dtps_suspend(). * * 1.7.2 Arguments and notes * * The first argument is the cookie as passed to dtrace_register(). The * second argument is the identifier of the probe to be resumed. The * third argument is the probe argument as passed to dtrace_probe_create(). * dtps_resume will only be called on an enabled probe. Providers that * provide a dtps_resume entry point will want to take roughly the action * that it takes for dtps_enable. * * 1.7.3 Return value * * None. * * 1.7.4 Caller's context * * Interrupts are disabled. The DTrace framework is in a state such that the * specified probe cannot be disabled or destroyed for the duration of * dtps_resume(). As interrupts are disabled, the provider is afforded * little latitude; the provider is expected to do no more than a store to * memory. * * 1.8 void dtps_getargdesc(void *arg, dtrace_id_t id, void *parg, * dtrace_argdesc_t *desc) * * 1.8.1 Overview * * Called to retrieve the argument description for an args[X] variable. * * 1.8.2 Arguments and notes * * The first argument is the cookie as passed to dtrace_register(). The * second argument is the identifier of the current probe. The third * argument is the probe argument as passed to dtrace_probe_create(). The * fourth argument is a pointer to the argument description. This * description is both an input and output parameter: it contains the * index of the desired argument in the dtargd_ndx field, and expects * the other fields to be filled in upon return. If there is no argument * corresponding to the specified index, the dtargd_ndx field should be set * to DTRACE_ARGNONE. * * 1.8.3 Return value * * None. The dtargd_ndx, dtargd_native, dtargd_xlate and dtargd_mapping * members of the dtrace_argdesc_t structure are all output values. * * 1.8.4 Caller's context * * dtps_getargdesc() is called from ioctl() context. mod_lock is held, and * the DTrace framework is locked in such a way that providers may not * register or unregister. This means that the provider may not call any * DTrace API that affects its registration with the framework, including * dtrace_register(), dtrace_unregister(), dtrace_invalidate(), and * dtrace_condense(). * * 1.9 uint64_t dtps_getargval(void *arg, dtrace_id_t id, void *parg, * int argno, int aframes) * * 1.9.1 Overview * * Called to retrieve a value for an argX or args[X] variable. * * 1.9.2 Arguments and notes * * The first argument is the cookie as passed to dtrace_register(). The * second argument is the identifier of the current probe. The third * argument is the probe argument as passed to dtrace_probe_create(). The * fourth argument is the number of the argument (the X in the example in * 1.9.1). The fifth argument is the number of stack frames that were used * to get from the actual place in the code that fired the probe to * dtrace_probe() itself, the so-called artificial frames. This argument may * be used to descend an appropriate number of frames to find the correct * values. If this entry point is left NULL, the dtrace_getarg() built-in * function is used. * * 1.9.3 Return value * * The value of the argument. * * 1.9.4 Caller's context * * This is called from within dtrace_probe() meaning that interrupts * are disabled. No locks should be taken within this entry point. * * 1.10 int dtps_usermode(void *arg, dtrace_id_t id, void *parg) * * 1.10.1 Overview * * Called to determine if the probe was fired in a user context. * * 1.10.2 Arguments and notes * * The first argument is the cookie as passed to dtrace_register(). The * second argument is the identifier of the current probe. The third * argument is the probe argument as passed to dtrace_probe_create(). This * entry point must not be left NULL for providers whose probes allow for * mixed mode tracing, that is to say those probes that can fire during * kernel- _or_ user-mode execution * * 1.10.3 Return value * * A bitwise OR that encapsulates both the mode (either DTRACE_MODE_KERNEL * or DTRACE_MODE_USER) and the policy when the privilege of the enabling * is insufficient for that mode (a combination of DTRACE_MODE_NOPRIV_DROP, * DTRACE_MODE_NOPRIV_RESTRICT, and DTRACE_MODE_LIMITEDPRIV_RESTRICT). If * DTRACE_MODE_NOPRIV_DROP bit is set, insufficient privilege will result * in the probe firing being silently ignored for the enabling; if the * DTRACE_NODE_NOPRIV_RESTRICT bit is set, insufficient privilege will not * prevent probe processing for the enabling, but restrictions will be in * place that induce a UPRIV fault upon attempt to examine probe arguments * or current process state. If the DTRACE_MODE_LIMITEDPRIV_RESTRICT bit * is set, similar restrictions will be placed upon operation if the * privilege is sufficient to process the enabling, but does not otherwise * entitle the enabling to all zones. The DTRACE_MODE_NOPRIV_DROP and * DTRACE_MODE_NOPRIV_RESTRICT are mutually exclusive (and one of these * two policies must be specified), but either may be combined (or not) * with DTRACE_MODE_LIMITEDPRIV_RESTRICT. * * 1.10.4 Caller's context * * This is called from within dtrace_probe() meaning that interrupts * are disabled. No locks should be taken within this entry point. * * 1.11 void dtps_destroy(void *arg, dtrace_id_t id, void *parg) * * 1.11.1 Overview * * Called to destroy the specified probe. * * 1.11.2 Arguments and notes * * The first argument is the cookie as passed to dtrace_register(). The * second argument is the identifier of the probe to be destroyed. The third * argument is the probe argument as passed to dtrace_probe_create(). The * provider should free all state associated with the probe. The framework * guarantees that dtps_destroy() is only called for probes that have either * been disabled via dtps_disable() or were never enabled via dtps_enable(). * Once dtps_disable() has been called for a probe, no further call will be * made specifying the probe. * * 1.11.3 Return value * * None. * * 1.11.4 Caller's context * * The DTrace framework is locked in such a way that it may not be called * back into at all. mod_lock is held. cpu_lock is not held, and may not be * acquired. * * * 2 Provider-to-Framework API * * 2.1 Overview * * The Provider-to-Framework API provides the mechanism for the provider to * register itself with the DTrace framework, to create probes, to lookup * probes and (most importantly) to fire probes. The Provider-to-Framework * consists of: * * dtrace_register() <-- Register a provider with the DTrace framework * dtrace_unregister() <-- Remove a provider's DTrace registration * dtrace_invalidate() <-- Invalidate the specified provider * dtrace_condense() <-- Remove a provider's unenabled probes * dtrace_attached() <-- Indicates whether or not DTrace has attached * dtrace_probe_create() <-- Create a DTrace probe * dtrace_probe_lookup() <-- Lookup a DTrace probe based on its name * dtrace_probe_arg() <-- Return the probe argument for a specific probe * dtrace_probe() <-- Fire the specified probe * * 2.2 int dtrace_register(const char *name, const dtrace_pattr_t *pap, * uint32_t priv, cred_t *cr, const dtrace_pops_t *pops, void *arg, * dtrace_provider_id_t *idp) * * 2.2.1 Overview * * dtrace_register() registers the calling provider with the DTrace * framework. It should generally be called by DTrace providers in their * attach(9E) entry point. * * 2.2.2 Arguments and Notes * * The first argument is the name of the provider. The second argument is a * pointer to the stability attributes for the provider. The third argument * is the privilege flags for the provider, and must be some combination of: * * DTRACE_PRIV_NONE <= All users may enable probes from this provider * * DTRACE_PRIV_PROC <= Any user with privilege of PRIV_DTRACE_PROC may * enable probes from this provider * * DTRACE_PRIV_USER <= Any user with privilege of PRIV_DTRACE_USER may * enable probes from this provider * * DTRACE_PRIV_KERNEL <= Any user with privilege of PRIV_DTRACE_KERNEL * may enable probes from this provider * * DTRACE_PRIV_OWNER <= This flag places an additional constraint on * the privilege requirements above. These probes * require either (a) a user ID matching the user * ID of the cred passed in the fourth argument * or (b) the PRIV_PROC_OWNER privilege. * * DTRACE_PRIV_ZONEOWNER<= This flag places an additional constraint on * the privilege requirements above. These probes * require either (a) a zone ID matching the zone * ID of the cred passed in the fourth argument * or (b) the PRIV_PROC_ZONE privilege. * * Note that these flags designate the _visibility_ of the probes, not * the conditions under which they may or may not fire. * * The fourth argument is the credential that is associated with the * provider. This argument should be NULL if the privilege flags don't * include DTRACE_PRIV_OWNER or DTRACE_PRIV_ZONEOWNER. If non-NULL, the * framework stashes the uid and zoneid represented by this credential * for use at probe-time, in implicit predicates. These limit visibility * of the probes to users and/or zones which have sufficient privilege to * access them. * * The fifth argument is a DTrace provider operations vector, which provides * the implementation for the Framework-to-Provider API. (See Section 1, * above.) This must be non-NULL, and each member must be non-NULL. The * exceptions to this are (1) the dtps_provide() and dtps_provide_module() * members (if the provider so desires, _one_ of these members may be left * NULL -- denoting that the provider only implements the other) and (2) * the dtps_suspend() and dtps_resume() members, which must either both be * NULL or both be non-NULL. * * The sixth argument is a cookie to be specified as the first argument for * each function in the Framework-to-Provider API. This argument may have * any value. * * The final argument is a pointer to dtrace_provider_id_t. If * dtrace_register() successfully completes, the provider identifier will be * stored in the memory pointed to be this argument. This argument must be * non-NULL. * * 2.2.3 Return value * * On success, dtrace_register() returns 0 and stores the new provider's * identifier into the memory pointed to by the idp argument. On failure, * dtrace_register() returns an errno: * * EINVAL The arguments passed to dtrace_register() were somehow invalid. * This may because a parameter that must be non-NULL was NULL, * because the name was invalid (either empty or an illegal * provider name) or because the attributes were invalid. * * No other failure code is returned. * * 2.2.4 Caller's context * * dtrace_register() may induce calls to dtrace_provide(); the provider must * hold no locks across dtrace_register() that may also be acquired by * dtrace_provide(). cpu_lock and mod_lock must not be held. * * 2.3 int dtrace_unregister(dtrace_provider_t id) * * 2.3.1 Overview * * Unregisters the specified provider from the DTrace framework. It should * generally be called by DTrace providers in their detach(9E) entry point. * * 2.3.2 Arguments and Notes * * The only argument is the provider identifier, as returned from a * successful call to dtrace_register(). As a result of calling * dtrace_unregister(), the DTrace framework will call back into the provider * via the dtps_destroy() entry point. Once dtrace_unregister() successfully * completes, however, the DTrace framework will no longer make calls through * the Framework-to-Provider API. * * 2.3.3 Return value * * On success, dtrace_unregister returns 0. On failure, dtrace_unregister() * returns an errno: * * EBUSY There are currently processes that have the DTrace pseudodevice * open, or there exists an anonymous enabling that hasn't yet * been claimed. * * No other failure code is returned. * * 2.3.4 Caller's context * * Because a call to dtrace_unregister() may induce calls through the * Framework-to-Provider API, the caller may not hold any lock across * dtrace_register() that is also acquired in any of the Framework-to- * Provider API functions. Additionally, mod_lock may not be held. * * 2.4 void dtrace_invalidate(dtrace_provider_id_t id) * * 2.4.1 Overview * * Invalidates the specified provider. All subsequent probe lookups for the * specified provider will fail, but its probes will not be removed. * * 2.4.2 Arguments and note * * The only argument is the provider identifier, as returned from a * successful call to dtrace_register(). In general, a provider's probes * always remain valid; dtrace_invalidate() is a mechanism for invalidating * an entire provider, regardless of whether or not probes are enabled or * not. Note that dtrace_invalidate() will _not_ prevent already enabled * probes from firing -- it will merely prevent any new enablings of the * provider's probes. * * 2.5 int dtrace_condense(dtrace_provider_id_t id) * * 2.5.1 Overview * * Removes all the unenabled probes for the given provider. This function is * not unlike dtrace_unregister(), except that it doesn't remove the * provider just as many of its associated probes as it can. * * 2.5.2 Arguments and Notes * * As with dtrace_unregister(), the sole argument is the provider identifier * as returned from a successful call to dtrace_register(). As a result of * calling dtrace_condense(), the DTrace framework will call back into the * given provider's dtps_destroy() entry point for each of the provider's * unenabled probes. * * 2.5.3 Return value * * Currently, dtrace_condense() always returns 0. However, consumers of this * function should check the return value as appropriate; its behavior may * change in the future. * * 2.5.4 Caller's context * * As with dtrace_unregister(), the caller may not hold any lock across * dtrace_condense() that is also acquired in the provider's entry points. * Also, mod_lock may not be held. * * 2.6 int dtrace_attached() * * 2.6.1 Overview * * Indicates whether or not DTrace has attached. * * 2.6.2 Arguments and Notes * * For most providers, DTrace makes initial contact beyond registration. * That is, once a provider has registered with DTrace, it waits to hear * from DTrace to create probes. However, some providers may wish to * proactively create probes without first being told by DTrace to do so. * If providers wish to do this, they must first call dtrace_attached() to * determine if DTrace itself has attached. If dtrace_attached() returns 0, * the provider must not make any other Provider-to-Framework API call. * * 2.6.3 Return value * * dtrace_attached() returns 1 if DTrace has attached, 0 otherwise. * * 2.7 int dtrace_probe_create(dtrace_provider_t id, const char *mod, * const char *func, const char *name, int aframes, void *arg) * * 2.7.1 Overview * * Creates a probe with specified module name, function name, and name. * * 2.7.2 Arguments and Notes * * The first argument is the provider identifier, as returned from a * successful call to dtrace_register(). The second, third, and fourth * arguments are the module name, function name, and probe name, * respectively. Of these, module name and function name may both be NULL * (in which case the probe is considered to be unanchored), or they may both * be non-NULL. The name must be non-NULL, and must point to a non-empty * string. * * The fifth argument is the number of artificial stack frames that will be * found on the stack when dtrace_probe() is called for the new probe. These * artificial frames will be automatically be pruned should the stack() or * stackdepth() functions be called as part of one of the probe's ECBs. If * the parameter doesn't add an artificial frame, this parameter should be * zero. * * The final argument is a probe argument that will be passed back to the * provider when a probe-specific operation is called. (e.g., via * dtps_enable(), dtps_disable(), etc.) * * Note that it is up to the provider to be sure that the probe that it * creates does not already exist -- if the provider is unsure of the probe's * existence, it should assure its absence with dtrace_probe_lookup() before * calling dtrace_probe_create(). * * 2.7.3 Return value * * dtrace_probe_create() always succeeds, and always returns the identifier * of the newly-created probe. * * 2.7.4 Caller's context * * While dtrace_probe_create() is generally expected to be called from * dtps_provide() and/or dtps_provide_module(), it may be called from other * non-DTrace contexts. Neither cpu_lock nor mod_lock may be held. * * 2.8 dtrace_id_t dtrace_probe_lookup(dtrace_provider_t id, const char *mod, * const char *func, const char *name) * * 2.8.1 Overview * * Looks up a probe based on provdider and one or more of module name, * function name and probe name. * * 2.8.2 Arguments and Notes * * The first argument is the provider identifier, as returned from a * successful call to dtrace_register(). The second, third, and fourth * arguments are the module name, function name, and probe name, * respectively. Any of these may be NULL; dtrace_probe_lookup() will return * the identifier of the first probe that is provided by the specified * provider and matches all of the non-NULL matching criteria. * dtrace_probe_lookup() is generally used by a provider to be check the * existence of a probe before creating it with dtrace_probe_create(). * * 2.8.3 Return value * * If the probe exists, returns its identifier. If the probe does not exist, * return DTRACE_IDNONE. * * 2.8.4 Caller's context * * While dtrace_probe_lookup() is generally expected to be called from * dtps_provide() and/or dtps_provide_module(), it may also be called from * other non-DTrace contexts. Neither cpu_lock nor mod_lock may be held. * * 2.9 void *dtrace_probe_arg(dtrace_provider_t id, dtrace_id_t probe) * * 2.9.1 Overview * * Returns the probe argument associated with the specified probe. * * 2.9.2 Arguments and Notes * * The first argument is the provider identifier, as returned from a * successful call to dtrace_register(). The second argument is a probe * identifier, as returned from dtrace_probe_lookup() or * dtrace_probe_create(). This is useful if a probe has multiple * provider-specific components to it: the provider can create the probe * once with provider-specific state, and then add to the state by looking * up the probe based on probe identifier. * * 2.9.3 Return value * * Returns the argument associated with the specified probe. If the * specified probe does not exist, or if the specified probe is not provided * by the specified provider, NULL is returned. * * 2.9.4 Caller's context * * While dtrace_probe_arg() is generally expected to be called from * dtps_provide() and/or dtps_provide_module(), it may also be called from * other non-DTrace contexts. Neither cpu_lock nor mod_lock may be held. * * 2.10 void dtrace_probe(dtrace_id_t probe, uintptr_t arg0, uintptr_t arg1, * uintptr_t arg2, uintptr_t arg3, uintptr_t arg4) * * 2.10.1 Overview * * The epicenter of DTrace: fires the specified probes with the specified * arguments. * * 2.10.2 Arguments and Notes * * The first argument is a probe identifier as returned by * dtrace_probe_create() or dtrace_probe_lookup(). The second through sixth * arguments are the values to which the D variables "arg0" through "arg4" * will be mapped. * * dtrace_probe() should be called whenever the specified probe has fired -- * however the provider defines it. * * 2.10.3 Return value * * None. * * 2.10.4 Caller's context * * dtrace_probe() may be called in virtually any context: kernel, user, * interrupt, high-level interrupt, with arbitrary adaptive locks held, with * dispatcher locks held, with interrupts disabled, etc. The only latitude * that must be afforded to DTrace is the ability to make calls within * itself (and to its in-kernel subroutines) and the ability to access * arbitrary (but mapped) memory. On some platforms, this constrains * context. For example, on UltraSPARC, dtrace_probe() cannot be called * from any context in which TL is greater than zero. dtrace_probe() may * also not be called from any routine which may be called by dtrace_probe() * -- which includes functions in the DTrace framework and some in-kernel * DTrace subroutines. All such functions "dtrace_"; providers that * instrument the kernel arbitrarily should be sure to not instrument these * routines. */ typedef struct dtrace_pops { void (*dtps_provide)(void *arg, dtrace_probedesc_t *spec); void (*dtps_provide_module)(void *arg, modctl_t *mp); void (*dtps_enable)(void *arg, dtrace_id_t id, void *parg); void (*dtps_disable)(void *arg, dtrace_id_t id, void *parg); void (*dtps_suspend)(void *arg, dtrace_id_t id, void *parg); void (*dtps_resume)(void *arg, dtrace_id_t id, void *parg); void (*dtps_getargdesc)(void *arg, dtrace_id_t id, void *parg, dtrace_argdesc_t *desc); uint64_t (*dtps_getargval)(void *arg, dtrace_id_t id, void *parg, int argno, int aframes); int (*dtps_usermode)(void *arg, dtrace_id_t id, void *parg); void (*dtps_destroy)(void *arg, dtrace_id_t id, void *parg); } dtrace_pops_t; #define DTRACE_MODE_KERNEL 0x01 #define DTRACE_MODE_USER 0x02 #define DTRACE_MODE_NOPRIV_DROP 0x10 #define DTRACE_MODE_NOPRIV_RESTRICT 0x20 #define DTRACE_MODE_LIMITEDPRIV_RESTRICT 0x40 typedef uintptr_t dtrace_provider_id_t; extern int dtrace_register(const char *, const dtrace_pattr_t *, uint32_t, cred_t *, const dtrace_pops_t *, void *, dtrace_provider_id_t *); extern int dtrace_unregister(dtrace_provider_id_t); extern int dtrace_condense(dtrace_provider_id_t); extern void dtrace_invalidate(dtrace_provider_id_t); extern dtrace_id_t dtrace_probe_lookup(dtrace_provider_id_t, char *, char *, char *); extern dtrace_id_t dtrace_probe_create(dtrace_provider_id_t, const char *, const char *, const char *, int, void *); extern void *dtrace_probe_arg(dtrace_provider_id_t, dtrace_id_t); extern void dtrace_probe(dtrace_id_t, uintptr_t arg0, uintptr_t arg1, uintptr_t arg2, uintptr_t arg3, uintptr_t arg4); /* * DTrace Meta Provider API * * The following functions are implemented by the DTrace framework and are * used to implement meta providers. Meta providers plug into the DTrace * framework and are used to instantiate new providers on the fly. At * present, there is only one type of meta provider and only one meta * provider may be registered with the DTrace framework at a time. The * sole meta provider type provides user-land static tracing facilities * by taking meta probe descriptions and adding a corresponding provider * into the DTrace framework. * * 1 Framework-to-Provider * * 1.1 Overview * * The Framework-to-Provider API is represented by the dtrace_mops structure * that the meta provider passes to the framework when registering itself as * a meta provider. This structure consists of the following members: * * dtms_create_probe() <-- Add a new probe to a created provider * dtms_provide_pid() <-- Create a new provider for a given process * dtms_remove_pid() <-- Remove a previously created provider * * 1.2 void dtms_create_probe(void *arg, void *parg, * dtrace_helper_probedesc_t *probedesc); * * 1.2.1 Overview * * Called by the DTrace framework to create a new probe in a provider * created by this meta provider. * * 1.2.2 Arguments and notes * * The first argument is the cookie as passed to dtrace_meta_register(). * The second argument is the provider cookie for the associated provider; * this is obtained from the return value of dtms_provide_pid(). The third * argument is the helper probe description. * * 1.2.3 Return value * * None * * 1.2.4 Caller's context * * dtms_create_probe() is called from either ioctl() or module load context. * The DTrace framework is locked in such a way that meta providers may not * register or unregister. This means that the meta provider cannot call * dtrace_meta_register() or dtrace_meta_unregister(). However, the context is * such that the provider may (and is expected to) call provider-related * DTrace provider APIs including dtrace_probe_create(). * * 1.3 void *dtms_provide_pid(void *arg, dtrace_meta_provider_t *mprov, * pid_t pid) * * 1.3.1 Overview * * Called by the DTrace framework to instantiate a new provider given the * description of the provider and probes in the mprov argument. The * meta provider should call dtrace_register() to insert the new provider * into the DTrace framework. * * 1.3.2 Arguments and notes * * The first argument is the cookie as passed to dtrace_meta_register(). * The second argument is a pointer to a structure describing the new * helper provider. The third argument is the process identifier for * process associated with this new provider. Note that the name of the * provider as passed to dtrace_register() should be the contatenation of * the dtmpb_provname member of the mprov argument and the processs * identifier as a string. * * 1.3.3 Return value * * The cookie for the provider that the meta provider creates. This is * the same value that it passed to dtrace_register(). * * 1.3.4 Caller's context * * dtms_provide_pid() is called from either ioctl() or module load context. * The DTrace framework is locked in such a way that meta providers may not * register or unregister. This means that the meta provider cannot call * dtrace_meta_register() or dtrace_meta_unregister(). However, the context * is such that the provider may -- and is expected to -- call * provider-related DTrace provider APIs including dtrace_register(). * * 1.4 void dtms_remove_pid(void *arg, dtrace_meta_provider_t *mprov, * pid_t pid) * * 1.4.1 Overview * * Called by the DTrace framework to remove a provider that had previously * been instantiated via the dtms_provide_pid() entry point. The meta * provider need not remove the provider immediately, but this entry * point indicates that the provider should be removed as soon as possible * using the dtrace_unregister() API. * * 1.4.2 Arguments and notes * * The first argument is the cookie as passed to dtrace_meta_register(). * The second argument is a pointer to a structure describing the helper * provider. The third argument is the process identifier for process * associated with this new provider. * * 1.4.3 Return value * * None * * 1.4.4 Caller's context * * dtms_remove_pid() is called from either ioctl() or exit() context. * The DTrace framework is locked in such a way that meta providers may not * register or unregister. This means that the meta provider cannot call * dtrace_meta_register() or dtrace_meta_unregister(). However, the context * is such that the provider may -- and is expected to -- call * provider-related DTrace provider APIs including dtrace_unregister(). */ typedef struct dtrace_helper_probedesc { char *dthpb_mod; /* probe module */ char *dthpb_func; /* probe function */ char *dthpb_name; /* probe name */ uint64_t dthpb_base; /* base address */ uint32_t *dthpb_offs; /* offsets array */ uint32_t *dthpb_enoffs; /* is-enabled offsets array */ uint32_t dthpb_noffs; /* offsets count */ uint32_t dthpb_nenoffs; /* is-enabled offsets count */ uint8_t *dthpb_args; /* argument mapping array */ uint8_t dthpb_xargc; /* translated argument count */ uint8_t dthpb_nargc; /* native argument count */ char *dthpb_xtypes; /* translated types strings */ char *dthpb_ntypes; /* native types strings */ } dtrace_helper_probedesc_t; typedef struct dtrace_helper_provdesc { char *dthpv_provname; /* provider name */ dtrace_pattr_t dthpv_pattr; /* stability attributes */ } dtrace_helper_provdesc_t; typedef struct dtrace_mops { void (*dtms_create_probe)(void *, void *, dtrace_helper_probedesc_t *); void *(*dtms_provide_pid)(void *, dtrace_helper_provdesc_t *, pid_t); void (*dtms_remove_pid)(void *, dtrace_helper_provdesc_t *, pid_t); } dtrace_mops_t; typedef uintptr_t dtrace_meta_provider_id_t; extern int dtrace_meta_register(const char *, const dtrace_mops_t *, void *, dtrace_meta_provider_id_t *); extern int dtrace_meta_unregister(dtrace_meta_provider_id_t); /* * DTrace Kernel Hooks * * The following functions are implemented by the base kernel and form a set of * hooks used by the DTrace framework. DTrace hooks are implemented in either * uts/common/os/dtrace_subr.c, an ISA-specific assembly file, or in a * uts//os/dtrace_subr.c corresponding to each hardware platform. */ typedef enum dtrace_vtime_state { DTRACE_VTIME_INACTIVE = 0, /* No DTrace, no TNF */ DTRACE_VTIME_ACTIVE, /* DTrace virtual time, no TNF */ DTRACE_VTIME_INACTIVE_TNF, /* No DTrace, TNF active */ DTRACE_VTIME_ACTIVE_TNF /* DTrace virtual time _and_ TNF */ } dtrace_vtime_state_t; #ifdef illumos extern dtrace_vtime_state_t dtrace_vtime_active; #endif extern void dtrace_vtime_switch(kthread_t *next); extern void dtrace_vtime_enable_tnf(void); extern void dtrace_vtime_disable_tnf(void); extern void dtrace_vtime_enable(void); extern void dtrace_vtime_disable(void); struct regs; struct reg; #ifdef illumos extern int (*dtrace_pid_probe_ptr)(struct reg *); extern int (*dtrace_return_probe_ptr)(struct reg *); extern void (*dtrace_fasttrap_fork_ptr)(proc_t *, proc_t *); extern void (*dtrace_fasttrap_exec_ptr)(proc_t *); extern void (*dtrace_fasttrap_exit_ptr)(proc_t *); extern void dtrace_fasttrap_fork(proc_t *, proc_t *); #endif typedef uintptr_t dtrace_icookie_t; typedef void (*dtrace_xcall_t)(void *); extern dtrace_icookie_t dtrace_interrupt_disable(void); extern void dtrace_interrupt_enable(dtrace_icookie_t); extern void dtrace_membar_producer(void); extern void dtrace_membar_consumer(void); extern void (*dtrace_cpu_init)(processorid_t); #ifdef illumos extern void (*dtrace_modload)(modctl_t *); extern void (*dtrace_modunload)(modctl_t *); #endif extern void (*dtrace_helpers_cleanup)(void); extern void (*dtrace_helpers_fork)(proc_t *parent, proc_t *child); extern void (*dtrace_cpustart_init)(void); extern void (*dtrace_cpustart_fini)(void); extern void (*dtrace_closef)(void); extern void (*dtrace_debugger_init)(void); extern void (*dtrace_debugger_fini)(void); extern dtrace_cacheid_t dtrace_predcache_id; #ifdef illumos extern hrtime_t dtrace_gethrtime(void); #else void dtrace_debug_printf(const char *, ...) __printflike(1, 2); #endif extern void dtrace_sync(void); extern void dtrace_toxic_ranges(void (*)(uintptr_t, uintptr_t)); extern void dtrace_xcall(processorid_t, dtrace_xcall_t, void *); extern void dtrace_vpanic(const char *, __va_list); extern void dtrace_panic(const char *, ...); extern int dtrace_safe_defer_signal(void); extern void dtrace_safe_synchronous_signal(void); extern int dtrace_mach_aframes(void); #if defined(__i386) || defined(__amd64) extern int dtrace_instr_size(uchar_t *instr); extern int dtrace_instr_size_isa(uchar_t *, model_t, int *); extern void dtrace_invop_callsite(void); #endif extern void dtrace_invop_add(int (*)(uintptr_t, uintptr_t *, uintptr_t)); extern void dtrace_invop_remove(int (*)(uintptr_t, uintptr_t *, uintptr_t)); #ifdef __sparc extern int dtrace_blksuword32(uintptr_t, uint32_t *, int); extern void dtrace_getfsr(uint64_t *); #endif #ifndef illumos extern void dtrace_helpers_duplicate(proc_t *, proc_t *); extern void dtrace_helpers_destroy(proc_t *); #endif #define DTRACE_CPUFLAG_ISSET(flag) \ (cpu_core[curcpu].cpuc_dtrace_flags & (flag)) #define DTRACE_CPUFLAG_SET(flag) \ (cpu_core[curcpu].cpuc_dtrace_flags |= (flag)) #define DTRACE_CPUFLAG_CLEAR(flag) \ (cpu_core[curcpu].cpuc_dtrace_flags &= ~(flag)) #endif /* _KERNEL */ #endif /* _ASM */ #if defined(__i386) || defined(__amd64) #define DTRACE_INVOP_PUSHL_EBP 1 #define DTRACE_INVOP_POPL_EBP 2 #define DTRACE_INVOP_LEAVE 3 #define DTRACE_INVOP_NOP 4 #define DTRACE_INVOP_RET 5 #elif defined(__powerpc__) #define DTRACE_INVOP_RET 1 #define DTRACE_INVOP_BCTR 2 #define DTRACE_INVOP_BLR 3 #define DTRACE_INVOP_JUMP 4 #define DTRACE_INVOP_MFLR_R0 5 #define DTRACE_INVOP_NOP 6 #elif defined(__arm__) +#define DTRACE_INVOP_SHIFT 4 +#define DTRACE_INVOP_MASK ((1 << DTRACE_INVOP_SHIFT) - 1) +#define DTRACE_INVOP_DATA(x) ((x) >> DTRACE_INVOP_SHIFT) + #define DTRACE_INVOP_PUSHM 1 #define DTRACE_INVOP_POPM 2 #define DTRACE_INVOP_B 3 #endif #ifdef __cplusplus } #endif #endif /* _SYS_DTRACE_H */ Index: projects/cxl_iscsi/sys/cddl/contrib/opensolaris =================================================================== --- projects/cxl_iscsi/sys/cddl/contrib/opensolaris (revision 279716) +++ projects/cxl_iscsi/sys/cddl/contrib/opensolaris (revision 279717) Property changes on: projects/cxl_iscsi/sys/cddl/contrib/opensolaris ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/cddl/contrib/opensolaris:r278538-279716 Index: projects/cxl_iscsi/sys/cddl/dev/dtrace/arm/dtrace_asm.S =================================================================== --- projects/cxl_iscsi/sys/cddl/dev/dtrace/arm/dtrace_asm.S (revision 279716) +++ projects/cxl_iscsi/sys/cddl/dev/dtrace/arm/dtrace_asm.S (revision 279717) @@ -1,219 +1,217 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License, Version 1.0 only * (the "License"). You may not use this file except in compliance * with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END * * $FreeBSD$ */ /* * Copyright 2004 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #define _ASM #define _LOCORE #define LOCORE #include #include #include #include #include "assym.s" /* void dtrace_membar_producer(void) */ ENTRY(dtrace_membar_producer) RET END(dtrace_membar_producer) /* void dtrace_membar_consumer(void) */ ENTRY(dtrace_membar_consumer) RET END(dtrace_membar_consumer) /* dtrace_icookie_t dtrace_interrupt_disable(void) */ ENTRY(dtrace_interrupt_disable) mrs r0, cpsr mov r1, r0 orr r1, r1, #(PSR_I | PSR_F) msr cpsr_c, r1 RET END(dtrace_interrupt_disable) /* void dtrace_interrupt_enable(dtrace_icookie_t cookie) */ ENTRY(dtrace_interrupt_enable) and r0, r0, #(PSR_I | PSR_F) mrs r1, cpsr bic r1, r1, #(PSR_I | PSR_F) orr r1, r1, r0 msr cpsr_c, r1 RET END(dtrace_interrupt_enable) /* uint8_t dtrace_fuword8_nocheck(void *addr) */ ENTRY(dtrace_fuword8_nocheck) ldrb r3, [r0] mov r0, r3 RET END(dtrace_fuword8_nocheck) /* uint16_t dtrace_fuword16_nocheck(void *addr) */ ENTRY(dtrace_fuword16_nocheck) ldrh r3, [r0] mov r0, r3 RET END(dtrace_fuword16_nocheck) /* uint32_t dtrace_fuword32_nocheck(void *addr) */ ENTRY(dtrace_fuword32_nocheck) ldr r3, [r0] mov r0, r3 RET END(dtrace_fuword32_nocheck) /* uint64_t dtrace_fuword64_nocheck(void *addr) */ ENTRY(dtrace_fuword64_nocheck) ldm r0, {r2, r3} mov r0, r2 mov r1, r3 #if defined(__BIG_ENDIAN__) /* big endian */ mov r0, r3 mov r1, r2 #else /* little endian */ mov r0, r2 mov r1, r3 #endif RET END(dtrace_fuword64_nocheck) /* void dtrace_copy(uintptr_t uaddr, uintptr_t kaddr, size_t size) */ ENTRY(dtrace_copy) stmfd sp!, {r4-r5} /* stack is 8 byte aligned */ teq r2, #0x00000000 mov r5, #0x00000000 beq 2f 1: ldrb r4, [r0], #0x0001 add r5, r5, #0x00000001 strb r4, [r1], #0x0001 teqne r5, r2 bne 1b 2: ldmfd sp!, {r4-r5} /* stack is 8 byte aligned */ RET END(dtrace_copy) /* void dtrace_copystr(uintptr_t uaddr, uintptr_t kaddr, size_t size, volatile uint16_t *flags) XXX: Check for flags? */ ENTRY(dtrace_copystr) stmfd sp!, {r4-r5} /* stack is 8 byte aligned */ teq r2, #0x00000000 mov r5, #0x00000000 beq 2f 1: ldrb r4, [r0], #0x0001 add r5, r5, #0x00000001 teq r4, #0x00000000 strb r4, [r1], #0x0001 teqne r5, r2 bne 1b 2: ldmfd sp!, {r4-r5} /* stack is 8 byte aligned */ RET END(dtrace_copystr) /* void vpanic(const char *format, va_list alist) */ ENTRY(vpanic) /* Initial stack layout: */ vpanic_common: RET END(vpanic) /* void dtrace_vpanic(const char *format, va_list alist) */ ENTRY(dtrace_vpanic) /* Initial stack layout: */ b vpanic RET END(dtrace_vpanic) /* Initial stack layout: */ /* uintptr_t dtrace_caller(int aframes) */ ENTRY(dtrace_caller) mov r0, #-1 RET END(dtrace_caller) /* uint32_t dtrace_cas32(uint32_t *target, uint32_t cmp, uint32_t new) void * dtrace_casptr(volatile void *target, volatile void *cmp, volatile void *new) */ ENTRY(dtrace_cas32) EENTRY(dtrace_casptr) 1: ldrex r3, [r0] /* Load target */ cmp r3, r1 /* Check if *target == cmp */ bne 2f /* No, return */ - strex r3, r2, [r0] /* Store new to target */ - cmp r3, #0 /* Did the store succeed? */ + strex ip, r2, [r0] /* Store new to target */ + cmp ip, #0 /* Did the store succeed? */ bne 1b /* No, try again */ - mov r0, r2 /* Return the new value of the store */ -2: movne r0, r3 /* The first compare failed, return */ - /* the value loaded from memory */ +2: mov r0, r3 /* Return the value loaded from target */ RET EEND(dtrace_casptr) END(dtrace_cas32) Index: projects/cxl_iscsi/sys/cddl/dev/dtrace/arm/dtrace_subr.c =================================================================== --- projects/cxl_iscsi/sys/cddl/dev/dtrace/arm/dtrace_subr.c (revision 279716) +++ projects/cxl_iscsi/sys/cddl/dev/dtrace/arm/dtrace_subr.c (revision 279717) @@ -1,261 +1,346 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License, Version 1.0 only * (the "License"). You may not use this file except in compliance * with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END * * $FreeBSD$ * */ /* * Copyright 2005 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define DELAYBRANCH(x) ((int)(x) < 0) - + +#define BIT_PC 15 +#define BIT_LR 14 +#define BIT_SP 13 + extern uintptr_t dtrace_in_probe_addr; extern int dtrace_in_probe; extern dtrace_id_t dtrace_probeid_error; extern int (*dtrace_invop_jump_addr)(struct trapframe *); int dtrace_invop(uintptr_t, uintptr_t *, uintptr_t); void dtrace_invop_init(void); void dtrace_invop_uninit(void); typedef struct dtrace_invop_hdlr { int (*dtih_func)(uintptr_t, uintptr_t *, uintptr_t); struct dtrace_invop_hdlr *dtih_next; } dtrace_invop_hdlr_t; dtrace_invop_hdlr_t *dtrace_invop_hdlr; int dtrace_invop(uintptr_t addr, uintptr_t *stack, uintptr_t eax) { dtrace_invop_hdlr_t *hdlr; int rval; for (hdlr = dtrace_invop_hdlr; hdlr != NULL; hdlr = hdlr->dtih_next) if ((rval = hdlr->dtih_func(addr, stack, eax)) != 0) return (rval); return (0); } void dtrace_invop_add(int (*func)(uintptr_t, uintptr_t *, uintptr_t)) { dtrace_invop_hdlr_t *hdlr; hdlr = kmem_alloc(sizeof (dtrace_invop_hdlr_t), KM_SLEEP); hdlr->dtih_func = func; hdlr->dtih_next = dtrace_invop_hdlr; dtrace_invop_hdlr = hdlr; } void dtrace_invop_remove(int (*func)(uintptr_t, uintptr_t *, uintptr_t)) { dtrace_invop_hdlr_t *hdlr = dtrace_invop_hdlr, *prev = NULL; for (;;) { if (hdlr == NULL) panic("attempt to remove non-existent invop handler"); if (hdlr->dtih_func == func) break; prev = hdlr; hdlr = hdlr->dtih_next; } if (prev == NULL) { ASSERT(dtrace_invop_hdlr == hdlr); dtrace_invop_hdlr = hdlr->dtih_next; } else { ASSERT(dtrace_invop_hdlr != hdlr); prev->dtih_next = hdlr->dtih_next; } kmem_free(hdlr, 0); } /*ARGSUSED*/ void dtrace_toxic_ranges(void (*func)(uintptr_t base, uintptr_t limit)) { printf("IMPLEMENT ME: dtrace_toxic_ranges\n"); } void dtrace_xcall(processorid_t cpu, dtrace_xcall_t func, void *arg) { cpuset_t cpus; if (cpu == DTRACE_CPUALL) cpus = all_cpus; else CPU_SETOF(cpu, &cpus); smp_rendezvous_cpus(cpus, smp_no_rendevous_barrier, func, smp_no_rendevous_barrier, arg); } static void dtrace_sync_func(void) { } void dtrace_sync(void) { dtrace_xcall(DTRACE_CPUALL, (dtrace_xcall_t)dtrace_sync_func, NULL); } /* * DTrace needs a high resolution time function which can * be called from a probe context and guaranteed not to have * instrumented with probes itself. * * Returns nanoseconds since boot. */ uint64_t dtrace_gethrtime() { struct timespec curtime; nanouptime(&curtime); return (curtime.tv_sec * 1000000000UL + curtime.tv_nsec); } uint64_t dtrace_gethrestime(void) { struct timespec curtime; getnanotime(&curtime); return (curtime.tv_sec * 1000000000UL + curtime.tv_nsec); } /* Function to handle DTrace traps during probes. See amd64/amd64/trap.c */ int dtrace_trap(struct trapframe *frame, u_int type) { /* * A trap can occur while DTrace executes a probe. Before * executing the probe, DTrace blocks re-scheduling and sets * a flag in it's per-cpu flags to indicate that it doesn't * want to fault. On returning from the probe, the no-fault * flag is cleared and finally re-scheduling is enabled. * * Check if DTrace has enabled 'no-fault' mode: * */ if ((cpu_core[curcpu].cpuc_dtrace_flags & CPU_DTRACE_NOFAULT) != 0) { /* * There are only a couple of trap types that are expected. * All the rest will be handled in the usual way. */ switch (type) { /* Page fault. */ case FAULT_ALIGN: /* Flag a bad address. */ cpu_core[curcpu].cpuc_dtrace_flags |= CPU_DTRACE_BADADDR; cpu_core[curcpu].cpuc_dtrace_illval = 0; /* * Offset the instruction pointer to the instruction * following the one causing the fault. */ frame->tf_pc += sizeof(int); return (1); default: /* Handle all other traps in the usual way. */ break; } } /* Handle the trap in the usual way. */ return (0); } void dtrace_probe_error(dtrace_state_t *state, dtrace_epid_t epid, int which, int fault, int fltoffs, uintptr_t illval) { dtrace_probe(dtrace_probeid_error, (uint64_t)(uintptr_t)state, (uintptr_t)epid, (uintptr_t)which, (uintptr_t)fault, (uintptr_t)fltoffs); } static int dtrace_invop_start(struct trapframe *frame) { - printf("IMPLEMENT ME: %s\n", __func__); - switch (dtrace_invop(frame->tf_pc, (uintptr_t *)frame, frame->tf_pc)) { + register_t *r0, *sp; + int data, invop, reg, update_sp; + + invop = dtrace_invop(frame->tf_pc, (uintptr_t *)frame, frame->tf_pc); + switch (invop & DTRACE_INVOP_MASK) { case DTRACE_INVOP_PUSHM: - // TODO: + sp = (register_t *)frame->tf_svc_sp; + r0 = &frame->tf_r0; + data = DTRACE_INVOP_DATA(invop); + + /* + * Store the pc, lr, and sp. These have their own + * entries in the struct. + */ + if (data & (1 << BIT_PC)) { + sp--; + *sp = frame->tf_pc; + } + if (data & (1 << BIT_LR)) { + sp--; + *sp = frame->tf_svc_lr; + } + if (data & (1 << BIT_SP)) { + sp--; + *sp = frame->tf_svc_sp; + } + + /* Store the general registers */ + for (reg = 12; reg >= 0; reg--) { + if (data & (1 << reg)) { + sp--; + *sp = r0[reg]; + } + } + + /* Update the stack pointer and program counter to continue */ + frame->tf_svc_sp = (register_t)sp; + frame->tf_pc += 4; break; case DTRACE_INVOP_POPM: - // TODO: + sp = (register_t *)frame->tf_svc_sp; + r0 = &frame->tf_r0; + data = DTRACE_INVOP_DATA(invop); + + /* Read the general registers */ + for (reg = 0; reg <= 12; reg++) { + if (data & (1 << reg)) { + r0[reg] = *sp; + sp++; + } + } + + /* + * Set the stack pointer. If we don't update it here we will + * need to update it at the end as the instruction would do + */ + update_sp = 1; + if (data & (1 << BIT_SP)) { + frame->tf_svc_sp = *sp; + *sp++; + update_sp = 0; + } + + /* Update the link register, we need to use the correct copy */ + if (data & (1 << BIT_LR)) { + frame->tf_svc_lr = *sp; + *sp++; + } + /* + * And the program counter. If it's not in the list skip over + * it when we return so to not hit this again. + */ + if (data & (1 << BIT_PC)) { + frame->tf_pc = *sp; + *sp++; + } else + frame->tf_pc += 4; + + /* Update the stack pointer if we haven't already done so */ + if (update_sp) + frame->tf_svc_sp = (register_t)sp; break; case DTRACE_INVOP_B: - // TODO + data = DTRACE_INVOP_DATA(invop) & 0x00ffffff; + /* Sign extend the data */ + if ((data & (1 << 23)) != 0) + data |= 0xff000000; + /* The data is the number of 4-byte words to change the pc */ + data *= 4; + data += 8; + frame->tf_pc += data; break; default: return (-1); break; } return (0); } void dtrace_invop_init(void) { dtrace_invop_jump_addr = dtrace_invop_start; } void dtrace_invop_uninit(void) { dtrace_invop_jump_addr = 0; } Index: projects/cxl_iscsi/sys/cddl/dev/dtrace/powerpc/dtrace_subr.c =================================================================== --- projects/cxl_iscsi/sys/cddl/dev/dtrace/powerpc/dtrace_subr.c (revision 279716) +++ projects/cxl_iscsi/sys/cddl/dev/dtrace/powerpc/dtrace_subr.c (revision 279717) @@ -1,360 +1,360 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License, Version 1.0 only * (the "License"). You may not use this file except in compliance * with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END * * $FreeBSD$ * */ /* * Copyright 2005 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #define DELAYBRANCH(x) ((int)(x) < 0) extern uintptr_t dtrace_in_probe_addr; extern int dtrace_in_probe; extern dtrace_id_t dtrace_probeid_error; extern int (*dtrace_invop_jump_addr)(struct trapframe *); extern void dtrace_getnanotime(struct timespec *tsp); int dtrace_invop(uintptr_t, uintptr_t *, uintptr_t); void dtrace_invop_init(void); void dtrace_invop_uninit(void); typedef struct dtrace_invop_hdlr { int (*dtih_func)(uintptr_t, uintptr_t *, uintptr_t); struct dtrace_invop_hdlr *dtih_next; } dtrace_invop_hdlr_t; dtrace_invop_hdlr_t *dtrace_invop_hdlr; int dtrace_invop(uintptr_t addr, uintptr_t *stack, uintptr_t arg0) { dtrace_invop_hdlr_t *hdlr; int rval; for (hdlr = dtrace_invop_hdlr; hdlr != NULL; hdlr = hdlr->dtih_next) if ((rval = hdlr->dtih_func(addr, stack, arg0)) != 0) return (rval); return (0); } void dtrace_invop_add(int (*func)(uintptr_t, uintptr_t *, uintptr_t)) { dtrace_invop_hdlr_t *hdlr; hdlr = kmem_alloc(sizeof (dtrace_invop_hdlr_t), KM_SLEEP); hdlr->dtih_func = func; hdlr->dtih_next = dtrace_invop_hdlr; dtrace_invop_hdlr = hdlr; } void dtrace_invop_remove(int (*func)(uintptr_t, uintptr_t *, uintptr_t)) { dtrace_invop_hdlr_t *hdlr = dtrace_invop_hdlr, *prev = NULL; for (;;) { if (hdlr == NULL) panic("attempt to remove non-existent invop handler"); if (hdlr->dtih_func == func) break; prev = hdlr; hdlr = hdlr->dtih_next; } if (prev == NULL) { ASSERT(dtrace_invop_hdlr == hdlr); dtrace_invop_hdlr = hdlr->dtih_next; } else { ASSERT(dtrace_invop_hdlr != hdlr); prev->dtih_next = hdlr->dtih_next; } kmem_free(hdlr, 0); } /*ARGSUSED*/ void dtrace_toxic_ranges(void (*func)(uintptr_t base, uintptr_t limit)) { /* * No toxic regions? */ } void dtrace_xcall(processorid_t cpu, dtrace_xcall_t func, void *arg) { cpuset_t cpus; if (cpu == DTRACE_CPUALL) cpus = all_cpus; else CPU_SETOF(cpu, &cpus); smp_rendezvous_cpus(cpus, smp_no_rendevous_barrier, func, smp_no_rendevous_barrier, arg); } static void dtrace_sync_func(void) { } void dtrace_sync(void) { dtrace_xcall(DTRACE_CPUALL, (dtrace_xcall_t)dtrace_sync_func, NULL); } static int64_t tgt_cpu_tsc; static int64_t hst_cpu_tsc; static int64_t timebase_skew[MAXCPU]; static uint64_t nsec_scale; /* See below for the explanation of this macro. */ /* This is taken from the amd64 dtrace_subr, to provide a synchronized timer * between multiple processors in dtrace. Since PowerPC Timebases can be much * lower than x86, the scale shift is 26 instead of 28, allowing for a 15.63MHz * timebase. */ #define SCALE_SHIFT 26 static void dtrace_gethrtime_init_cpu(void *arg) { uintptr_t cpu = (uintptr_t) arg; if (cpu == curcpu) tgt_cpu_tsc = mftb(); else hst_cpu_tsc = mftb(); } static void dtrace_gethrtime_init(void *arg) { struct pcpu *pc; uint64_t tb_f; cpuset_t map; int i; tb_f = cpu_tickrate(); /* * The following line checks that nsec_scale calculated below * doesn't overflow 32-bit unsigned integer, so that it can multiply * another 32-bit integer without overflowing 64-bit. * Thus minimum supported Timebase frequency is 15.63MHz. */ KASSERT(tb_f > (NANOSEC >> (32 - SCALE_SHIFT)), ("Timebase frequency is too low")); /* * We scale up NANOSEC/tb_f ratio to preserve as much precision * as possible. * 2^26 factor was chosen quite arbitrarily from practical * considerations: * - it supports TSC frequencies as low as 15.63MHz (see above); */ nsec_scale = ((uint64_t)NANOSEC << SCALE_SHIFT) / tb_f; /* The current CPU is the reference one. */ sched_pin(); timebase_skew[curcpu] = 0; CPU_FOREACH(i) { if (i == curcpu) continue; pc = pcpu_find(i); CPU_SETOF(PCPU_GET(cpuid), &map); CPU_SET(pc->pc_cpuid, &map); smp_rendezvous_cpus(map, NULL, dtrace_gethrtime_init_cpu, smp_no_rendevous_barrier, (void *)(uintptr_t) i); timebase_skew[i] = tgt_cpu_tsc - hst_cpu_tsc; } sched_unpin(); } SYSINIT(dtrace_gethrtime_init, SI_SUB_SMP, SI_ORDER_ANY, dtrace_gethrtime_init, NULL); /* * DTrace needs a high resolution time function which can * be called from a probe context and guaranteed not to have * instrumented with probes itself. * * Returns nanoseconds since boot. */ uint64_t dtrace_gethrtime() { uint64_t timebase; uint32_t lo; uint32_t hi; /* * We split timebase value into lower and higher 32-bit halves and separately * scale them with nsec_scale, then we scale them down by 2^28 * (see nsec_scale calculations) taking into account 32-bit shift of * the higher half and finally add. */ timebase = mftb() - timebase_skew[curcpu]; lo = timebase; hi = timebase >> 32; return (((lo * nsec_scale) >> SCALE_SHIFT) + ((hi * nsec_scale) << (32 - SCALE_SHIFT))); } uint64_t dtrace_gethrestime(void) { struct timespec curtime; dtrace_getnanotime(&curtime); return (curtime.tv_sec * 1000000000UL + curtime.tv_nsec); } /* Function to handle DTrace traps during probes. See powerpc/powerpc/trap.c */ int dtrace_trap(struct trapframe *frame, u_int type) { /* * A trap can occur while DTrace executes a probe. Before * executing the probe, DTrace blocks re-scheduling and sets * a flag in its per-cpu flags to indicate that it doesn't * want to fault. On returning from the probe, the no-fault * flag is cleared and finally re-scheduling is enabled. * * Check if DTrace has enabled 'no-fault' mode: */ if ((cpu_core[curcpu].cpuc_dtrace_flags & CPU_DTRACE_NOFAULT) != 0) { /* * There are only a couple of trap types that are expected. * All the rest will be handled in the usual way. */ switch (type) { /* Page fault. */ case EXC_DSI: case EXC_DSE: /* Flag a bad address. */ cpu_core[curcpu].cpuc_dtrace_flags |= CPU_DTRACE_BADADDR; - cpu_core[curcpu].cpuc_dtrace_illval = frame->cpu.aim.dar; + cpu_core[curcpu].cpuc_dtrace_illval = frame->dar; /* * Offset the instruction pointer to the instruction * following the one causing the fault. */ frame->srr0 += sizeof(int); return (1); case EXC_ISI: case EXC_ISE: /* Flag a bad address. */ cpu_core[curcpu].cpuc_dtrace_flags |= CPU_DTRACE_BADADDR; cpu_core[curcpu].cpuc_dtrace_illval = frame->srr0; /* * Offset the instruction pointer to the instruction * following the one causing the fault. */ frame->srr0 += sizeof(int); return (1); default: /* Handle all other traps in the usual way. */ break; } } /* Handle the trap in the usual way. */ return (0); } void dtrace_probe_error(dtrace_state_t *state, dtrace_epid_t epid, int which, int fault, int fltoffs, uintptr_t illval) { dtrace_probe(dtrace_probeid_error, (uint64_t)(uintptr_t)state, (uintptr_t)epid, (uintptr_t)which, (uintptr_t)fault, (uintptr_t)fltoffs); } static int dtrace_invop_start(struct trapframe *frame) { switch (dtrace_invop(frame->srr0, (uintptr_t *)frame, frame->fixreg[3])) { case DTRACE_INVOP_JUMP: break; case DTRACE_INVOP_BCTR: frame->srr0 = frame->ctr; break; case DTRACE_INVOP_BLR: frame->srr0 = frame->lr; break; case DTRACE_INVOP_MFLR_R0: frame->fixreg[0] = frame->lr; frame->srr0 = frame->srr0 + 4; break; default: return (-1); break; } return (0); } void dtrace_invop_init(void) { dtrace_invop_jump_addr = dtrace_invop_start; } void dtrace_invop_uninit(void) { dtrace_invop_jump_addr = 0; } Index: projects/cxl_iscsi/sys/cddl/dev/fbt/arm/fbt_isa.c =================================================================== --- projects/cxl_iscsi/sys/cddl/dev/fbt/arm/fbt_isa.c (revision 279716) +++ projects/cxl_iscsi/sys/cddl/dev/fbt/arm/fbt_isa.c (revision 279717) @@ -1,192 +1,199 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END * * Portions Copyright 2006-2008 John Birrell jb@freebsd.org * Portions Copyright 2013 Justin Hibbits jhibbits@freebsd.org * Portions Copyright 2013 Howard Su howardsu@freebsd.org * * $FreeBSD$ * */ /* * Copyright 2006 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #include #include #include #include "fbt.h" -#define FBT_PATCHVAL 0xe06a0cfe /* illegal instruction */ +#define FBT_PATCHVAL 0xe7f000f0 /* Specified undefined instruction */ #define FBT_PUSHM 0xe92d0000 #define FBT_POPM 0xe8bd0000 #define FBT_JUMP 0xea000000 #define FBT_ENTRY "entry" #define FBT_RETURN "return" int fbt_invop(uintptr_t addr, uintptr_t *stack, uintptr_t rval) { struct trapframe *frame = (struct trapframe *)stack; solaris_cpu_t *cpu = &solaris_cpu[curcpu]; fbt_probe_t *fbt = fbt_probetab[FBT_ADDR2NDX(addr)]; for (; fbt != NULL; fbt = fbt->fbtp_hashnext) { if ((uintptr_t)fbt->fbtp_patchpoint == addr) { fbt->fbtp_invop_cnt++; cpu->cpu_dtrace_caller = addr; /* TODO: Need 5th parameter from stack */ dtrace_probe(fbt->fbtp_id, frame->tf_r0, frame->tf_r1, frame->tf_r2, frame->tf_r3, 0); cpu->cpu_dtrace_caller = 0; - return (fbt->fbtp_rval); + return (fbt->fbtp_rval | (fbt->fbtp_savedval << DTRACE_INVOP_SHIFT)); } } return (0); } void fbt_patch_tracepoint(fbt_probe_t *fbt, fbt_patchval_t val) { *fbt->fbtp_patchpoint = val; cpu_icache_sync_range((vm_offset_t)fbt->fbtp_patchpoint, 4); } int fbt_provide_module_function(linker_file_t lf, int symindx, linker_symval_t *symval, void *opaque) { char *modname = opaque; const char *name = symval->name; fbt_probe_t *fbt, *retfbt; uint32_t *instr, *limit; int popm; if (strncmp(name, "dtrace_", 7) == 0 && strncmp(name, "dtrace_safe_", 12) != 0) { /* * Anything beginning with "dtrace_" may be called * from probe context unless it explicitly indicates * that it won't be called from probe context by * using the prefix "dtrace_safe_". */ return (0); } if (name[0] == '_' && name[1] == '_') + return (0); + + /* + * Architecture-specific exclusion list, largely to do with FBT trap + * processing, to prevent reentrance. + */ + if (strcmp(name, "undefinedinstruction") == 0) return (0); instr = (uint32_t *)symval->value; limit = (uint32_t *)(symval->value + symval->size); for (; instr < limit; instr++) if ((*instr & 0xffff0000) == FBT_PUSHM && (*instr & 0x4000) != 0) break; if (instr >= limit) return (0); fbt = malloc(sizeof (fbt_probe_t), M_FBT, M_WAITOK | M_ZERO); fbt->fbtp_name = name; fbt->fbtp_id = dtrace_probe_create(fbt_id, modname, name, FBT_ENTRY, 3, fbt); fbt->fbtp_patchpoint = instr; fbt->fbtp_ctl = lf; fbt->fbtp_loadcnt = lf->loadcnt; fbt->fbtp_savedval = *instr; fbt->fbtp_patchval = FBT_PATCHVAL; fbt->fbtp_rval = DTRACE_INVOP_PUSHM; fbt->fbtp_symindx = symindx; fbt->fbtp_hashnext = fbt_probetab[FBT_ADDR2NDX(instr)]; fbt_probetab[FBT_ADDR2NDX(instr)] = fbt; lf->fbt_nentries++; popm = FBT_POPM | ((*instr) & 0x3FFF) | 0x8000; retfbt = NULL; again: for (; instr < limit; instr++) { if (*instr == popm) break; else if ((*instr & 0xff000000) == FBT_JUMP) { uint32_t *target, *start; int offset; offset = (*instr & 0xffffff); offset <<= 8; offset /= 64; target = instr + (2 + offset); start = (uint32_t *)symval->value; if (target >= limit || target < start) break; instr++; /* skip delay slot */ } } if (instr >= limit) return (0); /* * We have a winner! */ fbt = malloc(sizeof (fbt_probe_t), M_FBT, M_WAITOK | M_ZERO); fbt->fbtp_name = name; if (retfbt == NULL) { fbt->fbtp_id = dtrace_probe_create(fbt_id, modname, name, FBT_RETURN, 5, fbt); } else { retfbt->fbtp_next = fbt; fbt->fbtp_id = retfbt->fbtp_id; } retfbt = fbt; fbt->fbtp_patchpoint = instr; fbt->fbtp_ctl = lf; fbt->fbtp_loadcnt = lf->loadcnt; fbt->fbtp_symindx = symindx; if ((*instr & 0xff000000) == FBT_JUMP) fbt->fbtp_rval = DTRACE_INVOP_B; else fbt->fbtp_rval = DTRACE_INVOP_POPM; fbt->fbtp_savedval = *instr; fbt->fbtp_patchval = FBT_PATCHVAL; fbt->fbtp_hashnext = fbt_probetab[FBT_ADDR2NDX(instr)]; fbt_probetab[FBT_ADDR2NDX(instr)] = fbt; lf->fbt_nentries++; instr++; goto again; } Index: projects/cxl_iscsi/sys/dev/acpica/acpi.c =================================================================== --- projects/cxl_iscsi/sys/dev/acpica/acpi.c (revision 279716) +++ projects/cxl_iscsi/sys/dev/acpica/acpi.c (revision 279717) @@ -1,4027 +1,4027 @@ /*- * Copyright (c) 2000 Takanori Watanabe * Copyright (c) 2000 Mitsuru IWASAKI * Copyright (c) 2000, 2001 Michael Smith * Copyright (c) 2000 BSDi * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include "opt_acpi.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #if defined(__i386__) || defined(__amd64__) #include #endif #include #include #include #include #include #include #include #include #include #include #include static MALLOC_DEFINE(M_ACPIDEV, "acpidev", "ACPI devices"); /* Hooks for the ACPI CA debugging infrastructure */ #define _COMPONENT ACPI_BUS ACPI_MODULE_NAME("ACPI") static d_open_t acpiopen; static d_close_t acpiclose; static d_ioctl_t acpiioctl; static struct cdevsw acpi_cdevsw = { .d_version = D_VERSION, .d_open = acpiopen, .d_close = acpiclose, .d_ioctl = acpiioctl, .d_name = "acpi", }; struct acpi_interface { ACPI_STRING *data; int num; }; /* Global mutex for locking access to the ACPI subsystem. */ struct mtx acpi_mutex; struct callout acpi_sleep_timer; /* Bitmap of device quirks. */ int acpi_quirks; /* Supported sleep states. */ static BOOLEAN acpi_sleep_states[ACPI_S_STATE_COUNT]; static void acpi_lookup(void *arg, const char *name, device_t *dev); static int acpi_modevent(struct module *mod, int event, void *junk); static int acpi_probe(device_t dev); static int acpi_attach(device_t dev); static int acpi_suspend(device_t dev); static int acpi_resume(device_t dev); static int acpi_shutdown(device_t dev); static device_t acpi_add_child(device_t bus, u_int order, const char *name, int unit); static int acpi_print_child(device_t bus, device_t child); static void acpi_probe_nomatch(device_t bus, device_t child); static void acpi_driver_added(device_t dev, driver_t *driver); static int acpi_read_ivar(device_t dev, device_t child, int index, uintptr_t *result); static int acpi_write_ivar(device_t dev, device_t child, int index, uintptr_t value); static struct resource_list *acpi_get_rlist(device_t dev, device_t child); static void acpi_reserve_resources(device_t dev); static int acpi_sysres_alloc(device_t dev); static int acpi_set_resource(device_t dev, device_t child, int type, int rid, u_long start, u_long count); static struct resource *acpi_alloc_resource(device_t bus, device_t child, int type, int *rid, u_long start, u_long end, u_long count, u_int flags); static int acpi_adjust_resource(device_t bus, device_t child, int type, struct resource *r, u_long start, u_long end); static int acpi_release_resource(device_t bus, device_t child, int type, int rid, struct resource *r); static void acpi_delete_resource(device_t bus, device_t child, int type, int rid); static uint32_t acpi_isa_get_logicalid(device_t dev); static int acpi_isa_get_compatid(device_t dev, uint32_t *cids, int count); static char *acpi_device_id_probe(device_t bus, device_t dev, char **ids); static ACPI_STATUS acpi_device_eval_obj(device_t bus, device_t dev, ACPI_STRING pathname, ACPI_OBJECT_LIST *parameters, ACPI_BUFFER *ret); static ACPI_STATUS acpi_device_scan_cb(ACPI_HANDLE h, UINT32 level, void *context, void **retval); static ACPI_STATUS acpi_device_scan_children(device_t bus, device_t dev, int max_depth, acpi_scan_cb_t user_fn, void *arg); static int acpi_set_powerstate(device_t child, int state); static int acpi_isa_pnp_probe(device_t bus, device_t child, struct isa_pnp_id *ids); static void acpi_probe_children(device_t bus); static void acpi_probe_order(ACPI_HANDLE handle, int *order); static ACPI_STATUS acpi_probe_child(ACPI_HANDLE handle, UINT32 level, void *context, void **status); static void acpi_sleep_enable(void *arg); static ACPI_STATUS acpi_sleep_disable(struct acpi_softc *sc); static ACPI_STATUS acpi_EnterSleepState(struct acpi_softc *sc, int state); static void acpi_shutdown_final(void *arg, int howto); static void acpi_enable_fixed_events(struct acpi_softc *sc); static BOOLEAN acpi_has_hid(ACPI_HANDLE handle); static void acpi_resync_clock(struct acpi_softc *sc); static int acpi_wake_sleep_prep(ACPI_HANDLE handle, int sstate); static int acpi_wake_run_prep(ACPI_HANDLE handle, int sstate); static int acpi_wake_prep_walk(int sstate); static int acpi_wake_sysctl_walk(device_t dev); static int acpi_wake_set_sysctl(SYSCTL_HANDLER_ARGS); static void acpi_system_eventhandler_sleep(void *arg, int state); static void acpi_system_eventhandler_wakeup(void *arg, int state); static int acpi_sname2sstate(const char *sname); static const char *acpi_sstate2sname(int sstate); static int acpi_supported_sleep_state_sysctl(SYSCTL_HANDLER_ARGS); static int acpi_sleep_state_sysctl(SYSCTL_HANDLER_ARGS); static int acpi_debug_objects_sysctl(SYSCTL_HANDLER_ARGS); static int acpi_pm_func(u_long cmd, void *arg, ...); static int acpi_child_location_str_method(device_t acdev, device_t child, char *buf, size_t buflen); static int acpi_child_pnpinfo_str_method(device_t acdev, device_t child, char *buf, size_t buflen); #if defined(__i386__) || defined(__amd64__) static void acpi_enable_pcie(void); #endif static void acpi_hint_device_unit(device_t acdev, device_t child, const char *name, int *unitp); static void acpi_reset_interfaces(device_t dev); static device_method_t acpi_methods[] = { /* Device interface */ DEVMETHOD(device_probe, acpi_probe), DEVMETHOD(device_attach, acpi_attach), DEVMETHOD(device_shutdown, acpi_shutdown), DEVMETHOD(device_detach, bus_generic_detach), DEVMETHOD(device_suspend, acpi_suspend), DEVMETHOD(device_resume, acpi_resume), /* Bus interface */ DEVMETHOD(bus_add_child, acpi_add_child), DEVMETHOD(bus_print_child, acpi_print_child), DEVMETHOD(bus_probe_nomatch, acpi_probe_nomatch), DEVMETHOD(bus_driver_added, acpi_driver_added), DEVMETHOD(bus_read_ivar, acpi_read_ivar), DEVMETHOD(bus_write_ivar, acpi_write_ivar), DEVMETHOD(bus_get_resource_list, acpi_get_rlist), DEVMETHOD(bus_set_resource, acpi_set_resource), DEVMETHOD(bus_get_resource, bus_generic_rl_get_resource), DEVMETHOD(bus_alloc_resource, acpi_alloc_resource), DEVMETHOD(bus_adjust_resource, acpi_adjust_resource), DEVMETHOD(bus_release_resource, acpi_release_resource), DEVMETHOD(bus_delete_resource, acpi_delete_resource), DEVMETHOD(bus_child_pnpinfo_str, acpi_child_pnpinfo_str_method), DEVMETHOD(bus_child_location_str, acpi_child_location_str_method), DEVMETHOD(bus_activate_resource, bus_generic_activate_resource), DEVMETHOD(bus_deactivate_resource, bus_generic_deactivate_resource), DEVMETHOD(bus_setup_intr, bus_generic_setup_intr), DEVMETHOD(bus_teardown_intr, bus_generic_teardown_intr), DEVMETHOD(bus_hint_device_unit, acpi_hint_device_unit), DEVMETHOD(bus_get_domain, acpi_get_domain), /* ACPI bus */ DEVMETHOD(acpi_id_probe, acpi_device_id_probe), DEVMETHOD(acpi_evaluate_object, acpi_device_eval_obj), DEVMETHOD(acpi_pwr_for_sleep, acpi_device_pwr_for_sleep), DEVMETHOD(acpi_scan_children, acpi_device_scan_children), /* ISA emulation */ DEVMETHOD(isa_pnp_probe, acpi_isa_pnp_probe), DEVMETHOD_END }; static driver_t acpi_driver = { "acpi", acpi_methods, sizeof(struct acpi_softc), }; static devclass_t acpi_devclass; DRIVER_MODULE(acpi, nexus, acpi_driver, acpi_devclass, acpi_modevent, 0); MODULE_VERSION(acpi, 1); ACPI_SERIAL_DECL(acpi, "ACPI root bus"); /* Local pools for managing system resources for ACPI child devices. */ static struct rman acpi_rman_io, acpi_rman_mem; #define ACPI_MINIMUM_AWAKETIME 5 /* Holds the description of the acpi0 device. */ static char acpi_desc[ACPI_OEM_ID_SIZE + ACPI_OEM_TABLE_ID_SIZE + 2]; SYSCTL_NODE(_debug, OID_AUTO, acpi, CTLFLAG_RD, NULL, "ACPI debugging"); static char acpi_ca_version[12]; SYSCTL_STRING(_debug_acpi, OID_AUTO, acpi_ca_version, CTLFLAG_RD, acpi_ca_version, 0, "Version of Intel ACPI-CA"); /* * Allow overriding _OSI methods. */ static char acpi_install_interface[256]; TUNABLE_STR("hw.acpi.install_interface", acpi_install_interface, sizeof(acpi_install_interface)); static char acpi_remove_interface[256]; TUNABLE_STR("hw.acpi.remove_interface", acpi_remove_interface, sizeof(acpi_remove_interface)); /* Allow users to dump Debug objects without ACPI debugger. */ static int acpi_debug_objects; TUNABLE_INT("debug.acpi.enable_debug_objects", &acpi_debug_objects); SYSCTL_PROC(_debug_acpi, OID_AUTO, enable_debug_objects, CTLFLAG_RW | CTLTYPE_INT, NULL, 0, acpi_debug_objects_sysctl, "I", "Enable Debug objects"); /* Allow the interpreter to ignore common mistakes in BIOS. */ static int acpi_interpreter_slack = 1; TUNABLE_INT("debug.acpi.interpreter_slack", &acpi_interpreter_slack); SYSCTL_INT(_debug_acpi, OID_AUTO, interpreter_slack, CTLFLAG_RDTUN, &acpi_interpreter_slack, 1, "Turn on interpreter slack mode."); /* Ignore register widths set by FADT and use default widths instead. */ static int acpi_ignore_reg_width = 1; TUNABLE_INT("debug.acpi.default_register_width", &acpi_ignore_reg_width); SYSCTL_INT(_debug_acpi, OID_AUTO, default_register_width, CTLFLAG_RDTUN, &acpi_ignore_reg_width, 1, "Ignore register widths set by FADT"); #ifdef __amd64__ /* Reset system clock while resuming. XXX Remove once tested. */ static int acpi_reset_clock = 1; TUNABLE_INT("debug.acpi.reset_clock", &acpi_reset_clock); SYSCTL_INT(_debug_acpi, OID_AUTO, reset_clock, CTLFLAG_RW, &acpi_reset_clock, 1, "Reset system clock while resuming."); #endif /* Allow users to override quirks. */ TUNABLE_INT("debug.acpi.quirks", &acpi_quirks); static int acpi_susp_bounce; SYSCTL_INT(_debug_acpi, OID_AUTO, suspend_bounce, CTLFLAG_RW, &acpi_susp_bounce, 0, "Don't actually suspend, just test devices."); /* * ACPI can only be loaded as a module by the loader; activating it after * system bootstrap time is not useful, and can be fatal to the system. * It also cannot be unloaded, since the entire system bus hierarchy hangs * off it. */ static int acpi_modevent(struct module *mod, int event, void *junk) { switch (event) { case MOD_LOAD: if (!cold) { printf("The ACPI driver cannot be loaded after boot.\n"); return (EPERM); } break; case MOD_UNLOAD: if (!cold && power_pm_get_type() == POWER_PM_TYPE_ACPI) return (EBUSY); break; default: break; } return (0); } /* * Perform early initialization. */ ACPI_STATUS acpi_Startup(void) { static int started = 0; ACPI_STATUS status; int val; ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__); /* Only run the startup code once. The MADT driver also calls this. */ if (started) return_VALUE (AE_OK); started = 1; /* * Pre-allocate space for RSDT/XSDT and DSDT tables and allow resizing * if more tables exist. */ if (ACPI_FAILURE(status = AcpiInitializeTables(NULL, 2, TRUE))) { printf("ACPI: Table initialisation failed: %s\n", AcpiFormatException(status)); return_VALUE (status); } /* Set up any quirks we have for this system. */ if (acpi_quirks == ACPI_Q_OK) acpi_table_quirks(&acpi_quirks); /* If the user manually set the disabled hint to 0, force-enable ACPI. */ if (resource_int_value("acpi", 0, "disabled", &val) == 0 && val == 0) acpi_quirks &= ~ACPI_Q_BROKEN; if (acpi_quirks & ACPI_Q_BROKEN) { printf("ACPI disabled by blacklist. Contact your BIOS vendor.\n"); status = AE_SUPPORT; } return_VALUE (status); } /* * Detect ACPI and perform early initialisation. */ int acpi_identify(void) { ACPI_TABLE_RSDP *rsdp; ACPI_TABLE_HEADER *rsdt; ACPI_PHYSICAL_ADDRESS paddr; struct sbuf sb; ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__); if (!cold) return (ENXIO); /* Check that we haven't been disabled with a hint. */ if (resource_disabled("acpi", 0)) return (ENXIO); /* Check for other PM systems. */ if (power_pm_get_type() != POWER_PM_TYPE_NONE && power_pm_get_type() != POWER_PM_TYPE_ACPI) { printf("ACPI identify failed, other PM system enabled.\n"); return (ENXIO); } /* Initialize root tables. */ if (ACPI_FAILURE(acpi_Startup())) { printf("ACPI: Try disabling either ACPI or apic support.\n"); return (ENXIO); } if ((paddr = AcpiOsGetRootPointer()) == 0 || (rsdp = AcpiOsMapMemory(paddr, sizeof(ACPI_TABLE_RSDP))) == NULL) return (ENXIO); if (rsdp->Revision > 1 && rsdp->XsdtPhysicalAddress != 0) paddr = (ACPI_PHYSICAL_ADDRESS)rsdp->XsdtPhysicalAddress; else paddr = (ACPI_PHYSICAL_ADDRESS)rsdp->RsdtPhysicalAddress; AcpiOsUnmapMemory(rsdp, sizeof(ACPI_TABLE_RSDP)); if ((rsdt = AcpiOsMapMemory(paddr, sizeof(ACPI_TABLE_HEADER))) == NULL) return (ENXIO); sbuf_new(&sb, acpi_desc, sizeof(acpi_desc), SBUF_FIXEDLEN); sbuf_bcat(&sb, rsdt->OemId, ACPI_OEM_ID_SIZE); sbuf_trim(&sb); sbuf_putc(&sb, ' '); sbuf_bcat(&sb, rsdt->OemTableId, ACPI_OEM_TABLE_ID_SIZE); sbuf_trim(&sb); sbuf_finish(&sb); sbuf_delete(&sb); AcpiOsUnmapMemory(rsdt, sizeof(ACPI_TABLE_HEADER)); snprintf(acpi_ca_version, sizeof(acpi_ca_version), "%x", ACPI_CA_VERSION); return (0); } /* * Fetch some descriptive data from ACPI to put in our attach message. */ static int acpi_probe(device_t dev) { ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__); device_set_desc(dev, acpi_desc); return_VALUE (BUS_PROBE_NOWILDCARD); } static int acpi_attach(device_t dev) { struct acpi_softc *sc; ACPI_STATUS status; int error, state; UINT32 flags; UINT8 TypeA, TypeB; char *env; ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__); sc = device_get_softc(dev); sc->acpi_dev = dev; callout_init(&sc->susp_force_to, TRUE); error = ENXIO; /* Initialize resource manager. */ acpi_rman_io.rm_type = RMAN_ARRAY; acpi_rman_io.rm_start = 0; acpi_rman_io.rm_end = 0xffff; acpi_rman_io.rm_descr = "ACPI I/O ports"; if (rman_init(&acpi_rman_io) != 0) panic("acpi rman_init IO ports failed"); acpi_rman_mem.rm_type = RMAN_ARRAY; acpi_rman_mem.rm_start = 0; acpi_rman_mem.rm_end = ~0ul; acpi_rman_mem.rm_descr = "ACPI I/O memory addresses"; if (rman_init(&acpi_rman_mem) != 0) panic("acpi rman_init memory failed"); /* Initialise the ACPI mutex */ mtx_init(&acpi_mutex, "ACPI global lock", NULL, MTX_DEF); /* * Set the globals from our tunables. This is needed because ACPI-CA * uses UINT8 for some values and we have no tunable_byte. */ AcpiGbl_EnableInterpreterSlack = acpi_interpreter_slack ? TRUE : FALSE; AcpiGbl_EnableAmlDebugObject = acpi_debug_objects ? TRUE : FALSE; AcpiGbl_UseDefaultRegisterWidths = acpi_ignore_reg_width ? TRUE : FALSE; #ifndef ACPI_DEBUG /* * Disable all debugging layers and levels. */ AcpiDbgLayer = 0; AcpiDbgLevel = 0; #endif /* Start up the ACPI CA subsystem. */ status = AcpiInitializeSubsystem(); if (ACPI_FAILURE(status)) { device_printf(dev, "Could not initialize Subsystem: %s\n", AcpiFormatException(status)); goto out; } /* Override OS interfaces if the user requested. */ acpi_reset_interfaces(dev); /* Load ACPI name space. */ status = AcpiLoadTables(); if (ACPI_FAILURE(status)) { device_printf(dev, "Could not load Namespace: %s\n", AcpiFormatException(status)); goto out; } #if defined(__i386__) || defined(__amd64__) /* Handle MCFG table if present. */ acpi_enable_pcie(); #endif /* * Note that some systems (specifically, those with namespace evaluation * issues that require the avoidance of parts of the namespace) must * avoid running _INI and _STA on everything, as well as dodging the final * object init pass. * * For these devices, we set ACPI_NO_DEVICE_INIT and ACPI_NO_OBJECT_INIT). * * XXX We should arrange for the object init pass after we have attached * all our child devices, but on many systems it works here. */ flags = 0; if (testenv("debug.acpi.avoid")) flags = ACPI_NO_DEVICE_INIT | ACPI_NO_OBJECT_INIT; /* Bring the hardware and basic handlers online. */ if (ACPI_FAILURE(status = AcpiEnableSubsystem(flags))) { device_printf(dev, "Could not enable ACPI: %s\n", AcpiFormatException(status)); goto out; } /* * Call the ECDT probe function to provide EC functionality before * the namespace has been evaluated. * * XXX This happens before the sysresource devices have been probed and * attached so its resources come from nexus0. In practice, this isn't * a problem but should be addressed eventually. */ acpi_ec_ecdt_probe(dev); /* Bring device objects and regions online. */ if (ACPI_FAILURE(status = AcpiInitializeObjects(flags))) { device_printf(dev, "Could not initialize ACPI objects: %s\n", AcpiFormatException(status)); goto out; } /* * Setup our sysctl tree. * * XXX: This doesn't check to make sure that none of these fail. */ sysctl_ctx_init(&sc->acpi_sysctl_ctx); sc->acpi_sysctl_tree = SYSCTL_ADD_NODE(&sc->acpi_sysctl_ctx, SYSCTL_STATIC_CHILDREN(_hw), OID_AUTO, device_get_name(dev), CTLFLAG_RD, 0, ""); SYSCTL_ADD_PROC(&sc->acpi_sysctl_ctx, SYSCTL_CHILDREN(sc->acpi_sysctl_tree), OID_AUTO, "supported_sleep_state", CTLTYPE_STRING | CTLFLAG_RD, 0, 0, acpi_supported_sleep_state_sysctl, "A", ""); SYSCTL_ADD_PROC(&sc->acpi_sysctl_ctx, SYSCTL_CHILDREN(sc->acpi_sysctl_tree), OID_AUTO, "power_button_state", CTLTYPE_STRING | CTLFLAG_RW, &sc->acpi_power_button_sx, 0, acpi_sleep_state_sysctl, "A", ""); SYSCTL_ADD_PROC(&sc->acpi_sysctl_ctx, SYSCTL_CHILDREN(sc->acpi_sysctl_tree), OID_AUTO, "sleep_button_state", CTLTYPE_STRING | CTLFLAG_RW, &sc->acpi_sleep_button_sx, 0, acpi_sleep_state_sysctl, "A", ""); SYSCTL_ADD_PROC(&sc->acpi_sysctl_ctx, SYSCTL_CHILDREN(sc->acpi_sysctl_tree), OID_AUTO, "lid_switch_state", CTLTYPE_STRING | CTLFLAG_RW, &sc->acpi_lid_switch_sx, 0, acpi_sleep_state_sysctl, "A", ""); SYSCTL_ADD_PROC(&sc->acpi_sysctl_ctx, SYSCTL_CHILDREN(sc->acpi_sysctl_tree), OID_AUTO, "standby_state", CTLTYPE_STRING | CTLFLAG_RW, &sc->acpi_standby_sx, 0, acpi_sleep_state_sysctl, "A", ""); SYSCTL_ADD_PROC(&sc->acpi_sysctl_ctx, SYSCTL_CHILDREN(sc->acpi_sysctl_tree), OID_AUTO, "suspend_state", CTLTYPE_STRING | CTLFLAG_RW, &sc->acpi_suspend_sx, 0, acpi_sleep_state_sysctl, "A", ""); SYSCTL_ADD_INT(&sc->acpi_sysctl_ctx, SYSCTL_CHILDREN(sc->acpi_sysctl_tree), OID_AUTO, "sleep_delay", CTLFLAG_RW, &sc->acpi_sleep_delay, 0, "sleep delay in seconds"); SYSCTL_ADD_INT(&sc->acpi_sysctl_ctx, SYSCTL_CHILDREN(sc->acpi_sysctl_tree), OID_AUTO, "s4bios", CTLFLAG_RW, &sc->acpi_s4bios, 0, "S4BIOS mode"); SYSCTL_ADD_INT(&sc->acpi_sysctl_ctx, SYSCTL_CHILDREN(sc->acpi_sysctl_tree), OID_AUTO, "verbose", CTLFLAG_RW, &sc->acpi_verbose, 0, "verbose mode"); SYSCTL_ADD_INT(&sc->acpi_sysctl_ctx, SYSCTL_CHILDREN(sc->acpi_sysctl_tree), OID_AUTO, "disable_on_reboot", CTLFLAG_RW, &sc->acpi_do_disable, 0, "Disable ACPI when rebooting/halting system"); SYSCTL_ADD_INT(&sc->acpi_sysctl_ctx, SYSCTL_CHILDREN(sc->acpi_sysctl_tree), OID_AUTO, "handle_reboot", CTLFLAG_RW, &sc->acpi_handle_reboot, 0, "Use ACPI Reset Register to reboot"); /* * Default to 1 second before sleeping to give some machines time to * stabilize. */ sc->acpi_sleep_delay = 1; if (bootverbose) sc->acpi_verbose = 1; if ((env = kern_getenv("hw.acpi.verbose")) != NULL) { if (strcmp(env, "0") != 0) sc->acpi_verbose = 1; freeenv(env); } /* Only enable reboot by default if the FADT says it is available. */ if (AcpiGbl_FADT.Flags & ACPI_FADT_RESET_REGISTER) sc->acpi_handle_reboot = 1; /* Only enable S4BIOS by default if the FACS says it is available. */ if (AcpiGbl_FACS->Flags & ACPI_FACS_S4_BIOS_PRESENT) sc->acpi_s4bios = 1; /* Probe all supported sleep states. */ acpi_sleep_states[ACPI_STATE_S0] = TRUE; for (state = ACPI_STATE_S1; state < ACPI_S_STATE_COUNT; state++) if (ACPI_SUCCESS(AcpiEvaluateObject(ACPI_ROOT_OBJECT, __DECONST(char *, AcpiGbl_SleepStateNames[state]), NULL, NULL)) && ACPI_SUCCESS(AcpiGetSleepTypeData(state, &TypeA, &TypeB))) acpi_sleep_states[state] = TRUE; /* * Dispatch the default sleep state to devices. The lid switch is set * to UNKNOWN by default to avoid surprising users. */ sc->acpi_power_button_sx = acpi_sleep_states[ACPI_STATE_S5] ? ACPI_STATE_S5 : ACPI_STATE_UNKNOWN; sc->acpi_lid_switch_sx = ACPI_STATE_UNKNOWN; sc->acpi_standby_sx = acpi_sleep_states[ACPI_STATE_S1] ? ACPI_STATE_S1 : ACPI_STATE_UNKNOWN; sc->acpi_suspend_sx = acpi_sleep_states[ACPI_STATE_S3] ? ACPI_STATE_S3 : ACPI_STATE_UNKNOWN; /* Pick the first valid sleep state for the sleep button default. */ sc->acpi_sleep_button_sx = ACPI_STATE_UNKNOWN; for (state = ACPI_STATE_S1; state <= ACPI_STATE_S4; state++) if (acpi_sleep_states[state]) { sc->acpi_sleep_button_sx = state; break; } acpi_enable_fixed_events(sc); /* * Scan the namespace and attach/initialise children. */ /* Register our shutdown handler. */ EVENTHANDLER_REGISTER(shutdown_final, acpi_shutdown_final, sc, SHUTDOWN_PRI_LAST); /* * Register our acpi event handlers. * XXX should be configurable eg. via userland policy manager. */ EVENTHANDLER_REGISTER(acpi_sleep_event, acpi_system_eventhandler_sleep, sc, ACPI_EVENT_PRI_LAST); EVENTHANDLER_REGISTER(acpi_wakeup_event, acpi_system_eventhandler_wakeup, sc, ACPI_EVENT_PRI_LAST); /* Flag our initial states. */ sc->acpi_enabled = TRUE; sc->acpi_sstate = ACPI_STATE_S0; sc->acpi_sleep_disabled = TRUE; /* Create the control device */ sc->acpi_dev_t = make_dev(&acpi_cdevsw, 0, UID_ROOT, GID_WHEEL, 0644, "acpi"); sc->acpi_dev_t->si_drv1 = sc; if ((error = acpi_machdep_init(dev))) goto out; /* Register ACPI again to pass the correct argument of pm_func. */ power_pm_register(POWER_PM_TYPE_ACPI, acpi_pm_func, sc); if (!acpi_disabled("bus")) { EVENTHANDLER_REGISTER(dev_lookup, acpi_lookup, NULL, 1000); acpi_probe_children(dev); } /* Update all GPEs and enable runtime GPEs. */ status = AcpiUpdateAllGpes(); if (ACPI_FAILURE(status)) device_printf(dev, "Could not update all GPEs: %s\n", AcpiFormatException(status)); /* Allow sleep request after a while. */ callout_init_mtx(&acpi_sleep_timer, &acpi_mutex, 0); callout_reset(&acpi_sleep_timer, hz * ACPI_MINIMUM_AWAKETIME, acpi_sleep_enable, sc); error = 0; out: return_VALUE (error); } static void acpi_set_power_children(device_t dev, int state) { device_t child; device_t *devlist; int dstate, i, numdevs; if (device_get_children(dev, &devlist, &numdevs) != 0) return; /* * Retrieve and set D-state for the sleep state if _SxD is present. * Skip children who aren't attached since they are handled separately. */ for (i = 0; i < numdevs; i++) { child = devlist[i]; dstate = state; if (device_is_attached(child) && acpi_device_pwr_for_sleep(dev, child, &dstate) == 0) acpi_set_powerstate(child, dstate); } free(devlist, M_TEMP); } static int acpi_suspend(device_t dev) { int error; GIANT_REQUIRED; error = bus_generic_suspend(dev); if (error == 0) acpi_set_power_children(dev, ACPI_STATE_D3); return (error); } static int acpi_resume(device_t dev) { GIANT_REQUIRED; acpi_set_power_children(dev, ACPI_STATE_D0); return (bus_generic_resume(dev)); } static int acpi_shutdown(device_t dev) { GIANT_REQUIRED; /* Allow children to shutdown first. */ bus_generic_shutdown(dev); /* * Enable any GPEs that are able to power-on the system (i.e., RTC). * Also, disable any that are not valid for this state (most). */ acpi_wake_prep_walk(ACPI_STATE_S5); return (0); } /* * Handle a new device being added */ static device_t acpi_add_child(device_t bus, u_int order, const char *name, int unit) { struct acpi_device *ad; device_t child; if ((ad = malloc(sizeof(*ad), M_ACPIDEV, M_NOWAIT | M_ZERO)) == NULL) return (NULL); resource_list_init(&ad->ad_rl); child = device_add_child_ordered(bus, order, name, unit); if (child != NULL) device_set_ivars(child, ad); else free(ad, M_ACPIDEV); return (child); } static int acpi_print_child(device_t bus, device_t child) { struct acpi_device *adev = device_get_ivars(child); struct resource_list *rl = &adev->ad_rl; int retval = 0; retval += bus_print_child_header(bus, child); retval += resource_list_print_type(rl, "port", SYS_RES_IOPORT, "%#lx"); retval += resource_list_print_type(rl, "iomem", SYS_RES_MEMORY, "%#lx"); retval += resource_list_print_type(rl, "irq", SYS_RES_IRQ, "%ld"); retval += resource_list_print_type(rl, "drq", SYS_RES_DRQ, "%ld"); if (device_get_flags(child)) retval += printf(" flags %#x", device_get_flags(child)); retval += bus_print_child_domain(bus, child); retval += bus_print_child_footer(bus, child); return (retval); } /* * If this device is an ACPI child but no one claimed it, attempt * to power it off. We'll power it back up when a driver is added. * * XXX Disabled for now since many necessary devices (like fdc and * ATA) don't claim the devices we created for them but still expect * them to be powered up. */ static void acpi_probe_nomatch(device_t bus, device_t child) { #ifdef ACPI_ENABLE_POWERDOWN_NODRIVER acpi_set_powerstate(child, ACPI_STATE_D3); #endif } /* * If a new driver has a chance to probe a child, first power it up. * * XXX Disabled for now (see acpi_probe_nomatch for details). */ static void acpi_driver_added(device_t dev, driver_t *driver) { device_t child, *devlist; int i, numdevs; DEVICE_IDENTIFY(driver, dev); if (device_get_children(dev, &devlist, &numdevs)) return; for (i = 0; i < numdevs; i++) { child = devlist[i]; if (device_get_state(child) == DS_NOTPRESENT) { #ifdef ACPI_ENABLE_POWERDOWN_NODRIVER acpi_set_powerstate(child, ACPI_STATE_D0); if (device_probe_and_attach(child) != 0) acpi_set_powerstate(child, ACPI_STATE_D3); #else device_probe_and_attach(child); #endif } } free(devlist, M_TEMP); } /* Location hint for devctl(8) */ static int acpi_child_location_str_method(device_t cbdev, device_t child, char *buf, size_t buflen) { struct acpi_device *dinfo = device_get_ivars(child); char buf2[32]; int pxm; if (dinfo->ad_handle) { snprintf(buf, buflen, "handle=%s", acpi_name(dinfo->ad_handle)); if (ACPI_SUCCESS(acpi_GetInteger(dinfo->ad_handle, "_PXM", &pxm))) { snprintf(buf2, 32, " _PXM=%d", pxm); strlcat(buf, buf2, buflen); } } else { snprintf(buf, buflen, "unknown"); } return (0); } /* PnP information for devctl(8) */ static int acpi_child_pnpinfo_str_method(device_t cbdev, device_t child, char *buf, size_t buflen) { struct acpi_device *dinfo = device_get_ivars(child); ACPI_DEVICE_INFO *adinfo; if (ACPI_FAILURE(AcpiGetObjectInfo(dinfo->ad_handle, &adinfo))) { snprintf(buf, buflen, "unknown"); return (0); } snprintf(buf, buflen, "_HID=%s _UID=%lu", (adinfo->Valid & ACPI_VALID_HID) ? adinfo->HardwareId.String : "none", (adinfo->Valid & ACPI_VALID_UID) ? strtoul(adinfo->UniqueId.String, NULL, 10) : 0UL); AcpiOsFree(adinfo); return (0); } /* * Handle per-device ivars */ static int acpi_read_ivar(device_t dev, device_t child, int index, uintptr_t *result) { struct acpi_device *ad; if ((ad = device_get_ivars(child)) == NULL) { device_printf(child, "device has no ivars\n"); return (ENOENT); } /* ACPI and ISA compatibility ivars */ switch(index) { case ACPI_IVAR_HANDLE: *(ACPI_HANDLE *)result = ad->ad_handle; break; case ACPI_IVAR_PRIVATE: *(void **)result = ad->ad_private; break; case ACPI_IVAR_FLAGS: *(int *)result = ad->ad_flags; break; case ISA_IVAR_VENDORID: case ISA_IVAR_SERIAL: case ISA_IVAR_COMPATID: *(int *)result = -1; break; case ISA_IVAR_LOGICALID: *(int *)result = acpi_isa_get_logicalid(child); break; default: return (ENOENT); } return (0); } static int acpi_write_ivar(device_t dev, device_t child, int index, uintptr_t value) { struct acpi_device *ad; if ((ad = device_get_ivars(child)) == NULL) { device_printf(child, "device has no ivars\n"); return (ENOENT); } switch(index) { case ACPI_IVAR_HANDLE: ad->ad_handle = (ACPI_HANDLE)value; break; case ACPI_IVAR_PRIVATE: ad->ad_private = (void *)value; break; case ACPI_IVAR_FLAGS: ad->ad_flags = (int)value; break; default: panic("bad ivar write request (%d)", index); return (ENOENT); } return (0); } /* * Handle child resource allocation/removal */ static struct resource_list * acpi_get_rlist(device_t dev, device_t child) { struct acpi_device *ad; ad = device_get_ivars(child); return (&ad->ad_rl); } static int acpi_match_resource_hint(device_t dev, int type, long value) { struct acpi_device *ad = device_get_ivars(dev); struct resource_list *rl = &ad->ad_rl; struct resource_list_entry *rle; STAILQ_FOREACH(rle, rl, link) { if (rle->type != type) continue; if (rle->start <= value && rle->end >= value) return (1); } return (0); } /* * Wire device unit numbers based on resource matches in hints. */ static void acpi_hint_device_unit(device_t acdev, device_t child, const char *name, int *unitp) { const char *s; long value; int line, matches, unit; /* * Iterate over all the hints for the devices with the specified * name to see if one's resources are a subset of this device. */ line = 0; for (;;) { if (resource_find_dev(&line, name, &unit, "at", NULL) != 0) break; /* Must have an "at" for acpi or isa. */ resource_string_value(name, unit, "at", &s); if (!(strcmp(s, "acpi0") == 0 || strcmp(s, "acpi") == 0 || strcmp(s, "isa0") == 0 || strcmp(s, "isa") == 0)) continue; /* * Check for matching resources. We must have at least one match. * Since I/O and memory resources cannot be shared, if we get a * match on either of those, ignore any mismatches in IRQs or DRQs. * * XXX: We may want to revisit this to be more lenient and wire * as long as it gets one match. */ matches = 0; if (resource_long_value(name, unit, "port", &value) == 0) { /* * Floppy drive controllers are notorious for having a * wide variety of resources not all of which include the * first port that is specified by the hint (typically * 0x3f0) (see the comment above fdc_isa_alloc_resources() * in fdc_isa.c). However, they do all seem to include * port + 2 (e.g. 0x3f2) so for a floppy device, look for * 'value + 2' in the port resources instead of the hint * value. */ if (strcmp(name, "fdc") == 0) value += 2; if (acpi_match_resource_hint(child, SYS_RES_IOPORT, value)) matches++; else continue; } if (resource_long_value(name, unit, "maddr", &value) == 0) { if (acpi_match_resource_hint(child, SYS_RES_MEMORY, value)) matches++; else continue; } if (matches > 0) goto matched; if (resource_long_value(name, unit, "irq", &value) == 0) { if (acpi_match_resource_hint(child, SYS_RES_IRQ, value)) matches++; else continue; } if (resource_long_value(name, unit, "drq", &value) == 0) { if (acpi_match_resource_hint(child, SYS_RES_DRQ, value)) matches++; else continue; } matched: if (matches > 0) { /* We have a winner! */ *unitp = unit; break; } } } /* - * Fech the NUMA domain for the given device. + * Fetch the NUMA domain for the given device. * * If a device has a _PXM method, map that to a NUMA domain. * * If none is found, then it'll call the parent method. * If there's no domain, return ENOENT. */ int acpi_get_domain(device_t dev, device_t child, int *domain) { #if MAXMEMDOM > 1 ACPI_HANDLE h; int d, pxm; h = acpi_get_handle(child); if ((h != NULL) && ACPI_SUCCESS(acpi_GetInteger(h, "_PXM", &pxm))) { d = acpi_map_pxm_to_vm_domainid(pxm); if (d < 0) return (ENOENT); *domain = d; return (0); } #endif /* No _PXM node; go up a level */ return (bus_generic_get_domain(dev, child, domain)); } /* * Pre-allocate/manage all memory and IO resources. Since rman can't handle * duplicates, we merge any in the sysresource attach routine. */ static int acpi_sysres_alloc(device_t dev) { struct resource *res; struct resource_list *rl; struct resource_list_entry *rle; struct rman *rm; char *sysres_ids[] = { "PNP0C01", "PNP0C02", NULL }; device_t *children; int child_count, i; /* * Probe/attach any sysresource devices. This would be unnecessary if we * had multi-pass probe/attach. */ if (device_get_children(dev, &children, &child_count) != 0) return (ENXIO); for (i = 0; i < child_count; i++) { if (ACPI_ID_PROBE(dev, children[i], sysres_ids) != NULL) device_probe_and_attach(children[i]); } free(children, M_TEMP); rl = BUS_GET_RESOURCE_LIST(device_get_parent(dev), dev); STAILQ_FOREACH(rle, rl, link) { if (rle->res != NULL) { device_printf(dev, "duplicate resource for %lx\n", rle->start); continue; } /* Only memory and IO resources are valid here. */ switch (rle->type) { case SYS_RES_IOPORT: rm = &acpi_rman_io; break; case SYS_RES_MEMORY: rm = &acpi_rman_mem; break; default: continue; } /* Pre-allocate resource and add to our rman pool. */ res = BUS_ALLOC_RESOURCE(device_get_parent(dev), dev, rle->type, &rle->rid, rle->start, rle->start + rle->count - 1, rle->count, 0); if (res != NULL) { rman_manage_region(rm, rman_get_start(res), rman_get_end(res)); rle->res = res; } else device_printf(dev, "reservation of %lx, %lx (%d) failed\n", rle->start, rle->count, rle->type); } return (0); } static char *pcilink_ids[] = { "PNP0C0F", NULL }; static char *sysres_ids[] = { "PNP0C01", "PNP0C02", NULL }; /* * Reserve declared resources for devices found during attach once system * resources have been allocated. */ static void acpi_reserve_resources(device_t dev) { struct resource_list_entry *rle; struct resource_list *rl; struct acpi_device *ad; struct acpi_softc *sc; device_t *children; int child_count, i; sc = device_get_softc(dev); if (device_get_children(dev, &children, &child_count) != 0) return; for (i = 0; i < child_count; i++) { ad = device_get_ivars(children[i]); rl = &ad->ad_rl; /* Don't reserve system resources. */ if (ACPI_ID_PROBE(dev, children[i], sysres_ids) != NULL) continue; STAILQ_FOREACH(rle, rl, link) { /* * Don't reserve IRQ resources. There are many sticky things * to get right otherwise (e.g. IRQs for psm, atkbd, and HPET * when using legacy routing). */ if (rle->type == SYS_RES_IRQ) continue; /* * Don't reserve the resource if it is already allocated. * The acpi_ec(4) driver can allocate its resources early * if ECDT is present. */ if (rle->res != NULL) continue; /* * Try to reserve the resource from our parent. If this * fails because the resource is a system resource, just * let it be. The resource range is already reserved so * that other devices will not use it. If the driver * needs to allocate the resource, then * acpi_alloc_resource() will sub-alloc from the system * resource. */ resource_list_reserve(rl, dev, children[i], rle->type, &rle->rid, rle->start, rle->end, rle->count, 0); } } free(children, M_TEMP); sc->acpi_resources_reserved = 1; } static int acpi_set_resource(device_t dev, device_t child, int type, int rid, u_long start, u_long count) { struct acpi_softc *sc = device_get_softc(dev); struct acpi_device *ad = device_get_ivars(child); struct resource_list *rl = &ad->ad_rl; ACPI_DEVICE_INFO *devinfo; u_long end; /* Ignore IRQ resources for PCI link devices. */ if (type == SYS_RES_IRQ && ACPI_ID_PROBE(dev, child, pcilink_ids) != NULL) return (0); /* * Ignore most resources for PCI root bridges. Some BIOSes * incorrectly enumerate the memory ranges they decode as plain * memory resources instead of as ResourceProducer ranges. Other * BIOSes incorrectly list system resource entries for I/O ranges * under the PCI bridge. Do allow the one known-correct case on * x86 of a PCI bridge claiming the I/O ports used for PCI config * access. */ if (type == SYS_RES_MEMORY || type == SYS_RES_IOPORT) { if (ACPI_SUCCESS(AcpiGetObjectInfo(ad->ad_handle, &devinfo))) { if ((devinfo->Flags & ACPI_PCI_ROOT_BRIDGE) != 0) { #if defined(__i386__) || defined(__amd64__) if (!(type == SYS_RES_IOPORT && start == CONF1_ADDR_PORT)) #endif { AcpiOsFree(devinfo); return (0); } } AcpiOsFree(devinfo); } } /* If the resource is already allocated, fail. */ if (resource_list_busy(rl, type, rid)) return (EBUSY); /* If the resource is already reserved, release it. */ if (resource_list_reserved(rl, type, rid)) resource_list_unreserve(rl, dev, child, type, rid); /* Add the resource. */ end = (start + count - 1); resource_list_add(rl, type, rid, start, end, count); /* Don't reserve resources until the system resources are allocated. */ if (!sc->acpi_resources_reserved) return (0); /* Don't reserve system resources. */ if (ACPI_ID_PROBE(dev, child, sysres_ids) != NULL) return (0); /* * Don't reserve IRQ resources. There are many sticky things to * get right otherwise (e.g. IRQs for psm, atkbd, and HPET when * using legacy routing). */ if (type == SYS_RES_IRQ) return (0); /* * Reserve the resource. * * XXX: Ignores failure for now. Failure here is probably a * BIOS/firmware bug? */ resource_list_reserve(rl, dev, child, type, &rid, start, end, count, 0); return (0); } static struct resource * acpi_alloc_resource(device_t bus, device_t child, int type, int *rid, u_long start, u_long end, u_long count, u_int flags) { ACPI_RESOURCE ares; struct acpi_device *ad; struct resource_list_entry *rle; struct resource_list *rl; struct resource *res; int isdefault = (start == 0UL && end == ~0UL); /* * First attempt at allocating the resource. For direct children, * use resource_list_alloc() to handle reserved resources. For * other devices, pass the request up to our parent. */ if (bus == device_get_parent(child)) { ad = device_get_ivars(child); rl = &ad->ad_rl; /* * Simulate the behavior of the ISA bus for direct children * devices. That is, if a non-default range is specified for * a resource that doesn't exist, use bus_set_resource() to * add the resource before allocating it. Note that these * resources will not be reserved. */ if (!isdefault && resource_list_find(rl, type, *rid) == NULL) resource_list_add(rl, type, *rid, start, end, count); res = resource_list_alloc(rl, bus, child, type, rid, start, end, count, flags); if (res != NULL && type == SYS_RES_IRQ) { /* * Since bus_config_intr() takes immediate effect, we cannot * configure the interrupt associated with a device when we * parse the resources but have to defer it until a driver * actually allocates the interrupt via bus_alloc_resource(). * * XXX: Should we handle the lookup failing? */ if (ACPI_SUCCESS(acpi_lookup_irq_resource(child, *rid, res, &ares))) acpi_config_intr(child, &ares); } /* * If this is an allocation of the "default" range for a given * RID, fetch the exact bounds for this resource from the * resource list entry to try to allocate the range from the * system resource regions. */ if (res == NULL && isdefault) { rle = resource_list_find(rl, type, *rid); if (rle != NULL) { start = rle->start; end = rle->end; count = rle->count; } } } else res = BUS_ALLOC_RESOURCE(device_get_parent(bus), child, type, rid, start, end, count, flags); /* * If the first attempt failed and this is an allocation of a * specific range, try to satisfy the request via a suballocation * from our system resource regions. */ if (res == NULL && start + count - 1 == end) res = acpi_alloc_sysres(child, type, rid, start, end, count, flags); return (res); } /* * Attempt to allocate a specific resource range from the system * resource ranges. Note that we only handle memory and I/O port * system resources. */ struct resource * acpi_alloc_sysres(device_t child, int type, int *rid, u_long start, u_long end, u_long count, u_int flags) { struct rman *rm; struct resource *res; switch (type) { case SYS_RES_IOPORT: rm = &acpi_rman_io; break; case SYS_RES_MEMORY: rm = &acpi_rman_mem; break; default: return (NULL); } KASSERT(start + count - 1 == end, ("wildcard resource range")); res = rman_reserve_resource(rm, start, end, count, flags & ~RF_ACTIVE, child); if (res == NULL) return (NULL); rman_set_rid(res, *rid); /* If requested, activate the resource using the parent's method. */ if (flags & RF_ACTIVE) if (bus_activate_resource(child, type, *rid, res) != 0) { rman_release_resource(res); return (NULL); } return (res); } static int acpi_is_resource_managed(int type, struct resource *r) { /* We only handle memory and IO resources through rman. */ switch (type) { case SYS_RES_IOPORT: return (rman_is_region_manager(r, &acpi_rman_io)); case SYS_RES_MEMORY: return (rman_is_region_manager(r, &acpi_rman_mem)); } return (0); } static int acpi_adjust_resource(device_t bus, device_t child, int type, struct resource *r, u_long start, u_long end) { if (acpi_is_resource_managed(type, r)) return (rman_adjust_resource(r, start, end)); return (bus_generic_adjust_resource(bus, child, type, r, start, end)); } static int acpi_release_resource(device_t bus, device_t child, int type, int rid, struct resource *r) { int ret; /* * If this resource belongs to one of our internal managers, * deactivate it and release it to the local pool. */ if (acpi_is_resource_managed(type, r)) { if (rman_get_flags(r) & RF_ACTIVE) { ret = bus_deactivate_resource(child, type, rid, r); if (ret != 0) return (ret); } return (rman_release_resource(r)); } return (bus_generic_rl_release_resource(bus, child, type, rid, r)); } static void acpi_delete_resource(device_t bus, device_t child, int type, int rid) { struct resource_list *rl; rl = acpi_get_rlist(bus, child); if (resource_list_busy(rl, type, rid)) { device_printf(bus, "delete_resource: Resource still owned by child" " (type=%d, rid=%d)\n", type, rid); return; } resource_list_unreserve(rl, bus, child, type, rid); resource_list_delete(rl, type, rid); } /* Allocate an IO port or memory resource, given its GAS. */ int acpi_bus_alloc_gas(device_t dev, int *type, int *rid, ACPI_GENERIC_ADDRESS *gas, struct resource **res, u_int flags) { int error, res_type; error = ENOMEM; if (type == NULL || rid == NULL || gas == NULL || res == NULL) return (EINVAL); /* We only support memory and IO spaces. */ switch (gas->SpaceId) { case ACPI_ADR_SPACE_SYSTEM_MEMORY: res_type = SYS_RES_MEMORY; break; case ACPI_ADR_SPACE_SYSTEM_IO: res_type = SYS_RES_IOPORT; break; default: return (EOPNOTSUPP); } /* * If the register width is less than 8, assume the BIOS author means * it is a bit field and just allocate a byte. */ if (gas->BitWidth && gas->BitWidth < 8) gas->BitWidth = 8; /* Validate the address after we're sure we support the space. */ if (gas->Address == 0 || gas->BitWidth == 0) return (EINVAL); bus_set_resource(dev, res_type, *rid, gas->Address, gas->BitWidth / 8); *res = bus_alloc_resource_any(dev, res_type, rid, RF_ACTIVE | flags); if (*res != NULL) { *type = res_type; error = 0; } else bus_delete_resource(dev, res_type, *rid); return (error); } /* Probe _HID and _CID for compatible ISA PNP ids. */ static uint32_t acpi_isa_get_logicalid(device_t dev) { ACPI_DEVICE_INFO *devinfo; ACPI_HANDLE h; uint32_t pnpid; ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__); /* Fetch and validate the HID. */ if ((h = acpi_get_handle(dev)) == NULL || ACPI_FAILURE(AcpiGetObjectInfo(h, &devinfo))) return_VALUE (0); pnpid = (devinfo->Valid & ACPI_VALID_HID) != 0 && devinfo->HardwareId.Length >= ACPI_EISAID_STRING_SIZE ? PNP_EISAID(devinfo->HardwareId.String) : 0; AcpiOsFree(devinfo); return_VALUE (pnpid); } static int acpi_isa_get_compatid(device_t dev, uint32_t *cids, int count) { ACPI_DEVICE_INFO *devinfo; ACPI_PNP_DEVICE_ID *ids; ACPI_HANDLE h; uint32_t *pnpid; int i, valid; ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__); pnpid = cids; /* Fetch and validate the CID */ if ((h = acpi_get_handle(dev)) == NULL || ACPI_FAILURE(AcpiGetObjectInfo(h, &devinfo))) return_VALUE (0); if ((devinfo->Valid & ACPI_VALID_CID) == 0) { AcpiOsFree(devinfo); return_VALUE (0); } if (devinfo->CompatibleIdList.Count < count) count = devinfo->CompatibleIdList.Count; ids = devinfo->CompatibleIdList.Ids; for (i = 0, valid = 0; i < count; i++) if (ids[i].Length >= ACPI_EISAID_STRING_SIZE && strncmp(ids[i].String, "PNP", 3) == 0) { *pnpid++ = PNP_EISAID(ids[i].String); valid++; } AcpiOsFree(devinfo); return_VALUE (valid); } static char * acpi_device_id_probe(device_t bus, device_t dev, char **ids) { ACPI_HANDLE h; ACPI_OBJECT_TYPE t; int i; h = acpi_get_handle(dev); if (ids == NULL || h == NULL) return (NULL); t = acpi_get_type(dev); if (t != ACPI_TYPE_DEVICE && t != ACPI_TYPE_PROCESSOR) return (NULL); /* Try to match one of the array of IDs with a HID or CID. */ for (i = 0; ids[i] != NULL; i++) { if (acpi_MatchHid(h, ids[i])) return (ids[i]); } return (NULL); } static ACPI_STATUS acpi_device_eval_obj(device_t bus, device_t dev, ACPI_STRING pathname, ACPI_OBJECT_LIST *parameters, ACPI_BUFFER *ret) { ACPI_HANDLE h; if (dev == NULL) h = ACPI_ROOT_OBJECT; else if ((h = acpi_get_handle(dev)) == NULL) return (AE_BAD_PARAMETER); return (AcpiEvaluateObject(h, pathname, parameters, ret)); } int acpi_device_pwr_for_sleep(device_t bus, device_t dev, int *dstate) { struct acpi_softc *sc; ACPI_HANDLE handle; ACPI_STATUS status; char sxd[8]; handle = acpi_get_handle(dev); /* * XXX If we find these devices, don't try to power them down. * The serial and IRDA ports on my T23 hang the system when * set to D3 and it appears that such legacy devices may * need special handling in their drivers. */ if (dstate == NULL || handle == NULL || acpi_MatchHid(handle, "PNP0500") || acpi_MatchHid(handle, "PNP0501") || acpi_MatchHid(handle, "PNP0502") || acpi_MatchHid(handle, "PNP0510") || acpi_MatchHid(handle, "PNP0511")) return (ENXIO); /* * Override next state with the value from _SxD, if present. * Note illegal _S0D is evaluated because some systems expect this. */ sc = device_get_softc(bus); snprintf(sxd, sizeof(sxd), "_S%dD", sc->acpi_sstate); status = acpi_GetInteger(handle, sxd, dstate); if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) { device_printf(dev, "failed to get %s on %s: %s\n", sxd, acpi_name(handle), AcpiFormatException(status)); return (ENXIO); } return (0); } /* Callback arg for our implementation of walking the namespace. */ struct acpi_device_scan_ctx { acpi_scan_cb_t user_fn; void *arg; ACPI_HANDLE parent; }; static ACPI_STATUS acpi_device_scan_cb(ACPI_HANDLE h, UINT32 level, void *arg, void **retval) { struct acpi_device_scan_ctx *ctx; device_t dev, old_dev; ACPI_STATUS status; ACPI_OBJECT_TYPE type; /* * Skip this device if we think we'll have trouble with it or it is * the parent where the scan began. */ ctx = (struct acpi_device_scan_ctx *)arg; if (acpi_avoid(h) || h == ctx->parent) return (AE_OK); /* If this is not a valid device type (e.g., a method), skip it. */ if (ACPI_FAILURE(AcpiGetType(h, &type))) return (AE_OK); if (type != ACPI_TYPE_DEVICE && type != ACPI_TYPE_PROCESSOR && type != ACPI_TYPE_THERMAL && type != ACPI_TYPE_POWER) return (AE_OK); /* * Call the user function with the current device. If it is unchanged * afterwards, return. Otherwise, we update the handle to the new dev. */ old_dev = acpi_get_device(h); dev = old_dev; status = ctx->user_fn(h, &dev, level, ctx->arg); if (ACPI_FAILURE(status) || old_dev == dev) return (status); /* Remove the old child and its connection to the handle. */ if (old_dev != NULL) { device_delete_child(device_get_parent(old_dev), old_dev); AcpiDetachData(h, acpi_fake_objhandler); } /* Recreate the handle association if the user created a device. */ if (dev != NULL) AcpiAttachData(h, acpi_fake_objhandler, dev); return (AE_OK); } static ACPI_STATUS acpi_device_scan_children(device_t bus, device_t dev, int max_depth, acpi_scan_cb_t user_fn, void *arg) { ACPI_HANDLE h; struct acpi_device_scan_ctx ctx; if (acpi_disabled("children")) return (AE_OK); if (dev == NULL) h = ACPI_ROOT_OBJECT; else if ((h = acpi_get_handle(dev)) == NULL) return (AE_BAD_PARAMETER); ctx.user_fn = user_fn; ctx.arg = arg; ctx.parent = h; return (AcpiWalkNamespace(ACPI_TYPE_ANY, h, max_depth, acpi_device_scan_cb, NULL, &ctx, NULL)); } /* * Even though ACPI devices are not PCI, we use the PCI approach for setting * device power states since it's close enough to ACPI. */ static int acpi_set_powerstate(device_t child, int state) { ACPI_HANDLE h; ACPI_STATUS status; h = acpi_get_handle(child); if (state < ACPI_STATE_D0 || state > ACPI_D_STATES_MAX) return (EINVAL); if (h == NULL) return (0); /* Ignore errors if the power methods aren't present. */ status = acpi_pwr_switch_consumer(h, state); if (ACPI_SUCCESS(status)) { if (bootverbose) device_printf(child, "set ACPI power state D%d on %s\n", state, acpi_name(h)); } else if (status != AE_NOT_FOUND) device_printf(child, "failed to set ACPI power state D%d on %s: %s\n", state, acpi_name(h), AcpiFormatException(status)); return (0); } static int acpi_isa_pnp_probe(device_t bus, device_t child, struct isa_pnp_id *ids) { int result, cid_count, i; uint32_t lid, cids[8]; ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__); /* * ISA-style drivers attached to ACPI may persist and * probe manually if we return ENOENT. We never want * that to happen, so don't ever return it. */ result = ENXIO; /* Scan the supplied IDs for a match */ lid = acpi_isa_get_logicalid(child); cid_count = acpi_isa_get_compatid(child, cids, 8); while (ids && ids->ip_id) { if (lid == ids->ip_id) { result = 0; goto out; } for (i = 0; i < cid_count; i++) { if (cids[i] == ids->ip_id) { result = 0; goto out; } } ids++; } out: if (result == 0 && ids->ip_desc) device_set_desc(child, ids->ip_desc); return_VALUE (result); } #if defined(__i386__) || defined(__amd64__) /* * Look for a MCFG table. If it is present, use the settings for * domain (segment) 0 to setup PCI config space access via the memory * map. */ static void acpi_enable_pcie(void) { ACPI_TABLE_HEADER *hdr; ACPI_MCFG_ALLOCATION *alloc, *end; ACPI_STATUS status; status = AcpiGetTable(ACPI_SIG_MCFG, 1, &hdr); if (ACPI_FAILURE(status)) return; end = (ACPI_MCFG_ALLOCATION *)((char *)hdr + hdr->Length); alloc = (ACPI_MCFG_ALLOCATION *)((ACPI_TABLE_MCFG *)hdr + 1); while (alloc < end) { if (alloc->PciSegment == 0) { pcie_cfgregopen(alloc->Address, alloc->StartBusNumber, alloc->EndBusNumber); return; } alloc++; } } #endif /* * Scan all of the ACPI namespace and attach child devices. * * We should only expect to find devices in the \_PR, \_TZ, \_SI, and * \_SB scopes, and \_PR and \_TZ became obsolete in the ACPI 2.0 spec. * However, in violation of the spec, some systems place their PCI link * devices in \, so we have to walk the whole namespace. We check the * type of namespace nodes, so this should be ok. */ static void acpi_probe_children(device_t bus) { ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__); /* * Scan the namespace and insert placeholders for all the devices that * we find. We also probe/attach any early devices. * * Note that we use AcpiWalkNamespace rather than AcpiGetDevices because * we want to create nodes for all devices, not just those that are * currently present. (This assumes that we don't want to create/remove * devices as they appear, which might be smarter.) */ ACPI_DEBUG_PRINT((ACPI_DB_OBJECTS, "namespace scan\n")); AcpiWalkNamespace(ACPI_TYPE_ANY, ACPI_ROOT_OBJECT, 100, acpi_probe_child, NULL, bus, NULL); /* Pre-allocate resources for our rman from any sysresource devices. */ acpi_sysres_alloc(bus); /* Reserve resources already allocated to children. */ acpi_reserve_resources(bus); /* Create any static children by calling device identify methods. */ ACPI_DEBUG_PRINT((ACPI_DB_OBJECTS, "device identify routines\n")); bus_generic_probe(bus); /* Probe/attach all children, created statically and from the namespace. */ ACPI_DEBUG_PRINT((ACPI_DB_OBJECTS, "acpi bus_generic_attach\n")); bus_generic_attach(bus); /* Attach wake sysctls. */ acpi_wake_sysctl_walk(bus); ACPI_DEBUG_PRINT((ACPI_DB_OBJECTS, "done attaching children\n")); return_VOID; } /* * Determine the probe order for a given device. */ static void acpi_probe_order(ACPI_HANDLE handle, int *order) { ACPI_OBJECT_TYPE type; /* * 0. CPUs * 1. I/O port and memory system resource holders * 2. Clocks and timers (to handle early accesses) * 3. Embedded controllers (to handle early accesses) * 4. PCI Link Devices */ AcpiGetType(handle, &type); if (type == ACPI_TYPE_PROCESSOR) *order = 0; else if (acpi_MatchHid(handle, "PNP0C01") || acpi_MatchHid(handle, "PNP0C02")) *order = 1; else if (acpi_MatchHid(handle, "PNP0100") || acpi_MatchHid(handle, "PNP0103") || acpi_MatchHid(handle, "PNP0B00")) *order = 2; else if (acpi_MatchHid(handle, "PNP0C09")) *order = 3; else if (acpi_MatchHid(handle, "PNP0C0F")) *order = 4; } /* * Evaluate a child device and determine whether we might attach a device to * it. */ static ACPI_STATUS acpi_probe_child(ACPI_HANDLE handle, UINT32 level, void *context, void **status) { struct acpi_prw_data prw; ACPI_OBJECT_TYPE type; ACPI_HANDLE h; device_t bus, child; char *handle_str; int order; ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__); if (acpi_disabled("children")) return_ACPI_STATUS (AE_OK); /* Skip this device if we think we'll have trouble with it. */ if (acpi_avoid(handle)) return_ACPI_STATUS (AE_OK); bus = (device_t)context; if (ACPI_SUCCESS(AcpiGetType(handle, &type))) { handle_str = acpi_name(handle); switch (type) { case ACPI_TYPE_DEVICE: /* * Since we scan from \, be sure to skip system scope objects. * \_SB_ and \_TZ_ are defined in ACPICA as devices to work around * BIOS bugs. For example, \_SB_ is to allow \_SB_._INI to be run * during the intialization and \_TZ_ is to support Notify() on it. */ if (strcmp(handle_str, "\\_SB_") == 0 || strcmp(handle_str, "\\_TZ_") == 0) break; if (acpi_parse_prw(handle, &prw) == 0) AcpiSetupGpeForWake(handle, prw.gpe_handle, prw.gpe_bit); /* * Ignore devices that do not have a _HID or _CID. They should * be discovered by other buses (e.g. the PCI bus driver). */ if (!acpi_has_hid(handle)) break; /* FALLTHROUGH */ case ACPI_TYPE_PROCESSOR: case ACPI_TYPE_THERMAL: case ACPI_TYPE_POWER: /* * Create a placeholder device for this node. Sort the * placeholder so that the probe/attach passes will run * breadth-first. Orders less than ACPI_DEV_BASE_ORDER * are reserved for special objects (i.e., system * resources). */ ACPI_DEBUG_PRINT((ACPI_DB_OBJECTS, "scanning '%s'\n", handle_str)); order = level * 10 + ACPI_DEV_BASE_ORDER; acpi_probe_order(handle, &order); child = BUS_ADD_CHILD(bus, order, NULL, -1); if (child == NULL) break; /* Associate the handle with the device_t and vice versa. */ acpi_set_handle(child, handle); AcpiAttachData(handle, acpi_fake_objhandler, child); /* * Check that the device is present. If it's not present, * leave it disabled (so that we have a device_t attached to * the handle, but we don't probe it). * * XXX PCI link devices sometimes report "present" but not * "functional" (i.e. if disabled). Go ahead and probe them * anyway since we may enable them later. */ if (type == ACPI_TYPE_DEVICE && !acpi_DeviceIsPresent(child)) { /* Never disable PCI link devices. */ if (acpi_MatchHid(handle, "PNP0C0F")) break; /* * Docking stations should remain enabled since the system * may be undocked at boot. */ if (ACPI_SUCCESS(AcpiGetHandle(handle, "_DCK", &h))) break; device_disable(child); break; } /* * Get the device's resource settings and attach them. * Note that if the device has _PRS but no _CRS, we need * to decide when it's appropriate to try to configure the * device. Ignore the return value here; it's OK for the * device not to have any resources. */ acpi_parse_resources(child, handle, &acpi_res_parse_set, NULL); break; } } return_ACPI_STATUS (AE_OK); } /* * AcpiAttachData() requires an object handler but never uses it. This is a * placeholder object handler so we can store a device_t in an ACPI_HANDLE. */ void acpi_fake_objhandler(ACPI_HANDLE h, void *data) { } static void acpi_shutdown_final(void *arg, int howto) { struct acpi_softc *sc = (struct acpi_softc *)arg; register_t intr; ACPI_STATUS status; /* * XXX Shutdown code should only run on the BSP (cpuid 0). * Some chipsets do not power off the system correctly if called from * an AP. */ if ((howto & RB_POWEROFF) != 0) { status = AcpiEnterSleepStatePrep(ACPI_STATE_S5); if (ACPI_FAILURE(status)) { device_printf(sc->acpi_dev, "AcpiEnterSleepStatePrep failed - %s\n", AcpiFormatException(status)); return; } device_printf(sc->acpi_dev, "Powering system off\n"); intr = intr_disable(); status = AcpiEnterSleepState(ACPI_STATE_S5); if (ACPI_FAILURE(status)) { intr_restore(intr); device_printf(sc->acpi_dev, "power-off failed - %s\n", AcpiFormatException(status)); } else { DELAY(1000000); intr_restore(intr); device_printf(sc->acpi_dev, "power-off failed - timeout\n"); } } else if ((howto & RB_HALT) == 0 && sc->acpi_handle_reboot) { /* Reboot using the reset register. */ status = AcpiReset(); if (ACPI_SUCCESS(status)) { DELAY(1000000); device_printf(sc->acpi_dev, "reset failed - timeout\n"); } else if (status != AE_NOT_EXIST) device_printf(sc->acpi_dev, "reset failed - %s\n", AcpiFormatException(status)); } else if (sc->acpi_do_disable && panicstr == NULL) { /* * Only disable ACPI if the user requested. On some systems, writing * the disable value to SMI_CMD hangs the system. */ device_printf(sc->acpi_dev, "Shutting down\n"); AcpiTerminate(); } } static void acpi_enable_fixed_events(struct acpi_softc *sc) { static int first_time = 1; /* Enable and clear fixed events and install handlers. */ if ((AcpiGbl_FADT.Flags & ACPI_FADT_POWER_BUTTON) == 0) { AcpiClearEvent(ACPI_EVENT_POWER_BUTTON); AcpiInstallFixedEventHandler(ACPI_EVENT_POWER_BUTTON, acpi_event_power_button_sleep, sc); if (first_time) device_printf(sc->acpi_dev, "Power Button (fixed)\n"); } if ((AcpiGbl_FADT.Flags & ACPI_FADT_SLEEP_BUTTON) == 0) { AcpiClearEvent(ACPI_EVENT_SLEEP_BUTTON); AcpiInstallFixedEventHandler(ACPI_EVENT_SLEEP_BUTTON, acpi_event_sleep_button_sleep, sc); if (first_time) device_printf(sc->acpi_dev, "Sleep Button (fixed)\n"); } first_time = 0; } /* * Returns true if the device is actually present and should * be attached to. This requires the present, enabled, UI-visible * and diagnostics-passed bits to be set. */ BOOLEAN acpi_DeviceIsPresent(device_t dev) { ACPI_DEVICE_INFO *devinfo; ACPI_HANDLE h; BOOLEAN present; if ((h = acpi_get_handle(dev)) == NULL || ACPI_FAILURE(AcpiGetObjectInfo(h, &devinfo))) return (FALSE); /* If no _STA method, must be present */ present = (devinfo->Valid & ACPI_VALID_STA) == 0 || ACPI_DEVICE_PRESENT(devinfo->CurrentStatus) ? TRUE : FALSE; AcpiOsFree(devinfo); return (present); } /* * Returns true if the battery is actually present and inserted. */ BOOLEAN acpi_BatteryIsPresent(device_t dev) { ACPI_DEVICE_INFO *devinfo; ACPI_HANDLE h; BOOLEAN present; if ((h = acpi_get_handle(dev)) == NULL || ACPI_FAILURE(AcpiGetObjectInfo(h, &devinfo))) return (FALSE); /* If no _STA method, must be present */ present = (devinfo->Valid & ACPI_VALID_STA) == 0 || ACPI_BATTERY_PRESENT(devinfo->CurrentStatus) ? TRUE : FALSE; AcpiOsFree(devinfo); return (present); } /* * Returns true if a device has at least one valid device ID. */ static BOOLEAN acpi_has_hid(ACPI_HANDLE h) { ACPI_DEVICE_INFO *devinfo; BOOLEAN ret; if (h == NULL || ACPI_FAILURE(AcpiGetObjectInfo(h, &devinfo))) return (FALSE); ret = FALSE; if ((devinfo->Valid & ACPI_VALID_HID) != 0) ret = TRUE; else if ((devinfo->Valid & ACPI_VALID_CID) != 0) if (devinfo->CompatibleIdList.Count > 0) ret = TRUE; AcpiOsFree(devinfo); return (ret); } /* * Match a HID string against a handle */ BOOLEAN acpi_MatchHid(ACPI_HANDLE h, const char *hid) { ACPI_DEVICE_INFO *devinfo; BOOLEAN ret; int i; if (hid == NULL || h == NULL || ACPI_FAILURE(AcpiGetObjectInfo(h, &devinfo))) return (FALSE); ret = FALSE; if ((devinfo->Valid & ACPI_VALID_HID) != 0 && strcmp(hid, devinfo->HardwareId.String) == 0) ret = TRUE; else if ((devinfo->Valid & ACPI_VALID_CID) != 0) for (i = 0; i < devinfo->CompatibleIdList.Count; i++) { if (strcmp(hid, devinfo->CompatibleIdList.Ids[i].String) == 0) { ret = TRUE; break; } } AcpiOsFree(devinfo); return (ret); } /* * Return the handle of a named object within our scope, ie. that of (parent) * or one if its parents. */ ACPI_STATUS acpi_GetHandleInScope(ACPI_HANDLE parent, char *path, ACPI_HANDLE *result) { ACPI_HANDLE r; ACPI_STATUS status; /* Walk back up the tree to the root */ for (;;) { status = AcpiGetHandle(parent, path, &r); if (ACPI_SUCCESS(status)) { *result = r; return (AE_OK); } /* XXX Return error here? */ if (status != AE_NOT_FOUND) return (AE_OK); if (ACPI_FAILURE(AcpiGetParent(parent, &r))) return (AE_NOT_FOUND); parent = r; } } /* * Allocate a buffer with a preset data size. */ ACPI_BUFFER * acpi_AllocBuffer(int size) { ACPI_BUFFER *buf; if ((buf = malloc(size + sizeof(*buf), M_ACPIDEV, M_NOWAIT)) == NULL) return (NULL); buf->Length = size; buf->Pointer = (void *)(buf + 1); return (buf); } ACPI_STATUS acpi_SetInteger(ACPI_HANDLE handle, char *path, UINT32 number) { ACPI_OBJECT arg1; ACPI_OBJECT_LIST args; arg1.Type = ACPI_TYPE_INTEGER; arg1.Integer.Value = number; args.Count = 1; args.Pointer = &arg1; return (AcpiEvaluateObject(handle, path, &args, NULL)); } /* * Evaluate a path that should return an integer. */ ACPI_STATUS acpi_GetInteger(ACPI_HANDLE handle, char *path, UINT32 *number) { ACPI_STATUS status; ACPI_BUFFER buf; ACPI_OBJECT param; if (handle == NULL) handle = ACPI_ROOT_OBJECT; /* * Assume that what we've been pointed at is an Integer object, or * a method that will return an Integer. */ buf.Pointer = ¶m; buf.Length = sizeof(param); status = AcpiEvaluateObject(handle, path, NULL, &buf); if (ACPI_SUCCESS(status)) { if (param.Type == ACPI_TYPE_INTEGER) *number = param.Integer.Value; else status = AE_TYPE; } /* * In some applications, a method that's expected to return an Integer * may instead return a Buffer (probably to simplify some internal * arithmetic). We'll try to fetch whatever it is, and if it's a Buffer, * convert it into an Integer as best we can. * * This is a hack. */ if (status == AE_BUFFER_OVERFLOW) { if ((buf.Pointer = AcpiOsAllocate(buf.Length)) == NULL) { status = AE_NO_MEMORY; } else { status = AcpiEvaluateObject(handle, path, NULL, &buf); if (ACPI_SUCCESS(status)) status = acpi_ConvertBufferToInteger(&buf, number); AcpiOsFree(buf.Pointer); } } return (status); } ACPI_STATUS acpi_ConvertBufferToInteger(ACPI_BUFFER *bufp, UINT32 *number) { ACPI_OBJECT *p; UINT8 *val; int i; p = (ACPI_OBJECT *)bufp->Pointer; if (p->Type == ACPI_TYPE_INTEGER) { *number = p->Integer.Value; return (AE_OK); } if (p->Type != ACPI_TYPE_BUFFER) return (AE_TYPE); if (p->Buffer.Length > sizeof(int)) return (AE_BAD_DATA); *number = 0; val = p->Buffer.Pointer; for (i = 0; i < p->Buffer.Length; i++) *number += val[i] << (i * 8); return (AE_OK); } /* * Iterate over the elements of an a package object, calling the supplied * function for each element. * * XXX possible enhancement might be to abort traversal on error. */ ACPI_STATUS acpi_ForeachPackageObject(ACPI_OBJECT *pkg, void (*func)(ACPI_OBJECT *comp, void *arg), void *arg) { ACPI_OBJECT *comp; int i; if (pkg == NULL || pkg->Type != ACPI_TYPE_PACKAGE) return (AE_BAD_PARAMETER); /* Iterate over components */ i = 0; comp = pkg->Package.Elements; for (; i < pkg->Package.Count; i++, comp++) func(comp, arg); return (AE_OK); } /* * Find the (index)th resource object in a set. */ ACPI_STATUS acpi_FindIndexedResource(ACPI_BUFFER *buf, int index, ACPI_RESOURCE **resp) { ACPI_RESOURCE *rp; int i; rp = (ACPI_RESOURCE *)buf->Pointer; i = index; while (i-- > 0) { /* Range check */ if (rp > (ACPI_RESOURCE *)((u_int8_t *)buf->Pointer + buf->Length)) return (AE_BAD_PARAMETER); /* Check for terminator */ if (rp->Type == ACPI_RESOURCE_TYPE_END_TAG || rp->Length == 0) return (AE_NOT_FOUND); rp = ACPI_NEXT_RESOURCE(rp); } if (resp != NULL) *resp = rp; return (AE_OK); } /* * Append an ACPI_RESOURCE to an ACPI_BUFFER. * * Given a pointer to an ACPI_RESOURCE structure, expand the ACPI_BUFFER * provided to contain it. If the ACPI_BUFFER is empty, allocate a sensible * backing block. If the ACPI_RESOURCE is NULL, return an empty set of * resources. */ #define ACPI_INITIAL_RESOURCE_BUFFER_SIZE 512 ACPI_STATUS acpi_AppendBufferResource(ACPI_BUFFER *buf, ACPI_RESOURCE *res) { ACPI_RESOURCE *rp; void *newp; /* Initialise the buffer if necessary. */ if (buf->Pointer == NULL) { buf->Length = ACPI_INITIAL_RESOURCE_BUFFER_SIZE; if ((buf->Pointer = AcpiOsAllocate(buf->Length)) == NULL) return (AE_NO_MEMORY); rp = (ACPI_RESOURCE *)buf->Pointer; rp->Type = ACPI_RESOURCE_TYPE_END_TAG; rp->Length = ACPI_RS_SIZE_MIN; } if (res == NULL) return (AE_OK); /* * Scan the current buffer looking for the terminator. * This will either find the terminator or hit the end * of the buffer and return an error. */ rp = (ACPI_RESOURCE *)buf->Pointer; for (;;) { /* Range check, don't go outside the buffer */ if (rp >= (ACPI_RESOURCE *)((u_int8_t *)buf->Pointer + buf->Length)) return (AE_BAD_PARAMETER); if (rp->Type == ACPI_RESOURCE_TYPE_END_TAG || rp->Length == 0) break; rp = ACPI_NEXT_RESOURCE(rp); } /* * Check the size of the buffer and expand if required. * * Required size is: * size of existing resources before terminator + * size of new resource and header + * size of terminator. * * Note that this loop should really only run once, unless * for some reason we are stuffing a *really* huge resource. */ while ((((u_int8_t *)rp - (u_int8_t *)buf->Pointer) + res->Length + ACPI_RS_SIZE_NO_DATA + ACPI_RS_SIZE_MIN) >= buf->Length) { if ((newp = AcpiOsAllocate(buf->Length * 2)) == NULL) return (AE_NO_MEMORY); bcopy(buf->Pointer, newp, buf->Length); rp = (ACPI_RESOURCE *)((u_int8_t *)newp + ((u_int8_t *)rp - (u_int8_t *)buf->Pointer)); AcpiOsFree(buf->Pointer); buf->Pointer = newp; buf->Length += buf->Length; } /* Insert the new resource. */ bcopy(res, rp, res->Length + ACPI_RS_SIZE_NO_DATA); /* And add the terminator. */ rp = ACPI_NEXT_RESOURCE(rp); rp->Type = ACPI_RESOURCE_TYPE_END_TAG; rp->Length = ACPI_RS_SIZE_MIN; return (AE_OK); } /* * Set interrupt model. */ ACPI_STATUS acpi_SetIntrModel(int model) { return (acpi_SetInteger(ACPI_ROOT_OBJECT, "_PIC", model)); } /* * Walk subtables of a table and call a callback routine for each * subtable. The caller should provide the first subtable and a * pointer to the end of the table. This can be used to walk tables * such as MADT and SRAT that use subtable entries. */ void acpi_walk_subtables(void *first, void *end, acpi_subtable_handler *handler, void *arg) { ACPI_SUBTABLE_HEADER *entry; for (entry = first; (void *)entry < end; ) { /* Avoid an infinite loop if we hit a bogus entry. */ if (entry->Length < sizeof(ACPI_SUBTABLE_HEADER)) return; handler(entry, arg); entry = ACPI_ADD_PTR(ACPI_SUBTABLE_HEADER, entry, entry->Length); } } /* * DEPRECATED. This interface has serious deficiencies and will be * removed. * * Immediately enter the sleep state. In the old model, acpiconf(8) ran * rc.suspend and rc.resume so we don't have to notify devd(8) to do this. */ ACPI_STATUS acpi_SetSleepState(struct acpi_softc *sc, int state) { static int once; if (!once) { device_printf(sc->acpi_dev, "warning: acpi_SetSleepState() deprecated, need to update your software\n"); once = 1; } return (acpi_EnterSleepState(sc, state)); } #if defined(__amd64__) || defined(__i386__) static void acpi_sleep_force_task(void *context) { struct acpi_softc *sc = (struct acpi_softc *)context; if (ACPI_FAILURE(acpi_EnterSleepState(sc, sc->acpi_next_sstate))) device_printf(sc->acpi_dev, "force sleep state S%d failed\n", sc->acpi_next_sstate); } static void acpi_sleep_force(void *arg) { struct acpi_softc *sc = (struct acpi_softc *)arg; device_printf(sc->acpi_dev, "suspend request timed out, forcing sleep now\n"); /* * XXX Suspending from callout causes freezes in DEVICE_SUSPEND(). * Suspend from acpi_task thread instead. */ if (ACPI_FAILURE(AcpiOsExecute(OSL_NOTIFY_HANDLER, acpi_sleep_force_task, sc))) device_printf(sc->acpi_dev, "AcpiOsExecute() for sleeping failed\n"); } #endif /* * Request that the system enter the given suspend state. All /dev/apm * devices and devd(8) will be notified. Userland then has a chance to * save state and acknowledge the request. The system sleeps once all * acks are in. */ int acpi_ReqSleepState(struct acpi_softc *sc, int state) { #if defined(__amd64__) || defined(__i386__) struct apm_clone_data *clone; ACPI_STATUS status; if (state < ACPI_STATE_S1 || state > ACPI_S_STATES_MAX) return (EINVAL); if (!acpi_sleep_states[state]) return (EOPNOTSUPP); /* If a suspend request is already in progress, just return. */ if (sc->acpi_next_sstate != 0) { return (0); } /* Wait until sleep is enabled. */ while (sc->acpi_sleep_disabled) { AcpiOsSleep(1000); } ACPI_LOCK(acpi); sc->acpi_next_sstate = state; /* S5 (soft-off) should be entered directly with no waiting. */ if (state == ACPI_STATE_S5) { ACPI_UNLOCK(acpi); status = acpi_EnterSleepState(sc, state); return (ACPI_SUCCESS(status) ? 0 : ENXIO); } /* Record the pending state and notify all apm devices. */ STAILQ_FOREACH(clone, &sc->apm_cdevs, entries) { clone->notify_status = APM_EV_NONE; if ((clone->flags & ACPI_EVF_DEVD) == 0) { selwakeuppri(&clone->sel_read, PZERO); KNOTE_LOCKED(&clone->sel_read.si_note, 0); } } /* If devd(8) is not running, immediately enter the sleep state. */ if (!devctl_process_running()) { ACPI_UNLOCK(acpi); status = acpi_EnterSleepState(sc, state); return (ACPI_SUCCESS(status) ? 0 : ENXIO); } /* * Set a timeout to fire if userland doesn't ack the suspend request * in time. This way we still eventually go to sleep if we were * overheating or running low on battery, even if userland is hung. * We cancel this timeout once all userland acks are in or the * suspend request is aborted. */ callout_reset(&sc->susp_force_to, 10 * hz, acpi_sleep_force, sc); ACPI_UNLOCK(acpi); /* Now notify devd(8) also. */ acpi_UserNotify("Suspend", ACPI_ROOT_OBJECT, state); return (0); #else /* This platform does not support acpi suspend/resume. */ return (EOPNOTSUPP); #endif } /* * Acknowledge (or reject) a pending sleep state. The caller has * prepared for suspend and is now ready for it to proceed. If the * error argument is non-zero, it indicates suspend should be cancelled * and gives an errno value describing why. Once all votes are in, * we suspend the system. */ int acpi_AckSleepState(struct apm_clone_data *clone, int error) { #if defined(__amd64__) || defined(__i386__) struct acpi_softc *sc; int ret, sleeping; /* If no pending sleep state, return an error. */ ACPI_LOCK(acpi); sc = clone->acpi_sc; if (sc->acpi_next_sstate == 0) { ACPI_UNLOCK(acpi); return (ENXIO); } /* Caller wants to abort suspend process. */ if (error) { sc->acpi_next_sstate = 0; callout_stop(&sc->susp_force_to); device_printf(sc->acpi_dev, "listener on %s cancelled the pending suspend\n", devtoname(clone->cdev)); ACPI_UNLOCK(acpi); return (0); } /* * Mark this device as acking the suspend request. Then, walk through * all devices, seeing if they agree yet. We only count devices that * are writable since read-only devices couldn't ack the request. */ sleeping = TRUE; clone->notify_status = APM_EV_ACKED; STAILQ_FOREACH(clone, &sc->apm_cdevs, entries) { if ((clone->flags & ACPI_EVF_WRITE) != 0 && clone->notify_status != APM_EV_ACKED) { sleeping = FALSE; break; } } /* If all devices have voted "yes", we will suspend now. */ if (sleeping) callout_stop(&sc->susp_force_to); ACPI_UNLOCK(acpi); ret = 0; if (sleeping) { if (ACPI_FAILURE(acpi_EnterSleepState(sc, sc->acpi_next_sstate))) ret = ENODEV; } return (ret); #else /* This platform does not support acpi suspend/resume. */ return (EOPNOTSUPP); #endif } static void acpi_sleep_enable(void *arg) { struct acpi_softc *sc = (struct acpi_softc *)arg; ACPI_LOCK_ASSERT(acpi); /* Reschedule if the system is not fully up and running. */ if (!AcpiGbl_SystemAwakeAndRunning) { callout_schedule(&acpi_sleep_timer, hz * ACPI_MINIMUM_AWAKETIME); return; } sc->acpi_sleep_disabled = FALSE; } static ACPI_STATUS acpi_sleep_disable(struct acpi_softc *sc) { ACPI_STATUS status; /* Fail if the system is not fully up and running. */ if (!AcpiGbl_SystemAwakeAndRunning) return (AE_ERROR); ACPI_LOCK(acpi); status = sc->acpi_sleep_disabled ? AE_ERROR : AE_OK; sc->acpi_sleep_disabled = TRUE; ACPI_UNLOCK(acpi); return (status); } enum acpi_sleep_state { ACPI_SS_NONE, ACPI_SS_GPE_SET, ACPI_SS_DEV_SUSPEND, ACPI_SS_SLP_PREP, ACPI_SS_SLEPT, }; /* * Enter the desired system sleep state. * * Currently we support S1-S5 but S4 is only S4BIOS */ static ACPI_STATUS acpi_EnterSleepState(struct acpi_softc *sc, int state) { register_t intr; ACPI_STATUS status; ACPI_EVENT_STATUS power_button_status; enum acpi_sleep_state slp_state; int sleep_result; ACPI_FUNCTION_TRACE_U32((char *)(uintptr_t)__func__, state); if (state < ACPI_STATE_S1 || state > ACPI_S_STATES_MAX) return_ACPI_STATUS (AE_BAD_PARAMETER); if (!acpi_sleep_states[state]) { device_printf(sc->acpi_dev, "Sleep state S%d not supported by BIOS\n", state); return (AE_SUPPORT); } /* Re-entry once we're suspending is not allowed. */ status = acpi_sleep_disable(sc); if (ACPI_FAILURE(status)) { device_printf(sc->acpi_dev, "suspend request ignored (not ready yet)\n"); return (status); } if (state == ACPI_STATE_S5) { /* * Shut down cleanly and power off. This will call us back through the * shutdown handlers. */ shutdown_nice(RB_POWEROFF); return_ACPI_STATUS (AE_OK); } EVENTHANDLER_INVOKE(power_suspend_early); stop_all_proc(); EVENTHANDLER_INVOKE(power_suspend); if (smp_started) { thread_lock(curthread); sched_bind(curthread, 0); thread_unlock(curthread); } /* * Be sure to hold Giant across DEVICE_SUSPEND/RESUME since non-MPSAFE * drivers need this. */ mtx_lock(&Giant); slp_state = ACPI_SS_NONE; sc->acpi_sstate = state; /* Enable any GPEs as appropriate and requested by the user. */ acpi_wake_prep_walk(state); slp_state = ACPI_SS_GPE_SET; /* * Inform all devices that we are going to sleep. If at least one * device fails, DEVICE_SUSPEND() automatically resumes the tree. * * XXX Note that a better two-pass approach with a 'veto' pass * followed by a "real thing" pass would be better, but the current * bus interface does not provide for this. */ if (DEVICE_SUSPEND(root_bus) != 0) { device_printf(sc->acpi_dev, "device_suspend failed\n"); goto backout; } slp_state = ACPI_SS_DEV_SUSPEND; /* If testing device suspend only, back out of everything here. */ if (acpi_susp_bounce) goto backout; status = AcpiEnterSleepStatePrep(state); if (ACPI_FAILURE(status)) { device_printf(sc->acpi_dev, "AcpiEnterSleepStatePrep failed - %s\n", AcpiFormatException(status)); goto backout; } slp_state = ACPI_SS_SLP_PREP; if (sc->acpi_sleep_delay > 0) DELAY(sc->acpi_sleep_delay * 1000000); intr = intr_disable(); if (state != ACPI_STATE_S1) { sleep_result = acpi_sleep_machdep(sc, state); acpi_wakeup_machdep(sc, state, sleep_result, 0); /* * XXX According to ACPI specification SCI_EN bit should be restored * by ACPI platform (BIOS, firmware) to its pre-sleep state. * Unfortunately some BIOSes fail to do that and that leads to * unexpected and serious consequences during wake up like a system * getting stuck in SMI handlers. * This hack is picked up from Linux, which claims that it follows * Windows behavior. */ if (sleep_result == 1 && state != ACPI_STATE_S4) AcpiWriteBitRegister(ACPI_BITREG_SCI_ENABLE, ACPI_ENABLE_EVENT); AcpiLeaveSleepStatePrep(state); if (sleep_result == 1 && state == ACPI_STATE_S3) { /* * Prevent mis-interpretation of the wakeup by power button * as a request for power off. * Ideally we should post an appropriate wakeup event, * perhaps using acpi_event_power_button_wake or alike. * * Clearing of power button status after wakeup is mandated * by ACPI specification in section "Fixed Power Button". * * XXX As of ACPICA 20121114 AcpiGetEventStatus provides * status as 0/1 corressponding to inactive/active despite * its type being ACPI_EVENT_STATUS. In other words, * we should not test for ACPI_EVENT_FLAG_SET for time being. */ if (ACPI_SUCCESS(AcpiGetEventStatus(ACPI_EVENT_POWER_BUTTON, &power_button_status)) && power_button_status != 0) { AcpiClearEvent(ACPI_EVENT_POWER_BUTTON); device_printf(sc->acpi_dev, "cleared fixed power button status\n"); } } intr_restore(intr); /* call acpi_wakeup_machdep() again with interrupt enabled */ acpi_wakeup_machdep(sc, state, sleep_result, 1); if (sleep_result == -1) goto backout; /* Re-enable ACPI hardware on wakeup from sleep state 4. */ if (state == ACPI_STATE_S4) AcpiEnable(); } else { status = AcpiEnterSleepState(state); AcpiLeaveSleepStatePrep(state); intr_restore(intr); if (ACPI_FAILURE(status)) { device_printf(sc->acpi_dev, "AcpiEnterSleepState failed - %s\n", AcpiFormatException(status)); goto backout; } } slp_state = ACPI_SS_SLEPT; /* * Back out state according to how far along we got in the suspend * process. This handles both the error and success cases. */ backout: if (slp_state >= ACPI_SS_GPE_SET) { acpi_wake_prep_walk(state); sc->acpi_sstate = ACPI_STATE_S0; } if (slp_state >= ACPI_SS_DEV_SUSPEND) DEVICE_RESUME(root_bus); if (slp_state >= ACPI_SS_SLP_PREP) AcpiLeaveSleepState(state); if (slp_state >= ACPI_SS_SLEPT) { acpi_resync_clock(sc); acpi_enable_fixed_events(sc); } sc->acpi_next_sstate = 0; mtx_unlock(&Giant); if (smp_started) { thread_lock(curthread); sched_unbind(curthread); thread_unlock(curthread); } resume_all_proc(); EVENTHANDLER_INVOKE(power_resume); /* Allow another sleep request after a while. */ callout_schedule(&acpi_sleep_timer, hz * ACPI_MINIMUM_AWAKETIME); /* Run /etc/rc.resume after we are back. */ if (devctl_process_running()) acpi_UserNotify("Resume", ACPI_ROOT_OBJECT, state); return_ACPI_STATUS (status); } static void acpi_resync_clock(struct acpi_softc *sc) { #ifdef __amd64__ if (!acpi_reset_clock) return; /* * Warm up timecounter again and reset system clock. */ (void)timecounter->tc_get_timecount(timecounter); (void)timecounter->tc_get_timecount(timecounter); inittodr(time_second + sc->acpi_sleep_delay); #endif } /* Enable or disable the device's wake GPE. */ int acpi_wake_set_enable(device_t dev, int enable) { struct acpi_prw_data prw; ACPI_STATUS status; int flags; /* Make sure the device supports waking the system and get the GPE. */ if (acpi_parse_prw(acpi_get_handle(dev), &prw) != 0) return (ENXIO); flags = acpi_get_flags(dev); if (enable) { status = AcpiSetGpeWakeMask(prw.gpe_handle, prw.gpe_bit, ACPI_GPE_ENABLE); if (ACPI_FAILURE(status)) { device_printf(dev, "enable wake failed\n"); return (ENXIO); } acpi_set_flags(dev, flags | ACPI_FLAG_WAKE_ENABLED); } else { status = AcpiSetGpeWakeMask(prw.gpe_handle, prw.gpe_bit, ACPI_GPE_DISABLE); if (ACPI_FAILURE(status)) { device_printf(dev, "disable wake failed\n"); return (ENXIO); } acpi_set_flags(dev, flags & ~ACPI_FLAG_WAKE_ENABLED); } return (0); } static int acpi_wake_sleep_prep(ACPI_HANDLE handle, int sstate) { struct acpi_prw_data prw; device_t dev; /* Check that this is a wake-capable device and get its GPE. */ if (acpi_parse_prw(handle, &prw) != 0) return (ENXIO); dev = acpi_get_device(handle); /* * The destination sleep state must be less than (i.e., higher power) * or equal to the value specified by _PRW. If this GPE cannot be * enabled for the next sleep state, then disable it. If it can and * the user requested it be enabled, turn on any required power resources * and set _PSW. */ if (sstate > prw.lowest_wake) { AcpiSetGpeWakeMask(prw.gpe_handle, prw.gpe_bit, ACPI_GPE_DISABLE); if (bootverbose) device_printf(dev, "wake_prep disabled wake for %s (S%d)\n", acpi_name(handle), sstate); } else if (dev && (acpi_get_flags(dev) & ACPI_FLAG_WAKE_ENABLED) != 0) { acpi_pwr_wake_enable(handle, 1); acpi_SetInteger(handle, "_PSW", 1); if (bootverbose) device_printf(dev, "wake_prep enabled for %s (S%d)\n", acpi_name(handle), sstate); } return (0); } static int acpi_wake_run_prep(ACPI_HANDLE handle, int sstate) { struct acpi_prw_data prw; device_t dev; /* * Check that this is a wake-capable device and get its GPE. Return * now if the user didn't enable this device for wake. */ if (acpi_parse_prw(handle, &prw) != 0) return (ENXIO); dev = acpi_get_device(handle); if (dev == NULL || (acpi_get_flags(dev) & ACPI_FLAG_WAKE_ENABLED) == 0) return (0); /* * If this GPE couldn't be enabled for the previous sleep state, it was * disabled before going to sleep so re-enable it. If it was enabled, * clear _PSW and turn off any power resources it used. */ if (sstate > prw.lowest_wake) { AcpiSetGpeWakeMask(prw.gpe_handle, prw.gpe_bit, ACPI_GPE_ENABLE); if (bootverbose) device_printf(dev, "run_prep re-enabled %s\n", acpi_name(handle)); } else { acpi_SetInteger(handle, "_PSW", 0); acpi_pwr_wake_enable(handle, 0); if (bootverbose) device_printf(dev, "run_prep cleaned up for %s\n", acpi_name(handle)); } return (0); } static ACPI_STATUS acpi_wake_prep(ACPI_HANDLE handle, UINT32 level, void *context, void **status) { int sstate; /* If suspending, run the sleep prep function, otherwise wake. */ sstate = *(int *)context; if (AcpiGbl_SystemAwakeAndRunning) acpi_wake_sleep_prep(handle, sstate); else acpi_wake_run_prep(handle, sstate); return (AE_OK); } /* Walk the tree rooted at acpi0 to prep devices for suspend/resume. */ static int acpi_wake_prep_walk(int sstate) { ACPI_HANDLE sb_handle; if (ACPI_SUCCESS(AcpiGetHandle(ACPI_ROOT_OBJECT, "\\_SB_", &sb_handle))) AcpiWalkNamespace(ACPI_TYPE_DEVICE, sb_handle, 100, acpi_wake_prep, NULL, &sstate, NULL); return (0); } /* Walk the tree rooted at acpi0 to attach per-device wake sysctls. */ static int acpi_wake_sysctl_walk(device_t dev) { int error, i, numdevs; device_t *devlist; device_t child; ACPI_STATUS status; error = device_get_children(dev, &devlist, &numdevs); if (error != 0 || numdevs == 0) { if (numdevs == 0) free(devlist, M_TEMP); return (error); } for (i = 0; i < numdevs; i++) { child = devlist[i]; acpi_wake_sysctl_walk(child); if (!device_is_attached(child)) continue; status = AcpiEvaluateObject(acpi_get_handle(child), "_PRW", NULL, NULL); if (ACPI_SUCCESS(status)) { SYSCTL_ADD_PROC(device_get_sysctl_ctx(child), SYSCTL_CHILDREN(device_get_sysctl_tree(child)), OID_AUTO, "wake", CTLTYPE_INT | CTLFLAG_RW, child, 0, acpi_wake_set_sysctl, "I", "Device set to wake the system"); } } free(devlist, M_TEMP); return (0); } /* Enable or disable wake from userland. */ static int acpi_wake_set_sysctl(SYSCTL_HANDLER_ARGS) { int enable, error; device_t dev; dev = (device_t)arg1; enable = (acpi_get_flags(dev) & ACPI_FLAG_WAKE_ENABLED) ? 1 : 0; error = sysctl_handle_int(oidp, &enable, 0, req); if (error != 0 || req->newptr == NULL) return (error); if (enable != 0 && enable != 1) return (EINVAL); return (acpi_wake_set_enable(dev, enable)); } /* Parse a device's _PRW into a structure. */ int acpi_parse_prw(ACPI_HANDLE h, struct acpi_prw_data *prw) { ACPI_STATUS status; ACPI_BUFFER prw_buffer; ACPI_OBJECT *res, *res2; int error, i, power_count; if (h == NULL || prw == NULL) return (EINVAL); /* * The _PRW object (7.2.9) is only required for devices that have the * ability to wake the system from a sleeping state. */ error = EINVAL; prw_buffer.Pointer = NULL; prw_buffer.Length = ACPI_ALLOCATE_BUFFER; status = AcpiEvaluateObject(h, "_PRW", NULL, &prw_buffer); if (ACPI_FAILURE(status)) return (ENOENT); res = (ACPI_OBJECT *)prw_buffer.Pointer; if (res == NULL) return (ENOENT); if (!ACPI_PKG_VALID(res, 2)) goto out; /* * Element 1 of the _PRW object: * The lowest power system sleeping state that can be entered while still * providing wake functionality. The sleeping state being entered must * be less than (i.e., higher power) or equal to this value. */ if (acpi_PkgInt32(res, 1, &prw->lowest_wake) != 0) goto out; /* * Element 0 of the _PRW object: */ switch (res->Package.Elements[0].Type) { case ACPI_TYPE_INTEGER: /* * If the data type of this package element is numeric, then this * _PRW package element is the bit index in the GPEx_EN, in the * GPE blocks described in the FADT, of the enable bit that is * enabled for the wake event. */ prw->gpe_handle = NULL; prw->gpe_bit = res->Package.Elements[0].Integer.Value; error = 0; break; case ACPI_TYPE_PACKAGE: /* * If the data type of this package element is a package, then this * _PRW package element is itself a package containing two * elements. The first is an object reference to the GPE Block * device that contains the GPE that will be triggered by the wake * event. The second element is numeric and it contains the bit * index in the GPEx_EN, in the GPE Block referenced by the * first element in the package, of the enable bit that is enabled for * the wake event. * * For example, if this field is a package then it is of the form: * Package() {\_SB.PCI0.ISA.GPE, 2} */ res2 = &res->Package.Elements[0]; if (!ACPI_PKG_VALID(res2, 2)) goto out; prw->gpe_handle = acpi_GetReference(NULL, &res2->Package.Elements[0]); if (prw->gpe_handle == NULL) goto out; if (acpi_PkgInt32(res2, 1, &prw->gpe_bit) != 0) goto out; error = 0; break; default: goto out; } /* Elements 2 to N of the _PRW object are power resources. */ power_count = res->Package.Count - 2; if (power_count > ACPI_PRW_MAX_POWERRES) { printf("ACPI device %s has too many power resources\n", acpi_name(h)); power_count = 0; } prw->power_res_count = power_count; for (i = 0; i < power_count; i++) prw->power_res[i] = res->Package.Elements[i]; out: if (prw_buffer.Pointer != NULL) AcpiOsFree(prw_buffer.Pointer); return (error); } /* * ACPI Event Handlers */ /* System Event Handlers (registered by EVENTHANDLER_REGISTER) */ static void acpi_system_eventhandler_sleep(void *arg, int state) { struct acpi_softc *sc = (struct acpi_softc *)arg; int ret; ACPI_FUNCTION_TRACE_U32((char *)(uintptr_t)__func__, state); /* Check if button action is disabled or unknown. */ if (state == ACPI_STATE_UNKNOWN) return; /* Request that the system prepare to enter the given suspend state. */ ret = acpi_ReqSleepState(sc, state); if (ret != 0) device_printf(sc->acpi_dev, "request to enter state S%d failed (err %d)\n", state, ret); return_VOID; } static void acpi_system_eventhandler_wakeup(void *arg, int state) { ACPI_FUNCTION_TRACE_U32((char *)(uintptr_t)__func__, state); /* Currently, nothing to do for wakeup. */ return_VOID; } /* * ACPICA Event Handlers (FixedEvent, also called from button notify handler) */ static void acpi_invoke_sleep_eventhandler(void *context) { EVENTHANDLER_INVOKE(acpi_sleep_event, *(int *)context); } static void acpi_invoke_wake_eventhandler(void *context) { EVENTHANDLER_INVOKE(acpi_wakeup_event, *(int *)context); } UINT32 acpi_event_power_button_sleep(void *context) { struct acpi_softc *sc = (struct acpi_softc *)context; ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__); if (ACPI_FAILURE(AcpiOsExecute(OSL_NOTIFY_HANDLER, acpi_invoke_sleep_eventhandler, &sc->acpi_power_button_sx))) return_VALUE (ACPI_INTERRUPT_NOT_HANDLED); return_VALUE (ACPI_INTERRUPT_HANDLED); } UINT32 acpi_event_power_button_wake(void *context) { struct acpi_softc *sc = (struct acpi_softc *)context; ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__); if (ACPI_FAILURE(AcpiOsExecute(OSL_NOTIFY_HANDLER, acpi_invoke_wake_eventhandler, &sc->acpi_power_button_sx))) return_VALUE (ACPI_INTERRUPT_NOT_HANDLED); return_VALUE (ACPI_INTERRUPT_HANDLED); } UINT32 acpi_event_sleep_button_sleep(void *context) { struct acpi_softc *sc = (struct acpi_softc *)context; ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__); if (ACPI_FAILURE(AcpiOsExecute(OSL_NOTIFY_HANDLER, acpi_invoke_sleep_eventhandler, &sc->acpi_sleep_button_sx))) return_VALUE (ACPI_INTERRUPT_NOT_HANDLED); return_VALUE (ACPI_INTERRUPT_HANDLED); } UINT32 acpi_event_sleep_button_wake(void *context) { struct acpi_softc *sc = (struct acpi_softc *)context; ACPI_FUNCTION_TRACE((char *)(uintptr_t)__func__); if (ACPI_FAILURE(AcpiOsExecute(OSL_NOTIFY_HANDLER, acpi_invoke_wake_eventhandler, &sc->acpi_sleep_button_sx))) return_VALUE (ACPI_INTERRUPT_NOT_HANDLED); return_VALUE (ACPI_INTERRUPT_HANDLED); } /* * XXX This static buffer is suboptimal. There is no locking so only * use this for single-threaded callers. */ char * acpi_name(ACPI_HANDLE handle) { ACPI_BUFFER buf; static char data[256]; buf.Length = sizeof(data); buf.Pointer = data; if (handle && ACPI_SUCCESS(AcpiGetName(handle, ACPI_FULL_PATHNAME, &buf))) return (data); return ("(unknown)"); } /* * Debugging/bug-avoidance. Avoid trying to fetch info on various * parts of the namespace. */ int acpi_avoid(ACPI_HANDLE handle) { char *cp, *env, *np; int len; np = acpi_name(handle); if (*np == '\\') np++; if ((env = kern_getenv("debug.acpi.avoid")) == NULL) return (0); /* Scan the avoid list checking for a match */ cp = env; for (;;) { while (*cp != 0 && isspace(*cp)) cp++; if (*cp == 0) break; len = 0; while (cp[len] != 0 && !isspace(cp[len])) len++; if (!strncmp(cp, np, len)) { freeenv(env); return(1); } cp += len; } freeenv(env); return (0); } /* * Debugging/bug-avoidance. Disable ACPI subsystem components. */ int acpi_disabled(char *subsys) { char *cp, *env; int len; if ((env = kern_getenv("debug.acpi.disabled")) == NULL) return (0); if (strcmp(env, "all") == 0) { freeenv(env); return (1); } /* Scan the disable list, checking for a match. */ cp = env; for (;;) { while (*cp != '\0' && isspace(*cp)) cp++; if (*cp == '\0') break; len = 0; while (cp[len] != '\0' && !isspace(cp[len])) len++; if (strncmp(cp, subsys, len) == 0) { freeenv(env); return (1); } cp += len; } freeenv(env); return (0); } static void acpi_lookup(void *arg, const char *name, device_t *dev) { ACPI_HANDLE handle; if (*dev != NULL) return; /* * Allow any handle name that is specified as an absolute path and * starts with '\'. We could restrict this to \_SB and friends, * but see acpi_probe_children() for notes on why we scan the entire * namespace for devices. * * XXX: The pathname argument to AcpiGetHandle() should be fixed to * be const. */ if (name[0] != '\\') return; if (ACPI_FAILURE(AcpiGetHandle(ACPI_ROOT_OBJECT, __DECONST(char *, name), &handle))) return; *dev = acpi_get_device(handle); } /* * Control interface. * * We multiplex ioctls for all participating ACPI devices here. Individual * drivers wanting to be accessible via /dev/acpi should use the * register/deregister interface to make their handlers visible. */ struct acpi_ioctl_hook { TAILQ_ENTRY(acpi_ioctl_hook) link; u_long cmd; acpi_ioctl_fn fn; void *arg; }; static TAILQ_HEAD(,acpi_ioctl_hook) acpi_ioctl_hooks; static int acpi_ioctl_hooks_initted; int acpi_register_ioctl(u_long cmd, acpi_ioctl_fn fn, void *arg) { struct acpi_ioctl_hook *hp; if ((hp = malloc(sizeof(*hp), M_ACPIDEV, M_NOWAIT)) == NULL) return (ENOMEM); hp->cmd = cmd; hp->fn = fn; hp->arg = arg; ACPI_LOCK(acpi); if (acpi_ioctl_hooks_initted == 0) { TAILQ_INIT(&acpi_ioctl_hooks); acpi_ioctl_hooks_initted = 1; } TAILQ_INSERT_TAIL(&acpi_ioctl_hooks, hp, link); ACPI_UNLOCK(acpi); return (0); } void acpi_deregister_ioctl(u_long cmd, acpi_ioctl_fn fn) { struct acpi_ioctl_hook *hp; ACPI_LOCK(acpi); TAILQ_FOREACH(hp, &acpi_ioctl_hooks, link) if (hp->cmd == cmd && hp->fn == fn) break; if (hp != NULL) { TAILQ_REMOVE(&acpi_ioctl_hooks, hp, link); free(hp, M_ACPIDEV); } ACPI_UNLOCK(acpi); } static int acpiopen(struct cdev *dev, int flag, int fmt, struct thread *td) { return (0); } static int acpiclose(struct cdev *dev, int flag, int fmt, struct thread *td) { return (0); } static int acpiioctl(struct cdev *dev, u_long cmd, caddr_t addr, int flag, struct thread *td) { struct acpi_softc *sc; struct acpi_ioctl_hook *hp; int error, state; error = 0; hp = NULL; sc = dev->si_drv1; /* * Scan the list of registered ioctls, looking for handlers. */ ACPI_LOCK(acpi); if (acpi_ioctl_hooks_initted) TAILQ_FOREACH(hp, &acpi_ioctl_hooks, link) { if (hp->cmd == cmd) break; } ACPI_UNLOCK(acpi); if (hp) return (hp->fn(cmd, addr, hp->arg)); /* * Core ioctls are not permitted for non-writable user. * Currently, other ioctls just fetch information. * Not changing system behavior. */ if ((flag & FWRITE) == 0) return (EPERM); /* Core system ioctls. */ switch (cmd) { case ACPIIO_REQSLPSTATE: state = *(int *)addr; if (state != ACPI_STATE_S5) return (acpi_ReqSleepState(sc, state)); device_printf(sc->acpi_dev, "power off via acpi ioctl not supported\n"); error = EOPNOTSUPP; break; case ACPIIO_ACKSLPSTATE: error = *(int *)addr; error = acpi_AckSleepState(sc->acpi_clone, error); break; case ACPIIO_SETSLPSTATE: /* DEPRECATED */ state = *(int *)addr; if (state < ACPI_STATE_S0 || state > ACPI_S_STATES_MAX) return (EINVAL); if (!acpi_sleep_states[state]) return (EOPNOTSUPP); if (ACPI_FAILURE(acpi_SetSleepState(sc, state))) error = ENXIO; break; default: error = ENXIO; break; } return (error); } static int acpi_sname2sstate(const char *sname) { int sstate; if (toupper(sname[0]) == 'S') { sstate = sname[1] - '0'; if (sstate >= ACPI_STATE_S0 && sstate <= ACPI_STATE_S5 && sname[2] == '\0') return (sstate); } else if (strcasecmp(sname, "NONE") == 0) return (ACPI_STATE_UNKNOWN); return (-1); } static const char * acpi_sstate2sname(int sstate) { static const char *snames[] = { "S0", "S1", "S2", "S3", "S4", "S5" }; if (sstate >= ACPI_STATE_S0 && sstate <= ACPI_STATE_S5) return (snames[sstate]); else if (sstate == ACPI_STATE_UNKNOWN) return ("NONE"); return (NULL); } static int acpi_supported_sleep_state_sysctl(SYSCTL_HANDLER_ARGS) { int error; struct sbuf sb; UINT8 state; sbuf_new(&sb, NULL, 32, SBUF_AUTOEXTEND); for (state = ACPI_STATE_S1; state < ACPI_S_STATE_COUNT; state++) if (acpi_sleep_states[state]) sbuf_printf(&sb, "%s ", acpi_sstate2sname(state)); sbuf_trim(&sb); sbuf_finish(&sb); error = sysctl_handle_string(oidp, sbuf_data(&sb), sbuf_len(&sb), req); sbuf_delete(&sb); return (error); } static int acpi_sleep_state_sysctl(SYSCTL_HANDLER_ARGS) { char sleep_state[10]; int error, new_state, old_state; old_state = *(int *)oidp->oid_arg1; strlcpy(sleep_state, acpi_sstate2sname(old_state), sizeof(sleep_state)); error = sysctl_handle_string(oidp, sleep_state, sizeof(sleep_state), req); if (error == 0 && req->newptr != NULL) { new_state = acpi_sname2sstate(sleep_state); if (new_state < ACPI_STATE_S1) return (EINVAL); if (new_state < ACPI_S_STATE_COUNT && !acpi_sleep_states[new_state]) return (EOPNOTSUPP); if (new_state != old_state) *(int *)oidp->oid_arg1 = new_state; } return (error); } /* Inform devctl(4) when we receive a Notify. */ void acpi_UserNotify(const char *subsystem, ACPI_HANDLE h, uint8_t notify) { char notify_buf[16]; ACPI_BUFFER handle_buf; ACPI_STATUS status; if (subsystem == NULL) return; handle_buf.Pointer = NULL; handle_buf.Length = ACPI_ALLOCATE_BUFFER; status = AcpiNsHandleToPathname(h, &handle_buf); if (ACPI_FAILURE(status)) return; snprintf(notify_buf, sizeof(notify_buf), "notify=0x%02x", notify); devctl_notify("ACPI", subsystem, handle_buf.Pointer, notify_buf); AcpiOsFree(handle_buf.Pointer); } #ifdef ACPI_DEBUG /* * Support for parsing debug options from the kernel environment. * * Bits may be set in the AcpiDbgLayer and AcpiDbgLevel debug registers * by specifying the names of the bits in the debug.acpi.layer and * debug.acpi.level environment variables. Bits may be unset by * prefixing the bit name with !. */ struct debugtag { char *name; UINT32 value; }; static struct debugtag dbg_layer[] = { {"ACPI_UTILITIES", ACPI_UTILITIES}, {"ACPI_HARDWARE", ACPI_HARDWARE}, {"ACPI_EVENTS", ACPI_EVENTS}, {"ACPI_TABLES", ACPI_TABLES}, {"ACPI_NAMESPACE", ACPI_NAMESPACE}, {"ACPI_PARSER", ACPI_PARSER}, {"ACPI_DISPATCHER", ACPI_DISPATCHER}, {"ACPI_EXECUTER", ACPI_EXECUTER}, {"ACPI_RESOURCES", ACPI_RESOURCES}, {"ACPI_CA_DEBUGGER", ACPI_CA_DEBUGGER}, {"ACPI_OS_SERVICES", ACPI_OS_SERVICES}, {"ACPI_CA_DISASSEMBLER", ACPI_CA_DISASSEMBLER}, {"ACPI_ALL_COMPONENTS", ACPI_ALL_COMPONENTS}, {"ACPI_AC_ADAPTER", ACPI_AC_ADAPTER}, {"ACPI_BATTERY", ACPI_BATTERY}, {"ACPI_BUS", ACPI_BUS}, {"ACPI_BUTTON", ACPI_BUTTON}, {"ACPI_EC", ACPI_EC}, {"ACPI_FAN", ACPI_FAN}, {"ACPI_POWERRES", ACPI_POWERRES}, {"ACPI_PROCESSOR", ACPI_PROCESSOR}, {"ACPI_THERMAL", ACPI_THERMAL}, {"ACPI_TIMER", ACPI_TIMER}, {"ACPI_ALL_DRIVERS", ACPI_ALL_DRIVERS}, {NULL, 0} }; static struct debugtag dbg_level[] = { {"ACPI_LV_INIT", ACPI_LV_INIT}, {"ACPI_LV_DEBUG_OBJECT", ACPI_LV_DEBUG_OBJECT}, {"ACPI_LV_INFO", ACPI_LV_INFO}, {"ACPI_LV_REPAIR", ACPI_LV_REPAIR}, {"ACPI_LV_ALL_EXCEPTIONS", ACPI_LV_ALL_EXCEPTIONS}, /* Trace verbosity level 1 [Standard Trace Level] */ {"ACPI_LV_INIT_NAMES", ACPI_LV_INIT_NAMES}, {"ACPI_LV_PARSE", ACPI_LV_PARSE}, {"ACPI_LV_LOAD", ACPI_LV_LOAD}, {"ACPI_LV_DISPATCH", ACPI_LV_DISPATCH}, {"ACPI_LV_EXEC", ACPI_LV_EXEC}, {"ACPI_LV_NAMES", ACPI_LV_NAMES}, {"ACPI_LV_OPREGION", ACPI_LV_OPREGION}, {"ACPI_LV_BFIELD", ACPI_LV_BFIELD}, {"ACPI_LV_TABLES", ACPI_LV_TABLES}, {"ACPI_LV_VALUES", ACPI_LV_VALUES}, {"ACPI_LV_OBJECTS", ACPI_LV_OBJECTS}, {"ACPI_LV_RESOURCES", ACPI_LV_RESOURCES}, {"ACPI_LV_USER_REQUESTS", ACPI_LV_USER_REQUESTS}, {"ACPI_LV_PACKAGE", ACPI_LV_PACKAGE}, {"ACPI_LV_VERBOSITY1", ACPI_LV_VERBOSITY1}, /* Trace verbosity level 2 [Function tracing and memory allocation] */ {"ACPI_LV_ALLOCATIONS", ACPI_LV_ALLOCATIONS}, {"ACPI_LV_FUNCTIONS", ACPI_LV_FUNCTIONS}, {"ACPI_LV_OPTIMIZATIONS", ACPI_LV_OPTIMIZATIONS}, {"ACPI_LV_VERBOSITY2", ACPI_LV_VERBOSITY2}, {"ACPI_LV_ALL", ACPI_LV_ALL}, /* Trace verbosity level 3 [Threading, I/O, and Interrupts] */ {"ACPI_LV_MUTEX", ACPI_LV_MUTEX}, {"ACPI_LV_THREADS", ACPI_LV_THREADS}, {"ACPI_LV_IO", ACPI_LV_IO}, {"ACPI_LV_INTERRUPTS", ACPI_LV_INTERRUPTS}, {"ACPI_LV_VERBOSITY3", ACPI_LV_VERBOSITY3}, /* Exceptionally verbose output -- also used in the global "DebugLevel" */ {"ACPI_LV_AML_DISASSEMBLE", ACPI_LV_AML_DISASSEMBLE}, {"ACPI_LV_VERBOSE_INFO", ACPI_LV_VERBOSE_INFO}, {"ACPI_LV_FULL_TABLES", ACPI_LV_FULL_TABLES}, {"ACPI_LV_EVENTS", ACPI_LV_EVENTS}, {"ACPI_LV_VERBOSE", ACPI_LV_VERBOSE}, {NULL, 0} }; static void acpi_parse_debug(char *cp, struct debugtag *tag, UINT32 *flag) { char *ep; int i, l; int set; while (*cp) { if (isspace(*cp)) { cp++; continue; } ep = cp; while (*ep && !isspace(*ep)) ep++; if (*cp == '!') { set = 0; cp++; if (cp == ep) continue; } else { set = 1; } l = ep - cp; for (i = 0; tag[i].name != NULL; i++) { if (!strncmp(cp, tag[i].name, l)) { if (set) *flag |= tag[i].value; else *flag &= ~tag[i].value; } } cp = ep; } } static void acpi_set_debugging(void *junk) { char *layer, *level; if (cold) { AcpiDbgLayer = 0; AcpiDbgLevel = 0; } layer = kern_getenv("debug.acpi.layer"); level = kern_getenv("debug.acpi.level"); if (layer == NULL && level == NULL) return; printf("ACPI set debug"); if (layer != NULL) { if (strcmp("NONE", layer) != 0) printf(" layer '%s'", layer); acpi_parse_debug(layer, &dbg_layer[0], &AcpiDbgLayer); freeenv(layer); } if (level != NULL) { if (strcmp("NONE", level) != 0) printf(" level '%s'", level); acpi_parse_debug(level, &dbg_level[0], &AcpiDbgLevel); freeenv(level); } printf("\n"); } SYSINIT(acpi_debugging, SI_SUB_TUNABLES, SI_ORDER_ANY, acpi_set_debugging, NULL); static int acpi_debug_sysctl(SYSCTL_HANDLER_ARGS) { int error, *dbg; struct debugtag *tag; struct sbuf sb; char temp[128]; if (sbuf_new(&sb, NULL, 128, SBUF_AUTOEXTEND) == NULL) return (ENOMEM); if (strcmp(oidp->oid_arg1, "debug.acpi.layer") == 0) { tag = &dbg_layer[0]; dbg = &AcpiDbgLayer; } else { tag = &dbg_level[0]; dbg = &AcpiDbgLevel; } /* Get old values if this is a get request. */ ACPI_SERIAL_BEGIN(acpi); if (*dbg == 0) { sbuf_cpy(&sb, "NONE"); } else if (req->newptr == NULL) { for (; tag->name != NULL; tag++) { if ((*dbg & tag->value) == tag->value) sbuf_printf(&sb, "%s ", tag->name); } } sbuf_trim(&sb); sbuf_finish(&sb); strlcpy(temp, sbuf_data(&sb), sizeof(temp)); sbuf_delete(&sb); error = sysctl_handle_string(oidp, temp, sizeof(temp), req); /* Check for error or no change */ if (error == 0 && req->newptr != NULL) { *dbg = 0; kern_setenv((char *)oidp->oid_arg1, temp); acpi_set_debugging(NULL); } ACPI_SERIAL_END(acpi); return (error); } SYSCTL_PROC(_debug_acpi, OID_AUTO, layer, CTLFLAG_RW | CTLTYPE_STRING, "debug.acpi.layer", 0, acpi_debug_sysctl, "A", ""); SYSCTL_PROC(_debug_acpi, OID_AUTO, level, CTLFLAG_RW | CTLTYPE_STRING, "debug.acpi.level", 0, acpi_debug_sysctl, "A", ""); #endif /* ACPI_DEBUG */ static int acpi_debug_objects_sysctl(SYSCTL_HANDLER_ARGS) { int error; int old; old = acpi_debug_objects; error = sysctl_handle_int(oidp, &acpi_debug_objects, 0, req); if (error != 0 || req->newptr == NULL) return (error); if (old == acpi_debug_objects || (old && acpi_debug_objects)) return (0); ACPI_SERIAL_BEGIN(acpi); AcpiGbl_EnableAmlDebugObject = acpi_debug_objects ? TRUE : FALSE; ACPI_SERIAL_END(acpi); return (0); } static int acpi_parse_interfaces(char *str, struct acpi_interface *iface) { char *p; size_t len; int i, j; p = str; while (isspace(*p) || *p == ',') p++; len = strlen(p); if (len == 0) return (0); p = strdup(p, M_TEMP); for (i = 0; i < len; i++) if (p[i] == ',') p[i] = '\0'; i = j = 0; while (i < len) if (isspace(p[i]) || p[i] == '\0') i++; else { i += strlen(p + i) + 1; j++; } if (j == 0) { free(p, M_TEMP); return (0); } iface->data = malloc(sizeof(*iface->data) * j, M_TEMP, M_WAITOK); iface->num = j; i = j = 0; while (i < len) if (isspace(p[i]) || p[i] == '\0') i++; else { iface->data[j] = p + i; i += strlen(p + i) + 1; j++; } return (j); } static void acpi_free_interfaces(struct acpi_interface *iface) { free(iface->data[0], M_TEMP); free(iface->data, M_TEMP); } static void acpi_reset_interfaces(device_t dev) { struct acpi_interface list; ACPI_STATUS status; int i; if (acpi_parse_interfaces(acpi_install_interface, &list) > 0) { for (i = 0; i < list.num; i++) { status = AcpiInstallInterface(list.data[i]); if (ACPI_FAILURE(status)) device_printf(dev, "failed to install _OSI(\"%s\"): %s\n", list.data[i], AcpiFormatException(status)); else if (bootverbose) device_printf(dev, "installed _OSI(\"%s\")\n", list.data[i]); } acpi_free_interfaces(&list); } if (acpi_parse_interfaces(acpi_remove_interface, &list) > 0) { for (i = 0; i < list.num; i++) { status = AcpiRemoveInterface(list.data[i]); if (ACPI_FAILURE(status)) device_printf(dev, "failed to remove _OSI(\"%s\"): %s\n", list.data[i], AcpiFormatException(status)); else if (bootverbose) device_printf(dev, "removed _OSI(\"%s\")\n", list.data[i]); } acpi_free_interfaces(&list); } } static int acpi_pm_func(u_long cmd, void *arg, ...) { int state, acpi_state; int error; struct acpi_softc *sc; va_list ap; error = 0; switch (cmd) { case POWER_CMD_SUSPEND: sc = (struct acpi_softc *)arg; if (sc == NULL) { error = EINVAL; goto out; } va_start(ap, arg); state = va_arg(ap, int); va_end(ap); switch (state) { case POWER_SLEEP_STATE_STANDBY: acpi_state = sc->acpi_standby_sx; break; case POWER_SLEEP_STATE_SUSPEND: acpi_state = sc->acpi_suspend_sx; break; case POWER_SLEEP_STATE_HIBERNATE: acpi_state = ACPI_STATE_S4; break; default: error = EINVAL; goto out; } if (ACPI_FAILURE(acpi_EnterSleepState(sc, acpi_state))) error = ENXIO; break; default: error = EINVAL; goto out; } out: return (error); } static void acpi_pm_register(void *arg) { if (!cold || resource_disabled("acpi", 0)) return; power_pm_register(POWER_PM_TYPE_ACPI, acpi_pm_func, NULL); } SYSINIT(power, SI_SUB_KLD, SI_ORDER_ANY, acpi_pm_register, 0); Index: projects/cxl_iscsi/sys/dev/ahci/ahci_pci.c =================================================================== --- projects/cxl_iscsi/sys/dev/ahci/ahci_pci.c (revision 279716) +++ projects/cxl_iscsi/sys/dev/ahci/ahci_pci.c (revision 279717) @@ -1,522 +1,522 @@ /*- * Copyright (c) 2009-2012 Alexander Motin * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer, * without modification, immediately at the beginning of the file. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "ahci.h" static int force_ahci = 1; TUNABLE_INT("hw.ahci.force", &force_ahci); static const struct { uint32_t id; uint8_t rev; const char *name; int quirks; } ahci_ids[] = { {0x43801002, 0x00, "AMD SB600", AHCI_Q_NOMSI | AHCI_Q_ATI_PMP_BUG | AHCI_Q_MAXIO_64K}, {0x43901002, 0x00, "AMD SB7x0/SB8x0/SB9x0", AHCI_Q_ATI_PMP_BUG | AHCI_Q_1MSI}, {0x43911002, 0x00, "AMD SB7x0/SB8x0/SB9x0", AHCI_Q_ATI_PMP_BUG | AHCI_Q_1MSI}, {0x43921002, 0x00, "AMD SB7x0/SB8x0/SB9x0", AHCI_Q_ATI_PMP_BUG | AHCI_Q_1MSI}, {0x43931002, 0x00, "AMD SB7x0/SB8x0/SB9x0", AHCI_Q_ATI_PMP_BUG | AHCI_Q_1MSI}, {0x43941002, 0x00, "AMD SB7x0/SB8x0/SB9x0", AHCI_Q_ATI_PMP_BUG | AHCI_Q_1MSI}, /* Not sure SB8x0/SB9x0 needs this quirk. Be conservative though */ {0x43951002, 0x00, "AMD SB8x0/SB9x0", AHCI_Q_ATI_PMP_BUG}, {0x78001022, 0x00, "AMD Hudson-2", 0}, {0x78011022, 0x00, "AMD Hudson-2", 0}, {0x78021022, 0x00, "AMD Hudson-2", 0}, {0x78031022, 0x00, "AMD Hudson-2", 0}, {0x78041022, 0x00, "AMD Hudson-2", 0}, {0x06111b21, 0x00, "ASMedia ASM2106", 0}, {0x06121b21, 0x00, "ASMedia ASM1061", 0}, {0x26528086, 0x00, "Intel ICH6", AHCI_Q_NOFORCE}, {0x26538086, 0x00, "Intel ICH6M", AHCI_Q_NOFORCE}, {0x26818086, 0x00, "Intel ESB2", 0}, {0x26828086, 0x00, "Intel ESB2", 0}, {0x26838086, 0x00, "Intel ESB2", 0}, {0x27c18086, 0x00, "Intel ICH7", 0}, {0x27c38086, 0x00, "Intel ICH7", 0}, {0x27c58086, 0x00, "Intel ICH7M", 0}, {0x27c68086, 0x00, "Intel ICH7M", 0}, {0x28218086, 0x00, "Intel ICH8", 0}, {0x28228086, 0x00, "Intel ICH8", 0}, {0x28248086, 0x00, "Intel ICH8", 0}, {0x28298086, 0x00, "Intel ICH8M", 0}, {0x282a8086, 0x00, "Intel ICH8M", 0}, {0x29228086, 0x00, "Intel ICH9", 0}, {0x29238086, 0x00, "Intel ICH9", 0}, {0x29248086, 0x00, "Intel ICH9", 0}, {0x29258086, 0x00, "Intel ICH9", 0}, {0x29278086, 0x00, "Intel ICH9", 0}, {0x29298086, 0x00, "Intel ICH9M", 0}, {0x292a8086, 0x00, "Intel ICH9M", 0}, {0x292b8086, 0x00, "Intel ICH9M", 0}, {0x292c8086, 0x00, "Intel ICH9M", 0}, {0x292f8086, 0x00, "Intel ICH9M", 0}, {0x294d8086, 0x00, "Intel ICH9", 0}, {0x294e8086, 0x00, "Intel ICH9M", 0}, {0x3a058086, 0x00, "Intel ICH10", 0}, {0x3a228086, 0x00, "Intel ICH10", 0}, {0x3a258086, 0x00, "Intel ICH10", 0}, {0x3b228086, 0x00, "Intel 5 Series/3400 Series", 0}, {0x3b238086, 0x00, "Intel 5 Series/3400 Series", 0}, {0x3b258086, 0x00, "Intel 5 Series/3400 Series", 0}, {0x3b298086, 0x00, "Intel 5 Series/3400 Series", 0}, {0x3b2c8086, 0x00, "Intel 5 Series/3400 Series", 0}, {0x3b2f8086, 0x00, "Intel 5 Series/3400 Series", 0}, {0x1c028086, 0x00, "Intel Cougar Point", 0}, {0x1c038086, 0x00, "Intel Cougar Point", 0}, {0x1c048086, 0x00, "Intel Cougar Point", 0}, {0x1c058086, 0x00, "Intel Cougar Point", 0}, {0x1d028086, 0x00, "Intel Patsburg", 0}, {0x1d048086, 0x00, "Intel Patsburg", 0}, {0x1d068086, 0x00, "Intel Patsburg", 0}, {0x28268086, 0x00, "Intel Patsburg (RAID)", 0}, {0x1e028086, 0x00, "Intel Panther Point", 0}, {0x1e038086, 0x00, "Intel Panther Point", 0}, {0x1e048086, 0x00, "Intel Panther Point (RAID)", 0}, {0x1e058086, 0x00, "Intel Panther Point (RAID)", 0}, {0x1e068086, 0x00, "Intel Panther Point (RAID)", 0}, {0x1e078086, 0x00, "Intel Panther Point (RAID)", 0}, {0x1e0e8086, 0x00, "Intel Panther Point (RAID)", 0}, {0x1e0f8086, 0x00, "Intel Panther Point (RAID)", 0}, {0x1f228086, 0x00, "Intel Avoton", 0}, {0x1f238086, 0x00, "Intel Avoton", 0}, {0x1f248086, 0x00, "Intel Avoton (RAID)", 0}, {0x1f258086, 0x00, "Intel Avoton (RAID)", 0}, {0x1f268086, 0x00, "Intel Avoton (RAID)", 0}, {0x1f278086, 0x00, "Intel Avoton (RAID)", 0}, {0x1f2e8086, 0x00, "Intel Avoton (RAID)", 0}, {0x1f2f8086, 0x00, "Intel Avoton (RAID)", 0}, {0x1f328086, 0x00, "Intel Avoton", 0}, {0x1f338086, 0x00, "Intel Avoton", 0}, {0x1f348086, 0x00, "Intel Avoton (RAID)", 0}, {0x1f358086, 0x00, "Intel Avoton (RAID)", 0}, {0x1f368086, 0x00, "Intel Avoton (RAID)", 0}, {0x1f378086, 0x00, "Intel Avoton (RAID)", 0}, {0x1f3e8086, 0x00, "Intel Avoton (RAID)", 0}, {0x1f3f8086, 0x00, "Intel Avoton (RAID)", 0}, {0x23a38086, 0x00, "Intel Coleto Creek", 0}, {0x28238086, 0x00, "Intel Wellsburg (RAID)", 0}, {0x28278086, 0x00, "Intel Wellsburg (RAID)", 0}, {0x8c028086, 0x00, "Intel Lynx Point", 0}, {0x8c038086, 0x00, "Intel Lynx Point", 0}, {0x8c048086, 0x00, "Intel Lynx Point (RAID)", 0}, {0x8c058086, 0x00, "Intel Lynx Point (RAID)", 0}, {0x8c068086, 0x00, "Intel Lynx Point (RAID)", 0}, {0x8c078086, 0x00, "Intel Lynx Point (RAID)", 0}, {0x8c0e8086, 0x00, "Intel Lynx Point (RAID)", 0}, {0x8c0f8086, 0x00, "Intel Lynx Point (RAID)", 0}, {0x8c828086, 0x00, "Intel Wildcat Point", 0}, {0x8c838086, 0x00, "Intel Wildcat Point", 0}, {0x8c848086, 0x00, "Intel Wildcat Point (RAID)", 0}, {0x8c858086, 0x00, "Intel Wildcat Point (RAID)", 0}, {0x8c868086, 0x00, "Intel Wildcat Point (RAID)", 0}, {0x8c878086, 0x00, "Intel Wildcat Point (RAID)", 0}, {0x8c8e8086, 0x00, "Intel Wildcat Point (RAID)", 0}, {0x8c8f8086, 0x00, "Intel Wildcat Point (RAID)", 0}, {0x8d028086, 0x00, "Intel Wellsburg", 0}, {0x8d048086, 0x00, "Intel Wellsburg (RAID)", 0}, {0x8d068086, 0x00, "Intel Wellsburg (RAID)", 0}, {0x8d628086, 0x00, "Intel Wellsburg", 0}, {0x8d648086, 0x00, "Intel Wellsburg (RAID)", 0}, {0x8d668086, 0x00, "Intel Wellsburg (RAID)", 0}, {0x8d6e8086, 0x00, "Intel Wellsburg (RAID)", 0}, {0x9c028086, 0x00, "Intel Lynx Point-LP", 0}, {0x9c038086, 0x00, "Intel Lynx Point-LP", 0}, {0x9c048086, 0x00, "Intel Lynx Point-LP (RAID)", 0}, {0x9c058086, 0x00, "Intel Lynx Point-LP (RAID)", 0}, {0x9c068086, 0x00, "Intel Lynx Point-LP (RAID)", 0}, {0x9c078086, 0x00, "Intel Lynx Point-LP (RAID)", 0}, {0x9c0e8086, 0x00, "Intel Lynx Point-LP (RAID)", 0}, {0x9c0f8086, 0x00, "Intel Lynx Point-LP (RAID)", 0}, {0x23238086, 0x00, "Intel DH89xxCC", 0}, {0x2360197b, 0x00, "JMicron JMB360", 0}, {0x2361197b, 0x00, "JMicron JMB361", AHCI_Q_NOFORCE}, {0x2362197b, 0x00, "JMicron JMB362", 0}, {0x2363197b, 0x00, "JMicron JMB363", AHCI_Q_NOFORCE}, {0x2365197b, 0x00, "JMicron JMB365", AHCI_Q_NOFORCE}, {0x2366197b, 0x00, "JMicron JMB366", AHCI_Q_NOFORCE}, {0x2368197b, 0x00, "JMicron JMB368", AHCI_Q_NOFORCE}, {0x611111ab, 0x00, "Marvell 88SE6111", AHCI_Q_NOFORCE | AHCI_Q_1CH | AHCI_Q_EDGEIS}, {0x612111ab, 0x00, "Marvell 88SE6121", AHCI_Q_NOFORCE | AHCI_Q_2CH | AHCI_Q_EDGEIS | AHCI_Q_NONCQ | AHCI_Q_NOCOUNT}, {0x614111ab, 0x00, "Marvell 88SE6141", AHCI_Q_NOFORCE | AHCI_Q_4CH | AHCI_Q_EDGEIS | AHCI_Q_NONCQ | AHCI_Q_NOCOUNT}, {0x614511ab, 0x00, "Marvell 88SE6145", AHCI_Q_NOFORCE | AHCI_Q_4CH | AHCI_Q_EDGEIS | AHCI_Q_NONCQ | AHCI_Q_NOCOUNT}, {0x91201b4b, 0x00, "Marvell 88SE912x", AHCI_Q_EDGEIS}, {0x91231b4b, 0x11, "Marvell 88SE912x", AHCI_Q_ALTSIG}, {0x91231b4b, 0x00, "Marvell 88SE912x", AHCI_Q_EDGEIS|AHCI_Q_SATA2}, {0x91251b4b, 0x00, "Marvell 88SE9125", 0}, {0x91281b4b, 0x00, "Marvell 88SE9128", AHCI_Q_ALTSIG}, {0x91301b4b, 0x00, "Marvell 88SE9130", AHCI_Q_ALTSIG}, {0x91721b4b, 0x00, "Marvell 88SE9172", 0}, {0x91821b4b, 0x00, "Marvell 88SE9182", 0}, {0x91831b4b, 0x00, "Marvell 88SS9183", 0}, {0x91a01b4b, 0x00, "Marvell 88SE91Ax", 0}, {0x92151b4b, 0x00, "Marvell 88SE9215", 0}, {0x92201b4b, 0x00, "Marvell 88SE9220", AHCI_Q_ALTSIG}, {0x92301b4b, 0x00, "Marvell 88SE9230", AHCI_Q_ALTSIG}, {0x92351b4b, 0x00, "Marvell 88SE9235", 0}, {0x06201103, 0x00, "HighPoint RocketRAID 620", 0}, {0x06201b4b, 0x00, "HighPoint RocketRAID 620", 0}, {0x06221103, 0x00, "HighPoint RocketRAID 622", 0}, {0x06221b4b, 0x00, "HighPoint RocketRAID 622", 0}, {0x06401103, 0x00, "HighPoint RocketRAID 640", 0}, {0x06401b4b, 0x00, "HighPoint RocketRAID 640", 0}, {0x06441103, 0x00, "HighPoint RocketRAID 644", 0}, {0x06441b4b, 0x00, "HighPoint RocketRAID 644", 0}, {0x06411103, 0x00, "HighPoint RocketRAID 640L", 0}, {0x06421103, 0x00, "HighPoint RocketRAID 642L", 0}, {0x06451103, 0x00, "HighPoint RocketRAID 644L", 0}, {0x044c10de, 0x00, "NVIDIA MCP65", AHCI_Q_NOAA}, {0x044d10de, 0x00, "NVIDIA MCP65", AHCI_Q_NOAA}, {0x044e10de, 0x00, "NVIDIA MCP65", AHCI_Q_NOAA}, {0x044f10de, 0x00, "NVIDIA MCP65", AHCI_Q_NOAA}, {0x045c10de, 0x00, "NVIDIA MCP65", AHCI_Q_NOAA}, {0x045d10de, 0x00, "NVIDIA MCP65", AHCI_Q_NOAA}, {0x045e10de, 0x00, "NVIDIA MCP65", AHCI_Q_NOAA}, {0x045f10de, 0x00, "NVIDIA MCP65", AHCI_Q_NOAA}, {0x055010de, 0x00, "NVIDIA MCP67", AHCI_Q_NOAA}, {0x055110de, 0x00, "NVIDIA MCP67", AHCI_Q_NOAA}, {0x055210de, 0x00, "NVIDIA MCP67", AHCI_Q_NOAA}, {0x055310de, 0x00, "NVIDIA MCP67", AHCI_Q_NOAA}, {0x055410de, 0x00, "NVIDIA MCP67", AHCI_Q_NOAA}, {0x055510de, 0x00, "NVIDIA MCP67", AHCI_Q_NOAA}, {0x055610de, 0x00, "NVIDIA MCP67", AHCI_Q_NOAA}, {0x055710de, 0x00, "NVIDIA MCP67", AHCI_Q_NOAA}, {0x055810de, 0x00, "NVIDIA MCP67", AHCI_Q_NOAA}, {0x055910de, 0x00, "NVIDIA MCP67", AHCI_Q_NOAA}, {0x055A10de, 0x00, "NVIDIA MCP67", AHCI_Q_NOAA}, {0x055B10de, 0x00, "NVIDIA MCP67", AHCI_Q_NOAA}, {0x058410de, 0x00, "NVIDIA MCP67", AHCI_Q_NOAA}, {0x07f010de, 0x00, "NVIDIA MCP73", AHCI_Q_NOAA}, {0x07f110de, 0x00, "NVIDIA MCP73", AHCI_Q_NOAA}, {0x07f210de, 0x00, "NVIDIA MCP73", AHCI_Q_NOAA}, {0x07f310de, 0x00, "NVIDIA MCP73", AHCI_Q_NOAA}, {0x07f410de, 0x00, "NVIDIA MCP73", AHCI_Q_NOAA}, {0x07f510de, 0x00, "NVIDIA MCP73", AHCI_Q_NOAA}, {0x07f610de, 0x00, "NVIDIA MCP73", AHCI_Q_NOAA}, {0x07f710de, 0x00, "NVIDIA MCP73", AHCI_Q_NOAA}, {0x07f810de, 0x00, "NVIDIA MCP73", AHCI_Q_NOAA}, {0x07f910de, 0x00, "NVIDIA MCP73", AHCI_Q_NOAA}, {0x07fa10de, 0x00, "NVIDIA MCP73", AHCI_Q_NOAA}, {0x07fb10de, 0x00, "NVIDIA MCP73", AHCI_Q_NOAA}, {0x0ad010de, 0x00, "NVIDIA MCP77", AHCI_Q_NOAA}, {0x0ad110de, 0x00, "NVIDIA MCP77", AHCI_Q_NOAA}, {0x0ad210de, 0x00, "NVIDIA MCP77", AHCI_Q_NOAA}, {0x0ad310de, 0x00, "NVIDIA MCP77", AHCI_Q_NOAA}, {0x0ad410de, 0x00, "NVIDIA MCP77", AHCI_Q_NOAA}, {0x0ad510de, 0x00, "NVIDIA MCP77", AHCI_Q_NOAA}, {0x0ad610de, 0x00, "NVIDIA MCP77", AHCI_Q_NOAA}, {0x0ad710de, 0x00, "NVIDIA MCP77", AHCI_Q_NOAA}, {0x0ad810de, 0x00, "NVIDIA MCP77", AHCI_Q_NOAA}, {0x0ad910de, 0x00, "NVIDIA MCP77", AHCI_Q_NOAA}, {0x0ada10de, 0x00, "NVIDIA MCP77", AHCI_Q_NOAA}, {0x0adb10de, 0x00, "NVIDIA MCP77", AHCI_Q_NOAA}, {0x0ab410de, 0x00, "NVIDIA MCP79", AHCI_Q_NOAA}, {0x0ab510de, 0x00, "NVIDIA MCP79", AHCI_Q_NOAA}, {0x0ab610de, 0x00, "NVIDIA MCP79", AHCI_Q_NOAA}, {0x0ab710de, 0x00, "NVIDIA MCP79", AHCI_Q_NOAA}, {0x0ab810de, 0x00, "NVIDIA MCP79", AHCI_Q_NOAA}, {0x0ab910de, 0x00, "NVIDIA MCP79", AHCI_Q_NOAA}, {0x0aba10de, 0x00, "NVIDIA MCP79", AHCI_Q_NOAA}, {0x0abb10de, 0x00, "NVIDIA MCP79", AHCI_Q_NOAA}, {0x0abc10de, 0x00, "NVIDIA MCP79", AHCI_Q_NOAA}, {0x0abd10de, 0x00, "NVIDIA MCP79", AHCI_Q_NOAA}, {0x0abe10de, 0x00, "NVIDIA MCP79", AHCI_Q_NOAA}, {0x0abf10de, 0x00, "NVIDIA MCP79", AHCI_Q_NOAA}, {0x0d8410de, 0x00, "NVIDIA MCP89", AHCI_Q_NOAA}, {0x0d8510de, 0x00, "NVIDIA MCP89", AHCI_Q_NOFORCE|AHCI_Q_NOAA}, {0x0d8610de, 0x00, "NVIDIA MCP89", AHCI_Q_NOAA}, {0x0d8710de, 0x00, "NVIDIA MCP89", AHCI_Q_NOAA}, {0x0d8810de, 0x00, "NVIDIA MCP89", AHCI_Q_NOAA}, {0x0d8910de, 0x00, "NVIDIA MCP89", AHCI_Q_NOAA}, {0x0d8a10de, 0x00, "NVIDIA MCP89", AHCI_Q_NOAA}, {0x0d8b10de, 0x00, "NVIDIA MCP89", AHCI_Q_NOAA}, {0x0d8c10de, 0x00, "NVIDIA MCP89", AHCI_Q_NOAA}, {0x0d8d10de, 0x00, "NVIDIA MCP89", AHCI_Q_NOAA}, {0x0d8e10de, 0x00, "NVIDIA MCP89", AHCI_Q_NOAA}, {0x0d8f10de, 0x00, "NVIDIA MCP89", AHCI_Q_NOAA}, {0x3781105a, 0x00, "Promise TX8660", 0}, {0x33491106, 0x00, "VIA VT8251", AHCI_Q_NOPMP|AHCI_Q_NONCQ}, {0x62871106, 0x00, "VIA VT8251", AHCI_Q_NOPMP|AHCI_Q_NONCQ}, {0x11841039, 0x00, "SiS 966", 0}, {0x11851039, 0x00, "SiS 968", 0}, {0x01861039, 0x00, "SiS 968", 0}, - {0xa01c177d, 0x00, "ThunderX SATA", AHCI_Q_ABAR0}, + {0xa01c177d, 0x00, "ThunderX", AHCI_Q_ABAR0|AHCI_Q_1MSI}, {0x00000000, 0x00, NULL, 0} }; static int ahci_pci_ctlr_reset(device_t dev) { if (pci_read_config(dev, PCIR_DEVVENDOR, 4) == 0x28298086 && (pci_read_config(dev, 0x92, 1) & 0xfe) == 0x04) pci_write_config(dev, 0x92, 0x01, 1); return ahci_ctlr_reset(dev); } static int ahci_probe(device_t dev) { char buf[64]; int i, valid = 0; uint32_t devid = pci_get_devid(dev); uint8_t revid = pci_get_revid(dev); /* * Ensure it is not a PCI bridge (some vendors use * the same PID and VID in PCI bridge and AHCI cards). */ if (pci_get_class(dev) == PCIC_BRIDGE) return (ENXIO); /* Is this a possible AHCI candidate? */ if (pci_get_class(dev) == PCIC_STORAGE && pci_get_subclass(dev) == PCIS_STORAGE_SATA && pci_get_progif(dev) == PCIP_STORAGE_SATA_AHCI_1_0) valid = 1; /* Is this a known AHCI chip? */ for (i = 0; ahci_ids[i].id != 0; i++) { if (ahci_ids[i].id == devid && ahci_ids[i].rev <= revid && (valid || (force_ahci == 1 && !(ahci_ids[i].quirks & AHCI_Q_NOFORCE)))) { /* Do not attach JMicrons with single PCI function. */ if (pci_get_vendor(dev) == 0x197b && (pci_read_config(dev, 0xdf, 1) & 0x40) == 0) return (ENXIO); snprintf(buf, sizeof(buf), "%s AHCI SATA controller", ahci_ids[i].name); device_set_desc_copy(dev, buf); return (BUS_PROBE_VENDOR); } } if (!valid) return (ENXIO); device_set_desc_copy(dev, "AHCI SATA controller"); return (BUS_PROBE_VENDOR); } static int ahci_ata_probe(device_t dev) { char buf[64]; int i; uint32_t devid = pci_get_devid(dev); uint8_t revid = pci_get_revid(dev); if ((intptr_t)device_get_ivars(dev) >= 0) return (ENXIO); /* Is this a known AHCI chip? */ for (i = 0; ahci_ids[i].id != 0; i++) { if (ahci_ids[i].id == devid && ahci_ids[i].rev <= revid) { snprintf(buf, sizeof(buf), "%s AHCI SATA controller", ahci_ids[i].name); device_set_desc_copy(dev, buf); return (BUS_PROBE_VENDOR); } } device_set_desc_copy(dev, "AHCI SATA controller"); return (BUS_PROBE_VENDOR); } static int ahci_pci_attach(device_t dev) { struct ahci_controller *ctlr = device_get_softc(dev); int error, i; uint32_t devid = pci_get_devid(dev); uint8_t revid = pci_get_revid(dev); i = 0; while (ahci_ids[i].id != 0 && (ahci_ids[i].id != devid || ahci_ids[i].rev > revid)) i++; ctlr->quirks = ahci_ids[i].quirks; /* Limit speed for my onboard JMicron external port. * It is not eSATA really, limit to SATA 1 */ if (pci_get_devid(dev) == 0x2363197b && pci_get_subvendor(dev) == 0x1043 && pci_get_subdevice(dev) == 0x81e4) ctlr->quirks |= AHCI_Q_SATA1_UNIT0; ctlr->vendorid = pci_get_vendor(dev); ctlr->deviceid = pci_get_device(dev); ctlr->subvendorid = pci_get_subvendor(dev); ctlr->subdeviceid = pci_get_subdevice(dev); /* Default AHCI Base Address is BAR(5), Cavium uses BAR(0) */ if (ctlr->quirks & AHCI_Q_ABAR0) ctlr->r_rid = PCIR_BAR(0); else ctlr->r_rid = PCIR_BAR(5); if (!(ctlr->r_mem = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &ctlr->r_rid, RF_ACTIVE))) return ENXIO; pci_enable_busmaster(dev); /* Reset controller */ if ((error = ahci_pci_ctlr_reset(dev)) != 0) { bus_release_resource(dev, SYS_RES_MEMORY, ctlr->r_rid, ctlr->r_mem); return (error); }; /* Setup interrupts. */ /* Setup MSI register parameters */ /* Process hints. */ if (ctlr->quirks & AHCI_Q_NOMSI) ctlr->msi = 0; else if (ctlr->quirks & AHCI_Q_1MSI) ctlr->msi = 1; else ctlr->msi = 2; resource_int_value(device_get_name(dev), device_get_unit(dev), "msi", &ctlr->msi); ctlr->numirqs = 1; if (ctlr->msi < 0) ctlr->msi = 0; else if (ctlr->msi == 1) ctlr->msi = min(1, pci_msi_count(dev)); else if (ctlr->msi > 1) { ctlr->msi = 2; ctlr->numirqs = pci_msi_count(dev); } /* Allocate MSI if needed/present. */ if (ctlr->msi && pci_alloc_msi(dev, &ctlr->numirqs) != 0) { ctlr->msi = 0; ctlr->numirqs = 1; } error = ahci_attach(dev); if (error != 0) if (ctlr->msi) pci_release_msi(dev); return error; } static int ahci_pci_detach(device_t dev) { ahci_detach(dev); pci_release_msi(dev); return (0); } static int ahci_pci_suspend(device_t dev) { struct ahci_controller *ctlr = device_get_softc(dev); bus_generic_suspend(dev); /* Disable interupts, so the state change(s) doesn't trigger */ ATA_OUTL(ctlr->r_mem, AHCI_GHC, ATA_INL(ctlr->r_mem, AHCI_GHC) & (~AHCI_GHC_IE)); return 0; } static int ahci_pci_resume(device_t dev) { int res; if ((res = ahci_pci_ctlr_reset(dev)) != 0) return (res); ahci_ctlr_setup(dev); return (bus_generic_resume(dev)); } devclass_t ahci_devclass; static device_method_t ahci_methods[] = { DEVMETHOD(device_probe, ahci_probe), DEVMETHOD(device_attach, ahci_pci_attach), DEVMETHOD(device_detach, ahci_pci_detach), DEVMETHOD(device_suspend, ahci_pci_suspend), DEVMETHOD(device_resume, ahci_pci_resume), DEVMETHOD(bus_print_child, ahci_print_child), DEVMETHOD(bus_alloc_resource, ahci_alloc_resource), DEVMETHOD(bus_release_resource, ahci_release_resource), DEVMETHOD(bus_setup_intr, ahci_setup_intr), DEVMETHOD(bus_teardown_intr,ahci_teardown_intr), DEVMETHOD(bus_child_location_str, ahci_child_location_str), DEVMETHOD(bus_get_dma_tag, ahci_get_dma_tag), DEVMETHOD_END }; static driver_t ahci_driver = { "ahci", ahci_methods, sizeof(struct ahci_controller) }; DRIVER_MODULE(ahci, pci, ahci_driver, ahci_devclass, NULL, NULL); static device_method_t ahci_ata_methods[] = { DEVMETHOD(device_probe, ahci_ata_probe), DEVMETHOD(device_attach, ahci_pci_attach), DEVMETHOD(device_detach, ahci_pci_detach), DEVMETHOD(device_suspend, ahci_pci_suspend), DEVMETHOD(device_resume, ahci_pci_resume), DEVMETHOD(bus_print_child, ahci_print_child), DEVMETHOD(bus_alloc_resource, ahci_alloc_resource), DEVMETHOD(bus_release_resource, ahci_release_resource), DEVMETHOD(bus_setup_intr, ahci_setup_intr), DEVMETHOD(bus_teardown_intr,ahci_teardown_intr), DEVMETHOD(bus_child_location_str, ahci_child_location_str), DEVMETHOD_END }; static driver_t ahci_ata_driver = { "ahci", ahci_ata_methods, sizeof(struct ahci_controller) }; DRIVER_MODULE(ahci, atapci, ahci_ata_driver, ahci_devclass, NULL, NULL); Index: projects/cxl_iscsi/sys/dev/cxgbe/t4_netmap.c =================================================================== --- projects/cxl_iscsi/sys/dev/cxgbe/t4_netmap.c (revision 279716) +++ projects/cxl_iscsi/sys/dev/cxgbe/t4_netmap.c (revision 279717) @@ -1,1182 +1,1230 @@ /*- * Copyright (c) 2014 Chelsio Communications, Inc. * All rights reserved. * Written by: Navdeep Parhar * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #ifdef DEV_NETMAP #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "common/common.h" #include "common/t4_regs.h" #include "common/t4_regs_values.h" extern int fl_pad; /* XXXNM */ extern int spg_len; /* XXXNM */ extern int fl_pktshift; /* XXXNM */ +SYSCTL_NODE(_hw, OID_AUTO, cxgbe, CTLFLAG_RD, 0, "cxgbe netmap parameters"); + +/* + * 0 = normal netmap rx + * 1 = black hole + * 2 = supermassive black hole (buffer packing enabled) + */ +int black_hole = 0; +SYSCTL_INT(_hw_cxgbe, OID_AUTO, nm_black_hole, CTLFLAG_RDTUN, &black_hole, 0, + "Sink incoming packets."); + +int rx_ndesc = 256; +SYSCTL_INT(_hw_cxgbe, OID_AUTO, nm_rx_ndesc, CTLFLAG_RWTUN, + &rx_ndesc, 0, "# of rx descriptors after which the hw cidx is updated."); + +int holdoff_tmr_idx = 2; +SYSCTL_INT(_hw_cxgbe, OID_AUTO, nm_holdoff_tmr_idx, CTLFLAG_RWTUN, + &holdoff_tmr_idx, 0, "Holdoff timer index for netmap rx queues."); + /* netmap ifnet routines */ static void cxgbe_nm_init(void *); static int cxgbe_nm_ioctl(struct ifnet *, unsigned long, caddr_t); static int cxgbe_nm_transmit(struct ifnet *, struct mbuf *); static void cxgbe_nm_qflush(struct ifnet *); static int cxgbe_nm_init_synchronized(struct port_info *); static int cxgbe_nm_uninit_synchronized(struct port_info *); static void cxgbe_nm_init(void *arg) { struct port_info *pi = arg; struct adapter *sc = pi->adapter; if (begin_synchronized_op(sc, pi, SLEEP_OK | INTR_OK, "t4nminit") != 0) return; cxgbe_nm_init_synchronized(pi); end_synchronized_op(sc, 0); return; } static int cxgbe_nm_init_synchronized(struct port_info *pi) { struct adapter *sc = pi->adapter; struct ifnet *ifp = pi->nm_ifp; int rc = 0; ASSERT_SYNCHRONIZED_OP(sc); if (ifp->if_drv_flags & IFF_DRV_RUNNING) return (0); /* already running */ if (!(sc->flags & FULL_INIT_DONE) && ((rc = adapter_full_init(sc)) != 0)) return (rc); /* error message displayed already */ if (!(pi->flags & PORT_INIT_DONE) && ((rc = port_full_init(pi)) != 0)) return (rc); /* error message displayed already */ rc = update_mac_settings(ifp, XGMAC_ALL); if (rc) return (rc); /* error message displayed already */ ifp->if_drv_flags |= IFF_DRV_RUNNING; return (rc); } static int cxgbe_nm_uninit_synchronized(struct port_info *pi) { #ifdef INVARIANTS struct adapter *sc = pi->adapter; #endif struct ifnet *ifp = pi->nm_ifp; ASSERT_SYNCHRONIZED_OP(sc); ifp->if_drv_flags &= ~IFF_DRV_RUNNING; return (0); } static int cxgbe_nm_ioctl(struct ifnet *ifp, unsigned long cmd, caddr_t data) { int rc = 0, mtu, flags; struct port_info *pi = ifp->if_softc; struct adapter *sc = pi->adapter; struct ifreq *ifr = (struct ifreq *)data; uint32_t mask; MPASS(pi->nm_ifp == ifp); switch (cmd) { case SIOCSIFMTU: mtu = ifr->ifr_mtu; if ((mtu < ETHERMIN) || (mtu > ETHERMTU_JUMBO)) return (EINVAL); rc = begin_synchronized_op(sc, pi, SLEEP_OK | INTR_OK, "t4nmtu"); if (rc) return (rc); ifp->if_mtu = mtu; if (ifp->if_drv_flags & IFF_DRV_RUNNING) rc = update_mac_settings(ifp, XGMAC_MTU); end_synchronized_op(sc, 0); break; case SIOCSIFFLAGS: rc = begin_synchronized_op(sc, pi, SLEEP_OK | INTR_OK, "t4nflg"); if (rc) return (rc); if (ifp->if_flags & IFF_UP) { if (ifp->if_drv_flags & IFF_DRV_RUNNING) { flags = pi->nmif_flags; if ((ifp->if_flags ^ flags) & (IFF_PROMISC | IFF_ALLMULTI)) { rc = update_mac_settings(ifp, XGMAC_PROMISC | XGMAC_ALLMULTI); } } else rc = cxgbe_nm_init_synchronized(pi); pi->nmif_flags = ifp->if_flags; } else if (ifp->if_drv_flags & IFF_DRV_RUNNING) rc = cxgbe_nm_uninit_synchronized(pi); end_synchronized_op(sc, 0); break; case SIOCADDMULTI: case SIOCDELMULTI: /* these two are called with a mutex held :-( */ rc = begin_synchronized_op(sc, pi, HOLD_LOCK, "t4nmulti"); if (rc) return (rc); if (ifp->if_drv_flags & IFF_DRV_RUNNING) rc = update_mac_settings(ifp, XGMAC_MCADDRS); end_synchronized_op(sc, LOCK_HELD); break; case SIOCSIFCAP: mask = ifr->ifr_reqcap ^ ifp->if_capenable; if (mask & IFCAP_TXCSUM) { ifp->if_capenable ^= IFCAP_TXCSUM; ifp->if_hwassist ^= (CSUM_TCP | CSUM_UDP | CSUM_IP); } if (mask & IFCAP_TXCSUM_IPV6) { ifp->if_capenable ^= IFCAP_TXCSUM_IPV6; ifp->if_hwassist ^= (CSUM_UDP_IPV6 | CSUM_TCP_IPV6); } if (mask & IFCAP_RXCSUM) ifp->if_capenable ^= IFCAP_RXCSUM; if (mask & IFCAP_RXCSUM_IPV6) ifp->if_capenable ^= IFCAP_RXCSUM_IPV6; break; case SIOCSIFMEDIA: case SIOCGIFMEDIA: ifmedia_ioctl(ifp, ifr, &pi->nm_media, cmd); break; default: rc = ether_ioctl(ifp, cmd, data); } return (rc); } static int cxgbe_nm_transmit(struct ifnet *ifp, struct mbuf *m) { m_freem(m); return (0); } static void cxgbe_nm_qflush(struct ifnet *ifp) { return; } static int alloc_nm_rxq_hwq(struct port_info *pi, struct sge_nm_rxq *nm_rxq, int cong) { int rc, cntxt_id, i; __be32 v; struct adapter *sc = pi->adapter; struct netmap_adapter *na = NA(pi->nm_ifp); struct fw_iq_cmd c; MPASS(na != NULL); MPASS(nm_rxq->iq_desc != NULL); MPASS(nm_rxq->fl_desc != NULL); bzero(nm_rxq->iq_desc, pi->qsize_rxq * IQ_ESIZE); bzero(nm_rxq->fl_desc, na->num_rx_desc * EQ_ESIZE + spg_len); bzero(&c, sizeof(c)); c.op_to_vfn = htobe32(V_FW_CMD_OP(FW_IQ_CMD) | F_FW_CMD_REQUEST | F_FW_CMD_WRITE | F_FW_CMD_EXEC | V_FW_IQ_CMD_PFN(sc->pf) | V_FW_IQ_CMD_VFN(0)); c.alloc_to_len16 = htobe32(F_FW_IQ_CMD_ALLOC | F_FW_IQ_CMD_IQSTART | FW_LEN16(c)); if (pi->flags & INTR_NM_RXQ) { KASSERT(nm_rxq->intr_idx < sc->intr_count, ("%s: invalid direct intr_idx %d", __func__, nm_rxq->intr_idx)); v = V_FW_IQ_CMD_IQANDSTINDEX(nm_rxq->intr_idx); } else { CXGBE_UNIMPLEMENTED(__func__); /* XXXNM: needs review */ v = V_FW_IQ_CMD_IQANDSTINDEX(nm_rxq->intr_idx) | F_FW_IQ_CMD_IQANDST; } c.type_to_iqandstindex = htobe32(v | V_FW_IQ_CMD_TYPE(FW_IQ_TYPE_FL_INT_CAP) | V_FW_IQ_CMD_VIID(pi->nm_viid) | V_FW_IQ_CMD_IQANUD(X_UPDATEDELIVERY_INTERRUPT)); c.iqdroprss_to_iqesize = htobe16(V_FW_IQ_CMD_IQPCIECH(pi->tx_chan) | F_FW_IQ_CMD_IQGTSMODE | V_FW_IQ_CMD_IQINTCNTTHRESH(0) | V_FW_IQ_CMD_IQESIZE(ilog2(IQ_ESIZE) - 4)); c.iqsize = htobe16(pi->qsize_rxq); c.iqaddr = htobe64(nm_rxq->iq_ba); if (cong >= 0) { c.iqns_to_fl0congen = htobe32(F_FW_IQ_CMD_IQFLINTCONGEN | V_FW_IQ_CMD_FL0CNGCHMAP(cong) | F_FW_IQ_CMD_FL0CONGCIF | F_FW_IQ_CMD_FL0CONGEN); } c.iqns_to_fl0congen |= htobe32(V_FW_IQ_CMD_FL0HOSTFCMODE(X_HOSTFCMODE_NONE) | F_FW_IQ_CMD_FL0FETCHRO | F_FW_IQ_CMD_FL0DATARO | - (fl_pad ? F_FW_IQ_CMD_FL0PADEN : 0)); + (fl_pad ? F_FW_IQ_CMD_FL0PADEN : 0) | + (black_hole == 2 ? F_FW_IQ_CMD_FL0PACKEN : 0)); c.fl0dcaen_to_fl0cidxfthresh = htobe16(V_FW_IQ_CMD_FL0FBMIN(X_FETCHBURSTMIN_64B) | V_FW_IQ_CMD_FL0FBMAX(X_FETCHBURSTMAX_512B)); - c.fl0size = htobe16(na->num_rx_desc + spg_len / EQ_ESIZE); + c.fl0size = htobe16(na->num_rx_desc / 8 + spg_len / EQ_ESIZE); c.fl0addr = htobe64(nm_rxq->fl_ba); rc = -t4_wr_mbox(sc, sc->mbox, &c, sizeof(c), &c); if (rc != 0) { device_printf(sc->dev, "failed to create netmap ingress queue: %d\n", rc); return (rc); } nm_rxq->iq_cidx = 0; MPASS(nm_rxq->iq_sidx == pi->qsize_rxq - spg_len / IQ_ESIZE); nm_rxq->iq_gen = F_RSPD_GEN; nm_rxq->iq_cntxt_id = be16toh(c.iqid); nm_rxq->iq_abs_id = be16toh(c.physiqid); cntxt_id = nm_rxq->iq_cntxt_id - sc->sge.iq_start; if (cntxt_id >= sc->sge.niq) { panic ("%s: nm_rxq->iq_cntxt_id (%d) more than the max (%d)", __func__, cntxt_id, sc->sge.niq - 1); } sc->sge.iqmap[cntxt_id] = (void *)nm_rxq; nm_rxq->fl_cntxt_id = be16toh(c.fl0id); nm_rxq->fl_pidx = nm_rxq->fl_cidx = 0; MPASS(nm_rxq->fl_sidx == na->num_rx_desc); cntxt_id = nm_rxq->fl_cntxt_id - sc->sge.eq_start; if (cntxt_id >= sc->sge.neq) { panic("%s: nm_rxq->fl_cntxt_id (%d) more than the max (%d)", __func__, cntxt_id, sc->sge.neq - 1); } sc->sge.eqmap[cntxt_id] = (void *)nm_rxq; nm_rxq->fl_db_val = F_DBPRIO | V_QID(nm_rxq->fl_cntxt_id) | V_PIDX(0); if (is_t5(sc)) nm_rxq->fl_db_val |= F_DBTYPE; if (is_t5(sc) && cong >= 0) { uint32_t param, val; param = V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_DMAQ) | V_FW_PARAMS_PARAM_X(FW_PARAMS_PARAM_DMAQ_CONM_CTXT) | V_FW_PARAMS_PARAM_YZ(nm_rxq->iq_cntxt_id); param = V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_DMAQ) | V_FW_PARAMS_PARAM_X(FW_PARAMS_PARAM_DMAQ_CONM_CTXT) | V_FW_PARAMS_PARAM_YZ(nm_rxq->iq_cntxt_id); if (cong == 0) val = 1 << 19; else { val = 2 << 19; for (i = 0; i < 4; i++) { if (cong & (1 << i)) val |= 1 << (i << 2); } } rc = -t4_set_params(sc, sc->mbox, sc->pf, 0, 1, ¶m, &val); if (rc != 0) { /* report error but carry on */ device_printf(sc->dev, "failed to set congestion manager context for " "ingress queue %d: %d\n", nm_rxq->iq_cntxt_id, rc); } } t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), - V_SEINTARM(V_QINTR_TIMER_IDX(1)) | - V_INGRESSQID(nm_rxq->iq_cntxt_id)); + V_INGRESSQID(nm_rxq->iq_cntxt_id) | + V_SEINTARM(V_QINTR_TIMER_IDX(holdoff_tmr_idx))); return (rc); } static int free_nm_rxq_hwq(struct port_info *pi, struct sge_nm_rxq *nm_rxq) { struct adapter *sc = pi->adapter; int rc; rc = -t4_iq_free(sc, sc->mbox, sc->pf, 0, FW_IQ_TYPE_FL_INT_CAP, nm_rxq->iq_cntxt_id, nm_rxq->fl_cntxt_id, 0xffff); if (rc != 0) device_printf(sc->dev, "%s: failed for iq %d, fl %d: %d\n", __func__, nm_rxq->iq_cntxt_id, nm_rxq->fl_cntxt_id, rc); return (rc); } static int alloc_nm_txq_hwq(struct port_info *pi, struct sge_nm_txq *nm_txq) { int rc, cntxt_id; size_t len; struct adapter *sc = pi->adapter; struct netmap_adapter *na = NA(pi->nm_ifp); struct fw_eq_eth_cmd c; MPASS(na != NULL); MPASS(nm_txq->desc != NULL); len = na->num_tx_desc * EQ_ESIZE + spg_len; bzero(nm_txq->desc, len); bzero(&c, sizeof(c)); c.op_to_vfn = htobe32(V_FW_CMD_OP(FW_EQ_ETH_CMD) | F_FW_CMD_REQUEST | F_FW_CMD_WRITE | F_FW_CMD_EXEC | V_FW_EQ_ETH_CMD_PFN(sc->pf) | V_FW_EQ_ETH_CMD_VFN(0)); c.alloc_to_len16 = htobe32(F_FW_EQ_ETH_CMD_ALLOC | F_FW_EQ_ETH_CMD_EQSTART | FW_LEN16(c)); c.autoequiqe_to_viid = htobe32(F_FW_EQ_ETH_CMD_AUTOEQUEQE | V_FW_EQ_ETH_CMD_VIID(pi->nm_viid)); c.fetchszm_to_iqid = htobe32(V_FW_EQ_ETH_CMD_HOSTFCMODE(X_HOSTFCMODE_NONE) | V_FW_EQ_ETH_CMD_PCIECHN(pi->tx_chan) | F_FW_EQ_ETH_CMD_FETCHRO | V_FW_EQ_ETH_CMD_IQID(sc->sge.nm_rxq[nm_txq->iqidx].iq_cntxt_id)); c.dcaen_to_eqsize = htobe32(V_FW_EQ_ETH_CMD_FBMIN(X_FETCHBURSTMIN_64B) | V_FW_EQ_ETH_CMD_FBMAX(X_FETCHBURSTMAX_512B) | V_FW_EQ_ETH_CMD_EQSIZE(len / EQ_ESIZE)); c.eqaddr = htobe64(nm_txq->ba); rc = -t4_wr_mbox(sc, sc->mbox, &c, sizeof(c), &c); if (rc != 0) { device_printf(pi->dev, "failed to create netmap egress queue: %d\n", rc); return (rc); } nm_txq->cntxt_id = G_FW_EQ_ETH_CMD_EQID(be32toh(c.eqid_pkd)); cntxt_id = nm_txq->cntxt_id - sc->sge.eq_start; if (cntxt_id >= sc->sge.neq) panic("%s: nm_txq->cntxt_id (%d) more than the max (%d)", __func__, cntxt_id, sc->sge.neq - 1); sc->sge.eqmap[cntxt_id] = (void *)nm_txq; nm_txq->pidx = nm_txq->cidx = 0; MPASS(nm_txq->sidx == na->num_tx_desc); nm_txq->equiqidx = nm_txq->equeqidx = nm_txq->dbidx = 0; nm_txq->doorbells = sc->doorbells; if (isset(&nm_txq->doorbells, DOORBELL_UDB) || isset(&nm_txq->doorbells, DOORBELL_UDBWC) || isset(&nm_txq->doorbells, DOORBELL_WCWR)) { uint32_t s_qpp = sc->sge.eq_s_qpp; uint32_t mask = (1 << s_qpp) - 1; volatile uint8_t *udb; udb = sc->udbs_base + UDBS_DB_OFFSET; udb += (nm_txq->cntxt_id >> s_qpp) << PAGE_SHIFT; nm_txq->udb_qid = nm_txq->cntxt_id & mask; if (nm_txq->udb_qid >= PAGE_SIZE / UDBS_SEG_SIZE) clrbit(&nm_txq->doorbells, DOORBELL_WCWR); else { udb += nm_txq->udb_qid << UDBS_SEG_SHIFT; nm_txq->udb_qid = 0; } nm_txq->udb = (volatile void *)udb; } return (rc); } static int free_nm_txq_hwq(struct port_info *pi, struct sge_nm_txq *nm_txq) { struct adapter *sc = pi->adapter; int rc; rc = -t4_eth_eq_free(sc, sc->mbox, sc->pf, 0, nm_txq->cntxt_id); if (rc != 0) device_printf(sc->dev, "%s: failed for eq %d: %d\n", __func__, nm_txq->cntxt_id, rc); return (rc); } static int cxgbe_netmap_on(struct adapter *sc, struct port_info *pi, struct ifnet *ifp, struct netmap_adapter *na) { struct netmap_slot *slot; struct sge_nm_rxq *nm_rxq; struct sge_nm_txq *nm_txq; int rc, i, j, hwidx; struct hw_buf_info *hwb; uint16_t *rss; ASSERT_SYNCHRONIZED_OP(sc); if ((pi->flags & PORT_INIT_DONE) == 0 || (ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) return (EAGAIN); hwb = &sc->sge.hw_buf_info[0]; for (i = 0; i < SGE_FLBUF_SIZES; i++, hwb++) { if (hwb->size == NETMAP_BUF_SIZE(na)) break; } if (i >= SGE_FLBUF_SIZES) { if_printf(ifp, "no hwidx for netmap buffer size %d.\n", NETMAP_BUF_SIZE(na)); return (ENXIO); } hwidx = i; /* Must set caps before calling netmap_reset */ nm_set_native_flags(na); for_each_nm_rxq(pi, i, nm_rxq) { alloc_nm_rxq_hwq(pi, nm_rxq, tnl_cong(pi)); nm_rxq->fl_hwidx = hwidx; slot = netmap_reset(na, NR_RX, i, 0); MPASS(slot != NULL); /* XXXNM: error check, not assert */ /* We deal with 8 bufs at a time */ MPASS((na->num_rx_desc & 7) == 0); MPASS(na->num_rx_desc == nm_rxq->fl_sidx); - for (j = 0; j < nm_rxq->fl_sidx - 8; j++) { + for (j = 0; j < nm_rxq->fl_sidx; j++) { uint64_t ba; PNMB(na, &slot[j], &ba); + MPASS(ba != 0); nm_rxq->fl_desc[j] = htobe64(ba | hwidx); } - nm_rxq->fl_pidx = j; + j = nm_rxq->fl_pidx = nm_rxq->fl_sidx - 8; MPASS((j & 7) == 0); j /= 8; /* driver pidx to hardware pidx */ wmb(); t4_write_reg(sc, MYPF_REG(A_SGE_PF_KDOORBELL), nm_rxq->fl_db_val | V_PIDX(j)); } for_each_nm_txq(pi, i, nm_txq) { alloc_nm_txq_hwq(pi, nm_txq); slot = netmap_reset(na, NR_TX, i, 0); MPASS(slot != NULL); /* XXXNM: error check, not assert */ } rss = malloc(pi->nm_rss_size * sizeof (*rss), M_CXGBE, M_ZERO | M_WAITOK); for (i = 0; i < pi->nm_rss_size;) { for_each_nm_rxq(pi, j, nm_rxq) { rss[i++] = nm_rxq->iq_abs_id; if (i == pi->nm_rss_size) break; } } rc = -t4_config_rss_range(sc, sc->mbox, pi->nm_viid, 0, pi->nm_rss_size, rss, pi->nm_rss_size); if (rc != 0) if_printf(ifp, "netmap rss_config failed: %d\n", rc); free(rss, M_CXGBE); rc = -t4_enable_vi(sc, sc->mbox, pi->nm_viid, true, true); if (rc != 0) if_printf(ifp, "netmap enable_vi failed: %d\n", rc); return (rc); } static int cxgbe_netmap_off(struct adapter *sc, struct port_info *pi, struct ifnet *ifp, struct netmap_adapter *na) { int rc, i; struct sge_nm_txq *nm_txq; struct sge_nm_rxq *nm_rxq; ASSERT_SYNCHRONIZED_OP(sc); rc = -t4_enable_vi(sc, sc->mbox, pi->nm_viid, false, false); if (rc != 0) if_printf(ifp, "netmap disable_vi failed: %d\n", rc); nm_clear_native_flags(na); for_each_nm_txq(pi, i, nm_txq) { struct sge_qstat *spg = (void *)&nm_txq->desc[nm_txq->sidx]; /* Wait for hw pidx to catch up ... */ while (be16toh(nm_txq->pidx) != spg->pidx) pause("nmpidx", 1); /* ... and then for the cidx. */ while (spg->pidx != spg->cidx) pause("nmcidx", 1); free_nm_txq_hwq(pi, nm_txq); } for_each_nm_rxq(pi, i, nm_rxq) { free_nm_rxq_hwq(pi, nm_rxq); } return (rc); } static int cxgbe_netmap_reg(struct netmap_adapter *na, int on) { struct ifnet *ifp = na->ifp; struct port_info *pi = ifp->if_softc; struct adapter *sc = pi->adapter; int rc; rc = begin_synchronized_op(sc, pi, SLEEP_OK | INTR_OK, "t4nmreg"); if (rc != 0) return (rc); if (on) rc = cxgbe_netmap_on(sc, pi, ifp, na); else rc = cxgbe_netmap_off(sc, pi, ifp, na); end_synchronized_op(sc, 0); return (rc); } /* How many packets can a single type1 WR carry in n descriptors */ static inline int ndesc_to_npkt(const int n) { MPASS(n > 0 && n <= SGE_MAX_WR_NDESC); return (n * 2 - 1); } #define MAX_NPKT_IN_TYPE1_WR (ndesc_to_npkt(SGE_MAX_WR_NDESC)) /* Space (in descriptors) needed for a type1 WR that carries n packets */ static inline int npkt_to_ndesc(const int n) { MPASS(n > 0 && n <= MAX_NPKT_IN_TYPE1_WR); return ((n + 2) / 2); } /* Space (in 16B units) needed for a type1 WR that carries n packets */ static inline int npkt_to_len16(const int n) { MPASS(n > 0 && n <= MAX_NPKT_IN_TYPE1_WR); return (n * 2 + 1); } #define NMIDXDIFF(q, idx) IDXDIFF((q)->pidx, (q)->idx, (q)->sidx) static void ring_nm_txq_db(struct adapter *sc, struct sge_nm_txq *nm_txq) { int n; u_int db = nm_txq->doorbells; MPASS(nm_txq->pidx != nm_txq->dbidx); n = NMIDXDIFF(nm_txq, dbidx); if (n > 1) clrbit(&db, DOORBELL_WCWR); wmb(); switch (ffs(db) - 1) { case DOORBELL_UDB: *nm_txq->udb = htole32(V_QID(nm_txq->udb_qid) | V_PIDX(n)); break; case DOORBELL_WCWR: { volatile uint64_t *dst, *src; /* * Queues whose 128B doorbell segment fits in the page do not * use relative qid (udb_qid is always 0). Only queues with * doorbell segments can do WCWR. */ KASSERT(nm_txq->udb_qid == 0 && n == 1, ("%s: inappropriate doorbell (0x%x, %d, %d) for nm_txq %p", __func__, nm_txq->doorbells, n, nm_txq->pidx, nm_txq)); dst = (volatile void *)((uintptr_t)nm_txq->udb + UDBS_WR_OFFSET - UDBS_DB_OFFSET); src = (void *)&nm_txq->desc[nm_txq->dbidx]; while (src != (void *)&nm_txq->desc[nm_txq->dbidx + 1]) *dst++ = *src++; wmb(); break; } case DOORBELL_UDBWC: *nm_txq->udb = htole32(V_QID(nm_txq->udb_qid) | V_PIDX(n)); wmb(); break; case DOORBELL_KDB: t4_write_reg(sc, MYPF_REG(A_SGE_PF_KDOORBELL), V_QID(nm_txq->cntxt_id) | V_PIDX(n)); break; } nm_txq->dbidx = nm_txq->pidx; } int lazy_tx_credit_flush = 1; /* * Write work requests to send 'npkt' frames and ring the doorbell to send them * on their way. No need to check for wraparound. */ static void cxgbe_nm_tx(struct adapter *sc, struct sge_nm_txq *nm_txq, struct netmap_kring *kring, int npkt, int npkt_remaining, int txcsum) { struct netmap_ring *ring = kring->ring; struct netmap_slot *slot; const u_int lim = kring->nkr_num_slots - 1; struct fw_eth_tx_pkts_wr *wr = (void *)&nm_txq->desc[nm_txq->pidx]; uint16_t len; uint64_t ba; struct cpl_tx_pkt_core *cpl; struct ulptx_sgl *usgl; int i, n; while (npkt) { n = min(npkt, MAX_NPKT_IN_TYPE1_WR); len = 0; wr = (void *)&nm_txq->desc[nm_txq->pidx]; wr->op_pkd = htobe32(V_FW_WR_OP(FW_ETH_TX_PKTS_WR)); wr->equiq_to_len16 = htobe32(V_FW_WR_LEN16(npkt_to_len16(n))); wr->npkt = n; wr->r3 = 0; wr->type = 1; cpl = (void *)(wr + 1); for (i = 0; i < n; i++) { slot = &ring->slot[kring->nr_hwcur]; PNMB(kring->na, slot, &ba); + MPASS(ba != 0); cpl->ctrl0 = nm_txq->cpl_ctrl0; cpl->pack = 0; cpl->len = htobe16(slot->len); /* * netmap(4) says "netmap does not use features such as * checksum offloading, TCP segmentation offloading, * encryption, VLAN encapsulation/decapsulation, etc." * * So the ncxl interfaces have tx hardware checksumming * disabled by default. But you can override netmap by * enabling IFCAP_TXCSUM on the interface manully. */ cpl->ctrl1 = txcsum ? 0 : htobe64(F_TXPKT_IPCSUM_DIS | F_TXPKT_L4CSUM_DIS); usgl = (void *)(cpl + 1); usgl->cmd_nsge = htobe32(V_ULPTX_CMD(ULP_TX_SC_DSGL) | V_ULPTX_NSGE(1)); usgl->len0 = htobe32(slot->len); usgl->addr0 = htobe64(ba); slot->flags &= ~(NS_REPORT | NS_BUF_CHANGED); cpl = (void *)(usgl + 1); MPASS(slot->len + len <= UINT16_MAX); len += slot->len; kring->nr_hwcur = nm_next(kring->nr_hwcur, lim); } wr->plen = htobe16(len); npkt -= n; nm_txq->pidx += npkt_to_ndesc(n); MPASS(nm_txq->pidx <= nm_txq->sidx); if (__predict_false(nm_txq->pidx == nm_txq->sidx)) { /* * This routine doesn't know how to write WRs that wrap * around. Make sure it wasn't asked to. */ MPASS(npkt == 0); nm_txq->pidx = 0; } if (npkt == 0 && npkt_remaining == 0) { /* All done. */ if (lazy_tx_credit_flush == 0) { wr->equiq_to_len16 |= htobe32(F_FW_WR_EQUEQ | F_FW_WR_EQUIQ); nm_txq->equeqidx = nm_txq->pidx; nm_txq->equiqidx = nm_txq->pidx; } ring_nm_txq_db(sc, nm_txq); return; } if (NMIDXDIFF(nm_txq, equiqidx) >= nm_txq->sidx / 2) { wr->equiq_to_len16 |= htobe32(F_FW_WR_EQUEQ | F_FW_WR_EQUIQ); nm_txq->equeqidx = nm_txq->pidx; nm_txq->equiqidx = nm_txq->pidx; } else if (NMIDXDIFF(nm_txq, equeqidx) >= 64) { wr->equiq_to_len16 |= htobe32(F_FW_WR_EQUEQ); nm_txq->equeqidx = nm_txq->pidx; } if (NMIDXDIFF(nm_txq, dbidx) >= 2 * SGE_MAX_WR_NDESC) ring_nm_txq_db(sc, nm_txq); } /* Will get called again. */ MPASS(npkt_remaining); } /* How many contiguous free descriptors starting at pidx */ static inline int contiguous_ndesc_available(struct sge_nm_txq *nm_txq) { if (nm_txq->cidx > nm_txq->pidx) return (nm_txq->cidx - nm_txq->pidx - 1); else if (nm_txq->cidx > 0) return (nm_txq->sidx - nm_txq->pidx); else return (nm_txq->sidx - nm_txq->pidx - 1); } static int reclaim_nm_tx_desc(struct sge_nm_txq *nm_txq) { struct sge_qstat *spg = (void *)&nm_txq->desc[nm_txq->sidx]; uint16_t hw_cidx = spg->cidx; /* snapshot */ struct fw_eth_tx_pkts_wr *wr; int n = 0; hw_cidx = be16toh(hw_cidx); while (nm_txq->cidx != hw_cidx) { wr = (void *)&nm_txq->desc[nm_txq->cidx]; MPASS(wr->op_pkd == htobe32(V_FW_WR_OP(FW_ETH_TX_PKTS_WR))); MPASS(wr->type == 1); MPASS(wr->npkt > 0 && wr->npkt <= MAX_NPKT_IN_TYPE1_WR); n += wr->npkt; nm_txq->cidx += npkt_to_ndesc(wr->npkt); /* * We never sent a WR that wrapped around so the credits coming * back, WR by WR, should never cause the cidx to wrap around * either. */ MPASS(nm_txq->cidx <= nm_txq->sidx); if (__predict_false(nm_txq->cidx == nm_txq->sidx)) nm_txq->cidx = 0; } return (n); } static int cxgbe_netmap_txsync(struct netmap_kring *kring, int flags) { struct netmap_adapter *na = kring->na; struct ifnet *ifp = na->ifp; struct port_info *pi = ifp->if_softc; struct adapter *sc = pi->adapter; struct sge_nm_txq *nm_txq = &sc->sge.nm_txq[pi->first_nm_txq + kring->ring_id]; const u_int head = kring->rhead; u_int reclaimed = 0; int n, d, npkt_remaining, ndesc_remaining, txcsum; /* * Tx was at kring->nr_hwcur last time around and now we need to advance * to kring->rhead. Note that the driver's pidx moves independent of * netmap's kring->nr_hwcur (pidx counts descriptors and the relation * between descriptors and frames isn't 1:1). */ npkt_remaining = head >= kring->nr_hwcur ? head - kring->nr_hwcur : kring->nkr_num_slots - kring->nr_hwcur + head; txcsum = ifp->if_capenable & (IFCAP_TXCSUM | IFCAP_TXCSUM_IPV6); while (npkt_remaining) { reclaimed += reclaim_nm_tx_desc(nm_txq); ndesc_remaining = contiguous_ndesc_available(nm_txq); /* Can't run out of descriptors with packets still remaining */ MPASS(ndesc_remaining > 0); /* # of desc needed to tx all remaining packets */ d = (npkt_remaining / MAX_NPKT_IN_TYPE1_WR) * SGE_MAX_WR_NDESC; if (npkt_remaining % MAX_NPKT_IN_TYPE1_WR) d += npkt_to_ndesc(npkt_remaining % MAX_NPKT_IN_TYPE1_WR); if (d <= ndesc_remaining) n = npkt_remaining; else { /* Can't send all, calculate how many can be sent */ n = (ndesc_remaining / SGE_MAX_WR_NDESC) * MAX_NPKT_IN_TYPE1_WR; if (ndesc_remaining % SGE_MAX_WR_NDESC) n += ndesc_to_npkt(ndesc_remaining % SGE_MAX_WR_NDESC); } /* Send n packets and update nm_txq->pidx and kring->nr_hwcur */ npkt_remaining -= n; cxgbe_nm_tx(sc, nm_txq, kring, n, npkt_remaining, txcsum); } MPASS(npkt_remaining == 0); MPASS(kring->nr_hwcur == head); MPASS(nm_txq->dbidx == nm_txq->pidx); /* * Second part: reclaim buffers for completed transmissions. */ if (reclaimed || flags & NAF_FORCE_RECLAIM || nm_kr_txempty(kring)) { reclaimed += reclaim_nm_tx_desc(nm_txq); kring->nr_hwtail += reclaimed; if (kring->nr_hwtail >= kring->nkr_num_slots) kring->nr_hwtail -= kring->nkr_num_slots; } nm_txsync_finalize(kring); return (0); } static int cxgbe_netmap_rxsync(struct netmap_kring *kring, int flags) { struct netmap_adapter *na = kring->na; struct netmap_ring *ring = kring->ring; struct ifnet *ifp = na->ifp; struct port_info *pi = ifp->if_softc; struct adapter *sc = pi->adapter; struct sge_nm_rxq *nm_rxq = &sc->sge.nm_rxq[pi->first_nm_rxq + kring->ring_id]; u_int const head = nm_rxsync_prologue(kring); u_int n; int force_update = (flags & NAF_FORCE_READ) || kring->nr_kflags & NKR_PENDINTR; + if (black_hole) + return (0); /* No updates ever. */ + if (netmap_no_pendintr || force_update) { kring->nr_hwtail = atomic_load_acq_32(&nm_rxq->fl_cidx); kring->nr_kflags &= ~NKR_PENDINTR; } /* Userspace done with buffers from kring->nr_hwcur to head */ n = head >= kring->nr_hwcur ? head - kring->nr_hwcur : kring->nkr_num_slots - kring->nr_hwcur + head; n &= ~7U; if (n > 0) { u_int fl_pidx = nm_rxq->fl_pidx; struct netmap_slot *slot = &ring->slot[fl_pidx]; uint64_t ba; int i, dbinc = 0, hwidx = nm_rxq->fl_hwidx; /* * We always deal with 8 buffers at a time. We must have * stopped at an 8B boundary (fl_pidx) last time around and we * must have a multiple of 8B buffers to give to the freelist. */ MPASS((fl_pidx & 7) == 0); MPASS((n & 7) == 0); IDXINCR(kring->nr_hwcur, n, kring->nkr_num_slots); IDXINCR(nm_rxq->fl_pidx, n, nm_rxq->fl_sidx); while (n > 0) { for (i = 0; i < 8; i++, fl_pidx++, slot++) { PNMB(na, slot, &ba); + MPASS(ba != 0); nm_rxq->fl_desc[fl_pidx] = htobe64(ba | hwidx); slot->flags &= ~NS_BUF_CHANGED; MPASS(fl_pidx <= nm_rxq->fl_sidx); } n -= 8; if (fl_pidx == nm_rxq->fl_sidx) { fl_pidx = 0; slot = &ring->slot[0]; } if (++dbinc == 8 && n >= 32) { wmb(); t4_write_reg(sc, MYPF_REG(A_SGE_PF_KDOORBELL), nm_rxq->fl_db_val | V_PIDX(dbinc)); dbinc = 0; } } MPASS(nm_rxq->fl_pidx == fl_pidx); if (dbinc > 0) { wmb(); t4_write_reg(sc, MYPF_REG(A_SGE_PF_KDOORBELL), nm_rxq->fl_db_val | V_PIDX(dbinc)); } } nm_rxsync_finalize(kring); return (0); } /* * Create an ifnet solely for netmap use and register it with the kernel. */ int create_netmap_ifnet(struct port_info *pi) { struct adapter *sc = pi->adapter; struct netmap_adapter na; struct ifnet *ifp; device_t dev = pi->dev; uint8_t mac[ETHER_ADDR_LEN]; int rc; if (pi->nnmtxq <= 0 || pi->nnmrxq <= 0) return (0); MPASS(pi->nm_ifp == NULL); /* * Allocate a virtual interface exclusively for netmap use. Give it the * MAC address normally reserved for use by a TOE interface. (The TOE * driver on FreeBSD doesn't use it). */ rc = t4_alloc_vi_func(sc, sc->mbox, pi->tx_chan, sc->pf, 0, 1, &mac[0], &pi->nm_rss_size, FW_VI_FUNC_OFLD, 0); if (rc < 0) { device_printf(dev, "unable to allocate netmap virtual " "interface for port %d: %d\n", pi->port_id, -rc); return (-rc); } pi->nm_viid = rc; pi->nm_xact_addr_filt = -1; ifp = if_alloc(IFT_ETHER); if (ifp == NULL) { device_printf(dev, "Cannot allocate netmap ifnet\n"); return (ENOMEM); } pi->nm_ifp = ifp; ifp->if_softc = pi; if_initname(ifp, is_t4(pi->adapter) ? "ncxgbe" : "ncxl", device_get_unit(dev)); ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST; ifp->if_init = cxgbe_nm_init; ifp->if_ioctl = cxgbe_nm_ioctl; ifp->if_transmit = cxgbe_nm_transmit; ifp->if_qflush = cxgbe_nm_qflush; /* * netmap(4) says "netmap does not use features such as checksum * offloading, TCP segmentation offloading, encryption, VLAN * encapsulation/decapsulation, etc." * * By default we comply with the statement above. But we do declare the * ifnet capable of L3/L4 checksumming so that a user can override * netmap and have the hardware do the L3/L4 checksums. */ ifp->if_capabilities = IFCAP_HWCSUM | IFCAP_JUMBO_MTU | IFCAP_HWCSUM_IPV6; ifp->if_capenable = 0; ifp->if_hwassist = 0; /* nm_media has already been setup by the caller */ ether_ifattach(ifp, mac); /* * Register with netmap in the kernel. */ bzero(&na, sizeof(na)); na.ifp = pi->nm_ifp; na.na_flags = NAF_BDG_MAYSLEEP; /* Netmap doesn't know about the space reserved for the status page. */ na.num_tx_desc = pi->qsize_txq - spg_len / EQ_ESIZE; /* * The freelist's cidx/pidx drives netmap's rx cidx/pidx. So * num_rx_desc is based on the number of buffers that can be held in the * freelist, and not the number of entries in the iq. (These two are * not exactly the same due to the space taken up by the status page). */ na.num_rx_desc = (pi->qsize_rxq / 8) * 8; na.nm_txsync = cxgbe_netmap_txsync; na.nm_rxsync = cxgbe_netmap_rxsync; na.nm_register = cxgbe_netmap_reg; na.num_tx_rings = pi->nnmtxq; na.num_rx_rings = pi->nnmrxq; netmap_attach(&na); /* This adds IFCAP_NETMAP to if_capabilities */ return (0); } int destroy_netmap_ifnet(struct port_info *pi) { struct adapter *sc = pi->adapter; if (pi->nm_ifp == NULL) return (0); netmap_detach(pi->nm_ifp); ifmedia_removeall(&pi->nm_media); ether_ifdetach(pi->nm_ifp); if_free(pi->nm_ifp); t4_free_vi(sc, sc->mbox, sc->pf, 0, pi->nm_viid); return (0); } static void handle_nm_fw6_msg(struct adapter *sc, struct ifnet *ifp, const struct cpl_fw6_msg *cpl) { const struct cpl_sge_egr_update *egr; uint32_t oq; struct sge_nm_txq *nm_txq; if (cpl->type != FW_TYPE_RSSCPL && cpl->type != FW6_TYPE_RSSCPL) panic("%s: FW_TYPE 0x%x on nm_rxq.", __func__, cpl->type); /* data[0] is RSS header */ egr = (const void *)&cpl->data[1]; oq = be32toh(egr->opcode_qid); MPASS(G_CPL_OPCODE(oq) == CPL_SGE_EGR_UPDATE); nm_txq = (void *)sc->sge.eqmap[G_EGR_QID(oq) - sc->sge.eq_start]; netmap_tx_irq(ifp, nm_txq->nid); } void t4_nm_intr(void *arg) { struct sge_nm_rxq *nm_rxq = arg; struct port_info *pi = nm_rxq->pi; struct adapter *sc = pi->adapter; struct ifnet *ifp = pi->nm_ifp; struct netmap_adapter *na = NA(ifp); struct netmap_kring *kring = &na->rx_rings[nm_rxq->nid]; struct netmap_ring *ring = kring->ring; struct iq_desc *d = &nm_rxq->iq_desc[nm_rxq->iq_cidx]; uint32_t lq; - u_int n = 0; - int processed = 0; + u_int n = 0, work = 0; uint8_t opcode; uint32_t fl_cidx = atomic_load_acq_32(&nm_rxq->fl_cidx); + u_int fl_credits = fl_cidx & 7; while ((d->rsp.u.type_gen & F_RSPD_GEN) == nm_rxq->iq_gen) { rmb(); lq = be32toh(d->rsp.pldbuflen_qid); opcode = d->rss.opcode; switch (G_RSPD_TYPE(d->rsp.u.type_gen)) { case X_RSPD_TYPE_FLBUF: - /* No buffer packing so new buf every time */ - MPASS(lq & F_RSPD_NEWBUF); + if (black_hole != 2) { + /* No buffer packing so new buf every time */ + MPASS(lq & F_RSPD_NEWBUF); + } /* fall through */ case X_RSPD_TYPE_CPL: MPASS(opcode < NUM_CPL_CMDS); switch (opcode) { case CPL_FW4_MSG: case CPL_FW6_MSG: handle_nm_fw6_msg(sc, ifp, (const void *)&d->cpl[0]); break; case CPL_RX_PKT: ring->slot[fl_cidx].len = G_RSPD_LEN(lq) - fl_pktshift; ring->slot[fl_cidx].flags = kring->nkr_slot_flags; - if (__predict_false(++fl_cidx == nm_rxq->fl_sidx)) + fl_cidx += (lq & F_RSPD_NEWBUF) ? 1 : 0; + fl_credits += (lq & F_RSPD_NEWBUF) ? 1 : 0; + if (__predict_false(fl_cidx == nm_rxq->fl_sidx)) fl_cidx = 0; break; default: panic("%s: unexpected opcode 0x%x on nm_rxq %p", __func__, opcode, nm_rxq); } break; case X_RSPD_TYPE_INTR: /* Not equipped to handle forwarded interrupts. */ panic("%s: netmap queue received interrupt for iq %u\n", __func__, lq); default: panic("%s: illegal response type %d on nm_rxq %p", __func__, G_RSPD_TYPE(d->rsp.u.type_gen), nm_rxq); } d++; if (__predict_false(++nm_rxq->iq_cidx == nm_rxq->iq_sidx)) { nm_rxq->iq_cidx = 0; d = &nm_rxq->iq_desc[0]; nm_rxq->iq_gen ^= F_RSPD_GEN; } - if (__predict_false(++n == 64)) { /* XXXNM: tune */ + if (__predict_false(++n == rx_ndesc)) { + atomic_store_rel_32(&nm_rxq->fl_cidx, fl_cidx); + if (black_hole && fl_credits >= 8) { + fl_credits /= 8; + IDXINCR(nm_rxq->fl_pidx, fl_credits * 8, + nm_rxq->fl_sidx); + t4_write_reg(sc, MYPF_REG(A_SGE_PF_KDOORBELL), + nm_rxq->fl_db_val | V_PIDX(fl_credits)); + fl_credits = fl_cidx & 7; + } else if (!black_hole) { + netmap_rx_irq(ifp, nm_rxq->nid, &work); + MPASS(work != 0); + } t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(n) | V_INGRESSQID(nm_rxq->iq_cntxt_id) | V_SEINTARM(V_QINTR_TIMER_IDX(X_TIMERREG_UPDATE_CIDX))); n = 0; } } - if (fl_cidx != nm_rxq->fl_cidx) { - atomic_store_rel_32(&nm_rxq->fl_cidx, fl_cidx); - netmap_rx_irq(ifp, nm_rxq->nid, &processed); - } + + atomic_store_rel_32(&nm_rxq->fl_cidx, fl_cidx); + if (black_hole) { + fl_credits /= 8; + IDXINCR(nm_rxq->fl_pidx, fl_credits * 8, nm_rxq->fl_sidx); + t4_write_reg(sc, MYPF_REG(A_SGE_PF_KDOORBELL), + nm_rxq->fl_db_val | V_PIDX(fl_credits)); + } else + netmap_rx_irq(ifp, nm_rxq->nid, &work); + t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(n) | V_INGRESSQID((u32)nm_rxq->iq_cntxt_id) | - V_SEINTARM(V_QINTR_TIMER_IDX(1))); + V_SEINTARM(V_QINTR_TIMER_IDX(holdoff_tmr_idx))); } #endif Index: projects/cxl_iscsi/sys/dev/drm2/drm_irq.c =================================================================== --- projects/cxl_iscsi/sys/dev/drm2/drm_irq.c (revision 279716) +++ projects/cxl_iscsi/sys/dev/drm2/drm_irq.c (revision 279717) @@ -1,1249 +1,1252 @@ /*- * Copyright 2003 Eric Anholt * All Rights Reserved. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), * to deal in the Software without restriction, including without limitation * the rights to use, copy, modify, merge, publish, distribute, sublicense, * and/or sell copies of the Software, and to permit persons to whom the * Software is furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice (including the next * paragraph) shall be included in all copies or substantial portions of the * Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL * ERIC ANHOLT BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER * IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. * * Authors: * Eric Anholt * */ #include __FBSDID("$FreeBSD$"); /** @file drm_irq.c * Support code for handling setup/teardown of interrupt handlers and * handing interrupt handlers off to the drivers. */ #include #include MALLOC_DEFINE(DRM_MEM_VBLANK, "drm_vblank", "DRM VBLANK Handling Data"); /* Access macro for slots in vblank timestamp ringbuffer. */ #define vblanktimestamp(dev, crtc, count) ( \ (dev)->_vblank_time[(crtc) * DRM_VBLANKTIME_RBSIZE + \ ((count) % DRM_VBLANKTIME_RBSIZE)]) /* Retry timestamp calculation up to 3 times to satisfy * drm_timestamp_precision before giving up. */ #define DRM_TIMESTAMP_MAXRETRIES 3 /* Threshold in nanoseconds for detection of redundant * vblank irq in drm_handle_vblank(). 1 msec should be ok. */ #define DRM_REDUNDANT_VBLIRQ_THRESH_NS 1000000 int drm_irq_by_busid(struct drm_device *dev, void *data, struct drm_file *file_priv) { struct drm_irq_busid *irq = data; if ((irq->busnum >> 8) != dev->pci_domain || (irq->busnum & 0xff) != dev->pci_bus || irq->devnum != dev->pci_slot || irq->funcnum != dev->pci_func) return EINVAL; irq->irq = dev->irq; DRM_DEBUG("%d:%d:%d => IRQ %d\n", irq->busnum, irq->devnum, irq->funcnum, irq->irq); return 0; } static void drm_irq_handler_wrap(void *arg) { struct drm_device *dev = arg; mtx_lock(&dev->irq_lock); dev->driver->irq_handler(arg); mtx_unlock(&dev->irq_lock); } int drm_irq_install(struct drm_device *dev) { int retcode; if (dev->irq == 0 || dev->dev_private == NULL) return (EINVAL); DRM_DEBUG("irq=%d\n", dev->irq); DRM_LOCK(dev); if (dev->irq_enabled) { DRM_UNLOCK(dev); return EBUSY; } dev->irq_enabled = 1; dev->context_flag = 0; /* Before installing handler */ if (dev->driver->irq_preinstall) dev->driver->irq_preinstall(dev); DRM_UNLOCK(dev); /* Install handler */ retcode = bus_setup_intr(dev->device, dev->irqr, INTR_TYPE_TTY | INTR_MPSAFE, NULL, (dev->driver->driver_features & DRIVER_LOCKLESS_IRQ) != 0 ? drm_irq_handler_wrap : dev->driver->irq_handler, dev, &dev->irqh); if (retcode != 0) goto err; /* After installing handler */ DRM_LOCK(dev); if (dev->driver->irq_postinstall) dev->driver->irq_postinstall(dev); DRM_UNLOCK(dev); return (0); err: device_printf(dev->device, "Error setting interrupt: %d\n", retcode); dev->irq_enabled = 0; return (retcode); } int drm_irq_uninstall(struct drm_device *dev) { int i; if (!dev->irq_enabled) return EINVAL; dev->irq_enabled = 0; /* * Wake up any waiters so they don't hang. */ if (dev->num_crtcs) { mtx_lock(&dev->vbl_lock); for (i = 0; i < dev->num_crtcs; i++) { wakeup(&dev->_vblank_count[i]); dev->vblank_enabled[i] = 0; dev->last_vblank[i] = dev->driver->get_vblank_counter(dev, i); } mtx_unlock(&dev->vbl_lock); } DRM_DEBUG("irq=%d\n", dev->irq); if (dev->driver->irq_uninstall) dev->driver->irq_uninstall(dev); DRM_UNLOCK(dev); bus_teardown_intr(dev->device, dev->irqr, dev->irqh); DRM_LOCK(dev); return 0; } int drm_control(struct drm_device *dev, void *data, struct drm_file *file_priv) { struct drm_control *ctl = data; int err; switch (ctl->func) { case DRM_INST_HANDLER: /* Handle drivers whose DRM used to require IRQ setup but the * no longer does. */ if (!drm_core_check_feature(dev, DRIVER_HAVE_IRQ)) return 0; if (drm_core_check_feature(dev, DRIVER_MODESET)) return 0; if (dev->if_version < DRM_IF_VERSION(1, 2) && ctl->irq != dev->irq) return EINVAL; return drm_irq_install(dev); case DRM_UNINST_HANDLER: if (!drm_core_check_feature(dev, DRIVER_HAVE_IRQ)) return 0; if (drm_core_check_feature(dev, DRIVER_MODESET)) return 0; DRM_LOCK(dev); err = drm_irq_uninstall(dev); DRM_UNLOCK(dev); return err; default: return EINVAL; } } #define NSEC_PER_USEC 1000L #define NSEC_PER_SEC 1000000000L int64_t timeval_to_ns(const struct timeval *tv) { return ((int64_t)tv->tv_sec * NSEC_PER_SEC) + tv->tv_usec * NSEC_PER_USEC; } struct timeval ns_to_timeval(const int64_t nsec) { struct timeval tv; long rem; if (nsec == 0) { tv.tv_sec = 0; tv.tv_usec = 0; return (tv); } tv.tv_sec = nsec / NSEC_PER_SEC; rem = nsec % NSEC_PER_SEC; if (rem < 0) { tv.tv_sec--; rem += NSEC_PER_SEC; } tv.tv_usec = rem / 1000; return (tv); } /* * Clear vblank timestamp buffer for a crtc. */ static void clear_vblank_timestamps(struct drm_device *dev, int crtc) { memset(&dev->_vblank_time[crtc * DRM_VBLANKTIME_RBSIZE], 0, DRM_VBLANKTIME_RBSIZE * sizeof(struct timeval)); } static int64_t abs64(int64_t x) { return (x < 0 ? -x : x); } /* * Disable vblank irq's on crtc, make sure that last vblank count * of hardware and corresponding consistent software vblank counter * are preserved, even if there are any spurious vblank irq's after * disable. */ static void vblank_disable_and_save(struct drm_device *dev, int crtc) { u32 vblcount; int64_t diff_ns; int vblrc; struct timeval tvblank; /* Prevent vblank irq processing while disabling vblank irqs, * so no updates of timestamps or count can happen after we've * disabled. Needed to prevent races in case of delayed irq's. */ mtx_lock(&dev->vblank_time_lock); dev->driver->disable_vblank(dev, crtc); dev->vblank_enabled[crtc] = 0; /* No further vblank irq's will be processed after * this point. Get current hardware vblank count and * vblank timestamp, repeat until they are consistent. * * FIXME: There is still a race condition here and in * drm_update_vblank_count() which can cause off-by-one * reinitialization of software vblank counter. If gpu * vblank counter doesn't increment exactly at the leading * edge of a vblank interval, then we can lose 1 count if * we happen to execute between start of vblank and the * delayed gpu counter increment. */ do { dev->last_vblank[crtc] = dev->driver->get_vblank_counter(dev, crtc); vblrc = drm_get_last_vbltimestamp(dev, crtc, &tvblank, 0); } while (dev->last_vblank[crtc] != dev->driver->get_vblank_counter(dev, crtc)); /* Compute time difference to stored timestamp of last vblank * as updated by last invocation of drm_handle_vblank() in vblank irq. */ vblcount = atomic_read(&dev->_vblank_count[crtc]); diff_ns = timeval_to_ns(&tvblank) - timeval_to_ns(&vblanktimestamp(dev, crtc, vblcount)); /* If there is at least 1 msec difference between the last stored * timestamp and tvblank, then we are currently executing our * disable inside a new vblank interval, the tvblank timestamp * corresponds to this new vblank interval and the irq handler * for this vblank didn't run yet and won't run due to our disable. * Therefore we need to do the job of drm_handle_vblank() and * increment the vblank counter by one to account for this vblank. * * Skip this step if there isn't any high precision timestamp * available. In that case we can't account for this and just * hope for the best. */ if ((vblrc > 0) && (abs64(diff_ns) > 1000000)) { atomic_inc(&dev->_vblank_count[crtc]); } /* Invalidate all timestamps while vblank irq's are off. */ clear_vblank_timestamps(dev, crtc); mtx_unlock(&dev->vblank_time_lock); } static void vblank_disable_fn(void * arg) { struct drm_device *dev = (struct drm_device *)arg; int i; if (!dev->vblank_disable_allowed) return; for (i = 0; i < dev->num_crtcs; i++) { mtx_lock(&dev->vbl_lock); if (atomic_read(&dev->vblank_refcount[i]) == 0 && dev->vblank_enabled[i]) { DRM_DEBUG("disabling vblank on crtc %d\n", i); vblank_disable_and_save(dev, i); } mtx_unlock(&dev->vbl_lock); } } void drm_vblank_cleanup(struct drm_device *dev) { /* Bail if the driver didn't call drm_vblank_init() */ if (dev->num_crtcs == 0) return; callout_stop(&dev->vblank_disable_callout); vblank_disable_fn(dev); free(dev->_vblank_count, DRM_MEM_VBLANK); free(dev->vblank_refcount, DRM_MEM_VBLANK); free(dev->vblank_enabled, DRM_MEM_VBLANK); free(dev->last_vblank, DRM_MEM_VBLANK); free(dev->last_vblank_wait, DRM_MEM_VBLANK); free(dev->vblank_inmodeset, DRM_MEM_VBLANK); free(dev->_vblank_time, DRM_MEM_VBLANK); dev->num_crtcs = 0; } int drm_vblank_init(struct drm_device *dev, int num_crtcs) { int i; callout_init(&dev->vblank_disable_callout, CALLOUT_MPSAFE); #if 0 mtx_init(&dev->vbl_lock, "drmvbl", NULL, MTX_DEF); #endif mtx_init(&dev->vblank_time_lock, "drmvtl", NULL, MTX_DEF); dev->num_crtcs = num_crtcs; dev->_vblank_count = malloc(sizeof(atomic_t) * num_crtcs, DRM_MEM_VBLANK, M_WAITOK); dev->vblank_refcount = malloc(sizeof(atomic_t) * num_crtcs, DRM_MEM_VBLANK, M_WAITOK); dev->vblank_enabled = malloc(num_crtcs * sizeof(int), DRM_MEM_VBLANK, M_WAITOK | M_ZERO); dev->last_vblank = malloc(num_crtcs * sizeof(u32), DRM_MEM_VBLANK, M_WAITOK | M_ZERO); dev->last_vblank_wait = malloc(num_crtcs * sizeof(u32), DRM_MEM_VBLANK, M_WAITOK | M_ZERO); dev->vblank_inmodeset = malloc(num_crtcs * sizeof(int), DRM_MEM_VBLANK, M_WAITOK | M_ZERO); dev->_vblank_time = malloc(num_crtcs * DRM_VBLANKTIME_RBSIZE * sizeof(struct timeval), DRM_MEM_VBLANK, M_WAITOK | M_ZERO); DRM_INFO("Supports vblank timestamp caching Rev 1 (10.10.2010).\n"); /* Driver specific high-precision vblank timestamping supported? */ if (dev->driver->get_vblank_timestamp) DRM_INFO("Driver supports precise vblank timestamp query.\n"); else DRM_INFO("No driver support for vblank timestamp query.\n"); /* Zero per-crtc vblank stuff */ for (i = 0; i < num_crtcs; i++) { atomic_set(&dev->_vblank_count[i], 0); atomic_set(&dev->vblank_refcount[i], 0); } dev->vblank_disable_allowed = 0; return 0; } void drm_calc_timestamping_constants(struct drm_crtc *crtc) { int64_t linedur_ns = 0, pixeldur_ns = 0, framedur_ns = 0; uint64_t dotclock; /* Dot clock in Hz: */ dotclock = (uint64_t) crtc->hwmode.clock * 1000; /* Fields of interlaced scanout modes are only halve a frame duration. * Double the dotclock to get halve the frame-/line-/pixelduration. */ if (crtc->hwmode.flags & DRM_MODE_FLAG_INTERLACE) dotclock *= 2; /* Valid dotclock? */ if (dotclock > 0) { /* Convert scanline length in pixels and video dot clock to * line duration, frame duration and pixel duration in * nanoseconds: */ pixeldur_ns = (int64_t)1000000000 / dotclock; linedur_ns = ((uint64_t)crtc->hwmode.crtc_htotal * 1000000000) / dotclock; framedur_ns = (int64_t)crtc->hwmode.crtc_vtotal * linedur_ns; } else DRM_ERROR("crtc %d: Can't calculate constants, dotclock = 0!\n", crtc->base.id); crtc->pixeldur_ns = pixeldur_ns; crtc->linedur_ns = linedur_ns; crtc->framedur_ns = framedur_ns; DRM_DEBUG("crtc %d: hwmode: htotal %d, vtotal %d, vdisplay %d\n", crtc->base.id, crtc->hwmode.crtc_htotal, crtc->hwmode.crtc_vtotal, crtc->hwmode.crtc_vdisplay); DRM_DEBUG("crtc %d: clock %d kHz framedur %d linedur %d, pixeldur %d\n", crtc->base.id, (int) dotclock/1000, (int) framedur_ns, (int) linedur_ns, (int) pixeldur_ns); } /** * drm_calc_vbltimestamp_from_scanoutpos - helper routine for kms * drivers. Implements calculation of exact vblank timestamps from * given drm_display_mode timings and current video scanout position * of a crtc. This can be called from within get_vblank_timestamp() * implementation of a kms driver to implement the actual timestamping. * * Should return timestamps conforming to the OML_sync_control OpenML * extension specification. The timestamp corresponds to the end of * the vblank interval, aka start of scanout of topmost-leftmost display * pixel in the following video frame. * * Requires support for optional dev->driver->get_scanout_position() * in kms driver, plus a bit of setup code to provide a drm_display_mode * that corresponds to the true scanout timing. * * The current implementation only handles standard video modes. It * returns as no operation if a doublescan or interlaced video mode is * active. Higher level code is expected to handle this. * * @dev: DRM device. * @crtc: Which crtc's vblank timestamp to retrieve. * @max_error: Desired maximum allowable error in timestamps (nanosecs). * On return contains true maximum error of timestamp. * @vblank_time: Pointer to struct timeval which should receive the timestamp. * @flags: Flags to pass to driver: * 0 = Default. * DRM_CALLED_FROM_VBLIRQ = If function is called from vbl irq handler. * @refcrtc: drm_crtc* of crtc which defines scanout timing. * * Returns negative value on error, failure or if not supported in current * video mode: * * -EINVAL - Invalid crtc. * -EAGAIN - Temporary unavailable, e.g., called before initial modeset. * -ENOTSUPP - Function not supported in current display mode. * -EIO - Failed, e.g., due to failed scanout position query. * * Returns or'ed positive status flags on success: * * DRM_VBLANKTIME_SCANOUTPOS_METHOD - Signal this method used for timestamping. * DRM_VBLANKTIME_INVBL - Timestamp taken while scanout was in vblank interval. * */ int drm_calc_vbltimestamp_from_scanoutpos(struct drm_device *dev, int crtc, int *max_error, struct timeval *vblank_time, unsigned flags, struct drm_crtc *refcrtc) { struct timeval stime, raw_time; struct drm_display_mode *mode; int vbl_status, vtotal, vdisplay; int vpos, hpos, i; int64_t framedur_ns, linedur_ns, pixeldur_ns, delta_ns, duration_ns; bool invbl; if (crtc < 0 || crtc >= dev->num_crtcs) { DRM_ERROR("Invalid crtc %d\n", crtc); return -EINVAL; } /* Scanout position query not supported? Should not happen. */ if (!dev->driver->get_scanout_position) { DRM_ERROR("Called from driver w/o get_scanout_position()!?\n"); return -EIO; } mode = &refcrtc->hwmode; vtotal = mode->crtc_vtotal; vdisplay = mode->crtc_vdisplay; /* Durations of frames, lines, pixels in nanoseconds. */ framedur_ns = refcrtc->framedur_ns; linedur_ns = refcrtc->linedur_ns; pixeldur_ns = refcrtc->pixeldur_ns; /* If mode timing undefined, just return as no-op: * Happens during initial modesetting of a crtc. */ if (vtotal <= 0 || vdisplay <= 0 || framedur_ns == 0) { DRM_DEBUG("crtc %d: Noop due to uninitialized mode.\n", crtc); return -EAGAIN; } /* Get current scanout position with system timestamp. * Repeat query up to DRM_TIMESTAMP_MAXRETRIES times * if single query takes longer than max_error nanoseconds. * * This guarantees a tight bound on maximum error if * code gets preempted or delayed for some reason. */ for (i = 0; i < DRM_TIMESTAMP_MAXRETRIES; i++) { /* Disable preemption to make it very likely to * succeed in the first iteration. */ critical_enter(); /* Get system timestamp before query. */ getmicrouptime(&stime); /* Get vertical and horizontal scanout pos. vpos, hpos. */ vbl_status = dev->driver->get_scanout_position(dev, crtc, &vpos, &hpos); /* Get system timestamp after query. */ getmicrouptime(&raw_time); critical_exit(); /* Return as no-op if scanout query unsupported or failed. */ if (!(vbl_status & DRM_SCANOUTPOS_VALID)) { DRM_DEBUG("crtc %d : scanoutpos query failed [%d].\n", crtc, vbl_status); return -EIO; } duration_ns = timeval_to_ns(&raw_time) - timeval_to_ns(&stime); /* Accept result with < max_error nsecs timing uncertainty. */ if (duration_ns <= (int64_t) *max_error) break; } /* Noisy system timing? */ if (i == DRM_TIMESTAMP_MAXRETRIES) { DRM_DEBUG("crtc %d: Noisy timestamp %d us > %d us [%d reps].\n", crtc, (int) duration_ns/1000, *max_error/1000, i); } /* Return upper bound of timestamp precision error. */ *max_error = (int) duration_ns; /* Check if in vblank area: * vpos is >=0 in video scanout area, but negative * within vblank area, counting down the number of lines until * start of scanout. */ invbl = vbl_status & DRM_SCANOUTPOS_INVBL; /* Convert scanout position into elapsed time at raw_time query * since start of scanout at first display scanline. delta_ns * can be negative if start of scanout hasn't happened yet. */ delta_ns = (int64_t)vpos * linedur_ns + (int64_t)hpos * pixeldur_ns; /* Is vpos outside nominal vblank area, but less than * 1/100 of a frame height away from start of vblank? * If so, assume this isn't a massively delayed vblank * interrupt, but a vblank interrupt that fired a few * microseconds before true start of vblank. Compensate * by adding a full frame duration to the final timestamp. * Happens, e.g., on ATI R500, R600. * * We only do this if DRM_CALLED_FROM_VBLIRQ. */ if ((flags & DRM_CALLED_FROM_VBLIRQ) && !invbl && ((vdisplay - vpos) < vtotal / 100)) { delta_ns = delta_ns - framedur_ns; /* Signal this correction as "applied". */ vbl_status |= 0x8; } /* Subtract time delta from raw timestamp to get final * vblank_time timestamp for end of vblank. */ *vblank_time = ns_to_timeval(timeval_to_ns(&raw_time) - delta_ns); DRM_DEBUG("crtc %d : v %d p(%d,%d)@ %jd.%jd -> %jd.%jd [e %d us, %d rep]\n", crtc, (int)vbl_status, hpos, vpos, (uintmax_t)raw_time.tv_sec, (uintmax_t)raw_time.tv_usec, (uintmax_t)vblank_time->tv_sec, (uintmax_t)vblank_time->tv_usec, (int)duration_ns/1000, i); vbl_status = DRM_VBLANKTIME_SCANOUTPOS_METHOD; if (invbl) vbl_status |= DRM_VBLANKTIME_INVBL; return vbl_status; } /** * drm_get_last_vbltimestamp - retrieve raw timestamp for the most recent * vblank interval. * * @dev: DRM device * @crtc: which crtc's vblank timestamp to retrieve * @tvblank: Pointer to target struct timeval which should receive the timestamp * @flags: Flags to pass to driver: * 0 = Default. * DRM_CALLED_FROM_VBLIRQ = If function is called from vbl irq handler. * * Fetches the system timestamp corresponding to the time of the most recent * vblank interval on specified crtc. May call into kms-driver to * compute the timestamp with a high-precision GPU specific method. * * Returns zero if timestamp originates from uncorrected do_gettimeofday() * call, i.e., it isn't very precisely locked to the true vblank. * * Returns non-zero if timestamp is considered to be very precise. */ u32 drm_get_last_vbltimestamp(struct drm_device *dev, int crtc, struct timeval *tvblank, unsigned flags) { int ret; /* Define requested maximum error on timestamps (nanoseconds). */ int max_error = (int) drm_timestamp_precision * 1000; /* Query driver if possible and precision timestamping enabled. */ if (dev->driver->get_vblank_timestamp && (max_error > 0)) { ret = dev->driver->get_vblank_timestamp(dev, crtc, &max_error, tvblank, flags); if (ret > 0) return (u32) ret; } /* GPU high precision timestamp query unsupported or failed. * Return gettimeofday timestamp as best estimate. */ microtime(tvblank); return 0; } /** * drm_vblank_count - retrieve "cooked" vblank counter value * @dev: DRM device * @crtc: which counter to retrieve * * Fetches the "cooked" vblank count value that represents the number of * vblank events since the system was booted, including lost events due to * modesetting activity. */ u32 drm_vblank_count(struct drm_device *dev, int crtc) { return atomic_read(&dev->_vblank_count[crtc]); } /** * drm_vblank_count_and_time - retrieve "cooked" vblank counter value * and the system timestamp corresponding to that vblank counter value. * * @dev: DRM device * @crtc: which counter to retrieve * @vblanktime: Pointer to struct timeval to receive the vblank timestamp. * * Fetches the "cooked" vblank count value that represents the number of * vblank events since the system was booted, including lost events due to * modesetting activity. Returns corresponding system timestamp of the time * of the vblank interval that corresponds to the current value vblank counter * value. */ u32 drm_vblank_count_and_time(struct drm_device *dev, int crtc, struct timeval *vblanktime) { u32 cur_vblank; /* Read timestamp from slot of _vblank_time ringbuffer * that corresponds to current vblank count. Retry if * count has incremented during readout. This works like * a seqlock. */ do { cur_vblank = atomic_read(&dev->_vblank_count[crtc]); *vblanktime = vblanktimestamp(dev, crtc, cur_vblank); rmb(); } while (cur_vblank != atomic_read(&dev->_vblank_count[crtc])); return cur_vblank; } /** * drm_update_vblank_count - update the master vblank counter * @dev: DRM device * @crtc: counter to update * * Call back into the driver to update the appropriate vblank counter * (specified by @crtc). Deal with wraparound, if it occurred, and * update the last read value so we can deal with wraparound on the next * call if necessary. * * Only necessary when going from off->on, to account for frames we * didn't get an interrupt for. * * Note: caller must hold dev->vbl_lock since this reads & writes * device vblank fields. */ static void drm_update_vblank_count(struct drm_device *dev, int crtc) { u32 cur_vblank, diff, tslot, rc; struct timeval t_vblank; /* * Interrupts were disabled prior to this call, so deal with counter * wrap if needed. * NOTE! It's possible we lost a full dev->max_vblank_count events * here if the register is small or we had vblank interrupts off for * a long time. * * We repeat the hardware vblank counter & timestamp query until * we get consistent results. This to prevent races between gpu * updating its hardware counter while we are retrieving the * corresponding vblank timestamp. */ do { cur_vblank = dev->driver->get_vblank_counter(dev, crtc); rc = drm_get_last_vbltimestamp(dev, crtc, &t_vblank, 0); } while (cur_vblank != dev->driver->get_vblank_counter(dev, crtc)); /* Deal with counter wrap */ diff = cur_vblank - dev->last_vblank[crtc]; if (cur_vblank < dev->last_vblank[crtc]) { diff += dev->max_vblank_count; DRM_DEBUG("last_vblank[%d]=0x%x, cur_vblank=0x%x => diff=0x%x\n", crtc, dev->last_vblank[crtc], cur_vblank, diff); } DRM_DEBUG("enabling vblank interrupts on crtc %d, missed %d\n", crtc, diff); /* Reinitialize corresponding vblank timestamp if high-precision query * available. Skip this step if query unsupported or failed. Will * reinitialize delayed at next vblank interrupt in that case. */ if (rc) { tslot = atomic_read(&dev->_vblank_count[crtc]) + diff; vblanktimestamp(dev, crtc, tslot) = t_vblank; } atomic_add(diff, &dev->_vblank_count[crtc]); } /** * drm_vblank_get - get a reference count on vblank events * @dev: DRM device * @crtc: which CRTC to own * * Acquire a reference count on vblank events to avoid having them disabled * while in use. * * RETURNS * Zero on success, nonzero on failure. */ int drm_vblank_get(struct drm_device *dev, int crtc) { int ret = 0; mtx_lock(&dev->vbl_lock); /* Going from 0->1 means we have to enable interrupts again */ if (atomic_add_return(1, &dev->vblank_refcount[crtc]) == 1) { mtx_lock(&dev->vblank_time_lock); if (!dev->vblank_enabled[crtc]) { /* Enable vblank irqs under vblank_time_lock protection. * All vblank count & timestamp updates are held off * until we are done reinitializing master counter and * timestamps. Filtercode in drm_handle_vblank() will * prevent double-accounting of same vblank interval. */ ret = -dev->driver->enable_vblank(dev, crtc); DRM_DEBUG("enabling vblank on crtc %d, ret: %d\n", crtc, ret); if (ret) atomic_dec(&dev->vblank_refcount[crtc]); else { dev->vblank_enabled[crtc] = 1; drm_update_vblank_count(dev, crtc); } } mtx_unlock(&dev->vblank_time_lock); } else { if (!dev->vblank_enabled[crtc]) { atomic_dec(&dev->vblank_refcount[crtc]); ret = EINVAL; } } mtx_unlock(&dev->vbl_lock); return ret; } /** * drm_vblank_put - give up ownership of vblank events * @dev: DRM device * @crtc: which counter to give up * * Release ownership of a given vblank counter, turning off interrupts * if possible. Disable interrupts after drm_vblank_offdelay milliseconds. */ void drm_vblank_put(struct drm_device *dev, int crtc) { KASSERT(atomic_read(&dev->vblank_refcount[crtc]) != 0, ("Too many drm_vblank_put for crtc %d", crtc)); /* Last user schedules interrupt disable */ if (atomic_dec_and_test(&dev->vblank_refcount[crtc]) && (drm_vblank_offdelay > 0)) callout_reset(&dev->vblank_disable_callout, (drm_vblank_offdelay * DRM_HZ) / 1000, vblank_disable_fn, dev); } void drm_vblank_off(struct drm_device *dev, int crtc) { struct drm_pending_vblank_event *e, *t; struct timeval now; unsigned int seq; mtx_lock(&dev->vbl_lock); vblank_disable_and_save(dev, crtc); mtx_lock(&dev->event_lock); wakeup(&dev->_vblank_count[crtc]); /* Send any queued vblank events, lest the natives grow disquiet */ seq = drm_vblank_count_and_time(dev, crtc, &now); list_for_each_entry_safe(e, t, &dev->vblank_event_list, base.link) { if (e->pipe != crtc) continue; DRM_DEBUG("Sending premature vblank event on disable: \ wanted %d, current %d\n", e->event.sequence, seq); e->event.sequence = seq; e->event.tv_sec = now.tv_sec; e->event.tv_usec = now.tv_usec; drm_vblank_put(dev, e->pipe); list_move_tail(&e->base.link, &e->base.file_priv->event_list); drm_event_wakeup(&e->base); CTR3(KTR_DRM, "vblank_event_delivered %d %d %d", e->base.pid, e->pipe, e->event.sequence); } mtx_unlock(&dev->event_lock); mtx_unlock(&dev->vbl_lock); } /** * drm_vblank_pre_modeset - account for vblanks across mode sets * @dev: DRM device * @crtc: CRTC in question * @post: post or pre mode set? * * Account for vblank events across mode setting events, which will likely * reset the hardware frame counter. */ void drm_vblank_pre_modeset(struct drm_device *dev, int crtc) { - /* vblank is not initialized (IRQ not installed ?) */ + /* vblank is not initialized (IRQ not installed ?), or has been freed */ if (!dev->num_crtcs) return; /* * To avoid all the problems that might happen if interrupts * were enabled/disabled around or between these calls, we just * have the kernel take a reference on the CRTC (just once though * to avoid corrupting the count if multiple, mismatch calls occur), * so that interrupts remain enabled in the interim. */ if (!dev->vblank_inmodeset[crtc]) { dev->vblank_inmodeset[crtc] = 0x1; if (drm_vblank_get(dev, crtc) == 0) dev->vblank_inmodeset[crtc] |= 0x2; } } void drm_vblank_post_modeset(struct drm_device *dev, int crtc) { + /* vblank is not initialized (IRQ not installed ?), or has been freed */ + if (!dev->num_crtcs) + return; if (dev->vblank_inmodeset[crtc]) { mtx_lock(&dev->vbl_lock); dev->vblank_disable_allowed = 1; mtx_unlock(&dev->vbl_lock); if (dev->vblank_inmodeset[crtc] & 0x2) drm_vblank_put(dev, crtc); dev->vblank_inmodeset[crtc] = 0; } } /** * drm_modeset_ctl - handle vblank event counter changes across mode switch * @DRM_IOCTL_ARGS: standard ioctl arguments * * Applications should call the %_DRM_PRE_MODESET and %_DRM_POST_MODESET * ioctls around modesetting so that any lost vblank events are accounted for. * * Generally the counter will reset across mode sets. If interrupts are * enabled around this call, we don't have to do anything since the counter * will have already been incremented. */ int drm_modeset_ctl(struct drm_device *dev, void *data, struct drm_file *file_priv) { struct drm_modeset_ctl *modeset = data; unsigned int crtc; /* If drm_vblank_init() hasn't been called yet, just no-op */ if (!dev->num_crtcs) return 0; crtc = modeset->crtc; if (crtc >= dev->num_crtcs) return -EINVAL; switch (modeset->cmd) { case _DRM_PRE_MODESET: drm_vblank_pre_modeset(dev, crtc); break; case _DRM_POST_MODESET: drm_vblank_post_modeset(dev, crtc); break; default: return -EINVAL; break; } return 0; } static void drm_vblank_event_destroy(struct drm_pending_event *e) { free(e, DRM_MEM_VBLANK); } static int drm_queue_vblank_event(struct drm_device *dev, int pipe, union drm_wait_vblank *vblwait, struct drm_file *file_priv) { struct drm_pending_vblank_event *e; struct timeval now; unsigned int seq; int ret; e = malloc(sizeof *e, DRM_MEM_VBLANK, M_WAITOK | M_ZERO); e->pipe = pipe; e->base.pid = curproc->p_pid; e->event.base.type = DRM_EVENT_VBLANK; e->event.base.length = sizeof e->event; e->event.user_data = vblwait->request.signal; e->base.event = &e->event.base; e->base.file_priv = file_priv; e->base.destroy = drm_vblank_event_destroy; mtx_lock(&dev->event_lock); if (file_priv->event_space < sizeof e->event) { ret = EBUSY; goto err_unlock; } file_priv->event_space -= sizeof e->event; seq = drm_vblank_count_and_time(dev, pipe, &now); if ((vblwait->request.type & _DRM_VBLANK_NEXTONMISS) && (seq - vblwait->request.sequence) <= (1 << 23)) { vblwait->request.sequence = seq + 1; vblwait->reply.sequence = vblwait->request.sequence; } DRM_DEBUG("event on vblank count %d, current %d, crtc %d\n", vblwait->request.sequence, seq, pipe); CTR4(KTR_DRM, "vblank_event_queued %d %d rt %x %d", curproc->p_pid, pipe, vblwait->request.type, vblwait->request.sequence); e->event.sequence = vblwait->request.sequence; if ((seq - vblwait->request.sequence) <= (1 << 23)) { e->event.sequence = seq; e->event.tv_sec = now.tv_sec; e->event.tv_usec = now.tv_usec; drm_vblank_put(dev, pipe); list_add_tail(&e->base.link, &e->base.file_priv->event_list); drm_event_wakeup(&e->base); vblwait->reply.sequence = seq; CTR3(KTR_DRM, "vblank_event_wakeup p1 %d %d %d", curproc->p_pid, pipe, vblwait->request.sequence); } else { /* drm_handle_vblank_events will call drm_vblank_put */ list_add_tail(&e->base.link, &dev->vblank_event_list); vblwait->reply.sequence = vblwait->request.sequence; } mtx_unlock(&dev->event_lock); return 0; err_unlock: mtx_unlock(&dev->event_lock); free(e, DRM_MEM_VBLANK); drm_vblank_put(dev, pipe); return ret; } /** * Wait for VBLANK. * * \param inode device inode. * \param file_priv DRM file private. * \param cmd command. * \param data user argument, pointing to a drm_wait_vblank structure. * \return zero on success or a negative number on failure. * * This function enables the vblank interrupt on the pipe requested, then * sleeps waiting for the requested sequence number to occur, and drops * the vblank interrupt refcount afterwards. (vblank irq disable follows that * after a timeout with no further vblank waits scheduled). */ int drm_wait_vblank(struct drm_device *dev, void *data, struct drm_file *file_priv) { union drm_wait_vblank *vblwait = data; int ret; unsigned int flags, seq, crtc, high_crtc; if (/*(!drm_dev_to_irq(dev)) || */(!dev->irq_enabled)) return (EINVAL); if (vblwait->request.type & _DRM_VBLANK_SIGNAL) return (EINVAL); if (vblwait->request.type & ~(_DRM_VBLANK_TYPES_MASK | _DRM_VBLANK_FLAGS_MASK | _DRM_VBLANK_HIGH_CRTC_MASK)) { DRM_ERROR("Unsupported type value 0x%x, supported mask 0x%x\n", vblwait->request.type, (_DRM_VBLANK_TYPES_MASK | _DRM_VBLANK_FLAGS_MASK | _DRM_VBLANK_HIGH_CRTC_MASK)); return (EINVAL); } flags = vblwait->request.type & _DRM_VBLANK_FLAGS_MASK; high_crtc = (vblwait->request.type & _DRM_VBLANK_HIGH_CRTC_MASK); if (high_crtc) crtc = high_crtc >> _DRM_VBLANK_HIGH_CRTC_SHIFT; else crtc = flags & _DRM_VBLANK_SECONDARY ? 1 : 0; if (crtc >= dev->num_crtcs) return (EINVAL); ret = drm_vblank_get(dev, crtc); if (ret) { DRM_DEBUG("failed to acquire vblank counter, %d\n", ret); return (ret); } seq = drm_vblank_count(dev, crtc); switch (vblwait->request.type & _DRM_VBLANK_TYPES_MASK) { case _DRM_VBLANK_RELATIVE: vblwait->request.sequence += seq; vblwait->request.type &= ~_DRM_VBLANK_RELATIVE; case _DRM_VBLANK_ABSOLUTE: break; default: ret = (EINVAL); goto done; } if (flags & _DRM_VBLANK_EVENT) { /* must hold on to the vblank ref until the event fires * drm_vblank_put will be called asynchronously */ return drm_queue_vblank_event(dev, crtc, vblwait, file_priv); } if ((flags & _DRM_VBLANK_NEXTONMISS) && (seq - vblwait->request.sequence) <= (1<<23)) { vblwait->request.sequence = seq + 1; } dev->last_vblank_wait[crtc] = vblwait->request.sequence; mtx_lock(&dev->vblank_time_lock); while (((drm_vblank_count(dev, crtc) - vblwait->request.sequence) > (1 << 23)) && dev->irq_enabled) { /* * The wakeups from the drm_irq_uninstall() and * drm_vblank_off() may be lost there since vbl_lock * is not held. Then, the timeout will wake us; the 3 * seconds delay should not be a problem for * application when crtc is disabled or irq * uninstalled anyway. */ ret = msleep(&dev->_vblank_count[crtc], &dev->vblank_time_lock, PCATCH, "drmvbl", 3 * hz); if (ret != 0) break; } mtx_unlock(&dev->vblank_time_lock); if (ret != EINTR) { struct timeval now; long reply_seq; reply_seq = drm_vblank_count_and_time(dev, crtc, &now); CTR5(KTR_DRM, "wait_vblank %d %d rt %x success %d %d", curproc->p_pid, crtc, vblwait->request.type, vblwait->request.sequence, reply_seq); vblwait->reply.sequence = reply_seq; vblwait->reply.tval_sec = now.tv_sec; vblwait->reply.tval_usec = now.tv_usec; } else { CTR5(KTR_DRM, "wait_vblank %d %d rt %x error %d %d", curproc->p_pid, crtc, vblwait->request.type, ret, vblwait->request.sequence); } done: drm_vblank_put(dev, crtc); return ret; } void drm_handle_vblank_events(struct drm_device *dev, int crtc) { struct drm_pending_vblank_event *e, *t; struct timeval now; unsigned int seq; seq = drm_vblank_count_and_time(dev, crtc, &now); CTR2(KTR_DRM, "drm_handle_vblank_events %d %d", seq, crtc); mtx_lock(&dev->event_lock); list_for_each_entry_safe(e, t, &dev->vblank_event_list, base.link) { if (e->pipe != crtc) continue; if ((seq - e->event.sequence) > (1<<23)) continue; e->event.sequence = seq; e->event.tv_sec = now.tv_sec; e->event.tv_usec = now.tv_usec; drm_vblank_put(dev, e->pipe); list_move_tail(&e->base.link, &e->base.file_priv->event_list); drm_event_wakeup(&e->base); CTR3(KTR_DRM, "vblank_event_wakeup p2 %d %d %d", e->base.pid, e->pipe, e->event.sequence); } mtx_unlock(&dev->event_lock); } /** * drm_handle_vblank - handle a vblank event * @dev: DRM device * @crtc: where this event occurred * * Drivers should call this routine in their vblank interrupt handlers to * update the vblank counter and send any signals that may be pending. */ bool drm_handle_vblank(struct drm_device *dev, int crtc) { u32 vblcount; int64_t diff_ns; struct timeval tvblank; if (!dev->num_crtcs) return false; /* Need timestamp lock to prevent concurrent execution with * vblank enable/disable, as this would cause inconsistent * or corrupted timestamps and vblank counts. */ mtx_lock(&dev->vblank_time_lock); /* Vblank irq handling disabled. Nothing to do. */ if (!dev->vblank_enabled[crtc]) { mtx_unlock(&dev->vblank_time_lock); return false; } /* Fetch corresponding timestamp for this vblank interval from * driver and store it in proper slot of timestamp ringbuffer. */ /* Get current timestamp and count. */ vblcount = atomic_read(&dev->_vblank_count[crtc]); drm_get_last_vbltimestamp(dev, crtc, &tvblank, DRM_CALLED_FROM_VBLIRQ); /* Compute time difference to timestamp of last vblank */ diff_ns = timeval_to_ns(&tvblank) - timeval_to_ns(&vblanktimestamp(dev, crtc, vblcount)); /* Update vblank timestamp and count if at least * DRM_REDUNDANT_VBLIRQ_THRESH_NS nanoseconds * difference between last stored timestamp and current * timestamp. A smaller difference means basically * identical timestamps. Happens if this vblank has * been already processed and this is a redundant call, * e.g., due to spurious vblank interrupts. We need to * ignore those for accounting. */ if (abs64(diff_ns) > DRM_REDUNDANT_VBLIRQ_THRESH_NS) { /* Store new timestamp in ringbuffer. */ vblanktimestamp(dev, crtc, vblcount + 1) = tvblank; /* Increment cooked vblank count. This also atomically commits * the timestamp computed above. */ atomic_inc(&dev->_vblank_count[crtc]); } else { DRM_DEBUG("crtc %d: Redundant vblirq ignored. diff_ns = %d\n", crtc, (int) diff_ns); } wakeup(&dev->_vblank_count[crtc]); drm_handle_vblank_events(dev, crtc); mtx_unlock(&dev->vblank_time_lock); return true; } Index: projects/cxl_iscsi/sys/dev/flash/at45d.c =================================================================== --- projects/cxl_iscsi/sys/dev/flash/at45d.c (revision 279716) +++ projects/cxl_iscsi/sys/dev/flash/at45d.c (revision 279717) @@ -1,448 +1,450 @@ /*- * Copyright (c) 2006 M. Warner Losh * Copyright (c) 2011-2012 Ian Lepore * Copyright (c) 2012 Marius Strobl * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "spibus_if.h" struct at45d_flash_ident { const char *name; uint32_t jedec; uint16_t pagecount; uint16_t pageoffset; uint16_t pagesize; uint16_t pagesize2n; }; struct at45d_softc { struct bio_queue_head bio_queue; struct mtx sc_mtx; struct disk *disk; struct proc *p; struct intr_config_hook config_intrhook; device_t dev; uint16_t pagecount; uint16_t pageoffset; uint16_t pagesize; }; #define AT45D_LOCK(_sc) mtx_lock(&(_sc)->sc_mtx) #define AT45D_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_mtx) #define AT45D_LOCK_INIT(_sc) \ mtx_init(&_sc->sc_mtx, device_get_nameunit(_sc->dev), \ "at45d", MTX_DEF) #define AT45D_LOCK_DESTROY(_sc) mtx_destroy(&_sc->sc_mtx); #define AT45D_ASSERT_LOCKED(_sc) mtx_assert(&_sc->sc_mtx, MA_OWNED); #define AT45D_ASSERT_UNLOCKED(_sc) mtx_assert(&_sc->sc_mtx, MA_NOTOWNED); /* bus entry points */ static device_attach_t at45d_attach; static device_detach_t at45d_detach; static device_probe_t at45d_probe; /* disk routines */ static int at45d_close(struct disk *dp); static int at45d_open(struct disk *dp); static void at45d_strategy(struct bio *bp); static void at45d_task(void *arg); /* helper routines */ static void at45d_delayed_attach(void *xsc); static int at45d_get_mfg_info(device_t dev, uint8_t *resp); static int at45d_get_status(device_t dev, uint8_t *status); static int at45d_wait_ready(device_t dev, uint8_t *status); #define BUFFER_TRANSFER 0x53 #define BUFFER_COMPARE 0x60 #define PROGRAM_THROUGH_BUFFER 0x82 #define MANUFACTURER_ID 0x9f #define STATUS_REGISTER_READ 0xd7 #define CONTINUOUS_ARRAY_READ 0xe8 /* * A sectorsize2n != 0 is used to indicate that a device optionally supports * 2^N byte pages. If support for the latter is enabled, the sector offset * has to be reduced by one. */ static const struct at45d_flash_ident at45d_flash_devices[] = { { "AT45DB011B", 0x1f2200, 512, 9, 264, 256 }, { "AT45DB021B", 0x1f2300, 1024, 9, 264, 256 }, { "AT45DB041x", 0x1f2400, 2028, 9, 264, 256 }, { "AT45DB081B", 0x1f2500, 4096, 9, 264, 256 }, { "AT45DB161x", 0x1f2600, 4096, 10, 528, 512 }, { "AT45DB321x", 0x1f2700, 8192, 10, 528, 0 }, { "AT45DB321x", 0x1f2701, 8192, 10, 528, 512 }, { "AT45DB642x", 0x1f2800, 8192, 11, 1056, 1024 } }; static int at45d_get_status(device_t dev, uint8_t *status) { uint8_t rxBuf[8], txBuf[8]; struct spi_command cmd; int err; memset(&cmd, 0, sizeof(cmd)); memset(txBuf, 0, sizeof(txBuf)); memset(rxBuf, 0, sizeof(rxBuf)); txBuf[0] = STATUS_REGISTER_READ; cmd.tx_cmd = txBuf; cmd.rx_cmd = rxBuf; cmd.rx_cmd_sz = cmd.tx_cmd_sz = 2; err = SPIBUS_TRANSFER(device_get_parent(dev), dev, &cmd); *status = rxBuf[1]; return (err); } static int at45d_get_mfg_info(device_t dev, uint8_t *resp) { uint8_t rxBuf[8], txBuf[8]; struct spi_command cmd; int err; memset(&cmd, 0, sizeof(cmd)); memset(txBuf, 0, sizeof(txBuf)); memset(rxBuf, 0, sizeof(rxBuf)); txBuf[0] = MANUFACTURER_ID; cmd.tx_cmd = &txBuf; cmd.rx_cmd = &rxBuf; cmd.tx_cmd_sz = cmd.rx_cmd_sz = 5; err = SPIBUS_TRANSFER(device_get_parent(dev), dev, &cmd); if (err) return (err); memcpy(resp, rxBuf + 1, 4); return (0); } static int at45d_wait_ready(device_t dev, uint8_t *status) { struct timeval now, tout; int err; getmicrouptime(&tout); tout.tv_sec += 3; do { getmicrouptime(&now); if (now.tv_sec > tout.tv_sec) err = ETIMEDOUT; else err = at45d_get_status(dev, status); } while (err == 0 && (*status & 0x80) == 0); return (err); } static int at45d_probe(device_t dev) { device_set_desc(dev, "AT45D Flash Family"); return (0); } static int at45d_attach(device_t dev) { struct at45d_softc *sc; sc = device_get_softc(dev); sc->dev = dev; AT45D_LOCK_INIT(sc); /* We'll see what kind of flash we have later... */ sc->config_intrhook.ich_func = at45d_delayed_attach; sc->config_intrhook.ich_arg = sc; - if (config_intrhook_establish(&sc->config_intrhook) != 0) + if (config_intrhook_establish(&sc->config_intrhook) != 0) { device_printf(dev, "config_intrhook_establish failed\n"); + return (ENOMEM); + } return (0); } static int at45d_detach(device_t dev) { return (EBUSY) /* XXX */; } static void at45d_delayed_attach(void *xsc) { struct at45d_softc *sc; const struct at45d_flash_ident *ident; u_int i; uint32_t jedec; uint16_t pagesize; uint8_t buf[4], status; sc = xsc; ident = NULL; jedec = 0; if (at45d_wait_ready(sc->dev, &status) != 0) device_printf(sc->dev, "Error waiting for device-ready.\n"); else if (at45d_get_mfg_info(sc->dev, buf) != 0) device_printf(sc->dev, "Failed to get ID.\n"); else { jedec = buf[0] << 16 | buf[1] << 8 | buf[2]; for (i = 0; i < nitems(at45d_flash_devices); i++) { if (at45d_flash_devices[i].jedec == jedec) { ident = &at45d_flash_devices[i]; break; } } } if (ident == NULL) device_printf(sc->dev, "JEDEC 0x%x not in list.\n", jedec); else { sc->pagecount = ident->pagecount; sc->pageoffset = ident->pageoffset; if (ident->pagesize2n != 0 && (status & 0x01) != 0) { sc->pageoffset -= 1; pagesize = ident->pagesize2n; } else pagesize = ident->pagesize; sc->pagesize = pagesize; sc->disk = disk_alloc(); sc->disk->d_open = at45d_open; sc->disk->d_close = at45d_close; sc->disk->d_strategy = at45d_strategy; sc->disk->d_name = "flash/spi"; sc->disk->d_drv1 = sc; sc->disk->d_maxsize = DFLTPHYS; sc->disk->d_sectorsize = pagesize; sc->disk->d_mediasize = pagesize * ident->pagecount; sc->disk->d_unit = device_get_unit(sc->dev); disk_create(sc->disk, DISK_VERSION); bioq_init(&sc->bio_queue); kproc_create(&at45d_task, sc, &sc->p, 0, 0, "task: at45d flash"); device_printf(sc->dev, "%s, %d bytes per page, %d pages\n", ident->name, pagesize, ident->pagecount); } config_intrhook_disestablish(&sc->config_intrhook); } static int at45d_open(struct disk *dp) { return (0); } static int at45d_close(struct disk *dp) { return (0); } static void at45d_strategy(struct bio *bp) { struct at45d_softc *sc; sc = (struct at45d_softc *)bp->bio_disk->d_drv1; AT45D_LOCK(sc); bioq_disksort(&sc->bio_queue, bp); wakeup(sc); AT45D_UNLOCK(sc); } static void at45d_task(void *arg) { uint8_t rxBuf[8], txBuf[8]; struct at45d_softc *sc; struct bio *bp; struct spi_command cmd; device_t dev, pdev; caddr_t buf; u_long len, resid; u_int addr, berr, err, offset, page; uint8_t status; sc = (struct at45d_softc*)arg; dev = sc->dev; pdev = device_get_parent(dev); memset(&cmd, 0, sizeof(cmd)); memset(txBuf, 0, sizeof(txBuf)); memset(rxBuf, 0, sizeof(rxBuf)); cmd.tx_cmd = txBuf; cmd.rx_cmd = rxBuf; for (;;) { AT45D_LOCK(sc); do { bp = bioq_takefirst(&sc->bio_queue); if (bp == NULL) msleep(sc, &sc->sc_mtx, PRIBIO, "jobqueue", 0); } while (bp == NULL); AT45D_UNLOCK(sc); berr = 0; buf = bp->bio_data; len = resid = bp->bio_bcount; page = bp->bio_offset / sc->pagesize; offset = bp->bio_offset % sc->pagesize; switch (bp->bio_cmd) { case BIO_READ: txBuf[0] = CONTINUOUS_ARRAY_READ; cmd.tx_cmd_sz = cmd.rx_cmd_sz = 8; cmd.tx_data = cmd.rx_data = buf; break; case BIO_WRITE: cmd.tx_cmd_sz = cmd.rx_cmd_sz = 4; cmd.tx_data = cmd.rx_data = buf; if (resid + offset > sc->pagesize) len = sc->pagesize - offset; break; default: berr = EINVAL; goto out; } /* * NB: for BIO_READ, this loop is only traversed once. */ while (resid > 0) { if (page > sc->pagecount) { berr = EINVAL; goto out; } addr = page << sc->pageoffset; if (bp->bio_cmd == BIO_WRITE) { if (len != sc->pagesize) { txBuf[0] = BUFFER_TRANSFER; txBuf[1] = ((addr >> 16) & 0xff); txBuf[2] = ((addr >> 8) & 0xff); txBuf[3] = 0; cmd.tx_data_sz = cmd.rx_data_sz = 0; err = SPIBUS_TRANSFER(pdev, dev, &cmd); if (err == 0) err = at45d_wait_ready(dev, &status); if (err != 0) { berr = EIO; goto out; } } txBuf[0] = PROGRAM_THROUGH_BUFFER; } addr += offset; txBuf[1] = ((addr >> 16) & 0xff); txBuf[2] = ((addr >> 8) & 0xff); txBuf[3] = (addr & 0xff); cmd.tx_data_sz = cmd.rx_data_sz = len; err = SPIBUS_TRANSFER(pdev, dev, &cmd); if (err == 0 && bp->bio_cmd != BIO_READ) err = at45d_wait_ready(dev, &status); if (err != 0) { berr = EIO; goto out; } if (bp->bio_cmd == BIO_WRITE) { addr = page << sc->pageoffset; txBuf[0] = BUFFER_COMPARE; txBuf[1] = ((addr >> 16) & 0xff); txBuf[2] = ((addr >> 8) & 0xff); txBuf[3] = 0; cmd.tx_data_sz = cmd.rx_data_sz = 0; err = SPIBUS_TRANSFER(pdev, dev, &cmd); if (err == 0) err = at45d_wait_ready(dev, &status); if (err != 0 || (status & 0x40) != 0) { device_printf(dev, "comparing page " "%d failed (status=0x%x)\n", addr, status); berr = EIO; goto out; } } page++; buf += len; offset = 0; resid -= len; if (resid > sc->pagesize) len = sc->pagesize; else len = resid; cmd.tx_data = cmd.rx_data = buf; } out: if (berr != 0) { bp->bio_flags |= BIO_ERROR; bp->bio_error = berr; } bp->bio_resid = resid; biodone(bp); } } static devclass_t at45d_devclass; static device_method_t at45d_methods[] = { /* Device interface */ DEVMETHOD(device_probe, at45d_probe), DEVMETHOD(device_attach, at45d_attach), DEVMETHOD(device_detach, at45d_detach), DEVMETHOD_END }; static driver_t at45d_driver = { "at45d", at45d_methods, sizeof(struct at45d_softc), }; DRIVER_MODULE(at45d, spibus, at45d_driver, at45d_devclass, NULL, NULL); Index: projects/cxl_iscsi/sys/dev/gpio/gpiobus.c =================================================================== --- projects/cxl_iscsi/sys/dev/gpio/gpiobus.c (revision 279716) +++ projects/cxl_iscsi/sys/dev/gpio/gpiobus.c (revision 279717) @@ -1,700 +1,715 @@ /*- * Copyright (c) 2009 Oleksandr Tymoshenko * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include "gpiobus_if.h" #undef GPIOBUS_DEBUG #ifdef GPIOBUS_DEBUG #define dprintf printf #else #define dprintf(x, arg...) #endif static void gpiobus_print_pins(struct gpiobus_ivar *, char *, size_t); static int gpiobus_parse_pins(struct gpiobus_softc *, device_t, int); static int gpiobus_probe(device_t); static int gpiobus_attach(device_t); static int gpiobus_detach(device_t); static int gpiobus_suspend(device_t); static int gpiobus_resume(device_t); +static void gpiobus_probe_nomatch(device_t, device_t); static int gpiobus_print_child(device_t, device_t); static int gpiobus_child_location_str(device_t, device_t, char *, size_t); static int gpiobus_child_pnpinfo_str(device_t, device_t, char *, size_t); static device_t gpiobus_add_child(device_t, u_int, const char *, int); static void gpiobus_hinted_child(device_t, const char *, int); /* * GPIOBUS interface */ static int gpiobus_acquire_bus(device_t, device_t, int); static void gpiobus_release_bus(device_t, device_t); static int gpiobus_pin_setflags(device_t, device_t, uint32_t, uint32_t); static int gpiobus_pin_getflags(device_t, device_t, uint32_t, uint32_t*); static int gpiobus_pin_getcaps(device_t, device_t, uint32_t, uint32_t*); static int gpiobus_pin_set(device_t, device_t, uint32_t, unsigned int); static int gpiobus_pin_get(device_t, device_t, uint32_t, unsigned int*); static int gpiobus_pin_toggle(device_t, device_t, uint32_t); int gpio_check_flags(uint32_t caps, uint32_t flags) { /* Check for unwanted flags. */ if ((flags & caps) == 0 || (flags & caps) != flags) return (EINVAL); /* Cannot mix input/output together. */ if (flags & GPIO_PIN_INPUT && flags & GPIO_PIN_OUTPUT) return (EINVAL); /* Cannot mix pull-up/pull-down together. */ if (flags & GPIO_PIN_PULLUP && flags & GPIO_PIN_PULLDOWN) return (EINVAL); return (0); } static void gpiobus_print_pins(struct gpiobus_ivar *devi, char *buf, size_t buflen) { char tmp[128]; int i, range_start, range_stop, need_coma; if (devi->npins == 0) return; need_coma = 0; range_start = range_stop = devi->pins[0]; for (i = 1; i < devi->npins; i++) { if (devi->pins[i] != (range_stop + 1)) { if (need_coma) strlcat(buf, ",", buflen); memset(tmp, 0, sizeof(tmp)); if (range_start != range_stop) snprintf(tmp, sizeof(tmp) - 1, "%d-%d", range_start, range_stop); else snprintf(tmp, sizeof(tmp) - 1, "%d", range_start); strlcat(buf, tmp, buflen); range_start = range_stop = devi->pins[i]; need_coma = 1; } else range_stop++; } if (need_coma) strlcat(buf, ",", buflen); memset(tmp, 0, sizeof(tmp)); if (range_start != range_stop) snprintf(tmp, sizeof(tmp) - 1, "%d-%d", range_start, range_stop); else snprintf(tmp, sizeof(tmp) - 1, "%d", range_start); strlcat(buf, tmp, buflen); } device_t gpiobus_attach_bus(device_t dev) { device_t busdev; busdev = device_add_child(dev, "gpiobus", -1); if (busdev == NULL) return (NULL); if (device_add_child(dev, "gpioc", -1) == NULL) { device_delete_child(dev, busdev); return (NULL); } #ifdef FDT ofw_gpiobus_register_provider(dev); #endif bus_generic_attach(dev); return (busdev); } int gpiobus_detach_bus(device_t dev) { #ifdef FDT ofw_gpiobus_unregister_provider(dev); #endif return (bus_generic_detach(dev)); } int gpiobus_init_softc(device_t dev) { struct gpiobus_softc *sc; sc = GPIOBUS_SOFTC(dev); sc->sc_busdev = dev; sc->sc_dev = device_get_parent(dev); sc->sc_intr_rman.rm_type = RMAN_ARRAY; sc->sc_intr_rman.rm_descr = "GPIO Interrupts"; if (rman_init(&sc->sc_intr_rman) != 0 || rman_manage_region(&sc->sc_intr_rman, 0, ~0) != 0) panic("%s: failed to set up rman.", __func__); if (GPIO_PIN_MAX(sc->sc_dev, &sc->sc_npins) != 0) return (ENXIO); KASSERT(sc->sc_npins != 0, ("GPIO device with no pins")); /* Pins = GPIO_PIN_MAX() + 1 */ sc->sc_npins++; sc->sc_pins_mapped = malloc(sizeof(int) * sc->sc_npins, M_DEVBUF, M_NOWAIT | M_ZERO); if (sc->sc_pins_mapped == NULL) return (ENOMEM); /* Initialize the bus lock. */ GPIOBUS_LOCK_INIT(sc); return (0); } int gpiobus_alloc_ivars(struct gpiobus_ivar *devi) { /* Allocate pins and flags memory. */ devi->pins = malloc(sizeof(uint32_t) * devi->npins, M_DEVBUF, M_NOWAIT | M_ZERO); if (devi->pins == NULL) return (ENOMEM); devi->flags = malloc(sizeof(uint32_t) * devi->npins, M_DEVBUF, M_NOWAIT | M_ZERO); if (devi->flags == NULL) { free(devi->pins, M_DEVBUF); return (ENOMEM); } return (0); } void gpiobus_free_ivars(struct gpiobus_ivar *devi) { if (devi->flags) { free(devi->flags, M_DEVBUF); devi->flags = NULL; } if (devi->pins) { free(devi->pins, M_DEVBUF); devi->pins = NULL; } } int -gpiobus_map_pin(device_t bus, device_t child, uint32_t pin) +gpiobus_map_pin(device_t bus, uint32_t pin) { struct gpiobus_softc *sc; sc = device_get_softc(bus); /* Consistency check. */ if (pin >= sc->sc_npins) { - device_printf(child, + device_printf(bus, "invalid pin %d, max: %d\n", pin, sc->sc_npins - 1); return (-1); } /* Mark pin as mapped and give warning if it's already mapped. */ if (sc->sc_pins_mapped[pin]) { - device_printf(child, - "warning: pin %d is already mapped\n", pin); + device_printf(bus, "warning: pin %d is already mapped\n", pin); return (-1); } sc->sc_pins_mapped[pin] = 1; return (0); } static int gpiobus_parse_pins(struct gpiobus_softc *sc, device_t child, int mask) { struct gpiobus_ivar *devi = GPIOBUS_IVAR(child); int i, npins; npins = 0; for (i = 0; i < 32; i++) { if (mask & (1 << i)) npins++; } if (npins == 0) { device_printf(child, "empty pin mask\n"); return (EINVAL); } devi->npins = npins; if (gpiobus_alloc_ivars(devi) != 0) { device_printf(child, "cannot allocate device ivars\n"); return (EINVAL); } npins = 0; for (i = 0; i < 32; i++) { if ((mask & (1 << i)) == 0) continue; /* Reserve the GPIO pin. */ - if (gpiobus_map_pin(sc->sc_busdev, child, i) != 0) { + if (gpiobus_map_pin(sc->sc_busdev, i) != 0) { gpiobus_free_ivars(devi); return (EINVAL); } devi->pins[npins++] = i; } return (0); } static int gpiobus_probe(device_t dev) { device_set_desc(dev, "GPIO bus"); return (BUS_PROBE_GENERIC); } static int gpiobus_attach(device_t dev) { int err; err = gpiobus_init_softc(dev); if (err != 0) return (err); /* * Get parent's pins and mark them as unmapped */ bus_generic_probe(dev); bus_enumerate_hinted_children(dev); return (bus_generic_attach(dev)); } /* * Since this is not a self-enumerating bus, and since we always add * children in attach, we have to always delete children here. */ static int gpiobus_detach(device_t dev) { struct gpiobus_softc *sc; struct gpiobus_ivar *devi; device_t *devlist; int i, err, ndevs; sc = GPIOBUS_SOFTC(dev); KASSERT(mtx_initialized(&sc->sc_mtx), ("gpiobus mutex not initialized")); GPIOBUS_LOCK_DESTROY(sc); if ((err = bus_generic_detach(dev)) != 0) return (err); if ((err = device_get_children(dev, &devlist, &ndevs)) != 0) return (err); for (i = 0; i < ndevs; i++) { device_delete_child(dev, devlist[i]); devi = GPIOBUS_IVAR(devlist[i]); gpiobus_free_ivars(devi); } free(devlist, M_TEMP); if (sc->sc_pins_mapped) { free(sc->sc_pins_mapped, M_DEVBUF); sc->sc_pins_mapped = NULL; } return (0); } static int gpiobus_suspend(device_t dev) { return (bus_generic_suspend(dev)); } static int gpiobus_resume(device_t dev) { return (bus_generic_resume(dev)); } +static void +gpiobus_probe_nomatch(device_t dev, device_t child) +{ + char pins[128]; + struct gpiobus_ivar *devi; + + devi = GPIOBUS_IVAR(child); + memset(pins, 0, sizeof(pins)); + gpiobus_print_pins(devi, pins, sizeof(pins)); + device_printf(dev, " at pin(s) %s", pins); + resource_list_print_type(&devi->rl, "irq", SYS_RES_IRQ, "%ld"); + printf("\n"); +} + static int gpiobus_print_child(device_t dev, device_t child) { char pins[128]; int retval = 0; struct gpiobus_ivar *devi; devi = GPIOBUS_IVAR(child); memset(pins, 0, sizeof(pins)); retval += bus_print_child_header(dev, child); retval += printf(" at pin(s) "); gpiobus_print_pins(devi, pins, sizeof(pins)); retval += printf("%s", pins); resource_list_print_type(&devi->rl, "irq", SYS_RES_IRQ, "%ld"); retval += bus_print_child_footer(dev, child); return (retval); } static int gpiobus_child_location_str(device_t bus, device_t child, char *buf, size_t buflen) { struct gpiobus_ivar *devi; devi = GPIOBUS_IVAR(child); strlcpy(buf, "pin(s)=", buflen); gpiobus_print_pins(devi, buf, buflen); return (0); } static int gpiobus_child_pnpinfo_str(device_t bus, device_t child, char *buf, size_t buflen) { *buf = '\0'; return (0); } static device_t gpiobus_add_child(device_t dev, u_int order, const char *name, int unit) { device_t child; struct gpiobus_ivar *devi; child = device_add_child_ordered(dev, order, name, unit); if (child == NULL) return (child); devi = malloc(sizeof(struct gpiobus_ivar), M_DEVBUF, M_NOWAIT | M_ZERO); if (devi == NULL) { device_delete_child(dev, child); return (0); } resource_list_init(&devi->rl); device_set_ivars(child, devi); return (child); } static void gpiobus_hinted_child(device_t bus, const char *dname, int dunit) { struct gpiobus_softc *sc = GPIOBUS_SOFTC(bus); struct gpiobus_ivar *devi; device_t child; int irq, pins; child = BUS_ADD_CHILD(bus, 0, dname, dunit); devi = GPIOBUS_IVAR(child); resource_int_value(dname, dunit, "pins", &pins); if (gpiobus_parse_pins(sc, child, pins)) device_delete_child(bus, child); if (resource_int_value(dname, dunit, "irq", &irq) == 0) { if (bus_set_resource(child, SYS_RES_IRQ, 0, irq, 1) != 0) device_printf(bus, "warning: bus_set_resource() failed\n"); } } static int gpiobus_set_resource(device_t dev, device_t child, int type, int rid, u_long start, u_long count) { struct gpiobus_ivar *devi; struct resource_list_entry *rle; dprintf("%s: entry (%p, %p, %d, %d, %p, %ld)\n", __func__, dev, child, type, rid, (void *)(intptr_t)start, count); devi = GPIOBUS_IVAR(child); rle = resource_list_add(&devi->rl, type, rid, start, start + count - 1, count); if (rle == NULL) return (ENXIO); return (0); } static struct resource * gpiobus_alloc_resource(device_t bus, device_t child, int type, int *rid, u_long start, u_long end, u_long count, u_int flags) { struct gpiobus_softc *sc; struct resource *rv; struct resource_list *rl; struct resource_list_entry *rle; int isdefault; if (type != SYS_RES_IRQ) return (NULL); isdefault = (start == 0UL && end == ~0UL && count == 1); rle = NULL; if (isdefault) { rl = BUS_GET_RESOURCE_LIST(bus, child); if (rl == NULL) return (NULL); rle = resource_list_find(rl, type, *rid); if (rle == NULL) return (NULL); if (rle->res != NULL) panic("%s: resource entry is busy", __func__); start = rle->start; count = rle->count; end = rle->end; } sc = device_get_softc(bus); rv = rman_reserve_resource(&sc->sc_intr_rman, start, end, count, flags, child); if (rv == NULL) return (NULL); rman_set_rid(rv, *rid); if ((flags & RF_ACTIVE) != 0 && bus_activate_resource(child, type, *rid, rv) != 0) { rman_release_resource(rv); return (NULL); } return (rv); } static int gpiobus_release_resource(device_t bus __unused, device_t child, int type, int rid, struct resource *r) { int error; if (rman_get_flags(r) & RF_ACTIVE) { error = bus_deactivate_resource(child, type, rid, r); if (error) return (error); } return (rman_release_resource(r)); } static struct resource_list * gpiobus_get_resource_list(device_t bus __unused, device_t child) { struct gpiobus_ivar *ivar; ivar = GPIOBUS_IVAR(child); return (&ivar->rl); } static int gpiobus_acquire_bus(device_t busdev, device_t child, int how) { struct gpiobus_softc *sc; sc = device_get_softc(busdev); GPIOBUS_ASSERT_UNLOCKED(sc); GPIOBUS_LOCK(sc); if (sc->sc_owner != NULL) { if (how == GPIOBUS_DONTWAIT) { GPIOBUS_UNLOCK(sc); return (EWOULDBLOCK); } while (sc->sc_owner != NULL) mtx_sleep(sc, &sc->sc_mtx, 0, "gpiobuswait", 0); } sc->sc_owner = child; GPIOBUS_UNLOCK(sc); return (0); } static void gpiobus_release_bus(device_t busdev, device_t child) { struct gpiobus_softc *sc; sc = device_get_softc(busdev); GPIOBUS_ASSERT_UNLOCKED(sc); GPIOBUS_LOCK(sc); if (sc->sc_owner == NULL) panic("gpiobus: releasing unowned bus."); if (sc->sc_owner != child) panic("gpiobus: you don't own the bus. game over."); sc->sc_owner = NULL; wakeup(sc); GPIOBUS_UNLOCK(sc); } static int gpiobus_pin_setflags(device_t dev, device_t child, uint32_t pin, uint32_t flags) { struct gpiobus_softc *sc = GPIOBUS_SOFTC(dev); struct gpiobus_ivar *devi = GPIOBUS_IVAR(child); uint32_t caps; if (pin >= devi->npins) return (EINVAL); if (GPIO_PIN_GETCAPS(sc->sc_dev, devi->pins[pin], &caps) != 0) return (EINVAL); if (gpio_check_flags(caps, flags) != 0) return (EINVAL); return (GPIO_PIN_SETFLAGS(sc->sc_dev, devi->pins[pin], flags)); } static int gpiobus_pin_getflags(device_t dev, device_t child, uint32_t pin, uint32_t *flags) { struct gpiobus_softc *sc = GPIOBUS_SOFTC(dev); struct gpiobus_ivar *devi = GPIOBUS_IVAR(child); if (pin >= devi->npins) return (EINVAL); return GPIO_PIN_GETFLAGS(sc->sc_dev, devi->pins[pin], flags); } static int gpiobus_pin_getcaps(device_t dev, device_t child, uint32_t pin, uint32_t *caps) { struct gpiobus_softc *sc = GPIOBUS_SOFTC(dev); struct gpiobus_ivar *devi = GPIOBUS_IVAR(child); if (pin >= devi->npins) return (EINVAL); return GPIO_PIN_GETCAPS(sc->sc_dev, devi->pins[pin], caps); } static int gpiobus_pin_set(device_t dev, device_t child, uint32_t pin, unsigned int value) { struct gpiobus_softc *sc = GPIOBUS_SOFTC(dev); struct gpiobus_ivar *devi = GPIOBUS_IVAR(child); if (pin >= devi->npins) return (EINVAL); return GPIO_PIN_SET(sc->sc_dev, devi->pins[pin], value); } static int gpiobus_pin_get(device_t dev, device_t child, uint32_t pin, unsigned int *value) { struct gpiobus_softc *sc = GPIOBUS_SOFTC(dev); struct gpiobus_ivar *devi = GPIOBUS_IVAR(child); if (pin >= devi->npins) return (EINVAL); return GPIO_PIN_GET(sc->sc_dev, devi->pins[pin], value); } static int gpiobus_pin_toggle(device_t dev, device_t child, uint32_t pin) { struct gpiobus_softc *sc = GPIOBUS_SOFTC(dev); struct gpiobus_ivar *devi = GPIOBUS_IVAR(child); if (pin >= devi->npins) return (EINVAL); return GPIO_PIN_TOGGLE(sc->sc_dev, devi->pins[pin]); } static device_method_t gpiobus_methods[] = { /* Device interface */ DEVMETHOD(device_probe, gpiobus_probe), DEVMETHOD(device_attach, gpiobus_attach), DEVMETHOD(device_detach, gpiobus_detach), DEVMETHOD(device_shutdown, bus_generic_shutdown), DEVMETHOD(device_suspend, gpiobus_suspend), DEVMETHOD(device_resume, gpiobus_resume), /* Bus interface */ DEVMETHOD(bus_setup_intr, bus_generic_setup_intr), DEVMETHOD(bus_config_intr, bus_generic_config_intr), DEVMETHOD(bus_teardown_intr, bus_generic_teardown_intr), DEVMETHOD(bus_set_resource, gpiobus_set_resource), DEVMETHOD(bus_alloc_resource, gpiobus_alloc_resource), DEVMETHOD(bus_release_resource, gpiobus_release_resource), DEVMETHOD(bus_activate_resource, bus_generic_activate_resource), DEVMETHOD(bus_deactivate_resource, bus_generic_deactivate_resource), DEVMETHOD(bus_get_resource_list, gpiobus_get_resource_list), DEVMETHOD(bus_add_child, gpiobus_add_child), + DEVMETHOD(bus_probe_nomatch, gpiobus_probe_nomatch), DEVMETHOD(bus_print_child, gpiobus_print_child), DEVMETHOD(bus_child_pnpinfo_str, gpiobus_child_pnpinfo_str), DEVMETHOD(bus_child_location_str, gpiobus_child_location_str), DEVMETHOD(bus_hinted_child, gpiobus_hinted_child), /* GPIO protocol */ DEVMETHOD(gpiobus_acquire_bus, gpiobus_acquire_bus), DEVMETHOD(gpiobus_release_bus, gpiobus_release_bus), DEVMETHOD(gpiobus_pin_getflags, gpiobus_pin_getflags), DEVMETHOD(gpiobus_pin_getcaps, gpiobus_pin_getcaps), DEVMETHOD(gpiobus_pin_setflags, gpiobus_pin_setflags), DEVMETHOD(gpiobus_pin_get, gpiobus_pin_get), DEVMETHOD(gpiobus_pin_set, gpiobus_pin_set), DEVMETHOD(gpiobus_pin_toggle, gpiobus_pin_toggle), DEVMETHOD_END }; driver_t gpiobus_driver = { "gpiobus", gpiobus_methods, sizeof(struct gpiobus_softc) }; devclass_t gpiobus_devclass; DRIVER_MODULE(gpiobus, gpio, gpiobus_driver, gpiobus_devclass, 0, 0); MODULE_VERSION(gpiobus, 1); Index: projects/cxl_iscsi/sys/dev/gpio/gpiobusvar.h =================================================================== --- projects/cxl_iscsi/sys/dev/gpio/gpiobusvar.h (revision 279716) +++ projects/cxl_iscsi/sys/dev/gpio/gpiobusvar.h (revision 279717) @@ -1,117 +1,117 @@ /*- * Copyright (c) 2009 Oleksandr Tymoshenko * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ * */ #ifndef __GPIOBUS_H__ #define __GPIOBUS_H__ #include "opt_platform.h" #include #include #include #ifdef FDT #include #endif #include "gpio_if.h" #ifdef FDT #define GPIOBUS_IVAR(d) (struct gpiobus_ivar *) \ &((struct ofw_gpiobus_devinfo *)device_get_ivars(d))->opd_dinfo #else #define GPIOBUS_IVAR(d) (struct gpiobus_ivar *) device_get_ivars(d) #endif #define GPIOBUS_SOFTC(d) (struct gpiobus_softc *) device_get_softc(d) #define GPIOBUS_LOCK(_sc) mtx_lock(&(_sc)->sc_mtx) #define GPIOBUS_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_mtx) #define GPIOBUS_LOCK_INIT(_sc) mtx_init(&_sc->sc_mtx, \ device_get_nameunit(_sc->sc_dev), "gpiobus", MTX_DEF) #define GPIOBUS_LOCK_DESTROY(_sc) mtx_destroy(&_sc->sc_mtx) #define GPIOBUS_ASSERT_LOCKED(_sc) mtx_assert(&_sc->sc_mtx, MA_OWNED) #define GPIOBUS_ASSERT_UNLOCKED(_sc) mtx_assert(&_sc->sc_mtx, MA_NOTOWNED) #define GPIOBUS_WAIT 1 #define GPIOBUS_DONTWAIT 2 struct gpiobus_softc { struct mtx sc_mtx; /* bus mutex */ struct rman sc_intr_rman; /* isr resources */ device_t sc_busdev; /* bus device */ device_t sc_owner; /* bus owner */ device_t sc_dev; /* driver device */ int sc_npins; /* total pins on bus */ int *sc_pins_mapped; /* mark mapped pins */ }; struct gpiobus_pin { device_t dev; /* gpio device */ uint32_t flags; /* pin flags */ uint32_t pin; /* pin number */ }; struct gpiobus_ivar { struct resource_list rl; /* isr resource list */ uint32_t npins; /* pins total */ uint32_t *flags; /* pins flags */ uint32_t *pins; /* pins map */ }; #ifdef FDT struct ofw_gpiobus_devinfo { struct gpiobus_ivar opd_dinfo; struct ofw_bus_devinfo opd_obdinfo; }; static __inline int gpio_map_gpios(device_t bus, phandle_t dev, phandle_t gparent, int gcells, pcell_t *gpios, uint32_t *pin, uint32_t *flags) { return (GPIO_MAP_GPIOS(bus, dev, gparent, gcells, gpios, pin, flags)); } device_t ofw_gpiobus_add_fdt_child(device_t, const char *, phandle_t); int ofw_gpiobus_parse_gpios(device_t, char *, struct gpiobus_pin **); void ofw_gpiobus_register_provider(device_t); void ofw_gpiobus_unregister_provider(device_t); #endif int gpio_check_flags(uint32_t, uint32_t); device_t gpiobus_attach_bus(device_t); int gpiobus_detach_bus(device_t); int gpiobus_init_softc(device_t); int gpiobus_alloc_ivars(struct gpiobus_ivar *); void gpiobus_free_ivars(struct gpiobus_ivar *); -int gpiobus_map_pin(device_t, device_t, uint32_t); +int gpiobus_map_pin(device_t, uint32_t); extern driver_t gpiobus_driver; #endif /* __GPIOBUS_H__ */ Index: projects/cxl_iscsi/sys/dev/gpio/ofw_gpiobus.c =================================================================== --- projects/cxl_iscsi/sys/dev/gpio/ofw_gpiobus.c (revision 279716) +++ projects/cxl_iscsi/sys/dev/gpio/ofw_gpiobus.c (revision 279717) @@ -1,395 +1,403 @@ /*- * Copyright (c) 2009, Nathan Whitehorn * Copyright (c) 2013, Luiz Otavio O Souza * Copyright (c) 2013 The FreeBSD Foundation * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice unmodified, this list of conditions, and the following * disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include static struct ofw_gpiobus_devinfo *ofw_gpiobus_setup_devinfo(device_t, device_t, phandle_t); -static void ofw_gpiobus_destroy_devinfo(struct ofw_gpiobus_devinfo *); +static void ofw_gpiobus_destroy_devinfo(device_t, struct ofw_gpiobus_devinfo *); static int ofw_gpiobus_parse_gpios_impl(device_t, phandle_t, char *, struct gpiobus_softc *, struct gpiobus_pin **); device_t ofw_gpiobus_add_fdt_child(device_t bus, const char *drvname, phandle_t child) { device_t childdev; struct ofw_gpiobus_devinfo *dinfo; /* * Set up the GPIO child and OFW bus layer devinfo and add it to bus. */ childdev = device_add_child(bus, drvname, -1); if (childdev == NULL) return (NULL); dinfo = ofw_gpiobus_setup_devinfo(bus, childdev, child); if (dinfo == NULL) { device_delete_child(bus, childdev); return (NULL); } if (device_probe_and_attach(childdev) != 0) { - ofw_gpiobus_destroy_devinfo(dinfo); + ofw_gpiobus_destroy_devinfo(bus, dinfo); device_delete_child(bus, childdev); return (NULL); } return (childdev); } int ofw_gpiobus_parse_gpios(device_t consumer, char *pname, struct gpiobus_pin **pins) { return (ofw_gpiobus_parse_gpios_impl(consumer, ofw_bus_get_node(consumer), pname, NULL, pins)); } void ofw_gpiobus_register_provider(device_t provider) { phandle_t node; node = ofw_bus_get_node(provider); OF_device_register_xref(OF_xref_from_node(node), provider); } void ofw_gpiobus_unregister_provider(device_t provider) { phandle_t node; node = ofw_bus_get_node(provider); OF_device_register_xref(OF_xref_from_node(node), NULL); } static struct ofw_gpiobus_devinfo * ofw_gpiobus_setup_devinfo(device_t bus, device_t child, phandle_t node) { int i, npins; struct gpiobus_ivar *devi; struct gpiobus_pin *pins; struct gpiobus_softc *sc; struct ofw_gpiobus_devinfo *dinfo; sc = device_get_softc(bus); dinfo = malloc(sizeof(*dinfo), M_DEVBUF, M_NOWAIT | M_ZERO); if (dinfo == NULL) return (NULL); if (ofw_bus_gen_setup_devinfo(&dinfo->opd_obdinfo, node) != 0) { free(dinfo, M_DEVBUF); return (NULL); } /* Parse the gpios property for the child. */ npins = ofw_gpiobus_parse_gpios_impl(child, node, "gpios", sc, &pins); - if (npins <= 0) - goto fail; + if (npins <= 0) { + ofw_bus_gen_destroy_devinfo(&dinfo->opd_obdinfo); + free(dinfo, M_DEVBUF); + return (NULL); + } + /* Initialize the irq resource list. */ + resource_list_init(&dinfo->opd_dinfo.rl); + /* Allocate the child ivars and copy the parsed pin data. */ devi = &dinfo->opd_dinfo; devi->npins = (uint32_t)npins; if (gpiobus_alloc_ivars(devi) != 0) { free(pins, M_DEVBUF); - goto fail; + ofw_gpiobus_destroy_devinfo(bus, dinfo); + return (NULL); } for (i = 0; i < devi->npins; i++) { devi->flags[i] = pins[i].flags; devi->pins[i] = pins[i].pin; } free(pins, M_DEVBUF); - /* And now the interrupt resources. */ - resource_list_init(&dinfo->opd_dinfo.rl); + /* Parse the interrupt resources. */ if (ofw_bus_intr_to_rl(bus, node, &dinfo->opd_dinfo.rl) != 0) { - gpiobus_free_ivars(devi); - goto fail; + ofw_gpiobus_destroy_devinfo(bus, dinfo); + return (NULL); } device_set_ivars(child, dinfo); return (dinfo); - -fail: - ofw_bus_gen_destroy_devinfo(&dinfo->opd_obdinfo); - free(dinfo, M_DEVBUF); - return (NULL); } static void -ofw_gpiobus_destroy_devinfo(struct ofw_gpiobus_devinfo *dinfo) +ofw_gpiobus_destroy_devinfo(device_t bus, struct ofw_gpiobus_devinfo *dinfo) { + int i; struct gpiobus_ivar *devi; + struct gpiobus_softc *sc; + sc = device_get_softc(bus); devi = &dinfo->opd_dinfo; + for (i = 0; i < devi->npins; i++) { + if (devi->pins[i] > sc->sc_npins) + continue; + sc->sc_pins_mapped[devi->pins[i]] = 0; + } gpiobus_free_ivars(devi); resource_list_free(&dinfo->opd_dinfo.rl); ofw_bus_gen_destroy_devinfo(&dinfo->opd_obdinfo); free(dinfo, M_DEVBUF); } static int ofw_gpiobus_parse_gpios_impl(device_t consumer, phandle_t cnode, char *pname, struct gpiobus_softc *bussc, struct gpiobus_pin **pins) { int gpiocells, i, j, ncells, npins; pcell_t *gpios; phandle_t gpio; ncells = OF_getencprop_alloc(cnode, pname, sizeof(*gpios), (void **)&gpios); if (ncells == -1) { device_printf(consumer, "Warning: No %s specified in fdt data; " "device may not function.\n", pname); return (-1); } /* * The gpio-specifier is controller independent, the first pcell has * the reference to the GPIO controller phandler. * Count the number of encoded gpio-specifiers on the first pass. */ i = 0; npins = 0; while (i < ncells) { /* Allow NULL specifiers. */ if (gpios[i] == 0) { npins++; i++; continue; } gpio = OF_node_from_xref(gpios[i]); /* If we have bussc, ignore devices from other gpios. */ if (bussc != NULL) if (ofw_bus_get_node(bussc->sc_dev) != gpio) return (0); /* * Check for gpio-controller property and read the #gpio-cells * for this GPIO controller. */ if (!OF_hasprop(gpio, "gpio-controller") || OF_getencprop(gpio, "#gpio-cells", &gpiocells, sizeof(gpiocells)) < 0) { device_printf(consumer, "gpio reference is not a gpio-controller.\n"); free(gpios, M_OFWPROP); return (-1); } if (ncells - i < gpiocells + 1) { device_printf(consumer, "%s cells doesn't match #gpio-cells.\n", pname); return (-1); } npins++; i += gpiocells + 1; } if (npins == 0 || pins == NULL) { if (npins == 0) device_printf(consumer, "no pin specified in %s.\n", pname); free(gpios, M_OFWPROP); return (npins); } *pins = malloc(sizeof(struct gpiobus_pin) * npins, M_DEVBUF, M_NOWAIT | M_ZERO); if (*pins == NULL) { free(gpios, M_OFWPROP); return (-1); } /* Decode the gpio specifier on the second pass. */ i = 0; j = 0; while (i < ncells) { /* Allow NULL specifiers. */ if (gpios[i] == 0) { j++; i++; continue; } gpio = OF_node_from_xref(gpios[i]); /* Read gpio-cells property for this GPIO controller. */ if (OF_getencprop(gpio, "#gpio-cells", &gpiocells, sizeof(gpiocells)) < 0) { device_printf(consumer, "gpio does not have the #gpio-cells property.\n"); goto fail; } /* Return the device reference for the GPIO controller. */ (*pins)[j].dev = OF_device_from_xref(gpios[i]); if ((*pins)[j].dev == NULL) { device_printf(consumer, "no device registered for the gpio controller.\n"); goto fail; } /* * If the gpiobus softc is NULL we use the GPIO_GET_BUS() to * retrieve it. The GPIO_GET_BUS() method is only valid after * the child is probed and attached. */ if (bussc == NULL) { if (GPIO_GET_BUS((*pins)[j].dev) == NULL) { device_printf(consumer, "no gpiobus reference for %s.\n", device_get_nameunit((*pins)[j].dev)); goto fail; } bussc = device_get_softc(GPIO_GET_BUS((*pins)[j].dev)); } /* Get the GPIO pin number and flags. */ if (gpio_map_gpios((*pins)[j].dev, cnode, gpio, gpiocells, &gpios[i + 1], &(*pins)[j].pin, &(*pins)[j].flags) != 0) { device_printf(consumer, "cannot map the gpios specifier.\n"); goto fail; } /* Reserve the GPIO pin. */ - if (gpiobus_map_pin(bussc->sc_busdev, consumer, - (*pins)[j].pin) != 0) + if (gpiobus_map_pin(bussc->sc_busdev, (*pins)[j].pin) != 0) goto fail; j++; i += gpiocells + 1; } free(gpios, M_OFWPROP); return (npins); fail: free(gpios, M_OFWPROP); free(*pins, M_DEVBUF); return (-1); } static int ofw_gpiobus_probe(device_t dev) { if (ofw_bus_get_node(dev) == -1) return (ENXIO); device_set_desc(dev, "OFW GPIO bus"); return (0); } static int ofw_gpiobus_attach(device_t dev) { int err; phandle_t child; err = gpiobus_init_softc(dev); if (err != 0) return (err); bus_generic_probe(dev); bus_enumerate_hinted_children(dev); /* * Attach the children represented in the device tree. */ for (child = OF_child(ofw_bus_get_node(dev)); child != 0; child = OF_peer(child)) { if (!OF_hasprop(child, "gpios")) continue; if (ofw_gpiobus_add_fdt_child(dev, NULL, child) == NULL) continue; } return (bus_generic_attach(dev)); } static device_t ofw_gpiobus_add_child(device_t dev, u_int order, const char *name, int unit) { device_t child; struct ofw_gpiobus_devinfo *devi; child = device_add_child_ordered(dev, order, name, unit); if (child == NULL) return (child); devi = malloc(sizeof(struct ofw_gpiobus_devinfo), M_DEVBUF, M_NOWAIT | M_ZERO); if (devi == NULL) { device_delete_child(dev, child); return (0); } /* * NULL all the OFW-related parts of the ivars for non-OFW * children. */ devi->opd_obdinfo.obd_node = -1; devi->opd_obdinfo.obd_name = NULL; devi->opd_obdinfo.obd_compat = NULL; devi->opd_obdinfo.obd_type = NULL; devi->opd_obdinfo.obd_model = NULL; device_set_ivars(child, devi); return (child); } static const struct ofw_bus_devinfo * ofw_gpiobus_get_devinfo(device_t bus, device_t dev) { struct ofw_gpiobus_devinfo *dinfo; dinfo = device_get_ivars(dev); return (&dinfo->opd_obdinfo); } static device_method_t ofw_gpiobus_methods[] = { /* Device interface */ DEVMETHOD(device_probe, ofw_gpiobus_probe), DEVMETHOD(device_attach, ofw_gpiobus_attach), /* Bus interface */ DEVMETHOD(bus_child_pnpinfo_str, ofw_bus_gen_child_pnpinfo_str), DEVMETHOD(bus_add_child, ofw_gpiobus_add_child), /* ofw_bus interface */ DEVMETHOD(ofw_bus_get_devinfo, ofw_gpiobus_get_devinfo), DEVMETHOD(ofw_bus_get_compat, ofw_bus_gen_get_compat), DEVMETHOD(ofw_bus_get_model, ofw_bus_gen_get_model), DEVMETHOD(ofw_bus_get_name, ofw_bus_gen_get_name), DEVMETHOD(ofw_bus_get_node, ofw_bus_gen_get_node), DEVMETHOD(ofw_bus_get_type, ofw_bus_gen_get_type), DEVMETHOD_END }; static devclass_t ofwgpiobus_devclass; DEFINE_CLASS_1(gpiobus, ofw_gpiobus_driver, ofw_gpiobus_methods, sizeof(struct gpiobus_softc), gpiobus_driver); DRIVER_MODULE(ofw_gpiobus, gpio, ofw_gpiobus_driver, ofwgpiobus_devclass, 0, 0); MODULE_VERSION(ofw_gpiobus, 1); MODULE_DEPEND(ofw_gpiobus, gpiobus, 1, 1, 1); Index: projects/cxl_iscsi/sys/dev/mps/mps_sas.c =================================================================== --- projects/cxl_iscsi/sys/dev/mps/mps_sas.c (revision 279716) +++ projects/cxl_iscsi/sys/dev/mps/mps_sas.c (revision 279717) @@ -1,3704 +1,3691 @@ /*- * Copyright (c) 2009 Yahoo! Inc. * Copyright (c) 2011-2015 LSI Corp. * Copyright (c) 2013-2015 Avago Technologies * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * Avago Technologies (LSI) MPT-Fusion Host Adapter FreeBSD * * $FreeBSD$ */ #include __FBSDID("$FreeBSD$"); /* Communications core for Avago Technologies (LSI) MPT2 */ /* TODO Move headers to mpsvar */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #if __FreeBSD_version >= 900026 #include #endif #include #include #include #include #include #include #include #include #include #include #include #define MPSSAS_DISCOVERY_TIMEOUT 20 #define MPSSAS_MAX_DISCOVERY_TIMEOUTS 10 /* 200 seconds */ /* * static array to check SCSI OpCode for EEDP protection bits */ #define PRO_R MPI2_SCSIIO_EEDPFLAGS_CHECK_REMOVE_OP #define PRO_W MPI2_SCSIIO_EEDPFLAGS_INSERT_OP #define PRO_V MPI2_SCSIIO_EEDPFLAGS_INSERT_OP static uint8_t op_code_prot[256] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, PRO_R, 0, PRO_W, 0, 0, 0, PRO_W, PRO_V, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, PRO_W, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, PRO_R, 0, PRO_W, 0, 0, 0, PRO_W, PRO_V, 0, 0, 0, PRO_W, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, PRO_R, 0, PRO_W, 0, 0, 0, PRO_W, PRO_V, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; MALLOC_DEFINE(M_MPSSAS, "MPSSAS", "MPS SAS memory"); static void mpssas_remove_device(struct mps_softc *, struct mps_command *); static void mpssas_remove_complete(struct mps_softc *, struct mps_command *); static void mpssas_action(struct cam_sim *sim, union ccb *ccb); static void mpssas_poll(struct cam_sim *sim); static int mpssas_send_abort(struct mps_softc *sc, struct mps_command *tm, struct mps_command *cm); static void mpssas_scsiio_timeout(void *data); static void mpssas_abort_complete(struct mps_softc *sc, struct mps_command *cm); static void mpssas_direct_drive_io(struct mpssas_softc *sassc, struct mps_command *cm, union ccb *ccb); static void mpssas_action_scsiio(struct mpssas_softc *, union ccb *); static void mpssas_scsiio_complete(struct mps_softc *, struct mps_command *); static void mpssas_action_resetdev(struct mpssas_softc *, union ccb *); #if __FreeBSD_version >= 900026 static void mpssas_smpio_complete(struct mps_softc *sc, struct mps_command *cm); static void mpssas_send_smpcmd(struct mpssas_softc *sassc, union ccb *ccb, uint64_t sasaddr); static void mpssas_action_smpio(struct mpssas_softc *sassc, union ccb *ccb); #endif //FreeBSD_version >= 900026 static void mpssas_resetdev_complete(struct mps_softc *, struct mps_command *); static void mpssas_async(void *callback_arg, uint32_t code, struct cam_path *path, void *arg); #if (__FreeBSD_version < 901503) || \ ((__FreeBSD_version >= 1000000) && (__FreeBSD_version < 1000006)) static void mpssas_check_eedp(struct mps_softc *sc, struct cam_path *path, struct ccb_getdev *cgd); static void mpssas_read_cap_done(struct cam_periph *periph, union ccb *done_ccb); #endif static int mpssas_send_portenable(struct mps_softc *sc); static void mpssas_portenable_complete(struct mps_softc *sc, struct mps_command *cm); struct mpssas_target * mpssas_find_target_by_handle(struct mpssas_softc *sassc, int start, uint16_t handle) { struct mpssas_target *target; int i; for (i = start; i < sassc->maxtargets; i++) { target = &sassc->targets[i]; if (target->handle == handle) return (target); } return (NULL); } /* we need to freeze the simq during attach and diag reset, to avoid failing * commands before device handles have been found by discovery. Since * discovery involves reading config pages and possibly sending commands, * discovery actions may continue even after we receive the end of discovery * event, so refcount discovery actions instead of assuming we can unfreeze * the simq when we get the event. */ void mpssas_startup_increment(struct mpssas_softc *sassc) { MPS_FUNCTRACE(sassc->sc); if ((sassc->flags & MPSSAS_IN_STARTUP) != 0) { if (sassc->startup_refcount++ == 0) { /* just starting, freeze the simq */ mps_dprint(sassc->sc, MPS_INIT, "%s freezing simq\n", __func__); #if __FreeBSD_version >= 1000039 xpt_hold_boot(); #endif xpt_freeze_simq(sassc->sim, 1); } mps_dprint(sassc->sc, MPS_INIT, "%s refcount %u\n", __func__, sassc->startup_refcount); } } void mpssas_release_simq_reinit(struct mpssas_softc *sassc) { if (sassc->flags & MPSSAS_QUEUE_FROZEN) { sassc->flags &= ~MPSSAS_QUEUE_FROZEN; xpt_release_simq(sassc->sim, 1); mps_dprint(sassc->sc, MPS_INFO, "Unfreezing SIM queue\n"); } } void mpssas_startup_decrement(struct mpssas_softc *sassc) { MPS_FUNCTRACE(sassc->sc); if ((sassc->flags & MPSSAS_IN_STARTUP) != 0) { if (--sassc->startup_refcount == 0) { /* finished all discovery-related actions, release * the simq and rescan for the latest topology. */ mps_dprint(sassc->sc, MPS_INIT, "%s releasing simq\n", __func__); sassc->flags &= ~MPSSAS_IN_STARTUP; xpt_release_simq(sassc->sim, 1); #if __FreeBSD_version >= 1000039 xpt_release_boot(); #else mpssas_rescan_target(sassc->sc, NULL); #endif } mps_dprint(sassc->sc, MPS_INIT, "%s refcount %u\n", __func__, sassc->startup_refcount); } } /* The firmware requires us to stop sending commands when we're doing task * management, so refcount the TMs and keep the simq frozen when any are in * use. */ struct mps_command * mpssas_alloc_tm(struct mps_softc *sc) { struct mps_command *tm; tm = mps_alloc_high_priority_command(sc); return tm; } void mpssas_free_tm(struct mps_softc *sc, struct mps_command *tm) { if (tm == NULL) return; /* * For TM's the devq is frozen for the device. Unfreeze it here and * free the resources used for freezing the devq. Must clear the * INRESET flag as well or scsi I/O will not work. */ if (tm->cm_targ != NULL) { tm->cm_targ->flags &= ~MPSSAS_TARGET_INRESET; } if (tm->cm_ccb) { mps_dprint(sc, MPS_INFO, "Unfreezing devq for target ID %d\n", tm->cm_targ->tid); xpt_release_devq(tm->cm_ccb->ccb_h.path, 1, TRUE); xpt_free_path(tm->cm_ccb->ccb_h.path); xpt_free_ccb(tm->cm_ccb); } mps_free_high_priority_command(sc, tm); } void mpssas_rescan_target(struct mps_softc *sc, struct mpssas_target *targ) { struct mpssas_softc *sassc = sc->sassc; path_id_t pathid; target_id_t targetid; union ccb *ccb; MPS_FUNCTRACE(sc); pathid = cam_sim_path(sassc->sim); if (targ == NULL) targetid = CAM_TARGET_WILDCARD; else targetid = targ - sassc->targets; /* * Allocate a CCB and schedule a rescan. */ ccb = xpt_alloc_ccb_nowait(); if (ccb == NULL) { mps_dprint(sc, MPS_ERROR, "unable to alloc CCB for rescan\n"); return; } if (xpt_create_path(&ccb->ccb_h.path, NULL, pathid, targetid, CAM_LUN_WILDCARD) != CAM_REQ_CMP) { mps_dprint(sc, MPS_ERROR, "unable to create path for rescan\n"); xpt_free_ccb(ccb); return; } if (targetid == CAM_TARGET_WILDCARD) ccb->ccb_h.func_code = XPT_SCAN_BUS; else ccb->ccb_h.func_code = XPT_SCAN_TGT; mps_dprint(sc, MPS_TRACE, "%s targetid %u\n", __func__, targetid); xpt_rescan(ccb); } static void mpssas_log_command(struct mps_command *cm, u_int level, const char *fmt, ...) { struct sbuf sb; va_list ap; char str[192]; char path_str[64]; if (cm == NULL) return; /* No need to be in here if debugging isn't enabled */ if ((cm->cm_sc->mps_debug & level) == 0) return; sbuf_new(&sb, str, sizeof(str), 0); va_start(ap, fmt); if (cm->cm_ccb != NULL) { xpt_path_string(cm->cm_ccb->csio.ccb_h.path, path_str, sizeof(path_str)); sbuf_cat(&sb, path_str); if (cm->cm_ccb->ccb_h.func_code == XPT_SCSI_IO) { scsi_command_string(&cm->cm_ccb->csio, &sb); sbuf_printf(&sb, "length %d ", cm->cm_ccb->csio.dxfer_len); } } else { sbuf_printf(&sb, "(noperiph:%s%d:%u:%u:%u): ", cam_sim_name(cm->cm_sc->sassc->sim), cam_sim_unit(cm->cm_sc->sassc->sim), cam_sim_bus(cm->cm_sc->sassc->sim), cm->cm_targ ? cm->cm_targ->tid : 0xFFFFFFFF, cm->cm_lun); } sbuf_printf(&sb, "SMID %u ", cm->cm_desc.Default.SMID); sbuf_vprintf(&sb, fmt, ap); sbuf_finish(&sb); mps_dprint_field(cm->cm_sc, level, "%s", sbuf_data(&sb)); va_end(ap); } static void mpssas_remove_volume(struct mps_softc *sc, struct mps_command *tm) { MPI2_SCSI_TASK_MANAGE_REPLY *reply; struct mpssas_target *targ; uint16_t handle; MPS_FUNCTRACE(sc); reply = (MPI2_SCSI_TASK_MANAGE_REPLY *)tm->cm_reply; handle = (uint16_t)(uintptr_t)tm->cm_complete_data; targ = tm->cm_targ; if (reply == NULL) { /* XXX retry the remove after the diag reset completes? */ mps_dprint(sc, MPS_FAULT, "%s NULL reply resetting device 0x%04x\n", __func__, handle); mpssas_free_tm(sc, tm); return; } if (reply->IOCStatus != MPI2_IOCSTATUS_SUCCESS) { mps_dprint(sc, MPS_FAULT, "IOCStatus = 0x%x while resetting device 0x%x\n", reply->IOCStatus, handle); mpssas_free_tm(sc, tm); return; } mps_dprint(sc, MPS_XINFO, "Reset aborted %u commands\n", reply->TerminationCount); mps_free_reply(sc, tm->cm_reply_data); tm->cm_reply = NULL; /* Ensures the reply won't get re-freed */ mps_dprint(sc, MPS_XINFO, "clearing target %u handle 0x%04x\n", targ->tid, handle); /* * Don't clear target if remove fails because things will get confusing. * Leave the devname and sasaddr intact so that we know to avoid reusing * this target id if possible, and so we can assign the same target id * to this device if it comes back in the future. */ if (reply->IOCStatus == MPI2_IOCSTATUS_SUCCESS) { targ = tm->cm_targ; targ->handle = 0x0; targ->encl_handle = 0x0; targ->encl_slot = 0x0; targ->exp_dev_handle = 0x0; targ->phy_num = 0x0; targ->linkrate = 0x0; targ->devinfo = 0x0; targ->flags = 0x0; } mpssas_free_tm(sc, tm); } /* * No Need to call "MPI2_SAS_OP_REMOVE_DEVICE" For Volume removal. * Otherwise Volume Delete is same as Bare Drive Removal. */ void mpssas_prepare_volume_remove(struct mpssas_softc *sassc, uint16_t handle) { MPI2_SCSI_TASK_MANAGE_REQUEST *req; struct mps_softc *sc; struct mps_command *cm; struct mpssas_target *targ = NULL; MPS_FUNCTRACE(sassc->sc); sc = sassc->sc; #ifdef WD_SUPPORT /* * If this is a WD controller, determine if the disk should be exposed * to the OS or not. If disk should be exposed, return from this * function without doing anything. */ if (sc->WD_available && (sc->WD_hide_expose == MPS_WD_EXPOSE_ALWAYS)) { return; } #endif //WD_SUPPORT targ = mpssas_find_target_by_handle(sassc, 0, handle); if (targ == NULL) { /* FIXME: what is the action? */ /* We don't know about this device? */ mps_dprint(sc, MPS_ERROR, "%s %d : invalid handle 0x%x \n", __func__,__LINE__, handle); return; } targ->flags |= MPSSAS_TARGET_INREMOVAL; cm = mpssas_alloc_tm(sc); if (cm == NULL) { mps_dprint(sc, MPS_ERROR, "%s: command alloc failure\n", __func__); return; } mpssas_rescan_target(sc, targ); req = (MPI2_SCSI_TASK_MANAGE_REQUEST *)cm->cm_req; req->DevHandle = targ->handle; req->Function = MPI2_FUNCTION_SCSI_TASK_MGMT; req->TaskType = MPI2_SCSITASKMGMT_TASKTYPE_TARGET_RESET; /* SAS Hard Link Reset / SATA Link Reset */ req->MsgFlags = MPI2_SCSITASKMGMT_MSGFLAGS_LINK_RESET; cm->cm_targ = targ; cm->cm_data = NULL; cm->cm_desc.HighPriority.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY; cm->cm_complete = mpssas_remove_volume; cm->cm_complete_data = (void *)(uintptr_t)handle; mps_dprint(sc, MPS_INFO, "%s: Sending reset for target ID %d\n", __func__, targ->tid); mpssas_prepare_for_tm(sc, cm, targ, CAM_LUN_WILDCARD); mps_map_command(sc, cm); } /* * The MPT2 firmware performs debounce on the link to avoid transient link * errors and false removals. When it does decide that link has been lost * and a device need to go away, it expects that the host will perform a * target reset and then an op remove. The reset has the side-effect of * aborting any outstanding requests for the device, which is required for * the op-remove to succeed. It's not clear if the host should check for * the device coming back alive after the reset. */ void mpssas_prepare_remove(struct mpssas_softc *sassc, uint16_t handle) { MPI2_SCSI_TASK_MANAGE_REQUEST *req; struct mps_softc *sc; struct mps_command *cm; struct mpssas_target *targ = NULL; MPS_FUNCTRACE(sassc->sc); sc = sassc->sc; targ = mpssas_find_target_by_handle(sassc, 0, handle); if (targ == NULL) { /* FIXME: what is the action? */ /* We don't know about this device? */ mps_dprint(sc, MPS_ERROR, "%s : invalid handle 0x%x \n", __func__, handle); return; } targ->flags |= MPSSAS_TARGET_INREMOVAL; cm = mpssas_alloc_tm(sc); if (cm == NULL) { mps_dprint(sc, MPS_ERROR, "%s: command alloc failure\n", __func__); return; } mpssas_rescan_target(sc, targ); req = (MPI2_SCSI_TASK_MANAGE_REQUEST *)cm->cm_req; memset(req, 0, sizeof(*req)); req->DevHandle = htole16(targ->handle); req->Function = MPI2_FUNCTION_SCSI_TASK_MGMT; req->TaskType = MPI2_SCSITASKMGMT_TASKTYPE_TARGET_RESET; /* SAS Hard Link Reset / SATA Link Reset */ req->MsgFlags = MPI2_SCSITASKMGMT_MSGFLAGS_LINK_RESET; cm->cm_targ = targ; cm->cm_data = NULL; cm->cm_desc.HighPriority.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY; cm->cm_complete = mpssas_remove_device; cm->cm_complete_data = (void *)(uintptr_t)handle; mps_dprint(sc, MPS_INFO, "%s: Sending reset for target ID %d\n", __func__, targ->tid); mpssas_prepare_for_tm(sc, cm, targ, CAM_LUN_WILDCARD); mps_map_command(sc, cm); } static void mpssas_remove_device(struct mps_softc *sc, struct mps_command *tm) { MPI2_SCSI_TASK_MANAGE_REPLY *reply; MPI2_SAS_IOUNIT_CONTROL_REQUEST *req; struct mpssas_target *targ; struct mps_command *next_cm; uint16_t handle; MPS_FUNCTRACE(sc); reply = (MPI2_SCSI_TASK_MANAGE_REPLY *)tm->cm_reply; handle = (uint16_t)(uintptr_t)tm->cm_complete_data; targ = tm->cm_targ; /* * Currently there should be no way we can hit this case. It only * happens when we have a failure to allocate chain frames, and * task management commands don't have S/G lists. */ if ((tm->cm_flags & MPS_CM_FLAGS_ERROR_MASK) != 0) { mps_dprint(sc, MPS_ERROR, "%s: cm_flags = %#x for remove of handle %#04x! " "This should not happen!\n", __func__, tm->cm_flags, handle); mpssas_free_tm(sc, tm); return; } if (reply == NULL) { /* XXX retry the remove after the diag reset completes? */ mps_dprint(sc, MPS_FAULT, "%s NULL reply reseting device 0x%04x\n", __func__, handle); mpssas_free_tm(sc, tm); return; } if (le16toh(reply->IOCStatus) != MPI2_IOCSTATUS_SUCCESS) { mps_dprint(sc, MPS_FAULT, "IOCStatus = 0x%x while resetting device 0x%x\n", le16toh(reply->IOCStatus), handle); mpssas_free_tm(sc, tm); return; } mps_dprint(sc, MPS_XINFO, "Reset aborted %u commands\n", le32toh(reply->TerminationCount)); mps_free_reply(sc, tm->cm_reply_data); tm->cm_reply = NULL; /* Ensures the reply won't get re-freed */ /* Reuse the existing command */ req = (MPI2_SAS_IOUNIT_CONTROL_REQUEST *)tm->cm_req; memset(req, 0, sizeof(*req)); req->Function = MPI2_FUNCTION_SAS_IO_UNIT_CONTROL; req->Operation = MPI2_SAS_OP_REMOVE_DEVICE; req->DevHandle = htole16(handle); tm->cm_data = NULL; tm->cm_desc.Default.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_DEFAULT_TYPE; tm->cm_complete = mpssas_remove_complete; tm->cm_complete_data = (void *)(uintptr_t)handle; mps_map_command(sc, tm); mps_dprint(sc, MPS_XINFO, "clearing target %u handle 0x%04x\n", targ->tid, handle); TAILQ_FOREACH_SAFE(tm, &targ->commands, cm_link, next_cm) { union ccb *ccb; mps_dprint(sc, MPS_XINFO, "Completing missed command %p\n", tm); ccb = tm->cm_complete_data; mpssas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE); mpssas_scsiio_complete(sc, tm); } } static void mpssas_remove_complete(struct mps_softc *sc, struct mps_command *tm) { MPI2_SAS_IOUNIT_CONTROL_REPLY *reply; uint16_t handle; struct mpssas_target *targ; struct mpssas_lun *lun; MPS_FUNCTRACE(sc); reply = (MPI2_SAS_IOUNIT_CONTROL_REPLY *)tm->cm_reply; handle = (uint16_t)(uintptr_t)tm->cm_complete_data; /* * Currently there should be no way we can hit this case. It only * happens when we have a failure to allocate chain frames, and * task management commands don't have S/G lists. */ if ((tm->cm_flags & MPS_CM_FLAGS_ERROR_MASK) != 0) { mps_dprint(sc, MPS_XINFO, "%s: cm_flags = %#x for remove of handle %#04x! " "This should not happen!\n", __func__, tm->cm_flags, handle); mpssas_free_tm(sc, tm); return; } if (reply == NULL) { /* most likely a chip reset */ mps_dprint(sc, MPS_FAULT, "%s NULL reply removing device 0x%04x\n", __func__, handle); mpssas_free_tm(sc, tm); return; } mps_dprint(sc, MPS_XINFO, "%s on handle 0x%04x, IOCStatus= 0x%x\n", __func__, handle, le16toh(reply->IOCStatus)); /* * Don't clear target if remove fails because things will get confusing. * Leave the devname and sasaddr intact so that we know to avoid reusing * this target id if possible, and so we can assign the same target id * to this device if it comes back in the future. */ if (le16toh(reply->IOCStatus) == MPI2_IOCSTATUS_SUCCESS) { targ = tm->cm_targ; targ->handle = 0x0; targ->encl_handle = 0x0; targ->encl_slot = 0x0; targ->exp_dev_handle = 0x0; targ->phy_num = 0x0; targ->linkrate = 0x0; targ->devinfo = 0x0; targ->flags = 0x0; while(!SLIST_EMPTY(&targ->luns)) { lun = SLIST_FIRST(&targ->luns); SLIST_REMOVE_HEAD(&targ->luns, lun_link); free(lun, M_MPT2); } } mpssas_free_tm(sc, tm); } static int mpssas_register_events(struct mps_softc *sc) { u32 events[MPI2_EVENT_NOTIFY_EVENTMASK_WORDS]; bzero(events, 16); setbit(events, MPI2_EVENT_SAS_DEVICE_STATUS_CHANGE); setbit(events, MPI2_EVENT_SAS_DISCOVERY); setbit(events, MPI2_EVENT_SAS_BROADCAST_PRIMITIVE); setbit(events, MPI2_EVENT_SAS_INIT_DEVICE_STATUS_CHANGE); setbit(events, MPI2_EVENT_SAS_INIT_TABLE_OVERFLOW); setbit(events, MPI2_EVENT_SAS_TOPOLOGY_CHANGE_LIST); setbit(events, MPI2_EVENT_SAS_ENCL_DEVICE_STATUS_CHANGE); setbit(events, MPI2_EVENT_IR_CONFIGURATION_CHANGE_LIST); setbit(events, MPI2_EVENT_IR_VOLUME); setbit(events, MPI2_EVENT_IR_PHYSICAL_DISK); setbit(events, MPI2_EVENT_IR_OPERATION_STATUS); setbit(events, MPI2_EVENT_LOG_ENTRY_ADDED); mps_register_events(sc, events, mpssas_evt_handler, NULL, &sc->sassc->mpssas_eh); return (0); } int mps_attach_sas(struct mps_softc *sc) { struct mpssas_softc *sassc; cam_status status; int unit, error = 0; MPS_FUNCTRACE(sc); sassc = malloc(sizeof(struct mpssas_softc), M_MPT2, M_WAITOK|M_ZERO); if(!sassc) { device_printf(sc->mps_dev, "Cannot allocate memory %s %d\n", __func__, __LINE__); return (ENOMEM); } /* * XXX MaxTargets could change during a reinit. Since we don't * resize the targets[] array during such an event, cache the value * of MaxTargets here so that we don't get into trouble later. This * should move into the reinit logic. */ sassc->maxtargets = sc->facts->MaxTargets; sassc->targets = malloc(sizeof(struct mpssas_target) * sassc->maxtargets, M_MPT2, M_WAITOK|M_ZERO); if(!sassc->targets) { device_printf(sc->mps_dev, "Cannot allocate memory %s %d\n", __func__, __LINE__); free(sassc, M_MPT2); return (ENOMEM); } sc->sassc = sassc; sassc->sc = sc; if ((sassc->devq = cam_simq_alloc(sc->num_reqs)) == NULL) { mps_dprint(sc, MPS_ERROR, "Cannot allocate SIMQ\n"); error = ENOMEM; goto out; } unit = device_get_unit(sc->mps_dev); sassc->sim = cam_sim_alloc(mpssas_action, mpssas_poll, "mps", sassc, unit, &sc->mps_mtx, sc->num_reqs, sc->num_reqs, sassc->devq); if (sassc->sim == NULL) { mps_dprint(sc, MPS_ERROR, "Cannot allocate SIM\n"); error = EINVAL; goto out; } TAILQ_INIT(&sassc->ev_queue); /* Initialize taskqueue for Event Handling */ TASK_INIT(&sassc->ev_task, 0, mpssas_firmware_event_work, sc); sassc->ev_tq = taskqueue_create("mps_taskq", M_NOWAIT | M_ZERO, taskqueue_thread_enqueue, &sassc->ev_tq); taskqueue_start_threads(&sassc->ev_tq, 1, PRIBIO, "%s taskq", device_get_nameunit(sc->mps_dev)); mps_lock(sc); /* * XXX There should be a bus for every port on the adapter, but since * we're just going to fake the topology for now, we'll pretend that * everything is just a target on a single bus. */ if ((error = xpt_bus_register(sassc->sim, sc->mps_dev, 0)) != 0) { mps_dprint(sc, MPS_ERROR, "Error %d registering SCSI bus\n", error); mps_unlock(sc); goto out; } /* * Assume that discovery events will start right away. * * Hold off boot until discovery is complete. */ sassc->flags |= MPSSAS_IN_STARTUP | MPSSAS_IN_DISCOVERY; sc->sassc->startup_refcount = 0; mpssas_startup_increment(sassc); callout_init(&sassc->discovery_callout, 1 /*mpsafe*/); /* * Register for async events so we can determine the EEDP * capabilities of devices. */ status = xpt_create_path(&sassc->path, /*periph*/NULL, cam_sim_path(sc->sassc->sim), CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD); if (status != CAM_REQ_CMP) { mps_printf(sc, "Error %#x creating sim path\n", status); sassc->path = NULL; } else { int event; #if (__FreeBSD_version >= 1000006) || \ ((__FreeBSD_version >= 901503) && (__FreeBSD_version < 1000000)) event = AC_ADVINFO_CHANGED; #else event = AC_FOUND_DEVICE; #endif status = xpt_register_async(event, mpssas_async, sc, sassc->path); if (status != CAM_REQ_CMP) { mps_dprint(sc, MPS_ERROR, "Error %#x registering async handler for " "AC_ADVINFO_CHANGED events\n", status); xpt_free_path(sassc->path); sassc->path = NULL; } } if (status != CAM_REQ_CMP) { /* * EEDP use is the exception, not the rule. * Warn the user, but do not fail to attach. */ mps_printf(sc, "EEDP capabilities disabled.\n"); } mps_unlock(sc); mpssas_register_events(sc); out: if (error) mps_detach_sas(sc); return (error); } int mps_detach_sas(struct mps_softc *sc) { struct mpssas_softc *sassc; struct mpssas_lun *lun, *lun_tmp; struct mpssas_target *targ; int i; MPS_FUNCTRACE(sc); if (sc->sassc == NULL) return (0); sassc = sc->sassc; mps_deregister_events(sc, sassc->mpssas_eh); /* * Drain and free the event handling taskqueue with the lock * unheld so that any parallel processing tasks drain properly * without deadlocking. */ if (sassc->ev_tq != NULL) taskqueue_free(sassc->ev_tq); /* Make sure CAM doesn't wedge if we had to bail out early. */ mps_lock(sc); /* Deregister our async handler */ if (sassc->path != NULL) { xpt_register_async(0, mpssas_async, sc, sassc->path); xpt_free_path(sassc->path); sassc->path = NULL; } if (sassc->flags & MPSSAS_IN_STARTUP) xpt_release_simq(sassc->sim, 1); if (sassc->sim != NULL) { xpt_bus_deregister(cam_sim_path(sassc->sim)); cam_sim_free(sassc->sim, FALSE); } sassc->flags |= MPSSAS_SHUTDOWN; mps_unlock(sc); if (sassc->devq != NULL) cam_simq_free(sassc->devq); for(i=0; i< sassc->maxtargets ;i++) { targ = &sassc->targets[i]; SLIST_FOREACH_SAFE(lun, &targ->luns, lun_link, lun_tmp) { free(lun, M_MPT2); } } free(sassc->targets, M_MPT2); free(sassc, M_MPT2); sc->sassc = NULL; return (0); } void mpssas_discovery_end(struct mpssas_softc *sassc) { struct mps_softc *sc = sassc->sc; MPS_FUNCTRACE(sc); if (sassc->flags & MPSSAS_DISCOVERY_TIMEOUT_PENDING) callout_stop(&sassc->discovery_callout); } static void mpssas_action(struct cam_sim *sim, union ccb *ccb) { struct mpssas_softc *sassc; sassc = cam_sim_softc(sim); MPS_FUNCTRACE(sassc->sc); mps_dprint(sassc->sc, MPS_TRACE, "ccb func_code 0x%x\n", ccb->ccb_h.func_code); mtx_assert(&sassc->sc->mps_mtx, MA_OWNED); switch (ccb->ccb_h.func_code) { case XPT_PATH_INQ: { struct ccb_pathinq *cpi = &ccb->cpi; cpi->version_num = 1; cpi->hba_inquiry = PI_SDTR_ABLE|PI_TAG_ABLE|PI_WIDE_16; cpi->target_sprt = 0; #if __FreeBSD_version >= 1000039 cpi->hba_misc = PIM_NOBUSRESET | PIM_UNMAPPED | PIM_NOSCAN; #else cpi->hba_misc = PIM_NOBUSRESET | PIM_UNMAPPED; #endif cpi->hba_eng_cnt = 0; cpi->max_target = sassc->maxtargets - 1; cpi->max_lun = 255; cpi->initiator_id = sassc->maxtargets - 1; strncpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN); strncpy(cpi->hba_vid, "Avago Tech (LSI)", HBA_IDLEN); strncpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN); cpi->unit_number = cam_sim_unit(sim); cpi->bus_id = cam_sim_bus(sim); cpi->base_transfer_speed = 150000; cpi->transport = XPORT_SAS; cpi->transport_version = 0; cpi->protocol = PROTO_SCSI; cpi->protocol_version = SCSI_REV_SPC; #if __FreeBSD_version >= 800001 /* * XXX KDM where does this number come from? */ cpi->maxio = 256 * 1024; #endif mpssas_set_ccbstatus(ccb, CAM_REQ_CMP); break; } case XPT_GET_TRAN_SETTINGS: { struct ccb_trans_settings *cts; struct ccb_trans_settings_sas *sas; struct ccb_trans_settings_scsi *scsi; struct mpssas_target *targ; cts = &ccb->cts; sas = &cts->xport_specific.sas; scsi = &cts->proto_specific.scsi; KASSERT(cts->ccb_h.target_id < sassc->maxtargets, ("Target %d out of bounds in XPT_GET_TRANS_SETTINGS\n", cts->ccb_h.target_id)); targ = &sassc->targets[cts->ccb_h.target_id]; if (targ->handle == 0x0) { mpssas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE); break; } cts->protocol_version = SCSI_REV_SPC2; cts->transport = XPORT_SAS; cts->transport_version = 0; sas->valid = CTS_SAS_VALID_SPEED; switch (targ->linkrate) { case 0x08: sas->bitrate = 150000; break; case 0x09: sas->bitrate = 300000; break; case 0x0a: sas->bitrate = 600000; break; default: sas->valid = 0; } cts->protocol = PROTO_SCSI; scsi->valid = CTS_SCSI_VALID_TQ; scsi->flags = CTS_SCSI_FLAGS_TAG_ENB; mpssas_set_ccbstatus(ccb, CAM_REQ_CMP); break; } case XPT_CALC_GEOMETRY: cam_calc_geometry(&ccb->ccg, /*extended*/1); mpssas_set_ccbstatus(ccb, CAM_REQ_CMP); break; case XPT_RESET_DEV: mps_dprint(sassc->sc, MPS_XINFO, "mpssas_action XPT_RESET_DEV\n"); mpssas_action_resetdev(sassc, ccb); return; case XPT_RESET_BUS: case XPT_ABORT: case XPT_TERM_IO: mps_dprint(sassc->sc, MPS_XINFO, "mpssas_action faking success for abort or reset\n"); mpssas_set_ccbstatus(ccb, CAM_REQ_CMP); break; case XPT_SCSI_IO: mpssas_action_scsiio(sassc, ccb); return; #if __FreeBSD_version >= 900026 case XPT_SMP_IO: mpssas_action_smpio(sassc, ccb); return; #endif default: mpssas_set_ccbstatus(ccb, CAM_FUNC_NOTAVAIL); break; } xpt_done(ccb); } static void mpssas_announce_reset(struct mps_softc *sc, uint32_t ac_code, target_id_t target_id, lun_id_t lun_id) { path_id_t path_id = cam_sim_path(sc->sassc->sim); struct cam_path *path; mps_dprint(sc, MPS_XINFO, "%s code %x target %d lun %jx\n", __func__, ac_code, target_id, (uintmax_t)lun_id); if (xpt_create_path(&path, NULL, path_id, target_id, lun_id) != CAM_REQ_CMP) { mps_dprint(sc, MPS_ERROR, "unable to create path for reset " "notification\n"); return; } xpt_async(ac_code, path, NULL); xpt_free_path(path); } static void mpssas_complete_all_commands(struct mps_softc *sc) { struct mps_command *cm; int i; int completed; MPS_FUNCTRACE(sc); mtx_assert(&sc->mps_mtx, MA_OWNED); /* complete all commands with a NULL reply */ for (i = 1; i < sc->num_reqs; i++) { cm = &sc->commands[i]; cm->cm_reply = NULL; completed = 0; if (cm->cm_flags & MPS_CM_FLAGS_POLLED) cm->cm_flags |= MPS_CM_FLAGS_COMPLETE; if (cm->cm_complete != NULL) { mpssas_log_command(cm, MPS_RECOVERY, "completing cm %p state %x ccb %p for diag reset\n", cm, cm->cm_state, cm->cm_ccb); cm->cm_complete(sc, cm); completed = 1; } if (cm->cm_flags & MPS_CM_FLAGS_WAKEUP) { mpssas_log_command(cm, MPS_RECOVERY, "waking up cm %p state %x ccb %p for diag reset\n", cm, cm->cm_state, cm->cm_ccb); wakeup(cm); completed = 1; } if (cm->cm_sc->io_cmds_active != 0) { cm->cm_sc->io_cmds_active--; } else { mps_dprint(cm->cm_sc, MPS_INFO, "Warning: " "io_cmds_active is out of sync - resynching to " "0\n"); } if ((completed == 0) && (cm->cm_state != MPS_CM_STATE_FREE)) { /* this should never happen, but if it does, log */ mpssas_log_command(cm, MPS_RECOVERY, "cm %p state %x flags 0x%x ccb %p during diag " "reset\n", cm, cm->cm_state, cm->cm_flags, cm->cm_ccb); } } } void mpssas_handle_reinit(struct mps_softc *sc) { int i; /* Go back into startup mode and freeze the simq, so that CAM * doesn't send any commands until after we've rediscovered all * targets and found the proper device handles for them. * * After the reset, portenable will trigger discovery, and after all * discovery-related activities have finished, the simq will be * released. */ mps_dprint(sc, MPS_INIT, "%s startup\n", __func__); sc->sassc->flags |= MPSSAS_IN_STARTUP; sc->sassc->flags |= MPSSAS_IN_DISCOVERY; mpssas_startup_increment(sc->sassc); /* notify CAM of a bus reset */ mpssas_announce_reset(sc, AC_BUS_RESET, CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD); /* complete and cleanup after all outstanding commands */ mpssas_complete_all_commands(sc); mps_dprint(sc, MPS_INIT, "%s startup %u after command completion\n", __func__, sc->sassc->startup_refcount); /* zero all the target handles, since they may change after the * reset, and we have to rediscover all the targets and use the new * handles. */ for (i = 0; i < sc->sassc->maxtargets; i++) { if (sc->sassc->targets[i].outstanding != 0) mps_dprint(sc, MPS_INIT, "target %u outstanding %u\n", i, sc->sassc->targets[i].outstanding); sc->sassc->targets[i].handle = 0x0; sc->sassc->targets[i].exp_dev_handle = 0x0; sc->sassc->targets[i].outstanding = 0; sc->sassc->targets[i].flags = MPSSAS_TARGET_INDIAGRESET; } } static void mpssas_tm_timeout(void *data) { struct mps_command *tm = data; struct mps_softc *sc = tm->cm_sc; mtx_assert(&sc->mps_mtx, MA_OWNED); mpssas_log_command(tm, MPS_INFO|MPS_RECOVERY, "task mgmt %p timed out\n", tm); mps_reinit(sc); } static void mpssas_logical_unit_reset_complete(struct mps_softc *sc, struct mps_command *tm) { MPI2_SCSI_TASK_MANAGE_REPLY *reply; MPI2_SCSI_TASK_MANAGE_REQUEST *req; unsigned int cm_count = 0; struct mps_command *cm; struct mpssas_target *targ; callout_stop(&tm->cm_callout); req = (MPI2_SCSI_TASK_MANAGE_REQUEST *)tm->cm_req; reply = (MPI2_SCSI_TASK_MANAGE_REPLY *)tm->cm_reply; targ = tm->cm_targ; /* * Currently there should be no way we can hit this case. It only * happens when we have a failure to allocate chain frames, and * task management commands don't have S/G lists. * XXXSL So should it be an assertion? */ if ((tm->cm_flags & MPS_CM_FLAGS_ERROR_MASK) != 0) { mps_dprint(sc, MPS_ERROR, "%s: cm_flags = %#x for LUN reset! " "This should not happen!\n", __func__, tm->cm_flags); mpssas_free_tm(sc, tm); return; } if (reply == NULL) { mpssas_log_command(tm, MPS_RECOVERY, "NULL reset reply for tm %p\n", tm); if ((sc->mps_flags & MPS_FLAGS_DIAGRESET) != 0) { /* this completion was due to a reset, just cleanup */ targ->tm = NULL; mpssas_free_tm(sc, tm); } else { /* we should have gotten a reply. */ mps_reinit(sc); } return; } mpssas_log_command(tm, MPS_RECOVERY, "logical unit reset status 0x%x code 0x%x count %u\n", le16toh(reply->IOCStatus), le32toh(reply->ResponseCode), le32toh(reply->TerminationCount)); /* See if there are any outstanding commands for this LUN. * This could be made more efficient by using a per-LU data * structure of some sort. */ TAILQ_FOREACH(cm, &targ->commands, cm_link) { if (cm->cm_lun == tm->cm_lun) cm_count++; } if (cm_count == 0) { mpssas_log_command(tm, MPS_RECOVERY|MPS_INFO, "logical unit %u finished recovery after reset\n", tm->cm_lun, tm); mpssas_announce_reset(sc, AC_SENT_BDR, tm->cm_targ->tid, tm->cm_lun); /* we've finished recovery for this logical unit. check and * see if some other logical unit has a timedout command * that needs to be processed. */ cm = TAILQ_FIRST(&targ->timedout_commands); if (cm) { mpssas_send_abort(sc, tm, cm); } else { targ->tm = NULL; mpssas_free_tm(sc, tm); } } else { /* if we still have commands for this LUN, the reset * effectively failed, regardless of the status reported. * Escalate to a target reset. */ mpssas_log_command(tm, MPS_RECOVERY, "logical unit reset complete for tm %p, but still have %u command(s)\n", tm, cm_count); mpssas_send_reset(sc, tm, MPI2_SCSITASKMGMT_TASKTYPE_TARGET_RESET); } } static void mpssas_target_reset_complete(struct mps_softc *sc, struct mps_command *tm) { MPI2_SCSI_TASK_MANAGE_REPLY *reply; MPI2_SCSI_TASK_MANAGE_REQUEST *req; struct mpssas_target *targ; callout_stop(&tm->cm_callout); req = (MPI2_SCSI_TASK_MANAGE_REQUEST *)tm->cm_req; reply = (MPI2_SCSI_TASK_MANAGE_REPLY *)tm->cm_reply; targ = tm->cm_targ; /* * Currently there should be no way we can hit this case. It only * happens when we have a failure to allocate chain frames, and * task management commands don't have S/G lists. */ if ((tm->cm_flags & MPS_CM_FLAGS_ERROR_MASK) != 0) { mps_dprint(sc, MPS_ERROR,"%s: cm_flags = %#x for target reset! " "This should not happen!\n", __func__, tm->cm_flags); mpssas_free_tm(sc, tm); return; } if (reply == NULL) { mpssas_log_command(tm, MPS_RECOVERY, "NULL reset reply for tm %p\n", tm); if ((sc->mps_flags & MPS_FLAGS_DIAGRESET) != 0) { /* this completion was due to a reset, just cleanup */ targ->tm = NULL; mpssas_free_tm(sc, tm); } else { /* we should have gotten a reply. */ mps_reinit(sc); } return; } mpssas_log_command(tm, MPS_RECOVERY, "target reset status 0x%x code 0x%x count %u\n", le16toh(reply->IOCStatus), le32toh(reply->ResponseCode), le32toh(reply->TerminationCount)); if (targ->outstanding == 0) { /* we've finished recovery for this target and all * of its logical units. */ mpssas_log_command(tm, MPS_RECOVERY|MPS_INFO, "recovery finished after target reset\n"); mpssas_announce_reset(sc, AC_SENT_BDR, tm->cm_targ->tid, CAM_LUN_WILDCARD); targ->tm = NULL; mpssas_free_tm(sc, tm); } else { /* after a target reset, if this target still has * outstanding commands, the reset effectively failed, * regardless of the status reported. escalate. */ mpssas_log_command(tm, MPS_RECOVERY, "target reset complete for tm %p, but still have %u command(s)\n", tm, targ->outstanding); mps_reinit(sc); } } #define MPS_RESET_TIMEOUT 30 int mpssas_send_reset(struct mps_softc *sc, struct mps_command *tm, uint8_t type) { MPI2_SCSI_TASK_MANAGE_REQUEST *req; struct mpssas_target *target; int err; target = tm->cm_targ; if (target->handle == 0) { mps_dprint(sc, MPS_ERROR,"%s null devhandle for target_id %d\n", __func__, target->tid); return -1; } req = (MPI2_SCSI_TASK_MANAGE_REQUEST *)tm->cm_req; req->DevHandle = htole16(target->handle); req->Function = MPI2_FUNCTION_SCSI_TASK_MGMT; req->TaskType = type; if (type == MPI2_SCSITASKMGMT_TASKTYPE_LOGICAL_UNIT_RESET) { /* XXX Need to handle invalid LUNs */ MPS_SET_LUN(req->LUN, tm->cm_lun); tm->cm_targ->logical_unit_resets++; mpssas_log_command(tm, MPS_RECOVERY|MPS_INFO, "sending logical unit reset\n"); tm->cm_complete = mpssas_logical_unit_reset_complete; mpssas_prepare_for_tm(sc, tm, target, tm->cm_lun); } else if (type == MPI2_SCSITASKMGMT_TASKTYPE_TARGET_RESET) { /* * Target reset method = * SAS Hard Link Reset / SATA Link Reset */ req->MsgFlags = MPI2_SCSITASKMGMT_MSGFLAGS_LINK_RESET; tm->cm_targ->target_resets++; mpssas_log_command(tm, MPS_RECOVERY|MPS_INFO, "sending target reset\n"); tm->cm_complete = mpssas_target_reset_complete; mpssas_prepare_for_tm(sc, tm, target, CAM_LUN_WILDCARD); } else { mps_dprint(sc, MPS_ERROR, "unexpected reset type 0x%x\n", type); return -1; } tm->cm_data = NULL; tm->cm_desc.HighPriority.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY; tm->cm_complete_data = (void *)tm; callout_reset(&tm->cm_callout, MPS_RESET_TIMEOUT * hz, mpssas_tm_timeout, tm); err = mps_map_command(sc, tm); if (err) mpssas_log_command(tm, MPS_RECOVERY, "error %d sending reset type %u\n", err, type); return err; } static void mpssas_abort_complete(struct mps_softc *sc, struct mps_command *tm) { struct mps_command *cm; MPI2_SCSI_TASK_MANAGE_REPLY *reply; MPI2_SCSI_TASK_MANAGE_REQUEST *req; struct mpssas_target *targ; callout_stop(&tm->cm_callout); req = (MPI2_SCSI_TASK_MANAGE_REQUEST *)tm->cm_req; reply = (MPI2_SCSI_TASK_MANAGE_REPLY *)tm->cm_reply; targ = tm->cm_targ; /* * Currently there should be no way we can hit this case. It only * happens when we have a failure to allocate chain frames, and * task management commands don't have S/G lists. */ if ((tm->cm_flags & MPS_CM_FLAGS_ERROR_MASK) != 0) { mpssas_log_command(tm, MPS_RECOVERY, "cm_flags = %#x for abort %p TaskMID %u!\n", tm->cm_flags, tm, le16toh(req->TaskMID)); mpssas_free_tm(sc, tm); return; } if (reply == NULL) { mpssas_log_command(tm, MPS_RECOVERY, "NULL abort reply for tm %p TaskMID %u\n", tm, le16toh(req->TaskMID)); if ((sc->mps_flags & MPS_FLAGS_DIAGRESET) != 0) { /* this completion was due to a reset, just cleanup */ targ->tm = NULL; mpssas_free_tm(sc, tm); } else { /* we should have gotten a reply. */ mps_reinit(sc); } return; } mpssas_log_command(tm, MPS_RECOVERY, "abort TaskMID %u status 0x%x code 0x%x count %u\n", le16toh(req->TaskMID), le16toh(reply->IOCStatus), le32toh(reply->ResponseCode), le32toh(reply->TerminationCount)); cm = TAILQ_FIRST(&tm->cm_targ->timedout_commands); if (cm == NULL) { /* if there are no more timedout commands, we're done with * error recovery for this target. */ mpssas_log_command(tm, MPS_RECOVERY, "finished recovery after aborting TaskMID %u\n", le16toh(req->TaskMID)); targ->tm = NULL; mpssas_free_tm(sc, tm); } else if (le16toh(req->TaskMID) != cm->cm_desc.Default.SMID) { /* abort success, but we have more timedout commands to abort */ mpssas_log_command(tm, MPS_RECOVERY, "continuing recovery after aborting TaskMID %u\n", le16toh(req->TaskMID)); mpssas_send_abort(sc, tm, cm); } else { /* we didn't get a command completion, so the abort * failed as far as we're concerned. escalate. */ mpssas_log_command(tm, MPS_RECOVERY, "abort failed for TaskMID %u tm %p\n", le16toh(req->TaskMID), tm); mpssas_send_reset(sc, tm, MPI2_SCSITASKMGMT_TASKTYPE_LOGICAL_UNIT_RESET); } } #define MPS_ABORT_TIMEOUT 5 static int mpssas_send_abort(struct mps_softc *sc, struct mps_command *tm, struct mps_command *cm) { MPI2_SCSI_TASK_MANAGE_REQUEST *req; struct mpssas_target *targ; int err; targ = cm->cm_targ; if (targ->handle == 0) { mps_dprint(sc, MPS_ERROR,"%s null devhandle for target_id %d\n", __func__, cm->cm_ccb->ccb_h.target_id); return -1; } mpssas_log_command(tm, MPS_RECOVERY|MPS_INFO, "Aborting command %p\n", cm); req = (MPI2_SCSI_TASK_MANAGE_REQUEST *)tm->cm_req; req->DevHandle = htole16(targ->handle); req->Function = MPI2_FUNCTION_SCSI_TASK_MGMT; req->TaskType = MPI2_SCSITASKMGMT_TASKTYPE_ABORT_TASK; /* XXX Need to handle invalid LUNs */ MPS_SET_LUN(req->LUN, cm->cm_ccb->ccb_h.target_lun); req->TaskMID = htole16(cm->cm_desc.Default.SMID); tm->cm_data = NULL; tm->cm_desc.HighPriority.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY; tm->cm_complete = mpssas_abort_complete; tm->cm_complete_data = (void *)tm; tm->cm_targ = cm->cm_targ; tm->cm_lun = cm->cm_lun; callout_reset(&tm->cm_callout, MPS_ABORT_TIMEOUT * hz, mpssas_tm_timeout, tm); targ->aborts++; mps_dprint(sc, MPS_INFO, "Sending reset from %s for target ID %d\n", __func__, targ->tid); mpssas_prepare_for_tm(sc, tm, targ, tm->cm_lun); err = mps_map_command(sc, tm); if (err) mpssas_log_command(tm, MPS_RECOVERY, "error %d sending abort for cm %p SMID %u\n", err, cm, req->TaskMID); return err; } static void mpssas_scsiio_timeout(void *data) { struct mps_softc *sc; struct mps_command *cm; struct mpssas_target *targ; cm = (struct mps_command *)data; sc = cm->cm_sc; MPS_FUNCTRACE(sc); mtx_assert(&sc->mps_mtx, MA_OWNED); mps_dprint(sc, MPS_XINFO, "Timeout checking cm %p\n", sc); /* * Run the interrupt handler to make sure it's not pending. This * isn't perfect because the command could have already completed * and been re-used, though this is unlikely. */ mps_intr_locked(sc); if (cm->cm_state == MPS_CM_STATE_FREE) { mpssas_log_command(cm, MPS_XINFO, "SCSI command %p almost timed out\n", cm); return; } if (cm->cm_ccb == NULL) { mps_dprint(sc, MPS_ERROR, "command timeout with NULL ccb\n"); return; } mpssas_log_command(cm, MPS_INFO, "command timeout cm %p ccb %p\n", cm, cm->cm_ccb); targ = cm->cm_targ; targ->timeouts++; /* XXX first, check the firmware state, to see if it's still * operational. if not, do a diag reset. */ mpssas_set_ccbstatus(cm->cm_ccb, CAM_CMD_TIMEOUT); cm->cm_state = MPS_CM_STATE_TIMEDOUT; TAILQ_INSERT_TAIL(&targ->timedout_commands, cm, cm_recovery); if (targ->tm != NULL) { /* target already in recovery, just queue up another * timedout command to be processed later. */ mps_dprint(sc, MPS_RECOVERY, "queued timedout cm %p for processing by tm %p\n", cm, targ->tm); } else if ((targ->tm = mpssas_alloc_tm(sc)) != NULL) { mps_dprint(sc, MPS_RECOVERY, "timedout cm %p allocated tm %p\n", cm, targ->tm); /* start recovery by aborting the first timedout command */ mpssas_send_abort(sc, targ->tm, cm); } else { /* XXX queue this target up for recovery once a TM becomes * available. The firmware only has a limited number of * HighPriority credits for the high priority requests used * for task management, and we ran out. * * Isilon: don't worry about this for now, since we have * more credits than disks in an enclosure, and limit * ourselves to one TM per target for recovery. */ mps_dprint(sc, MPS_RECOVERY, "timedout cm %p failed to allocate a tm\n", cm); } } static void mpssas_action_scsiio(struct mpssas_softc *sassc, union ccb *ccb) { MPI2_SCSI_IO_REQUEST *req; struct ccb_scsiio *csio; struct mps_softc *sc; struct mpssas_target *targ; struct mpssas_lun *lun; struct mps_command *cm; uint8_t i, lba_byte, *ref_tag_addr; uint16_t eedp_flags; uint32_t mpi_control; sc = sassc->sc; MPS_FUNCTRACE(sc); mtx_assert(&sc->mps_mtx, MA_OWNED); csio = &ccb->csio; KASSERT(csio->ccb_h.target_id < sassc->maxtargets, ("Target %d out of bounds in XPT_SCSI_IO\n", csio->ccb_h.target_id)); targ = &sassc->targets[csio->ccb_h.target_id]; mps_dprint(sc, MPS_TRACE, "ccb %p target flag %x\n", ccb, targ->flags); if (targ->handle == 0x0) { mps_dprint(sc, MPS_ERROR, "%s NULL handle for target %u\n", __func__, csio->ccb_h.target_id); mpssas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE); xpt_done(ccb); return; } if (targ->flags & MPS_TARGET_FLAGS_RAID_COMPONENT) { mps_dprint(sc, MPS_ERROR, "%s Raid component no SCSI IO " "supported %u\n", __func__, csio->ccb_h.target_id); mpssas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE); xpt_done(ccb); return; } /* * Sometimes, it is possible to get a command that is not "In * Progress" and was actually aborted by the upper layer. Check for * this here and complete the command without error. */ if (mpssas_get_ccbstatus(ccb) != CAM_REQ_INPROG) { mps_dprint(sc, MPS_TRACE, "%s Command is not in progress for " "target %u\n", __func__, csio->ccb_h.target_id); xpt_done(ccb); return; } /* * If devinfo is 0 this will be a volume. In that case don't tell CAM * that the volume has timed out. We want volumes to be enumerated * until they are deleted/removed, not just failed. */ if (targ->flags & MPSSAS_TARGET_INREMOVAL) { if (targ->devinfo == 0) mpssas_set_ccbstatus(ccb, CAM_REQ_CMP); else mpssas_set_ccbstatus(ccb, CAM_SEL_TIMEOUT); xpt_done(ccb); return; } if ((sc->mps_flags & MPS_FLAGS_SHUTDOWN) != 0) { mps_dprint(sc, MPS_INFO, "%s shutting down\n", __func__); mpssas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE); xpt_done(ccb); return; } /* * If target has a reset in progress, freeze the devq and return. The * devq will be released when the TM reset is finished. */ if (targ->flags & MPSSAS_TARGET_INRESET) { ccb->ccb_h.status = CAM_BUSY | CAM_DEV_QFRZN; mps_dprint(sc, MPS_INFO, "%s: Freezing devq for target ID %d\n", __func__, targ->tid); xpt_freeze_devq(ccb->ccb_h.path, 1); xpt_done(ccb); return; } cm = mps_alloc_command(sc); if (cm == NULL || (sc->mps_flags & MPS_FLAGS_DIAGRESET)) { if (cm != NULL) { mps_free_command(sc, cm); } if ((sassc->flags & MPSSAS_QUEUE_FROZEN) == 0) { xpt_freeze_simq(sassc->sim, 1); sassc->flags |= MPSSAS_QUEUE_FROZEN; } ccb->ccb_h.status &= ~CAM_SIM_QUEUED; ccb->ccb_h.status |= CAM_REQUEUE_REQ; xpt_done(ccb); return; } req = (MPI2_SCSI_IO_REQUEST *)cm->cm_req; bzero(req, sizeof(*req)); req->DevHandle = htole16(targ->handle); req->Function = MPI2_FUNCTION_SCSI_IO_REQUEST; req->MsgFlags = 0; req->SenseBufferLowAddress = htole32(cm->cm_sense_busaddr); req->SenseBufferLength = MPS_SENSE_LEN; req->SGLFlags = 0; req->ChainOffset = 0; req->SGLOffset0 = 24; /* 32bit word offset to the SGL */ req->SGLOffset1= 0; req->SGLOffset2= 0; req->SGLOffset3= 0; req->SkipCount = 0; req->DataLength = htole32(csio->dxfer_len); req->BidirectionalDataLength = 0; req->IoFlags = htole16(csio->cdb_len); req->EEDPFlags = 0; /* Note: BiDirectional transfers are not supported */ switch (csio->ccb_h.flags & CAM_DIR_MASK) { case CAM_DIR_IN: mpi_control = MPI2_SCSIIO_CONTROL_READ; cm->cm_flags |= MPS_CM_FLAGS_DATAIN; break; case CAM_DIR_OUT: mpi_control = MPI2_SCSIIO_CONTROL_WRITE; cm->cm_flags |= MPS_CM_FLAGS_DATAOUT; break; case CAM_DIR_NONE: default: mpi_control = MPI2_SCSIIO_CONTROL_NODATATRANSFER; break; } if (csio->cdb_len == 32) mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; /* * It looks like the hardware doesn't require an explicit tag * number for each transaction. SAM Task Management not supported * at the moment. */ switch (csio->tag_action) { case MSG_HEAD_OF_Q_TAG: mpi_control |= MPI2_SCSIIO_CONTROL_HEADOFQ; break; case MSG_ORDERED_Q_TAG: mpi_control |= MPI2_SCSIIO_CONTROL_ORDEREDQ; break; case MSG_ACA_TASK: mpi_control |= MPI2_SCSIIO_CONTROL_ACAQ; break; case CAM_TAG_ACTION_NONE: case MSG_SIMPLE_Q_TAG: default: mpi_control |= MPI2_SCSIIO_CONTROL_SIMPLEQ; break; } mpi_control |= sc->mapping_table[csio->ccb_h.target_id].TLR_bits; req->Control = htole32(mpi_control); if (MPS_SET_LUN(req->LUN, csio->ccb_h.target_lun) != 0) { mps_free_command(sc, cm); mpssas_set_ccbstatus(ccb, CAM_LUN_INVALID); xpt_done(ccb); return; } if (csio->ccb_h.flags & CAM_CDB_POINTER) bcopy(csio->cdb_io.cdb_ptr, &req->CDB.CDB32[0], csio->cdb_len); else bcopy(csio->cdb_io.cdb_bytes, &req->CDB.CDB32[0],csio->cdb_len); req->IoFlags = htole16(csio->cdb_len); /* * Check if EEDP is supported and enabled. If it is then check if the * SCSI opcode could be using EEDP. If so, make sure the LUN exists and * is formatted for EEDP support. If all of this is true, set CDB up * for EEDP transfer. */ eedp_flags = op_code_prot[req->CDB.CDB32[0]]; if (sc->eedp_enabled && eedp_flags) { SLIST_FOREACH(lun, &targ->luns, lun_link) { if (lun->lun_id == csio->ccb_h.target_lun) { break; } } if ((lun != NULL) && (lun->eedp_formatted)) { req->EEDPBlockSize = htole16(lun->eedp_block_size); eedp_flags |= (MPI2_SCSIIO_EEDPFLAGS_INC_PRI_REFTAG | MPI2_SCSIIO_EEDPFLAGS_CHECK_REFTAG | MPI2_SCSIIO_EEDPFLAGS_CHECK_GUARD); req->EEDPFlags = htole16(eedp_flags); /* * If CDB less than 32, fill in Primary Ref Tag with * low 4 bytes of LBA. If CDB is 32, tag stuff is * already there. Also, set protection bit. FreeBSD * currently does not support CDBs bigger than 16, but * the code doesn't hurt, and will be here for the * future. */ if (csio->cdb_len != 32) { lba_byte = (csio->cdb_len == 16) ? 6 : 2; ref_tag_addr = (uint8_t *)&req->CDB.EEDP32. PrimaryReferenceTag; for (i = 0; i < 4; i++) { *ref_tag_addr = req->CDB.CDB32[lba_byte + i]; ref_tag_addr++; } req->CDB.EEDP32.PrimaryReferenceTag = htole32(req->CDB.EEDP32.PrimaryReferenceTag); req->CDB.EEDP32.PrimaryApplicationTagMask = 0xFFFF; req->CDB.CDB32[1] = (req->CDB.CDB32[1] & 0x1F) | 0x20; } else { eedp_flags |= MPI2_SCSIIO_EEDPFLAGS_INC_PRI_APPTAG; req->EEDPFlags = htole16(eedp_flags); req->CDB.CDB32[10] = (req->CDB.CDB32[10] & 0x1F) | 0x20; } } } cm->cm_length = csio->dxfer_len; if (cm->cm_length != 0) { cm->cm_data = ccb; cm->cm_flags |= MPS_CM_FLAGS_USE_CCB; } else { cm->cm_data = NULL; } cm->cm_sge = &req->SGL; cm->cm_sglsize = (32 - 24) * 4; cm->cm_desc.SCSIIO.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_SCSI_IO; cm->cm_desc.SCSIIO.DevHandle = htole16(targ->handle); cm->cm_complete = mpssas_scsiio_complete; cm->cm_complete_data = ccb; cm->cm_targ = targ; cm->cm_lun = csio->ccb_h.target_lun; cm->cm_ccb = ccb; /* * If HBA is a WD and the command is not for a retry, try to build a * direct I/O message. If failed, or the command is for a retry, send * the I/O to the IR volume itself. */ if (sc->WD_valid_config) { if (ccb->ccb_h.sim_priv.entries[0].field == MPS_WD_RETRY) { mpssas_direct_drive_io(sassc, cm, ccb); } else { mpssas_set_ccbstatus(ccb, CAM_REQ_INPROG); } } callout_reset_sbt(&cm->cm_callout, SBT_1MS * ccb->ccb_h.timeout, 0, mpssas_scsiio_timeout, cm, 0); targ->issued++; targ->outstanding++; TAILQ_INSERT_TAIL(&targ->commands, cm, cm_link); ccb->ccb_h.status |= CAM_SIM_QUEUED; mpssas_log_command(cm, MPS_XINFO, "%s cm %p ccb %p outstanding %u\n", __func__, cm, ccb, targ->outstanding); mps_map_command(sc, cm); return; } static void mps_response_code(struct mps_softc *sc, u8 response_code) { char *desc; switch (response_code) { case MPI2_SCSITASKMGMT_RSP_TM_COMPLETE: desc = "task management request completed"; break; case MPI2_SCSITASKMGMT_RSP_INVALID_FRAME: desc = "invalid frame"; break; case MPI2_SCSITASKMGMT_RSP_TM_NOT_SUPPORTED: desc = "task management request not supported"; break; case MPI2_SCSITASKMGMT_RSP_TM_FAILED: desc = "task management request failed"; break; case MPI2_SCSITASKMGMT_RSP_TM_SUCCEEDED: desc = "task management request succeeded"; break; case MPI2_SCSITASKMGMT_RSP_TM_INVALID_LUN: desc = "invalid lun"; break; case 0xA: desc = "overlapped tag attempted"; break; case MPI2_SCSITASKMGMT_RSP_IO_QUEUED_ON_IOC: desc = "task queued, however not sent to target"; break; default: desc = "unknown"; break; } mps_dprint(sc, MPS_XINFO, "response_code(0x%01x): %s\n", response_code, desc); } /** * mps_sc_failed_io_info - translated non-succesfull SCSI_IO request */ static void mps_sc_failed_io_info(struct mps_softc *sc, struct ccb_scsiio *csio, Mpi2SCSIIOReply_t *mpi_reply) { u32 response_info; u8 *response_bytes; u16 ioc_status = le16toh(mpi_reply->IOCStatus) & MPI2_IOCSTATUS_MASK; u8 scsi_state = mpi_reply->SCSIState; u8 scsi_status = mpi_reply->SCSIStatus; char *desc_ioc_state = NULL; char *desc_scsi_status = NULL; char *desc_scsi_state = sc->tmp_string; u32 log_info = le32toh(mpi_reply->IOCLogInfo); if (log_info == 0x31170000) return; switch (ioc_status) { case MPI2_IOCSTATUS_SUCCESS: desc_ioc_state = "success"; break; case MPI2_IOCSTATUS_INVALID_FUNCTION: desc_ioc_state = "invalid function"; break; case MPI2_IOCSTATUS_SCSI_RECOVERED_ERROR: desc_ioc_state = "scsi recovered error"; break; case MPI2_IOCSTATUS_SCSI_INVALID_DEVHANDLE: desc_ioc_state = "scsi invalid dev handle"; break; case MPI2_IOCSTATUS_SCSI_DEVICE_NOT_THERE: desc_ioc_state = "scsi device not there"; break; case MPI2_IOCSTATUS_SCSI_DATA_OVERRUN: desc_ioc_state = "scsi data overrun"; break; case MPI2_IOCSTATUS_SCSI_DATA_UNDERRUN: desc_ioc_state = "scsi data underrun"; break; case MPI2_IOCSTATUS_SCSI_IO_DATA_ERROR: desc_ioc_state = "scsi io data error"; break; case MPI2_IOCSTATUS_SCSI_PROTOCOL_ERROR: desc_ioc_state = "scsi protocol error"; break; case MPI2_IOCSTATUS_SCSI_TASK_TERMINATED: desc_ioc_state = "scsi task terminated"; break; case MPI2_IOCSTATUS_SCSI_RESIDUAL_MISMATCH: desc_ioc_state = "scsi residual mismatch"; break; case MPI2_IOCSTATUS_SCSI_TASK_MGMT_FAILED: desc_ioc_state = "scsi task mgmt failed"; break; case MPI2_IOCSTATUS_SCSI_IOC_TERMINATED: desc_ioc_state = "scsi ioc terminated"; break; case MPI2_IOCSTATUS_SCSI_EXT_TERMINATED: desc_ioc_state = "scsi ext terminated"; break; case MPI2_IOCSTATUS_EEDP_GUARD_ERROR: desc_ioc_state = "eedp guard error"; break; case MPI2_IOCSTATUS_EEDP_REF_TAG_ERROR: desc_ioc_state = "eedp ref tag error"; break; case MPI2_IOCSTATUS_EEDP_APP_TAG_ERROR: desc_ioc_state = "eedp app tag error"; break; default: desc_ioc_state = "unknown"; break; } switch (scsi_status) { case MPI2_SCSI_STATUS_GOOD: desc_scsi_status = "good"; break; case MPI2_SCSI_STATUS_CHECK_CONDITION: desc_scsi_status = "check condition"; break; case MPI2_SCSI_STATUS_CONDITION_MET: desc_scsi_status = "condition met"; break; case MPI2_SCSI_STATUS_BUSY: desc_scsi_status = "busy"; break; case MPI2_SCSI_STATUS_INTERMEDIATE: desc_scsi_status = "intermediate"; break; case MPI2_SCSI_STATUS_INTERMEDIATE_CONDMET: desc_scsi_status = "intermediate condmet"; break; case MPI2_SCSI_STATUS_RESERVATION_CONFLICT: desc_scsi_status = "reservation conflict"; break; case MPI2_SCSI_STATUS_COMMAND_TERMINATED: desc_scsi_status = "command terminated"; break; case MPI2_SCSI_STATUS_TASK_SET_FULL: desc_scsi_status = "task set full"; break; case MPI2_SCSI_STATUS_ACA_ACTIVE: desc_scsi_status = "aca active"; break; case MPI2_SCSI_STATUS_TASK_ABORTED: desc_scsi_status = "task aborted"; break; default: desc_scsi_status = "unknown"; break; } desc_scsi_state[0] = '\0'; if (!scsi_state) desc_scsi_state = " "; if (scsi_state & MPI2_SCSI_STATE_RESPONSE_INFO_VALID) strcat(desc_scsi_state, "response info "); if (scsi_state & MPI2_SCSI_STATE_TERMINATED) strcat(desc_scsi_state, "state terminated "); if (scsi_state & MPI2_SCSI_STATE_NO_SCSI_STATUS) strcat(desc_scsi_state, "no status "); if (scsi_state & MPI2_SCSI_STATE_AUTOSENSE_FAILED) strcat(desc_scsi_state, "autosense failed "); if (scsi_state & MPI2_SCSI_STATE_AUTOSENSE_VALID) strcat(desc_scsi_state, "autosense valid "); mps_dprint(sc, MPS_XINFO, "\thandle(0x%04x), ioc_status(%s)(0x%04x)\n", le16toh(mpi_reply->DevHandle), desc_ioc_state, ioc_status); /* We can add more detail about underflow data here * TO-DO * */ mps_dprint(sc, MPS_XINFO, "\tscsi_status(%s)(0x%02x), " "scsi_state(%s)(0x%02x)\n", desc_scsi_status, scsi_status, desc_scsi_state, scsi_state); if (sc->mps_debug & MPS_XINFO && scsi_state & MPI2_SCSI_STATE_AUTOSENSE_VALID) { mps_dprint(sc, MPS_XINFO, "-> Sense Buffer Data : Start :\n"); scsi_sense_print(csio); mps_dprint(sc, MPS_XINFO, "-> Sense Buffer Data : End :\n"); } if (scsi_state & MPI2_SCSI_STATE_RESPONSE_INFO_VALID) { response_info = le32toh(mpi_reply->ResponseInfo); response_bytes = (u8 *)&response_info; mps_response_code(sc,response_bytes[0]); } } static void mpssas_scsiio_complete(struct mps_softc *sc, struct mps_command *cm) { MPI2_SCSI_IO_REPLY *rep; union ccb *ccb; struct ccb_scsiio *csio; struct mpssas_softc *sassc; struct scsi_vpd_supported_page_list *vpd_list = NULL; u8 *TLR_bits, TLR_on; int dir = 0, i; u16 alloc_len; struct mpssas_target *target; target_id_t target_id; MPS_FUNCTRACE(sc); mps_dprint(sc, MPS_TRACE, "cm %p SMID %u ccb %p reply %p outstanding %u\n", cm, cm->cm_desc.Default.SMID, cm->cm_ccb, cm->cm_reply, cm->cm_targ->outstanding); callout_stop(&cm->cm_callout); mtx_assert(&sc->mps_mtx, MA_OWNED); sassc = sc->sassc; ccb = cm->cm_complete_data; csio = &ccb->csio; target_id = csio->ccb_h.target_id; rep = (MPI2_SCSI_IO_REPLY *)cm->cm_reply; /* * XXX KDM if the chain allocation fails, does it matter if we do * the sync and unload here? It is simpler to do it in every case, * assuming it doesn't cause problems. */ if (cm->cm_data != NULL) { if (cm->cm_flags & MPS_CM_FLAGS_DATAIN) dir = BUS_DMASYNC_POSTREAD; else if (cm->cm_flags & MPS_CM_FLAGS_DATAOUT) dir = BUS_DMASYNC_POSTWRITE; bus_dmamap_sync(sc->buffer_dmat, cm->cm_dmamap, dir); bus_dmamap_unload(sc->buffer_dmat, cm->cm_dmamap); } cm->cm_targ->completed++; cm->cm_targ->outstanding--; TAILQ_REMOVE(&cm->cm_targ->commands, cm, cm_link); ccb->ccb_h.status &= ~(CAM_STATUS_MASK | CAM_SIM_QUEUED); if (cm->cm_state == MPS_CM_STATE_TIMEDOUT) { TAILQ_REMOVE(&cm->cm_targ->timedout_commands, cm, cm_recovery); if (cm->cm_reply != NULL) mpssas_log_command(cm, MPS_RECOVERY, "completed timedout cm %p ccb %p during recovery " "ioc %x scsi %x state %x xfer %u\n", cm, cm->cm_ccb, le16toh(rep->IOCStatus), rep->SCSIStatus, rep->SCSIState, le32toh(rep->TransferCount)); else mpssas_log_command(cm, MPS_RECOVERY, "completed timedout cm %p ccb %p during recovery\n", cm, cm->cm_ccb); } else if (cm->cm_targ->tm != NULL) { if (cm->cm_reply != NULL) mpssas_log_command(cm, MPS_RECOVERY, "completed cm %p ccb %p during recovery " "ioc %x scsi %x state %x xfer %u\n", cm, cm->cm_ccb, le16toh(rep->IOCStatus), rep->SCSIStatus, rep->SCSIState, le32toh(rep->TransferCount)); else mpssas_log_command(cm, MPS_RECOVERY, "completed cm %p ccb %p during recovery\n", cm, cm->cm_ccb); } else if ((sc->mps_flags & MPS_FLAGS_DIAGRESET) != 0) { mpssas_log_command(cm, MPS_RECOVERY, "reset completed cm %p ccb %p\n", cm, cm->cm_ccb); } if ((cm->cm_flags & MPS_CM_FLAGS_ERROR_MASK) != 0) { /* * We ran into an error after we tried to map the command, * so we're getting a callback without queueing the command * to the hardware. So we set the status here, and it will * be retained below. We'll go through the "fast path", * because there can be no reply when we haven't actually * gone out to the hardware. */ mpssas_set_ccbstatus(ccb, CAM_REQUEUE_REQ); /* * Currently the only error included in the mask is * MPS_CM_FLAGS_CHAIN_FAILED, which means we're out of * chain frames. We need to freeze the queue until we get * a command that completed without this error, which will * hopefully have some chain frames attached that we can * use. If we wanted to get smarter about it, we would * only unfreeze the queue in this condition when we're * sure that we're getting some chain frames back. That's * probably unnecessary. */ if ((sassc->flags & MPSSAS_QUEUE_FROZEN) == 0) { xpt_freeze_simq(sassc->sim, 1); sassc->flags |= MPSSAS_QUEUE_FROZEN; mps_dprint(sc, MPS_XINFO, "Error sending command, " "freezing SIM queue\n"); } } /* * If this is a Start Stop Unit command and it was issued by the driver * during shutdown, decrement the refcount to account for all of the * commands that were sent. All SSU commands should be completed before * shutdown completes, meaning SSU_refcount will be 0 after SSU_started * is TRUE. */ if (sc->SSU_started && (csio->cdb_io.cdb_bytes[0] == START_STOP_UNIT)) { mps_dprint(sc, MPS_INFO, "Decrementing SSU count.\n"); sc->SSU_refcount--; } /* Take the fast path to completion */ if (cm->cm_reply == NULL) { if (mpssas_get_ccbstatus(ccb) == CAM_REQ_INPROG) { if ((sc->mps_flags & MPS_FLAGS_DIAGRESET) != 0) mpssas_set_ccbstatus(ccb, CAM_SCSI_BUS_RESET); else { mpssas_set_ccbstatus(ccb, CAM_REQ_CMP); ccb->csio.scsi_status = SCSI_STATUS_OK; } if (sassc->flags & MPSSAS_QUEUE_FROZEN) { ccb->ccb_h.status |= CAM_RELEASE_SIMQ; sassc->flags &= ~MPSSAS_QUEUE_FROZEN; mps_dprint(sc, MPS_XINFO, "Unfreezing SIM queue\n"); } } /* * There are two scenarios where the status won't be * CAM_REQ_CMP. The first is if MPS_CM_FLAGS_ERROR_MASK is * set, the second is in the MPS_FLAGS_DIAGRESET above. */ if (mpssas_get_ccbstatus(ccb) != CAM_REQ_CMP) { /* * Freeze the dev queue so that commands are * executed in the correct order after error * recovery. */ ccb->ccb_h.status |= CAM_DEV_QFRZN; xpt_freeze_devq(ccb->ccb_h.path, /*count*/ 1); } mps_free_command(sc, cm); xpt_done(ccb); return; } mpssas_log_command(cm, MPS_XINFO, "ioc %x scsi %x state %x xfer %u\n", le16toh(rep->IOCStatus), rep->SCSIStatus, rep->SCSIState, le32toh(rep->TransferCount)); /* * If this is a Direct Drive I/O, reissue the I/O to the original IR * Volume if an error occurred (normal I/O retry). Use the original * CCB, but set a flag that this will be a retry so that it's sent to * the original volume. Free the command but reuse the CCB. */ if (cm->cm_flags & MPS_CM_FLAGS_DD_IO) { mps_free_command(sc, cm); ccb->ccb_h.sim_priv.entries[0].field = MPS_WD_RETRY; mpssas_action_scsiio(sassc, ccb); return; } else ccb->ccb_h.sim_priv.entries[0].field = 0; switch (le16toh(rep->IOCStatus) & MPI2_IOCSTATUS_MASK) { case MPI2_IOCSTATUS_SCSI_DATA_UNDERRUN: csio->resid = cm->cm_length - le32toh(rep->TransferCount); /* FALLTHROUGH */ case MPI2_IOCSTATUS_SUCCESS: case MPI2_IOCSTATUS_SCSI_RECOVERED_ERROR: if ((le16toh(rep->IOCStatus) & MPI2_IOCSTATUS_MASK) == MPI2_IOCSTATUS_SCSI_RECOVERED_ERROR) mpssas_log_command(cm, MPS_XINFO, "recovered error\n"); /* Completion failed at the transport level. */ if (rep->SCSIState & (MPI2_SCSI_STATE_NO_SCSI_STATUS | MPI2_SCSI_STATE_TERMINATED)) { mpssas_set_ccbstatus(ccb, CAM_REQ_CMP_ERR); break; } /* In a modern packetized environment, an autosense failure * implies that there's not much else that can be done to * recover the command. */ if (rep->SCSIState & MPI2_SCSI_STATE_AUTOSENSE_FAILED) { mpssas_set_ccbstatus(ccb, CAM_AUTOSENSE_FAIL); break; } /* * CAM doesn't care about SAS Response Info data, but if this is * the state check if TLR should be done. If not, clear the * TLR_bits for the target. */ if ((rep->SCSIState & MPI2_SCSI_STATE_RESPONSE_INFO_VALID) && ((le32toh(rep->ResponseInfo) & MPI2_SCSI_RI_MASK_REASONCODE) == MPS_SCSI_RI_INVALID_FRAME)) { sc->mapping_table[target_id].TLR_bits = (u8)MPI2_SCSIIO_CONTROL_NO_TLR; } /* * Intentionally override the normal SCSI status reporting * for these two cases. These are likely to happen in a * multi-initiator environment, and we want to make sure that * CAM retries these commands rather than fail them. */ if ((rep->SCSIStatus == MPI2_SCSI_STATUS_COMMAND_TERMINATED) || (rep->SCSIStatus == MPI2_SCSI_STATUS_TASK_ABORTED)) { mpssas_set_ccbstatus(ccb, CAM_REQ_ABORTED); break; } /* Handle normal status and sense */ csio->scsi_status = rep->SCSIStatus; if (rep->SCSIStatus == MPI2_SCSI_STATUS_GOOD) mpssas_set_ccbstatus(ccb, CAM_REQ_CMP); else mpssas_set_ccbstatus(ccb, CAM_SCSI_STATUS_ERROR); if (rep->SCSIState & MPI2_SCSI_STATE_AUTOSENSE_VALID) { int sense_len, returned_sense_len; returned_sense_len = min(le32toh(rep->SenseCount), sizeof(struct scsi_sense_data)); if (returned_sense_len < ccb->csio.sense_len) ccb->csio.sense_resid = ccb->csio.sense_len - returned_sense_len; else ccb->csio.sense_resid = 0; sense_len = min(returned_sense_len, ccb->csio.sense_len - ccb->csio.sense_resid); bzero(&ccb->csio.sense_data, sizeof(ccb->csio.sense_data)); bcopy(cm->cm_sense, &ccb->csio.sense_data, sense_len); ccb->ccb_h.status |= CAM_AUTOSNS_VALID; } /* * Check if this is an INQUIRY command. If it's a VPD inquiry, * and it's page code 0 (Supported Page List), and there is * inquiry data, and this is for a sequential access device, and * the device is an SSP target, and TLR is supported by the * controller, turn the TLR_bits value ON if page 0x90 is * supported. */ if ((csio->cdb_io.cdb_bytes[0] == INQUIRY) && (csio->cdb_io.cdb_bytes[1] & SI_EVPD) && (csio->cdb_io.cdb_bytes[2] == SVPD_SUPPORTED_PAGE_LIST) && ((csio->ccb_h.flags & CAM_DATA_MASK) == CAM_DATA_VADDR) && (csio->data_ptr != NULL) && ((csio->data_ptr[0] & 0x1f) == T_SEQUENTIAL) && (sc->control_TLR) && (sc->mapping_table[target_id].device_info & MPI2_SAS_DEVICE_INFO_SSP_TARGET)) { vpd_list = (struct scsi_vpd_supported_page_list *) csio->data_ptr; TLR_bits = &sc->mapping_table[target_id].TLR_bits; *TLR_bits = (u8)MPI2_SCSIIO_CONTROL_NO_TLR; TLR_on = (u8)MPI2_SCSIIO_CONTROL_TLR_ON; alloc_len = ((u16)csio->cdb_io.cdb_bytes[3] << 8) + csio->cdb_io.cdb_bytes[4]; alloc_len -= csio->resid; for (i = 0; i < MIN(vpd_list->length, alloc_len); i++) { if (vpd_list->list[i] == 0x90) { *TLR_bits = TLR_on; break; } } } /* * If this is a SATA direct-access end device, mark it so that * a SCSI StartStopUnit command will be sent to it when the * driver is being shutdown. */ if ((csio->cdb_io.cdb_bytes[0] == INQUIRY) && ((csio->data_ptr[0] & 0x1f) == T_DIRECT) && (sc->mapping_table[target_id].device_info & MPI2_SAS_DEVICE_INFO_SATA_DEVICE) && ((sc->mapping_table[target_id].device_info & MPI2_SAS_DEVICE_INFO_MASK_DEVICE_TYPE) == MPI2_SAS_DEVICE_INFO_END_DEVICE)) { target = &sassc->targets[target_id]; target->supports_SSU = TRUE; mps_dprint(sc, MPS_XINFO, "Target %d supports SSU\n", target_id); } break; case MPI2_IOCSTATUS_SCSI_INVALID_DEVHANDLE: case MPI2_IOCSTATUS_SCSI_DEVICE_NOT_THERE: /* * If devinfo is 0 this will be a volume. In that case don't * tell CAM that the volume is not there. We want volumes to * be enumerated until they are deleted/removed, not just * failed. */ if (cm->cm_targ->devinfo == 0) mpssas_set_ccbstatus(ccb, CAM_REQ_CMP); else mpssas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE); break; case MPI2_IOCSTATUS_INVALID_SGL: mps_print_scsiio_cmd(sc, cm); mpssas_set_ccbstatus(ccb, CAM_UNREC_HBA_ERROR); break; case MPI2_IOCSTATUS_SCSI_TASK_TERMINATED: /* * This is one of the responses that comes back when an I/O * has been aborted. If it is because of a timeout that we * initiated, just set the status to CAM_CMD_TIMEOUT. * Otherwise set it to CAM_REQ_ABORTED. The effect on the * command is the same (it gets retried, subject to the * retry counter), the only difference is what gets printed * on the console. */ if (cm->cm_state == MPS_CM_STATE_TIMEDOUT) mpssas_set_ccbstatus(ccb, CAM_CMD_TIMEOUT); else mpssas_set_ccbstatus(ccb, CAM_REQ_ABORTED); break; case MPI2_IOCSTATUS_SCSI_DATA_OVERRUN: /* resid is ignored for this condition */ csio->resid = 0; mpssas_set_ccbstatus(ccb, CAM_DATA_RUN_ERR); break; case MPI2_IOCSTATUS_SCSI_IOC_TERMINATED: case MPI2_IOCSTATUS_SCSI_EXT_TERMINATED: /* * Since these are generally external (i.e. hopefully * transient transport-related) errors, retry these without * decrementing the retry count. */ mpssas_set_ccbstatus(ccb, CAM_REQUEUE_REQ); mpssas_log_command(cm, MPS_INFO, "terminated ioc %x scsi %x state %x xfer %u\n", le16toh(rep->IOCStatus), rep->SCSIStatus, rep->SCSIState, le32toh(rep->TransferCount)); break; case MPI2_IOCSTATUS_INVALID_FUNCTION: case MPI2_IOCSTATUS_INTERNAL_ERROR: case MPI2_IOCSTATUS_INVALID_VPID: case MPI2_IOCSTATUS_INVALID_FIELD: case MPI2_IOCSTATUS_INVALID_STATE: case MPI2_IOCSTATUS_OP_STATE_NOT_SUPPORTED: case MPI2_IOCSTATUS_SCSI_IO_DATA_ERROR: case MPI2_IOCSTATUS_SCSI_PROTOCOL_ERROR: case MPI2_IOCSTATUS_SCSI_RESIDUAL_MISMATCH: case MPI2_IOCSTATUS_SCSI_TASK_MGMT_FAILED: default: mpssas_log_command(cm, MPS_XINFO, "completed ioc %x scsi %x state %x xfer %u\n", le16toh(rep->IOCStatus), rep->SCSIStatus, rep->SCSIState, le32toh(rep->TransferCount)); csio->resid = cm->cm_length; mpssas_set_ccbstatus(ccb, CAM_REQ_CMP_ERR); break; } mps_sc_failed_io_info(sc,csio,rep); if (sassc->flags & MPSSAS_QUEUE_FROZEN) { ccb->ccb_h.status |= CAM_RELEASE_SIMQ; sassc->flags &= ~MPSSAS_QUEUE_FROZEN; mps_dprint(sc, MPS_XINFO, "Command completed, " "unfreezing SIM queue\n"); } if (mpssas_get_ccbstatus(ccb) != CAM_REQ_CMP) { ccb->ccb_h.status |= CAM_DEV_QFRZN; xpt_freeze_devq(ccb->ccb_h.path, /*count*/ 1); } mps_free_command(sc, cm); xpt_done(ccb); } /* All Request reached here are Endian safe */ static void mpssas_direct_drive_io(struct mpssas_softc *sassc, struct mps_command *cm, union ccb *ccb) { pMpi2SCSIIORequest_t pIO_req; struct mps_softc *sc = sassc->sc; uint64_t virtLBA; uint32_t physLBA, stripe_offset, stripe_unit; uint32_t io_size, column; uint8_t *ptrLBA, lba_idx, physLBA_byte, *CDB; /* * If this is a valid SCSI command (Read6, Read10, Read16, Write6, * Write10, or Write16), build a direct I/O message. Otherwise, the I/O * will be sent to the IR volume itself. Since Read6 and Write6 are a * bit different than the 10/16 CDBs, handle them separately. */ pIO_req = (pMpi2SCSIIORequest_t)cm->cm_req; CDB = pIO_req->CDB.CDB32; /* * Handle 6 byte CDBs. */ if ((pIO_req->DevHandle == sc->DD_dev_handle) && ((CDB[0] == READ_6) || (CDB[0] == WRITE_6))) { /* * Get the transfer size in blocks. */ io_size = (cm->cm_length >> sc->DD_block_exponent); /* * Get virtual LBA given in the CDB. */ virtLBA = ((uint64_t)(CDB[1] & 0x1F) << 16) | ((uint64_t)CDB[2] << 8) | (uint64_t)CDB[3]; /* * Check that LBA range for I/O does not exceed volume's * MaxLBA. */ if ((virtLBA + (uint64_t)io_size - 1) <= sc->DD_max_lba) { /* * Check if the I/O crosses a stripe boundary. If not, * translate the virtual LBA to a physical LBA and set * the DevHandle for the PhysDisk to be used. If it * does cross a boundry, do normal I/O. To get the * right DevHandle to use, get the map number for the * column, then use that map number to look up the * DevHandle of the PhysDisk. */ stripe_offset = (uint32_t)virtLBA & (sc->DD_stripe_size - 1); if ((stripe_offset + io_size) <= sc->DD_stripe_size) { physLBA = (uint32_t)virtLBA >> sc->DD_stripe_exponent; stripe_unit = physLBA / sc->DD_num_phys_disks; column = physLBA % sc->DD_num_phys_disks; pIO_req->DevHandle = htole16(sc->DD_column_map[column].dev_handle); /* ???? Is this endian safe*/ cm->cm_desc.SCSIIO.DevHandle = pIO_req->DevHandle; physLBA = (stripe_unit << sc->DD_stripe_exponent) + stripe_offset; ptrLBA = &pIO_req->CDB.CDB32[1]; physLBA_byte = (uint8_t)(physLBA >> 16); *ptrLBA = physLBA_byte; ptrLBA = &pIO_req->CDB.CDB32[2]; physLBA_byte = (uint8_t)(physLBA >> 8); *ptrLBA = physLBA_byte; ptrLBA = &pIO_req->CDB.CDB32[3]; physLBA_byte = (uint8_t)physLBA; *ptrLBA = physLBA_byte; /* * Set flag that Direct Drive I/O is * being done. */ cm->cm_flags |= MPS_CM_FLAGS_DD_IO; } } return; } /* * Handle 10, 12 or 16 byte CDBs. */ if ((pIO_req->DevHandle == sc->DD_dev_handle) && ((CDB[0] == READ_10) || (CDB[0] == WRITE_10) || (CDB[0] == READ_16) || (CDB[0] == WRITE_16) || (CDB[0] == READ_12) || (CDB[0] == WRITE_12))) { /* * For 16-byte CDB's, verify that the upper 4 bytes of the CDB * are 0. If not, this is accessing beyond 2TB so handle it in * the else section. 10-byte and 12-byte CDB's are OK. * FreeBSD sends very rare 12 byte READ/WRITE, but driver is * ready to accept 12byte CDB for Direct IOs. */ if ((CDB[0] == READ_10 || CDB[0] == WRITE_10) || (CDB[0] == READ_12 || CDB[0] == WRITE_12) || !(CDB[2] | CDB[3] | CDB[4] | CDB[5])) { /* * Get the transfer size in blocks. */ io_size = (cm->cm_length >> sc->DD_block_exponent); /* * Get virtual LBA. Point to correct lower 4 bytes of * LBA in the CDB depending on command. */ lba_idx = ((CDB[0] == READ_12) || (CDB[0] == WRITE_12) || (CDB[0] == READ_10) || (CDB[0] == WRITE_10))? 2 : 6; virtLBA = ((uint64_t)CDB[lba_idx] << 24) | ((uint64_t)CDB[lba_idx + 1] << 16) | ((uint64_t)CDB[lba_idx + 2] << 8) | (uint64_t)CDB[lba_idx + 3]; /* * Check that LBA range for I/O does not exceed volume's * MaxLBA. */ if ((virtLBA + (uint64_t)io_size - 1) <= sc->DD_max_lba) { /* * Check if the I/O crosses a stripe boundary. * If not, translate the virtual LBA to a * physical LBA and set the DevHandle for the * PhysDisk to be used. If it does cross a * boundry, do normal I/O. To get the right * DevHandle to use, get the map number for the * column, then use that map number to look up * the DevHandle of the PhysDisk. */ stripe_offset = (uint32_t)virtLBA & (sc->DD_stripe_size - 1); if ((stripe_offset + io_size) <= sc->DD_stripe_size) { physLBA = (uint32_t)virtLBA >> sc->DD_stripe_exponent; stripe_unit = physLBA / sc->DD_num_phys_disks; column = physLBA % sc->DD_num_phys_disks; pIO_req->DevHandle = htole16(sc->DD_column_map[column]. dev_handle); cm->cm_desc.SCSIIO.DevHandle = pIO_req->DevHandle; physLBA = (stripe_unit << sc->DD_stripe_exponent) + stripe_offset; ptrLBA = &pIO_req->CDB.CDB32[lba_idx]; physLBA_byte = (uint8_t)(physLBA >> 24); *ptrLBA = physLBA_byte; ptrLBA = &pIO_req->CDB.CDB32[lba_idx + 1]; physLBA_byte = (uint8_t)(physLBA >> 16); *ptrLBA = physLBA_byte; ptrLBA = &pIO_req->CDB.CDB32[lba_idx + 2]; physLBA_byte = (uint8_t)(physLBA >> 8); *ptrLBA = physLBA_byte; ptrLBA = &pIO_req->CDB.CDB32[lba_idx + 3]; physLBA_byte = (uint8_t)physLBA; *ptrLBA = physLBA_byte; /* * Set flag that Direct Drive I/O is * being done. */ cm->cm_flags |= MPS_CM_FLAGS_DD_IO; } } } else { /* * 16-byte CDB and the upper 4 bytes of the CDB are not * 0. Get the transfer size in blocks. */ io_size = (cm->cm_length >> sc->DD_block_exponent); /* * Get virtual LBA. */ virtLBA = ((uint64_t)CDB[2] << 54) | ((uint64_t)CDB[3] << 48) | ((uint64_t)CDB[4] << 40) | ((uint64_t)CDB[5] << 32) | ((uint64_t)CDB[6] << 24) | ((uint64_t)CDB[7] << 16) | ((uint64_t)CDB[8] << 8) | (uint64_t)CDB[9]; /* * Check that LBA range for I/O does not exceed volume's * MaxLBA. */ if ((virtLBA + (uint64_t)io_size - 1) <= sc->DD_max_lba) { /* * Check if the I/O crosses a stripe boundary. * If not, translate the virtual LBA to a * physical LBA and set the DevHandle for the * PhysDisk to be used. If it does cross a * boundry, do normal I/O. To get the right * DevHandle to use, get the map number for the * column, then use that map number to look up * the DevHandle of the PhysDisk. */ stripe_offset = (uint32_t)virtLBA & (sc->DD_stripe_size - 1); if ((stripe_offset + io_size) <= sc->DD_stripe_size) { physLBA = (uint32_t)(virtLBA >> sc->DD_stripe_exponent); stripe_unit = physLBA / sc->DD_num_phys_disks; column = physLBA % sc->DD_num_phys_disks; pIO_req->DevHandle = htole16(sc->DD_column_map[column]. dev_handle); cm->cm_desc.SCSIIO.DevHandle = pIO_req->DevHandle; physLBA = (stripe_unit << sc->DD_stripe_exponent) + stripe_offset; /* * Set upper 4 bytes of LBA to 0. We * assume that the phys disks are less * than 2 TB's in size. Then, set the * lower 4 bytes. */ pIO_req->CDB.CDB32[2] = 0; pIO_req->CDB.CDB32[3] = 0; pIO_req->CDB.CDB32[4] = 0; pIO_req->CDB.CDB32[5] = 0; ptrLBA = &pIO_req->CDB.CDB32[6]; physLBA_byte = (uint8_t)(physLBA >> 24); *ptrLBA = physLBA_byte; ptrLBA = &pIO_req->CDB.CDB32[7]; physLBA_byte = (uint8_t)(physLBA >> 16); *ptrLBA = physLBA_byte; ptrLBA = &pIO_req->CDB.CDB32[8]; physLBA_byte = (uint8_t)(physLBA >> 8); *ptrLBA = physLBA_byte; ptrLBA = &pIO_req->CDB.CDB32[9]; physLBA_byte = (uint8_t)physLBA; *ptrLBA = physLBA_byte; /* * Set flag that Direct Drive I/O is * being done. */ cm->cm_flags |= MPS_CM_FLAGS_DD_IO; } } } } } #if __FreeBSD_version >= 900026 static void mpssas_smpio_complete(struct mps_softc *sc, struct mps_command *cm) { MPI2_SMP_PASSTHROUGH_REPLY *rpl; MPI2_SMP_PASSTHROUGH_REQUEST *req; uint64_t sasaddr; union ccb *ccb; ccb = cm->cm_complete_data; /* * Currently there should be no way we can hit this case. It only * happens when we have a failure to allocate chain frames, and SMP * commands require two S/G elements only. That should be handled * in the standard request size. */ if ((cm->cm_flags & MPS_CM_FLAGS_ERROR_MASK) != 0) { mps_dprint(sc, MPS_ERROR,"%s: cm_flags = %#x on SMP request!\n", __func__, cm->cm_flags); mpssas_set_ccbstatus(ccb, CAM_REQ_CMP_ERR); goto bailout; } rpl = (MPI2_SMP_PASSTHROUGH_REPLY *)cm->cm_reply; if (rpl == NULL) { mps_dprint(sc, MPS_ERROR, "%s: NULL cm_reply!\n", __func__); mpssas_set_ccbstatus(ccb, CAM_REQ_CMP_ERR); goto bailout; } req = (MPI2_SMP_PASSTHROUGH_REQUEST *)cm->cm_req; sasaddr = le32toh(req->SASAddress.Low); sasaddr |= ((uint64_t)(le32toh(req->SASAddress.High))) << 32; if ((le16toh(rpl->IOCStatus) & MPI2_IOCSTATUS_MASK) != MPI2_IOCSTATUS_SUCCESS || rpl->SASStatus != MPI2_SASSTATUS_SUCCESS) { mps_dprint(sc, MPS_XINFO, "%s: IOCStatus %04x SASStatus %02x\n", __func__, le16toh(rpl->IOCStatus), rpl->SASStatus); mpssas_set_ccbstatus(ccb, CAM_REQ_CMP_ERR); goto bailout; } mps_dprint(sc, MPS_XINFO, "%s: SMP request to SAS address " "%#jx completed successfully\n", __func__, (uintmax_t)sasaddr); if (ccb->smpio.smp_response[2] == SMP_FR_ACCEPTED) mpssas_set_ccbstatus(ccb, CAM_REQ_CMP); else mpssas_set_ccbstatus(ccb, CAM_SMP_STATUS_ERROR); bailout: /* * We sync in both directions because we had DMAs in the S/G list * in both directions. */ bus_dmamap_sync(sc->buffer_dmat, cm->cm_dmamap, BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(sc->buffer_dmat, cm->cm_dmamap); mps_free_command(sc, cm); xpt_done(ccb); } static void mpssas_send_smpcmd(struct mpssas_softc *sassc, union ccb *ccb, uint64_t sasaddr) { struct mps_command *cm; uint8_t *request, *response; MPI2_SMP_PASSTHROUGH_REQUEST *req; struct mps_softc *sc; struct sglist *sg; int error; sc = sassc->sc; sg = NULL; error = 0; /* * XXX We don't yet support physical addresses here. */ switch ((ccb->ccb_h.flags & CAM_DATA_MASK)) { case CAM_DATA_PADDR: case CAM_DATA_SG_PADDR: mps_dprint(sc, MPS_ERROR, "%s: physical addresses not supported\n", __func__); mpssas_set_ccbstatus(ccb, CAM_REQ_INVALID); xpt_done(ccb); return; case CAM_DATA_SG: /* * The chip does not support more than one buffer for the * request or response. */ if ((ccb->smpio.smp_request_sglist_cnt > 1) || (ccb->smpio.smp_response_sglist_cnt > 1)) { mps_dprint(sc, MPS_ERROR, "%s: multiple request or response " "buffer segments not supported for SMP\n", __func__); mpssas_set_ccbstatus(ccb, CAM_REQ_INVALID); xpt_done(ccb); return; } /* * The CAM_SCATTER_VALID flag was originally implemented * for the XPT_SCSI_IO CCB, which only has one data pointer. * We have two. So, just take that flag to mean that we * might have S/G lists, and look at the S/G segment count * to figure out whether that is the case for each individual * buffer. */ if (ccb->smpio.smp_request_sglist_cnt != 0) { bus_dma_segment_t *req_sg; req_sg = (bus_dma_segment_t *)ccb->smpio.smp_request; request = (uint8_t *)(uintptr_t)req_sg[0].ds_addr; } else request = ccb->smpio.smp_request; if (ccb->smpio.smp_response_sglist_cnt != 0) { bus_dma_segment_t *rsp_sg; rsp_sg = (bus_dma_segment_t *)ccb->smpio.smp_response; response = (uint8_t *)(uintptr_t)rsp_sg[0].ds_addr; } else response = ccb->smpio.smp_response; break; case CAM_DATA_VADDR: request = ccb->smpio.smp_request; response = ccb->smpio.smp_response; break; default: mpssas_set_ccbstatus(ccb, CAM_REQ_INVALID); xpt_done(ccb); return; } cm = mps_alloc_command(sc); if (cm == NULL) { mps_dprint(sc, MPS_ERROR, "%s: cannot allocate command\n", __func__); mpssas_set_ccbstatus(ccb, CAM_RESRC_UNAVAIL); xpt_done(ccb); return; } req = (MPI2_SMP_PASSTHROUGH_REQUEST *)cm->cm_req; bzero(req, sizeof(*req)); req->Function = MPI2_FUNCTION_SMP_PASSTHROUGH; /* Allow the chip to use any route to this SAS address. */ req->PhysicalPort = 0xff; req->RequestDataLength = htole16(ccb->smpio.smp_request_len); req->SGLFlags = MPI2_SGLFLAGS_SYSTEM_ADDRESS_SPACE | MPI2_SGLFLAGS_SGL_TYPE_MPI; mps_dprint(sc, MPS_XINFO, "%s: sending SMP request to SAS " "address %#jx\n", __func__, (uintmax_t)sasaddr); mpi_init_sge(cm, req, &req->SGL); /* * Set up a uio to pass into mps_map_command(). This allows us to * do one map command, and one busdma call in there. */ cm->cm_uio.uio_iov = cm->cm_iovec; cm->cm_uio.uio_iovcnt = 2; cm->cm_uio.uio_segflg = UIO_SYSSPACE; /* * The read/write flag isn't used by busdma, but set it just in * case. This isn't exactly accurate, either, since we're going in * both directions. */ cm->cm_uio.uio_rw = UIO_WRITE; cm->cm_iovec[0].iov_base = request; cm->cm_iovec[0].iov_len = le16toh(req->RequestDataLength); cm->cm_iovec[1].iov_base = response; cm->cm_iovec[1].iov_len = ccb->smpio.smp_response_len; cm->cm_uio.uio_resid = cm->cm_iovec[0].iov_len + cm->cm_iovec[1].iov_len; /* * Trigger a warning message in mps_data_cb() for the user if we * wind up exceeding two S/G segments. The chip expects one * segment for the request and another for the response. */ cm->cm_max_segs = 2; cm->cm_desc.Default.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_DEFAULT_TYPE; cm->cm_complete = mpssas_smpio_complete; cm->cm_complete_data = ccb; /* * Tell the mapping code that we're using a uio, and that this is * an SMP passthrough request. There is a little special-case * logic there (in mps_data_cb()) to handle the bidirectional * transfer. */ cm->cm_flags |= MPS_CM_FLAGS_USE_UIO | MPS_CM_FLAGS_SMP_PASS | MPS_CM_FLAGS_DATAIN | MPS_CM_FLAGS_DATAOUT; /* The chip data format is little endian. */ req->SASAddress.High = htole32(sasaddr >> 32); req->SASAddress.Low = htole32(sasaddr); /* * XXX Note that we don't have a timeout/abort mechanism here. * From the manual, it looks like task management requests only * work for SCSI IO and SATA passthrough requests. We may need to * have a mechanism to retry requests in the event of a chip reset * at least. Hopefully the chip will insure that any errors short * of that are relayed back to the driver. */ error = mps_map_command(sc, cm); if ((error != 0) && (error != EINPROGRESS)) { mps_dprint(sc, MPS_ERROR, "%s: error %d returned from mps_map_command()\n", __func__, error); goto bailout_error; } return; bailout_error: mps_free_command(sc, cm); mpssas_set_ccbstatus(ccb, CAM_RESRC_UNAVAIL); xpt_done(ccb); return; } static void mpssas_action_smpio(struct mpssas_softc *sassc, union ccb *ccb) { struct mps_softc *sc; struct mpssas_target *targ; uint64_t sasaddr = 0; sc = sassc->sc; /* * Make sure the target exists. */ KASSERT(ccb->ccb_h.target_id < sassc->maxtargets, ("Target %d out of bounds in XPT_SMP_IO\n", ccb->ccb_h.target_id)); targ = &sassc->targets[ccb->ccb_h.target_id]; if (targ->handle == 0x0) { mps_dprint(sc, MPS_ERROR, "%s: target %d does not exist!\n", __func__, ccb->ccb_h.target_id); mpssas_set_ccbstatus(ccb, CAM_SEL_TIMEOUT); xpt_done(ccb); return; } /* * If this device has an embedded SMP target, we'll talk to it * directly. * figure out what the expander's address is. */ if ((targ->devinfo & MPI2_SAS_DEVICE_INFO_SMP_TARGET) != 0) sasaddr = targ->sasaddr; /* * If we don't have a SAS address for the expander yet, try * grabbing it from the page 0x83 information cached in the * transport layer for this target. LSI expanders report the * expander SAS address as the port-associated SAS address in * Inquiry VPD page 0x83. Maxim expanders don't report it in page * 0x83. * * XXX KDM disable this for now, but leave it commented out so that * it is obvious that this is another possible way to get the SAS * address. * * The parent handle method below is a little more reliable, and * the other benefit is that it works for devices other than SES * devices. So you can send a SMP request to a da(4) device and it * will get routed to the expander that device is attached to. * (Assuming the da(4) device doesn't contain an SMP target...) */ #if 0 if (sasaddr == 0) sasaddr = xpt_path_sas_addr(ccb->ccb_h.path); #endif /* * If we still don't have a SAS address for the expander, look for * the parent device of this device, which is probably the expander. */ if (sasaddr == 0) { #ifdef OLD_MPS_PROBE struct mpssas_target *parent_target; #endif if (targ->parent_handle == 0x0) { mps_dprint(sc, MPS_ERROR, "%s: handle %d does not have a valid " "parent handle!\n", __func__, targ->handle); mpssas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE); goto bailout; } #ifdef OLD_MPS_PROBE parent_target = mpssas_find_target_by_handle(sassc, 0, targ->parent_handle); if (parent_target == NULL) { mps_dprint(sc, MPS_ERROR, "%s: handle %d does not have a valid " "parent target!\n", __func__, targ->handle); mpssas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE); goto bailout; } if ((parent_target->devinfo & MPI2_SAS_DEVICE_INFO_SMP_TARGET) == 0) { mps_dprint(sc, MPS_ERROR, "%s: handle %d parent %d does not " "have an SMP target!\n", __func__, targ->handle, parent_target->handle); mpssas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE); goto bailout; } sasaddr = parent_target->sasaddr; #else /* OLD_MPS_PROBE */ if ((targ->parent_devinfo & MPI2_SAS_DEVICE_INFO_SMP_TARGET) == 0) { mps_dprint(sc, MPS_ERROR, "%s: handle %d parent %d does not " "have an SMP target!\n", __func__, targ->handle, targ->parent_handle); mpssas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE); goto bailout; } if (targ->parent_sasaddr == 0x0) { mps_dprint(sc, MPS_ERROR, "%s: handle %d parent handle %d does " "not have a valid SAS address!\n", __func__, targ->handle, targ->parent_handle); mpssas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE); goto bailout; } sasaddr = targ->parent_sasaddr; #endif /* OLD_MPS_PROBE */ } if (sasaddr == 0) { mps_dprint(sc, MPS_INFO, "%s: unable to find SAS address for handle %d\n", __func__, targ->handle); mpssas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE); goto bailout; } mpssas_send_smpcmd(sassc, ccb, sasaddr); return; bailout: xpt_done(ccb); } #endif //__FreeBSD_version >= 900026 static void mpssas_action_resetdev(struct mpssas_softc *sassc, union ccb *ccb) { MPI2_SCSI_TASK_MANAGE_REQUEST *req; struct mps_softc *sc; struct mps_command *tm; struct mpssas_target *targ; MPS_FUNCTRACE(sassc->sc); mtx_assert(&sassc->sc->mps_mtx, MA_OWNED); KASSERT(ccb->ccb_h.target_id < sassc->maxtargets, ("Target %d out of bounds in XPT_RESET_DEV\n", ccb->ccb_h.target_id)); sc = sassc->sc; tm = mps_alloc_command(sc); if (tm == NULL) { mps_dprint(sc, MPS_ERROR, "command alloc failure in mpssas_action_resetdev\n"); mpssas_set_ccbstatus(ccb, CAM_RESRC_UNAVAIL); xpt_done(ccb); return; } targ = &sassc->targets[ccb->ccb_h.target_id]; req = (MPI2_SCSI_TASK_MANAGE_REQUEST *)tm->cm_req; req->DevHandle = htole16(targ->handle); req->Function = MPI2_FUNCTION_SCSI_TASK_MGMT; req->TaskType = MPI2_SCSITASKMGMT_TASKTYPE_TARGET_RESET; /* SAS Hard Link Reset / SATA Link Reset */ req->MsgFlags = MPI2_SCSITASKMGMT_MSGFLAGS_LINK_RESET; tm->cm_data = NULL; tm->cm_desc.HighPriority.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY; tm->cm_complete = mpssas_resetdev_complete; tm->cm_complete_data = ccb; tm->cm_targ = targ; targ->flags |= MPSSAS_TARGET_INRESET; mps_map_command(sc, tm); } static void mpssas_resetdev_complete(struct mps_softc *sc, struct mps_command *tm) { MPI2_SCSI_TASK_MANAGE_REPLY *resp; union ccb *ccb; MPS_FUNCTRACE(sc); mtx_assert(&sc->mps_mtx, MA_OWNED); resp = (MPI2_SCSI_TASK_MANAGE_REPLY *)tm->cm_reply; ccb = tm->cm_complete_data; /* * Currently there should be no way we can hit this case. It only * happens when we have a failure to allocate chain frames, and * task management commands don't have S/G lists. */ if ((tm->cm_flags & MPS_CM_FLAGS_ERROR_MASK) != 0) { MPI2_SCSI_TASK_MANAGE_REQUEST *req; req = (MPI2_SCSI_TASK_MANAGE_REQUEST *)tm->cm_req; mps_dprint(sc, MPS_ERROR, "%s: cm_flags = %#x for reset of handle %#04x! " "This should not happen!\n", __func__, tm->cm_flags, req->DevHandle); mpssas_set_ccbstatus(ccb, CAM_REQ_CMP_ERR); goto bailout; } mps_dprint(sc, MPS_XINFO, "%s: IOCStatus = 0x%x ResponseCode = 0x%x\n", __func__, le16toh(resp->IOCStatus), le32toh(resp->ResponseCode)); if (le32toh(resp->ResponseCode) == MPI2_SCSITASKMGMT_RSP_TM_COMPLETE) { mpssas_set_ccbstatus(ccb, CAM_REQ_CMP); mpssas_announce_reset(sc, AC_SENT_BDR, tm->cm_targ->tid, CAM_LUN_WILDCARD); } else mpssas_set_ccbstatus(ccb, CAM_REQ_CMP_ERR); bailout: mpssas_free_tm(sc, tm); xpt_done(ccb); } static void mpssas_poll(struct cam_sim *sim) { struct mpssas_softc *sassc; sassc = cam_sim_softc(sim); if (sassc->sc->mps_debug & MPS_TRACE) { /* frequent debug messages during a panic just slow * everything down too much. */ mps_printf(sassc->sc, "%s clearing MPS_TRACE\n", __func__); sassc->sc->mps_debug &= ~MPS_TRACE; } mps_intr_locked(sassc->sc); } static void mpssas_async(void *callback_arg, uint32_t code, struct cam_path *path, void *arg) { struct mps_softc *sc; sc = (struct mps_softc *)callback_arg; switch (code) { #if (__FreeBSD_version >= 1000006) || \ ((__FreeBSD_version >= 901503) && (__FreeBSD_version < 1000000)) case AC_ADVINFO_CHANGED: { struct mpssas_target *target; struct mpssas_softc *sassc; struct scsi_read_capacity_data_long rcap_buf; struct ccb_dev_advinfo cdai; struct mpssas_lun *lun; lun_id_t lunid; int found_lun; uintptr_t buftype; buftype = (uintptr_t)arg; found_lun = 0; sassc = sc->sassc; /* * We're only interested in read capacity data changes. */ if (buftype != CDAI_TYPE_RCAPLONG) break; /* * We should have a handle for this, but check to make sure. */ KASSERT(xpt_path_target_id(path) < sassc->maxtargets, ("Target %d out of bounds in mpssas_async\n", xpt_path_target_id(path))); target = &sassc->targets[xpt_path_target_id(path)]; if (target->handle == 0) break; lunid = xpt_path_lun_id(path); SLIST_FOREACH(lun, &target->luns, lun_link) { if (lun->lun_id == lunid) { found_lun = 1; break; } } if (found_lun == 0) { lun = malloc(sizeof(struct mpssas_lun), M_MPT2, M_NOWAIT | M_ZERO); if (lun == NULL) { mps_dprint(sc, MPS_ERROR, "Unable to alloc " "LUN for EEDP support.\n"); break; } lun->lun_id = lunid; SLIST_INSERT_HEAD(&target->luns, lun, lun_link); } bzero(&rcap_buf, sizeof(rcap_buf)); xpt_setup_ccb(&cdai.ccb_h, path, CAM_PRIORITY_NORMAL); cdai.ccb_h.func_code = XPT_DEV_ADVINFO; cdai.ccb_h.flags = CAM_DIR_IN; cdai.buftype = CDAI_TYPE_RCAPLONG; #if (__FreeBSD_version >= 1100061) || \ ((__FreeBSD_version >= 1001510) && (__FreeBSD_version < 1100000)) cdai.flags = CDAI_FLAG_NONE; #else cdai.flags = 0; #endif cdai.bufsiz = sizeof(rcap_buf); cdai.buf = (uint8_t *)&rcap_buf; xpt_action((union ccb *)&cdai); if ((cdai.ccb_h.status & CAM_DEV_QFRZN) != 0) cam_release_devq(cdai.ccb_h.path, 0, 0, 0, FALSE); if ((mpssas_get_ccbstatus((union ccb *)&cdai) == CAM_REQ_CMP) && (rcap_buf.prot & SRC16_PROT_EN)) { lun->eedp_formatted = TRUE; lun->eedp_block_size = scsi_4btoul(rcap_buf.length); } else { lun->eedp_formatted = FALSE; lun->eedp_block_size = 0; } break; } #else case AC_FOUND_DEVICE: { struct ccb_getdev *cgd; cgd = arg; mpssas_check_eedp(sc, path, cgd); break; } #endif default: break; } } #if (__FreeBSD_version < 901503) || \ ((__FreeBSD_version >= 1000000) && (__FreeBSD_version < 1000006)) static void mpssas_check_eedp(struct mps_softc *sc, struct cam_path *path, struct ccb_getdev *cgd) { struct mpssas_softc *sassc = sc->sassc; struct ccb_scsiio *csio; struct scsi_read_capacity_16 *scsi_cmd; struct scsi_read_capacity_eedp *rcap_buf; path_id_t pathid; target_id_t targetid; lun_id_t lunid; union ccb *ccb; struct cam_path *local_path; struct mpssas_target *target; struct mpssas_lun *lun; uint8_t found_lun; char path_str[64]; sassc = sc->sassc; pathid = cam_sim_path(sassc->sim); targetid = xpt_path_target_id(path); lunid = xpt_path_lun_id(path); KASSERT(targetid < sassc->maxtargets, ("Target %d out of bounds in mpssas_check_eedp\n", targetid)); target = &sassc->targets[targetid]; if (target->handle == 0x0) return; /* * Determine if the device is EEDP capable. * * If this flag is set in the inquiry data, * the device supports protection information, * and must support the 16 byte read * capacity command, otherwise continue without * sending read cap 16 */ if ((cgd->inq_data.spc3_flags & SPC3_SID_PROTECT) == 0) return; /* * Issue a READ CAPACITY 16 command. This info * is used to determine if the LUN is formatted * for EEDP support. */ ccb = xpt_alloc_ccb_nowait(); if (ccb == NULL) { mps_dprint(sc, MPS_ERROR, "Unable to alloc CCB " "for EEDP support.\n"); return; } if (xpt_create_path(&local_path, xpt_periph, pathid, targetid, lunid) != CAM_REQ_CMP) { mps_dprint(sc, MPS_ERROR, "Unable to create " "path for EEDP support\n"); xpt_free_ccb(ccb); return; } /* * If LUN is already in list, don't create a new * one. */ found_lun = FALSE; SLIST_FOREACH(lun, &target->luns, lun_link) { if (lun->lun_id == lunid) { found_lun = TRUE; break; } } if (!found_lun) { lun = malloc(sizeof(struct mpssas_lun), M_MPT2, M_NOWAIT | M_ZERO); if (lun == NULL) { mps_dprint(sc, MPS_ERROR, "Unable to alloc LUN for EEDP support.\n"); xpt_free_path(local_path); xpt_free_ccb(ccb); return; } lun->lun_id = lunid; SLIST_INSERT_HEAD(&target->luns, lun, lun_link); } xpt_path_string(local_path, path_str, sizeof(path_str)); - /* - * If this is a SATA direct-access end device, - * mark it so that a SCSI StartStopUnit command - * will be sent to it when the driver is being - * shutdown. - */ - if ((cgd.inq_data.device == T_DIRECT) && - (target->devinfo & MPI2_SAS_DEVICE_INFO_SATA_DEVICE) && - ((target->devinfo & MPI2_SAS_DEVICE_INFO_MASK_DEVICE_TYPE) == - MPI2_SAS_DEVICE_INFO_END_DEVICE)) { - lun->stop_at_shutdown = TRUE; - } - mps_dprint(sc, MPS_INFO, "Sending read cap: path %s handle %d\n", path_str, target->handle); /* * Issue a READ CAPACITY 16 command for the LUN. * The mpssas_read_cap_done function will load * the read cap info into the LUN struct. */ rcap_buf = malloc(sizeof(struct scsi_read_capacity_eedp), M_MPT2, M_NOWAIT | M_ZERO); if (rcap_buf == NULL) { mps_dprint(sc, MPS_FAULT, "Unable to alloc read capacity buffer for EEDP support.\n"); xpt_free_path(ccb->ccb_h.path); xpt_free_ccb(ccb); return; } xpt_setup_ccb(&ccb->ccb_h, local_path, CAM_PRIORITY_XPT); csio = &ccb->csio; csio->ccb_h.func_code = XPT_SCSI_IO; csio->ccb_h.flags = CAM_DIR_IN; csio->ccb_h.retry_count = 4; csio->ccb_h.cbfcnp = mpssas_read_cap_done; csio->ccb_h.timeout = 60000; csio->data_ptr = (uint8_t *)rcap_buf; csio->dxfer_len = sizeof(struct scsi_read_capacity_eedp); csio->sense_len = MPS_SENSE_LEN; csio->cdb_len = sizeof(*scsi_cmd); csio->tag_action = MSG_SIMPLE_Q_TAG; scsi_cmd = (struct scsi_read_capacity_16 *)&csio->cdb_io.cdb_bytes; bzero(scsi_cmd, sizeof(*scsi_cmd)); scsi_cmd->opcode = 0x9E; scsi_cmd->service_action = SRC16_SERVICE_ACTION; ((uint8_t *)scsi_cmd)[13] = sizeof(struct scsi_read_capacity_eedp); ccb->ccb_h.ppriv_ptr1 = sassc; xpt_action(ccb); } static void mpssas_read_cap_done(struct cam_periph *periph, union ccb *done_ccb) { struct mpssas_softc *sassc; struct mpssas_target *target; struct mpssas_lun *lun; struct scsi_read_capacity_eedp *rcap_buf; if (done_ccb == NULL) return; /* Driver need to release devq, it Scsi command is * generated by driver internally. * Currently there is a single place where driver * calls scsi command internally. In future if driver * calls more scsi command internally, it needs to release * devq internally, since those command will not go back to * cam_periph. */ if ((done_ccb->ccb_h.status & CAM_DEV_QFRZN) ) { done_ccb->ccb_h.status &= ~CAM_DEV_QFRZN; xpt_release_devq(done_ccb->ccb_h.path, /*count*/ 1, /*run_queue*/TRUE); } rcap_buf = (struct scsi_read_capacity_eedp *)done_ccb->csio.data_ptr; /* * Get the LUN ID for the path and look it up in the LUN list for the * target. */ sassc = (struct mpssas_softc *)done_ccb->ccb_h.ppriv_ptr1; KASSERT(done_ccb->ccb_h.target_id < sassc->maxtargets, ("Target %d out of bounds in mpssas_read_cap_done\n", done_ccb->ccb_h.target_id)); target = &sassc->targets[done_ccb->ccb_h.target_id]; SLIST_FOREACH(lun, &target->luns, lun_link) { if (lun->lun_id != done_ccb->ccb_h.target_lun) continue; /* * Got the LUN in the target's LUN list. Fill it in * with EEDP info. If the READ CAP 16 command had some * SCSI error (common if command is not supported), mark * the lun as not supporting EEDP and set the block size * to 0. */ if ((mpssas_get_ccbstatus(done_ccb) != CAM_REQ_CMP) || (done_ccb->csio.scsi_status != SCSI_STATUS_OK)) { lun->eedp_formatted = FALSE; lun->eedp_block_size = 0; break; } if (rcap_buf->protect & 0x01) { mps_dprint(sassc->sc, MPS_INFO, "LUN %d for " "target ID %d is formatted for EEDP " "support.\n", done_ccb->ccb_h.target_lun, done_ccb->ccb_h.target_id); lun->eedp_formatted = TRUE; lun->eedp_block_size = scsi_4btoul(rcap_buf->length); } break; } // Finished with this CCB and path. free(rcap_buf, M_MPT2); xpt_free_path(done_ccb->ccb_h.path); xpt_free_ccb(done_ccb); } #endif /* (__FreeBSD_version < 901503) || \ ((__FreeBSD_version >= 1000000) && (__FreeBSD_version < 1000006)) */ void mpssas_prepare_for_tm(struct mps_softc *sc, struct mps_command *tm, struct mpssas_target *target, lun_id_t lun_id) { union ccb *ccb; path_id_t path_id; /* * Set the INRESET flag for this target so that no I/O will be sent to * the target until the reset has completed. If an I/O request does * happen, the devq will be frozen. The CCB holds the path which is * used to release the devq. The devq is released and the CCB is freed * when the TM completes. */ ccb = xpt_alloc_ccb_nowait(); if (ccb) { path_id = cam_sim_path(sc->sassc->sim); if (xpt_create_path(&ccb->ccb_h.path, xpt_periph, path_id, target->tid, lun_id) != CAM_REQ_CMP) { xpt_free_ccb(ccb); } else { tm->cm_ccb = ccb; tm->cm_targ = target; target->flags |= MPSSAS_TARGET_INRESET; } } } int mpssas_startup(struct mps_softc *sc) { /* * Send the port enable message and set the wait_for_port_enable flag. * This flag helps to keep the simq frozen until all discovery events * are processed. */ sc->wait_for_port_enable = 1; mpssas_send_portenable(sc); return (0); } static int mpssas_send_portenable(struct mps_softc *sc) { MPI2_PORT_ENABLE_REQUEST *request; struct mps_command *cm; MPS_FUNCTRACE(sc); if ((cm = mps_alloc_command(sc)) == NULL) return (EBUSY); request = (MPI2_PORT_ENABLE_REQUEST *)cm->cm_req; request->Function = MPI2_FUNCTION_PORT_ENABLE; request->MsgFlags = 0; request->VP_ID = 0; cm->cm_desc.Default.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_DEFAULT_TYPE; cm->cm_complete = mpssas_portenable_complete; cm->cm_data = NULL; cm->cm_sge = NULL; mps_map_command(sc, cm); mps_dprint(sc, MPS_XINFO, "mps_send_portenable finished cm %p req %p complete %p\n", cm, cm->cm_req, cm->cm_complete); return (0); } static void mpssas_portenable_complete(struct mps_softc *sc, struct mps_command *cm) { MPI2_PORT_ENABLE_REPLY *reply; struct mpssas_softc *sassc; MPS_FUNCTRACE(sc); sassc = sc->sassc; /* * Currently there should be no way we can hit this case. It only * happens when we have a failure to allocate chain frames, and * port enable commands don't have S/G lists. */ if ((cm->cm_flags & MPS_CM_FLAGS_ERROR_MASK) != 0) { mps_dprint(sc, MPS_ERROR, "%s: cm_flags = %#x for port enable! " "This should not happen!\n", __func__, cm->cm_flags); } reply = (MPI2_PORT_ENABLE_REPLY *)cm->cm_reply; if (reply == NULL) mps_dprint(sc, MPS_FAULT, "Portenable NULL reply\n"); else if (le16toh(reply->IOCStatus & MPI2_IOCSTATUS_MASK) != MPI2_IOCSTATUS_SUCCESS) mps_dprint(sc, MPS_FAULT, "Portenable failed\n"); mps_free_command(sc, cm); if (sc->mps_ich.ich_arg != NULL) { mps_dprint(sc, MPS_XINFO, "disestablish config intrhook\n"); config_intrhook_disestablish(&sc->mps_ich); sc->mps_ich.ich_arg = NULL; } /* * Get WarpDrive info after discovery is complete but before the scan * starts. At this point, all devices are ready to be exposed to the * OS. If devices should be hidden instead, take them out of the * 'targets' array before the scan. The devinfo for a disk will have * some info and a volume's will be 0. Use that to remove disks. */ mps_wd_config_pages(sc); /* * Done waiting for port enable to complete. Decrement the refcount. * If refcount is 0, discovery is complete and a rescan of the bus can * take place. Since the simq was explicitly frozen before port * enable, it must be explicitly released here to keep the * freeze/release count in sync. */ sc->wait_for_port_enable = 0; sc->port_enable_complete = 1; wakeup(&sc->port_enable_complete); mpssas_startup_decrement(sassc); } int mpssas_check_id(struct mpssas_softc *sassc, int id) { struct mps_softc *sc = sassc->sc; char *ids; char *name; ids = &sc->exclude_ids[0]; while((name = strsep(&ids, ",")) != NULL) { if (name[0] == '\0') continue; if (strtol(name, NULL, 0) == (long)id) return (1); } return (0); } void mpssas_realloc_targets(struct mps_softc *sc, int maxtargets) { struct mpssas_softc *sassc; struct mpssas_lun *lun, *lun_tmp; struct mpssas_target *targ; int i; sassc = sc->sassc; /* * The number of targets is based on IOC Facts, so free all of * the allocated LUNs for each target and then the target buffer * itself. */ for (i=0; i< maxtargets; i++) { targ = &sassc->targets[i]; SLIST_FOREACH_SAFE(lun, &targ->luns, lun_link, lun_tmp) { free(lun, M_MPT2); } } free(sassc->targets, M_MPT2); sassc->targets = malloc(sizeof(struct mpssas_target) * maxtargets, M_MPT2, M_WAITOK|M_ZERO); if (!sassc->targets) { panic("%s failed to alloc targets with error %d\n", __func__, ENOMEM); } } Index: projects/cxl_iscsi/sys/dev/uart/uart_dev_pl011.c =================================================================== --- projects/cxl_iscsi/sys/dev/uart/uart_dev_pl011.c (revision 279716) +++ projects/cxl_iscsi/sys/dev/uart/uart_dev_pl011.c (revision 279717) @@ -1,489 +1,497 @@ /*- * Copyright (c) 2012 Semihalf. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include "uart_if.h" #include /* PL011 UART registers and masks*/ #define UART_DR 0x00 /* Data register */ #define DR_FE (1 << 8) /* Framing error */ #define DR_PE (1 << 9) /* Parity error */ #define DR_BE (1 << 10) /* Break error */ #define DR_OE (1 << 11) /* Overrun error */ #define UART_FR 0x06 /* Flag register */ #define FR_TXFF (1 << 5) /* Transmit FIFO/reg full */ #define FR_RXFF (1 << 6) /* Receive FIFO/reg full */ #define FR_TXFE (1 << 7) /* Transmit FIFO/reg empty */ #define UART_IBRD 0x09 /* Integer baud rate register */ #define IBRD_BDIVINT 0xffff /* Significant part of int. divisor value */ #define UART_FBRD 0x0a /* Fractional baud rate register */ #define FBRD_BDIVFRAC 0x3f /* Significant part of frac. divisor value */ #define UART_LCR_H 0x0b /* Line control register */ #define LCR_H_WLEN8 (0x3 << 5) #define LCR_H_WLEN7 (0x2 << 5) #define LCR_H_WLEN6 (0x1 << 5) #define LCR_H_FEN (1 << 4) /* FIFO mode enable */ #define LCR_H_STP2 (1 << 3) /* 2 stop frames at the end */ #define LCR_H_EPS (1 << 2) /* Even parity select */ #define LCR_H_PEN (1 << 1) /* Parity enable */ #define UART_CR 0x0c /* Control register */ #define CR_RXE (1 << 9) /* Receive enable */ #define CR_TXE (1 << 8) /* Transmit enable */ #define CR_UARTEN (1 << 0) /* UART enable */ #define UART_IMSC 0x0e /* Interrupt mask set/clear register */ #define IMSC_MASK_ALL 0x7ff /* Mask all interrupts */ #define UART_RIS 0x0f /* Raw interrupt status register */ #define UART_RXREADY (1 << 4) /* RX buffer full */ #define UART_TXEMPTY (1 << 5) /* TX buffer empty */ #define RIS_RTIM (1 << 6) /* Receive timeout */ #define RIS_FE (1 << 7) /* Framing error interrupt status */ #define RIS_PE (1 << 8) /* Parity error interrupt status */ #define RIS_BE (1 << 9) /* Break error interrupt status */ #define RIS_OE (1 << 10) /* Overrun interrupt status */ #define UART_MIS 0x10 /* Masked interrupt status register */ #define UART_ICR 0x11 /* Interrupt clear register */ /* * FIXME: actual register size is SoC-dependent, we need to handle it */ #define __uart_getreg(bas, reg) \ bus_space_read_4((bas)->bst, (bas)->bsh, uart_regofs(bas, reg)) #define __uart_setreg(bas, reg, value) \ bus_space_write_4((bas)->bst, (bas)->bsh, uart_regofs(bas, reg), value) /* * Low-level UART interface. */ static int uart_pl011_probe(struct uart_bas *bas); static void uart_pl011_init(struct uart_bas *bas, int, int, int, int); static void uart_pl011_term(struct uart_bas *bas); static void uart_pl011_putc(struct uart_bas *bas, int); static int uart_pl011_rxready(struct uart_bas *bas); static int uart_pl011_getc(struct uart_bas *bas, struct mtx *); static struct uart_ops uart_pl011_ops = { .probe = uart_pl011_probe, .init = uart_pl011_init, .term = uart_pl011_term, .putc = uart_pl011_putc, .rxready = uart_pl011_rxready, .getc = uart_pl011_getc, }; static int uart_pl011_probe(struct uart_bas *bas) { return (0); } static void uart_pl011_param(struct uart_bas *bas, int baudrate, int databits, int stopbits, int parity) { uint32_t ctrl, line; uint32_t baud; /* * Zero all settings to make sure * UART is disabled and not configured */ ctrl = line = 0x0; __uart_setreg(bas, UART_CR, ctrl); /* As we know UART is disabled we may setup the line */ switch (databits) { case 7: line |= LCR_H_WLEN7; break; case 6: line |= LCR_H_WLEN6; break; case 8: default: line |= LCR_H_WLEN8; break; } if (stopbits == 2) line |= LCR_H_STP2; else line &= ~LCR_H_STP2; if (parity) line |= LCR_H_PEN; else line &= ~LCR_H_PEN; /* Configure the rest */ line &= ~LCR_H_FEN; ctrl |= (CR_RXE | CR_TXE | CR_UARTEN); if (bas->rclk != 0 && baudrate != 0) { baud = bas->rclk * 4 / baudrate; __uart_setreg(bas, UART_IBRD, ((uint32_t)(baud >> 6)) & IBRD_BDIVINT); __uart_setreg(bas, UART_FBRD, (uint32_t)(baud & 0x3F) & FBRD_BDIVFRAC); } /* Add config. to line before reenabling UART */ __uart_setreg(bas, UART_LCR_H, (__uart_getreg(bas, UART_LCR_H) & ~0xff) | line); __uart_setreg(bas, UART_CR, ctrl); } static void uart_pl011_init(struct uart_bas *bas, int baudrate, int databits, int stopbits, int parity) { /* Mask all interrupts */ __uart_setreg(bas, UART_IMSC, __uart_getreg(bas, UART_IMSC) & ~IMSC_MASK_ALL); uart_pl011_param(bas, baudrate, databits, stopbits, parity); } static void uart_pl011_term(struct uart_bas *bas) { } static void uart_pl011_putc(struct uart_bas *bas, int c) { /* Wait when TX FIFO full. Push character otherwise. */ while (__uart_getreg(bas, UART_FR) & FR_TXFF) ; __uart_setreg(bas, UART_DR, c & 0xff); } static int uart_pl011_rxready(struct uart_bas *bas) { return (__uart_getreg(bas, UART_FR) & FR_RXFF); } static int uart_pl011_getc(struct uart_bas *bas, struct mtx *hwmtx) { int c; while (!uart_pl011_rxready(bas)) ; c = __uart_getreg(bas, UART_DR) & 0xff; return (c); } /* * High-level UART interface. */ struct uart_pl011_softc { struct uart_softc base; uint8_t fcr; uint8_t ier; uint8_t mcr; uint8_t ier_mask; uint8_t ier_rxbits; }; static int uart_pl011_bus_attach(struct uart_softc *); static int uart_pl011_bus_detach(struct uart_softc *); static int uart_pl011_bus_flush(struct uart_softc *, int); static int uart_pl011_bus_getsig(struct uart_softc *); static int uart_pl011_bus_ioctl(struct uart_softc *, int, intptr_t); static int uart_pl011_bus_ipend(struct uart_softc *); static int uart_pl011_bus_param(struct uart_softc *, int, int, int, int); static int uart_pl011_bus_probe(struct uart_softc *); static int uart_pl011_bus_receive(struct uart_softc *); static int uart_pl011_bus_setsig(struct uart_softc *, int); static int uart_pl011_bus_transmit(struct uart_softc *); static void uart_pl011_bus_grab(struct uart_softc *); static void uart_pl011_bus_ungrab(struct uart_softc *); static kobj_method_t uart_pl011_methods[] = { KOBJMETHOD(uart_attach, uart_pl011_bus_attach), KOBJMETHOD(uart_detach, uart_pl011_bus_detach), KOBJMETHOD(uart_flush, uart_pl011_bus_flush), KOBJMETHOD(uart_getsig, uart_pl011_bus_getsig), KOBJMETHOD(uart_ioctl, uart_pl011_bus_ioctl), KOBJMETHOD(uart_ipend, uart_pl011_bus_ipend), KOBJMETHOD(uart_param, uart_pl011_bus_param), KOBJMETHOD(uart_probe, uart_pl011_bus_probe), KOBJMETHOD(uart_receive, uart_pl011_bus_receive), KOBJMETHOD(uart_setsig, uart_pl011_bus_setsig), KOBJMETHOD(uart_transmit, uart_pl011_bus_transmit), KOBJMETHOD(uart_grab, uart_pl011_bus_grab), KOBJMETHOD(uart_ungrab, uart_pl011_bus_ungrab), { 0, 0 } }; struct uart_class uart_pl011_class = { "uart_pl011", uart_pl011_methods, sizeof(struct uart_pl011_softc), .uc_ops = &uart_pl011_ops, .uc_range = 0x48, .uc_rclk = 0 }; static int uart_pl011_bus_attach(struct uart_softc *sc) { struct uart_bas *bas; int reg; bas = &sc->sc_bas; /* Enable interrupts */ reg = (UART_RXREADY | RIS_RTIM | UART_TXEMPTY); __uart_setreg(bas, UART_IMSC, reg); /* Clear interrupts */ __uart_setreg(bas, UART_ICR, IMSC_MASK_ALL); return (0); } static int uart_pl011_bus_detach(struct uart_softc *sc) { return (0); } static int uart_pl011_bus_flush(struct uart_softc *sc, int what) { return (0); } static int uart_pl011_bus_getsig(struct uart_softc *sc) { return (0); } static int uart_pl011_bus_ioctl(struct uart_softc *sc, int request, intptr_t data) { struct uart_bas *bas; int error; bas = &sc->sc_bas; error = 0; uart_lock(sc->sc_hwmtx); switch (request) { case UART_IOCTL_BREAK: break; case UART_IOCTL_BAUD: *(int*)data = 115200; break; default: error = EINVAL; break; } uart_unlock(sc->sc_hwmtx); return (error); } static int uart_pl011_bus_ipend(struct uart_softc *sc) { struct uart_bas *bas; uint32_t ints; int ipend; int reg; bas = &sc->sc_bas; uart_lock(sc->sc_hwmtx); ints = __uart_getreg(bas, UART_MIS); ipend = 0; if (ints & (UART_RXREADY | RIS_RTIM)) ipend |= SER_INT_RXREADY; if (ints & RIS_BE) ipend |= SER_INT_BREAK; if (ints & RIS_OE) ipend |= SER_INT_OVERRUN; if (ints & UART_TXEMPTY) { if (sc->sc_txbusy) ipend |= SER_INT_TXIDLE; /* Disable TX interrupt */ reg = __uart_getreg(bas, UART_IMSC); reg &= ~(UART_TXEMPTY); __uart_setreg(bas, UART_IMSC, reg); } uart_unlock(sc->sc_hwmtx); return (ipend); } static int uart_pl011_bus_param(struct uart_softc *sc, int baudrate, int databits, int stopbits, int parity) { uart_lock(sc->sc_hwmtx); uart_pl011_param(&sc->sc_bas, baudrate, databits, stopbits, parity); uart_unlock(sc->sc_hwmtx); return (0); } static int uart_pl011_bus_probe(struct uart_softc *sc) { device_set_desc(sc->sc_dev, "PrimeCell UART (PL011)"); sc->sc_rxfifosz = 1; sc->sc_txfifosz = 1; return (0); } static int uart_pl011_bus_receive(struct uart_softc *sc) { struct uart_bas *bas; uint32_t ints, xc; int rx; bas = &sc->sc_bas; uart_lock(sc->sc_hwmtx); ints = __uart_getreg(bas, UART_MIS); while (ints & (UART_RXREADY | RIS_RTIM)) { if (uart_rx_full(sc)) { sc->sc_rxbuf[sc->sc_rxput] = UART_STAT_OVERRUN; break; } xc = __uart_getreg(bas, UART_DR); rx = xc & 0xff; if (xc & DR_FE) rx |= UART_STAT_FRAMERR; if (xc & DR_PE) rx |= UART_STAT_PARERR; __uart_setreg(bas, UART_ICR, (UART_RXREADY | RIS_RTIM)); uart_rx_put(sc, rx); ints = __uart_getreg(bas, UART_MIS); } uart_unlock(sc->sc_hwmtx); return (0); } static int uart_pl011_bus_setsig(struct uart_softc *sc, int sig) { return (0); } static int uart_pl011_bus_transmit(struct uart_softc *sc) { struct uart_bas *bas; int reg; int i; bas = &sc->sc_bas; uart_lock(sc->sc_hwmtx); for (i = 0; i < sc->sc_txdatasz; i++) { __uart_setreg(bas, UART_DR, sc->sc_txbuf[i]); uart_barrier(bas); } - sc->sc_txbusy = 1; - /* Enable TX interrupt */ - reg = __uart_getreg(bas, UART_IMSC); - reg |= (UART_TXEMPTY); - __uart_setreg(bas, UART_IMSC, reg); + /* If not empty wait until it is */ + if ((__uart_getreg(bas, UART_FR) & FR_TXFE) != FR_TXFE) { + sc->sc_txbusy = 1; + /* Enable TX interrupt */ + reg = __uart_getreg(bas, UART_IMSC); + reg |= (UART_TXEMPTY); + __uart_setreg(bas, UART_IMSC, reg); + } + uart_unlock(sc->sc_hwmtx); + + /* No interrupt expected, schedule the next fifo write */ + if (!sc->sc_txbusy) + uart_sched_softih(sc, SER_INT_TXIDLE); return (0); } static void uart_pl011_bus_grab(struct uart_softc *sc) { struct uart_bas *bas; bas = &sc->sc_bas; uart_lock(sc->sc_hwmtx); __uart_setreg(bas, UART_IMSC, /* Switch to RX polling while grabbed */ ~UART_RXREADY & __uart_getreg(bas, UART_IMSC)); uart_unlock(sc->sc_hwmtx); } static void uart_pl011_bus_ungrab(struct uart_softc *sc) { struct uart_bas *bas; bas = &sc->sc_bas; uart_lock(sc->sc_hwmtx); __uart_setreg(bas, UART_IMSC, /* Switch to RX interrupts while not grabbed */ UART_RXREADY | __uart_getreg(bas, UART_IMSC)); uart_unlock(sc->sc_hwmtx); } Index: projects/cxl_iscsi/sys/dev/usb/controller/dwc_otg_fdt.c =================================================================== --- projects/cxl_iscsi/sys/dev/usb/controller/dwc_otg_fdt.c (revision 279716) +++ projects/cxl_iscsi/sys/dev/usb/controller/dwc_otg_fdt.c (revision 279717) @@ -1,234 +1,224 @@ /*- * Copyright (c) 2012 Hans Petter Selasky. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); -#include -#include #include -#include -#include #include -#include #include -#include -#include -#include #include -#include -#include -#include -#include +#include +#include #include -#include +#include +#include #include -#include #include #include #include #include #include #include #include #include #include #include #include #include static device_probe_t dwc_otg_probe; static device_attach_t dwc_otg_attach; static device_detach_t dwc_otg_detach; struct dwc_otg_super_softc { struct dwc_otg_softc sc_otg; /* must be first */ }; static int dwc_otg_probe(device_t dev) { if (!ofw_bus_status_okay(dev)) return (ENXIO); if (!ofw_bus_is_compatible(dev, "synopsys,designware-hs-otg2")) return (ENXIO); device_set_desc(dev, "DWC OTG 2.0 integrated USB controller"); return (0); } static int dwc_otg_attach(device_t dev) { struct dwc_otg_super_softc *sc = device_get_softc(dev); char usb_mode[24]; int err; int rid; /* initialise some bus fields */ sc->sc_otg.sc_bus.parent = dev; sc->sc_otg.sc_bus.devices = sc->sc_otg.sc_devices; sc->sc_otg.sc_bus.devices_max = DWC_OTG_MAX_DEVICES; sc->sc_otg.sc_bus.dma_bits = 32; /* get USB mode, if any */ if (OF_getprop(ofw_bus_get_node(dev), "dr_mode", &usb_mode, sizeof(usb_mode)) > 0) { /* ensure proper zero termination */ usb_mode[sizeof(usb_mode) - 1] = 0; if (strcasecmp(usb_mode, "host") == 0) sc->sc_otg.sc_mode = DWC_MODE_HOST; else if (strcasecmp(usb_mode, "peripheral") == 0) sc->sc_otg.sc_mode = DWC_MODE_DEVICE; else if (strcasecmp(usb_mode, "otg") != 0) { device_printf(dev, "Invalid FDT dr_mode: %s\n", usb_mode); } } /* get all DMA memory */ if (usb_bus_mem_alloc_all(&sc->sc_otg.sc_bus, USB_GET_DMA_TAG(dev), NULL)) { return (ENOMEM); } rid = 0; sc->sc_otg.sc_io_res = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &rid, RF_ACTIVE); if (!(sc->sc_otg.sc_io_res)) { err = ENOMEM; goto error; } sc->sc_otg.sc_io_tag = rman_get_bustag(sc->sc_otg.sc_io_res); sc->sc_otg.sc_io_hdl = rman_get_bushandle(sc->sc_otg.sc_io_res); sc->sc_otg.sc_io_size = rman_get_size(sc->sc_otg.sc_io_res); rid = 0; sc->sc_otg.sc_irq_res = bus_alloc_resource_any(dev, SYS_RES_IRQ, &rid, RF_ACTIVE); if (sc->sc_otg.sc_irq_res == NULL) goto error; sc->sc_otg.sc_bus.bdev = device_add_child(dev, "usbus", -1); if (sc->sc_otg.sc_bus.bdev == NULL) goto error; device_set_ivars(sc->sc_otg.sc_bus.bdev, &sc->sc_otg.sc_bus); err = bus_setup_intr(dev, sc->sc_otg.sc_irq_res, INTR_TYPE_TTY | INTR_MPSAFE, &dwc_otg_filter_interrupt, &dwc_otg_interrupt, sc, &sc->sc_otg.sc_intr_hdl); if (err) { sc->sc_otg.sc_intr_hdl = NULL; goto error; } err = dwc_otg_init(&sc->sc_otg); if (err == 0) { err = device_probe_and_attach(sc->sc_otg.sc_bus.bdev); } if (err) goto error; return (0); error: dwc_otg_detach(dev); return (ENXIO); } static int dwc_otg_detach(device_t dev) { struct dwc_otg_super_softc *sc = device_get_softc(dev); device_t bdev; int err; if (sc->sc_otg.sc_bus.bdev) { bdev = sc->sc_otg.sc_bus.bdev; device_detach(bdev); device_delete_child(dev, bdev); } /* during module unload there are lots of children leftover */ device_delete_children(dev); if (sc->sc_otg.sc_irq_res && sc->sc_otg.sc_intr_hdl) { /* * only call dwc_otg_uninit() after dwc_otg_init() */ dwc_otg_uninit(&sc->sc_otg); err = bus_teardown_intr(dev, sc->sc_otg.sc_irq_res, sc->sc_otg.sc_intr_hdl); sc->sc_otg.sc_intr_hdl = NULL; } /* free IRQ channel, if any */ if (sc->sc_otg.sc_irq_res) { bus_release_resource(dev, SYS_RES_IRQ, 0, sc->sc_otg.sc_irq_res); sc->sc_otg.sc_irq_res = NULL; } /* free memory resource, if any */ if (sc->sc_otg.sc_io_res) { bus_release_resource(dev, SYS_RES_MEMORY, 0, sc->sc_otg.sc_io_res); sc->sc_otg.sc_io_res = NULL; } usb_bus_mem_free_all(&sc->sc_otg.sc_bus, NULL); return (0); } static device_method_t dwc_otg_methods[] = { /* Device interface */ DEVMETHOD(device_probe, dwc_otg_probe), DEVMETHOD(device_attach, dwc_otg_attach), DEVMETHOD(device_detach, dwc_otg_detach), DEVMETHOD(device_suspend, bus_generic_suspend), DEVMETHOD(device_resume, bus_generic_resume), DEVMETHOD(device_shutdown, bus_generic_shutdown), DEVMETHOD_END }; static driver_t dwc_otg_driver = { .name = "dwcotg", .methods = dwc_otg_methods, .size = sizeof(struct dwc_otg_super_softc), }; static devclass_t dwc_otg_devclass; DRIVER_MODULE(dwcotg, simplebus, dwc_otg_driver, dwc_otg_devclass, 0, 0); MODULE_DEPEND(dwcotg, usb, 1, 1, 1); Index: projects/cxl_iscsi/sys/dev/usb/controller/xhci_pci.c =================================================================== --- projects/cxl_iscsi/sys/dev/usb/controller/xhci_pci.c (revision 279716) +++ projects/cxl_iscsi/sys/dev/usb/controller/xhci_pci.c (revision 279717) @@ -1,373 +1,384 @@ /*- * Copyright (c) 2010 Hans Petter Selasky. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "usb_if.h" static device_probe_t xhci_pci_probe; static device_attach_t xhci_pci_attach; static device_detach_t xhci_pci_detach; static usb_take_controller_t xhci_pci_take_controller; static device_method_t xhci_device_methods[] = { /* device interface */ DEVMETHOD(device_probe, xhci_pci_probe), DEVMETHOD(device_attach, xhci_pci_attach), DEVMETHOD(device_detach, xhci_pci_detach), DEVMETHOD(device_suspend, bus_generic_suspend), DEVMETHOD(device_resume, bus_generic_resume), DEVMETHOD(device_shutdown, bus_generic_shutdown), DEVMETHOD(usb_take_controller, xhci_pci_take_controller), DEVMETHOD_END }; static driver_t xhci_driver = { .name = "xhci", .methods = xhci_device_methods, .size = sizeof(struct xhci_softc), }; static devclass_t xhci_devclass; DRIVER_MODULE(xhci, pci, xhci_driver, xhci_devclass, 0, 0); MODULE_DEPEND(xhci, usb, 1, 1, 1); static const char * xhci_pci_match(device_t self) { uint32_t device_id = pci_get_devid(self); switch (device_id) { case 0x01941033: return ("NEC uPD720200 USB 3.0 controller"); case 0x10421b21: return ("ASMedia ASM1042 USB 3.0 controller"); case 0x0f358086: return ("Intel Intel BayTrail USB 3.0 controller"); case 0x9c318086: case 0x1e318086: return ("Intel Panther Point USB 3.0 controller"); case 0x8c318086: return ("Intel Lynx Point USB 3.0 controller"); case 0x8cb18086: return ("Intel Wildcat Point USB 3.0 controller"); default: break; } if ((pci_get_class(self) == PCIC_SERIALBUS) && (pci_get_subclass(self) == PCIS_SERIALBUS_USB) && (pci_get_progif(self) == PCIP_SERIALBUS_USB_XHCI)) { return ("XHCI (generic) USB 3.0 controller"); } return (NULL); /* dunno */ } static int xhci_pci_probe(device_t self) { const char *desc = xhci_pci_match(self); if (desc) { device_set_desc(self, desc); return (0); } else { return (ENXIO); } } static int xhci_use_msi = 1; TUNABLE_INT("hw.usb.xhci.msi", &xhci_use_msi); static void xhci_interrupt_poll(void *_sc) { struct xhci_softc *sc = _sc; USB_BUS_UNLOCK(&sc->sc_bus); xhci_interrupt(sc); USB_BUS_LOCK(&sc->sc_bus); usb_callout_reset(&sc->sc_callout, 1, (void *)&xhci_interrupt_poll, sc); } static int xhci_pci_port_route(device_t self, uint32_t set, uint32_t clear) { uint32_t temp; uint32_t usb3_mask; uint32_t usb2_mask; temp = pci_read_config(self, PCI_XHCI_INTEL_USB3_PSSEN, 4) | pci_read_config(self, PCI_XHCI_INTEL_XUSB2PR, 4); temp |= set; temp &= ~clear; /* Don't set bits which the hardware doesn't support */ usb3_mask = pci_read_config(self, PCI_XHCI_INTEL_USB3PRM, 4); usb2_mask = pci_read_config(self, PCI_XHCI_INTEL_USB2PRM, 4); pci_write_config(self, PCI_XHCI_INTEL_USB3_PSSEN, temp & usb3_mask, 4); pci_write_config(self, PCI_XHCI_INTEL_XUSB2PR, temp & usb2_mask, 4); device_printf(self, "Port routing mask set to 0x%08x\n", temp); return (0); } static int xhci_pci_attach(device_t self) { struct xhci_softc *sc = device_get_softc(self); int count, err, rid; + uint8_t usedma32; rid = PCI_XHCI_CBMEM; sc->sc_io_res = bus_alloc_resource_any(self, SYS_RES_MEMORY, &rid, RF_ACTIVE); if (!sc->sc_io_res) { device_printf(self, "Could not map memory\n"); return (ENOMEM); } sc->sc_io_tag = rman_get_bustag(sc->sc_io_res); sc->sc_io_hdl = rman_get_bushandle(sc->sc_io_res); sc->sc_io_size = rman_get_size(sc->sc_io_res); - if (xhci_init(sc, self, 0)) { + /* check for USB 3.0 controllers which don't support 64-bit DMA */ + switch (pci_get_devid(self)) { + case 0x01941033: /* NEC uPD720200 USB 3.0 controller */ + usedma32 = 1; + break; + default: + usedma32 = 0; + break; + } + + if (xhci_init(sc, self, usedma32)) { device_printf(self, "Could not initialize softc\n"); bus_release_resource(self, SYS_RES_MEMORY, PCI_XHCI_CBMEM, sc->sc_io_res); return (ENXIO); } pci_enable_busmaster(self); usb_callout_init_mtx(&sc->sc_callout, &sc->sc_bus.bus_mtx, 0); rid = 0; if (xhci_use_msi) { count = 1; if (pci_alloc_msi(self, &count) == 0) { if (bootverbose) device_printf(self, "MSI enabled\n"); rid = 1; } } sc->sc_irq_res = bus_alloc_resource_any(self, SYS_RES_IRQ, &rid, RF_ACTIVE | (rid != 0 ? 0 : RF_SHAREABLE)); if (sc->sc_irq_res == NULL) { pci_release_msi(self); device_printf(self, "Could not allocate IRQ\n"); /* goto error; FALLTHROUGH - use polling */ } sc->sc_bus.bdev = device_add_child(self, "usbus", -1); if (sc->sc_bus.bdev == NULL) { device_printf(self, "Could not add USB device\n"); goto error; } device_set_ivars(sc->sc_bus.bdev, &sc->sc_bus); sprintf(sc->sc_vendor, "0x%04x", pci_get_vendor(self)); if (sc->sc_irq_res != NULL) { err = bus_setup_intr(self, sc->sc_irq_res, INTR_TYPE_BIO | INTR_MPSAFE, NULL, (driver_intr_t *)xhci_interrupt, sc, &sc->sc_intr_hdl); if (err != 0) { bus_release_resource(self, SYS_RES_IRQ, rman_get_rid(sc->sc_irq_res), sc->sc_irq_res); sc->sc_irq_res = NULL; pci_release_msi(self); device_printf(self, "Could not setup IRQ, err=%d\n", err); sc->sc_intr_hdl = NULL; } } if (sc->sc_irq_res == NULL || sc->sc_intr_hdl == NULL) { if (xhci_use_polling() != 0) { device_printf(self, "Interrupt polling at %dHz\n", hz); USB_BUS_LOCK(&sc->sc_bus); xhci_interrupt_poll(sc); USB_BUS_UNLOCK(&sc->sc_bus); } else goto error; } /* On Intel chipsets reroute ports from EHCI to XHCI controller. */ switch (pci_get_devid(self)) { case 0x0f358086: /* BayTrail */ case 0x9c318086: /* Panther Point */ case 0x1e318086: /* Panther Point */ case 0x8c318086: /* Lynx Point */ case 0x8cb18086: /* Wildcat Point */ sc->sc_port_route = &xhci_pci_port_route; sc->sc_imod_default = XHCI_IMOD_DEFAULT_LP; break; default: break; } xhci_pci_take_controller(self); err = xhci_halt_controller(sc); if (err == 0) err = xhci_start_controller(sc); if (err == 0) err = device_probe_and_attach(sc->sc_bus.bdev); if (err) { device_printf(self, "XHCI halt/start/probe failed err=%d\n", err); goto error; } return (0); error: xhci_pci_detach(self); return (ENXIO); } static int xhci_pci_detach(device_t self) { struct xhci_softc *sc = device_get_softc(self); device_t bdev; if (sc->sc_bus.bdev != NULL) { bdev = sc->sc_bus.bdev; device_detach(bdev); device_delete_child(self, bdev); } /* during module unload there are lots of children leftover */ device_delete_children(self); usb_callout_drain(&sc->sc_callout); xhci_halt_controller(sc); pci_disable_busmaster(self); if (sc->sc_irq_res && sc->sc_intr_hdl) { bus_teardown_intr(self, sc->sc_irq_res, sc->sc_intr_hdl); sc->sc_intr_hdl = NULL; } if (sc->sc_irq_res) { bus_release_resource(self, SYS_RES_IRQ, rman_get_rid(sc->sc_irq_res), sc->sc_irq_res); sc->sc_irq_res = NULL; pci_release_msi(self); } if (sc->sc_io_res) { bus_release_resource(self, SYS_RES_MEMORY, PCI_XHCI_CBMEM, sc->sc_io_res); sc->sc_io_res = NULL; } xhci_uninit(sc); return (0); } static int xhci_pci_take_controller(device_t self) { struct xhci_softc *sc = device_get_softc(self); uint32_t cparams; uint32_t eecp; uint32_t eec; uint16_t to; uint8_t bios_sem; cparams = XREAD4(sc, capa, XHCI_HCSPARAMS0); eec = -1; /* Synchronise with the BIOS if it owns the controller. */ for (eecp = XHCI_HCS0_XECP(cparams) << 2; eecp != 0 && XHCI_XECP_NEXT(eec); eecp += XHCI_XECP_NEXT(eec) << 2) { eec = XREAD4(sc, capa, eecp); if (XHCI_XECP_ID(eec) != XHCI_ID_USB_LEGACY) continue; bios_sem = XREAD1(sc, capa, eecp + XHCI_XECP_BIOS_SEM); if (bios_sem == 0) continue; device_printf(sc->sc_bus.bdev, "waiting for BIOS " "to give up control\n"); XWRITE1(sc, capa, eecp + XHCI_XECP_OS_SEM, 1); to = 500; while (1) { bios_sem = XREAD1(sc, capa, eecp + XHCI_XECP_BIOS_SEM); if (bios_sem == 0) break; if (--to == 0) { device_printf(sc->sc_bus.bdev, "timed out waiting for BIOS\n"); break; } usb_pause_mtx(NULL, hz / 100); /* wait 10ms */ } } return (0); } Index: projects/cxl_iscsi/sys/dev/virtio/block/virtio_blk.c =================================================================== --- projects/cxl_iscsi/sys/dev/virtio/block/virtio_blk.c (revision 279716) +++ projects/cxl_iscsi/sys/dev/virtio/block/virtio_blk.c (revision 279717) @@ -1,1400 +1,1401 @@ /*- * Copyright (c) 2011, Bryan Venteicher * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice unmodified, this list of conditions, and the following * disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /* Driver for VirtIO block devices. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "virtio_if.h" struct vtblk_request { struct virtio_blk_outhdr vbr_hdr; struct bio *vbr_bp; uint8_t vbr_ack; TAILQ_ENTRY(vtblk_request) vbr_link; }; enum vtblk_cache_mode { VTBLK_CACHE_WRITETHROUGH, VTBLK_CACHE_WRITEBACK, VTBLK_CACHE_MAX }; struct vtblk_softc { device_t vtblk_dev; struct mtx vtblk_mtx; uint64_t vtblk_features; uint32_t vtblk_flags; #define VTBLK_FLAG_INDIRECT 0x0001 #define VTBLK_FLAG_READONLY 0x0002 #define VTBLK_FLAG_DETACH 0x0004 #define VTBLK_FLAG_SUSPEND 0x0008 #define VTBLK_FLAG_BARRIER 0x0010 #define VTBLK_FLAG_WC_CONFIG 0x0020 struct virtqueue *vtblk_vq; struct sglist *vtblk_sglist; struct disk *vtblk_disk; struct bio_queue_head vtblk_bioq; TAILQ_HEAD(, vtblk_request) vtblk_req_free; TAILQ_HEAD(, vtblk_request) vtblk_req_ready; struct vtblk_request *vtblk_req_ordered; int vtblk_max_nsegs; int vtblk_request_count; enum vtblk_cache_mode vtblk_write_cache; struct bio_queue vtblk_dump_queue; struct vtblk_request vtblk_dump_request; }; static struct virtio_feature_desc vtblk_feature_desc[] = { { VIRTIO_BLK_F_BARRIER, "HostBarrier" }, { VIRTIO_BLK_F_SIZE_MAX, "MaxSegSize" }, { VIRTIO_BLK_F_SEG_MAX, "MaxNumSegs" }, { VIRTIO_BLK_F_GEOMETRY, "DiskGeometry" }, { VIRTIO_BLK_F_RO, "ReadOnly" }, { VIRTIO_BLK_F_BLK_SIZE, "BlockSize" }, { VIRTIO_BLK_F_SCSI, "SCSICmds" }, { VIRTIO_BLK_F_WCE, "WriteCache" }, { VIRTIO_BLK_F_TOPOLOGY, "Topology" }, { VIRTIO_BLK_F_CONFIG_WCE, "ConfigWCE" }, { 0, NULL } }; static int vtblk_modevent(module_t, int, void *); static int vtblk_probe(device_t); static int vtblk_attach(device_t); static int vtblk_detach(device_t); static int vtblk_suspend(device_t); static int vtblk_resume(device_t); static int vtblk_shutdown(device_t); static int vtblk_config_change(device_t); static int vtblk_open(struct disk *); static int vtblk_close(struct disk *); static int vtblk_ioctl(struct disk *, u_long, void *, int, struct thread *); static int vtblk_dump(void *, void *, vm_offset_t, off_t, size_t); static void vtblk_strategy(struct bio *); static void vtblk_negotiate_features(struct vtblk_softc *); static void vtblk_setup_features(struct vtblk_softc *); static int vtblk_maximum_segments(struct vtblk_softc *, struct virtio_blk_config *); static int vtblk_alloc_virtqueue(struct vtblk_softc *); static void vtblk_resize_disk(struct vtblk_softc *, uint64_t); static void vtblk_alloc_disk(struct vtblk_softc *, struct virtio_blk_config *); static void vtblk_create_disk(struct vtblk_softc *); static int vtblk_request_prealloc(struct vtblk_softc *); static void vtblk_request_free(struct vtblk_softc *); static struct vtblk_request * vtblk_request_dequeue(struct vtblk_softc *); static void vtblk_request_enqueue(struct vtblk_softc *, struct vtblk_request *); static struct vtblk_request * vtblk_request_next_ready(struct vtblk_softc *); static void vtblk_request_requeue_ready(struct vtblk_softc *, struct vtblk_request *); static struct vtblk_request * vtblk_request_next(struct vtblk_softc *); static struct vtblk_request * vtblk_request_bio(struct vtblk_softc *); static int vtblk_request_execute(struct vtblk_softc *, struct vtblk_request *); static int vtblk_request_error(struct vtblk_request *); static void vtblk_queue_completed(struct vtblk_softc *, struct bio_queue *); static void vtblk_done_completed(struct vtblk_softc *, struct bio_queue *); static void vtblk_drain_vq(struct vtblk_softc *); static void vtblk_drain(struct vtblk_softc *); static void vtblk_startio(struct vtblk_softc *); static void vtblk_bio_done(struct vtblk_softc *, struct bio *, int); static void vtblk_read_config(struct vtblk_softc *, struct virtio_blk_config *); static void vtblk_ident(struct vtblk_softc *); static int vtblk_poll_request(struct vtblk_softc *, struct vtblk_request *); static int vtblk_quiesce(struct vtblk_softc *); static void vtblk_vq_intr(void *); static void vtblk_stop(struct vtblk_softc *); static void vtblk_dump_quiesce(struct vtblk_softc *); static int vtblk_dump_write(struct vtblk_softc *, void *, off_t, size_t); static int vtblk_dump_flush(struct vtblk_softc *); static void vtblk_dump_complete(struct vtblk_softc *); static void vtblk_set_write_cache(struct vtblk_softc *, int); static int vtblk_write_cache_enabled(struct vtblk_softc *sc, struct virtio_blk_config *); static int vtblk_write_cache_sysctl(SYSCTL_HANDLER_ARGS); static void vtblk_setup_sysctl(struct vtblk_softc *); static int vtblk_tunable_int(struct vtblk_softc *, const char *, int); /* Tunables. */ static int vtblk_no_ident = 0; TUNABLE_INT("hw.vtblk.no_ident", &vtblk_no_ident); static int vtblk_writecache_mode = -1; TUNABLE_INT("hw.vtblk.writecache_mode", &vtblk_writecache_mode); /* Features desired/implemented by this driver. */ #define VTBLK_FEATURES \ (VIRTIO_BLK_F_BARRIER | \ VIRTIO_BLK_F_SIZE_MAX | \ VIRTIO_BLK_F_SEG_MAX | \ VIRTIO_BLK_F_GEOMETRY | \ VIRTIO_BLK_F_RO | \ VIRTIO_BLK_F_BLK_SIZE | \ VIRTIO_BLK_F_WCE | \ + VIRTIO_BLK_F_TOPOLOGY | \ VIRTIO_BLK_F_CONFIG_WCE | \ VIRTIO_RING_F_INDIRECT_DESC) #define VTBLK_MTX(_sc) &(_sc)->vtblk_mtx #define VTBLK_LOCK_INIT(_sc, _name) \ mtx_init(VTBLK_MTX((_sc)), (_name), \ "VirtIO Block Lock", MTX_DEF) #define VTBLK_LOCK(_sc) mtx_lock(VTBLK_MTX((_sc))) #define VTBLK_UNLOCK(_sc) mtx_unlock(VTBLK_MTX((_sc))) #define VTBLK_LOCK_DESTROY(_sc) mtx_destroy(VTBLK_MTX((_sc))) #define VTBLK_LOCK_ASSERT(_sc) mtx_assert(VTBLK_MTX((_sc)), MA_OWNED) #define VTBLK_LOCK_ASSERT_NOTOWNED(_sc) \ mtx_assert(VTBLK_MTX((_sc)), MA_NOTOWNED) #define VTBLK_DISK_NAME "vtbd" #define VTBLK_QUIESCE_TIMEOUT (30 * hz) /* * Each block request uses at least two segments - one for the header * and one for the status. */ #define VTBLK_MIN_SEGMENTS 2 static device_method_t vtblk_methods[] = { /* Device methods. */ DEVMETHOD(device_probe, vtblk_probe), DEVMETHOD(device_attach, vtblk_attach), DEVMETHOD(device_detach, vtblk_detach), DEVMETHOD(device_suspend, vtblk_suspend), DEVMETHOD(device_resume, vtblk_resume), DEVMETHOD(device_shutdown, vtblk_shutdown), /* VirtIO methods. */ DEVMETHOD(virtio_config_change, vtblk_config_change), DEVMETHOD_END }; static driver_t vtblk_driver = { "vtblk", vtblk_methods, sizeof(struct vtblk_softc) }; static devclass_t vtblk_devclass; DRIVER_MODULE(virtio_blk, virtio_mmio, vtblk_driver, vtblk_devclass, vtblk_modevent, 0); DRIVER_MODULE(virtio_blk, virtio_pci, vtblk_driver, vtblk_devclass, vtblk_modevent, 0); MODULE_VERSION(virtio_blk, 1); MODULE_DEPEND(virtio_blk, virtio, 1, 1, 1); static int vtblk_modevent(module_t mod, int type, void *unused) { int error; error = 0; switch (type) { case MOD_LOAD: case MOD_QUIESCE: case MOD_UNLOAD: case MOD_SHUTDOWN: break; default: error = EOPNOTSUPP; break; } return (error); } static int vtblk_probe(device_t dev) { if (virtio_get_device_type(dev) != VIRTIO_ID_BLOCK) return (ENXIO); device_set_desc(dev, "VirtIO Block Adapter"); return (BUS_PROBE_DEFAULT); } static int vtblk_attach(device_t dev) { struct vtblk_softc *sc; struct virtio_blk_config blkcfg; int error; virtio_set_feature_desc(dev, vtblk_feature_desc); sc = device_get_softc(dev); sc->vtblk_dev = dev; VTBLK_LOCK_INIT(sc, device_get_nameunit(dev)); bioq_init(&sc->vtblk_bioq); TAILQ_INIT(&sc->vtblk_dump_queue); TAILQ_INIT(&sc->vtblk_req_free); TAILQ_INIT(&sc->vtblk_req_ready); vtblk_setup_sysctl(sc); vtblk_setup_features(sc); vtblk_read_config(sc, &blkcfg); /* * With the current sglist(9) implementation, it is not easy * for us to support a maximum segment size as adjacent * segments are coalesced. For now, just make sure it's larger * than the maximum supported transfer size. */ if (virtio_with_feature(dev, VIRTIO_BLK_F_SIZE_MAX)) { if (blkcfg.size_max < MAXPHYS) { error = ENOTSUP; device_printf(dev, "host requires unsupported " "maximum segment size feature\n"); goto fail; } } sc->vtblk_max_nsegs = vtblk_maximum_segments(sc, &blkcfg); if (sc->vtblk_max_nsegs <= VTBLK_MIN_SEGMENTS) { error = EINVAL; device_printf(dev, "fewer than minimum number of segments " "allowed: %d\n", sc->vtblk_max_nsegs); goto fail; } sc->vtblk_sglist = sglist_alloc(sc->vtblk_max_nsegs, M_NOWAIT); if (sc->vtblk_sglist == NULL) { error = ENOMEM; device_printf(dev, "cannot allocate sglist\n"); goto fail; } error = vtblk_alloc_virtqueue(sc); if (error) { device_printf(dev, "cannot allocate virtqueue\n"); goto fail; } error = vtblk_request_prealloc(sc); if (error) { device_printf(dev, "cannot preallocate requests\n"); goto fail; } vtblk_alloc_disk(sc, &blkcfg); error = virtio_setup_intr(dev, INTR_TYPE_BIO | INTR_ENTROPY); if (error) { device_printf(dev, "cannot setup virtqueue interrupt\n"); goto fail; } vtblk_create_disk(sc); virtqueue_enable_intr(sc->vtblk_vq); fail: if (error) vtblk_detach(dev); return (error); } static int vtblk_detach(device_t dev) { struct vtblk_softc *sc; sc = device_get_softc(dev); VTBLK_LOCK(sc); sc->vtblk_flags |= VTBLK_FLAG_DETACH; if (device_is_attached(dev)) vtblk_stop(sc); VTBLK_UNLOCK(sc); vtblk_drain(sc); if (sc->vtblk_disk != NULL) { disk_destroy(sc->vtblk_disk); sc->vtblk_disk = NULL; } if (sc->vtblk_sglist != NULL) { sglist_free(sc->vtblk_sglist); sc->vtblk_sglist = NULL; } VTBLK_LOCK_DESTROY(sc); return (0); } static int vtblk_suspend(device_t dev) { struct vtblk_softc *sc; int error; sc = device_get_softc(dev); VTBLK_LOCK(sc); sc->vtblk_flags |= VTBLK_FLAG_SUSPEND; /* XXX BMV: virtio_stop(), etc needed here? */ error = vtblk_quiesce(sc); if (error) sc->vtblk_flags &= ~VTBLK_FLAG_SUSPEND; VTBLK_UNLOCK(sc); return (error); } static int vtblk_resume(device_t dev) { struct vtblk_softc *sc; sc = device_get_softc(dev); VTBLK_LOCK(sc); /* XXX BMV: virtio_reinit(), etc needed here? */ sc->vtblk_flags &= ~VTBLK_FLAG_SUSPEND; vtblk_startio(sc); VTBLK_UNLOCK(sc); return (0); } static int vtblk_shutdown(device_t dev) { return (0); } static int vtblk_config_change(device_t dev) { struct vtblk_softc *sc; struct virtio_blk_config blkcfg; uint64_t capacity; sc = device_get_softc(dev); vtblk_read_config(sc, &blkcfg); /* Capacity is always in 512-byte units. */ capacity = blkcfg.capacity * 512; if (sc->vtblk_disk->d_mediasize != capacity) vtblk_resize_disk(sc, capacity); return (0); } static int vtblk_open(struct disk *dp) { struct vtblk_softc *sc; if ((sc = dp->d_drv1) == NULL) return (ENXIO); return (sc->vtblk_flags & VTBLK_FLAG_DETACH ? ENXIO : 0); } static int vtblk_close(struct disk *dp) { struct vtblk_softc *sc; if ((sc = dp->d_drv1) == NULL) return (ENXIO); return (0); } static int vtblk_ioctl(struct disk *dp, u_long cmd, void *addr, int flag, struct thread *td) { struct vtblk_softc *sc; if ((sc = dp->d_drv1) == NULL) return (ENXIO); return (ENOTTY); } static int vtblk_dump(void *arg, void *virtual, vm_offset_t physical, off_t offset, size_t length) { struct disk *dp; struct vtblk_softc *sc; int error; dp = arg; error = 0; if ((sc = dp->d_drv1) == NULL) return (ENXIO); VTBLK_LOCK(sc); vtblk_dump_quiesce(sc); if (length > 0) error = vtblk_dump_write(sc, virtual, offset, length); if (error || (virtual == NULL && offset == 0)) vtblk_dump_complete(sc); VTBLK_UNLOCK(sc); return (error); } static void vtblk_strategy(struct bio *bp) { struct vtblk_softc *sc; if ((sc = bp->bio_disk->d_drv1) == NULL) { vtblk_bio_done(NULL, bp, EINVAL); return; } /* * Fail any write if RO. Unfortunately, there does not seem to * be a better way to report our readonly'ness to GEOM above. */ if (sc->vtblk_flags & VTBLK_FLAG_READONLY && (bp->bio_cmd == BIO_WRITE || bp->bio_cmd == BIO_FLUSH)) { vtblk_bio_done(sc, bp, EROFS); return; } VTBLK_LOCK(sc); if (sc->vtblk_flags & VTBLK_FLAG_DETACH) { VTBLK_UNLOCK(sc); vtblk_bio_done(sc, bp, ENXIO); return; } bioq_insert_tail(&sc->vtblk_bioq, bp); vtblk_startio(sc); VTBLK_UNLOCK(sc); } static void vtblk_negotiate_features(struct vtblk_softc *sc) { device_t dev; uint64_t features; dev = sc->vtblk_dev; features = VTBLK_FEATURES; sc->vtblk_features = virtio_negotiate_features(dev, features); } static void vtblk_setup_features(struct vtblk_softc *sc) { device_t dev; dev = sc->vtblk_dev; vtblk_negotiate_features(sc); if (virtio_with_feature(dev, VIRTIO_RING_F_INDIRECT_DESC)) sc->vtblk_flags |= VTBLK_FLAG_INDIRECT; if (virtio_with_feature(dev, VIRTIO_BLK_F_RO)) sc->vtblk_flags |= VTBLK_FLAG_READONLY; if (virtio_with_feature(dev, VIRTIO_BLK_F_BARRIER)) sc->vtblk_flags |= VTBLK_FLAG_BARRIER; if (virtio_with_feature(dev, VIRTIO_BLK_F_CONFIG_WCE)) sc->vtblk_flags |= VTBLK_FLAG_WC_CONFIG; } static int vtblk_maximum_segments(struct vtblk_softc *sc, struct virtio_blk_config *blkcfg) { device_t dev; int nsegs; dev = sc->vtblk_dev; nsegs = VTBLK_MIN_SEGMENTS; if (virtio_with_feature(dev, VIRTIO_BLK_F_SEG_MAX)) { nsegs += MIN(blkcfg->seg_max, MAXPHYS / PAGE_SIZE + 1); if (sc->vtblk_flags & VTBLK_FLAG_INDIRECT) nsegs = MIN(nsegs, VIRTIO_MAX_INDIRECT); } else nsegs += 1; return (nsegs); } static int vtblk_alloc_virtqueue(struct vtblk_softc *sc) { device_t dev; struct vq_alloc_info vq_info; dev = sc->vtblk_dev; VQ_ALLOC_INFO_INIT(&vq_info, sc->vtblk_max_nsegs, vtblk_vq_intr, sc, &sc->vtblk_vq, "%s request", device_get_nameunit(dev)); return (virtio_alloc_virtqueues(dev, 0, 1, &vq_info)); } static void vtblk_resize_disk(struct vtblk_softc *sc, uint64_t new_capacity) { device_t dev; struct disk *dp; int error; dev = sc->vtblk_dev; dp = sc->vtblk_disk; dp->d_mediasize = new_capacity; if (bootverbose) { device_printf(dev, "resized to %juMB (%ju %u byte sectors)\n", (uintmax_t) dp->d_mediasize >> 20, (uintmax_t) dp->d_mediasize / dp->d_sectorsize, dp->d_sectorsize); } error = disk_resize(dp, M_NOWAIT); if (error) { device_printf(dev, "disk_resize(9) failed, error: %d\n", error); } } static void vtblk_alloc_disk(struct vtblk_softc *sc, struct virtio_blk_config *blkcfg) { device_t dev; struct disk *dp; dev = sc->vtblk_dev; sc->vtblk_disk = dp = disk_alloc(); dp->d_open = vtblk_open; dp->d_close = vtblk_close; dp->d_ioctl = vtblk_ioctl; dp->d_strategy = vtblk_strategy; dp->d_name = VTBLK_DISK_NAME; dp->d_unit = device_get_unit(dev); dp->d_drv1 = sc; dp->d_flags = DISKFLAG_CANFLUSHCACHE | DISKFLAG_UNMAPPED_BIO | DISKFLAG_DIRECT_COMPLETION; dp->d_hba_vendor = virtio_get_vendor(dev); dp->d_hba_device = virtio_get_device(dev); dp->d_hba_subvendor = virtio_get_subvendor(dev); dp->d_hba_subdevice = virtio_get_subdevice(dev); if ((sc->vtblk_flags & VTBLK_FLAG_READONLY) == 0) dp->d_dump = vtblk_dump; /* Capacity is always in 512-byte units. */ dp->d_mediasize = blkcfg->capacity * 512; if (virtio_with_feature(dev, VIRTIO_BLK_F_BLK_SIZE)) dp->d_sectorsize = blkcfg->blk_size; else dp->d_sectorsize = 512; /* * The VirtIO maximum I/O size is given in terms of segments. * However, FreeBSD limits I/O size by logical buffer size, not * by physically contiguous pages. Therefore, we have to assume * no pages are contiguous. This may impose an artificially low * maximum I/O size. But in practice, since QEMU advertises 128 * segments, this gives us a maximum IO size of 125 * PAGE_SIZE, * which is typically greater than MAXPHYS. Eventually we should * just advertise MAXPHYS and split buffers that are too big. * * Note we must subtract one additional segment in case of non * page aligned buffers. */ dp->d_maxsize = (sc->vtblk_max_nsegs - VTBLK_MIN_SEGMENTS - 1) * PAGE_SIZE; if (dp->d_maxsize < PAGE_SIZE) dp->d_maxsize = PAGE_SIZE; /* XXX */ if (virtio_with_feature(dev, VIRTIO_BLK_F_GEOMETRY)) { dp->d_fwsectors = blkcfg->geometry.sectors; dp->d_fwheads = blkcfg->geometry.heads; } if (virtio_with_feature(dev, VIRTIO_BLK_F_TOPOLOGY)) { dp->d_stripesize = dp->d_sectorsize * (1 << blkcfg->topology.physical_block_exp); dp->d_stripeoffset = (dp->d_stripesize - blkcfg->topology.alignment_offset * dp->d_sectorsize) % dp->d_stripesize; } if (vtblk_write_cache_enabled(sc, blkcfg) != 0) sc->vtblk_write_cache = VTBLK_CACHE_WRITEBACK; else sc->vtblk_write_cache = VTBLK_CACHE_WRITETHROUGH; } static void vtblk_create_disk(struct vtblk_softc *sc) { struct disk *dp; dp = sc->vtblk_disk; vtblk_ident(sc); device_printf(sc->vtblk_dev, "%juMB (%ju %u byte sectors)\n", (uintmax_t) dp->d_mediasize >> 20, (uintmax_t) dp->d_mediasize / dp->d_sectorsize, dp->d_sectorsize); disk_create(dp, DISK_VERSION); } static int vtblk_request_prealloc(struct vtblk_softc *sc) { struct vtblk_request *req; int i, nreqs; nreqs = virtqueue_size(sc->vtblk_vq); /* * Preallocate sufficient requests to keep the virtqueue full. Each * request consumes VTBLK_MIN_SEGMENTS or more descriptors so reduce * the number allocated when indirect descriptors are not available. */ if ((sc->vtblk_flags & VTBLK_FLAG_INDIRECT) == 0) nreqs /= VTBLK_MIN_SEGMENTS; for (i = 0; i < nreqs; i++) { req = malloc(sizeof(struct vtblk_request), M_DEVBUF, M_NOWAIT); if (req == NULL) return (ENOMEM); MPASS(sglist_count(&req->vbr_hdr, sizeof(req->vbr_hdr)) == 1); MPASS(sglist_count(&req->vbr_ack, sizeof(req->vbr_ack)) == 1); sc->vtblk_request_count++; vtblk_request_enqueue(sc, req); } return (0); } static void vtblk_request_free(struct vtblk_softc *sc) { struct vtblk_request *req; MPASS(TAILQ_EMPTY(&sc->vtblk_req_ready)); while ((req = vtblk_request_dequeue(sc)) != NULL) { sc->vtblk_request_count--; free(req, M_DEVBUF); } KASSERT(sc->vtblk_request_count == 0, ("%s: leaked %d requests", __func__, sc->vtblk_request_count)); } static struct vtblk_request * vtblk_request_dequeue(struct vtblk_softc *sc) { struct vtblk_request *req; req = TAILQ_FIRST(&sc->vtblk_req_free); if (req != NULL) { TAILQ_REMOVE(&sc->vtblk_req_free, req, vbr_link); bzero(req, sizeof(struct vtblk_request)); } return (req); } static void vtblk_request_enqueue(struct vtblk_softc *sc, struct vtblk_request *req) { TAILQ_INSERT_HEAD(&sc->vtblk_req_free, req, vbr_link); } static struct vtblk_request * vtblk_request_next_ready(struct vtblk_softc *sc) { struct vtblk_request *req; req = TAILQ_FIRST(&sc->vtblk_req_ready); if (req != NULL) TAILQ_REMOVE(&sc->vtblk_req_ready, req, vbr_link); return (req); } static void vtblk_request_requeue_ready(struct vtblk_softc *sc, struct vtblk_request *req) { /* NOTE: Currently, there will be at most one request in the queue. */ TAILQ_INSERT_HEAD(&sc->vtblk_req_ready, req, vbr_link); } static struct vtblk_request * vtblk_request_next(struct vtblk_softc *sc) { struct vtblk_request *req; req = vtblk_request_next_ready(sc); if (req != NULL) return (req); return (vtblk_request_bio(sc)); } static struct vtblk_request * vtblk_request_bio(struct vtblk_softc *sc) { struct bio_queue_head *bioq; struct vtblk_request *req; struct bio *bp; bioq = &sc->vtblk_bioq; if (bioq_first(bioq) == NULL) return (NULL); req = vtblk_request_dequeue(sc); if (req == NULL) return (NULL); bp = bioq_takefirst(bioq); req->vbr_bp = bp; req->vbr_ack = -1; req->vbr_hdr.ioprio = 1; switch (bp->bio_cmd) { case BIO_FLUSH: req->vbr_hdr.type = VIRTIO_BLK_T_FLUSH; break; case BIO_READ: req->vbr_hdr.type = VIRTIO_BLK_T_IN; req->vbr_hdr.sector = bp->bio_offset / 512; break; case BIO_WRITE: req->vbr_hdr.type = VIRTIO_BLK_T_OUT; req->vbr_hdr.sector = bp->bio_offset / 512; break; default: panic("%s: bio with unhandled cmd: %d", __func__, bp->bio_cmd); } if (bp->bio_flags & BIO_ORDERED) req->vbr_hdr.type |= VIRTIO_BLK_T_BARRIER; return (req); } static int vtblk_request_execute(struct vtblk_softc *sc, struct vtblk_request *req) { struct virtqueue *vq; struct sglist *sg; struct bio *bp; int ordered, readable, writable, error; vq = sc->vtblk_vq; sg = sc->vtblk_sglist; bp = req->vbr_bp; ordered = 0; writable = 0; /* * Some hosts (such as bhyve) do not implement the barrier feature, * so we emulate it in the driver by allowing the barrier request * to be the only one in flight. */ if ((sc->vtblk_flags & VTBLK_FLAG_BARRIER) == 0) { if (sc->vtblk_req_ordered != NULL) return (EBUSY); if (bp->bio_flags & BIO_ORDERED) { if (!virtqueue_empty(vq)) return (EBUSY); ordered = 1; req->vbr_hdr.type &= ~VIRTIO_BLK_T_BARRIER; } } sglist_reset(sg); sglist_append(sg, &req->vbr_hdr, sizeof(struct virtio_blk_outhdr)); if (bp->bio_cmd == BIO_READ || bp->bio_cmd == BIO_WRITE) { error = sglist_append_bio(sg, bp); if (error || sg->sg_nseg == sg->sg_maxseg) { panic("%s: bio %p data buffer too big %d", __func__, bp, error); } /* BIO_READ means the host writes into our buffer. */ if (bp->bio_cmd == BIO_READ) writable = sg->sg_nseg - 1; } writable++; sglist_append(sg, &req->vbr_ack, sizeof(uint8_t)); readable = sg->sg_nseg - writable; error = virtqueue_enqueue(vq, req, sg, readable, writable); if (error == 0 && ordered) sc->vtblk_req_ordered = req; return (error); } static int vtblk_request_error(struct vtblk_request *req) { int error; switch (req->vbr_ack) { case VIRTIO_BLK_S_OK: error = 0; break; case VIRTIO_BLK_S_UNSUPP: error = ENOTSUP; break; default: error = EIO; break; } return (error); } static void vtblk_queue_completed(struct vtblk_softc *sc, struct bio_queue *queue) { struct vtblk_request *req; struct bio *bp; while ((req = virtqueue_dequeue(sc->vtblk_vq, NULL)) != NULL) { if (sc->vtblk_req_ordered != NULL) { MPASS(sc->vtblk_req_ordered == req); sc->vtblk_req_ordered = NULL; } bp = req->vbr_bp; bp->bio_error = vtblk_request_error(req); TAILQ_INSERT_TAIL(queue, bp, bio_queue); vtblk_request_enqueue(sc, req); } } static void vtblk_done_completed(struct vtblk_softc *sc, struct bio_queue *queue) { struct bio *bp, *tmp; TAILQ_FOREACH_SAFE(bp, queue, bio_queue, tmp) { if (bp->bio_error != 0) disk_err(bp, "hard error", -1, 1); vtblk_bio_done(sc, bp, bp->bio_error); } } static void vtblk_drain_vq(struct vtblk_softc *sc) { struct virtqueue *vq; struct vtblk_request *req; int last; vq = sc->vtblk_vq; last = 0; while ((req = virtqueue_drain(vq, &last)) != NULL) { vtblk_bio_done(sc, req->vbr_bp, ENXIO); vtblk_request_enqueue(sc, req); } sc->vtblk_req_ordered = NULL; KASSERT(virtqueue_empty(vq), ("virtqueue not empty")); } static void vtblk_drain(struct vtblk_softc *sc) { struct bio_queue queue; struct bio_queue_head *bioq; struct vtblk_request *req; struct bio *bp; bioq = &sc->vtblk_bioq; TAILQ_INIT(&queue); if (sc->vtblk_vq != NULL) { vtblk_queue_completed(sc, &queue); vtblk_done_completed(sc, &queue); vtblk_drain_vq(sc); } while ((req = vtblk_request_next_ready(sc)) != NULL) { vtblk_bio_done(sc, req->vbr_bp, ENXIO); vtblk_request_enqueue(sc, req); } while (bioq_first(bioq) != NULL) { bp = bioq_takefirst(bioq); vtblk_bio_done(sc, bp, ENXIO); } vtblk_request_free(sc); } static void vtblk_startio(struct vtblk_softc *sc) { struct virtqueue *vq; struct vtblk_request *req; int enq; VTBLK_LOCK_ASSERT(sc); vq = sc->vtblk_vq; enq = 0; if (sc->vtblk_flags & VTBLK_FLAG_SUSPEND) return; while (!virtqueue_full(vq)) { req = vtblk_request_next(sc); if (req == NULL) break; if (vtblk_request_execute(sc, req) != 0) { vtblk_request_requeue_ready(sc, req); break; } enq++; } if (enq > 0) virtqueue_notify(vq); } static void vtblk_bio_done(struct vtblk_softc *sc, struct bio *bp, int error) { /* Because of GEOM direct dispatch, we cannot hold any locks. */ if (sc != NULL) VTBLK_LOCK_ASSERT_NOTOWNED(sc); if (error) { bp->bio_resid = bp->bio_bcount; bp->bio_error = error; bp->bio_flags |= BIO_ERROR; } biodone(bp); } #define VTBLK_GET_CONFIG(_dev, _feature, _field, _cfg) \ if (virtio_with_feature(_dev, _feature)) { \ virtio_read_device_config(_dev, \ offsetof(struct virtio_blk_config, _field), \ &(_cfg)->_field, sizeof((_cfg)->_field)); \ } static void vtblk_read_config(struct vtblk_softc *sc, struct virtio_blk_config *blkcfg) { device_t dev; dev = sc->vtblk_dev; bzero(blkcfg, sizeof(struct virtio_blk_config)); /* The capacity is always available. */ virtio_read_device_config(dev, offsetof(struct virtio_blk_config, capacity), &blkcfg->capacity, sizeof(blkcfg->capacity)); /* Read the configuration if the feature was negotiated. */ VTBLK_GET_CONFIG(dev, VIRTIO_BLK_F_SIZE_MAX, size_max, blkcfg); VTBLK_GET_CONFIG(dev, VIRTIO_BLK_F_SEG_MAX, seg_max, blkcfg); VTBLK_GET_CONFIG(dev, VIRTIO_BLK_F_GEOMETRY, geometry, blkcfg); VTBLK_GET_CONFIG(dev, VIRTIO_BLK_F_BLK_SIZE, blk_size, blkcfg); VTBLK_GET_CONFIG(dev, VIRTIO_BLK_F_TOPOLOGY, topology, blkcfg); VTBLK_GET_CONFIG(dev, VIRTIO_BLK_F_CONFIG_WCE, writeback, blkcfg); } #undef VTBLK_GET_CONFIG static void vtblk_ident(struct vtblk_softc *sc) { struct bio buf; struct disk *dp; struct vtblk_request *req; int len, error; dp = sc->vtblk_disk; len = MIN(VIRTIO_BLK_ID_BYTES, DISK_IDENT_SIZE); if (vtblk_tunable_int(sc, "no_ident", vtblk_no_ident) != 0) return; req = vtblk_request_dequeue(sc); if (req == NULL) return; req->vbr_ack = -1; req->vbr_hdr.type = VIRTIO_BLK_T_GET_ID; req->vbr_hdr.ioprio = 1; req->vbr_hdr.sector = 0; req->vbr_bp = &buf; bzero(&buf, sizeof(struct bio)); buf.bio_cmd = BIO_READ; buf.bio_data = dp->d_ident; buf.bio_bcount = len; VTBLK_LOCK(sc); error = vtblk_poll_request(sc, req); VTBLK_UNLOCK(sc); vtblk_request_enqueue(sc, req); if (error) { device_printf(sc->vtblk_dev, "error getting device identifier: %d\n", error); } } static int vtblk_poll_request(struct vtblk_softc *sc, struct vtblk_request *req) { struct virtqueue *vq; int error; vq = sc->vtblk_vq; if (!virtqueue_empty(vq)) return (EBUSY); error = vtblk_request_execute(sc, req); if (error) return (error); virtqueue_notify(vq); virtqueue_poll(vq, NULL); error = vtblk_request_error(req); if (error && bootverbose) { device_printf(sc->vtblk_dev, "%s: IO error: %d\n", __func__, error); } return (error); } static int vtblk_quiesce(struct vtblk_softc *sc) { int error; VTBLK_LOCK_ASSERT(sc); error = 0; while (!virtqueue_empty(sc->vtblk_vq)) { if (mtx_sleep(&sc->vtblk_vq, VTBLK_MTX(sc), PRIBIO, "vtblkq", VTBLK_QUIESCE_TIMEOUT) == EWOULDBLOCK) { error = EBUSY; break; } } return (error); } static void vtblk_vq_intr(void *xsc) { struct vtblk_softc *sc; struct virtqueue *vq; struct bio_queue queue; sc = xsc; vq = sc->vtblk_vq; TAILQ_INIT(&queue); VTBLK_LOCK(sc); again: if (sc->vtblk_flags & VTBLK_FLAG_DETACH) goto out; vtblk_queue_completed(sc, &queue); vtblk_startio(sc); if (virtqueue_enable_intr(vq) != 0) { virtqueue_disable_intr(vq); goto again; } if (sc->vtblk_flags & VTBLK_FLAG_SUSPEND) wakeup(&sc->vtblk_vq); out: VTBLK_UNLOCK(sc); vtblk_done_completed(sc, &queue); } static void vtblk_stop(struct vtblk_softc *sc) { virtqueue_disable_intr(sc->vtblk_vq); virtio_stop(sc->vtblk_dev); } static void vtblk_dump_quiesce(struct vtblk_softc *sc) { /* * Spin here until all the requests in-flight at the time of the * dump are completed and queued. The queued requests will be * biodone'd once the dump is finished. */ while (!virtqueue_empty(sc->vtblk_vq)) vtblk_queue_completed(sc, &sc->vtblk_dump_queue); } static int vtblk_dump_write(struct vtblk_softc *sc, void *virtual, off_t offset, size_t length) { struct bio buf; struct vtblk_request *req; req = &sc->vtblk_dump_request; req->vbr_ack = -1; req->vbr_hdr.type = VIRTIO_BLK_T_OUT; req->vbr_hdr.ioprio = 1; req->vbr_hdr.sector = offset / 512; req->vbr_bp = &buf; bzero(&buf, sizeof(struct bio)); buf.bio_cmd = BIO_WRITE; buf.bio_data = virtual; buf.bio_bcount = length; return (vtblk_poll_request(sc, req)); } static int vtblk_dump_flush(struct vtblk_softc *sc) { struct bio buf; struct vtblk_request *req; req = &sc->vtblk_dump_request; req->vbr_ack = -1; req->vbr_hdr.type = VIRTIO_BLK_T_FLUSH; req->vbr_hdr.ioprio = 1; req->vbr_hdr.sector = 0; req->vbr_bp = &buf; bzero(&buf, sizeof(struct bio)); buf.bio_cmd = BIO_FLUSH; return (vtblk_poll_request(sc, req)); } static void vtblk_dump_complete(struct vtblk_softc *sc) { vtblk_dump_flush(sc); VTBLK_UNLOCK(sc); vtblk_done_completed(sc, &sc->vtblk_dump_queue); VTBLK_LOCK(sc); } static void vtblk_set_write_cache(struct vtblk_softc *sc, int wc) { /* Set either writeback (1) or writethrough (0) mode. */ virtio_write_dev_config_1(sc->vtblk_dev, offsetof(struct virtio_blk_config, writeback), wc); } static int vtblk_write_cache_enabled(struct vtblk_softc *sc, struct virtio_blk_config *blkcfg) { int wc; if (sc->vtblk_flags & VTBLK_FLAG_WC_CONFIG) { wc = vtblk_tunable_int(sc, "writecache_mode", vtblk_writecache_mode); if (wc >= 0 && wc < VTBLK_CACHE_MAX) vtblk_set_write_cache(sc, wc); else wc = blkcfg->writeback; } else wc = virtio_with_feature(sc->vtblk_dev, VIRTIO_BLK_F_WCE); return (wc); } static int vtblk_write_cache_sysctl(SYSCTL_HANDLER_ARGS) { struct vtblk_softc *sc; int wc, error; sc = oidp->oid_arg1; wc = sc->vtblk_write_cache; error = sysctl_handle_int(oidp, &wc, 0, req); if (error || req->newptr == NULL) return (error); if ((sc->vtblk_flags & VTBLK_FLAG_WC_CONFIG) == 0) return (EPERM); if (wc < 0 || wc >= VTBLK_CACHE_MAX) return (EINVAL); VTBLK_LOCK(sc); sc->vtblk_write_cache = wc; vtblk_set_write_cache(sc, sc->vtblk_write_cache); VTBLK_UNLOCK(sc); return (0); } static void vtblk_setup_sysctl(struct vtblk_softc *sc) { device_t dev; struct sysctl_ctx_list *ctx; struct sysctl_oid *tree; struct sysctl_oid_list *child; dev = sc->vtblk_dev; ctx = device_get_sysctl_ctx(dev); tree = device_get_sysctl_tree(dev); child = SYSCTL_CHILDREN(tree); SYSCTL_ADD_PROC(ctx, child, OID_AUTO, "writecache_mode", CTLTYPE_INT | CTLFLAG_RW, sc, 0, vtblk_write_cache_sysctl, "I", "Write cache mode (writethrough (0) or writeback (1))"); } static int vtblk_tunable_int(struct vtblk_softc *sc, const char *knob, int def) { char path[64]; snprintf(path, sizeof(path), "hw.vtblk.%d.%s", device_get_unit(sc->vtblk_dev), knob); TUNABLE_INT_FETCH(path, &def); return (def); } Index: projects/cxl_iscsi/sys/dev/virtio/block/virtio_blk.h =================================================================== --- projects/cxl_iscsi/sys/dev/virtio/block/virtio_blk.h (revision 279716) +++ projects/cxl_iscsi/sys/dev/virtio/block/virtio_blk.h (revision 279717) @@ -1,130 +1,130 @@ /*- * This header is BSD licensed so anyone can use the definitions to implement * compatible drivers/servers. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of IBM nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL IBM OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #ifndef _VIRTIO_BLK_H #define _VIRTIO_BLK_H /* Feature bits */ #define VIRTIO_BLK_F_BARRIER 0x0001 /* Does host support barriers? */ #define VIRTIO_BLK_F_SIZE_MAX 0x0002 /* Indicates maximum segment size */ #define VIRTIO_BLK_F_SEG_MAX 0x0004 /* Indicates maximum # of segments */ #define VIRTIO_BLK_F_GEOMETRY 0x0010 /* Legacy geometry available */ #define VIRTIO_BLK_F_RO 0x0020 /* Disk is read-only */ #define VIRTIO_BLK_F_BLK_SIZE 0x0040 /* Block size of disk is available*/ #define VIRTIO_BLK_F_SCSI 0x0080 /* Supports scsi command passthru */ #define VIRTIO_BLK_F_WCE 0x0200 /* Writeback mode enabled after reset */ #define VIRTIO_BLK_F_TOPOLOGY 0x0400 /* Topology information is available */ #define VIRTIO_BLK_F_CONFIG_WCE 0x0800 /* Writeback mode available in config */ #define VIRTIO_BLK_ID_BYTES 20 /* ID string length */ struct virtio_blk_config { /* The capacity (in 512-byte sectors). */ uint64_t capacity; /* The maximum segment size (if VIRTIO_BLK_F_SIZE_MAX) */ uint32_t size_max; /* The maximum number of segments (if VIRTIO_BLK_F_SEG_MAX) */ uint32_t seg_max; /* Geometry of the device (if VIRTIO_BLK_F_GEOMETRY) */ struct virtio_blk_geometry { uint16_t cylinders; uint8_t heads; uint8_t sectors; } geometry; /* Block size of device (if VIRTIO_BLK_F_BLK_SIZE) */ uint32_t blk_size; /* Topology of the device (if VIRTIO_BLK_F_TOPOLOGY) */ struct virtio_blk_topology { uint8_t physical_block_exp; uint8_t alignment_offset; uint16_t min_io_size; - uint16_t opt_io_size; + uint32_t opt_io_size; } topology; /* Writeback mode (if VIRTIO_BLK_F_CONFIG_WCE) */ uint8_t writeback; } __packed; /* * Command types * * Usage is a bit tricky as some bits are used as flags and some are not. * * Rules: * VIRTIO_BLK_T_OUT may be combined with VIRTIO_BLK_T_SCSI_CMD or * VIRTIO_BLK_T_BARRIER. VIRTIO_BLK_T_FLUSH is a command of its own * and may not be combined with any of the other flags. */ /* These two define direction. */ #define VIRTIO_BLK_T_IN 0 #define VIRTIO_BLK_T_OUT 1 /* This bit says it's a scsi command, not an actual read or write. */ #define VIRTIO_BLK_T_SCSI_CMD 2 /* Cache flush command */ #define VIRTIO_BLK_T_FLUSH 4 /* Get device ID command */ #define VIRTIO_BLK_T_GET_ID 8 /* Barrier before this op. */ #define VIRTIO_BLK_T_BARRIER 0x80000000 /* ID string length */ #define VIRTIO_BLK_ID_BYTES 20 /* This is the first element of the read scatter-gather list. */ struct virtio_blk_outhdr { /* VIRTIO_BLK_T* */ uint32_t type; /* io priority. */ uint32_t ioprio; /* Sector (ie. 512 byte offset) */ uint64_t sector; }; struct virtio_scsi_inhdr { uint32_t errors; uint32_t data_len; uint32_t sense_len; uint32_t residual; }; /* And this is the final byte of the write scatter-gather list. */ #define VIRTIO_BLK_S_OK 0 #define VIRTIO_BLK_S_IOERR 1 #define VIRTIO_BLK_S_UNSUPP 2 #endif /* _VIRTIO_BLK_H */ Index: projects/cxl_iscsi/sys/i386/i386/mp_machdep.c =================================================================== --- projects/cxl_iscsi/sys/i386/i386/mp_machdep.c (revision 279716) +++ projects/cxl_iscsi/sys/i386/i386/mp_machdep.c (revision 279717) @@ -1,1682 +1,1682 @@ /*- * Copyright (c) 1996, by Steve Passe * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. The name of the developer may NOT be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include "opt_apic.h" #include "opt_cpu.h" #include "opt_kstack_pages.h" #include "opt_pmap.h" #include "opt_sched.h" #include "opt_smp.h" #if !defined(lint) #if !defined(SMP) #error How did you get here? #endif #ifndef DEV_APIC #error The apic device is required for SMP, add "device apic" to your config file. #endif #if defined(CPU_DISABLE_CMPXCHG) && !defined(COMPILING_LINT) #error SMP not supported with CPU_DISABLE_CMPXCHG #endif #endif /* not lint */ #include #include #include #include /* cngetc() */ #include #ifdef GPROF #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define WARMBOOT_TARGET 0 #define WARMBOOT_OFF (KERNBASE + 0x0467) #define WARMBOOT_SEG (KERNBASE + 0x0469) #define CMOS_REG (0x70) #define CMOS_DATA (0x71) #define BIOS_RESET (0x0f) #define BIOS_WARM (0x0a) /* * this code MUST be enabled here and in mpboot.s. * it follows the very early stages of AP boot by placing values in CMOS ram. * it NORMALLY will never be needed and thus the primitive method for enabling. * #define CHECK_POINTS */ #if defined(CHECK_POINTS) && !defined(PC98) #define CHECK_READ(A) (outb(CMOS_REG, (A)), inb(CMOS_DATA)) #define CHECK_WRITE(A,D) (outb(CMOS_REG, (A)), outb(CMOS_DATA, (D))) #define CHECK_INIT(D); \ CHECK_WRITE(0x34, (D)); \ CHECK_WRITE(0x35, (D)); \ CHECK_WRITE(0x36, (D)); \ CHECK_WRITE(0x37, (D)); \ CHECK_WRITE(0x38, (D)); \ CHECK_WRITE(0x39, (D)); #define CHECK_PRINT(S); \ printf("%s: %d, %d, %d, %d, %d, %d\n", \ (S), \ CHECK_READ(0x34), \ CHECK_READ(0x35), \ CHECK_READ(0x36), \ CHECK_READ(0x37), \ CHECK_READ(0x38), \ CHECK_READ(0x39)); #else /* CHECK_POINTS */ #define CHECK_INIT(D) #define CHECK_PRINT(S) #define CHECK_WRITE(A, D) #endif /* CHECK_POINTS */ /* lock region used by kernel profiling */ int mcount_lock; int mp_naps; /* # of Applications processors */ int boot_cpu_id = -1; /* designated BSP */ extern struct pcpu __pcpu[]; /* AP uses this during bootstrap. Do not staticize. */ char *bootSTK; static int bootAP; /* Free these after use */ void *bootstacks[MAXCPU]; static void *dpcpu; struct pcb stoppcbs[MAXCPU]; struct susppcb **susppcbs; /* Variables needed for SMP tlb shootdown. */ vm_offset_t smp_tlb_addr1; vm_offset_t smp_tlb_addr2; volatile int smp_tlb_wait; #ifdef COUNT_IPIS /* Interrupt counts. */ static u_long *ipi_preempt_counts[MAXCPU]; static u_long *ipi_ast_counts[MAXCPU]; u_long *ipi_invltlb_counts[MAXCPU]; u_long *ipi_invlrng_counts[MAXCPU]; u_long *ipi_invlpg_counts[MAXCPU]; u_long *ipi_invlcache_counts[MAXCPU]; u_long *ipi_rendezvous_counts[MAXCPU]; u_long *ipi_lazypmap_counts[MAXCPU]; static u_long *ipi_hardclock_counts[MAXCPU]; #endif /* Default cpu_ops implementation. */ struct cpu_ops cpu_ops; /* * Local data and functions. */ static volatile cpuset_t ipi_nmi_pending; /* used to hold the AP's until we are ready to release them */ static struct mtx ap_boot_mtx; /* Set to 1 once we're ready to let the APs out of the pen. */ static volatile int aps_ready = 0; /* * Store data from cpu_add() until later in the boot when we actually setup * the APs. */ struct cpu_info { int cpu_present:1; int cpu_bsp:1; int cpu_disabled:1; int cpu_hyperthread:1; } static cpu_info[MAX_APIC_ID + 1]; int cpu_apic_ids[MAXCPU]; int apic_cpuids[MAX_APIC_ID + 1]; /* Holds pending bitmap based IPIs per CPU */ volatile u_int cpu_ipi_pending[MAXCPU]; static u_int boot_address; static int cpu_logical; /* logical cpus per core */ static int cpu_cores; /* cores per package */ static void assign_cpu_ids(void); static void install_ap_tramp(void); static void set_interrupt_apic_ids(void); static int start_all_aps(void); static int start_ap(int apic_id); static void release_aps(void *dummy); static u_int hyperthreading_cpus; /* logical cpus sharing L1 cache */ static int hyperthreading_allowed = 1; static void mem_range_AP_init(void) { if (mem_range_softc.mr_op && mem_range_softc.mr_op->initAP) mem_range_softc.mr_op->initAP(&mem_range_softc); } static void topo_probe_amd(void) { int core_id_bits; int id; /* AMD processors do not support HTT. */ cpu_logical = 1; if ((amd_feature2 & AMDID2_CMP) == 0) { cpu_cores = 1; return; } core_id_bits = (cpu_procinfo2 & AMDID_COREID_SIZE) >> AMDID_COREID_SIZE_SHIFT; if (core_id_bits == 0) { cpu_cores = (cpu_procinfo2 & AMDID_CMP_CORES) + 1; return; } /* Fam 10h and newer should get here. */ for (id = 0; id <= MAX_APIC_ID; id++) { /* Check logical CPU availability. */ if (!cpu_info[id].cpu_present || cpu_info[id].cpu_disabled) continue; /* Check if logical CPU has the same package ID. */ if ((id >> core_id_bits) != (boot_cpu_id >> core_id_bits)) continue; cpu_cores++; } } /* * Round up to the next power of two, if necessary, and then * take log2. * Returns -1 if argument is zero. */ static __inline int mask_width(u_int x) { return (fls(x << (1 - powerof2(x))) - 1); } static void topo_probe_0x4(void) { u_int p[4]; int pkg_id_bits; int core_id_bits; int max_cores; int max_logical; int id; /* Both zero and one here mean one logical processor per package. */ max_logical = (cpu_feature & CPUID_HTT) != 0 ? (cpu_procinfo & CPUID_HTT_CORES) >> 16 : 1; if (max_logical <= 1) return; /* * Because of uniformity assumption we examine only * those logical processors that belong to the same * package as BSP. Further, we count number of * logical processors that belong to the same core * as BSP thus deducing number of threads per core. */ if (cpu_high >= 0x4) { cpuid_count(0x04, 0, p); max_cores = ((p[0] >> 26) & 0x3f) + 1; } else max_cores = 1; core_id_bits = mask_width(max_logical/max_cores); if (core_id_bits < 0) return; pkg_id_bits = core_id_bits + mask_width(max_cores); for (id = 0; id <= MAX_APIC_ID; id++) { /* Check logical CPU availability. */ if (!cpu_info[id].cpu_present || cpu_info[id].cpu_disabled) continue; /* Check if logical CPU has the same package ID. */ if ((id >> pkg_id_bits) != (boot_cpu_id >> pkg_id_bits)) continue; cpu_cores++; /* Check if logical CPU has the same package and core IDs. */ if ((id >> core_id_bits) == (boot_cpu_id >> core_id_bits)) cpu_logical++; } KASSERT(cpu_cores >= 1 && cpu_logical >= 1, ("topo_probe_0x4 couldn't find BSP")); cpu_cores /= cpu_logical; hyperthreading_cpus = cpu_logical; } static void topo_probe_0xb(void) { u_int p[4]; int bits; int cnt; int i; int logical; int type; int x; /* We only support three levels for now. */ for (i = 0; i < 3; i++) { cpuid_count(0x0b, i, p); /* Fall back if CPU leaf 11 doesn't really exist. */ if (i == 0 && p[1] == 0) { topo_probe_0x4(); return; } bits = p[0] & 0x1f; logical = p[1] &= 0xffff; type = (p[2] >> 8) & 0xff; if (type == 0 || logical == 0) break; /* * Because of uniformity assumption we examine only * those logical processors that belong to the same * package as BSP. */ for (cnt = 0, x = 0; x <= MAX_APIC_ID; x++) { if (!cpu_info[x].cpu_present || cpu_info[x].cpu_disabled) continue; if (x >> bits == boot_cpu_id >> bits) cnt++; } if (type == CPUID_TYPE_SMT) cpu_logical = cnt; else if (type == CPUID_TYPE_CORE) cpu_cores = cnt; } if (cpu_logical == 0) cpu_logical = 1; cpu_cores /= cpu_logical; } /* * Both topology discovery code and code that consumes topology * information assume top-down uniformity of the topology. * That is, all physical packages must be identical and each * core in a package must have the same number of threads. * Topology information is queried only on BSP, on which this * code runs and for which it can query CPUID information. * Then topology is extrapolated on all packages using the * uniformity assumption. */ static void topo_probe(void) { static int cpu_topo_probed = 0; if (cpu_topo_probed) return; CPU_ZERO(&logical_cpus_mask); if (mp_ncpus <= 1) cpu_cores = cpu_logical = 1; else if (cpu_vendor_id == CPU_VENDOR_AMD) topo_probe_amd(); else if (cpu_vendor_id == CPU_VENDOR_INTEL) { /* * See Intel(R) 64 Architecture Processor * Topology Enumeration article for details. * * Note that 0x1 <= cpu_high < 4 case should be * compatible with topo_probe_0x4() logic when * CPUID.1:EBX[23:16] > 0 (cpu_cores will be 1) * or it should trigger the fallback otherwise. */ if (cpu_high >= 0xb) topo_probe_0xb(); else if (cpu_high >= 0x1) topo_probe_0x4(); } /* * Fallback: assume each logical CPU is in separate * physical package. That is, no multi-core, no SMT. */ if (cpu_cores == 0 || cpu_logical == 0) cpu_cores = cpu_logical = 1; cpu_topo_probed = 1; } struct cpu_group * cpu_topo(void) { int cg_flags; /* * Determine whether any threading flags are * necessry. */ topo_probe(); if (cpu_logical > 1 && hyperthreading_cpus) cg_flags = CG_FLAG_HTT; else if (cpu_logical > 1) cg_flags = CG_FLAG_SMT; else cg_flags = 0; if (mp_ncpus % (cpu_cores * cpu_logical) != 0) { printf("WARNING: Non-uniform processors.\n"); printf("WARNING: Using suboptimal topology.\n"); return (smp_topo_none()); } /* * No multi-core or hyper-threaded. */ if (cpu_logical * cpu_cores == 1) return (smp_topo_none()); /* * Only HTT no multi-core. */ if (cpu_logical > 1 && cpu_cores == 1) return (smp_topo_1level(CG_SHARE_L1, cpu_logical, cg_flags)); /* * Only multi-core no HTT. */ if (cpu_cores > 1 && cpu_logical == 1) return (smp_topo_1level(CG_SHARE_L2, cpu_cores, cg_flags)); /* * Both HTT and multi-core. */ return (smp_topo_2level(CG_SHARE_L2, cpu_cores, CG_SHARE_L1, cpu_logical, cg_flags)); } /* * Calculate usable address in base memory for AP trampoline code. */ u_int mp_bootaddress(u_int basemem) { boot_address = trunc_page(basemem); /* round down to 4k boundary */ if ((basemem - boot_address) < bootMP_size) boot_address -= PAGE_SIZE; /* not enough, lower by 4k */ return boot_address; } void cpu_add(u_int apic_id, char boot_cpu) { if (apic_id > MAX_APIC_ID) { panic("SMP: APIC ID %d too high", apic_id); return; } KASSERT(cpu_info[apic_id].cpu_present == 0, ("CPU %d added twice", apic_id)); cpu_info[apic_id].cpu_present = 1; if (boot_cpu) { KASSERT(boot_cpu_id == -1, ("CPU %d claims to be BSP, but CPU %d already is", apic_id, boot_cpu_id)); boot_cpu_id = apic_id; cpu_info[apic_id].cpu_bsp = 1; } if (mp_ncpus < MAXCPU) { mp_ncpus++; mp_maxid = mp_ncpus - 1; } if (bootverbose) printf("SMP: Added CPU %d (%s)\n", apic_id, boot_cpu ? "BSP" : "AP"); } void cpu_mp_setmaxid(void) { /* * mp_maxid should be already set by calls to cpu_add(). * Just sanity check its value here. */ if (mp_ncpus == 0) KASSERT(mp_maxid == 0, ("%s: mp_ncpus is zero, but mp_maxid is not", __func__)); else if (mp_ncpus == 1) mp_maxid = 0; else KASSERT(mp_maxid >= mp_ncpus - 1, ("%s: counters out of sync: max %d, count %d", __func__, mp_maxid, mp_ncpus)); } int cpu_mp_probe(void) { /* * Always record BSP in CPU map so that the mbuf init code works * correctly. */ CPU_SETOF(0, &all_cpus); if (mp_ncpus == 0) { /* * No CPUs were found, so this must be a UP system. Setup * the variables to represent a system with a single CPU * with an id of 0. */ mp_ncpus = 1; return (0); } /* At least one CPU was found. */ if (mp_ncpus == 1) { /* * One CPU was found, so this must be a UP system with * an I/O APIC. */ mp_maxid = 0; return (0); } /* At least two CPUs were found. */ return (1); } /* * Initialize the IPI handlers and start up the AP's. */ void cpu_mp_start(void) { int i; /* Initialize the logical ID to APIC ID table. */ for (i = 0; i < MAXCPU; i++) { cpu_apic_ids[i] = -1; cpu_ipi_pending[i] = 0; } /* Install an inter-CPU IPI for TLB invalidation */ setidt(IPI_INVLTLB, IDTVEC(invltlb), SDT_SYS386IGT, SEL_KPL, GSEL(GCODE_SEL, SEL_KPL)); setidt(IPI_INVLPG, IDTVEC(invlpg), SDT_SYS386IGT, SEL_KPL, GSEL(GCODE_SEL, SEL_KPL)); setidt(IPI_INVLRNG, IDTVEC(invlrng), SDT_SYS386IGT, SEL_KPL, GSEL(GCODE_SEL, SEL_KPL)); /* Install an inter-CPU IPI for cache invalidation. */ setidt(IPI_INVLCACHE, IDTVEC(invlcache), SDT_SYS386IGT, SEL_KPL, GSEL(GCODE_SEL, SEL_KPL)); /* Install an inter-CPU IPI for lazy pmap release */ setidt(IPI_LAZYPMAP, IDTVEC(lazypmap), SDT_SYS386IGT, SEL_KPL, GSEL(GCODE_SEL, SEL_KPL)); /* Install an inter-CPU IPI for all-CPU rendezvous */ setidt(IPI_RENDEZVOUS, IDTVEC(rendezvous), SDT_SYS386IGT, SEL_KPL, GSEL(GCODE_SEL, SEL_KPL)); /* Install generic inter-CPU IPI handler */ setidt(IPI_BITMAP_VECTOR, IDTVEC(ipi_intr_bitmap_handler), SDT_SYS386IGT, SEL_KPL, GSEL(GCODE_SEL, SEL_KPL)); /* Install an inter-CPU IPI for CPU stop/restart */ setidt(IPI_STOP, IDTVEC(cpustop), SDT_SYS386IGT, SEL_KPL, GSEL(GCODE_SEL, SEL_KPL)); /* Install an inter-CPU IPI for CPU suspend/resume */ setidt(IPI_SUSPEND, IDTVEC(cpususpend), SDT_SYS386IGT, SEL_KPL, GSEL(GCODE_SEL, SEL_KPL)); /* Set boot_cpu_id if needed. */ if (boot_cpu_id == -1) { boot_cpu_id = PCPU_GET(apic_id); cpu_info[boot_cpu_id].cpu_bsp = 1; } else KASSERT(boot_cpu_id == PCPU_GET(apic_id), ("BSP's APIC ID doesn't match boot_cpu_id")); /* Probe logical/physical core configuration. */ topo_probe(); assign_cpu_ids(); /* Start each Application Processor */ start_all_aps(); set_interrupt_apic_ids(); } /* * Print various information about the SMP system hardware and setup. */ void cpu_mp_announce(void) { const char *hyperthread; int i; printf("FreeBSD/SMP: %d package(s) x %d core(s)", mp_ncpus / (cpu_cores * cpu_logical), cpu_cores); if (hyperthreading_cpus > 1) printf(" x %d HTT threads", cpu_logical); else if (cpu_logical > 1) printf(" x %d SMT threads", cpu_logical); printf("\n"); /* List active CPUs first. */ printf(" cpu0 (BSP): APIC ID: %2d\n", boot_cpu_id); for (i = 1; i < mp_ncpus; i++) { if (cpu_info[cpu_apic_ids[i]].cpu_hyperthread) hyperthread = "/HT"; else hyperthread = ""; printf(" cpu%d (AP%s): APIC ID: %2d\n", i, hyperthread, cpu_apic_ids[i]); } /* List disabled CPUs last. */ for (i = 0; i <= MAX_APIC_ID; i++) { if (!cpu_info[i].cpu_present || !cpu_info[i].cpu_disabled) continue; if (cpu_info[i].cpu_hyperthread) hyperthread = "/HT"; else hyperthread = ""; printf(" cpu (AP%s): APIC ID: %2d (disabled)\n", hyperthread, i); } } /* * AP CPU's call this to initialize themselves. */ void init_secondary(void) { struct pcpu *pc; vm_offset_t addr; int gsel_tss; int x, myid; u_int cpuid, cr0; /* bootAP is set in start_ap() to our ID. */ myid = bootAP; /* Get per-cpu data */ pc = &__pcpu[myid]; /* prime data page for it to use */ pcpu_init(pc, myid, sizeof(struct pcpu)); dpcpu_init(dpcpu, myid); pc->pc_apic_id = cpu_apic_ids[myid]; pc->pc_prvspace = pc; pc->pc_curthread = 0; gdt_segs[GPRIV_SEL].ssd_base = (int) pc; gdt_segs[GPROC0_SEL].ssd_base = (int) &pc->pc_common_tss; for (x = 0; x < NGDT; x++) { ssdtosd(&gdt_segs[x], &gdt[myid * NGDT + x].sd); } r_gdt.rd_limit = NGDT * sizeof(gdt[0]) - 1; r_gdt.rd_base = (int) &gdt[myid * NGDT]; lgdt(&r_gdt); /* does magic intra-segment return */ lidt(&r_idt); lldt(_default_ldt); PCPU_SET(currentldt, _default_ldt); gsel_tss = GSEL(GPROC0_SEL, SEL_KPL); gdt[myid * NGDT + GPROC0_SEL].sd.sd_type = SDT_SYS386TSS; PCPU_SET(common_tss.tss_esp0, 0); /* not used until after switch */ PCPU_SET(common_tss.tss_ss0, GSEL(GDATA_SEL, SEL_KPL)); PCPU_SET(common_tss.tss_ioopt, (sizeof (struct i386tss)) << 16); PCPU_SET(tss_gdt, &gdt[myid * NGDT + GPROC0_SEL].sd); PCPU_SET(common_tssd, *PCPU_GET(tss_gdt)); ltr(gsel_tss); PCPU_SET(fsgs_gdt, &gdt[myid * NGDT + GUFS_SEL].sd); /* * Set to a known state: * Set by mpboot.s: CR0_PG, CR0_PE * Set by cpu_setregs: CR0_NE, CR0_MP, CR0_TS, CR0_WP, CR0_AM */ cr0 = rcr0(); cr0 &= ~(CR0_CD | CR0_NW | CR0_EM); load_cr0(cr0); CHECK_WRITE(0x38, 5); /* signal our startup to the BSP. */ mp_naps++; CHECK_WRITE(0x39, 6); /* Spin until the BSP releases the AP's. */ while (!aps_ready) ia32_pause(); /* BSP may have changed PTD while we were waiting */ invltlb(); for (addr = 0; addr < NKPT * NBPDR - 1; addr += PAGE_SIZE) invlpg(addr); #if defined(I586_CPU) && !defined(NO_F00F_HACK) lidt(&r_idt); #endif /* * On real hardware, switch to x2apic mode if possible. Do it * after aps_ready was signalled, to avoid manipulating the * mode while BSP might still want to send some IPI to us * (second startup IPI is ignored on modern hardware etc). */ lapic_xapic_mode(); /* Initialize the PAT MSR if present. */ pmap_init_pat(); /* set up CPU registers and state */ cpu_setregs(); /* set up SSE/NX */ initializecpu(); /* set up FPU state on the AP */ npxinit(false); if (cpu_ops.cpu_init) cpu_ops.cpu_init(); /* A quick check from sanity claus */ cpuid = PCPU_GET(cpuid); if (PCPU_GET(apic_id) != lapic_id()) { printf("SMP: cpuid = %d\n", cpuid); printf("SMP: actual apic_id = %d\n", lapic_id()); printf("SMP: correct apic_id = %d\n", PCPU_GET(apic_id)); panic("cpuid mismatch! boom!!"); } /* Initialize curthread. */ KASSERT(PCPU_GET(idlethread) != NULL, ("no idle thread")); PCPU_SET(curthread, PCPU_GET(idlethread)); mca_init(); mtx_lock_spin(&ap_boot_mtx); /* Init local apic for irq's */ lapic_setup(1); /* Set memory range attributes for this CPU to match the BSP */ mem_range_AP_init(); smp_cpus++; CTR1(KTR_SMP, "SMP: AP CPU #%d Launched", cpuid); printf("SMP: AP CPU #%d Launched!\n", cpuid); /* Determine if we are a logical CPU. */ /* XXX Calculation depends on cpu_logical being a power of 2, e.g. 2 */ if (cpu_logical > 1 && PCPU_GET(apic_id) % cpu_logical != 0) CPU_SET(cpuid, &logical_cpus_mask); if (bootverbose) lapic_dump("AP"); if (smp_cpus == mp_ncpus) { /* enable IPI's, tlb shootdown, freezes etc */ atomic_store_rel_int(&smp_started, 1); } mtx_unlock_spin(&ap_boot_mtx); /* Wait until all the AP's are up. */ while (smp_started == 0) ia32_pause(); /* Start per-CPU event timers. */ cpu_initclocks_ap(); /* Enter the scheduler. */ sched_throw(NULL); panic("scheduler returned us to %s", __func__); /* NOTREACHED */ } /******************************************************************* * local functions and data */ /* * We tell the I/O APIC code about all the CPUs we want to receive * interrupts. If we don't want certain CPUs to receive IRQs we * can simply not tell the I/O APIC code about them in this function. * We also do not tell it about the BSP since it tells itself about * the BSP internally to work with UP kernels and on UP machines. */ static void set_interrupt_apic_ids(void) { u_int i, apic_id; for (i = 0; i < MAXCPU; i++) { apic_id = cpu_apic_ids[i]; if (apic_id == -1) continue; if (cpu_info[apic_id].cpu_bsp) continue; if (cpu_info[apic_id].cpu_disabled) continue; /* Don't let hyperthreads service interrupts. */ - if (hyperthreading_cpus > 1 && - apic_id % hyperthreading_cpus != 0) + if (cpu_logical > 1 && + apic_id % cpu_logical != 0) continue; intr_add_cpu(i); } } /* * Assign logical CPU IDs to local APICs. */ static void assign_cpu_ids(void) { u_int i; TUNABLE_INT_FETCH("machdep.hyperthreading_allowed", &hyperthreading_allowed); /* Check for explicitly disabled CPUs. */ for (i = 0; i <= MAX_APIC_ID; i++) { if (!cpu_info[i].cpu_present || cpu_info[i].cpu_bsp) continue; if (hyperthreading_cpus > 1 && i % hyperthreading_cpus != 0) { cpu_info[i].cpu_hyperthread = 1; /* * Don't use HT CPU if it has been disabled by a * tunable. */ if (hyperthreading_allowed == 0) { cpu_info[i].cpu_disabled = 1; continue; } } /* Don't use this CPU if it has been disabled by a tunable. */ if (resource_disabled("lapic", i)) { cpu_info[i].cpu_disabled = 1; continue; } } if (hyperthreading_allowed == 0 && hyperthreading_cpus > 1) { hyperthreading_cpus = 0; cpu_logical = 1; } /* * Assign CPU IDs to local APIC IDs and disable any CPUs * beyond MAXCPU. CPU 0 is always assigned to the BSP. * * To minimize confusion for userland, we attempt to number * CPUs such that all threads and cores in a package are * grouped together. For now we assume that the BSP is always * the first thread in a package and just start adding APs * starting with the BSP's APIC ID. */ mp_ncpus = 1; cpu_apic_ids[0] = boot_cpu_id; apic_cpuids[boot_cpu_id] = 0; for (i = boot_cpu_id + 1; i != boot_cpu_id; i == MAX_APIC_ID ? i = 0 : i++) { if (!cpu_info[i].cpu_present || cpu_info[i].cpu_bsp || cpu_info[i].cpu_disabled) continue; if (mp_ncpus < MAXCPU) { cpu_apic_ids[mp_ncpus] = i; apic_cpuids[i] = mp_ncpus; mp_ncpus++; } else cpu_info[i].cpu_disabled = 1; } KASSERT(mp_maxid >= mp_ncpus - 1, ("%s: counters out of sync: max %d, count %d", __func__, mp_maxid, mp_ncpus)); } /* * start each AP in our list */ /* Lowest 1MB is already mapped: don't touch*/ #define TMPMAP_START 1 static int start_all_aps(void) { #ifndef PC98 u_char mpbiosreason; #endif u_int32_t mpbioswarmvec; int apic_id, cpu, i; mtx_init(&ap_boot_mtx, "ap boot", NULL, MTX_SPIN); /* install the AP 1st level boot code */ install_ap_tramp(); /* save the current value of the warm-start vector */ mpbioswarmvec = *((u_int32_t *) WARMBOOT_OFF); #ifndef PC98 outb(CMOS_REG, BIOS_RESET); mpbiosreason = inb(CMOS_DATA); #endif /* set up temporary P==V mapping for AP boot */ /* XXX this is a hack, we should boot the AP on its own stack/PTD */ for (i = TMPMAP_START; i < NKPT; i++) PTD[i] = PTD[KPTDI + i]; invltlb(); /* start each AP */ for (cpu = 1; cpu < mp_ncpus; cpu++) { apic_id = cpu_apic_ids[cpu]; /* allocate and set up a boot stack data page */ bootstacks[cpu] = (char *)kmem_malloc(kernel_arena, KSTACK_PAGES * PAGE_SIZE, M_WAITOK | M_ZERO); dpcpu = (void *)kmem_malloc(kernel_arena, DPCPU_SIZE, M_WAITOK | M_ZERO); /* setup a vector to our boot code */ *((volatile u_short *) WARMBOOT_OFF) = WARMBOOT_TARGET; *((volatile u_short *) WARMBOOT_SEG) = (boot_address >> 4); #ifndef PC98 outb(CMOS_REG, BIOS_RESET); outb(CMOS_DATA, BIOS_WARM); /* 'warm-start' */ #endif bootSTK = (char *)bootstacks[cpu] + KSTACK_PAGES * PAGE_SIZE - 4; bootAP = cpu; /* attempt to start the Application Processor */ CHECK_INIT(99); /* setup checkpoints */ if (!start_ap(apic_id)) { printf("AP #%d (PHY# %d) failed!\n", cpu, apic_id); CHECK_PRINT("trace"); /* show checkpoints */ /* better panic as the AP may be running loose */ printf("panic y/n? [y] "); if (cngetc() != 'n') panic("bye-bye"); } CHECK_PRINT("trace"); /* show checkpoints */ CPU_SET(cpu, &all_cpus); /* record AP in CPU map */ } /* restore the warmstart vector */ *(u_int32_t *) WARMBOOT_OFF = mpbioswarmvec; #ifndef PC98 outb(CMOS_REG, BIOS_RESET); outb(CMOS_DATA, mpbiosreason); #endif /* Undo V==P hack from above */ for (i = TMPMAP_START; i < NKPT; i++) PTD[i] = 0; pmap_invalidate_range(kernel_pmap, 0, NKPT * NBPDR - 1); /* number of APs actually started */ return mp_naps; } /* * load the 1st level AP boot code into base memory. */ /* targets for relocation */ extern void bigJump(void); extern void bootCodeSeg(void); extern void bootDataSeg(void); extern void MPentry(void); extern u_int MP_GDT; extern u_int mp_gdtbase; static void install_ap_tramp(void) { int x; int size = *(int *) ((u_long) & bootMP_size); vm_offset_t va = boot_address + KERNBASE; u_char *src = (u_char *) ((u_long) bootMP); u_char *dst = (u_char *) va; u_int boot_base = (u_int) bootMP; u_int8_t *dst8; u_int16_t *dst16; u_int32_t *dst32; KASSERT (size <= PAGE_SIZE, ("'size' do not fit into PAGE_SIZE, as expected.")); pmap_kenter(va, boot_address); pmap_invalidate_page (kernel_pmap, va); for (x = 0; x < size; ++x) *dst++ = *src++; /* * modify addresses in code we just moved to basemem. unfortunately we * need fairly detailed info about mpboot.s for this to work. changes * to mpboot.s might require changes here. */ /* boot code is located in KERNEL space */ dst = (u_char *) va; /* modify the lgdt arg */ dst32 = (u_int32_t *) (dst + ((u_int) & mp_gdtbase - boot_base)); *dst32 = boot_address + ((u_int) & MP_GDT - boot_base); /* modify the ljmp target for MPentry() */ dst32 = (u_int32_t *) (dst + ((u_int) bigJump - boot_base) + 1); *dst32 = ((u_int) MPentry - KERNBASE); /* modify the target for boot code segment */ dst16 = (u_int16_t *) (dst + ((u_int) bootCodeSeg - boot_base)); dst8 = (u_int8_t *) (dst16 + 1); *dst16 = (u_int) boot_address & 0xffff; *dst8 = ((u_int) boot_address >> 16) & 0xff; /* modify the target for boot data segment */ dst16 = (u_int16_t *) (dst + ((u_int) bootDataSeg - boot_base)); dst8 = (u_int8_t *) (dst16 + 1); *dst16 = (u_int) boot_address & 0xffff; *dst8 = ((u_int) boot_address >> 16) & 0xff; } /* * This function starts the AP (application processor) identified * by the APIC ID 'physicalCpu'. It does quite a "song and dance" * to accomplish this. This is necessary because of the nuances * of the different hardware we might encounter. It isn't pretty, * but it seems to work. */ static int start_ap(int apic_id) { int vector, ms; int cpus; /* calculate the vector */ vector = (boot_address >> 12) & 0xff; /* used as a watchpoint to signal AP startup */ cpus = mp_naps; ipi_startup(apic_id, vector); /* Wait up to 5 seconds for it to start. */ for (ms = 0; ms < 5000; ms++) { if (mp_naps > cpus) return 1; /* return SUCCESS */ DELAY(1000); } return 0; /* return FAILURE */ } #ifdef COUNT_XINVLTLB_HITS u_int xhits_gbl[MAXCPU]; u_int xhits_pg[MAXCPU]; u_int xhits_rng[MAXCPU]; static SYSCTL_NODE(_debug, OID_AUTO, xhits, CTLFLAG_RW, 0, ""); SYSCTL_OPAQUE(_debug_xhits, OID_AUTO, global, CTLFLAG_RW, &xhits_gbl, sizeof(xhits_gbl), "IU", ""); SYSCTL_OPAQUE(_debug_xhits, OID_AUTO, page, CTLFLAG_RW, &xhits_pg, sizeof(xhits_pg), "IU", ""); SYSCTL_OPAQUE(_debug_xhits, OID_AUTO, range, CTLFLAG_RW, &xhits_rng, sizeof(xhits_rng), "IU", ""); u_int ipi_global; u_int ipi_page; u_int ipi_range; u_int ipi_range_size; SYSCTL_INT(_debug_xhits, OID_AUTO, ipi_global, CTLFLAG_RW, &ipi_global, 0, ""); SYSCTL_INT(_debug_xhits, OID_AUTO, ipi_page, CTLFLAG_RW, &ipi_page, 0, ""); SYSCTL_INT(_debug_xhits, OID_AUTO, ipi_range, CTLFLAG_RW, &ipi_range, 0, ""); SYSCTL_INT(_debug_xhits, OID_AUTO, ipi_range_size, CTLFLAG_RW, &ipi_range_size, 0, ""); u_int ipi_masked_global; u_int ipi_masked_page; u_int ipi_masked_range; u_int ipi_masked_range_size; SYSCTL_INT(_debug_xhits, OID_AUTO, ipi_masked_global, CTLFLAG_RW, &ipi_masked_global, 0, ""); SYSCTL_INT(_debug_xhits, OID_AUTO, ipi_masked_page, CTLFLAG_RW, &ipi_masked_page, 0, ""); SYSCTL_INT(_debug_xhits, OID_AUTO, ipi_masked_range, CTLFLAG_RW, &ipi_masked_range, 0, ""); SYSCTL_INT(_debug_xhits, OID_AUTO, ipi_masked_range_size, CTLFLAG_RW, &ipi_masked_range_size, 0, ""); #endif /* COUNT_XINVLTLB_HITS */ /* * Init and startup IPI. */ void ipi_startup(int apic_id, int vector) { /* * This attempts to follow the algorithm described in the * Intel Multiprocessor Specification v1.4 in section B.4. * For each IPI, we allow the local APIC ~20us to deliver the * IPI. If that times out, we panic. */ /* * first we do an INIT IPI: this INIT IPI might be run, resetting * and running the target CPU. OR this INIT IPI might be latched (P5 * bug), CPU waiting for STARTUP IPI. OR this INIT IPI might be * ignored. */ lapic_ipi_raw(APIC_DEST_DESTFLD | APIC_TRIGMOD_LEVEL | APIC_LEVEL_ASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_INIT, apic_id); lapic_ipi_wait(20); /* Explicitly deassert the INIT IPI. */ lapic_ipi_raw(APIC_DEST_DESTFLD | APIC_TRIGMOD_LEVEL | APIC_LEVEL_DEASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_INIT, apic_id); DELAY(10000); /* wait ~10mS */ /* * next we do a STARTUP IPI: the previous INIT IPI might still be * latched, (P5 bug) this 1st STARTUP would then terminate * immediately, and the previously started INIT IPI would continue. OR * the previous INIT IPI has already run. and this STARTUP IPI will * run. OR the previous INIT IPI was ignored. and this STARTUP IPI * will run. */ lapic_ipi_raw(APIC_DEST_DESTFLD | APIC_TRIGMOD_EDGE | APIC_LEVEL_ASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_STARTUP | vector, apic_id); if (!lapic_ipi_wait(20)) panic("Failed to deliver first STARTUP IPI to APIC %d", apic_id); DELAY(200); /* wait ~200uS */ /* * finally we do a 2nd STARTUP IPI: this 2nd STARTUP IPI should run IF * the previous STARTUP IPI was cancelled by a latched INIT IPI. OR * this STARTUP IPI will be ignored, as only ONE STARTUP IPI is * recognized after hardware RESET or INIT IPI. */ lapic_ipi_raw(APIC_DEST_DESTFLD | APIC_TRIGMOD_EDGE | APIC_LEVEL_ASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_STARTUP | vector, apic_id); if (!lapic_ipi_wait(20)) panic("Failed to deliver second STARTUP IPI to APIC %d", apic_id); DELAY(200); /* wait ~200uS */ } /* * Send an IPI to specified CPU handling the bitmap logic. */ static void ipi_send_cpu(int cpu, u_int ipi) { u_int bitmap, old_pending, new_pending; KASSERT(cpu_apic_ids[cpu] != -1, ("IPI to non-existent CPU %d", cpu)); if (IPI_IS_BITMAPED(ipi)) { bitmap = 1 << ipi; ipi = IPI_BITMAP_VECTOR; do { old_pending = cpu_ipi_pending[cpu]; new_pending = old_pending | bitmap; } while (!atomic_cmpset_int(&cpu_ipi_pending[cpu], old_pending, new_pending)); if (old_pending) return; } lapic_ipi_vectored(ipi, cpu_apic_ids[cpu]); } /* * Flush the TLB on all other CPU's */ static void smp_tlb_shootdown(u_int vector, vm_offset_t addr1, vm_offset_t addr2) { u_int ncpu; ncpu = mp_ncpus - 1; /* does not shootdown self */ if (ncpu < 1) return; /* no other cpus */ if (!(read_eflags() & PSL_I)) panic("%s: interrupts disabled", __func__); mtx_lock_spin(&smp_ipi_mtx); smp_tlb_addr1 = addr1; smp_tlb_addr2 = addr2; atomic_store_rel_int(&smp_tlb_wait, 0); ipi_all_but_self(vector); while (smp_tlb_wait < ncpu) ia32_pause(); mtx_unlock_spin(&smp_ipi_mtx); } static void smp_targeted_tlb_shootdown(cpuset_t mask, u_int vector, vm_offset_t addr1, vm_offset_t addr2) { int cpu, ncpu, othercpus; othercpus = mp_ncpus - 1; if (CPU_ISFULLSET(&mask)) { if (othercpus < 1) return; } else { CPU_CLR(PCPU_GET(cpuid), &mask); if (CPU_EMPTY(&mask)) return; } if (!(read_eflags() & PSL_I)) panic("%s: interrupts disabled", __func__); mtx_lock_spin(&smp_ipi_mtx); smp_tlb_addr1 = addr1; smp_tlb_addr2 = addr2; atomic_store_rel_int(&smp_tlb_wait, 0); if (CPU_ISFULLSET(&mask)) { ncpu = othercpus; ipi_all_but_self(vector); } else { ncpu = 0; while ((cpu = CPU_FFS(&mask)) != 0) { cpu--; CPU_CLR(cpu, &mask); CTR3(KTR_SMP, "%s: cpu: %d ipi: %x", __func__, cpu, vector); ipi_send_cpu(cpu, vector); ncpu++; } } while (smp_tlb_wait < ncpu) ia32_pause(); mtx_unlock_spin(&smp_ipi_mtx); } void smp_cache_flush(void) { if (smp_started) smp_tlb_shootdown(IPI_INVLCACHE, 0, 0); } void smp_invltlb(void) { if (smp_started) { smp_tlb_shootdown(IPI_INVLTLB, 0, 0); #ifdef COUNT_XINVLTLB_HITS ipi_global++; #endif } } void smp_invlpg(vm_offset_t addr) { if (smp_started) { smp_tlb_shootdown(IPI_INVLPG, addr, 0); #ifdef COUNT_XINVLTLB_HITS ipi_page++; #endif } } void smp_invlpg_range(vm_offset_t addr1, vm_offset_t addr2) { if (smp_started) { smp_tlb_shootdown(IPI_INVLRNG, addr1, addr2); #ifdef COUNT_XINVLTLB_HITS ipi_range++; ipi_range_size += (addr2 - addr1) / PAGE_SIZE; #endif } } void smp_masked_invltlb(cpuset_t mask) { if (smp_started) { smp_targeted_tlb_shootdown(mask, IPI_INVLTLB, 0, 0); #ifdef COUNT_XINVLTLB_HITS ipi_masked_global++; #endif } } void smp_masked_invlpg(cpuset_t mask, vm_offset_t addr) { if (smp_started) { smp_targeted_tlb_shootdown(mask, IPI_INVLPG, addr, 0); #ifdef COUNT_XINVLTLB_HITS ipi_masked_page++; #endif } } void smp_masked_invlpg_range(cpuset_t mask, vm_offset_t addr1, vm_offset_t addr2) { if (smp_started) { smp_targeted_tlb_shootdown(mask, IPI_INVLRNG, addr1, addr2); #ifdef COUNT_XINVLTLB_HITS ipi_masked_range++; ipi_masked_range_size += (addr2 - addr1) / PAGE_SIZE; #endif } } void ipi_bitmap_handler(struct trapframe frame) { struct trapframe *oldframe; struct thread *td; int cpu = PCPU_GET(cpuid); u_int ipi_bitmap; critical_enter(); td = curthread; td->td_intr_nesting_level++; oldframe = td->td_intr_frame; td->td_intr_frame = &frame; ipi_bitmap = atomic_readandclear_int(&cpu_ipi_pending[cpu]); if (ipi_bitmap & (1 << IPI_PREEMPT)) { #ifdef COUNT_IPIS (*ipi_preempt_counts[cpu])++; #endif sched_preempt(td); } if (ipi_bitmap & (1 << IPI_AST)) { #ifdef COUNT_IPIS (*ipi_ast_counts[cpu])++; #endif /* Nothing to do for AST */ } if (ipi_bitmap & (1 << IPI_HARDCLOCK)) { #ifdef COUNT_IPIS (*ipi_hardclock_counts[cpu])++; #endif hardclockintr(); } td->td_intr_frame = oldframe; td->td_intr_nesting_level--; critical_exit(); } /* * send an IPI to a set of cpus. */ void ipi_selected(cpuset_t cpus, u_int ipi) { int cpu; /* * IPI_STOP_HARD maps to a NMI and the trap handler needs a bit * of help in order to understand what is the source. * Set the mask of receiving CPUs for this purpose. */ if (ipi == IPI_STOP_HARD) CPU_OR_ATOMIC(&ipi_nmi_pending, &cpus); while ((cpu = CPU_FFS(&cpus)) != 0) { cpu--; CPU_CLR(cpu, &cpus); CTR3(KTR_SMP, "%s: cpu: %d ipi: %x", __func__, cpu, ipi); ipi_send_cpu(cpu, ipi); } } /* * send an IPI to a specific CPU. */ void ipi_cpu(int cpu, u_int ipi) { /* * IPI_STOP_HARD maps to a NMI and the trap handler needs a bit * of help in order to understand what is the source. * Set the mask of receiving CPUs for this purpose. */ if (ipi == IPI_STOP_HARD) CPU_SET_ATOMIC(cpu, &ipi_nmi_pending); CTR3(KTR_SMP, "%s: cpu: %d ipi: %x", __func__, cpu, ipi); ipi_send_cpu(cpu, ipi); } /* * send an IPI to all CPUs EXCEPT myself */ void ipi_all_but_self(u_int ipi) { cpuset_t other_cpus; other_cpus = all_cpus; CPU_CLR(PCPU_GET(cpuid), &other_cpus); if (IPI_IS_BITMAPED(ipi)) { ipi_selected(other_cpus, ipi); return; } /* * IPI_STOP_HARD maps to a NMI and the trap handler needs a bit * of help in order to understand what is the source. * Set the mask of receiving CPUs for this purpose. */ if (ipi == IPI_STOP_HARD) CPU_OR_ATOMIC(&ipi_nmi_pending, &other_cpus); CTR2(KTR_SMP, "%s: ipi: %x", __func__, ipi); lapic_ipi_vectored(ipi, APIC_IPI_DEST_OTHERS); } int ipi_nmi_handler() { u_int cpuid; /* * As long as there is not a simple way to know about a NMI's * source, if the bitmask for the current CPU is present in * the global pending bitword an IPI_STOP_HARD has been issued * and should be handled. */ cpuid = PCPU_GET(cpuid); if (!CPU_ISSET(cpuid, &ipi_nmi_pending)) return (1); CPU_CLR_ATOMIC(cpuid, &ipi_nmi_pending); cpustop_handler(); return (0); } /* * Handle an IPI_STOP by saving our current context and spinning until we * are resumed. */ void cpustop_handler(void) { u_int cpu; cpu = PCPU_GET(cpuid); savectx(&stoppcbs[cpu]); /* Indicate that we are stopped */ CPU_SET_ATOMIC(cpu, &stopped_cpus); /* Wait for restart */ while (!CPU_ISSET(cpu, &started_cpus)) ia32_pause(); CPU_CLR_ATOMIC(cpu, &started_cpus); CPU_CLR_ATOMIC(cpu, &stopped_cpus); if (cpu == 0 && cpustop_restartfunc != NULL) { cpustop_restartfunc(); cpustop_restartfunc = NULL; } } /* * Handle an IPI_SUSPEND by saving our current context and spinning until we * are resumed. */ void cpususpend_handler(void) { u_int cpu; mtx_assert(&smp_ipi_mtx, MA_NOTOWNED); cpu = PCPU_GET(cpuid); if (savectx(&susppcbs[cpu]->sp_pcb)) { npxsuspend(susppcbs[cpu]->sp_fpususpend); wbinvd(); CPU_SET_ATOMIC(cpu, &suspended_cpus); } else { npxresume(susppcbs[cpu]->sp_fpususpend); pmap_init_pat(); initializecpu(); PCPU_SET(switchtime, 0); PCPU_SET(switchticks, ticks); /* Indicate that we are resumed */ CPU_CLR_ATOMIC(cpu, &suspended_cpus); } /* Wait for resume */ while (!CPU_ISSET(cpu, &started_cpus)) ia32_pause(); if (cpu_ops.cpu_resume) cpu_ops.cpu_resume(); /* Resume MCA and local APIC */ lapic_xapic_mode(); mca_resume(); lapic_setup(0); /* Indicate that we are resumed */ CPU_CLR_ATOMIC(cpu, &suspended_cpus); CPU_CLR_ATOMIC(cpu, &started_cpus); } /* * Handlers for TLB related IPIs */ void invltlb_handler(void) { uint64_t cr3; #ifdef COUNT_XINVLTLB_HITS xhits_gbl[PCPU_GET(cpuid)]++; #endif /* COUNT_XINVLTLB_HITS */ #ifdef COUNT_IPIS (*ipi_invltlb_counts[PCPU_GET(cpuid)])++; #endif /* COUNT_IPIS */ cr3 = rcr3(); load_cr3(cr3); atomic_add_int(&smp_tlb_wait, 1); } void invlpg_handler(void) { #ifdef COUNT_XINVLTLB_HITS xhits_pg[PCPU_GET(cpuid)]++; #endif /* COUNT_XINVLTLB_HITS */ #ifdef COUNT_IPIS (*ipi_invlpg_counts[PCPU_GET(cpuid)])++; #endif /* COUNT_IPIS */ invlpg(smp_tlb_addr1); atomic_add_int(&smp_tlb_wait, 1); } void invlrng_handler(void) { vm_offset_t addr; #ifdef COUNT_XINVLTLB_HITS xhits_rng[PCPU_GET(cpuid)]++; #endif /* COUNT_XINVLTLB_HITS */ #ifdef COUNT_IPIS (*ipi_invlrng_counts[PCPU_GET(cpuid)])++; #endif /* COUNT_IPIS */ addr = smp_tlb_addr1; do { invlpg(addr); addr += PAGE_SIZE; } while (addr < smp_tlb_addr2); atomic_add_int(&smp_tlb_wait, 1); } void invlcache_handler(void) { #ifdef COUNT_IPIS (*ipi_invlcache_counts[PCPU_GET(cpuid)])++; #endif /* COUNT_IPIS */ wbinvd(); atomic_add_int(&smp_tlb_wait, 1); } /* * This is called once the rest of the system is up and running and we're * ready to let the AP's out of the pen. */ static void release_aps(void *dummy __unused) { if (mp_ncpus == 1) return; atomic_store_rel_int(&aps_ready, 1); while (smp_started == 0) ia32_pause(); } SYSINIT(start_aps, SI_SUB_SMP, SI_ORDER_FIRST, release_aps, NULL); #ifdef COUNT_IPIS /* * Setup interrupt counters for IPI handlers. */ static void mp_ipi_intrcnt(void *dummy) { char buf[64]; int i; CPU_FOREACH(i) { snprintf(buf, sizeof(buf), "cpu%d:invltlb", i); intrcnt_add(buf, &ipi_invltlb_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:invlrng", i); intrcnt_add(buf, &ipi_invlrng_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:invlpg", i); intrcnt_add(buf, &ipi_invlpg_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:invlcache", i); intrcnt_add(buf, &ipi_invlcache_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:preempt", i); intrcnt_add(buf, &ipi_preempt_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:ast", i); intrcnt_add(buf, &ipi_ast_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:rendezvous", i); intrcnt_add(buf, &ipi_rendezvous_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:lazypmap", i); intrcnt_add(buf, &ipi_lazypmap_counts[i]); snprintf(buf, sizeof(buf), "cpu%d:hardclock", i); intrcnt_add(buf, &ipi_hardclock_counts[i]); } } SYSINIT(mp_ipi_intrcnt, SI_SUB_INTR, SI_ORDER_MIDDLE, mp_ipi_intrcnt, NULL); #endif Index: projects/cxl_iscsi/sys/mips/atheros/qca955x_chip.c =================================================================== --- projects/cxl_iscsi/sys/mips/atheros/qca955x_chip.c (revision 279716) +++ projects/cxl_iscsi/sys/mips/atheros/qca955x_chip.c (revision 279717) @@ -1,401 +1,402 @@ /*- * Copyright (c) 2015 Adrian Chadd * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include "opt_ddb.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include //#include #include #include #include #include #include static void qca955x_chip_detect_mem_size(void) { } static void qca955x_chip_detect_sys_frequency(void) { unsigned long ref_rate; unsigned long cpu_rate; unsigned long ddr_rate; unsigned long ahb_rate; uint32_t pll, out_div, ref_div, nint, frac, clk_ctrl, postdiv; uint32_t cpu_pll, ddr_pll; uint32_t bootstrap; bootstrap = ATH_READ_REG(QCA955X_RESET_REG_BOOTSTRAP); if (bootstrap & QCA955X_BOOTSTRAP_REF_CLK_40) ref_rate = 40 * 1000 * 1000; else ref_rate = 25 * 1000 * 1000; pll = ATH_READ_REG(QCA955X_PLL_CPU_CONFIG_REG); out_div = (pll >> QCA955X_PLL_CPU_CONFIG_OUTDIV_SHIFT) & QCA955X_PLL_CPU_CONFIG_OUTDIV_MASK; ref_div = (pll >> QCA955X_PLL_CPU_CONFIG_REFDIV_SHIFT) & QCA955X_PLL_CPU_CONFIG_REFDIV_MASK; nint = (pll >> QCA955X_PLL_CPU_CONFIG_NINT_SHIFT) & QCA955X_PLL_CPU_CONFIG_NINT_MASK; frac = (pll >> QCA955X_PLL_CPU_CONFIG_NFRAC_SHIFT) & QCA955X_PLL_CPU_CONFIG_NFRAC_MASK; cpu_pll = nint * ref_rate / ref_div; cpu_pll += frac * ref_rate / (ref_div * (1 << 6)); cpu_pll /= (1 << out_div); pll = ATH_READ_REG(QCA955X_PLL_DDR_CONFIG_REG); out_div = (pll >> QCA955X_PLL_DDR_CONFIG_OUTDIV_SHIFT) & QCA955X_PLL_DDR_CONFIG_OUTDIV_MASK; ref_div = (pll >> QCA955X_PLL_DDR_CONFIG_REFDIV_SHIFT) & QCA955X_PLL_DDR_CONFIG_REFDIV_MASK; nint = (pll >> QCA955X_PLL_DDR_CONFIG_NINT_SHIFT) & QCA955X_PLL_DDR_CONFIG_NINT_MASK; frac = (pll >> QCA955X_PLL_DDR_CONFIG_NFRAC_SHIFT) & QCA955X_PLL_DDR_CONFIG_NFRAC_MASK; ddr_pll = nint * ref_rate / ref_div; ddr_pll += frac * ref_rate / (ref_div * (1 << 10)); ddr_pll /= (1 << out_div); clk_ctrl = ATH_READ_REG(QCA955X_PLL_CLK_CTRL_REG); postdiv = (clk_ctrl >> QCA955X_PLL_CLK_CTRL_CPU_POST_DIV_SHIFT) & QCA955X_PLL_CLK_CTRL_CPU_POST_DIV_MASK; if (clk_ctrl & QCA955X_PLL_CLK_CTRL_CPU_PLL_BYPASS) cpu_rate = ref_rate; else if (clk_ctrl & QCA955X_PLL_CLK_CTRL_CPUCLK_FROM_CPUPLL) cpu_rate = ddr_pll / (postdiv + 1); else cpu_rate = cpu_pll / (postdiv + 1); postdiv = (clk_ctrl >> QCA955X_PLL_CLK_CTRL_DDR_POST_DIV_SHIFT) & QCA955X_PLL_CLK_CTRL_DDR_POST_DIV_MASK; if (clk_ctrl & QCA955X_PLL_CLK_CTRL_DDR_PLL_BYPASS) ddr_rate = ref_rate; else if (clk_ctrl & QCA955X_PLL_CLK_CTRL_DDRCLK_FROM_DDRPLL) ddr_rate = cpu_pll / (postdiv + 1); else ddr_rate = ddr_pll / (postdiv + 1); postdiv = (clk_ctrl >> QCA955X_PLL_CLK_CTRL_AHB_POST_DIV_SHIFT) & QCA955X_PLL_CLK_CTRL_AHB_POST_DIV_MASK; if (clk_ctrl & QCA955X_PLL_CLK_CTRL_AHB_PLL_BYPASS) ahb_rate = ref_rate; else if (clk_ctrl & QCA955X_PLL_CLK_CTRL_AHBCLK_FROM_DDRPLL) ahb_rate = ddr_pll / (postdiv + 1); else ahb_rate = cpu_pll / (postdiv + 1); u_ar71xx_ddr_freq = ddr_rate; u_ar71xx_cpu_freq = cpu_rate; u_ar71xx_ahb_freq = ahb_rate; u_ar71xx_wdt_freq = ref_rate; u_ar71xx_uart_freq = ref_rate; u_ar71xx_mdio_freq = ref_rate; u_ar71xx_refclk = ref_rate; } static void qca955x_chip_device_stop(uint32_t mask) { uint32_t reg; reg = ATH_READ_REG(QCA955X_RESET_REG_RESET_MODULE); ATH_WRITE_REG(QCA955X_RESET_REG_RESET_MODULE, reg | mask); } static void qca955x_chip_device_start(uint32_t mask) { uint32_t reg; reg = ATH_READ_REG(QCA955X_RESET_REG_RESET_MODULE); ATH_WRITE_REG(QCA955X_RESET_REG_RESET_MODULE, reg & ~mask); } static int qca955x_chip_device_stopped(uint32_t mask) { uint32_t reg; reg = ATH_READ_REG(QCA955X_RESET_REG_RESET_MODULE); return ((reg & mask) == mask); } static void qca955x_chip_set_mii_speed(uint32_t unit, uint32_t speed) { /* XXX TODO */ return; } static void qca955x_chip_set_pll_ge(int unit, int speed, uint32_t pll) { switch (unit) { case 0: ATH_WRITE_REG(QCA955X_PLL_ETH_XMII_CONTROL_REG, pll); break; case 1: ATH_WRITE_REG(QCA955X_PLL_ETH_SGMII_CONTROL_REG, pll); break; default: printf("%s: invalid PLL set for arge unit: %d\n", __func__, unit); return; } } static void qca955x_chip_ddr_flush_ge(int unit) { -#if 0 + switch (unit) { case 0: - ar71xx_ddr_flush(AR934X_DDR_REG_FLUSH_GE0); + ar71xx_ddr_flush(QCA955X_DDR_REG_FLUSH_GE0); break; case 1: - ar71xx_ddr_flush(AR934X_DDR_REG_FLUSH_GE1); + ar71xx_ddr_flush(QCA955X_DDR_REG_FLUSH_GE1); break; default: printf("%s: invalid DDR flush for arge unit: %d\n", __func__, unit); return; } -#endif } + +/* XXX TODO: USB flush, PCIe flush, wmac flush */ static void qca955x_chip_ddr_flush_ip2(void) { #if 0 ar71xx_ddr_flush(AR934X_DDR_REG_FLUSH_WMAC); #endif } static uint32_t qca955x_chip_get_eth_pll(unsigned int mac, int speed) { uint32_t pll; switch (speed) { case 10: pll = QCA955X_PLL_VAL_10; break; case 100: pll = QCA955X_PLL_VAL_100; break; case 1000: pll = QCA955X_PLL_VAL_1000; break; default: printf("%s%d: invalid speed %d\n", __func__, mac, speed); pll = 0; } return (pll); } static void qca955x_chip_reset_ethernet_switch(void) { #if 0 ar71xx_device_stop(AR934X_RESET_ETH_SWITCH); DELAY(100); ar71xx_device_start(AR934X_RESET_ETH_SWITCH); DELAY(100); #endif } static void qca955x_configure_gmac(uint32_t gmac_cfg) { uint32_t reg; reg = ATH_READ_REG(QCA955X_GMAC_REG_ETH_CFG); printf("%s: ETH_CFG=0x%08x\n", __func__, reg); reg &= ~(QCA955X_ETH_CFG_RGMII_EN | QCA955X_ETH_CFG_GE0_SGMII); reg |= gmac_cfg; ATH_WRITE_REG(QCA955X_GMAC_REG_ETH_CFG, reg); } static void qca955x_chip_init_usb_peripheral(void) { } static void qca955x_chip_set_mii_if(uint32_t unit, uint32_t mii_mode) { /* * XXX ! * * Nothing to see here; although gmac0 can have its * MII configuration changed, the register values * are slightly different. */ } /* * XXX TODO: fetch default MII divider configuration */ static void qca955x_chip_reset_wmac(void) { /* XXX TODO */ } static void qca955x_chip_init_gmac(void) { long gmac_cfg; if (resource_long_value("qca955x_gmac", 0, "gmac_cfg", &gmac_cfg) == 0) { printf("%s: gmac_cfg=0x%08lx\n", __func__, (long) gmac_cfg); qca955x_configure_gmac((uint32_t) gmac_cfg); } } /* * Reset the NAND Flash Controller. * * + active=1 means "make it active". * + active=0 means "make it inactive". */ static void qca955x_chip_reset_nfc(int active) { #if 0 if (active) { ar71xx_device_start(AR934X_RESET_NANDF); DELAY(100); ar71xx_device_start(AR934X_RESET_ETH_SWITCH_ANALOG); DELAY(250); } else { ar71xx_device_stop(AR934X_RESET_ETH_SWITCH_ANALOG); DELAY(250); ar71xx_device_stop(AR934X_RESET_NANDF); DELAY(100); } #endif } /* * Configure the GPIO output mux setup. * * The AR934x introduced an output mux which allowed * certain functions to be configured on any pin. * Specifically, the switch PHY link LEDs and * WMAC external RX LNA switches are not limited to * a specific GPIO pin. */ static void qca955x_chip_gpio_output_configure(int gpio, uint8_t func) { #if 0 uint32_t reg, s; uint32_t t; if (gpio > QCA955X_GPIO_COUNT) return; reg = AR934X_GPIO_REG_OUT_FUNC0 + 4 * (gpio / 4); s = 8 * (gpio % 4); /* read-modify-write */ t = ATH_READ_REG(AR71XX_GPIO_BASE + reg); t &= ~(0xff << s); t |= func << s; ATH_WRITE_REG(AR71XX_GPIO_BASE + reg, t); /* flush write */ ATH_READ_REG(AR71XX_GPIO_BASE + reg); #endif } struct ar71xx_cpu_def qca955x_chip_def = { &qca955x_chip_detect_mem_size, &qca955x_chip_detect_sys_frequency, &qca955x_chip_device_stop, &qca955x_chip_device_start, &qca955x_chip_device_stopped, &qca955x_chip_set_pll_ge, &qca955x_chip_set_mii_speed, &qca955x_chip_set_mii_if, &qca955x_chip_ddr_flush_ge, &qca955x_chip_get_eth_pll, &qca955x_chip_ddr_flush_ip2, &qca955x_chip_init_usb_peripheral, &qca955x_chip_reset_ethernet_switch, &qca955x_chip_reset_wmac, &qca955x_chip_init_gmac, &qca955x_chip_reset_nfc, &qca955x_chip_gpio_output_configure, }; Index: projects/cxl_iscsi/sys/mips/atheros/qca955xreg.h =================================================================== --- projects/cxl_iscsi/sys/mips/atheros/qca955xreg.h (revision 279716) +++ projects/cxl_iscsi/sys/mips/atheros/qca955xreg.h (revision 279717) @@ -1,201 +1,208 @@ /*- * Copyright (c) 2013 Adrian Chadd * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #ifndef __QCA955XREG_H__ #define __QCA955XREG_H__ #define BIT(x) (1 << (x)) /* Revision ID information */ #define REV_ID_MAJOR_QCA9556 0x0130 #define REV_ID_MAJOR_QCA9558 0x1130 #define QCA955X_REV_ID_REVISION_MASK 0xf /* Big enough to cover APB and SPI, and most peripherals */ /* * it needs to cover SPI because right now the if_ath_ahb * code uses rman to map in the SPI address into memory * to read data instead of us squirreling it away at early * boot-time and using the firmware interface. * * if_ath_ahb.c should use the same firmware interface * that if_ath_pci.c uses. */ #define QCA955X_APB_BASE 0x18000000 #define QCA955X_APB_SIZE 0x08000000 #define QCA955X_PCI_MEM_BASE0 0x10000000 #define QCA955X_PCI_MEM_BASE1 0x12000000 #define QCA955X_PCI_MEM_SIZE 0x02000000 #define QCA955X_PCI_CFG_BASE0 0x14000000 #define QCA955X_PCI_CFG_BASE1 0x16000000 #define QCA955X_PCI_CFG_SIZE 0x1000 #define QCA955X_PCI_CRP_BASE0 (AR71XX_APB_BASE + 0x000c0000) #define QCA955X_PCI_CRP_BASE1 (AR71XX_APB_BASE + 0x00250000) #define QCA955X_PCI_CRP_SIZE 0x1000 #define QCA955X_PCI_CTRL_BASE0 (AR71XX_APB_BASE + 0x000f0000) #define QCA955X_PCI_CTRL_BASE1 (AR71XX_APB_BASE + 0x00280000) #define QCA955X_PCI_CTRL_SIZE 0x100 #define QCA955X_WMAC_BASE (AR71XX_APB_BASE + 0x00100000) #define QCA955X_WMAC_SIZE 0x20000 #define QCA955X_EHCI0_BASE 0x1b000000 #define QCA955X_EHCI1_BASE 0x1b400000 #define QCA955X_EHCI_SIZE 0x1000 /* PLL block */ #define QCA955X_PLL_CPU_CONFIG_REG (AR71XX_PLL_CPU_BASE + 0x00) #define QCA955X_PLL_DDR_CONFIG_REG (AR71XX_PLL_CPU_BASE + 0x04) #define QCA955X_PLL_CLK_CTRL_REG (AR71XX_PLL_CPU_BASE + 0x08) #define QCA955X_PLL_CPU_CONFIG_NFRAC_SHIFT 0 #define QCA955X_PLL_CPU_CONFIG_NFRAC_MASK 0x3f #define QCA955X_PLL_CPU_CONFIG_NINT_SHIFT 6 #define QCA955X_PLL_CPU_CONFIG_NINT_MASK 0x3f #define QCA955X_PLL_CPU_CONFIG_REFDIV_SHIFT 12 #define QCA955X_PLL_CPU_CONFIG_REFDIV_MASK 0x1f #define QCA955X_PLL_CPU_CONFIG_OUTDIV_SHIFT 19 #define QCA955X_PLL_CPU_CONFIG_OUTDIV_MASK 0x3 #define QCA955X_PLL_DDR_CONFIG_NFRAC_SHIFT 0 #define QCA955X_PLL_DDR_CONFIG_NFRAC_MASK 0x3ff #define QCA955X_PLL_DDR_CONFIG_NINT_SHIFT 10 #define QCA955X_PLL_DDR_CONFIG_NINT_MASK 0x3f #define QCA955X_PLL_DDR_CONFIG_REFDIV_SHIFT 16 #define QCA955X_PLL_DDR_CONFIG_REFDIV_MASK 0x1f #define QCA955X_PLL_DDR_CONFIG_OUTDIV_SHIFT 23 #define QCA955X_PLL_DDR_CONFIG_OUTDIV_MASK 0x7 #define QCA955X_PLL_CLK_CTRL_CPU_PLL_BYPASS BIT(2) #define QCA955X_PLL_CLK_CTRL_DDR_PLL_BYPASS BIT(3) #define QCA955X_PLL_CLK_CTRL_AHB_PLL_BYPASS BIT(4) #define QCA955X_PLL_CLK_CTRL_CPU_POST_DIV_SHIFT 5 #define QCA955X_PLL_CLK_CTRL_CPU_POST_DIV_MASK 0x1f #define QCA955X_PLL_CLK_CTRL_DDR_POST_DIV_SHIFT 10 #define QCA955X_PLL_CLK_CTRL_DDR_POST_DIV_MASK 0x1f #define QCA955X_PLL_CLK_CTRL_AHB_POST_DIV_SHIFT 15 #define QCA955X_PLL_CLK_CTRL_AHB_POST_DIV_MASK 0x1f #define QCA955X_PLL_CLK_CTRL_CPUCLK_FROM_CPUPLL BIT(20) #define QCA955X_PLL_CLK_CTRL_DDRCLK_FROM_DDRPLL BIT(21) #define QCA955X_PLL_CLK_CTRL_AHBCLK_FROM_DDRPLL BIT(24) #define QCA955X_PLL_ETH_XMII_CONTROL_REG (AR71XX_PLL_CPU_BASE + 0x28) #define QCA955X_PLL_ETH_SGMII_CONTROL_REG (AR71XX_PLL_CPU_BASE + 0x48) /* Reset block */ #define QCA955X_RESET_REG_RESET_MODULE (AR71XX_RST_BLOCK_BASE + 0x1c) #define QCA955X_RESET_REG_BOOTSTRAP (AR71XX_RST_BLOCK_BASE + 0xb0) #define QCA955X_RESET_REG_EXT_INT_STATUS (AR71XX_RST_BLOCK_BASE + 0xac) #define QCA955X_BOOTSTRAP_REF_CLK_40 BIT(4) #define QCA955X_EXT_INT_WMAC_MISC BIT(0) #define QCA955X_EXT_INT_WMAC_TX BIT(1) #define QCA955X_EXT_INT_WMAC_RXLP BIT(2) #define QCA955X_EXT_INT_WMAC_RXHP BIT(3) #define QCA955X_EXT_INT_PCIE_RC1 BIT(4) #define QCA955X_EXT_INT_PCIE_RC1_INT0 BIT(5) #define QCA955X_EXT_INT_PCIE_RC1_INT1 BIT(6) #define QCA955X_EXT_INT_PCIE_RC1_INT2 BIT(7) #define QCA955X_EXT_INT_PCIE_RC1_INT3 BIT(8) #define QCA955X_EXT_INT_PCIE_RC2 BIT(12) #define QCA955X_EXT_INT_PCIE_RC2_INT0 BIT(13) #define QCA955X_EXT_INT_PCIE_RC2_INT1 BIT(14) #define QCA955X_EXT_INT_PCIE_RC2_INT2 BIT(15) #define QCA955X_EXT_INT_PCIE_RC2_INT3 BIT(16) #define QCA955X_EXT_INT_USB1 BIT(24) #define QCA955X_EXT_INT_USB2 BIT(28) #define QCA955X_EXT_INT_WMAC_ALL \ (QCA955X_EXT_INT_WMAC_MISC | QCA955X_EXT_INT_WMAC_TX | \ QCA955X_EXT_INT_WMAC_RXLP | QCA955X_EXT_INT_WMAC_RXHP) #define QCA955X_EXT_INT_PCIE_RC1_ALL \ (QCA955X_EXT_INT_PCIE_RC1 | QCA955X_EXT_INT_PCIE_RC1_INT0 | \ QCA955X_EXT_INT_PCIE_RC1_INT1 | QCA955X_EXT_INT_PCIE_RC1_INT2 | \ QCA955X_EXT_INT_PCIE_RC1_INT3) #define QCA955X_EXT_INT_PCIE_RC2_ALL \ (QCA955X_EXT_INT_PCIE_RC2 | QCA955X_EXT_INT_PCIE_RC2_INT0 | \ QCA955X_EXT_INT_PCIE_RC2_INT1 | QCA955X_EXT_INT_PCIE_RC2_INT2 | \ QCA955X_EXT_INT_PCIE_RC2_INT3) #define QCA955X_RESET_HOST BIT(31) #define QCA955X_RESET_SLIC BIT(30) #define QCA955X_RESET_HDMA BIT(29) #define QCA955X_RESET_EXTERNAL BIT(28) #define QCA955X_RESET_RTC BIT(27) #define QCA955X_RESET_PCIE_EP_INT BIT(26) #define QCA955X_RESET_CHKSUM_ACC BIT(25) #define QCA955X_RESET_FULL_CHIP BIT(24) #define QCA955X_RESET_GE1_MDIO BIT(23) #define QCA955X_RESET_GE0_MDIO BIT(22) #define QCA955X_RESET_CPU_NMI BIT(21) #define QCA955X_RESET_CPU_COLD BIT(20) #define QCA955X_RESET_HOST_RESET_INT BIT(19) #define QCA955X_RESET_PCIE_EP BIT(18) #define QCA955X_RESET_UART1 BIT(17) #define QCA955X_RESET_DDR BIT(16) #define QCA955X_RESET_USB_PHY_PLL_PWD_EXT BIT(15) #define QCA955X_RESET_NANDF BIT(14) #define QCA955X_RESET_GE1_MAC BIT(13) #define QCA955X_RESET_SGMII_ANALOG BIT(12) #define QCA955X_RESET_USB_PHY_ANALOG BIT(11) #define QCA955X_RESET_HOST_DMA_INT BIT(10) #define QCA955X_RESET_GE0_MAC BIT(9) #define QCA955X_RESET_SGMII BIT(8) #define QCA955X_RESET_PCIE_PHY BIT(7) #define QCA955X_RESET_PCIE BIT(6) #define QCA955X_RESET_USB_HOST BIT(5) #define QCA955X_RESET_USB_PHY BIT(4) #define QCA955X_RESET_USBSUS_OVERRIDE BIT(3) #define QCA955X_RESET_LUT BIT(2) #define QCA955X_RESET_MBOX BIT(1) #define QCA955X_RESET_I2S BIT(0) /* GPIO block */ #define QCA955X_GPIO_COUNT 24 #define QCA955X_GMAC_BASE (AR71XX_APB_BASE + 0x00070000) #define QCA955X_GMAC_SIZE 0x40 #define QCA955X_NFC_BASE 0x1b800200 #define QCA955X_NFC_SIZE 0xb8 /* GMAC Interface */ #define QCA955X_GMAC_REG_ETH_CFG (QCA955X_GMAC_BASE + 0x00) #define QCA955X_ETH_CFG_RGMII_EN BIT(0) #define QCA955X_ETH_CFG_GE0_SGMII BIT(6) /* XXX Same as AR934x values */ #define QCA955X_PLL_VAL_1000 0x16000000 #define QCA955X_PLL_VAL_100 0x00000101 #define QCA955X_PLL_VAL_10 0x00001616 +/* DDR block */ +#define QCA955X_DDR_REG_FLUSH_GE0 (AR71XX_APB_BASE + 0x9c) +#define QCA955X_DDR_REG_FLUSH_GE1 (AR71XX_APB_BASE + 0xa0) +#define QCA955X_DDR_REG_FLUSH_USB (AR71XX_APB_BASE + 0xa4) +#define QCA955X_DDR_REG_FLUSH_PCIE (AR71XX_APB_BASE + 0xa8) +#define QCA955X_DDR_REG_FLUSH_WMAC (AR71XX_APB_BASE + 0xac) + #endif /* __QCA955XREG_H__ */ Index: projects/cxl_iscsi/sys/mips/conf/AP135.hints =================================================================== --- projects/cxl_iscsi/sys/mips/conf/AP135.hints (revision 279716) +++ projects/cxl_iscsi/sys/mips/conf/AP135.hints (revision 279717) @@ -1,166 +1,174 @@ # This is a placeholder until the hardware support is complete. # I'm assuming this is an AP135-020. The AP136-010 in openwrt has # the ethernet ports wired up to the switch in the reverse way. # $FreeBSD$ # QCA955X_ETH_CFG_RGMII_EN (1 << 0) hint.qca955x_gmac.0.gmac_cfg=0x1 # mdiobus0 on arge0 hint.argemdio.0.at="nexus0" hint.argemdio.0.maddr=0x19000000 hint.argemdio.0.msize=0x1000 hint.argemdio.0.order=0 +# mdiobus1 on arge1 - required to bring up arge1? +hint.argemdio.1.at="nexus0" +hint.argemdio.1.maddr=0x1a000000 +hint.argemdio.1.msize=0x1000 +hint.argemdio.1.order=0 + # AR8327 - connected via mdiobus0 on arge0 hint.arswitch.0.at="mdio0" hint.arswitch.0.is_7240=0 # definitely not the internal switch! hint.arswitch.0.is_9340=0 # not the internal switch! hint.arswitch.0.numphys=5 # all ports are PHYs hint.arswitch.0.phy4cpu=0 hint.arswitch.0.is_rgmii=0 # not needed hint.arswitch.0.is_gmii=0 # not needed # This is where it gets a bit odd. port 0 and port 6 are CPU ports. # The current code only supports one CPU port. So hm, what should # we do to hook PAD6 up to be RGMII but a PHY, not a MAC? # The other trick - how do we get arge1 (hooked up to GMAC0) to work? # That's currently supposed to be hooked up to CPU port 0. # Other AR8327 configuration parameters # AP136-020 parameters # GMAC0 AR8327 -> GMAC1 (arge1) SoC, SGMII # AR8327_PAD_MAC_SGMII hint.arswitch.0.pad.0.mode=3 #hint.arswitch.0.pad.0.rxclk_delay_sel=0 hint.arswitch.0.pad.0.sgmii_delay_en=1 # GMAC6 AR8327 -> GMAC0 (arge0) SoC, RGMII # AR8327_PAD_MAC_RGMII # XXX I think this hooks it up to the internal MAC6 hint.arswitch.0.pad.6.mode=6 hint.arswitch.0.pad.6.txclk_delay_en=1 hint.arswitch.0.pad.6.rxclk_delay_en=1 # AR8327_CLK_DELAY_SEL1 hint.arswitch.0.pad.6.txclk_delay_sel=1 # AR8327_CLK_DELAY_SEL2 hint.arswitch.0.pad.6.rxclk_delay_sel=2 # XXX there's no LED management just yet! hint.arswitch.0.led.ctrl0=0x00000000 hint.arswitch.0.led.ctrl1=0xc737c737 hint.arswitch.0.led.ctrl2=0x00000000 hint.arswitch.0.led.ctrl3=0x00c30c00 hint.arswitch.0.led.open_drain=1 # force_link=1 is required for the rest of the parameters # to be configured. hint.arswitch.0.port.0.force_link=1 hint.arswitch.0.port.0.speed=1000 hint.arswitch.0.port.0.duplex=1 hint.arswitch.0.port.0.txpause=1 hint.arswitch.0.port.0.rxpause=1 # force_link=1 is required for the rest of the parameters # to be configured. hint.arswitch.0.port.6.force_link=1 hint.arswitch.0.port.6.speed=1000 hint.arswitch.0.port.6.duplex=1 hint.arswitch.0.port.6.txpause=1 hint.arswitch.0.port.6.rxpause=1 # arge0 - hooked up to AR8327 GMAC6, RGMII -hint.arge.0.phymask=0x10 +hint.arge.0.phymask=0x0 hint.arge.0.miimode=3 # RGMII +hint.arge.0.media=1000 +hint.arge.0.fduplex=1 hint.arge.0.pll_1000=0x56000000 # MAC for arge0 is the first 6 bytes of the ART hint.arge.0.eeprommac=0x1fff0000 # arge1 - lock up to 1000/full hint.arge.1.phymask=0x0 hint.arge.1.media=1000 hint.arge.1.fduplex=1 hint.arge.1.miimode=5 # SGMII hint.arge.1.pll_1000=0x03000101 # MAC for arge1 is the second 6 bytes of the ART hint.arge.1.eeprommac=0x1fff0006 # ath0: Where the ART is - last 64k in the flash hint.ath.0.eepromaddr=0x1fff0000 hint.ath.0.eepromsize=16384 # ath1: it's different; it's a PCIe attached device, so # we instead need to teach the PCIe bridge code about it # (ie, the 'early pci fixup' stuff that programs the PCIe # host registers on the NIC) and then we teach ath where # to find it. # ath1 hint - pcie slot 0 hint.pcib.0.bus.0.0.0.ath_fixup_addr=0x1fff4000 hint.pcib.0.bus.0.0.0.ath_fixup_size=16384 # ath0 - eeprom comes from here hint.ath.1.eeprom_firmware="pcib.0.bus.0.0.0.eeprom_firmware" # flash layout: # # bootargs=console=ttyS0,115200 root=31:02 rootfstype=jffs2 init=/sbin/init mtdparts=ath-nor0:256k(u-boot),64k(u-boot-env),6336k(rootfs),1408k(uImage),8256k(mib0),64k(ART) # 256KiB u-boot hint.map.0.at="flash/spi0" hint.map.0.start=0x00000000 hint.map.0.end=0x00040000 # 256k u-boot hint.map.0.name="u-boot" hint.map.0.readonly=1 # 64KiB u-boot-env hint.map.1.at="flash/spi0" hint.map.1.start=0x00040000 hint.map.1.end=0x00050000 # 64k u-boot-env hint.map.1.name="u-boot-env" hint.map.1.readonly=1 # 6336KiB rootfs hint.map.2.at="flash/spi0" hint.map.2.start=0x00050000 hint.map.2.end=0x00680000 # 6336k rootfs hint.map.2.name="rootfs" hint.map.2.readonly=1 # 1344KiB uImage hint.map.3.at="flash/spi0" hint.map.3.start=0x00680000 hint.map.3.end=0x007d0000 # 1408k uImage, 64k off the end.. hint.map.3.name="uImage" hint.map.3.readonly=1 # 64KiB cfg hint.map.4.at="flash/spi0" hint.map.4.start=0x007d0000 hint.map.4.end=0x007e0000 hint.map.4.name="cfg" hint.map.4.readonly=0 # 8256 KiB mib0 hint.map.5.at="flash/spi0" hint.map.5.start=0x007e0000 hint.map.5.end=0x00ff0000 # 64k mib0 hint.map.5.name="mib0" hint.map.5.readonly=1 # 64KiB ART hint.map.6.at="flash/spi0" hint.map.6.start=0x007f0000 hint.map.6.end=0x01000000 # 64k ART hint.map.6.name="ART" hint.map.6.readonly=1 Index: projects/cxl_iscsi/sys/modules/drm2/Makefile =================================================================== --- projects/cxl_iscsi/sys/modules/drm2/Makefile (revision 279716) +++ projects/cxl_iscsi/sys/modules/drm2/Makefile (revision 279717) @@ -1,30 +1,32 @@ # $FreeBSD$ SYSDIR?=${.CURDIR}/../.. .include "${SYSDIR}/conf/kern.opts.mk" +SUBDIR_PARALLEL= + .if ${MACHINE_CPUARCH} == "amd64" _i915kms= i915kms _radeonkms= radeonkms . if ${MK_SOURCELESS_UCODE} != "no" _radeonkmsfw= radeonkmsfw . endif .endif .if ${MACHINE_CPUARCH} == "i386" . if ${MACHINE} != "pc98" _i915kms= i915kms _radeonkms= radeonkms . if ${MK_SOURCELESS_UCODE} != "no" _radeonkmsfw= radeonkmsfw . endif . endif .endif SUBDIR = \ drm2 \ ${_i915kms} \ ${_radeonkms} \ ${_radeonkmsfw} .include Index: projects/cxl_iscsi/sys/modules/drm2/radeonkmsfw/Makefile =================================================================== --- projects/cxl_iscsi/sys/modules/drm2/radeonkmsfw/Makefile (revision 279716) +++ projects/cxl_iscsi/sys/modules/drm2/radeonkmsfw/Makefile (revision 279717) @@ -1,85 +1,87 @@ # $FreeBSD$ +SUBDIR_PARALLEL= + SUBDIR= \ ARUBA_me \ ARUBA_pfp \ ARUBA_rlc \ BARTS_mc \ BARTS_me \ BARTS_pfp \ BTC_rlc \ CAICOS_mc \ CAICOS_me \ CAICOS_pfp \ CAYMAN_mc \ CAYMAN_me \ CAYMAN_pfp \ CAYMAN_rlc \ CEDAR_me \ CEDAR_pfp \ CEDAR_rlc \ CYPRESS_me \ CYPRESS_pfp \ CYPRESS_rlc \ JUNIPER_me \ JUNIPER_pfp \ JUNIPER_rlc \ PALM_me \ PALM_pfp \ PITCAIRN_ce \ PITCAIRN_mc \ PITCAIRN_me \ PITCAIRN_pfp \ PITCAIRN_rlc \ R100_cp \ R200_cp \ R300_cp \ R420_cp \ R520_cp \ R600_me \ R600_pfp \ R600_rlc \ R700_rlc \ REDWOOD_me \ REDWOOD_pfp \ REDWOOD_rlc \ RS600_cp \ RS690_cp \ RS780_me \ RS780_pfp \ RV610_me \ RV610_pfp \ RV620_me \ RV620_pfp \ RV630_me \ RV630_pfp \ RV635_me \ RV635_pfp \ RV670_me \ RV670_pfp \ RV710_me \ RV710_pfp \ RV730_me \ RV730_pfp \ RV770_me \ RV770_pfp \ SUMO2_me \ SUMO2_pfp \ SUMO_me \ SUMO_pfp \ SUMO_rlc \ TAHITI_ce \ TAHITI_mc \ TAHITI_me \ TAHITI_pfp \ TAHITI_rlc \ TURKS_mc \ TURKS_me \ TURKS_pfp \ VERDE_ce \ VERDE_mc \ VERDE_me \ VERDE_pfp \ VERDE_rlc .include Index: projects/cxl_iscsi/sys/modules/dtb/imx6/Makefile =================================================================== --- projects/cxl_iscsi/sys/modules/dtb/imx6/Makefile (revision 279716) +++ projects/cxl_iscsi/sys/modules/dtb/imx6/Makefile (revision 279717) @@ -1,12 +1,11 @@ # $FreeBSD$ # All the dts files for imx6 systems we support. DTS= \ imx6dl-cubox-i.dts \ imx6q-cubox-i.dts \ imx6dl-hummingboard.dts \ imx6q-hummingboard.dts \ - imx6q-wandboard.dts \ imx6dl-wandboard.dts \ - imx6s-wandboard.dts + imx6q-wandboard.dts .include Index: projects/cxl_iscsi/sys/modules/dtrace/dtraceall/dtraceall.c =================================================================== --- projects/cxl_iscsi/sys/modules/dtrace/dtraceall/dtraceall.c (revision 279716) +++ projects/cxl_iscsi/sys/modules/dtrace/dtraceall/dtraceall.c (revision 279717) @@ -1,84 +1,84 @@ /* * Copyright (C) 2008 John Birrell * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include #include #include #include #include #include #include #include "opt_compat.h" #include "opt_nfs.h" static int dtraceall_modevent(module_t mod __unused, int type, void *data __unused) { int error = 0; switch (type) { case MOD_LOAD: break; case MOD_UNLOAD: break; case MOD_SHUTDOWN: break; default: error = EOPNOTSUPP; break; } return (error); } DEV_MODULE(dtraceall, dtraceall_modevent, NULL); MODULE_VERSION(dtraceall, 1); /* All the DTrace modules should be dependencies here: */ MODULE_DEPEND(dtraceall, opensolaris, 1, 1, 1); MODULE_DEPEND(dtraceall, dtrace, 1, 1, 1); MODULE_DEPEND(dtraceall, dtmalloc, 1, 1, 1); #if defined(NFSCL) MODULE_DEPEND(dtraceall, dtnfscl, 1, 1, 1); #endif -#if defined(__amd64__) || defined(__i386__) || defined(__powerpc__) +#if defined(__amd64__) || defined(__i386__) || defined(__powerpc__) || defined(__arm__) MODULE_DEPEND(dtraceall, fbt, 1, 1, 1); #endif #if defined(__amd64__) || defined(__i386__) MODULE_DEPEND(dtraceall, fasttrap, 1, 1, 1); #endif MODULE_DEPEND(dtraceall, lockstat, 1, 1, 1); MODULE_DEPEND(dtraceall, sdt, 1, 1, 1); MODULE_DEPEND(dtraceall, systrace, 1, 1, 1); #if defined(COMPAT_FREEBSD32) MODULE_DEPEND(dtraceall, systrace_freebsd32, 1, 1, 1); #endif MODULE_DEPEND(dtraceall, profile, 1, 1, 1); Index: projects/cxl_iscsi/sys/modules/usb/Makefile =================================================================== --- projects/cxl_iscsi/sys/modules/usb/Makefile (revision 279716) +++ projects/cxl_iscsi/sys/modules/usb/Makefile (revision 279717) @@ -1,104 +1,109 @@ # # $FreeBSD$ # # Copyright (c) 2008 Hans Petter Selasky. All rights reserved. # # Redistribution and use in source and binary forms, with or without # modification, are permitted provided that the following conditions # are met: # 1. Redistributions of source code must retain the above copyright # notice, this list of conditions and the following disclaimer. # 2. Redistributions in binary form must reproduce the above copyright # notice, this list of conditions and the following disclaimer in the # documentation and/or other materials provided with the distribution. # # THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND # ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE # IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE # ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE # FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL # DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS # OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) # HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT # LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY # OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF # SUCH DAMAGE. # SYSDIR?=${.CURDIR}/../.. .include "${SYSDIR}/conf/kern.opts.mk" # +# Allow USB modules to be built in parallel +# +SUBDIR_PARALLEL= + +# # Check for common USB debug flags to pass when building the USB # modules in this directory: # .if defined(USB_DEBUG) MAKE+=" DEBUG_FLAGS+=-DUSB_DEBUG" .endif .if defined(USB_DEBUG) && defined(USB_REQ_DEBUG) MAKE+=" DEBUG_FLAGS+=-DUSB_REQ_DEBUG" .endif # Modules that include binary-only blobs of microcode should be selectable by # MK_SOURCELESS_UCODE option (see below). SUBDIR = usb SUBDIR += ${_dwc_otg} ehci ${_musb} ohci uhci xhci ${_uss820dci} ${_at91dci} \ ${_atmegadci} ${_avr32dci} ${_rsu} ${_rsufw} ${_saf1761otg} SUBDIR += ${_rum} ${_run} ${_runfw} ${_uath} upgt usie ural ${_zyd} ${_urtw} SUBDIR += ${_urtwn} ${_urtwnfw} SUBDIR += atp uhid ukbd ums udbp ufm uep wsp uled SUBDIR += ucom u3g uark ubsa ubser uchcom ucycom ufoma uftdi ugensa uipaq ulpt \ umct umcs umodem umoscom uplcom uslcom uvisor uvscom SUBDIR += uether aue axe axge cdce cue ${_kue} mos rue smsc udav uhso ipheth SUBDIR += urndis SUBDIR += usfs umass urio SUBDIR += quirk template SUBDIR += ${_g_audio} ${_g_keyboard} ${_g_modem} ${_g_mouse} .if ${MK_USB_GADGET_EXAMPLES} == "yes" _g_audio= g_audio _g_keyboard= g_keyboard _g_modem= g_modem _g_mouse= g_mouse .endif .if ${MK_SOURCELESS_UCODE} != "no" _rum= rum _uath= uath _zyd= zyd _kue= kue _urtwn= urtwn _urtwnfw= urtwnfw _run= run _runfw= runfw _rsu= rsu _rsufw= rsufw .endif .if ${MACHINE_CPUARCH} == "amd64" _urtw= urtw .endif .if ${MACHINE_CPUARCH} == "arm" _at91dci= at91dci _atmegadci= atmegadci _dwc_otg= dwc_otg _musb= musb _uss820dci= uss820dci .endif .if ${MACHINE_CPUARCH} == "i386" _urtw= urtw .endif .if ${MACHINE_CPUARCH} == "avr32" _avr32dci= avr32dci .endif .if ${MACHINE_CPUARCH} == "mips" _saf1761otg= saf1761otg .endif .include Index: projects/cxl_iscsi/sys/net/if_media.c =================================================================== --- projects/cxl_iscsi/sys/net/if_media.c (revision 279716) +++ projects/cxl_iscsi/sys/net/if_media.c (revision 279717) @@ -1,564 +1,520 @@ /* $NetBSD: if_media.c,v 1.1 1997/03/17 02:55:15 thorpej Exp $ */ /* $FreeBSD$ */ /*- * Copyright (c) 1997 * Jonathan Stone and Jason R. Thorpe. All rights reserved. * * This software is derived from information provided by Matt Thomas. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by Jonathan Stone * and Jason R. Thorpe for the NetBSD Project. * 4. The names of the authors may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHORS ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ /* * BSD/OS-compatible network interface media selection. * * Where it is safe to do so, this code strays slightly from the BSD/OS * design. Software which uses the API (device drivers, basically) * shouldn't notice any difference. * * Many thanks to Matt Thomas for providing the information necessary * to implement this interface. */ #include #include #include #include #include #include #include #include #include /* * Compile-time options: * IFMEDIA_DEBUG: * turn on implementation-level debug printfs. * Useful for debugging newly-ported drivers. */ static struct ifmedia_entry *ifmedia_match(struct ifmedia *ifm, int flags, int mask); #ifdef IFMEDIA_DEBUG int ifmedia_debug = 0; SYSCTL_INT(_debug, OID_AUTO, ifmedia, CTLFLAG_RW, &ifmedia_debug, 0, "if_media debugging msgs"); static void ifmedia_printword(int); #endif /* * Initialize if_media struct for a specific interface instance. */ void ifmedia_init(ifm, dontcare_mask, change_callback, status_callback) struct ifmedia *ifm; int dontcare_mask; ifm_change_cb_t change_callback; ifm_stat_cb_t status_callback; { LIST_INIT(&ifm->ifm_list); ifm->ifm_cur = NULL; ifm->ifm_media = 0; ifm->ifm_mask = dontcare_mask; /* IF don't-care bits */ ifm->ifm_change = change_callback; ifm->ifm_status = status_callback; } void ifmedia_removeall(ifm) struct ifmedia *ifm; { struct ifmedia_entry *entry; for (entry = LIST_FIRST(&ifm->ifm_list); entry; entry = LIST_FIRST(&ifm->ifm_list)) { LIST_REMOVE(entry, ifm_list); free(entry, M_IFADDR); } } /* * Add a media configuration to the list of supported media * for a specific interface instance. */ void ifmedia_add(ifm, mword, data, aux) struct ifmedia *ifm; int mword; int data; void *aux; { register struct ifmedia_entry *entry; #ifdef IFMEDIA_DEBUG if (ifmedia_debug) { if (ifm == NULL) { printf("ifmedia_add: null ifm\n"); return; } printf("Adding entry for "); ifmedia_printword(mword); } #endif entry = malloc(sizeof(*entry), M_IFADDR, M_NOWAIT); if (entry == NULL) panic("ifmedia_add: can't malloc entry"); entry->ifm_media = mword; entry->ifm_data = data; entry->ifm_aux = aux; LIST_INSERT_HEAD(&ifm->ifm_list, entry, ifm_list); } /* * Add an array of media configurations to the list of * supported media for a specific interface instance. */ void ifmedia_list_add(ifm, lp, count) struct ifmedia *ifm; struct ifmedia_entry *lp; int count; { int i; for (i = 0; i < count; i++) ifmedia_add(ifm, lp[i].ifm_media, lp[i].ifm_data, lp[i].ifm_aux); } /* * Set the default active media. * * Called by device-specific code which is assumed to have already * selected the default media in hardware. We do _not_ call the * media-change callback. */ void ifmedia_set(ifm, target) struct ifmedia *ifm; int target; { struct ifmedia_entry *match; match = ifmedia_match(ifm, target, ifm->ifm_mask); if (match == NULL) { printf("ifmedia_set: no match for 0x%x/0x%x\n", target, ~ifm->ifm_mask); panic("ifmedia_set"); } ifm->ifm_cur = match; #ifdef IFMEDIA_DEBUG if (ifmedia_debug) { printf("ifmedia_set: target "); ifmedia_printword(target); printf("ifmedia_set: setting to "); ifmedia_printword(ifm->ifm_cur->ifm_media); } #endif } /* * Device-independent media ioctl support function. */ int ifmedia_ioctl(ifp, ifr, ifm, cmd) struct ifnet *ifp; struct ifreq *ifr; struct ifmedia *ifm; u_long cmd; { struct ifmedia_entry *match; struct ifmediareq *ifmr = (struct ifmediareq *) ifr; - int error = 0, sticky; + int error = 0; if (ifp == NULL || ifr == NULL || ifm == NULL) return(EINVAL); switch (cmd) { /* * Set the current media. */ case SIOCSIFMEDIA: { struct ifmedia_entry *oldentry; int oldmedia; int newmedia = ifr->ifr_media; match = ifmedia_match(ifm, newmedia, ifm->ifm_mask); if (match == NULL) { #ifdef IFMEDIA_DEBUG if (ifmedia_debug) { printf( "ifmedia_ioctl: no media found for 0x%x\n", newmedia); } #endif return (ENXIO); } /* * If no change, we're done. * XXX Automedia may invole software intervention. * Keep going in case the connected media changed. * Similarly, if best match changed (kernel debugger?). */ if ((IFM_SUBTYPE(newmedia) != IFM_AUTO) && (newmedia == ifm->ifm_media) && (match == ifm->ifm_cur)) return 0; /* * We found a match, now make the driver switch to it. * Make sure to preserve our old media type in case the * driver can't switch. */ #ifdef IFMEDIA_DEBUG if (ifmedia_debug) { printf("ifmedia_ioctl: switching %s to ", ifp->if_xname); ifmedia_printword(match->ifm_media); } #endif oldentry = ifm->ifm_cur; oldmedia = ifm->ifm_media; ifm->ifm_cur = match; ifm->ifm_media = newmedia; error = (*ifm->ifm_change)(ifp); if (error) { ifm->ifm_cur = oldentry; ifm->ifm_media = oldmedia; } break; } /* * Get list of available media and current media on interface. */ case SIOCGIFMEDIA: { struct ifmedia_entry *ep; - int *kptr, count; - int usermax; /* user requested max */ + int i; - kptr = NULL; /* XXX gcc */ + if (ifmr->ifm_count < 0) + return (EINVAL); ifmr->ifm_active = ifmr->ifm_current = ifm->ifm_cur ? ifm->ifm_cur->ifm_media : IFM_NONE; ifmr->ifm_mask = ifm->ifm_mask; ifmr->ifm_status = 0; (*ifm->ifm_status)(ifp, ifmr); - count = 0; - usermax = 0; - /* * If there are more interfaces on the list, count * them. This allows the caller to set ifmr->ifm_count * to 0 on the first call to know how much space to * allocate. */ + i = 0; LIST_FOREACH(ep, &ifm->ifm_list, ifm_list) - usermax++; - - /* - * Don't allow the user to ask for too many - * or a negative number. - */ - if (ifmr->ifm_count > usermax) - ifmr->ifm_count = usermax; - else if (ifmr->ifm_count < 0) - return (EINVAL); - - if (ifmr->ifm_count != 0) { - kptr = (int *)malloc(ifmr->ifm_count * sizeof(int), - M_TEMP, M_NOWAIT); - - if (kptr == NULL) - return (ENOMEM); - /* - * Get the media words from the interface's list. - */ - ep = LIST_FIRST(&ifm->ifm_list); - for (; ep != NULL && count < ifmr->ifm_count; - ep = LIST_NEXT(ep, ifm_list), count++) - kptr[count] = ep->ifm_media; - - if (ep != NULL) - error = E2BIG; /* oops! */ - } else { - count = usermax; - } - - /* - * We do the copyout on E2BIG, because that's - * just our way of telling userland that there - * are more. This is the behavior I've observed - * under BSD/OS 3.0 - */ - sticky = error; - if ((error == 0 || error == E2BIG) && ifmr->ifm_count != 0) { - error = copyout((caddr_t)kptr, - (caddr_t)ifmr->ifm_ulist, - ifmr->ifm_count * sizeof(int)); - } - - if (error == 0) - error = sticky; - - if (ifmr->ifm_count != 0) - free(kptr, M_TEMP); - - ifmr->ifm_count = count; + if (i++ < ifmr->ifm_count) { + error = copyout(&ep->ifm_media, + ifmr->ifm_ulist + i - 1, sizeof(int)); + if (error) + break; + } + if (error == 0 && i > ifmr->ifm_count) + error = ifmr->ifm_count ? E2BIG : 0; + ifmr->ifm_count = i; break; } default: return (EINVAL); } return (error); } /* * Find media entry matching a given ifm word. * */ static struct ifmedia_entry * ifmedia_match(ifm, target, mask) struct ifmedia *ifm; int target; int mask; { struct ifmedia_entry *match, *next; match = NULL; mask = ~mask; LIST_FOREACH(next, &ifm->ifm_list, ifm_list) { if ((next->ifm_media & mask) == (target & mask)) { #if defined(IFMEDIA_DEBUG) || defined(DIAGNOSTIC) if (match) { printf("ifmedia_match: multiple match for " "0x%x/0x%x\n", target, mask); } #endif match = next; } } return match; } /* * Compute the interface `baudrate' from the media, for the interface * metrics (used by routing daemons). */ static const struct ifmedia_baudrate ifmedia_baudrate_descriptions[] = IFM_BAUDRATE_DESCRIPTIONS; uint64_t ifmedia_baudrate(int mword) { int i; for (i = 0; ifmedia_baudrate_descriptions[i].ifmb_word != 0; i++) { if ((mword & (IFM_NMASK|IFM_TMASK)) == ifmedia_baudrate_descriptions[i].ifmb_word) return (ifmedia_baudrate_descriptions[i].ifmb_baudrate); } /* Not known. */ return (0); } #ifdef IFMEDIA_DEBUG struct ifmedia_description ifm_type_descriptions[] = IFM_TYPE_DESCRIPTIONS; struct ifmedia_description ifm_subtype_ethernet_descriptions[] = IFM_SUBTYPE_ETHERNET_DESCRIPTIONS; struct ifmedia_description ifm_subtype_ethernet_option_descriptions[] = IFM_SUBTYPE_ETHERNET_OPTION_DESCRIPTIONS; struct ifmedia_description ifm_subtype_tokenring_descriptions[] = IFM_SUBTYPE_TOKENRING_DESCRIPTIONS; struct ifmedia_description ifm_subtype_tokenring_option_descriptions[] = IFM_SUBTYPE_TOKENRING_OPTION_DESCRIPTIONS; struct ifmedia_description ifm_subtype_fddi_descriptions[] = IFM_SUBTYPE_FDDI_DESCRIPTIONS; struct ifmedia_description ifm_subtype_fddi_option_descriptions[] = IFM_SUBTYPE_FDDI_OPTION_DESCRIPTIONS; struct ifmedia_description ifm_subtype_ieee80211_descriptions[] = IFM_SUBTYPE_IEEE80211_DESCRIPTIONS; struct ifmedia_description ifm_subtype_ieee80211_option_descriptions[] = IFM_SUBTYPE_IEEE80211_OPTION_DESCRIPTIONS; struct ifmedia_description ifm_subtype_ieee80211_mode_descriptions[] = IFM_SUBTYPE_IEEE80211_MODE_DESCRIPTIONS; struct ifmedia_description ifm_subtype_atm_descriptions[] = IFM_SUBTYPE_ATM_DESCRIPTIONS; struct ifmedia_description ifm_subtype_atm_option_descriptions[] = IFM_SUBTYPE_ATM_OPTION_DESCRIPTIONS; struct ifmedia_description ifm_subtype_shared_descriptions[] = IFM_SUBTYPE_SHARED_DESCRIPTIONS; struct ifmedia_description ifm_shared_option_descriptions[] = IFM_SHARED_OPTION_DESCRIPTIONS; struct ifmedia_type_to_subtype { struct ifmedia_description *subtypes; struct ifmedia_description *options; struct ifmedia_description *modes; }; /* must be in the same order as IFM_TYPE_DESCRIPTIONS */ struct ifmedia_type_to_subtype ifmedia_types_to_subtypes[] = { { &ifm_subtype_ethernet_descriptions[0], &ifm_subtype_ethernet_option_descriptions[0], NULL, }, { &ifm_subtype_tokenring_descriptions[0], &ifm_subtype_tokenring_option_descriptions[0], NULL, }, { &ifm_subtype_fddi_descriptions[0], &ifm_subtype_fddi_option_descriptions[0], NULL, }, { &ifm_subtype_ieee80211_descriptions[0], &ifm_subtype_ieee80211_option_descriptions[0], &ifm_subtype_ieee80211_mode_descriptions[0] }, { &ifm_subtype_atm_descriptions[0], &ifm_subtype_atm_option_descriptions[0], NULL, }, }; /* * print a media word. */ static void ifmedia_printword(ifmw) int ifmw; { struct ifmedia_description *desc; struct ifmedia_type_to_subtype *ttos; int seen_option = 0; /* Find the top-level interface type. */ for (desc = ifm_type_descriptions, ttos = ifmedia_types_to_subtypes; desc->ifmt_string != NULL; desc++, ttos++) if (IFM_TYPE(ifmw) == desc->ifmt_word) break; if (desc->ifmt_string == NULL) { printf("\n"); return; } printf(desc->ifmt_string); /* Any mode. */ for (desc = ttos->modes; desc && desc->ifmt_string != NULL; desc++) if (IFM_MODE(ifmw) == desc->ifmt_word) { if (desc->ifmt_string != NULL) printf(" mode %s", desc->ifmt_string); break; } /* * Check for the shared subtype descriptions first, then the * type-specific ones. */ for (desc = ifm_subtype_shared_descriptions; desc->ifmt_string != NULL; desc++) if (IFM_SUBTYPE(ifmw) == desc->ifmt_word) goto got_subtype; for (desc = ttos->subtypes; desc->ifmt_string != NULL; desc++) if (IFM_SUBTYPE(ifmw) == desc->ifmt_word) break; if (desc->ifmt_string == NULL) { printf(" \n"); return; } got_subtype: printf(" %s", desc->ifmt_string); /* * Look for shared options. */ for (desc = ifm_shared_option_descriptions; desc->ifmt_string != NULL; desc++) { if (ifmw & desc->ifmt_word) { if (seen_option == 0) printf(" <"); printf("%s%s", seen_option++ ? "," : "", desc->ifmt_string); } } /* * Look for subtype-specific options. */ for (desc = ttos->options; desc->ifmt_string != NULL; desc++) { if (ifmw & desc->ifmt_word) { if (seen_option == 0) printf(" <"); printf("%s%s", seen_option++ ? "," : "", desc->ifmt_string); } } printf("%s\n", seen_option ? ">" : ""); } #endif /* IFMEDIA_DEBUG */ Index: projects/cxl_iscsi/sys/netinet6/in6.c =================================================================== --- projects/cxl_iscsi/sys/netinet6/in6.c (revision 279716) +++ projects/cxl_iscsi/sys/netinet6/in6.c (revision 279717) @@ -1,2402 +1,2403 @@ /*- * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: in6.c,v 1.259 2002/01/21 11:37:50 keiichi Exp $ */ /*- * Copyright (c) 1982, 1986, 1991, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)in.c 8.2 (Berkeley) 11/15/93 */ #include __FBSDID("$FreeBSD$"); #include "opt_compat.h" #include "opt_inet.h" #include "opt_inet6.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include VNET_DECLARE(int, icmp6_nodeinfo_oldmcprefix); #define V_icmp6_nodeinfo_oldmcprefix VNET(icmp6_nodeinfo_oldmcprefix) /* * Definitions of some costant IP6 addresses. */ const struct in6_addr in6addr_any = IN6ADDR_ANY_INIT; const struct in6_addr in6addr_loopback = IN6ADDR_LOOPBACK_INIT; const struct in6_addr in6addr_nodelocal_allnodes = IN6ADDR_NODELOCAL_ALLNODES_INIT; const struct in6_addr in6addr_linklocal_allnodes = IN6ADDR_LINKLOCAL_ALLNODES_INIT; const struct in6_addr in6addr_linklocal_allrouters = IN6ADDR_LINKLOCAL_ALLROUTERS_INIT; const struct in6_addr in6addr_linklocal_allv2routers = IN6ADDR_LINKLOCAL_ALLV2ROUTERS_INIT; const struct in6_addr in6mask0 = IN6MASK0; const struct in6_addr in6mask32 = IN6MASK32; const struct in6_addr in6mask64 = IN6MASK64; const struct in6_addr in6mask96 = IN6MASK96; const struct in6_addr in6mask128 = IN6MASK128; const struct sockaddr_in6 sa6_any = { sizeof(sa6_any), AF_INET6, 0, 0, IN6ADDR_ANY_INIT, 0 }; static int in6_notify_ifa(struct ifnet *, struct in6_ifaddr *, struct in6_aliasreq *, int); static void in6_unlink_ifa(struct in6_ifaddr *, struct ifnet *); static int in6_validate_ifra(struct ifnet *, struct in6_aliasreq *, struct in6_ifaddr *, int); static struct in6_ifaddr *in6_alloc_ifa(struct ifnet *, struct in6_aliasreq *, int flags); static int in6_update_ifa_internal(struct ifnet *, struct in6_aliasreq *, struct in6_ifaddr *, int, int); static int in6_broadcast_ifa(struct ifnet *, struct in6_aliasreq *, struct in6_ifaddr *, int); #define ifa2ia6(ifa) ((struct in6_ifaddr *)(ifa)) #define ia62ifa(ia6) (&((ia6)->ia_ifa)) void in6_newaddrmsg(struct in6_ifaddr *ia, int cmd) { struct sockaddr_dl gateway; struct sockaddr_in6 mask, addr; struct rtentry rt; /* * initialize for rtmsg generation */ bzero(&gateway, sizeof(gateway)); gateway.sdl_len = sizeof(gateway); gateway.sdl_family = AF_LINK; bzero(&rt, sizeof(rt)); rt.rt_gateway = (struct sockaddr *)&gateway; memcpy(&mask, &ia->ia_prefixmask, sizeof(ia->ia_prefixmask)); memcpy(&addr, &ia->ia_addr, sizeof(ia->ia_addr)); rt_mask(&rt) = (struct sockaddr *)&mask; rt_key(&rt) = (struct sockaddr *)&addr; rt.rt_flags = RTF_HOST | RTF_STATIC; if (cmd == RTM_ADD) rt.rt_flags |= RTF_UP; /* Announce arrival of local address to all FIBs. */ rt_newaddrmsg(cmd, &ia->ia_ifa, 0, &rt); } int in6_mask2len(struct in6_addr *mask, u_char *lim0) { int x = 0, y; u_char *lim = lim0, *p; /* ignore the scope_id part */ if (lim0 == NULL || lim0 - (u_char *)mask > sizeof(*mask)) lim = (u_char *)mask + sizeof(*mask); for (p = (u_char *)mask; p < lim; x++, p++) { if (*p != 0xff) break; } y = 0; if (p < lim) { for (y = 0; y < 8; y++) { if ((*p & (0x80 >> y)) == 0) break; } } /* * when the limit pointer is given, do a stricter check on the * remaining bits. */ if (p < lim) { if (y != 0 && (*p & (0x00ff >> y)) != 0) return (-1); for (p = p + 1; p < lim; p++) if (*p != 0) return (-1); } return x * 8 + y; } #ifdef COMPAT_FREEBSD32 struct in6_ndifreq32 { char ifname[IFNAMSIZ]; uint32_t ifindex; }; #define SIOCGDEFIFACE32_IN6 _IOWR('i', 86, struct in6_ndifreq32) #endif int in6_control(struct socket *so, u_long cmd, caddr_t data, struct ifnet *ifp, struct thread *td) { struct in6_ifreq *ifr = (struct in6_ifreq *)data; struct in6_ifaddr *ia = NULL; struct in6_aliasreq *ifra = (struct in6_aliasreq *)data; struct sockaddr_in6 *sa6; int carp_attached = 0; int error; u_long ocmd = cmd; /* * Compat to make pre-10.x ifconfig(8) operable. */ if (cmd == OSIOCAIFADDR_IN6) cmd = SIOCAIFADDR_IN6; switch (cmd) { case SIOCGETSGCNT_IN6: case SIOCGETMIFCNT_IN6: /* * XXX mrt_ioctl has a 3rd, unused, FIB argument in route.c. * We cannot see how that would be needed, so do not adjust the * KPI blindly; more likely should clean up the IPv4 variant. */ return (mrt6_ioctl ? mrt6_ioctl(cmd, data) : EOPNOTSUPP); } switch (cmd) { case SIOCAADDRCTL_POLICY: case SIOCDADDRCTL_POLICY: if (td != NULL) { error = priv_check(td, PRIV_NETINET_ADDRCTRL6); if (error) return (error); } return (in6_src_ioctl(cmd, data)); } if (ifp == NULL) return (EOPNOTSUPP); switch (cmd) { case SIOCSNDFLUSH_IN6: case SIOCSPFXFLUSH_IN6: case SIOCSRTRFLUSH_IN6: case SIOCSDEFIFACE_IN6: case SIOCSIFINFO_FLAGS: case SIOCSIFINFO_IN6: if (td != NULL) { error = priv_check(td, PRIV_NETINET_ND6); if (error) return (error); } /* FALLTHROUGH */ case OSIOCGIFINFO_IN6: case SIOCGIFINFO_IN6: case SIOCGDRLST_IN6: case SIOCGPRLST_IN6: case SIOCGNBRINFO_IN6: case SIOCGDEFIFACE_IN6: return (nd6_ioctl(cmd, data, ifp)); #ifdef COMPAT_FREEBSD32 case SIOCGDEFIFACE32_IN6: { struct in6_ndifreq ndif; struct in6_ndifreq32 *ndif32; error = nd6_ioctl(SIOCGDEFIFACE_IN6, (caddr_t)&ndif, ifp); if (error) return (error); ndif32 = (struct in6_ndifreq32 *)data; ndif32->ifindex = ndif.ifindex; return (0); } #endif } switch (cmd) { case SIOCSIFPREFIX_IN6: case SIOCDIFPREFIX_IN6: case SIOCAIFPREFIX_IN6: case SIOCCIFPREFIX_IN6: case SIOCSGIFPREFIX_IN6: case SIOCGIFPREFIX_IN6: log(LOG_NOTICE, "prefix ioctls are now invalidated. " "please use ifconfig.\n"); return (EOPNOTSUPP); } switch (cmd) { case SIOCSSCOPE6: if (td != NULL) { error = priv_check(td, PRIV_NETINET_SCOPE6); if (error) return (error); } /* FALLTHROUGH */ case SIOCGSCOPE6: case SIOCGSCOPE6DEF: return (scope6_ioctl(cmd, data, ifp)); } /* * Find address for this interface, if it exists. * * In netinet code, we have checked ifra_addr in SIOCSIF*ADDR operation * only, and used the first interface address as the target of other * operations (without checking ifra_addr). This was because netinet * code/API assumed at most 1 interface address per interface. * Since IPv6 allows a node to assign multiple addresses * on a single interface, we almost always look and check the * presence of ifra_addr, and reject invalid ones here. * It also decreases duplicated code among SIOC*_IN6 operations. */ switch (cmd) { case SIOCAIFADDR_IN6: case SIOCSIFPHYADDR_IN6: sa6 = &ifra->ifra_addr; break; case SIOCSIFADDR_IN6: case SIOCGIFADDR_IN6: case SIOCSIFDSTADDR_IN6: case SIOCSIFNETMASK_IN6: case SIOCGIFDSTADDR_IN6: case SIOCGIFNETMASK_IN6: case SIOCDIFADDR_IN6: case SIOCGIFPSRCADDR_IN6: case SIOCGIFPDSTADDR_IN6: case SIOCGIFAFLAG_IN6: case SIOCSNDFLUSH_IN6: case SIOCSPFXFLUSH_IN6: case SIOCSRTRFLUSH_IN6: case SIOCGIFALIFETIME_IN6: case SIOCSIFALIFETIME_IN6: case SIOCGIFSTAT_IN6: case SIOCGIFSTAT_ICMP6: sa6 = &ifr->ifr_addr; break; case SIOCSIFADDR: case SIOCSIFBRDADDR: case SIOCSIFDSTADDR: case SIOCSIFNETMASK: /* * Although we should pass any non-INET6 ioctl requests * down to driver, we filter some legacy INET requests. * Drivers trust SIOCSIFADDR et al to come from an already * privileged layer, and do not perform any credentials * checks or input validation. */ return (EINVAL); default: sa6 = NULL; break; } if (sa6 && sa6->sin6_family == AF_INET6) { if (sa6->sin6_scope_id != 0) error = sa6_embedscope(sa6, 0); else error = in6_setscope(&sa6->sin6_addr, ifp, NULL); if (error != 0) return (error); if (td != NULL && (error = prison_check_ip6(td->td_ucred, &sa6->sin6_addr)) != 0) return (error); ia = in6ifa_ifpwithaddr(ifp, &sa6->sin6_addr); } else ia = NULL; switch (cmd) { case SIOCSIFADDR_IN6: case SIOCSIFDSTADDR_IN6: case SIOCSIFNETMASK_IN6: /* * Since IPv6 allows a node to assign multiple addresses * on a single interface, SIOCSIFxxx ioctls are deprecated. */ /* we decided to obsolete this command (20000704) */ error = EINVAL; goto out; case SIOCDIFADDR_IN6: /* * for IPv4, we look for existing in_ifaddr here to allow * "ifconfig if0 delete" to remove the first IPv4 address on * the interface. For IPv6, as the spec allows multiple * interface address from the day one, we consider "remove the * first one" semantics to be not preferable. */ if (ia == NULL) { error = EADDRNOTAVAIL; goto out; } /* FALLTHROUGH */ case SIOCAIFADDR_IN6: /* * We always require users to specify a valid IPv6 address for * the corresponding operation. */ if (ifra->ifra_addr.sin6_family != AF_INET6 || ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6)) { error = EAFNOSUPPORT; goto out; } if (td != NULL) { error = priv_check(td, (cmd == SIOCDIFADDR_IN6) ? PRIV_NET_DELIFADDR : PRIV_NET_ADDIFADDR); if (error) goto out; } /* FALLTHROUGH */ case SIOCGIFSTAT_IN6: case SIOCGIFSTAT_ICMP6: if (ifp->if_afdata[AF_INET6] == NULL) { error = EPFNOSUPPORT; goto out; } break; case SIOCGIFADDR_IN6: /* This interface is basically deprecated. use SIOCGIFCONF. */ /* FALLTHROUGH */ case SIOCGIFAFLAG_IN6: case SIOCGIFNETMASK_IN6: case SIOCGIFDSTADDR_IN6: case SIOCGIFALIFETIME_IN6: /* must think again about its semantics */ if (ia == NULL) { error = EADDRNOTAVAIL; goto out; } break; case SIOCSIFALIFETIME_IN6: { struct in6_addrlifetime *lt; if (td != NULL) { error = priv_check(td, PRIV_NETINET_ALIFETIME6); if (error) goto out; } if (ia == NULL) { error = EADDRNOTAVAIL; goto out; } /* sanity for overflow - beware unsigned */ lt = &ifr->ifr_ifru.ifru_lifetime; if (lt->ia6t_vltime != ND6_INFINITE_LIFETIME && lt->ia6t_vltime + time_uptime < time_uptime) { error = EINVAL; goto out; } if (lt->ia6t_pltime != ND6_INFINITE_LIFETIME && lt->ia6t_pltime + time_uptime < time_uptime) { error = EINVAL; goto out; } break; } } switch (cmd) { case SIOCGIFADDR_IN6: ifr->ifr_addr = ia->ia_addr; if ((error = sa6_recoverscope(&ifr->ifr_addr)) != 0) goto out; break; case SIOCGIFDSTADDR_IN6: if ((ifp->if_flags & IFF_POINTOPOINT) == 0) { error = EINVAL; goto out; } /* * XXX: should we check if ifa_dstaddr is NULL and return * an error? */ ifr->ifr_dstaddr = ia->ia_dstaddr; if ((error = sa6_recoverscope(&ifr->ifr_dstaddr)) != 0) goto out; break; case SIOCGIFNETMASK_IN6: ifr->ifr_addr = ia->ia_prefixmask; break; case SIOCGIFAFLAG_IN6: ifr->ifr_ifru.ifru_flags6 = ia->ia6_flags; break; case SIOCGIFSTAT_IN6: COUNTER_ARRAY_COPY(((struct in6_ifextra *) ifp->if_afdata[AF_INET6])->in6_ifstat, &ifr->ifr_ifru.ifru_stat, sizeof(struct in6_ifstat) / sizeof(uint64_t)); break; case SIOCGIFSTAT_ICMP6: COUNTER_ARRAY_COPY(((struct in6_ifextra *) ifp->if_afdata[AF_INET6])->icmp6_ifstat, &ifr->ifr_ifru.ifru_icmp6stat, sizeof(struct icmp6_ifstat) / sizeof(uint64_t)); break; case SIOCGIFALIFETIME_IN6: ifr->ifr_ifru.ifru_lifetime = ia->ia6_lifetime; if (ia->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) { time_t maxexpire; struct in6_addrlifetime *retlt = &ifr->ifr_ifru.ifru_lifetime; /* * XXX: adjust expiration time assuming time_t is * signed. */ maxexpire = (-1) & ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1)); if (ia->ia6_lifetime.ia6t_vltime < maxexpire - ia->ia6_updatetime) { retlt->ia6t_expire = ia->ia6_updatetime + ia->ia6_lifetime.ia6t_vltime; } else retlt->ia6t_expire = maxexpire; } if (ia->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) { time_t maxexpire; struct in6_addrlifetime *retlt = &ifr->ifr_ifru.ifru_lifetime; /* * XXX: adjust expiration time assuming time_t is * signed. */ maxexpire = (-1) & ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1)); if (ia->ia6_lifetime.ia6t_pltime < maxexpire - ia->ia6_updatetime) { retlt->ia6t_preferred = ia->ia6_updatetime + ia->ia6_lifetime.ia6t_pltime; } else retlt->ia6t_preferred = maxexpire; } break; case SIOCSIFALIFETIME_IN6: ia->ia6_lifetime = ifr->ifr_ifru.ifru_lifetime; /* for sanity */ if (ia->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) { ia->ia6_lifetime.ia6t_expire = time_uptime + ia->ia6_lifetime.ia6t_vltime; } else ia->ia6_lifetime.ia6t_expire = 0; if (ia->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) { ia->ia6_lifetime.ia6t_preferred = time_uptime + ia->ia6_lifetime.ia6t_pltime; } else ia->ia6_lifetime.ia6t_preferred = 0; break; case SIOCAIFADDR_IN6: { struct nd_prefixctl pr0; struct nd_prefix *pr; /* * first, make or update the interface address structure, * and link it to the list. */ if ((error = in6_update_ifa(ifp, ifra, ia, 0)) != 0) goto out; if (ia != NULL) ifa_free(&ia->ia_ifa); if ((ia = in6ifa_ifpwithaddr(ifp, &ifra->ifra_addr.sin6_addr)) == NULL) { /* * this can happen when the user specify the 0 valid * lifetime. */ break; } if (cmd == ocmd && ifra->ifra_vhid > 0) { if (carp_attach_p != NULL) error = (*carp_attach_p)(&ia->ia_ifa, ifra->ifra_vhid); else error = EPROTONOSUPPORT; if (error) goto out; else carp_attached = 1; } /* * then, make the prefix on-link on the interface. * XXX: we'd rather create the prefix before the address, but * we need at least one address to install the corresponding * interface route, so we configure the address first. */ /* * convert mask to prefix length (prefixmask has already * been validated in in6_update_ifa(). */ bzero(&pr0, sizeof(pr0)); pr0.ndpr_ifp = ifp; pr0.ndpr_plen = in6_mask2len(&ifra->ifra_prefixmask.sin6_addr, NULL); if (pr0.ndpr_plen == 128) { /* we don't need to install a host route. */ goto aifaddr_out; } pr0.ndpr_prefix = ifra->ifra_addr; /* apply the mask for safety. */ IN6_MASK_ADDR(&pr0.ndpr_prefix.sin6_addr, &ifra->ifra_prefixmask.sin6_addr); /* * XXX: since we don't have an API to set prefix (not address) * lifetimes, we just use the same lifetimes as addresses. * The (temporarily) installed lifetimes can be overridden by * later advertised RAs (when accept_rtadv is non 0), which is * an intended behavior. */ pr0.ndpr_raf_onlink = 1; /* should be configurable? */ pr0.ndpr_raf_auto = ((ifra->ifra_flags & IN6_IFF_AUTOCONF) != 0); pr0.ndpr_vltime = ifra->ifra_lifetime.ia6t_vltime; pr0.ndpr_pltime = ifra->ifra_lifetime.ia6t_pltime; /* add the prefix if not yet. */ if ((pr = nd6_prefix_lookup(&pr0)) == NULL) { /* * nd6_prelist_add will install the corresponding * interface route. */ if ((error = nd6_prelist_add(&pr0, NULL, &pr)) != 0) { if (carp_attached) (*carp_detach_p)(&ia->ia_ifa); goto out; } if (pr == NULL) { if (carp_attached) (*carp_detach_p)(&ia->ia_ifa); log(LOG_ERR, "nd6_prelist_add succeeded but " "no prefix\n"); error = EINVAL; goto out; } } /* relate the address to the prefix */ if (ia->ia6_ndpr == NULL) { ia->ia6_ndpr = pr; pr->ndpr_refcnt++; /* * If this is the first autoconf address from the * prefix, create a temporary address as well * (when required). */ if ((ia->ia6_flags & IN6_IFF_AUTOCONF) && V_ip6_use_tempaddr && pr->ndpr_refcnt == 1) { int e; if ((e = in6_tmpifadd(ia, 1, 0)) != 0) { log(LOG_NOTICE, "in6_control: failed " "to create a temporary address, " "errno=%d\n", e); } } } /* * this might affect the status of autoconfigured addresses, * that is, this address might make other addresses detached. */ pfxlist_onlink_check(); aifaddr_out: if (error != 0 || ia == NULL) break; /* * Try to clear the flag when a new IPv6 address is added * onto an IFDISABLED interface and it succeeds. */ if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) { struct in6_ndireq nd; memset(&nd, 0, sizeof(nd)); nd.ndi.flags = ND_IFINFO(ifp)->flags; nd.ndi.flags &= ~ND6_IFF_IFDISABLED; if (nd6_ioctl(SIOCSIFINFO_FLAGS, (caddr_t)&nd, ifp) < 0) log(LOG_NOTICE, "SIOCAIFADDR_IN6: " "SIOCSIFINFO_FLAGS for -ifdisabled " "failed."); /* * Ignore failure of clearing the flag intentionally. * The failure means address duplication was detected. */ } EVENTHANDLER_INVOKE(ifaddr_event, ifp); break; } case SIOCDIFADDR_IN6: { struct nd_prefix *pr; /* * If the address being deleted is the only one that owns * the corresponding prefix, expire the prefix as well. * XXX: theoretically, we don't have to worry about such * relationship, since we separate the address management * and the prefix management. We do this, however, to provide * as much backward compatibility as possible in terms of * the ioctl operation. * Note that in6_purgeaddr() will decrement ndpr_refcnt. */ pr = ia->ia6_ndpr; in6_purgeaddr(&ia->ia_ifa); if (pr && pr->ndpr_refcnt == 0) prelist_remove(pr); EVENTHANDLER_INVOKE(ifaddr_event, ifp); break; } default: if (ifp->if_ioctl == NULL) { error = EOPNOTSUPP; goto out; } error = (*ifp->if_ioctl)(ifp, cmd, data); goto out; } error = 0; out: if (ia != NULL) ifa_free(&ia->ia_ifa); return (error); } /* * Join necessary multicast groups. Factored out from in6_update_ifa(). * This entire work should only be done once, for the default FIB. */ static int in6_update_ifa_join_mc(struct ifnet *ifp, struct in6_aliasreq *ifra, struct in6_ifaddr *ia, int flags, struct in6_multi **in6m_sol) { char ip6buf[INET6_ADDRSTRLEN]; struct in6_addr mltaddr; struct in6_multi_mship *imm; int delay, error; KASSERT(in6m_sol != NULL, ("%s: in6m_sol is NULL", __func__)); /* Join solicited multicast addr for new host id. */ bzero(&mltaddr, sizeof(struct in6_addr)); mltaddr.s6_addr32[0] = IPV6_ADDR_INT32_MLL; mltaddr.s6_addr32[2] = htonl(1); mltaddr.s6_addr32[3] = ifra->ifra_addr.sin6_addr.s6_addr32[3]; mltaddr.s6_addr8[12] = 0xff; if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0) { /* XXX: should not happen */ log(LOG_ERR, "%s: in6_setscope failed\n", __func__); goto cleanup; } delay = error = 0; if ((flags & IN6_IFAUPDATE_DADDELAY)) { /* * We need a random delay for DAD on the address being * configured. It also means delaying transmission of the * corresponding MLD report to avoid report collision. * [RFC 4861, Section 6.3.7] */ delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz); } imm = in6_joingroup(ifp, &mltaddr, &error, delay); if (imm == NULL) { nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s " "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr), if_name(ifp), error)); goto cleanup; } LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain); *in6m_sol = imm->i6mm_maddr; /* * Join link-local all-nodes address. */ mltaddr = in6addr_linklocal_allnodes; if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0) goto cleanup; /* XXX: should not fail */ imm = in6_joingroup(ifp, &mltaddr, &error, 0); if (imm == NULL) { nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s " "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr), if_name(ifp), error)); goto cleanup; } LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain); /* * Join node information group address. */ delay = 0; if ((flags & IN6_IFAUPDATE_DADDELAY)) { /* * The spec does not say anything about delay for this group, * but the same logic should apply. */ delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz); } if (in6_nigroup(ifp, NULL, -1, &mltaddr) == 0) { /* XXX jinmei */ imm = in6_joingroup(ifp, &mltaddr, &error, delay); if (imm == NULL) nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s " "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr), if_name(ifp), error)); /* XXX not very fatal, go on... */ else LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain); } if (V_icmp6_nodeinfo_oldmcprefix && in6_nigroup_oldmcprefix(ifp, NULL, -1, &mltaddr) == 0) { imm = in6_joingroup(ifp, &mltaddr, &error, delay); if (imm == NULL) nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s " "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr), if_name(ifp), error)); /* XXX not very fatal, go on... */ else LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain); } /* * Join interface-local all-nodes address. * (ff01::1%ifN, and ff01::%ifN/32) */ mltaddr = in6addr_nodelocal_allnodes; if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0) goto cleanup; /* XXX: should not fail */ imm = in6_joingroup(ifp, &mltaddr, &error, 0); if (imm == NULL) { nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s " "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr), if_name(ifp), error)); goto cleanup; } LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain); cleanup: return (error); } /* * Update parameters of an IPv6 interface address. * If necessary, a new entry is created and linked into address chains. * This function is separated from in6_control(). */ int in6_update_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra, struct in6_ifaddr *ia, int flags) { int error, hostIsNew = 0; if ((error = in6_validate_ifra(ifp, ifra, ia, flags)) != 0) return (error); if (ia == NULL) { hostIsNew = 1; if ((ia = in6_alloc_ifa(ifp, ifra, flags)) == NULL) return (ENOBUFS); } error = in6_update_ifa_internal(ifp, ifra, ia, hostIsNew, flags); if (error != 0) { if (hostIsNew != 0) { in6_unlink_ifa(ia, ifp); ifa_free(&ia->ia_ifa); } return (error); } if (hostIsNew) error = in6_broadcast_ifa(ifp, ifra, ia, flags); return (error); } /* * Fill in basic IPv6 address request info. */ void in6_prepare_ifra(struct in6_aliasreq *ifra, const struct in6_addr *addr, const struct in6_addr *mask) { memset(ifra, 0, sizeof(struct in6_aliasreq)); ifra->ifra_addr.sin6_family = AF_INET6; ifra->ifra_addr.sin6_len = sizeof(struct sockaddr_in6); if (addr != NULL) ifra->ifra_addr.sin6_addr = *addr; ifra->ifra_prefixmask.sin6_family = AF_INET6; ifra->ifra_prefixmask.sin6_len = sizeof(struct sockaddr_in6); if (mask != NULL) ifra->ifra_prefixmask.sin6_addr = *mask; } static int in6_validate_ifra(struct ifnet *ifp, struct in6_aliasreq *ifra, struct in6_ifaddr *ia, int flags) { int plen = -1; struct sockaddr_in6 dst6; struct in6_addrlifetime *lt; char ip6buf[INET6_ADDRSTRLEN]; /* Validate parameters */ if (ifp == NULL || ifra == NULL) /* this maybe redundant */ return (EINVAL); /* * The destination address for a p2p link must have a family * of AF_UNSPEC or AF_INET6. */ if ((ifp->if_flags & IFF_POINTOPOINT) != 0 && ifra->ifra_dstaddr.sin6_family != AF_INET6 && ifra->ifra_dstaddr.sin6_family != AF_UNSPEC) return (EAFNOSUPPORT); /* * Validate address */ if (ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6) || ifra->ifra_addr.sin6_family != AF_INET6) return (EINVAL); /* * validate ifra_prefixmask. don't check sin6_family, netmask * does not carry fields other than sin6_len. */ if (ifra->ifra_prefixmask.sin6_len > sizeof(struct sockaddr_in6)) return (EINVAL); /* * Because the IPv6 address architecture is classless, we require * users to specify a (non 0) prefix length (mask) for a new address. * We also require the prefix (when specified) mask is valid, and thus * reject a non-consecutive mask. */ if (ia == NULL && ifra->ifra_prefixmask.sin6_len == 0) return (EINVAL); if (ifra->ifra_prefixmask.sin6_len != 0) { plen = in6_mask2len(&ifra->ifra_prefixmask.sin6_addr, (u_char *)&ifra->ifra_prefixmask + ifra->ifra_prefixmask.sin6_len); if (plen <= 0) return (EINVAL); } else { /* * In this case, ia must not be NULL. We just use its prefix * length. */ plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); } /* * If the destination address on a p2p interface is specified, * and the address is a scoped one, validate/set the scope * zone identifier. */ dst6 = ifra->ifra_dstaddr; if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) != 0 && (dst6.sin6_family == AF_INET6)) { struct in6_addr in6_tmp; u_int32_t zoneid; in6_tmp = dst6.sin6_addr; if (in6_setscope(&in6_tmp, ifp, &zoneid)) return (EINVAL); /* XXX: should be impossible */ if (dst6.sin6_scope_id != 0) { if (dst6.sin6_scope_id != zoneid) return (EINVAL); } else /* user omit to specify the ID. */ dst6.sin6_scope_id = zoneid; /* convert into the internal form */ if (sa6_embedscope(&dst6, 0)) return (EINVAL); /* XXX: should be impossible */ } /* Modify original ifra_dstaddr to reflect changes */ ifra->ifra_dstaddr = dst6; /* * The destination address can be specified only for a p2p or a * loopback interface. If specified, the corresponding prefix length * must be 128. */ if (ifra->ifra_dstaddr.sin6_family == AF_INET6) { if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) == 0) { /* XXX: noisy message */ nd6log((LOG_INFO, "in6_update_ifa: a destination can " "be specified for a p2p or a loopback IF only\n")); return (EINVAL); } if (plen != 128) { nd6log((LOG_INFO, "in6_update_ifa: prefixlen should " "be 128 when dstaddr is specified\n")); return (EINVAL); } } /* lifetime consistency check */ lt = &ifra->ifra_lifetime; if (lt->ia6t_pltime > lt->ia6t_vltime) return (EINVAL); if (lt->ia6t_vltime == 0) { /* * the following log might be noisy, but this is a typical * configuration mistake or a tool's bug. */ nd6log((LOG_INFO, "in6_update_ifa: valid lifetime is 0 for %s\n", ip6_sprintf(ip6buf, &ifra->ifra_addr.sin6_addr))); if (ia == NULL) return (0); /* there's nothing to do */ } /* Check prefix mask */ if (ia != NULL && ifra->ifra_prefixmask.sin6_len != 0) { /* * We prohibit changing the prefix length of an existing * address, because * + such an operation should be rare in IPv6, and * + the operation would confuse prefix management. */ if (ia->ia_prefixmask.sin6_len != 0 && in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL) != plen) { nd6log((LOG_INFO, "in6_validate_ifa: the prefix length " "of an existing %s address should not be changed\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr))); return (EINVAL); } } return (0); } /* * Allocate a new ifaddr and link it into chains. */ static struct in6_ifaddr * in6_alloc_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra, int flags) { struct in6_ifaddr *ia; /* * When in6_alloc_ifa() is called in a process of a received * RA, it is called under an interrupt context. So, we should * call malloc with M_NOWAIT. */ ia = (struct in6_ifaddr *)ifa_alloc(sizeof(*ia), M_NOWAIT); if (ia == NULL) return (NULL); LIST_INIT(&ia->ia6_memberships); /* Initialize the address and masks, and put time stamp */ ia->ia_ifa.ifa_addr = (struct sockaddr *)&ia->ia_addr; ia->ia_addr.sin6_family = AF_INET6; ia->ia_addr.sin6_len = sizeof(ia->ia_addr); /* XXX: Can we assign ,sin6_addr and skip the rest? */ ia->ia_addr = ifra->ifra_addr; ia->ia6_createtime = time_uptime; if ((ifp->if_flags & (IFF_POINTOPOINT | IFF_LOOPBACK)) != 0) { /* * Some functions expect that ifa_dstaddr is not * NULL for p2p interfaces. */ ia->ia_ifa.ifa_dstaddr = (struct sockaddr *)&ia->ia_dstaddr; } else { ia->ia_ifa.ifa_dstaddr = NULL; } /* set prefix mask if any */ ia->ia_ifa.ifa_netmask = (struct sockaddr *)&ia->ia_prefixmask; if (ifra->ifra_prefixmask.sin6_len != 0) { ia->ia_prefixmask.sin6_family = AF_INET6; ia->ia_prefixmask.sin6_len = ifra->ifra_prefixmask.sin6_len; ia->ia_prefixmask.sin6_addr = ifra->ifra_prefixmask.sin6_addr; } ia->ia_ifp = ifp; ifa_ref(&ia->ia_ifa); /* if_addrhead */ IF_ADDR_WLOCK(ifp); TAILQ_INSERT_TAIL(&ifp->if_addrhead, &ia->ia_ifa, ifa_link); IF_ADDR_WUNLOCK(ifp); ifa_ref(&ia->ia_ifa); /* in6_ifaddrhead */ IN6_IFADDR_WLOCK(); TAILQ_INSERT_TAIL(&V_in6_ifaddrhead, ia, ia_link); LIST_INSERT_HEAD(IN6ADDR_HASH(&ia->ia_addr.sin6_addr), ia, ia6_hash); IN6_IFADDR_WUNLOCK(); return (ia); } /* * Update/configure interface address parameters: * * 1) Update lifetime * 2) Update interface metric ad flags * 3) Notify other subsystems */ static int in6_update_ifa_internal(struct ifnet *ifp, struct in6_aliasreq *ifra, struct in6_ifaddr *ia, int hostIsNew, int flags) { int error; /* update timestamp */ ia->ia6_updatetime = time_uptime; /* * Set lifetimes. We do not refer to ia6t_expire and ia6t_preferred * to see if the address is deprecated or invalidated, but initialize * these members for applications. */ ia->ia6_lifetime = ifra->ifra_lifetime; if (ia->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) { ia->ia6_lifetime.ia6t_expire = time_uptime + ia->ia6_lifetime.ia6t_vltime; } else ia->ia6_lifetime.ia6t_expire = 0; if (ia->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) { ia->ia6_lifetime.ia6t_preferred = time_uptime + ia->ia6_lifetime.ia6t_pltime; } else ia->ia6_lifetime.ia6t_preferred = 0; /* * backward compatibility - if IN6_IFF_DEPRECATED is set from the * userland, make it deprecated. */ if ((ifra->ifra_flags & IN6_IFF_DEPRECATED) != 0) { ia->ia6_lifetime.ia6t_pltime = 0; ia->ia6_lifetime.ia6t_preferred = time_uptime; } /* * configure address flags. */ ia->ia6_flags = ifra->ifra_flags; /* * Make the address tentative before joining multicast addresses, * so that corresponding MLD responses would not have a tentative * source address. */ ia->ia6_flags &= ~IN6_IFF_DUPLICATED; /* safety */ if (hostIsNew && in6if_do_dad(ifp)) ia->ia6_flags |= IN6_IFF_TENTATIVE; /* DAD should be performed after ND6_IFF_IFDISABLED is cleared. */ if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) ia->ia6_flags |= IN6_IFF_TENTATIVE; /* notify other subsystems */ error = in6_notify_ifa(ifp, ia, ifra, hostIsNew); return (error); } /* * Do link-level ifa job: * 1) Add lle entry for added address * 2) Notifies routing socket users about new address * 3) join appropriate multicast group * 4) start DAD if enabled */ static int in6_broadcast_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra, struct in6_ifaddr *ia, int flags) { struct in6_multi *in6m_sol; int error = 0; /* Add local address to lltable, if necessary (ex. on p2p link). */ if ((error = nd6_add_ifa_lle(ia)) != 0) { in6_purgeaddr(&ia->ia_ifa); ifa_free(&ia->ia_ifa); return (error); } /* Join necessary multicast groups. */ in6m_sol = NULL; if ((ifp->if_flags & IFF_MULTICAST) != 0) { error = in6_update_ifa_join_mc(ifp, ifra, ia, flags, &in6m_sol); if (error != 0) { in6_purgeaddr(&ia->ia_ifa); ifa_free(&ia->ia_ifa); return (error); } } /* * Perform DAD, if needed. * XXX It may be of use, if we can administratively disable DAD. */ if (in6if_do_dad(ifp) && ((ifra->ifra_flags & IN6_IFF_NODAD) == 0) && (ia->ia6_flags & IN6_IFF_TENTATIVE)) { int delay, mindelay, maxdelay; delay = 0; if ((flags & IN6_IFAUPDATE_DADDELAY)) { /* * We need to impose a delay before sending an NS * for DAD. Check if we also needed a delay for the * corresponding MLD message. If we did, the delay * should be larger than the MLD delay (this could be * relaxed a bit, but this simple logic is at least * safe). * XXX: Break data hiding guidelines and look at * state for the solicited multicast group. */ mindelay = 0; if (in6m_sol != NULL && in6m_sol->in6m_state == MLD_REPORTING_MEMBER) { mindelay = in6m_sol->in6m_timer; } maxdelay = MAX_RTR_SOLICITATION_DELAY * hz; if (maxdelay - mindelay == 0) delay = 0; else { delay = (arc4random() % (maxdelay - mindelay)) + mindelay; } } nd6_dad_start((struct ifaddr *)ia, delay); } ifa_free(&ia->ia_ifa); return (error); } void in6_purgeaddr(struct ifaddr *ifa) { struct ifnet *ifp = ifa->ifa_ifp; struct in6_ifaddr *ia = (struct in6_ifaddr *) ifa; struct in6_multi_mship *imm; int plen, error; if (ifa->ifa_carp) (*carp_detach_p)(ifa); /* * Remove the loopback route to the interface address. * The check for the current setting of "nd6_useloopback" * is not needed. */ if (ia->ia_flags & IFA_RTSELF) { error = ifa_del_loopback_route((struct ifaddr *)ia, (struct sockaddr *)&ia->ia_addr); if (error == 0) ia->ia_flags &= ~IFA_RTSELF; } /* stop DAD processing */ nd6_dad_stop(ifa); /* Remove local address entry from lltable. */ nd6_rem_ifa_lle(ia); /* Leave multicast groups. */ while ((imm = LIST_FIRST(&ia->ia6_memberships)) != NULL) { LIST_REMOVE(imm, i6mm_chain); in6_leavegroup(imm); } plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); /* XXX */ if ((ia->ia_flags & IFA_ROUTE) && plen == 128) { error = rtinit(&(ia->ia_ifa), RTM_DELETE, ia->ia_flags | (ia->ia_dstaddr.sin6_family == AF_INET6) ? RTF_HOST : 0); if (error != 0) log(LOG_INFO, "%s: err=%d, destination address delete " "failed\n", __func__, error); ia->ia_flags &= ~IFA_ROUTE; } in6_unlink_ifa(ia, ifp); } static void in6_unlink_ifa(struct in6_ifaddr *ia, struct ifnet *ifp) { char ip6buf[INET6_ADDRSTRLEN]; IF_ADDR_WLOCK(ifp); TAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifa_link); IF_ADDR_WUNLOCK(ifp); ifa_free(&ia->ia_ifa); /* if_addrhead */ /* * Defer the release of what might be the last reference to the * in6_ifaddr so that it can't be freed before the remainder of the * cleanup. */ IN6_IFADDR_WLOCK(); TAILQ_REMOVE(&V_in6_ifaddrhead, ia, ia_link); LIST_REMOVE(ia, ia6_hash); IN6_IFADDR_WUNLOCK(); /* * Release the reference to the base prefix. There should be a * positive reference. */ if (ia->ia6_ndpr == NULL) { nd6log((LOG_NOTICE, "in6_unlink_ifa: autoconf'ed address " "%s has no prefix\n", ip6_sprintf(ip6buf, IA6_IN6(ia)))); } else { ia->ia6_ndpr->ndpr_refcnt--; ia->ia6_ndpr = NULL; } /* * Also, if the address being removed is autoconf'ed, call * pfxlist_onlink_check() since the release might affect the status of * other (detached) addresses. */ if ((ia->ia6_flags & IN6_IFF_AUTOCONF)) { pfxlist_onlink_check(); } ifa_free(&ia->ia_ifa); /* in6_ifaddrhead */ } /* * Notifies other other subsystems about address change/arrival: * 1) Notifies device handler on first IPv6 address assignment * 2) Handle routing table changes for P2P links and route * 3) Handle routing table changes for address host route */ static int in6_notify_ifa(struct ifnet *ifp, struct in6_ifaddr *ia, struct in6_aliasreq *ifra, int hostIsNew) { int error = 0, plen, ifacount = 0; struct ifaddr *ifa; struct sockaddr_in6 *pdst; char ip6buf[INET6_ADDRSTRLEN]; /* * Give the interface a chance to initialize * if this is its first address, */ if (hostIsNew != 0) { IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; ifacount++; } IF_ADDR_RUNLOCK(ifp); } if (ifacount <= 1 && ifp->if_ioctl) { error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia); if (error) return (error); } /* * If a new destination address is specified, scrub the old one and * install the new destination. Note that the interface must be * p2p or loopback. */ pdst = &ifra->ifra_dstaddr; if (pdst->sin6_family == AF_INET6 && !IN6_ARE_ADDR_EQUAL(&pdst->sin6_addr, &ia->ia_dstaddr.sin6_addr)) { if ((ia->ia_flags & IFA_ROUTE) != 0 && (rtinit(&(ia->ia_ifa), (int)RTM_DELETE, RTF_HOST) != 0)) { nd6log((LOG_ERR, "in6_update_ifa_internal: failed to " "remove a route to the old destination: %s\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr))); /* proceed anyway... */ } else ia->ia_flags &= ~IFA_ROUTE; ia->ia_dstaddr = *pdst; } /* * If a new destination address is specified for a point-to-point * interface, install a route to the destination as an interface * direct route. * XXX: the logic below rejects assigning multiple addresses on a p2p * interface that share the same destination. */ plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); /* XXX */ if (!(ia->ia_flags & IFA_ROUTE) && plen == 128 && ia->ia_dstaddr.sin6_family == AF_INET6) { int rtflags = RTF_UP | RTF_HOST; /* * Handle the case for ::1 . */ if (ifp->if_flags & IFF_LOOPBACK) ia->ia_flags |= IFA_RTSELF; error = rtinit(&ia->ia_ifa, RTM_ADD, ia->ia_flags | rtflags); if (error) return (error); ia->ia_flags |= IFA_ROUTE; } /* * add a loopback route to self if not exists */ if (!(ia->ia_flags & IFA_RTSELF) && V_nd6_useloopback) { error = ifa_add_loopback_route((struct ifaddr *)ia, (struct sockaddr *)&ia->ia_addr); if (error == 0) ia->ia_flags |= IFA_RTSELF; } return (error); } /* * Find an IPv6 interface link-local address specific to an interface. * ifaddr is returned referenced. */ struct in6_ifaddr * in6ifa_ifpforlinklocal(struct ifnet *ifp, int ignoreflags) { struct ifaddr *ifa; IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; if (IN6_IS_ADDR_LINKLOCAL(IFA_IN6(ifa))) { if ((((struct in6_ifaddr *)ifa)->ia6_flags & ignoreflags) != 0) continue; ifa_ref(ifa); break; } } IF_ADDR_RUNLOCK(ifp); return ((struct in6_ifaddr *)ifa); } /* * find the internet address corresponding to a given address. * ifaddr is returned referenced. */ struct in6_ifaddr * in6ifa_ifwithaddr(const struct in6_addr *addr, uint32_t zoneid) { struct in6_ifaddr *ia; IN6_IFADDR_RLOCK(); LIST_FOREACH(ia, IN6ADDR_HASH(addr), ia6_hash) { if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), addr)) { if (zoneid != 0 && zoneid != ia->ia_addr.sin6_scope_id) continue; ifa_ref(&ia->ia_ifa); break; } } IN6_IFADDR_RUNLOCK(); return (ia); } /* * find the internet address corresponding to a given interface and address. * ifaddr is returned referenced. */ struct in6_ifaddr * in6ifa_ifpwithaddr(struct ifnet *ifp, struct in6_addr *addr) { struct ifaddr *ifa; IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; if (IN6_ARE_ADDR_EQUAL(addr, IFA_IN6(ifa))) { ifa_ref(ifa); break; } } IF_ADDR_RUNLOCK(ifp); return ((struct in6_ifaddr *)ifa); } /* * Find a link-local scoped address on ifp and return it if any. */ struct in6_ifaddr * in6ifa_llaonifp(struct ifnet *ifp) { struct sockaddr_in6 *sin6; struct ifaddr *ifa; if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) return (NULL); if_addr_rlock(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; sin6 = (struct sockaddr_in6 *)ifa->ifa_addr; if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr) || IN6_IS_ADDR_MC_INTFACELOCAL(&sin6->sin6_addr) || IN6_IS_ADDR_MC_NODELOCAL(&sin6->sin6_addr)) break; } if_addr_runlock(ifp); return ((struct in6_ifaddr *)ifa); } /* * Convert IP6 address to printable (loggable) representation. Caller * has to make sure that ip6buf is at least INET6_ADDRSTRLEN long. */ static char digits[] = "0123456789abcdef"; char * ip6_sprintf(char *ip6buf, const struct in6_addr *addr) { int i, cnt = 0, maxcnt = 0, idx = 0, index = 0; char *cp; const u_int16_t *a = (const u_int16_t *)addr; const u_int8_t *d; int dcolon = 0, zero = 0; cp = ip6buf; for (i = 0; i < 8; i++) { if (*(a + i) == 0) { cnt++; if (cnt == 1) idx = i; } else if (maxcnt < cnt) { maxcnt = cnt; index = idx; cnt = 0; } } if (maxcnt < cnt) { maxcnt = cnt; index = idx; } for (i = 0; i < 8; i++) { if (dcolon == 1) { if (*a == 0) { if (i == 7) *cp++ = ':'; a++; continue; } else dcolon = 2; } if (*a == 0) { if (dcolon == 0 && *(a + 1) == 0 && i == index) { if (i == 0) *cp++ = ':'; *cp++ = ':'; dcolon = 1; } else { *cp++ = '0'; *cp++ = ':'; } a++; continue; } d = (const u_char *)a; /* Try to eliminate leading zeros in printout like in :0001. */ zero = 1; *cp = digits[*d >> 4]; if (*cp != '0') { zero = 0; cp++; } *cp = digits[*d++ & 0xf]; if (zero == 0 || (*cp != '0')) { zero = 0; cp++; } *cp = digits[*d >> 4]; if (zero == 0 || (*cp != '0')) { zero = 0; cp++; } *cp++ = digits[*d & 0xf]; *cp++ = ':'; a++; } *--cp = '\0'; return (ip6buf); } int in6_localaddr(struct in6_addr *in6) { struct in6_ifaddr *ia; if (IN6_IS_ADDR_LOOPBACK(in6) || IN6_IS_ADDR_LINKLOCAL(in6)) return 1; IN6_IFADDR_RLOCK(); TAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) { if (IN6_ARE_MASKED_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr, &ia->ia_prefixmask.sin6_addr)) { IN6_IFADDR_RUNLOCK(); return 1; } } IN6_IFADDR_RUNLOCK(); return (0); } /* * Return 1 if an internet address is for the local host and configured * on one of its interfaces. */ int in6_localip(struct in6_addr *in6) { struct in6_ifaddr *ia; IN6_IFADDR_RLOCK(); LIST_FOREACH(ia, IN6ADDR_HASH(in6), ia6_hash) { if (IN6_ARE_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr)) { IN6_IFADDR_RUNLOCK(); return (1); } } IN6_IFADDR_RUNLOCK(); return (0); } int in6_is_addr_deprecated(struct sockaddr_in6 *sa6) { struct in6_ifaddr *ia; IN6_IFADDR_RLOCK(); LIST_FOREACH(ia, IN6ADDR_HASH(&sa6->sin6_addr), ia6_hash) { if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), &sa6->sin6_addr)) { if (ia->ia6_flags & IN6_IFF_DEPRECATED) { IN6_IFADDR_RUNLOCK(); return (1); /* true */ } break; } } IN6_IFADDR_RUNLOCK(); return (0); /* false */ } /* * return length of part which dst and src are equal * hard coding... */ int in6_matchlen(struct in6_addr *src, struct in6_addr *dst) { int match = 0; u_char *s = (u_char *)src, *d = (u_char *)dst; u_char *lim = s + 16, r; while (s < lim) if ((r = (*d++ ^ *s++)) != 0) { while (r < 128) { match++; r <<= 1; } break; } else match += 8; return match; } /* XXX: to be scope conscious */ int in6_are_prefix_equal(struct in6_addr *p1, struct in6_addr *p2, int len) { int bytelen, bitlen; /* sanity check */ if (0 > len || len > 128) { log(LOG_ERR, "in6_are_prefix_equal: invalid prefix length(%d)\n", len); return (0); } bytelen = len / 8; bitlen = len % 8; if (bcmp(&p1->s6_addr, &p2->s6_addr, bytelen)) return (0); if (bitlen != 0 && p1->s6_addr[bytelen] >> (8 - bitlen) != p2->s6_addr[bytelen] >> (8 - bitlen)) return (0); return (1); } void in6_prefixlen2mask(struct in6_addr *maskp, int len) { u_char maskarray[8] = {0x80, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc, 0xfe, 0xff}; int bytelen, bitlen, i; /* sanity check */ if (0 > len || len > 128) { log(LOG_ERR, "in6_prefixlen2mask: invalid prefix length(%d)\n", len); return; } bzero(maskp, sizeof(*maskp)); bytelen = len / 8; bitlen = len % 8; for (i = 0; i < bytelen; i++) maskp->s6_addr[i] = 0xff; if (bitlen) maskp->s6_addr[bytelen] = maskarray[bitlen - 1]; } /* * return the best address out of the same scope. if no address was * found, return the first valid address from designated IF. */ struct in6_ifaddr * in6_ifawithifp(struct ifnet *ifp, struct in6_addr *dst) { int dst_scope = in6_addrscope(dst), blen = -1, tlen; struct ifaddr *ifa; struct in6_ifaddr *besta = 0; struct in6_ifaddr *dep[2]; /* last-resort: deprecated */ dep[0] = dep[1] = NULL; /* * We first look for addresses in the same scope. * If there is one, return it. * If two or more, return one which matches the dst longest. * If none, return one of global addresses assigned other ifs. */ IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST) continue; /* XXX: is there any case to allow anycast? */ if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY) continue; /* don't use this interface */ if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED) continue; if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) { if (V_ip6_use_deprecated) dep[0] = (struct in6_ifaddr *)ifa; continue; } if (dst_scope == in6_addrscope(IFA_IN6(ifa))) { /* * call in6_matchlen() as few as possible */ if (besta) { if (blen == -1) blen = in6_matchlen(&besta->ia_addr.sin6_addr, dst); tlen = in6_matchlen(IFA_IN6(ifa), dst); if (tlen > blen) { blen = tlen; besta = (struct in6_ifaddr *)ifa; } } else besta = (struct in6_ifaddr *)ifa; } } if (besta) { ifa_ref(&besta->ia_ifa); IF_ADDR_RUNLOCK(ifp); return (besta); } TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST) continue; /* XXX: is there any case to allow anycast? */ if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY) continue; /* don't use this interface */ if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED) continue; if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) { if (V_ip6_use_deprecated) dep[1] = (struct in6_ifaddr *)ifa; continue; } if (ifa != NULL) ifa_ref(ifa); IF_ADDR_RUNLOCK(ifp); return (struct in6_ifaddr *)ifa; } /* use the last-resort values, that are, deprecated addresses */ if (dep[0]) { ifa_ref((struct ifaddr *)dep[0]); IF_ADDR_RUNLOCK(ifp); return dep[0]; } if (dep[1]) { ifa_ref((struct ifaddr *)dep[1]); IF_ADDR_RUNLOCK(ifp); return dep[1]; } IF_ADDR_RUNLOCK(ifp); return NULL; } /* * perform DAD when interface becomes IFF_UP. */ void in6_if_up(struct ifnet *ifp) { struct ifaddr *ifa; struct in6_ifaddr *ia; IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; ia = (struct in6_ifaddr *)ifa; if (ia->ia6_flags & IN6_IFF_TENTATIVE) { /* * The TENTATIVE flag was likely set by hand * beforehand, implicitly indicating the need for DAD. * We may be able to skip the random delay in this * case, but we impose delays just in case. */ nd6_dad_start(ifa, arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz)); } } IF_ADDR_RUNLOCK(ifp); /* * special cases, like 6to4, are handled in in6_ifattach */ in6_ifattach(ifp, NULL); } int in6if_do_dad(struct ifnet *ifp) { if ((ifp->if_flags & IFF_LOOPBACK) != 0) return (0); - if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) + if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) || + (ND_IFINFO(ifp)->flags & ND6_IFF_NO_DAD)) return (0); /* * Our DAD routine requires the interface up and running. * However, some interfaces can be up before the RUNNING * status. Additionaly, users may try to assign addresses * before the interface becomes up (or running). * We simply skip DAD in such a case as a work around. * XXX: we should rather mark "tentative" on such addresses, * and do DAD after the interface becomes ready. */ if (!((ifp->if_flags & IFF_UP) && (ifp->if_drv_flags & IFF_DRV_RUNNING))) return (0); return (1); } /* * Calculate max IPv6 MTU through all the interfaces and store it * to in6_maxmtu. */ void in6_setmaxmtu(void) { unsigned long maxmtu = 0; struct ifnet *ifp; IFNET_RLOCK_NOSLEEP(); TAILQ_FOREACH(ifp, &V_ifnet, if_link) { /* this function can be called during ifnet initialization */ if (!ifp->if_afdata[AF_INET6]) continue; if ((ifp->if_flags & IFF_LOOPBACK) == 0 && IN6_LINKMTU(ifp) > maxmtu) maxmtu = IN6_LINKMTU(ifp); } IFNET_RUNLOCK_NOSLEEP(); if (maxmtu) /* update only when maxmtu is positive */ V_in6_maxmtu = maxmtu; } /* * Provide the length of interface identifiers to be used for the link attached * to the given interface. The length should be defined in "IPv6 over * xxx-link" document. Note that address architecture might also define * the length for a particular set of address prefixes, regardless of the * link type. As clarified in rfc2462bis, those two definitions should be * consistent, and those really are as of August 2004. */ int in6_if2idlen(struct ifnet *ifp) { switch (ifp->if_type) { case IFT_ETHER: /* RFC2464 */ #ifdef IFT_PROPVIRTUAL case IFT_PROPVIRTUAL: /* XXX: no RFC. treat it as ether */ #endif #ifdef IFT_L2VLAN case IFT_L2VLAN: /* ditto */ #endif #ifdef IFT_IEEE80211 case IFT_IEEE80211: /* ditto */ #endif #ifdef IFT_MIP case IFT_MIP: /* ditto */ #endif case IFT_INFINIBAND: return (64); case IFT_FDDI: /* RFC2467 */ return (64); case IFT_ISO88025: /* RFC2470 (IPv6 over Token Ring) */ return (64); case IFT_PPP: /* RFC2472 */ return (64); case IFT_ARCNET: /* RFC2497 */ return (64); case IFT_FRELAY: /* RFC2590 */ return (64); case IFT_IEEE1394: /* RFC3146 */ return (64); case IFT_GIF: return (64); /* draft-ietf-v6ops-mech-v2-07 */ case IFT_LOOP: return (64); /* XXX: is this really correct? */ default: /* * Unknown link type: * It might be controversial to use the today's common constant * of 64 for these cases unconditionally. For full compliance, * we should return an error in this case. On the other hand, * if we simply miss the standard for the link type or a new * standard is defined for a new link type, the IFID length * is very likely to be the common constant. As a compromise, * we always use the constant, but make an explicit notice * indicating the "unknown" case. */ printf("in6_if2idlen: unknown link type (%d)\n", ifp->if_type); return (64); } } #include struct in6_llentry { struct llentry base; struct sockaddr_in6 l3_addr6; }; /* * Deletes an address from the address table. * This function is called by the timer functions * such as arptimer() and nd6_llinfo_timer(), and * the caller does the locking. */ static void in6_lltable_free(struct lltable *llt, struct llentry *lle) { LLE_WUNLOCK(lle); LLE_LOCK_DESTROY(lle); free(lle, M_LLTABLE); } static struct llentry * in6_lltable_new(const struct sockaddr *l3addr, u_int flags) { struct in6_llentry *lle; lle = malloc(sizeof(struct in6_llentry), M_LLTABLE, M_NOWAIT | M_ZERO); if (lle == NULL) /* NB: caller generates msg */ return NULL; lle->l3_addr6 = *(const struct sockaddr_in6 *)l3addr; lle->base.lle_refcnt = 1; lle->base.lle_free = in6_lltable_free; LLE_LOCK_INIT(&lle->base); callout_init(&lle->base.ln_timer_ch, 1); return (&lle->base); } static void in6_lltable_prefix_free(struct lltable *llt, const struct sockaddr *prefix, const struct sockaddr *mask, u_int flags) { const struct sockaddr_in6 *pfx = (const struct sockaddr_in6 *)prefix; const struct sockaddr_in6 *msk = (const struct sockaddr_in6 *)mask; struct llentry *lle, *next; int i; /* * (flags & LLE_STATIC) means deleting all entries * including static ND6 entries. */ IF_AFDATA_WLOCK(llt->llt_ifp); for (i = 0; i < LLTBL_HASHTBL_SIZE; i++) { LIST_FOREACH_SAFE(lle, &llt->lle_head[i], lle_next, next) { if (IN6_ARE_MASKED_ADDR_EQUAL( &satosin6(L3_ADDR(lle))->sin6_addr, &pfx->sin6_addr, &msk->sin6_addr) && ((flags & LLE_STATIC) || !(lle->la_flags & LLE_STATIC))) { LLE_WLOCK(lle); if (callout_stop(&lle->la_timer)) LLE_REMREF(lle); llentry_free(lle); } } } IF_AFDATA_WUNLOCK(llt->llt_ifp); } static int in6_lltable_rtcheck(struct ifnet *ifp, u_int flags, const struct sockaddr *l3addr) { struct rtentry *rt; char ip6buf[INET6_ADDRSTRLEN]; KASSERT(l3addr->sa_family == AF_INET6, ("sin_family %d", l3addr->sa_family)); /* Our local addresses are always only installed on the default FIB. */ /* XXX rtalloc1 should take a const param */ rt = in6_rtalloc1(__DECONST(struct sockaddr *, l3addr), 0, 0, RT_DEFAULT_FIB); if (rt == NULL || (rt->rt_flags & RTF_GATEWAY) || rt->rt_ifp != ifp) { struct ifaddr *ifa; /* * Create an ND6 cache for an IPv6 neighbor * that is not covered by our own prefix. */ /* XXX ifaof_ifpforaddr should take a const param */ ifa = ifaof_ifpforaddr(__DECONST(struct sockaddr *, l3addr), ifp); if (ifa != NULL) { ifa_free(ifa); if (rt != NULL) RTFREE_LOCKED(rt); return 0; } log(LOG_INFO, "IPv6 address: \"%s\" is not on the network\n", ip6_sprintf(ip6buf, &((const struct sockaddr_in6 *)l3addr)->sin6_addr)); if (rt != NULL) RTFREE_LOCKED(rt); return EINVAL; } RTFREE_LOCKED(rt); return 0; } static struct llentry * in6_lltable_lookup(struct lltable *llt, u_int flags, const struct sockaddr *l3addr) { const struct sockaddr_in6 *sin6 = (const struct sockaddr_in6 *)l3addr; struct ifnet *ifp = llt->llt_ifp; struct llentry *lle; struct llentries *lleh; u_int hashkey; IF_AFDATA_LOCK_ASSERT(ifp); KASSERT(l3addr->sa_family == AF_INET6, ("sin_family %d", l3addr->sa_family)); hashkey = sin6->sin6_addr.s6_addr32[3]; lleh = &llt->lle_head[LLATBL_HASH(hashkey, LLTBL_HASHMASK)]; LIST_FOREACH(lle, lleh, lle_next) { struct sockaddr_in6 *sa6 = (struct sockaddr_in6 *)L3_ADDR(lle); if (lle->la_flags & LLE_DELETED) continue; if (bcmp(&sa6->sin6_addr, &sin6->sin6_addr, sizeof(struct in6_addr)) == 0) break; } if (lle == NULL) { if (!(flags & LLE_CREATE)) return (NULL); IF_AFDATA_WLOCK_ASSERT(ifp); /* * A route that covers the given address must have * been installed 1st because we are doing a resolution, * verify this. */ if (!(flags & LLE_IFADDR) && in6_lltable_rtcheck(ifp, flags, l3addr) != 0) return NULL; lle = in6_lltable_new(l3addr, flags); if (lle == NULL) { log(LOG_INFO, "lla_lookup: new lle malloc failed\n"); return NULL; } lle->la_flags = flags & ~LLE_CREATE; if ((flags & (LLE_CREATE | LLE_IFADDR)) == (LLE_CREATE | LLE_IFADDR)) { bcopy(IF_LLADDR(ifp), &lle->ll_addr, ifp->if_addrlen); lle->la_flags |= (LLE_VALID | LLE_STATIC); } lle->lle_tbl = llt; lle->lle_head = lleh; lle->la_flags |= LLE_LINKED; LIST_INSERT_HEAD(lleh, lle, lle_next); } else if (flags & LLE_DELETE) { if (!(lle->la_flags & LLE_IFADDR) || (flags & LLE_IFADDR)) { LLE_WLOCK(lle); lle->la_flags |= LLE_DELETED; #ifdef DIAGNOSTIC log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle); #endif if ((lle->la_flags & (LLE_STATIC | LLE_IFADDR)) == LLE_STATIC) llentry_free(lle); else LLE_WUNLOCK(lle); } lle = (void *)-1; } if (LLE_IS_VALID(lle)) { if (flags & LLE_EXCLUSIVE) LLE_WLOCK(lle); else LLE_RLOCK(lle); } return (lle); } static int in6_lltable_dump(struct lltable *llt, struct sysctl_req *wr) { struct ifnet *ifp = llt->llt_ifp; struct llentry *lle; /* XXX stack use */ struct { struct rt_msghdr rtm; struct sockaddr_in6 sin6; /* * ndp.c assumes that sdl is word aligned */ #ifdef __LP64__ uint32_t pad; #endif struct sockaddr_dl sdl; } ndpc; int i, error; if (ifp->if_flags & IFF_LOOPBACK) return 0; LLTABLE_LOCK_ASSERT(); error = 0; for (i = 0; i < LLTBL_HASHTBL_SIZE; i++) { LIST_FOREACH(lle, &llt->lle_head[i], lle_next) { struct sockaddr_dl *sdl; /* skip deleted or invalid entries */ if ((lle->la_flags & (LLE_DELETED|LLE_VALID)) != LLE_VALID) continue; /* Skip if jailed and not a valid IP of the prison. */ if (prison_if(wr->td->td_ucred, L3_ADDR(lle)) != 0) continue; /* * produce a msg made of: * struct rt_msghdr; * struct sockaddr_in6 (IPv6) * struct sockaddr_dl; */ bzero(&ndpc, sizeof(ndpc)); ndpc.rtm.rtm_msglen = sizeof(ndpc); ndpc.rtm.rtm_version = RTM_VERSION; ndpc.rtm.rtm_type = RTM_GET; ndpc.rtm.rtm_flags = RTF_UP; ndpc.rtm.rtm_addrs = RTA_DST | RTA_GATEWAY; ndpc.sin6.sin6_family = AF_INET6; ndpc.sin6.sin6_len = sizeof(ndpc.sin6); bcopy(L3_ADDR(lle), &ndpc.sin6, L3_ADDR_LEN(lle)); if (V_deembed_scopeid) sa6_recoverscope(&ndpc.sin6); /* publish */ if (lle->la_flags & LLE_PUB) ndpc.rtm.rtm_flags |= RTF_ANNOUNCE; sdl = &ndpc.sdl; sdl->sdl_family = AF_LINK; sdl->sdl_len = sizeof(*sdl); sdl->sdl_alen = ifp->if_addrlen; sdl->sdl_index = ifp->if_index; sdl->sdl_type = ifp->if_type; bcopy(&lle->ll_addr, LLADDR(sdl), ifp->if_addrlen); ndpc.rtm.rtm_rmx.rmx_expire = lle->la_flags & LLE_STATIC ? 0 : lle->la_expire; ndpc.rtm.rtm_flags |= (RTF_HOST | RTF_LLDATA); if (lle->la_flags & LLE_STATIC) ndpc.rtm.rtm_flags |= RTF_STATIC; ndpc.rtm.rtm_index = ifp->if_index; error = SYSCTL_OUT(wr, &ndpc, sizeof(ndpc)); if (error) break; } } return error; } void * in6_domifattach(struct ifnet *ifp) { struct in6_ifextra *ext; /* There are not IPv6-capable interfaces. */ switch (ifp->if_type) { case IFT_PFLOG: case IFT_PFSYNC: case IFT_USB: return (NULL); } ext = (struct in6_ifextra *)malloc(sizeof(*ext), M_IFADDR, M_WAITOK); bzero(ext, sizeof(*ext)); ext->in6_ifstat = malloc(sizeof(counter_u64_t) * sizeof(struct in6_ifstat) / sizeof(uint64_t), M_IFADDR, M_WAITOK); COUNTER_ARRAY_ALLOC(ext->in6_ifstat, sizeof(struct in6_ifstat) / sizeof(uint64_t), M_WAITOK); ext->icmp6_ifstat = malloc(sizeof(counter_u64_t) * sizeof(struct icmp6_ifstat) / sizeof(uint64_t), M_IFADDR, M_WAITOK); COUNTER_ARRAY_ALLOC(ext->icmp6_ifstat, sizeof(struct icmp6_ifstat) / sizeof(uint64_t), M_WAITOK); ext->nd_ifinfo = nd6_ifattach(ifp); ext->scope6_id = scope6_ifattach(ifp); ext->lltable = lltable_init(ifp, AF_INET6); if (ext->lltable != NULL) { ext->lltable->llt_prefix_free = in6_lltable_prefix_free; ext->lltable->llt_lookup = in6_lltable_lookup; ext->lltable->llt_dump = in6_lltable_dump; } ext->mld_ifinfo = mld_domifattach(ifp); return ext; } int in6_domifmtu(struct ifnet *ifp) { return (IN6_LINKMTU(ifp)); } void in6_domifdetach(struct ifnet *ifp, void *aux) { struct in6_ifextra *ext = (struct in6_ifextra *)aux; mld_domifdetach(ifp); scope6_ifdetach(ext->scope6_id); nd6_ifdetach(ext->nd_ifinfo); lltable_free(ext->lltable); COUNTER_ARRAY_FREE(ext->in6_ifstat, sizeof(struct in6_ifstat) / sizeof(uint64_t)); free(ext->in6_ifstat, M_IFADDR); COUNTER_ARRAY_FREE(ext->icmp6_ifstat, sizeof(struct icmp6_ifstat) / sizeof(uint64_t)); free(ext->icmp6_ifstat, M_IFADDR); free(ext, M_IFADDR); } /* * Convert sockaddr_in6 to sockaddr_in. Original sockaddr_in6 must be * v4 mapped addr or v4 compat addr */ void in6_sin6_2_sin(struct sockaddr_in *sin, struct sockaddr_in6 *sin6) { bzero(sin, sizeof(*sin)); sin->sin_len = sizeof(struct sockaddr_in); sin->sin_family = AF_INET; sin->sin_port = sin6->sin6_port; sin->sin_addr.s_addr = sin6->sin6_addr.s6_addr32[3]; } /* Convert sockaddr_in to sockaddr_in6 in v4 mapped addr format. */ void in6_sin_2_v4mapsin6(struct sockaddr_in *sin, struct sockaddr_in6 *sin6) { bzero(sin6, sizeof(*sin6)); sin6->sin6_len = sizeof(struct sockaddr_in6); sin6->sin6_family = AF_INET6; sin6->sin6_port = sin->sin_port; sin6->sin6_addr.s6_addr32[0] = 0; sin6->sin6_addr.s6_addr32[1] = 0; sin6->sin6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP; sin6->sin6_addr.s6_addr32[3] = sin->sin_addr.s_addr; } /* Convert sockaddr_in6 into sockaddr_in. */ void in6_sin6_2_sin_in_sock(struct sockaddr *nam) { struct sockaddr_in *sin_p; struct sockaddr_in6 sin6; /* * Save original sockaddr_in6 addr and convert it * to sockaddr_in. */ sin6 = *(struct sockaddr_in6 *)nam; sin_p = (struct sockaddr_in *)nam; in6_sin6_2_sin(sin_p, &sin6); } /* Convert sockaddr_in into sockaddr_in6 in v4 mapped addr format. */ void in6_sin_2_v4mapsin6_in_sock(struct sockaddr **nam) { struct sockaddr_in *sin_p; struct sockaddr_in6 *sin6_p; sin6_p = malloc(sizeof *sin6_p, M_SONAME, M_WAITOK); sin_p = (struct sockaddr_in *)*nam; in6_sin_2_v4mapsin6(sin_p, sin6_p); free(*nam, M_SONAME); *nam = (struct sockaddr *)sin6_p; } Index: projects/cxl_iscsi/sys/netinet6/in6_pcb.c =================================================================== --- projects/cxl_iscsi/sys/netinet6/in6_pcb.c (revision 279716) +++ projects/cxl_iscsi/sys/netinet6/in6_pcb.c (revision 279717) @@ -1,1287 +1,1281 @@ /*- * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * Copyright (c) 2010-2011 Juniper Networks, Inc. * All rights reserved. * * Portions of this software were developed by Robert N. M. Watson under * contract to Juniper Networks, Inc. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: in6_pcb.c,v 1.31 2001/05/21 05:45:10 jinmei Exp $ */ /*- * Copyright (c) 1982, 1986, 1991, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94 */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_pcbgroup.h" #include "opt_rss.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include static struct inpcb *in6_pcblookup_hash_locked(struct inpcbinfo *, struct in6_addr *, u_int, struct in6_addr *, u_int, int, struct ifnet *); int in6_pcbbind(register struct inpcb *inp, struct sockaddr *nam, struct ucred *cred) { struct socket *so = inp->inp_socket; struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL; struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; u_short lport = 0; int error, lookupflags = 0; int reuseport = (so->so_options & SO_REUSEPORT); INP_WLOCK_ASSERT(inp); INP_HASH_WLOCK_ASSERT(pcbinfo); if (TAILQ_EMPTY(&V_in6_ifaddrhead)) /* XXX broken! */ return (EADDRNOTAVAIL); if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) return (EINVAL); if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) lookupflags = INPLOOKUP_WILDCARD; if (nam == NULL) { if ((error = prison_local_ip6(cred, &inp->in6p_laddr, ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) return (error); } else { sin6 = (struct sockaddr_in6 *)nam; if (nam->sa_len != sizeof(*sin6)) return (EINVAL); /* * family check. */ if (nam->sa_family != AF_INET6) return (EAFNOSUPPORT); if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) return(error); if ((error = prison_local_ip6(cred, &sin6->sin6_addr, ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) return (error); lport = sin6->sin6_port; if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { /* * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; * allow compepte duplication of binding if * SO_REUSEPORT is set, or if SO_REUSEADDR is set * and a multicast address is bound on both * new and duplicated sockets. */ if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) != 0) reuseport = SO_REUSEADDR|SO_REUSEPORT; } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { struct ifaddr *ifa; sin6->sin6_port = 0; /* yech... */ if ((ifa = ifa_ifwithaddr((struct sockaddr *)sin6)) == NULL && (inp->inp_flags & INP_BINDANY) == 0) { return (EADDRNOTAVAIL); } /* * XXX: bind to an anycast address might accidentally * cause sending a packet with anycast source address. * We should allow to bind to a deprecated address, since * the application dares to use it. */ if (ifa != NULL && ((struct in6_ifaddr *)ifa)->ia6_flags & (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) { ifa_free(ifa); return (EADDRNOTAVAIL); } if (ifa != NULL) ifa_free(ifa); } if (lport) { struct inpcb *t; struct tcptw *tw; /* GROSS */ if (ntohs(lport) <= V_ipport_reservedhigh && ntohs(lport) >= V_ipport_reservedlow && priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT, 0)) return (EACCES); if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) && priv_check_cred(inp->inp_cred, PRIV_NETINET_REUSEPORT, 0) != 0) { t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr, lport, INPLOOKUP_WILDCARD, cred); if (t && ((inp->inp_flags2 & INP_BINDMULTI) == 0) && ((t->inp_flags & INP_TIMEWAIT) == 0) && (so->so_type != SOCK_STREAM || IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) && (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) || !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) || (t->inp_flags2 & INP_REUSEPORT) == 0) && (inp->inp_cred->cr_uid != t->inp_cred->cr_uid)) return (EADDRINUSE); /* * If the socket is a BINDMULTI socket, then * the credentials need to match and the * original socket also has to have been bound * with BINDMULTI. */ if (t && (! in_pcbbind_check_bindmulti(inp, t))) return (EADDRINUSE); #ifdef INET if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { struct sockaddr_in sin; in6_sin6_2_sin(&sin, sin6); t = in_pcblookup_local(pcbinfo, sin.sin_addr, lport, INPLOOKUP_WILDCARD, cred); if (t && ((inp->inp_flags2 & INP_BINDMULTI) == 0) && ((t->inp_flags & INP_TIMEWAIT) == 0) && (so->so_type != SOCK_STREAM || ntohl(t->inp_faddr.s_addr) == INADDR_ANY) && (inp->inp_cred->cr_uid != t->inp_cred->cr_uid)) return (EADDRINUSE); if (t && (! in_pcbbind_check_bindmulti(inp, t))) return (EADDRINUSE); } #endif } t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr, lport, lookupflags, cred); if (t && (t->inp_flags & INP_TIMEWAIT)) { /* * XXXRW: If an incpb has had its timewait * state recycled, we treat the address as * being in use (for now). This is better * than a panic, but not desirable. */ tw = intotw(t); if (tw == NULL || (reuseport & tw->tw_so_options) == 0) return (EADDRINUSE); } else if (t && (reuseport & inp_so_options(t)) == 0) { return (EADDRINUSE); } #ifdef INET if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { struct sockaddr_in sin; in6_sin6_2_sin(&sin, sin6); t = in_pcblookup_local(pcbinfo, sin.sin_addr, lport, lookupflags, cred); if (t && t->inp_flags & INP_TIMEWAIT) { tw = intotw(t); if (tw == NULL) return (EADDRINUSE); if ((reuseport & tw->tw_so_options) == 0 && (ntohl(t->inp_laddr.s_addr) != INADDR_ANY || ((inp->inp_vflag & INP_IPV6PROTO) == (t->inp_vflag & INP_IPV6PROTO)))) return (EADDRINUSE); } else if (t && (reuseport & inp_so_options(t)) == 0 && (ntohl(t->inp_laddr.s_addr) != INADDR_ANY || (t->inp_vflag & INP_IPV6PROTO) != 0)) return (EADDRINUSE); } #endif } inp->in6p_laddr = sin6->sin6_addr; } if (lport == 0) { if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0) { /* Undo an address bind that may have occurred. */ inp->in6p_laddr = in6addr_any; return (error); } } else { inp->inp_lport = lport; if (in_pcbinshash(inp) != 0) { inp->in6p_laddr = in6addr_any; inp->inp_lport = 0; return (EAGAIN); } } return (0); } /* * Transform old in6_pcbconnect() into an inner subroutine for new * in6_pcbconnect(): Do some validity-checking on the remote * address (in mbuf 'nam') and then determine local host address * (i.e., which interface) to use to access that remote host. * * This preserves definition of in6_pcbconnect(), while supporting a * slightly different version for T/TCP. (This is more than * a bit of a kludge, but cleaning up the internal interfaces would * have forced minor changes in every protocol). */ static int in6_pcbladdr(register struct inpcb *inp, struct sockaddr *nam, struct in6_addr *plocal_addr6) { register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; int error = 0; struct ifnet *ifp = NULL; int scope_ambiguous = 0; struct in6_addr in6a; INP_WLOCK_ASSERT(inp); INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); /* XXXRW: why? */ if (nam->sa_len != sizeof (*sin6)) return (EINVAL); if (sin6->sin6_family != AF_INET6) return (EAFNOSUPPORT); if (sin6->sin6_port == 0) return (EADDRNOTAVAIL); if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone) scope_ambiguous = 1; if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) return(error); if (!TAILQ_EMPTY(&V_in6_ifaddrhead)) { /* * If the destination address is UNSPECIFIED addr, * use the loopback addr, e.g ::1. */ if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) sin6->sin6_addr = in6addr_loopback; } if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0) return (error); error = in6_selectsrc(sin6, inp->in6p_outputopts, inp, NULL, inp->inp_cred, &ifp, &in6a); if (error) return (error); if (ifp && scope_ambiguous && (error = in6_setscope(&sin6->sin6_addr, ifp, NULL)) != 0) { return(error); } /* * Do not update this earlier, in case we return with an error. * * XXX: this in6_selectsrc result might replace the bound local * address with the address specified by setsockopt(IPV6_PKTINFO). * Is it the intended behavior? */ *plocal_addr6 = in6a; /* * Don't do pcblookup call here; return interface in * plocal_addr6 * and exit to caller, that will do the lookup. */ return (0); } /* * Outer subroutine: * Connect from a socket to a specified address. * Both address and port must be specified in argument sin. * If don't have a local address for this socket yet, * then pick one. */ int in6_pcbconnect_mbuf(register struct inpcb *inp, struct sockaddr *nam, struct ucred *cred, struct mbuf *m) { struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; struct in6_addr addr6; int error; INP_WLOCK_ASSERT(inp); INP_HASH_WLOCK_ASSERT(pcbinfo); /* * Call inner routine, to assign local interface address. * in6_pcbladdr() may automatically fill in sin6_scope_id. */ if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0) return (error); if (in6_pcblookup_hash_locked(pcbinfo, &sin6->sin6_addr, sin6->sin6_port, IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) ? &addr6 : &inp->in6p_laddr, inp->inp_lport, 0, NULL) != NULL) { return (EADDRINUSE); } if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { if (inp->inp_lport == 0) { error = in6_pcbbind(inp, (struct sockaddr *)0, cred); if (error) return (error); } inp->in6p_laddr = addr6; } inp->in6p_faddr = sin6->sin6_addr; inp->inp_fport = sin6->sin6_port; /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; if (inp->inp_flags & IN6P_AUTOFLOWLABEL) inp->inp_flow |= (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK); in_pcbrehash_mbuf(inp, m); return (0); } int in6_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred) { return (in6_pcbconnect_mbuf(inp, nam, cred, NULL)); } void in6_pcbdisconnect(struct inpcb *inp) { INP_WLOCK_ASSERT(inp); INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr)); inp->inp_fport = 0; /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; in_pcbrehash(inp); } struct sockaddr * in6_sockaddr(in_port_t port, struct in6_addr *addr_p) { struct sockaddr_in6 *sin6; sin6 = malloc(sizeof *sin6, M_SONAME, M_WAITOK); bzero(sin6, sizeof *sin6); sin6->sin6_family = AF_INET6; sin6->sin6_len = sizeof(*sin6); sin6->sin6_port = port; sin6->sin6_addr = *addr_p; (void)sa6_recoverscope(sin6); /* XXX: should catch errors */ return (struct sockaddr *)sin6; } struct sockaddr * in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p) { struct sockaddr_in sin; struct sockaddr_in6 *sin6_p; bzero(&sin, sizeof sin); sin.sin_family = AF_INET; sin.sin_len = sizeof(sin); sin.sin_port = port; sin.sin_addr = *addr_p; sin6_p = malloc(sizeof *sin6_p, M_SONAME, M_WAITOK); in6_sin_2_v4mapsin6(&sin, sin6_p); return (struct sockaddr *)sin6_p; } int in6_getsockaddr(struct socket *so, struct sockaddr **nam) { register struct inpcb *inp; struct in6_addr addr; in_port_t port; inp = sotoinpcb(so); KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL")); INP_RLOCK(inp); port = inp->inp_lport; addr = inp->in6p_laddr; INP_RUNLOCK(inp); *nam = in6_sockaddr(port, &addr); return 0; } int in6_getpeeraddr(struct socket *so, struct sockaddr **nam) { struct inpcb *inp; struct in6_addr addr; in_port_t port; inp = sotoinpcb(so); KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL")); INP_RLOCK(inp); port = inp->inp_fport; addr = inp->in6p_faddr; INP_RUNLOCK(inp); *nam = in6_sockaddr(port, &addr); return 0; } int in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam) { struct inpcb *inp; int error; inp = sotoinpcb(so); KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL")); #ifdef INET if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) { error = in_getsockaddr(so, nam); if (error == 0) in6_sin_2_v4mapsin6_in_sock(nam); } else #endif { /* scope issues will be handled in in6_getsockaddr(). */ error = in6_getsockaddr(so, nam); } return error; } int in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam) { struct inpcb *inp; int error; inp = sotoinpcb(so); KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL")); #ifdef INET if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) { error = in_getpeeraddr(so, nam); if (error == 0) in6_sin_2_v4mapsin6_in_sock(nam); } else #endif /* scope issues will be handled in in6_getpeeraddr(). */ error = in6_getpeeraddr(so, nam); return error; } /* * Pass some notification to all connections of a protocol * associated with address dst. The local address and/or port numbers * may be specified to limit the search. The "usual action" will be * taken, depending on the ctlinput cmd. The caller must filter any * cmds that are uninteresting (e.g., no error in the map). * Call the protocol specific routine (if any) to report * any errors for each matching socket. */ void in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst, u_int fport_arg, const struct sockaddr *src, u_int lport_arg, int cmd, void *cmdarg, struct inpcb *(*notify)(struct inpcb *, int)) { struct inpcb *inp, *inp_temp; struct sockaddr_in6 sa6_src, *sa6_dst; u_short fport = fport_arg, lport = lport_arg; u_int32_t flowinfo; int errno; if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6) return; sa6_dst = (struct sockaddr_in6 *)dst; if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) return; /* * note that src can be NULL when we get notify by local fragmentation. */ sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src; flowinfo = sa6_src.sin6_flowinfo; /* * Redirects go to all references to the destination, * and use in6_rtchange to invalidate the route cache. * Dead host indications: also use in6_rtchange to invalidate * the cache, and deliver the error to all the sockets. * Otherwise, if we have knowledge of the local port and address, * deliver only to that socket. */ if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) { fport = 0; lport = 0; bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr)); if (cmd != PRC_HOSTDEAD) notify = in6_rtchange; } errno = inet6ctlerrmap[cmd]; INP_INFO_WLOCK(pcbinfo); LIST_FOREACH_SAFE(inp, pcbinfo->ipi_listhead, inp_list, inp_temp) { INP_WLOCK(inp); if ((inp->inp_vflag & INP_IPV6) == 0) { INP_WUNLOCK(inp); continue; } /* * If the error designates a new path MTU for a destination * and the application (associated with this socket) wanted to - * know the value, notify. Note that we notify for all - * disconnected sockets if the corresponding application - * wanted. This is because some UDP applications keep sending - * sockets disconnected. + * know the value, notify. * XXX: should we avoid to notify the value to TCP sockets? */ - if (cmd == PRC_MSGSIZE && (inp->inp_flags & IN6P_MTU) != 0 && - (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || - IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &sa6_dst->sin6_addr))) { + if (cmd == PRC_MSGSIZE && cmdarg != NULL) ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst, - (u_int32_t *)cmdarg); - } + *(u_int32_t *)cmdarg); /* * Detect if we should notify the error. If no source and * destination ports are specifed, but non-zero flowinfo and * local address match, notify the error. This is the case * when the error is delivered with an encrypted buffer * by ESP. Otherwise, just compare addresses and ports * as usual. */ if (lport == 0 && fport == 0 && flowinfo && inp->inp_socket != NULL && flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) && IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr)) goto do_notify; else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &sa6_dst->sin6_addr) || inp->inp_socket == 0 || (lport && inp->inp_lport != lport) || (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) && !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr)) || (fport && inp->inp_fport != fport)) { INP_WUNLOCK(inp); continue; } do_notify: if (notify) { if ((*notify)(inp, errno)) INP_WUNLOCK(inp); } else INP_WUNLOCK(inp); } INP_INFO_WUNLOCK(pcbinfo); } /* * Lookup a PCB based on the local address and port. Caller must hold the * hash lock. No inpcb locks or references are acquired. */ struct inpcb * in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr, u_short lport, int lookupflags, struct ucred *cred) { register struct inpcb *inp; int matchwild = 3, wildcard; KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0, ("%s: invalid lookup flags %d", __func__, lookupflags)); INP_HASH_WLOCK_ASSERT(pcbinfo); if ((lookupflags & INPLOOKUP_WILDCARD) == 0) { struct inpcbhead *head; /* * Look for an unconnected (wildcard foreign addr) PCB that * matches the local address and port we're looking for. */ head = &pcbinfo->ipi_hashbase[INP_PCBHASH( INP6_PCBHASHKEY(&in6addr_any), lport, 0, pcbinfo->ipi_hashmask)]; LIST_FOREACH(inp, head, inp_hash) { /* XXX inp locking */ if ((inp->inp_vflag & INP_IPV6) == 0) continue; if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) && IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && inp->inp_lport == lport) { /* Found. */ if (cred == NULL || prison_equal_ip6(cred->cr_prison, inp->inp_cred->cr_prison)) return (inp); } } /* * Not found. */ return (NULL); } else { struct inpcbporthead *porthash; struct inpcbport *phd; struct inpcb *match = NULL; /* * Best fit PCB lookup. * * First see if this local port is in use by looking on the * port hash list. */ porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport, pcbinfo->ipi_porthashmask)]; LIST_FOREACH(phd, porthash, phd_hash) { if (phd->phd_port == lport) break; } if (phd != NULL) { /* * Port is in use by one or more PCBs. Look for best * fit. */ LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { wildcard = 0; if (cred != NULL && !prison_equal_ip6(cred->cr_prison, inp->inp_cred->cr_prison)) continue; /* XXX inp locking */ if ((inp->inp_vflag & INP_IPV6) == 0) continue; if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) wildcard++; if (!IN6_IS_ADDR_UNSPECIFIED( &inp->in6p_laddr)) { if (IN6_IS_ADDR_UNSPECIFIED(laddr)) wildcard++; else if (!IN6_ARE_ADDR_EQUAL( &inp->in6p_laddr, laddr)) continue; } else { if (!IN6_IS_ADDR_UNSPECIFIED(laddr)) wildcard++; } if (wildcard < matchwild) { match = inp; matchwild = wildcard; if (matchwild == 0) break; } } } return (match); } } void in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp) { struct inpcb *in6p; struct ip6_moptions *im6o; int i, gap; INP_INFO_RLOCK(pcbinfo); LIST_FOREACH(in6p, pcbinfo->ipi_listhead, inp_list) { INP_WLOCK(in6p); im6o = in6p->in6p_moptions; if ((in6p->inp_vflag & INP_IPV6) && im6o != NULL) { /* * Unselect the outgoing ifp for multicast if it * is being detached. */ if (im6o->im6o_multicast_ifp == ifp) im6o->im6o_multicast_ifp = NULL; /* * Drop multicast group membership if we joined * through the interface being detached. */ gap = 0; for (i = 0; i < im6o->im6o_num_memberships; i++) { if (im6o->im6o_membership[i]->in6m_ifp == ifp) { in6_mc_leave(im6o->im6o_membership[i], NULL); gap++; } else if (gap != 0) { im6o->im6o_membership[i - gap] = im6o->im6o_membership[i]; } } im6o->im6o_num_memberships -= gap; } INP_WUNLOCK(in6p); } INP_INFO_RUNLOCK(pcbinfo); } /* * Check for alternatives when higher level complains * about service problems. For now, invalidate cached * routing information. If the route was created dynamically * (by a redirect), time to try a default gateway again. */ void in6_losing(struct inpcb *in6p) { /* * We don't store route pointers in the routing table anymore */ return; } /* * After a routing change, flush old routing * and allocate a (hopefully) better one. */ struct inpcb * in6_rtchange(struct inpcb *inp, int errno) { /* * We don't store route pointers in the routing table anymore */ return inp; } #ifdef PCBGROUP /* * Lookup PCB in hash list, using pcbgroup tables. */ static struct inpcb * in6_pcblookup_group(struct inpcbinfo *pcbinfo, struct inpcbgroup *pcbgroup, struct in6_addr *faddr, u_int fport_arg, struct in6_addr *laddr, u_int lport_arg, int lookupflags, struct ifnet *ifp) { struct inpcbhead *head; struct inpcb *inp, *tmpinp; u_short fport = fport_arg, lport = lport_arg; /* * First look for an exact match. */ tmpinp = NULL; INP_GROUP_LOCK(pcbgroup); head = &pcbgroup->ipg_hashbase[INP_PCBHASH( INP6_PCBHASHKEY(faddr), lport, fport, pcbgroup->ipg_hashmask)]; LIST_FOREACH(inp, head, inp_pcbgrouphash) { /* XXX inp locking */ if ((inp->inp_vflag & INP_IPV6) == 0) continue; if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) && IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && inp->inp_fport == fport && inp->inp_lport == lport) { /* * XXX We should be able to directly return * the inp here, without any checks. * Well unless both bound with SO_REUSEPORT? */ if (prison_flag(inp->inp_cred, PR_IP6)) goto found; if (tmpinp == NULL) tmpinp = inp; } } if (tmpinp != NULL) { inp = tmpinp; goto found; } /* * Then look for a wildcard match in the pcbgroup. */ if ((lookupflags & INPLOOKUP_WILDCARD) != 0) { struct inpcb *local_wild = NULL, *local_exact = NULL; struct inpcb *jail_wild = NULL; int injail; /* * Order of socket selection - we always prefer jails. * 1. jailed, non-wild. * 2. jailed, wild. * 3. non-jailed, non-wild. * 4. non-jailed, wild. */ head = &pcbgroup->ipg_hashbase[ INP_PCBHASH(INADDR_ANY, lport, 0, pcbgroup->ipg_hashmask)]; LIST_FOREACH(inp, head, inp_pcbgrouphash) { /* XXX inp locking */ if ((inp->inp_vflag & INP_IPV6) == 0) continue; if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || inp->inp_lport != lport) { continue; } injail = prison_flag(inp->inp_cred, PR_IP6); if (injail) { if (prison_check_ip6(inp->inp_cred, laddr) != 0) continue; } else { if (local_exact != NULL) continue; } if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) { if (injail) goto found; else local_exact = inp; } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { if (injail) jail_wild = inp; else local_wild = inp; } } /* LIST_FOREACH */ inp = jail_wild; if (inp == NULL) inp = jail_wild; if (inp == NULL) inp = local_exact; if (inp == NULL) inp = local_wild; if (inp != NULL) goto found; } /* * Then look for a wildcard match, if requested. */ if ((lookupflags & INPLOOKUP_WILDCARD) != 0) { struct inpcb *local_wild = NULL, *local_exact = NULL; struct inpcb *jail_wild = NULL; int injail; /* * Order of socket selection - we always prefer jails. * 1. jailed, non-wild. * 2. jailed, wild. * 3. non-jailed, non-wild. * 4. non-jailed, wild. */ head = &pcbinfo->ipi_wildbase[INP_PCBHASH( INP6_PCBHASHKEY(&in6addr_any), lport, 0, pcbinfo->ipi_wildmask)]; LIST_FOREACH(inp, head, inp_pcbgroup_wild) { /* XXX inp locking */ if ((inp->inp_vflag & INP_IPV6) == 0) continue; if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || inp->inp_lport != lport) { continue; } injail = prison_flag(inp->inp_cred, PR_IP6); if (injail) { if (prison_check_ip6(inp->inp_cred, laddr) != 0) continue; } else { if (local_exact != NULL) continue; } if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) { if (injail) goto found; else local_exact = inp; } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { if (injail) jail_wild = inp; else local_wild = inp; } } /* LIST_FOREACH */ inp = jail_wild; if (inp == NULL) inp = jail_wild; if (inp == NULL) inp = local_exact; if (inp == NULL) inp = local_wild; if (inp != NULL) goto found; } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */ INP_GROUP_UNLOCK(pcbgroup); return (NULL); found: in_pcbref(inp); INP_GROUP_UNLOCK(pcbgroup); if (lookupflags & INPLOOKUP_WLOCKPCB) { INP_WLOCK(inp); if (in_pcbrele_wlocked(inp)) return (NULL); } else if (lookupflags & INPLOOKUP_RLOCKPCB) { INP_RLOCK(inp); if (in_pcbrele_rlocked(inp)) return (NULL); } else panic("%s: locking buf", __func__); return (inp); } #endif /* PCBGROUP */ /* * Lookup PCB in hash list. */ static struct inpcb * in6_pcblookup_hash_locked(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, u_int fport_arg, struct in6_addr *laddr, u_int lport_arg, int lookupflags, struct ifnet *ifp) { struct inpcbhead *head; struct inpcb *inp, *tmpinp; u_short fport = fport_arg, lport = lport_arg; KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0, ("%s: invalid lookup flags %d", __func__, lookupflags)); INP_HASH_LOCK_ASSERT(pcbinfo); /* * First look for an exact match. */ tmpinp = NULL; head = &pcbinfo->ipi_hashbase[INP_PCBHASH( INP6_PCBHASHKEY(faddr), lport, fport, pcbinfo->ipi_hashmask)]; LIST_FOREACH(inp, head, inp_hash) { /* XXX inp locking */ if ((inp->inp_vflag & INP_IPV6) == 0) continue; if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) && IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && inp->inp_fport == fport && inp->inp_lport == lport) { /* * XXX We should be able to directly return * the inp here, without any checks. * Well unless both bound with SO_REUSEPORT? */ if (prison_flag(inp->inp_cred, PR_IP6)) return (inp); if (tmpinp == NULL) tmpinp = inp; } } if (tmpinp != NULL) return (tmpinp); /* * Then look for a wildcard match, if requested. */ if ((lookupflags & INPLOOKUP_WILDCARD) != 0) { struct inpcb *local_wild = NULL, *local_exact = NULL; struct inpcb *jail_wild = NULL; int injail; /* * Order of socket selection - we always prefer jails. * 1. jailed, non-wild. * 2. jailed, wild. * 3. non-jailed, non-wild. * 4. non-jailed, wild. */ head = &pcbinfo->ipi_hashbase[INP_PCBHASH( INP6_PCBHASHKEY(&in6addr_any), lport, 0, pcbinfo->ipi_hashmask)]; LIST_FOREACH(inp, head, inp_hash) { /* XXX inp locking */ if ((inp->inp_vflag & INP_IPV6) == 0) continue; if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || inp->inp_lport != lport) { continue; } injail = prison_flag(inp->inp_cred, PR_IP6); if (injail) { if (prison_check_ip6(inp->inp_cred, laddr) != 0) continue; } else { if (local_exact != NULL) continue; } if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) { if (injail) return (inp); else local_exact = inp; } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { if (injail) jail_wild = inp; else local_wild = inp; } } /* LIST_FOREACH */ if (jail_wild != NULL) return (jail_wild); if (local_exact != NULL) return (local_exact); if (local_wild != NULL) return (local_wild); } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */ /* * Not found. */ return (NULL); } /* * Lookup PCB in hash list, using pcbinfo tables. This variation locks the * hash list lock, and will return the inpcb locked (i.e., requires * INPLOOKUP_LOCKPCB). */ static struct inpcb * in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags, struct ifnet *ifp) { struct inpcb *inp; INP_HASH_RLOCK(pcbinfo); inp = in6_pcblookup_hash_locked(pcbinfo, faddr, fport, laddr, lport, (lookupflags & ~(INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)), ifp); if (inp != NULL) { in_pcbref(inp); INP_HASH_RUNLOCK(pcbinfo); if (lookupflags & INPLOOKUP_WLOCKPCB) { INP_WLOCK(inp); if (in_pcbrele_wlocked(inp)) return (NULL); } else if (lookupflags & INPLOOKUP_RLOCKPCB) { INP_RLOCK(inp); if (in_pcbrele_rlocked(inp)) return (NULL); } else panic("%s: locking bug", __func__); } else INP_HASH_RUNLOCK(pcbinfo); return (inp); } /* * Public inpcb lookup routines, accepting a 4-tuple, and optionally, an mbuf * from which a pre-calculated hash value may be extracted. * * Possibly more of this logic should be in in6_pcbgroup.c. */ struct inpcb * in6_pcblookup(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags, struct ifnet *ifp) { #if defined(PCBGROUP) && !defined(RSS) struct inpcbgroup *pcbgroup; #endif KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0, ("%s: invalid lookup flags %d", __func__, lookupflags)); KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0, ("%s: LOCKPCB not set", __func__)); /* * When not using RSS, use connection groups in preference to the * reservation table when looking up 4-tuples. When using RSS, just * use the reservation table, due to the cost of the Toeplitz hash * in software. * * XXXRW: This policy belongs in the pcbgroup code, as in principle * we could be doing RSS with a non-Toeplitz hash that is affordable * in software. */ #if defined(PCBGROUP) && !defined(RSS) if (in_pcbgroup_enabled(pcbinfo)) { pcbgroup = in6_pcbgroup_bytuple(pcbinfo, laddr, lport, faddr, fport); return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr, fport, laddr, lport, lookupflags, ifp)); } #endif return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport, lookupflags, ifp)); } struct inpcb * in6_pcblookup_mbuf(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags, struct ifnet *ifp, struct mbuf *m) { #ifdef PCBGROUP struct inpcbgroup *pcbgroup; #endif KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0, ("%s: invalid lookup flags %d", __func__, lookupflags)); KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0, ("%s: LOCKPCB not set", __func__)); #ifdef PCBGROUP /* * If we can use a hardware-generated hash to look up the connection * group, use that connection group to find the inpcb. Otherwise * fall back on a software hash -- or the reservation table if we're * using RSS. * * XXXRW: As above, that policy belongs in the pcbgroup code. */ if (in_pcbgroup_enabled(pcbinfo) && M_HASHTYPE_TEST(m, M_HASHTYPE_NONE) == 0) { pcbgroup = in6_pcbgroup_byhash(pcbinfo, M_HASHTYPE_GET(m), m->m_pkthdr.flowid); if (pcbgroup != NULL) return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr, fport, laddr, lport, lookupflags, ifp)); #ifndef RSS pcbgroup = in6_pcbgroup_bytuple(pcbinfo, laddr, lport, faddr, fport); return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr, fport, laddr, lport, lookupflags, ifp)); #endif } #endif return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport, lookupflags, ifp)); } void init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m) { struct ip6_hdr *ip; ip = mtod(m, struct ip6_hdr *); bzero(sin6, sizeof(*sin6)); sin6->sin6_len = sizeof(*sin6); sin6->sin6_family = AF_INET6; sin6->sin6_addr = ip->ip6_src; (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */ return; } Index: projects/cxl_iscsi/sys/netinet6/ip6_input.c =================================================================== --- projects/cxl_iscsi/sys/netinet6/ip6_input.c (revision 279716) +++ projects/cxl_iscsi/sys/netinet6/ip6_input.c (revision 279717) @@ -1,1586 +1,1589 @@ /*- * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $ */ /*- * Copyright (c) 1982, 1986, 1988, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipfw.h" #include "opt_ipsec.h" #include "opt_route.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef INET #include #include #endif /* INET */ #include #include #include #include #include #include #include #include #ifdef IPSEC #include #include #include #endif /* IPSEC */ #include extern struct domain inet6domain; u_char ip6_protox[IPPROTO_MAX]; VNET_DEFINE(struct in6_ifaddrhead, in6_ifaddrhead); VNET_DEFINE(struct in6_ifaddrlisthead *, in6_ifaddrhashtbl); VNET_DEFINE(u_long, in6_ifaddrhmask); static struct netisr_handler ip6_nh = { .nh_name = "ip6", .nh_handler = ip6_input, .nh_proto = NETISR_IPV6, .nh_policy = NETISR_POLICY_FLOW, }; VNET_DECLARE(struct callout, in6_tmpaddrtimer_ch); #define V_in6_tmpaddrtimer_ch VNET(in6_tmpaddrtimer_ch) VNET_DEFINE(struct pfil_head, inet6_pfil_hook); VNET_PCPUSTAT_DEFINE(struct ip6stat, ip6stat); VNET_PCPUSTAT_SYSINIT(ip6stat); #ifdef VIMAGE VNET_PCPUSTAT_SYSUNINIT(ip6stat); #endif /* VIMAGE */ struct rwlock in6_ifaddr_lock; RW_SYSINIT(in6_ifaddr_lock, &in6_ifaddr_lock, "in6_ifaddr_lock"); static void ip6_init2(void *); static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *); #ifdef PULLDOWN_TEST static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int); #endif /* * IP6 initialization: fill in IP6 protocol switch table. * All protocols not implemented in kernel go to raw IP6 protocol handler. */ void ip6_init(void) { struct protosw *pr; int i; TUNABLE_INT_FETCH("net.inet6.ip6.auto_linklocal", &V_ip6_auto_linklocal); TUNABLE_INT_FETCH("net.inet6.ip6.accept_rtadv", &V_ip6_accept_rtadv); TUNABLE_INT_FETCH("net.inet6.ip6.no_radr", &V_ip6_no_radr); TAILQ_INIT(&V_in6_ifaddrhead); V_in6_ifaddrhashtbl = hashinit(IN6ADDR_NHASH, M_IFADDR, &V_in6_ifaddrhmask); /* Initialize packet filter hooks. */ V_inet6_pfil_hook.ph_type = PFIL_TYPE_AF; V_inet6_pfil_hook.ph_af = AF_INET6; if ((i = pfil_head_register(&V_inet6_pfil_hook)) != 0) printf("%s: WARNING: unable to register pfil hook, " "error %d\n", __func__, i); scope6_init(); addrsel_policy_init(); nd6_init(); frag6_init(); V_ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR; /* Skip global initialization stuff for non-default instances. */ if (!IS_DEFAULT_VNET(curvnet)) return; pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); if (pr == NULL) panic("ip6_init"); /* Initialize the entire ip6_protox[] array to IPPROTO_RAW. */ for (i = 0; i < IPPROTO_MAX; i++) ip6_protox[i] = pr - inet6sw; /* * Cycle through IP protocols and put them into the appropriate place * in ip6_protox[]. */ for (pr = inet6domain.dom_protosw; pr < inet6domain.dom_protoswNPROTOSW; pr++) if (pr->pr_domain->dom_family == PF_INET6 && pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) { /* Be careful to only index valid IP protocols. */ if (pr->pr_protocol < IPPROTO_MAX) ip6_protox[pr->pr_protocol] = pr - inet6sw; } netisr_register(&ip6_nh); } /* * The protocol to be inserted into ip6_protox[] must be already registered * in inet6sw[], either statically or through pf_proto_register(). */ int ip6proto_register(short ip6proto) { struct protosw *pr; /* Sanity checks. */ if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX) return (EPROTONOSUPPORT); /* * The protocol slot must not be occupied by another protocol * already. An index pointing to IPPROTO_RAW is unused. */ pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); if (pr == NULL) return (EPFNOSUPPORT); if (ip6_protox[ip6proto] != pr - inet6sw) /* IPPROTO_RAW */ return (EEXIST); /* * Find the protocol position in inet6sw[] and set the index. */ for (pr = inet6domain.dom_protosw; pr < inet6domain.dom_protoswNPROTOSW; pr++) { if (pr->pr_domain->dom_family == PF_INET6 && pr->pr_protocol && pr->pr_protocol == ip6proto) { ip6_protox[pr->pr_protocol] = pr - inet6sw; return (0); } } return (EPROTONOSUPPORT); } int ip6proto_unregister(short ip6proto) { struct protosw *pr; /* Sanity checks. */ if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX) return (EPROTONOSUPPORT); /* Check if the protocol was indeed registered. */ pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); if (pr == NULL) return (EPFNOSUPPORT); if (ip6_protox[ip6proto] == pr - inet6sw) /* IPPROTO_RAW */ return (ENOENT); /* Reset the protocol slot to IPPROTO_RAW. */ ip6_protox[ip6proto] = pr - inet6sw; return (0); } #ifdef VIMAGE void ip6_destroy() { int i; if ((i = pfil_head_unregister(&V_inet6_pfil_hook)) != 0) printf("%s: WARNING: unable to unregister pfil hook, " "error %d\n", __func__, i); hashdestroy(V_in6_ifaddrhashtbl, M_IFADDR, V_in6_ifaddrhmask); nd6_destroy(); callout_drain(&V_in6_tmpaddrtimer_ch); } #endif static int ip6_init2_vnet(const void *unused __unused) { /* nd6_timer_init */ callout_init(&V_nd6_timer_ch, 0); callout_reset(&V_nd6_timer_ch, hz, nd6_timer, curvnet); /* timer for regeneranation of temporary addresses randomize ID */ callout_init(&V_in6_tmpaddrtimer_ch, 0); callout_reset(&V_in6_tmpaddrtimer_ch, (V_ip6_temp_preferred_lifetime - V_ip6_desync_factor - V_ip6_temp_regen_advance) * hz, in6_tmpaddrtimer, curvnet); return (0); } static void ip6_init2(void *dummy) { ip6_init2_vnet(NULL); } /* cheat */ /* This must be after route_init(), which is now SI_ORDER_THIRD */ SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); static int ip6_input_hbh(struct mbuf *m, uint32_t *plen, uint32_t *rtalert, int *off, int *nxt, int *ours) { struct ip6_hdr *ip6; struct ip6_hbh *hbh; if (ip6_hopopts_input(plen, rtalert, &m, off)) { #if 0 /*touches NULL pointer*/ in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); #endif goto out; /* m have already been freed */ } /* adjust pointer */ ip6 = mtod(m, struct ip6_hdr *); /* * if the payload length field is 0 and the next header field * indicates Hop-by-Hop Options header, then a Jumbo Payload * option MUST be included. */ if (ip6->ip6_plen == 0 && *plen == 0) { /* * Note that if a valid jumbo payload option is * contained, ip6_hopopts_input() must set a valid * (non-zero) payload length to the variable plen. */ IP6STAT_INC(ip6s_badoptions); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); goto out; } #ifndef PULLDOWN_TEST /* ip6_hopopts_input() ensures that mbuf is contiguous */ hbh = (struct ip6_hbh *)(ip6 + 1); #else IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); if (hbh == NULL) { IP6STAT_INC(ip6s_tooshort); goto out; } #endif *nxt = hbh->ip6h_nxt; /* * If we are acting as a router and the packet contains a * router alert option, see if we know the option value. * Currently, we only support the option value for MLD, in which * case we should pass the packet to the multicast routing * daemon. */ if (*rtalert != ~0) { switch (*rtalert) { case IP6OPT_RTALERT_MLD: if (V_ip6_forwarding) *ours = 1; break; default: /* * RFC2711 requires unrecognized values must be * silently ignored. */ break; } } return (0); out: return (1); } void ip6_input(struct mbuf *m) { struct in6_addr odst; struct ip6_hdr *ip6; struct in6_ifaddr *ia; u_int32_t plen; u_int32_t rtalert = ~0; int off = sizeof(struct ip6_hdr), nest; int nxt, ours = 0; int srcrt = 0; #ifdef IPSEC /* * should the inner packet be considered authentic? * see comment in ah4_input(). * NB: m cannot be NULL when passed to the input routine */ m->m_flags &= ~M_AUTHIPHDR; m->m_flags &= ~M_AUTHIPDGM; #endif /* IPSEC */ if (m->m_flags & M_FASTFWD_OURS) { /* * Firewall changed destination to local. */ m->m_flags &= ~M_FASTFWD_OURS; ours = 1; ip6 = mtod(m, struct ip6_hdr *); goto hbhcheck; } /* * mbuf statistics */ if (m->m_flags & M_EXT) { if (m->m_next) IP6STAT_INC(ip6s_mext2m); else IP6STAT_INC(ip6s_mext1); } else { if (m->m_next) { if (m->m_flags & M_LOOP) { IP6STAT_INC(ip6s_m2m[V_loif->if_index]); } else if (m->m_pkthdr.rcvif->if_index < IP6S_M2MMAX) IP6STAT_INC( ip6s_m2m[m->m_pkthdr.rcvif->if_index]); else IP6STAT_INC(ip6s_m2m[0]); } else IP6STAT_INC(ip6s_m1); } /* drop the packet if IPv6 operation is disabled on the IF */ if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) goto bad; in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); IP6STAT_INC(ip6s_total); #ifndef PULLDOWN_TEST /* * L2 bridge code and some other code can return mbuf chain * that does not conform to KAME requirement. too bad. * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? */ if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { struct mbuf *n; if (m->m_pkthdr.len > MHLEN) n = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR); else n = m_gethdr(M_NOWAIT, MT_DATA); if (n == NULL) { m_freem(m); return; /* ENOBUFS */ } m_move_pkthdr(n, m); m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t)); n->m_len = n->m_pkthdr.len; m_freem(m); m = n; } IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */); #endif if (m->m_len < sizeof(struct ip6_hdr)) { struct ifnet *inifp; inifp = m->m_pkthdr.rcvif; if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { IP6STAT_INC(ip6s_toosmall); in6_ifstat_inc(inifp, ifs6_in_hdrerr); return; } } ip6 = mtod(m, struct ip6_hdr *); if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { IP6STAT_INC(ip6s_badvers); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); goto bad; } IP6STAT_INC(ip6s_nxthist[ip6->ip6_nxt]); IP_PROBE(receive, NULL, NULL, ip6, m->m_pkthdr.rcvif, NULL, ip6); /* * Check against address spoofing/corruption. */ if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { /* * XXX: "badscope" is not very suitable for a multicast source. */ IP6STAT_INC(ip6s_badscope); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); goto bad; } if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) && !(m->m_flags & M_LOOP)) { /* * In this case, the packet should come from the loopback * interface. However, we cannot just check the if_flags, * because ip6_mloopback() passes the "actual" interface * as the outgoing/incoming interface. */ IP6STAT_INC(ip6s_badscope); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); goto bad; } if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) && IPV6_ADDR_MC_SCOPE(&ip6->ip6_dst) == 0) { /* * RFC4291 2.7: * Nodes must not originate a packet to a multicast address * whose scop field contains the reserved value 0; if such * a packet is received, it must be silently dropped. */ IP6STAT_INC(ip6s_badscope); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); goto bad; } #ifdef ALTQ if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { /* packet is dropped by traffic conditioner */ return; } #endif /* * The following check is not documented in specs. A malicious * party may be able to use IPv4 mapped addr to confuse tcp/udp stack * and bypass security checks (act as if it was from 127.0.0.1 by using * IPv6 src ::ffff:127.0.0.1). Be cautious. * * This check chokes if we are in an SIIT cloud. As none of BSDs * support IPv4-less kernel compilation, we cannot support SIIT * environment at all. So, it makes more sense for us to reject any * malicious packets for non-SIIT environment, than try to do a * partial support for SIIT environment. */ if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { IP6STAT_INC(ip6s_badscope); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); goto bad; } #if 0 /* * Reject packets with IPv4 compatible addresses (auto tunnel). * * The code forbids auto tunnel relay case in RFC1933 (the check is * stronger than RFC1933). We may want to re-enable it if mech-xx * is revised to forbid relaying case. */ if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { IP6STAT_INC(ip6s_badscope); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); goto bad; } #endif #ifdef IPSEC /* * Bypass packet filtering for packets previously handled by IPsec. */ if (ip6_ipsec_filtertunnel(m)) goto passin; #endif /* IPSEC */ /* * Run through list of hooks for input packets. * * NB: Beware of the destination address changing * (e.g. by NAT rewriting). When this happens, * tell ip6_forward to do the right thing. */ odst = ip6->ip6_dst; /* Jump over all PFIL processing if hooks are not active. */ if (!PFIL_HOOKED(&V_inet6_pfil_hook)) goto passin; if (pfil_run_hooks(&V_inet6_pfil_hook, &m, m->m_pkthdr.rcvif, PFIL_IN, NULL)) return; if (m == NULL) /* consumed by filter */ return; ip6 = mtod(m, struct ip6_hdr *); srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); if (m->m_flags & M_FASTFWD_OURS) { m->m_flags &= ~M_FASTFWD_OURS; ours = 1; goto hbhcheck; } if ((m->m_flags & M_IP6_NEXTHOP) && m_tag_find(m, PACKET_TAG_IPFORWARD, NULL) != NULL) { /* * Directly ship the packet on. This allows forwarding * packets originally destined to us to some other directly * connected host. */ ip6_forward(m, 1); return; } passin: /* * Disambiguate address scope zones (if there is ambiguity). * We first make sure that the original source or destination address * is not in our internal form for scoped addresses. Such addresses * are not necessarily invalid spec-wise, but we cannot accept them due * to the usage conflict. * in6_setscope() then also checks and rejects the cases where src or * dst are the loopback address and the receiving interface * is not loopback. */ if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) { IP6STAT_INC(ip6s_badscope); /* XXX */ goto bad; } if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) || in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) { IP6STAT_INC(ip6s_badscope); goto bad; } /* * Multicast check. Assume packet is for us to avoid * prematurely taking locks. */ if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { ours = 1; in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); goto hbhcheck; } /* * Unicast check * XXX: For now we keep link-local IPv6 addresses with embedded * scope zone id, therefore we use zero zoneid here. */ ia = in6ifa_ifwithaddr(&ip6->ip6_dst, 0 /* XXX */); if (ia != NULL) { if (ia->ia6_flags & IN6_IFF_NOTREADY) { char ip6bufs[INET6_ADDRSTRLEN]; char ip6bufd[INET6_ADDRSTRLEN]; /* address is not ready, so discard the packet. */ nd6log((LOG_INFO, "ip6_input: packet to an unready address %s->%s\n", ip6_sprintf(ip6bufs, &ip6->ip6_src), ip6_sprintf(ip6bufd, &ip6->ip6_dst))); ifa_free(&ia->ia_ifa); goto bad; } /* Count the packet in the ip address stats */ counter_u64_add(ia->ia_ifa.ifa_ipackets, 1); counter_u64_add(ia->ia_ifa.ifa_ibytes, m->m_pkthdr.len); ifa_free(&ia->ia_ifa); ours = 1; goto hbhcheck; } /* * Now there is no reason to process the packet if it's not our own * and we're not a router. */ if (!V_ip6_forwarding) { IP6STAT_INC(ip6s_cantforward); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); goto bad; } hbhcheck: /* * Process Hop-by-Hop options header if it's contained. * m may be modified in ip6_hopopts_input(). * If a JumboPayload option is included, plen will also be modified. */ plen = (u_int32_t)ntohs(ip6->ip6_plen); if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { if (ip6_input_hbh(m, &plen, &rtalert, &off, &nxt, &ours) != 0) return; } else nxt = ip6->ip6_nxt; /* * Check that the amount of data in the buffers * is as at least much as the IPv6 header would have us expect. * Trim mbufs if longer than we expect. * Drop packet if shorter than we expect. */ if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { IP6STAT_INC(ip6s_tooshort); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); goto bad; } if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { if (m->m_len == m->m_pkthdr.len) { m->m_len = sizeof(struct ip6_hdr) + plen; m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; } else m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); } /* * Forward if desirable. */ if (V_ip6_mrouter && IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { /* * If we are acting as a multicast router, all * incoming multicast packets are passed to the * kernel-level multicast forwarding function. * The packet is returned (relatively) intact; if * ip6_mforward() returns a non-zero value, the packet * must be discarded, else it may be accepted below. * * XXX TODO: Check hlim and multicast scope here to avoid * unnecessarily calling into ip6_mforward(). */ if (ip6_mforward && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { IP6STAT_INC(ip6s_cantforward); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); goto bad; } } else if (!ours) { ip6_forward(m, srcrt); return; } ip6 = mtod(m, struct ip6_hdr *); /* * Malicious party may be able to use IPv4 mapped addr to confuse * tcp/udp stack and bypass security checks (act as if it was from * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. * * For SIIT end node behavior, you may want to disable the check. * However, you will become vulnerable to attacks using IPv4 mapped * source. */ if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { IP6STAT_INC(ip6s_badscope); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); goto bad; } /* * Tell launch routine the next header */ IP6STAT_INC(ip6s_delivered); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_deliver); nest = 0; while (nxt != IPPROTO_DONE) { if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) { IP6STAT_INC(ip6s_toomanyhdr); goto bad; } /* * protection against faulty packet - there should be * more sanity checks in header chain processing. */ if (m->m_pkthdr.len < off) { IP6STAT_INC(ip6s_tooshort); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); goto bad; } #ifdef IPSEC /* * enforce IPsec policy checking if we are seeing last header. * note that we do not visit this with protocols with pcb layer * code - like udp/tcp/raw ip. */ if (ip6_ipsec_input(m, nxt)) goto bad; #endif /* IPSEC */ /* * Use mbuf flags to propagate Router Alert option to * ICMPv6 layer, as hop-by-hop options have been stripped. */ if (nxt == IPPROTO_ICMPV6 && rtalert != ~0) m->m_flags |= M_RTALERT_MLD; nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); } return; bad: m_freem(m); } /* * Hop-by-Hop options header processing. If a valid jumbo payload option is * included, the real payload length will be stored in plenp. * * rtalertp - XXX: should be stored more smart way */ static int ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp, struct mbuf **mp, int *offp) { struct mbuf *m = *mp; int off = *offp, hbhlen; struct ip6_hbh *hbh; /* validation of the length of the header */ #ifndef PULLDOWN_TEST IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); hbhlen = (hbh->ip6h_len + 1) << 3; IP6_EXTHDR_CHECK(m, off, hbhlen, -1); hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); #else IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); if (hbh == NULL) { IP6STAT_INC(ip6s_tooshort); return -1; } hbhlen = (hbh->ip6h_len + 1) << 3; IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), hbhlen); if (hbh == NULL) { IP6STAT_INC(ip6s_tooshort); return -1; } #endif off += hbhlen; hbhlen -= sizeof(struct ip6_hbh); if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), hbhlen, rtalertp, plenp) < 0) return (-1); *offp = off; *mp = m; return (0); } /* * Search header for all Hop-by-hop options and process each option. * This function is separate from ip6_hopopts_input() in order to * handle a case where the sending node itself process its hop-by-hop * options header. In such a case, the function is called from ip6_output(). * * The function assumes that hbh header is located right after the IPv6 header * (RFC2460 p7), opthead is pointer into data content in m, and opthead to * opthead + hbhlen is located in contiguous memory region. */ int ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, u_int32_t *rtalertp, u_int32_t *plenp) { struct ip6_hdr *ip6; int optlen = 0; u_int8_t *opt = opthead; u_int16_t rtalert_val; u_int32_t jumboplen; const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { switch (*opt) { case IP6OPT_PAD1: optlen = 1; break; case IP6OPT_PADN: if (hbhlen < IP6OPT_MINLEN) { IP6STAT_INC(ip6s_toosmall); goto bad; } optlen = *(opt + 1) + 2; break; case IP6OPT_ROUTER_ALERT: /* XXX may need check for alignment */ if (hbhlen < IP6OPT_RTALERT_LEN) { IP6STAT_INC(ip6s_toosmall); goto bad; } if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { /* XXX stat */ icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, erroff + opt + 1 - opthead); return (-1); } optlen = IP6OPT_RTALERT_LEN; bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); *rtalertp = ntohs(rtalert_val); break; case IP6OPT_JUMBO: /* XXX may need check for alignment */ if (hbhlen < IP6OPT_JUMBO_LEN) { IP6STAT_INC(ip6s_toosmall); goto bad; } if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { /* XXX stat */ icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, erroff + opt + 1 - opthead); return (-1); } optlen = IP6OPT_JUMBO_LEN; /* * IPv6 packets that have non 0 payload length * must not contain a jumbo payload option. */ ip6 = mtod(m, struct ip6_hdr *); if (ip6->ip6_plen) { IP6STAT_INC(ip6s_badoptions); icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, erroff + opt - opthead); return (-1); } /* * We may see jumbolen in unaligned location, so * we'd need to perform bcopy(). */ bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); jumboplen = (u_int32_t)htonl(jumboplen); #if 1 /* * if there are multiple jumbo payload options, * *plenp will be non-zero and the packet will be * rejected. * the behavior may need some debate in ipngwg - * multiple options does not make sense, however, * there's no explicit mention in specification. */ if (*plenp != 0) { IP6STAT_INC(ip6s_badoptions); icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, erroff + opt + 2 - opthead); return (-1); } #endif /* * jumbo payload length must be larger than 65535. */ if (jumboplen <= IPV6_MAXPACKET) { IP6STAT_INC(ip6s_badoptions); icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, erroff + opt + 2 - opthead); return (-1); } *plenp = jumboplen; break; default: /* unknown option */ if (hbhlen < IP6OPT_MINLEN) { IP6STAT_INC(ip6s_toosmall); goto bad; } optlen = ip6_unknown_opt(opt, m, erroff + opt - opthead); if (optlen == -1) return (-1); optlen += 2; break; } } return (0); bad: m_freem(m); return (-1); } /* * Unknown option processing. * The third argument `off' is the offset from the IPv6 header to the option, * which is necessary if the IPv6 header the and option header and IPv6 header * is not contiguous in order to return an ICMPv6 error. */ int ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off) { struct ip6_hdr *ip6; switch (IP6OPT_TYPE(*optp)) { case IP6OPT_TYPE_SKIP: /* ignore the option */ return ((int)*(optp + 1)); case IP6OPT_TYPE_DISCARD: /* silently discard */ m_freem(m); return (-1); case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ IP6STAT_INC(ip6s_badoptions); icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); return (-1); case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ IP6STAT_INC(ip6s_badoptions); ip6 = mtod(m, struct ip6_hdr *); if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || (m->m_flags & (M_BCAST|M_MCAST))) m_freem(m); else icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); return (-1); } m_freem(m); /* XXX: NOTREACHED */ return (-1); } /* * Create the "control" list for this pcb. * These functions will not modify mbuf chain at all. * * With KAME mbuf chain restriction: * The routine will be called from upper layer handlers like tcp6_input(). * Thus the routine assumes that the caller (tcp6_input) have already * called IP6_EXTHDR_CHECK() and all the extension headers are located in the * very first mbuf on the mbuf chain. * * ip6_savecontrol_v4 will handle those options that are possible to be * set on a v4-mapped socket. * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those * options and handle the v6-only ones itself. */ struct mbuf ** ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp, int *v4only) { struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); #ifdef SO_TIMESTAMP if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) { struct timeval tv; microtime(&tv); *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), SCM_TIMESTAMP, SOL_SOCKET); if (*mp) mp = &(*mp)->m_next; } #endif #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y)) /* RFC 2292 sec. 5 */ if ((inp->inp_flags & IN6P_PKTINFO) != 0) { struct in6_pktinfo pi6; if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { #ifdef INET struct ip *ip; ip = mtod(m, struct ip *); pi6.ipi6_addr.s6_addr32[0] = 0; pi6.ipi6_addr.s6_addr32[1] = 0; pi6.ipi6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP; pi6.ipi6_addr.s6_addr32[3] = ip->ip_dst.s_addr; #else /* We won't hit this code */ bzero(&pi6.ipi6_addr, sizeof(struct in6_addr)); #endif } else { bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); in6_clearscope(&pi6.ipi6_addr); /* XXX */ } pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0; *mp = sbcreatecontrol((caddr_t) &pi6, sizeof(struct in6_pktinfo), IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; } if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) { int hlim; if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { #ifdef INET struct ip *ip; ip = mtod(m, struct ip *); hlim = ip->ip_ttl; #else /* We won't hit this code */ hlim = 0; #endif } else { hlim = ip6->ip6_hlim & 0xff; } *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int), IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; } if ((inp->inp_flags & IN6P_TCLASS) != 0) { int tclass; if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { #ifdef INET struct ip *ip; ip = mtod(m, struct ip *); tclass = ip->ip_tos; #else /* We won't hit this code */ tclass = 0; #endif } else { u_int32_t flowinfo; flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK); flowinfo >>= 20; tclass = flowinfo & 0xff; } *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(int), IPV6_TCLASS, IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; } if (v4only != NULL) { if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { *v4only = 1; } else { *v4only = 0; } } return (mp); } void ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp) { struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); int v4only = 0; mp = ip6_savecontrol_v4(in6p, m, mp, &v4only); if (v4only) return; /* * IPV6_HOPOPTS socket option. Recall that we required super-user * privilege for the option (see ip6_ctloutput), but it might be too * strict, since there might be some hop-by-hop options which can be * returned to normal user. * See also RFC 2292 section 6 (or RFC 3542 section 8). */ if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) { /* * Check if a hop-by-hop options header is contatined in the * received packet, and if so, store the options as ancillary * data. Note that a hop-by-hop options header must be * just after the IPv6 header, which is assured through the * IPv6 input processing. */ if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { struct ip6_hbh *hbh; int hbhlen = 0; #ifdef PULLDOWN_TEST struct mbuf *ext; #endif #ifndef PULLDOWN_TEST hbh = (struct ip6_hbh *)(ip6 + 1); hbhlen = (hbh->ip6h_len + 1) << 3; #else ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), ip6->ip6_nxt); if (ext == NULL) { IP6STAT_INC(ip6s_tooshort); return; } hbh = mtod(ext, struct ip6_hbh *); hbhlen = (hbh->ip6h_len + 1) << 3; if (hbhlen != ext->m_len) { m_freem(ext); IP6STAT_INC(ip6s_tooshort); return; } #endif /* * XXX: We copy the whole header even if a * jumbo payload option is included, the option which * is to be removed before returning according to * RFC2292. * Note: this constraint is removed in RFC3542 */ *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; #ifdef PULLDOWN_TEST m_freem(ext); #endif } } if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) { int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); /* * Search for destination options headers or routing * header(s) through the header chain, and stores each * header as ancillary data. * Note that the order of the headers remains in * the chain of ancillary data. */ while (1) { /* is explicit loop prevention necessary? */ struct ip6_ext *ip6e = NULL; int elen; #ifdef PULLDOWN_TEST struct mbuf *ext = NULL; #endif /* * if it is not an extension header, don't try to * pull it from the chain. */ switch (nxt) { case IPPROTO_DSTOPTS: case IPPROTO_ROUTING: case IPPROTO_HOPOPTS: case IPPROTO_AH: /* is it possible? */ break; default: goto loopend; } #ifndef PULLDOWN_TEST if (off + sizeof(*ip6e) > m->m_len) goto loopend; ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); if (nxt == IPPROTO_AH) elen = (ip6e->ip6e_len + 2) << 2; else elen = (ip6e->ip6e_len + 1) << 3; if (off + elen > m->m_len) goto loopend; #else ext = ip6_pullexthdr(m, off, nxt); if (ext == NULL) { IP6STAT_INC(ip6s_tooshort); return; } ip6e = mtod(ext, struct ip6_ext *); if (nxt == IPPROTO_AH) elen = (ip6e->ip6e_len + 2) << 2; else elen = (ip6e->ip6e_len + 1) << 3; if (elen != ext->m_len) { m_freem(ext); IP6STAT_INC(ip6s_tooshort); return; } #endif switch (nxt) { case IPPROTO_DSTOPTS: if (!(in6p->inp_flags & IN6P_DSTOPTS)) break; *mp = sbcreatecontrol((caddr_t)ip6e, elen, IS2292(in6p, IPV6_2292DSTOPTS, IPV6_DSTOPTS), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; break; case IPPROTO_ROUTING: if (!(in6p->inp_flags & IN6P_RTHDR)) break; *mp = sbcreatecontrol((caddr_t)ip6e, elen, IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; break; case IPPROTO_HOPOPTS: case IPPROTO_AH: /* is it possible? */ break; default: /* * other cases have been filtered in the above. * none will visit this case. here we supply * the code just in case (nxt overwritten or * other cases). */ #ifdef PULLDOWN_TEST m_freem(ext); #endif goto loopend; } /* proceed with the next header. */ off += elen; nxt = ip6e->ip6e_nxt; ip6e = NULL; #ifdef PULLDOWN_TEST m_freem(ext); ext = NULL; #endif } loopend: ; } } #undef IS2292 void -ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu) +ip6_notify_pmtu(struct inpcb *inp, struct sockaddr_in6 *dst, u_int32_t mtu) { struct socket *so; struct mbuf *m_mtu; struct ip6_mtuinfo mtuctl; - so = in6p->inp_socket; - - if (mtu == NULL) + KASSERT(inp != NULL, ("%s: inp == NULL", __func__)); + /* + * Notify the error by sending IPV6_PATHMTU ancillary data if + * application wanted to know the MTU value. + * NOTE: we notify disconnected sockets, because some udp + * applications keep sending sockets disconnected. + * NOTE: our implementation doesn't notify connected sockets that has + * foreign address that is different than given destination addresses + * (this is permitted by RFC 3542). + */ + if ((inp->inp_flags & IN6P_MTU) == 0 || ( + !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) && + !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &dst->sin6_addr))) return; -#ifdef DIAGNOSTIC - if (so == NULL) /* I believe this is impossible */ - panic("ip6_notify_pmtu: socket is NULL"); -#endif - - bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */ - mtuctl.ip6m_mtu = *mtu; + mtuctl.ip6m_mtu = mtu; mtuctl.ip6m_addr = *dst; if (sa6_recoverscope(&mtuctl.ip6m_addr)) return; if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl), IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) return; + so = inp->inp_socket; if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu) == 0) { m_freem(m_mtu); /* XXX: should count statistics */ } else sorwakeup(so); - - return; } #ifdef PULLDOWN_TEST /* * pull single extension header from mbuf chain. returns single mbuf that * contains the result, or NULL on error. */ static struct mbuf * ip6_pullexthdr(struct mbuf *m, size_t off, int nxt) { struct ip6_ext ip6e; size_t elen; struct mbuf *n; #ifdef DIAGNOSTIC switch (nxt) { case IPPROTO_DSTOPTS: case IPPROTO_ROUTING: case IPPROTO_HOPOPTS: case IPPROTO_AH: /* is it possible? */ break; default: printf("ip6_pullexthdr: invalid nxt=%d\n", nxt); } #endif m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); if (nxt == IPPROTO_AH) elen = (ip6e.ip6e_len + 2) << 2; else elen = (ip6e.ip6e_len + 1) << 3; if (elen > MLEN) n = m_getcl(M_NOWAIT, MT_DATA, 0); else n = m_get(M_NOWAIT, MT_DATA); if (n == NULL) return NULL; m_copydata(m, off, elen, mtod(n, caddr_t)); n->m_len = elen; return n; } #endif /* * Get pointer to the previous header followed by the header * currently processed. * XXX: This function supposes that * M includes all headers, * the next header field and the header length field of each header * are valid, and * the sum of each header length equals to OFF. * Because of these assumptions, this function must be called very * carefully. Moreover, it will not be used in the near future when * we develop `neater' mechanism to process extension headers. */ char * ip6_get_prevhdr(struct mbuf *m, int off) { struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); if (off == sizeof(struct ip6_hdr)) return (&ip6->ip6_nxt); else { int len, nxt; struct ip6_ext *ip6e = NULL; nxt = ip6->ip6_nxt; len = sizeof(struct ip6_hdr); while (len < off) { ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); switch (nxt) { case IPPROTO_FRAGMENT: len += sizeof(struct ip6_frag); break; case IPPROTO_AH: len += (ip6e->ip6e_len + 2) << 2; break; default: len += (ip6e->ip6e_len + 1) << 3; break; } nxt = ip6e->ip6e_nxt; } if (ip6e) return (&ip6e->ip6e_nxt); else return NULL; } } /* * get next header offset. m will be retained. */ int ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp) { struct ip6_hdr ip6; struct ip6_ext ip6e; struct ip6_frag fh; /* just in case */ if (m == NULL) panic("ip6_nexthdr: m == NULL"); if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) return -1; switch (proto) { case IPPROTO_IPV6: if (m->m_pkthdr.len < off + sizeof(ip6)) return -1; m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); if (nxtp) *nxtp = ip6.ip6_nxt; off += sizeof(ip6); return off; case IPPROTO_FRAGMENT: /* * terminate parsing if it is not the first fragment, * it does not make sense to parse through it. */ if (m->m_pkthdr.len < off + sizeof(fh)) return -1; m_copydata(m, off, sizeof(fh), (caddr_t)&fh); /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */ if (fh.ip6f_offlg & IP6F_OFF_MASK) return -1; if (nxtp) *nxtp = fh.ip6f_nxt; off += sizeof(struct ip6_frag); return off; case IPPROTO_AH: if (m->m_pkthdr.len < off + sizeof(ip6e)) return -1; m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); if (nxtp) *nxtp = ip6e.ip6e_nxt; off += (ip6e.ip6e_len + 2) << 2; return off; case IPPROTO_HOPOPTS: case IPPROTO_ROUTING: case IPPROTO_DSTOPTS: if (m->m_pkthdr.len < off + sizeof(ip6e)) return -1; m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); if (nxtp) *nxtp = ip6e.ip6e_nxt; off += (ip6e.ip6e_len + 1) << 3; return off; case IPPROTO_NONE: case IPPROTO_ESP: case IPPROTO_IPCOMP: /* give up */ return -1; default: return -1; } return -1; } /* * get offset for the last header in the chain. m will be kept untainted. */ int ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp) { int newoff; int nxt; if (!nxtp) { nxt = -1; nxtp = &nxt; } while (1) { newoff = ip6_nexthdr(m, off, proto, nxtp); if (newoff < 0) return off; else if (newoff < off) return -1; /* invalid */ else if (newoff == off) return newoff; off = newoff; proto = *nxtp; } } /* * System control for IP6 */ u_char inet6ctlerrmap[PRC_NCMDS] = { 0, 0, 0, 0, 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, EMSGSIZE, EHOSTUNREACH, 0, 0, 0, 0, 0, 0, ENOPROTOOPT }; Index: projects/cxl_iscsi/sys/netinet6/ip6_output.c =================================================================== --- projects/cxl_iscsi/sys/netinet6/ip6_output.c (revision 279716) +++ projects/cxl_iscsi/sys/netinet6/ip6_output.c (revision 279717) @@ -1,2980 +1,2973 @@ /*- * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: ip6_output.c,v 1.279 2002/01/26 06:12:30 jinmei Exp $ */ /*- * Copyright (c) 1982, 1986, 1988, 1990, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)ip_output.c 8.3 (Berkeley) 1/21/94 */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipfw.h" #include "opt_ipsec.h" #include "opt_sctp.h" #include "opt_route.h" #include "opt_rss.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef IPSEC #include #include #include #include #endif /* IPSEC */ #ifdef SCTP #include #include #endif #include #include #ifdef FLOWTABLE #include #endif extern int in6_mcast_loop; struct ip6_exthdrs { struct mbuf *ip6e_ip6; struct mbuf *ip6e_hbh; struct mbuf *ip6e_dest1; struct mbuf *ip6e_rthdr; struct mbuf *ip6e_dest2; }; static int ip6_pcbopt(int, u_char *, int, struct ip6_pktopts **, struct ucred *, int); static int ip6_pcbopts(struct ip6_pktopts **, struct mbuf *, struct socket *, struct sockopt *); static int ip6_getpcbopt(struct ip6_pktopts *, int, struct sockopt *); static int ip6_setpktopt(int, u_char *, int, struct ip6_pktopts *, struct ucred *, int, int, int); static int ip6_copyexthdr(struct mbuf **, caddr_t, int); static int ip6_insertfraghdr(struct mbuf *, struct mbuf *, int, struct ip6_frag **); static int ip6_insert_jumboopt(struct ip6_exthdrs *, u_int32_t); static int ip6_splithdr(struct mbuf *, struct ip6_exthdrs *); static int ip6_getpmtu(struct route_in6 *, struct route_in6 *, struct ifnet *, struct in6_addr *, u_long *, int *, u_int); static int copypktopts(struct ip6_pktopts *, struct ip6_pktopts *, int); /* * Make an extension header from option data. hp is the source, and * mp is the destination. */ #define MAKE_EXTHDR(hp, mp) \ do { \ if (hp) { \ struct ip6_ext *eh = (struct ip6_ext *)(hp); \ error = ip6_copyexthdr((mp), (caddr_t)(hp), \ ((eh)->ip6e_len + 1) << 3); \ if (error) \ goto freehdrs; \ } \ } while (/*CONSTCOND*/ 0) /* * Form a chain of extension headers. * m is the extension header mbuf * mp is the previous mbuf in the chain * p is the next header * i is the type of option. */ #define MAKE_CHAIN(m, mp, p, i)\ do {\ if (m) {\ if (!hdrsplit) \ panic("assumption failed: hdr not split"); \ *mtod((m), u_char *) = *(p);\ *(p) = (i);\ p = mtod((m), u_char *);\ (m)->m_next = (mp)->m_next;\ (mp)->m_next = (m);\ (mp) = (m);\ }\ } while (/*CONSTCOND*/ 0) void in6_delayed_cksum(struct mbuf *m, uint32_t plen, u_short offset) { u_short csum; csum = in_cksum_skip(m, offset + plen, offset); if (m->m_pkthdr.csum_flags & CSUM_UDP_IPV6 && csum == 0) csum = 0xffff; offset += m->m_pkthdr.csum_data; /* checksum offset */ if (offset + sizeof(u_short) > m->m_len) { printf("%s: delayed m_pullup, m->len: %d plen %u off %u " "csum_flags=%b\n", __func__, m->m_len, plen, offset, (int)m->m_pkthdr.csum_flags, CSUM_BITS); /* * XXX this should not happen, but if it does, the correct * behavior may be to insert the checksum in the appropriate * next mbuf in the chain. */ return; } *(u_short *)(m->m_data + offset) = csum; } int ip6_fragment(struct ifnet *ifp, struct mbuf *m0, int hlen, u_char nextproto, int mtu) { struct mbuf *m, **mnext, *m_frgpart; struct ip6_hdr *ip6, *mhip6; struct ip6_frag *ip6f; int off; int error; int tlen = m0->m_pkthdr.len; uint32_t id = htonl(ip6_randomid()); m = m0; ip6 = mtod(m, struct ip6_hdr *); mnext = &m->m_nextpkt; for (off = hlen; off < tlen; off += mtu) { m = m_gethdr(M_NOWAIT, MT_DATA); if (!m) { IP6STAT_INC(ip6s_odropped); return (ENOBUFS); } m->m_flags = m0->m_flags & M_COPYFLAGS; *mnext = m; mnext = &m->m_nextpkt; m->m_data += max_linkhdr; mhip6 = mtod(m, struct ip6_hdr *); *mhip6 = *ip6; m->m_len = sizeof(*mhip6); error = ip6_insertfraghdr(m0, m, hlen, &ip6f); if (error) { IP6STAT_INC(ip6s_odropped); return (error); } ip6f->ip6f_offlg = htons((u_short)((off - hlen) & ~7)); if (off + mtu >= tlen) mtu = tlen - off; else ip6f->ip6f_offlg |= IP6F_MORE_FRAG; mhip6->ip6_plen = htons((u_short)(mtu + hlen + sizeof(*ip6f) - sizeof(struct ip6_hdr))); if ((m_frgpart = m_copy(m0, off, mtu)) == 0) { IP6STAT_INC(ip6s_odropped); return (ENOBUFS); } m_cat(m, m_frgpart); m->m_pkthdr.len = mtu + hlen + sizeof(*ip6f); m->m_pkthdr.fibnum = m0->m_pkthdr.fibnum; m->m_pkthdr.rcvif = NULL; ip6f->ip6f_reserved = 0; ip6f->ip6f_ident = id; ip6f->ip6f_nxt = nextproto; IP6STAT_INC(ip6s_ofragments); in6_ifstat_inc(ifp, ifs6_out_fragcreat); } return (0); } /* * IP6 output. The packet in mbuf chain m contains a skeletal IP6 * header (with pri, len, nxt, hlim, src, dst). * This function may modify ver and hlim only. * The mbuf chain containing the packet will be freed. * The mbuf opt, if present, will not be freed. * If route_in6 ro is present and has ro_rt initialized, route lookup would be * skipped and ro->ro_rt would be used. If ro is present but ro->ro_rt is NULL, * then result of route lookup is stored in ro->ro_rt. * * type of "mtu": rt_mtu is u_long, ifnet.ifr_mtu is int, and * nd_ifinfo.linkmtu is u_int32_t. so we use u_long to hold largest one, * which is rt_mtu. * * ifpp - XXX: just for statistics */ /* * XXX TODO: no flowid is assigned for outbound flows? */ int ip6_output(struct mbuf *m0, struct ip6_pktopts *opt, struct route_in6 *ro, int flags, struct ip6_moptions *im6o, struct ifnet **ifpp, struct inpcb *inp) { struct ip6_hdr *ip6; struct ifnet *ifp, *origifp; struct mbuf *m = m0; struct mbuf *mprev = NULL; int hlen, tlen, len; struct route_in6 ip6route; struct rtentry *rt = NULL; struct sockaddr_in6 *dst, src_sa, dst_sa; struct in6_addr odst; int error = 0; struct in6_ifaddr *ia = NULL; u_long mtu; int alwaysfrag, dontfrag; u_int32_t optlen = 0, plen = 0, unfragpartlen = 0; struct ip6_exthdrs exthdrs; struct in6_addr finaldst, src0, dst0; u_int32_t zone; struct route_in6 *ro_pmtu = NULL; int hdrsplit = 0; int sw_csum, tso; int needfiblookup; uint32_t fibnum; struct m_tag *fwd_tag = NULL; ip6 = mtod(m, struct ip6_hdr *); if (ip6 == NULL) { printf ("ip6 is NULL"); goto bad; } if (inp != NULL) { M_SETFIB(m, inp->inp_inc.inc_fibnum); if ((flags & IP_NODEFAULTFLOWID) == 0) { /* unconditionally set flowid */ m->m_pkthdr.flowid = inp->inp_flowid; M_HASHTYPE_SET(m, inp->inp_flowtype); } } finaldst = ip6->ip6_dst; bzero(&exthdrs, sizeof(exthdrs)); if (opt) { /* Hop-by-Hop options header */ MAKE_EXTHDR(opt->ip6po_hbh, &exthdrs.ip6e_hbh); /* Destination options header(1st part) */ if (opt->ip6po_rthdr) { /* * Destination options header(1st part) * This only makes sense with a routing header. * See Section 9.2 of RFC 3542. * Disabling this part just for MIP6 convenience is * a bad idea. We need to think carefully about a * way to make the advanced API coexist with MIP6 * options, which might automatically be inserted in * the kernel. */ MAKE_EXTHDR(opt->ip6po_dest1, &exthdrs.ip6e_dest1); } /* Routing header */ MAKE_EXTHDR(opt->ip6po_rthdr, &exthdrs.ip6e_rthdr); /* Destination options header(2nd part) */ MAKE_EXTHDR(opt->ip6po_dest2, &exthdrs.ip6e_dest2); } #ifdef IPSEC /* * IPSec checking which handles several cases. * FAST IPSEC: We re-injected the packet. * XXX: need scope argument. */ switch(ip6_ipsec_output(&m, inp, &error)) { case 1: /* Bad packet */ goto freehdrs; case -1: /* IPSec done */ goto done; case 0: /* No IPSec */ default: break; } #endif /* IPSEC */ /* * Calculate the total length of the extension header chain. * Keep the length of the unfragmentable part for fragmentation. */ optlen = 0; if (exthdrs.ip6e_hbh) optlen += exthdrs.ip6e_hbh->m_len; if (exthdrs.ip6e_dest1) optlen += exthdrs.ip6e_dest1->m_len; if (exthdrs.ip6e_rthdr) optlen += exthdrs.ip6e_rthdr->m_len; unfragpartlen = optlen + sizeof(struct ip6_hdr); /* NOTE: we don't add AH/ESP length here (done in ip6_ipsec_output) */ if (exthdrs.ip6e_dest2) optlen += exthdrs.ip6e_dest2->m_len; /* * If there is at least one extension header, * separate IP6 header from the payload. */ if (optlen && !hdrsplit) { if ((error = ip6_splithdr(m, &exthdrs)) != 0) { m = NULL; goto freehdrs; } m = exthdrs.ip6e_ip6; hdrsplit++; } /* adjust pointer */ ip6 = mtod(m, struct ip6_hdr *); /* adjust mbuf packet header length */ m->m_pkthdr.len += optlen; plen = m->m_pkthdr.len - sizeof(*ip6); /* If this is a jumbo payload, insert a jumbo payload option. */ if (plen > IPV6_MAXPACKET) { if (!hdrsplit) { if ((error = ip6_splithdr(m, &exthdrs)) != 0) { m = NULL; goto freehdrs; } m = exthdrs.ip6e_ip6; hdrsplit++; } /* adjust pointer */ ip6 = mtod(m, struct ip6_hdr *); if ((error = ip6_insert_jumboopt(&exthdrs, plen)) != 0) goto freehdrs; ip6->ip6_plen = 0; } else ip6->ip6_plen = htons(plen); /* * Concatenate headers and fill in next header fields. * Here we have, on "m" * IPv6 payload * and we insert headers accordingly. Finally, we should be getting: * IPv6 hbh dest1 rthdr ah* [esp* dest2 payload] * * during the header composing process, "m" points to IPv6 header. * "mprev" points to an extension header prior to esp. */ u_char *nexthdrp = &ip6->ip6_nxt; mprev = m; /* * we treat dest2 specially. this makes IPsec processing * much easier. the goal here is to make mprev point the * mbuf prior to dest2. * * result: IPv6 dest2 payload * m and mprev will point to IPv6 header. */ if (exthdrs.ip6e_dest2) { if (!hdrsplit) panic("assumption failed: hdr not split"); exthdrs.ip6e_dest2->m_next = m->m_next; m->m_next = exthdrs.ip6e_dest2; *mtod(exthdrs.ip6e_dest2, u_char *) = ip6->ip6_nxt; ip6->ip6_nxt = IPPROTO_DSTOPTS; } /* * result: IPv6 hbh dest1 rthdr dest2 payload * m will point to IPv6 header. mprev will point to the * extension header prior to dest2 (rthdr in the above case). */ MAKE_CHAIN(exthdrs.ip6e_hbh, mprev, nexthdrp, IPPROTO_HOPOPTS); MAKE_CHAIN(exthdrs.ip6e_dest1, mprev, nexthdrp, IPPROTO_DSTOPTS); MAKE_CHAIN(exthdrs.ip6e_rthdr, mprev, nexthdrp, IPPROTO_ROUTING); /* * If there is a routing header, discard the packet. */ if (exthdrs.ip6e_rthdr) { error = EINVAL; goto bad; } /* Source address validation */ if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src) && (flags & IPV6_UNSPECSRC) == 0) { error = EOPNOTSUPP; IP6STAT_INC(ip6s_badscope); goto bad; } if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src)) { error = EOPNOTSUPP; IP6STAT_INC(ip6s_badscope); goto bad; } IP6STAT_INC(ip6s_localout); /* * Route packet. */ if (ro == 0) { ro = &ip6route; bzero((caddr_t)ro, sizeof(*ro)); } ro_pmtu = ro; if (opt && opt->ip6po_rthdr) ro = &opt->ip6po_route; dst = (struct sockaddr_in6 *)&ro->ro_dst; #ifdef FLOWTABLE if (ro->ro_rt == NULL) (void )flowtable_lookup(AF_INET6, m, (struct route *)ro); #endif fibnum = (inp != NULL) ? inp->inp_inc.inc_fibnum : M_GETFIB(m); again: /* * if specified, try to fill in the traffic class field. * do not override if a non-zero value is already set. * we check the diffserv field and the ecn field separately. */ if (opt && opt->ip6po_tclass >= 0) { int mask = 0; if ((ip6->ip6_flow & htonl(0xfc << 20)) == 0) mask |= 0xfc; if ((ip6->ip6_flow & htonl(0x03 << 20)) == 0) mask |= 0x03; if (mask != 0) ip6->ip6_flow |= htonl((opt->ip6po_tclass & mask) << 20); } /* fill in or override the hop limit field, if necessary. */ if (opt && opt->ip6po_hlim != -1) ip6->ip6_hlim = opt->ip6po_hlim & 0xff; else if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { if (im6o != NULL) ip6->ip6_hlim = im6o->im6o_multicast_hlim; else ip6->ip6_hlim = V_ip6_defmcasthlim; } /* adjust pointer */ ip6 = mtod(m, struct ip6_hdr *); if (ro->ro_rt && fwd_tag == NULL) { rt = ro->ro_rt; ifp = ro->ro_rt->rt_ifp; } else { if (fwd_tag == NULL) { bzero(&dst_sa, sizeof(dst_sa)); dst_sa.sin6_family = AF_INET6; dst_sa.sin6_len = sizeof(dst_sa); dst_sa.sin6_addr = ip6->ip6_dst; } error = in6_selectroute_fib(&dst_sa, opt, im6o, ro, &ifp, &rt, fibnum); if (error != 0) { if (ifp != NULL) in6_ifstat_inc(ifp, ifs6_out_discard); goto bad; } } if (rt == NULL) { /* * If in6_selectroute() does not return a route entry, * dst may not have been updated. */ *dst = dst_sa; /* XXX */ } /* * then rt (for unicast) and ifp must be non-NULL valid values. */ if ((flags & IPV6_FORWARDING) == 0) { /* XXX: the FORWARDING flag can be set for mrouting. */ in6_ifstat_inc(ifp, ifs6_out_request); } if (rt != NULL) { ia = (struct in6_ifaddr *)(rt->rt_ifa); counter_u64_add(rt->rt_pksent, 1); } /* * The outgoing interface must be in the zone of source and * destination addresses. */ origifp = ifp; src0 = ip6->ip6_src; if (in6_setscope(&src0, origifp, &zone)) goto badscope; bzero(&src_sa, sizeof(src_sa)); src_sa.sin6_family = AF_INET6; src_sa.sin6_len = sizeof(src_sa); src_sa.sin6_addr = ip6->ip6_src; if (sa6_recoverscope(&src_sa) || zone != src_sa.sin6_scope_id) goto badscope; dst0 = ip6->ip6_dst; if (in6_setscope(&dst0, origifp, &zone)) goto badscope; /* re-initialize to be sure */ bzero(&dst_sa, sizeof(dst_sa)); dst_sa.sin6_family = AF_INET6; dst_sa.sin6_len = sizeof(dst_sa); dst_sa.sin6_addr = ip6->ip6_dst; if (sa6_recoverscope(&dst_sa) || zone != dst_sa.sin6_scope_id) { goto badscope; } /* We should use ia_ifp to support the case of * sending packets to an address of our own. */ if (ia != NULL && ia->ia_ifp) ifp = ia->ia_ifp; /* scope check is done. */ goto routefound; badscope: IP6STAT_INC(ip6s_badscope); in6_ifstat_inc(origifp, ifs6_out_discard); if (error == 0) error = EHOSTUNREACH; /* XXX */ goto bad; routefound: if (rt && !IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { if (opt && opt->ip6po_nextroute.ro_rt) { /* * The nexthop is explicitly specified by the * application. We assume the next hop is an IPv6 * address. */ dst = (struct sockaddr_in6 *)opt->ip6po_nexthop; } else if ((rt->rt_flags & RTF_GATEWAY)) dst = (struct sockaddr_in6 *)rt->rt_gateway; } if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { m->m_flags &= ~(M_BCAST | M_MCAST); /* just in case */ } else { m->m_flags = (m->m_flags & ~M_BCAST) | M_MCAST; in6_ifstat_inc(ifp, ifs6_out_mcast); /* * Confirm that the outgoing interface supports multicast. */ if (!(ifp->if_flags & IFF_MULTICAST)) { IP6STAT_INC(ip6s_noroute); in6_ifstat_inc(ifp, ifs6_out_discard); error = ENETUNREACH; goto bad; } if ((im6o == NULL && in6_mcast_loop) || (im6o && im6o->im6o_multicast_loop)) { /* * Loop back multicast datagram if not expressly * forbidden to do so, even if we have not joined * the address; protocols will filter it later, * thus deferring a hash lookup and lock acquisition * at the expense of an m_copym(). */ ip6_mloopback(ifp, m, dst); } else { /* * If we are acting as a multicast router, perform * multicast forwarding as if the packet had just * arrived on the interface to which we are about * to send. The multicast forwarding function * recursively calls this function, using the * IPV6_FORWARDING flag to prevent infinite recursion. * * Multicasts that are looped back by ip6_mloopback(), * above, will be forwarded by the ip6_input() routine, * if necessary. */ if (V_ip6_mrouter && (flags & IPV6_FORWARDING) == 0) { /* * XXX: ip6_mforward expects that rcvif is NULL * when it is called from the originating path. * However, it may not always be the case. */ m->m_pkthdr.rcvif = NULL; if (ip6_mforward(ip6, ifp, m) != 0) { m_freem(m); goto done; } } } /* * Multicasts with a hoplimit of zero may be looped back, * above, but must not be transmitted on a network. * Also, multicasts addressed to the loopback interface * are not sent -- the above call to ip6_mloopback() will * loop back a copy if this host actually belongs to the * destination group on the loopback interface. */ if (ip6->ip6_hlim == 0 || (ifp->if_flags & IFF_LOOPBACK) || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst)) { m_freem(m); goto done; } } /* * Fill the outgoing inteface to tell the upper layer * to increment per-interface statistics. */ if (ifpp) *ifpp = ifp; /* Determine path MTU. */ if ((error = ip6_getpmtu(ro_pmtu, ro, ifp, &finaldst, &mtu, &alwaysfrag, fibnum)) != 0) goto bad; /* * The caller of this function may specify to use the minimum MTU * in some cases. * An advanced API option (IPV6_USE_MIN_MTU) can also override MTU * setting. The logic is a bit complicated; by default, unicast * packets will follow path MTU while multicast packets will be sent at * the minimum MTU. If IP6PO_MINMTU_ALL is specified, all packets * including unicast ones will be sent at the minimum MTU. Multicast * packets will always be sent at the minimum MTU unless * IP6PO_MINMTU_DISABLE is explicitly specified. * See RFC 3542 for more details. */ if (mtu > IPV6_MMTU) { if ((flags & IPV6_MINMTU)) mtu = IPV6_MMTU; else if (opt && opt->ip6po_minmtu == IP6PO_MINMTU_ALL) mtu = IPV6_MMTU; else if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) && (opt == NULL || opt->ip6po_minmtu != IP6PO_MINMTU_DISABLE)) { mtu = IPV6_MMTU; } } /* * clear embedded scope identifiers if necessary. * in6_clearscope will touch the addresses only when necessary. */ in6_clearscope(&ip6->ip6_src); in6_clearscope(&ip6->ip6_dst); /* * If the outgoing packet contains a hop-by-hop options header, * it must be examined and processed even by the source node. * (RFC 2460, section 4.) */ if (exthdrs.ip6e_hbh) { struct ip6_hbh *hbh = mtod(exthdrs.ip6e_hbh, struct ip6_hbh *); u_int32_t dummy; /* XXX unused */ u_int32_t plen = 0; /* XXX: ip6_process will check the value */ #ifdef DIAGNOSTIC if ((hbh->ip6h_len + 1) << 3 > exthdrs.ip6e_hbh->m_len) panic("ip6e_hbh is not contiguous"); #endif /* * XXX: if we have to send an ICMPv6 error to the sender, * we need the M_LOOP flag since icmp6_error() expects * the IPv6 and the hop-by-hop options header are * contiguous unless the flag is set. */ m->m_flags |= M_LOOP; m->m_pkthdr.rcvif = ifp; if (ip6_process_hopopts(m, (u_int8_t *)(hbh + 1), ((hbh->ip6h_len + 1) << 3) - sizeof(struct ip6_hbh), &dummy, &plen) < 0) { /* m was already freed at this point */ error = EINVAL;/* better error? */ goto done; } m->m_flags &= ~M_LOOP; /* XXX */ m->m_pkthdr.rcvif = NULL; } /* Jump over all PFIL processing if hooks are not active. */ if (!PFIL_HOOKED(&V_inet6_pfil_hook)) goto passout; odst = ip6->ip6_dst; /* Run through list of hooks for output packets. */ error = pfil_run_hooks(&V_inet6_pfil_hook, &m, ifp, PFIL_OUT, inp); if (error != 0 || m == NULL) goto done; ip6 = mtod(m, struct ip6_hdr *); needfiblookup = 0; /* See if destination IP address was changed by packet filter. */ if (!IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst)) { m->m_flags |= M_SKIP_FIREWALL; /* If destination is now ourself drop to ip6_input(). */ if (in6_localip(&ip6->ip6_dst)) { m->m_flags |= M_FASTFWD_OURS; if (m->m_pkthdr.rcvif == NULL) m->m_pkthdr.rcvif = V_loif; if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA_IPV6) { m->m_pkthdr.csum_flags |= CSUM_DATA_VALID_IPV6 | CSUM_PSEUDO_HDR; m->m_pkthdr.csum_data = 0xffff; } #ifdef SCTP if (m->m_pkthdr.csum_flags & CSUM_SCTP_IPV6) m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID; #endif error = netisr_queue(NETISR_IPV6, m); goto done; } else needfiblookup = 1; /* Redo the routing table lookup. */ } /* See if fib was changed by packet filter. */ if (fibnum != M_GETFIB(m)) { m->m_flags |= M_SKIP_FIREWALL; fibnum = M_GETFIB(m); RO_RTFREE(ro); needfiblookup = 1; } if (needfiblookup) goto again; /* See if local, if yes, send it to netisr. */ if (m->m_flags & M_FASTFWD_OURS) { if (m->m_pkthdr.rcvif == NULL) m->m_pkthdr.rcvif = V_loif; if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA_IPV6) { m->m_pkthdr.csum_flags |= CSUM_DATA_VALID_IPV6 | CSUM_PSEUDO_HDR; m->m_pkthdr.csum_data = 0xffff; } #ifdef SCTP if (m->m_pkthdr.csum_flags & CSUM_SCTP_IPV6) m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID; #endif error = netisr_queue(NETISR_IPV6, m); goto done; } /* Or forward to some other address? */ if ((m->m_flags & M_IP6_NEXTHOP) && (fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL) { dst = (struct sockaddr_in6 *)&ro->ro_dst; bcopy((fwd_tag+1), &dst_sa, sizeof(struct sockaddr_in6)); m->m_flags |= M_SKIP_FIREWALL; m->m_flags &= ~M_IP6_NEXTHOP; m_tag_delete(m, fwd_tag); goto again; } passout: /* * Send the packet to the outgoing interface. * If necessary, do IPv6 fragmentation before sending. * * the logic here is rather complex: * 1: normal case (dontfrag == 0, alwaysfrag == 0) * 1-a: send as is if tlen <= path mtu * 1-b: fragment if tlen > path mtu * * 2: if user asks us not to fragment (dontfrag == 1) * 2-a: send as is if tlen <= interface mtu * 2-b: error if tlen > interface mtu * * 3: if we always need to attach fragment header (alwaysfrag == 1) * always fragment * * 4: if dontfrag == 1 && alwaysfrag == 1 * error, as we cannot handle this conflicting request */ sw_csum = m->m_pkthdr.csum_flags; if (!hdrsplit) { tso = ((sw_csum & ifp->if_hwassist & CSUM_TSO) != 0) ? 1 : 0; sw_csum &= ~ifp->if_hwassist; } else tso = 0; /* * If we added extension headers, we will not do TSO and calculate the * checksums ourselves for now. * XXX-BZ Need a framework to know when the NIC can handle it, even * with ext. hdrs. */ if (sw_csum & CSUM_DELAY_DATA_IPV6) { sw_csum &= ~CSUM_DELAY_DATA_IPV6; in6_delayed_cksum(m, plen, sizeof(struct ip6_hdr)); } #ifdef SCTP if (sw_csum & CSUM_SCTP_IPV6) { sw_csum &= ~CSUM_SCTP_IPV6; sctp_delayed_cksum(m, sizeof(struct ip6_hdr)); } #endif m->m_pkthdr.csum_flags &= ifp->if_hwassist; tlen = m->m_pkthdr.len; if ((opt && (opt->ip6po_flags & IP6PO_DONTFRAG)) || tso) dontfrag = 1; else dontfrag = 0; if (dontfrag && alwaysfrag) { /* case 4 */ /* conflicting request - can't transmit */ error = EMSGSIZE; goto bad; } if (dontfrag && tlen > IN6_LINKMTU(ifp) && !tso) { /* case 2-b */ /* * Even if the DONTFRAG option is specified, we cannot send the * packet when the data length is larger than the MTU of the * outgoing interface. - * Notify the error by sending IPV6_PATHMTU ancillary data as - * well as returning an error code (the latter is not described - * in the API spec.) + * Notify the error by sending IPV6_PATHMTU ancillary data if + * application wanted to know the MTU value. Also return an + * error code (this is not described in the API spec). */ - u_int32_t mtu32; - struct ip6ctlparam ip6cp; - - mtu32 = (u_int32_t)mtu; - bzero(&ip6cp, sizeof(ip6cp)); - ip6cp.ip6c_cmdarg = (void *)&mtu32; - pfctlinput2(PRC_MSGSIZE, (struct sockaddr *)&ro_pmtu->ro_dst, - (void *)&ip6cp); - + if (inp != NULL) + ip6_notify_pmtu(inp, &dst_sa, (u_int32_t)mtu); error = EMSGSIZE; goto bad; } /* * transmit packet without fragmentation */ if (dontfrag || (!alwaysfrag && tlen <= mtu)) { /* case 1-a and 2-a */ struct in6_ifaddr *ia6; ip6 = mtod(m, struct ip6_hdr *); ia6 = in6_ifawithifp(ifp, &ip6->ip6_src); if (ia6) { /* Record statistics for this interface address. */ counter_u64_add(ia6->ia_ifa.ifa_opackets, 1); counter_u64_add(ia6->ia_ifa.ifa_obytes, m->m_pkthdr.len); ifa_free(&ia6->ia_ifa); } error = nd6_output(ifp, origifp, m, dst, ro->ro_rt); goto done; } /* * try to fragment the packet. case 1-b and 3 */ if (mtu < IPV6_MMTU) { /* path MTU cannot be less than IPV6_MMTU */ error = EMSGSIZE; in6_ifstat_inc(ifp, ifs6_out_fragfail); goto bad; } else if (ip6->ip6_plen == 0) { /* jumbo payload cannot be fragmented */ error = EMSGSIZE; in6_ifstat_inc(ifp, ifs6_out_fragfail); goto bad; } else { u_char nextproto; /* * Too large for the destination or interface; * fragment if possible. * Must be able to put at least 8 bytes per fragment. */ hlen = unfragpartlen; if (mtu > IPV6_MAXPACKET) mtu = IPV6_MAXPACKET; len = (mtu - hlen - sizeof(struct ip6_frag)) & ~7; if (len < 8) { error = EMSGSIZE; in6_ifstat_inc(ifp, ifs6_out_fragfail); goto bad; } /* * If the interface will not calculate checksums on * fragmented packets, then do it here. * XXX-BZ handle the hw offloading case. Need flags. */ if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA_IPV6) { in6_delayed_cksum(m, plen, hlen); m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA_IPV6; } #ifdef SCTP if (m->m_pkthdr.csum_flags & CSUM_SCTP_IPV6) { sctp_delayed_cksum(m, hlen); m->m_pkthdr.csum_flags &= ~CSUM_SCTP_IPV6; } #endif /* * Change the next header field of the last header in the * unfragmentable part. */ if (exthdrs.ip6e_rthdr) { nextproto = *mtod(exthdrs.ip6e_rthdr, u_char *); *mtod(exthdrs.ip6e_rthdr, u_char *) = IPPROTO_FRAGMENT; } else if (exthdrs.ip6e_dest1) { nextproto = *mtod(exthdrs.ip6e_dest1, u_char *); *mtod(exthdrs.ip6e_dest1, u_char *) = IPPROTO_FRAGMENT; } else if (exthdrs.ip6e_hbh) { nextproto = *mtod(exthdrs.ip6e_hbh, u_char *); *mtod(exthdrs.ip6e_hbh, u_char *) = IPPROTO_FRAGMENT; } else { nextproto = ip6->ip6_nxt; ip6->ip6_nxt = IPPROTO_FRAGMENT; } /* * Loop through length of segment after first fragment, * make new header and copy data of each part and link onto * chain. */ m0 = m; if ((error = ip6_fragment(ifp, m, hlen, nextproto, len))) goto sendorfree; in6_ifstat_inc(ifp, ifs6_out_fragok); } /* * Remove leading garbages. */ sendorfree: m = m0->m_nextpkt; m0->m_nextpkt = 0; m_freem(m0); for (m0 = m; m; m = m0) { m0 = m->m_nextpkt; m->m_nextpkt = 0; if (error == 0) { /* Record statistics for this interface address. */ if (ia) { counter_u64_add(ia->ia_ifa.ifa_opackets, 1); counter_u64_add(ia->ia_ifa.ifa_obytes, m->m_pkthdr.len); } error = nd6_output(ifp, origifp, m, dst, ro->ro_rt); } else m_freem(m); } if (error == 0) IP6STAT_INC(ip6s_fragmented); done: if (ro == &ip6route) RO_RTFREE(ro); if (ro_pmtu == &ip6route) RO_RTFREE(ro_pmtu); return (error); freehdrs: m_freem(exthdrs.ip6e_hbh); /* m_freem will check if mbuf is 0 */ m_freem(exthdrs.ip6e_dest1); m_freem(exthdrs.ip6e_rthdr); m_freem(exthdrs.ip6e_dest2); /* FALLTHROUGH */ bad: if (m) m_freem(m); goto done; } static int ip6_copyexthdr(struct mbuf **mp, caddr_t hdr, int hlen) { struct mbuf *m; if (hlen > MCLBYTES) return (ENOBUFS); /* XXX */ if (hlen > MLEN) m = m_getcl(M_NOWAIT, MT_DATA, 0); else m = m_get(M_NOWAIT, MT_DATA); if (m == NULL) return (ENOBUFS); m->m_len = hlen; if (hdr) bcopy(hdr, mtod(m, caddr_t), hlen); *mp = m; return (0); } /* * Insert jumbo payload option. */ static int ip6_insert_jumboopt(struct ip6_exthdrs *exthdrs, u_int32_t plen) { struct mbuf *mopt; u_char *optbuf; u_int32_t v; #define JUMBOOPTLEN 8 /* length of jumbo payload option and padding */ /* * If there is no hop-by-hop options header, allocate new one. * If there is one but it doesn't have enough space to store the * jumbo payload option, allocate a cluster to store the whole options. * Otherwise, use it to store the options. */ if (exthdrs->ip6e_hbh == 0) { mopt = m_get(M_NOWAIT, MT_DATA); if (mopt == NULL) return (ENOBUFS); mopt->m_len = JUMBOOPTLEN; optbuf = mtod(mopt, u_char *); optbuf[1] = 0; /* = ((JUMBOOPTLEN) >> 3) - 1 */ exthdrs->ip6e_hbh = mopt; } else { struct ip6_hbh *hbh; mopt = exthdrs->ip6e_hbh; if (M_TRAILINGSPACE(mopt) < JUMBOOPTLEN) { /* * XXX assumption: * - exthdrs->ip6e_hbh is not referenced from places * other than exthdrs. * - exthdrs->ip6e_hbh is not an mbuf chain. */ int oldoptlen = mopt->m_len; struct mbuf *n; /* * XXX: give up if the whole (new) hbh header does * not fit even in an mbuf cluster. */ if (oldoptlen + JUMBOOPTLEN > MCLBYTES) return (ENOBUFS); /* * As a consequence, we must always prepare a cluster * at this point. */ n = m_getcl(M_NOWAIT, MT_DATA, 0); if (n == NULL) return (ENOBUFS); n->m_len = oldoptlen + JUMBOOPTLEN; bcopy(mtod(mopt, caddr_t), mtod(n, caddr_t), oldoptlen); optbuf = mtod(n, caddr_t) + oldoptlen; m_freem(mopt); mopt = exthdrs->ip6e_hbh = n; } else { optbuf = mtod(mopt, u_char *) + mopt->m_len; mopt->m_len += JUMBOOPTLEN; } optbuf[0] = IP6OPT_PADN; optbuf[1] = 1; /* * Adjust the header length according to the pad and * the jumbo payload option. */ hbh = mtod(mopt, struct ip6_hbh *); hbh->ip6h_len += (JUMBOOPTLEN >> 3); } /* fill in the option. */ optbuf[2] = IP6OPT_JUMBO; optbuf[3] = 4; v = (u_int32_t)htonl(plen + JUMBOOPTLEN); bcopy(&v, &optbuf[4], sizeof(u_int32_t)); /* finally, adjust the packet header length */ exthdrs->ip6e_ip6->m_pkthdr.len += JUMBOOPTLEN; return (0); #undef JUMBOOPTLEN } /* * Insert fragment header and copy unfragmentable header portions. */ static int ip6_insertfraghdr(struct mbuf *m0, struct mbuf *m, int hlen, struct ip6_frag **frghdrp) { struct mbuf *n, *mlast; if (hlen > sizeof(struct ip6_hdr)) { n = m_copym(m0, sizeof(struct ip6_hdr), hlen - sizeof(struct ip6_hdr), M_NOWAIT); if (n == 0) return (ENOBUFS); m->m_next = n; } else n = m; /* Search for the last mbuf of unfragmentable part. */ for (mlast = n; mlast->m_next; mlast = mlast->m_next) ; if (M_WRITABLE(mlast) && M_TRAILINGSPACE(mlast) >= sizeof(struct ip6_frag)) { /* use the trailing space of the last mbuf for the fragment hdr */ *frghdrp = (struct ip6_frag *)(mtod(mlast, caddr_t) + mlast->m_len); mlast->m_len += sizeof(struct ip6_frag); m->m_pkthdr.len += sizeof(struct ip6_frag); } else { /* allocate a new mbuf for the fragment header */ struct mbuf *mfrg; mfrg = m_get(M_NOWAIT, MT_DATA); if (mfrg == NULL) return (ENOBUFS); mfrg->m_len = sizeof(struct ip6_frag); *frghdrp = mtod(mfrg, struct ip6_frag *); mlast->m_next = mfrg; } return (0); } static int ip6_getpmtu(struct route_in6 *ro_pmtu, struct route_in6 *ro, struct ifnet *ifp, struct in6_addr *dst, u_long *mtup, int *alwaysfragp, u_int fibnum) { u_int32_t mtu = 0; int alwaysfrag = 0; int error = 0; if (ro_pmtu != ro) { /* The first hop and the final destination may differ. */ struct sockaddr_in6 *sa6_dst = (struct sockaddr_in6 *)&ro_pmtu->ro_dst; if (ro_pmtu->ro_rt && ((ro_pmtu->ro_rt->rt_flags & RTF_UP) == 0 || !IN6_ARE_ADDR_EQUAL(&sa6_dst->sin6_addr, dst))) { RTFREE(ro_pmtu->ro_rt); ro_pmtu->ro_rt = (struct rtentry *)NULL; } if (ro_pmtu->ro_rt == NULL) { bzero(sa6_dst, sizeof(*sa6_dst)); sa6_dst->sin6_family = AF_INET6; sa6_dst->sin6_len = sizeof(struct sockaddr_in6); sa6_dst->sin6_addr = *dst; in6_rtalloc(ro_pmtu, fibnum); } } if (ro_pmtu->ro_rt) { u_int32_t ifmtu; struct in_conninfo inc; bzero(&inc, sizeof(inc)); inc.inc_flags |= INC_ISIPV6; inc.inc6_faddr = *dst; if (ifp == NULL) ifp = ro_pmtu->ro_rt->rt_ifp; ifmtu = IN6_LINKMTU(ifp); mtu = tcp_hc_getmtu(&inc); if (mtu) mtu = min(mtu, ro_pmtu->ro_rt->rt_mtu); else mtu = ro_pmtu->ro_rt->rt_mtu; if (mtu == 0) mtu = ifmtu; else if (mtu < IPV6_MMTU) { /* * RFC2460 section 5, last paragraph: * if we record ICMPv6 too big message with * mtu < IPV6_MMTU, transmit packets sized IPV6_MMTU * or smaller, with framgent header attached. * (fragment header is needed regardless from the * packet size, for translators to identify packets) */ alwaysfrag = 1; mtu = IPV6_MMTU; } } else if (ifp) { mtu = IN6_LINKMTU(ifp); } else error = EHOSTUNREACH; /* XXX */ *mtup = mtu; if (alwaysfragp) *alwaysfragp = alwaysfrag; return (error); } /* * IP6 socket option processing. */ int ip6_ctloutput(struct socket *so, struct sockopt *sopt) { int optdatalen, uproto; void *optdata; struct inpcb *in6p = sotoinpcb(so); int error, optval; int level, op, optname; int optlen; struct thread *td; #ifdef RSS uint32_t rss_bucket; int retval; #endif level = sopt->sopt_level; op = sopt->sopt_dir; optname = sopt->sopt_name; optlen = sopt->sopt_valsize; td = sopt->sopt_td; error = 0; optval = 0; uproto = (int)so->so_proto->pr_protocol; if (level != IPPROTO_IPV6) { error = EINVAL; if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_dir == SOPT_SET) { switch (sopt->sopt_name) { case SO_REUSEADDR: INP_WLOCK(in6p); if ((so->so_options & SO_REUSEADDR) != 0) in6p->inp_flags2 |= INP_REUSEADDR; else in6p->inp_flags2 &= ~INP_REUSEADDR; INP_WUNLOCK(in6p); error = 0; break; case SO_REUSEPORT: INP_WLOCK(in6p); if ((so->so_options & SO_REUSEPORT) != 0) in6p->inp_flags2 |= INP_REUSEPORT; else in6p->inp_flags2 &= ~INP_REUSEPORT; INP_WUNLOCK(in6p); error = 0; break; case SO_SETFIB: INP_WLOCK(in6p); in6p->inp_inc.inc_fibnum = so->so_fibnum; INP_WUNLOCK(in6p); error = 0; break; default: break; } } } else { /* level == IPPROTO_IPV6 */ switch (op) { case SOPT_SET: switch (optname) { case IPV6_2292PKTOPTIONS: #ifdef IPV6_PKTOPTIONS case IPV6_PKTOPTIONS: #endif { struct mbuf *m; error = soopt_getm(sopt, &m); /* XXX */ if (error != 0) break; error = soopt_mcopyin(sopt, m); /* XXX */ if (error != 0) break; error = ip6_pcbopts(&in6p->in6p_outputopts, m, so, sopt); m_freem(m); /* XXX */ break; } /* * Use of some Hop-by-Hop options or some * Destination options, might require special * privilege. That is, normal applications * (without special privilege) might be forbidden * from setting certain options in outgoing packets, * and might never see certain options in received * packets. [RFC 2292 Section 6] * KAME specific note: * KAME prevents non-privileged users from sending or * receiving ANY hbh/dst options in order to avoid * overhead of parsing options in the kernel. */ case IPV6_RECVHOPOPTS: case IPV6_RECVDSTOPTS: case IPV6_RECVRTHDRDSTOPTS: if (td != NULL) { error = priv_check(td, PRIV_NETINET_SETHDROPTS); if (error) break; } /* FALLTHROUGH */ case IPV6_UNICAST_HOPS: case IPV6_HOPLIMIT: case IPV6_RECVPKTINFO: case IPV6_RECVHOPLIMIT: case IPV6_RECVRTHDR: case IPV6_RECVPATHMTU: case IPV6_RECVTCLASS: case IPV6_V6ONLY: case IPV6_AUTOFLOWLABEL: case IPV6_BINDANY: case IPV6_BINDMULTI: #ifdef RSS case IPV6_RSS_LISTEN_BUCKET: #endif if (optname == IPV6_BINDANY && td != NULL) { error = priv_check(td, PRIV_NETINET_BINDANY); if (error) break; } if (optlen != sizeof(int)) { error = EINVAL; break; } error = sooptcopyin(sopt, &optval, sizeof optval, sizeof optval); if (error) break; switch (optname) { case IPV6_UNICAST_HOPS: if (optval < -1 || optval >= 256) error = EINVAL; else { /* -1 = kernel default */ in6p->in6p_hops = optval; if ((in6p->inp_vflag & INP_IPV4) != 0) in6p->inp_ip_ttl = optval; } break; #define OPTSET(bit) \ do { \ INP_WLOCK(in6p); \ if (optval) \ in6p->inp_flags |= (bit); \ else \ in6p->inp_flags &= ~(bit); \ INP_WUNLOCK(in6p); \ } while (/*CONSTCOND*/ 0) #define OPTSET2292(bit) \ do { \ INP_WLOCK(in6p); \ in6p->inp_flags |= IN6P_RFC2292; \ if (optval) \ in6p->inp_flags |= (bit); \ else \ in6p->inp_flags &= ~(bit); \ INP_WUNLOCK(in6p); \ } while (/*CONSTCOND*/ 0) #define OPTBIT(bit) (in6p->inp_flags & (bit) ? 1 : 0) #define OPTSET2(bit, val) do { \ INP_WLOCK(in6p); \ if (val) \ in6p->inp_flags2 |= bit; \ else \ in6p->inp_flags2 &= ~bit; \ INP_WUNLOCK(in6p); \ } while (0) #define OPTBIT2(bit) (in6p->inp_flags2 & (bit) ? 1 : 0) case IPV6_RECVPKTINFO: /* cannot mix with RFC2292 */ if (OPTBIT(IN6P_RFC2292)) { error = EINVAL; break; } OPTSET(IN6P_PKTINFO); break; case IPV6_HOPLIMIT: { struct ip6_pktopts **optp; /* cannot mix with RFC2292 */ if (OPTBIT(IN6P_RFC2292)) { error = EINVAL; break; } optp = &in6p->in6p_outputopts; error = ip6_pcbopt(IPV6_HOPLIMIT, (u_char *)&optval, sizeof(optval), optp, (td != NULL) ? td->td_ucred : NULL, uproto); break; } case IPV6_RECVHOPLIMIT: /* cannot mix with RFC2292 */ if (OPTBIT(IN6P_RFC2292)) { error = EINVAL; break; } OPTSET(IN6P_HOPLIMIT); break; case IPV6_RECVHOPOPTS: /* cannot mix with RFC2292 */ if (OPTBIT(IN6P_RFC2292)) { error = EINVAL; break; } OPTSET(IN6P_HOPOPTS); break; case IPV6_RECVDSTOPTS: /* cannot mix with RFC2292 */ if (OPTBIT(IN6P_RFC2292)) { error = EINVAL; break; } OPTSET(IN6P_DSTOPTS); break; case IPV6_RECVRTHDRDSTOPTS: /* cannot mix with RFC2292 */ if (OPTBIT(IN6P_RFC2292)) { error = EINVAL; break; } OPTSET(IN6P_RTHDRDSTOPTS); break; case IPV6_RECVRTHDR: /* cannot mix with RFC2292 */ if (OPTBIT(IN6P_RFC2292)) { error = EINVAL; break; } OPTSET(IN6P_RTHDR); break; case IPV6_RECVPATHMTU: /* * We ignore this option for TCP * sockets. * (RFC3542 leaves this case * unspecified.) */ if (uproto != IPPROTO_TCP) OPTSET(IN6P_MTU); break; case IPV6_V6ONLY: /* * make setsockopt(IPV6_V6ONLY) * available only prior to bind(2). * see ipng mailing list, Jun 22 2001. */ if (in6p->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) { error = EINVAL; break; } OPTSET(IN6P_IPV6_V6ONLY); if (optval) in6p->inp_vflag &= ~INP_IPV4; else in6p->inp_vflag |= INP_IPV4; break; case IPV6_RECVTCLASS: /* cannot mix with RFC2292 XXX */ if (OPTBIT(IN6P_RFC2292)) { error = EINVAL; break; } OPTSET(IN6P_TCLASS); break; case IPV6_AUTOFLOWLABEL: OPTSET(IN6P_AUTOFLOWLABEL); break; case IPV6_BINDANY: OPTSET(INP_BINDANY); break; case IPV6_BINDMULTI: OPTSET2(INP_BINDMULTI, optval); break; #ifdef RSS case IPV6_RSS_LISTEN_BUCKET: if ((optval >= 0) && (optval < rss_getnumbuckets())) { in6p->inp_rss_listen_bucket = optval; OPTSET2(INP_RSS_BUCKET_SET, 1); } else { error = EINVAL; } break; #endif } break; case IPV6_TCLASS: case IPV6_DONTFRAG: case IPV6_USE_MIN_MTU: case IPV6_PREFER_TEMPADDR: if (optlen != sizeof(optval)) { error = EINVAL; break; } error = sooptcopyin(sopt, &optval, sizeof optval, sizeof optval); if (error) break; { struct ip6_pktopts **optp; optp = &in6p->in6p_outputopts; error = ip6_pcbopt(optname, (u_char *)&optval, sizeof(optval), optp, (td != NULL) ? td->td_ucred : NULL, uproto); break; } case IPV6_2292PKTINFO: case IPV6_2292HOPLIMIT: case IPV6_2292HOPOPTS: case IPV6_2292DSTOPTS: case IPV6_2292RTHDR: /* RFC 2292 */ if (optlen != sizeof(int)) { error = EINVAL; break; } error = sooptcopyin(sopt, &optval, sizeof optval, sizeof optval); if (error) break; switch (optname) { case IPV6_2292PKTINFO: OPTSET2292(IN6P_PKTINFO); break; case IPV6_2292HOPLIMIT: OPTSET2292(IN6P_HOPLIMIT); break; case IPV6_2292HOPOPTS: /* * Check super-user privilege. * See comments for IPV6_RECVHOPOPTS. */ if (td != NULL) { error = priv_check(td, PRIV_NETINET_SETHDROPTS); if (error) return (error); } OPTSET2292(IN6P_HOPOPTS); break; case IPV6_2292DSTOPTS: if (td != NULL) { error = priv_check(td, PRIV_NETINET_SETHDROPTS); if (error) return (error); } OPTSET2292(IN6P_DSTOPTS|IN6P_RTHDRDSTOPTS); /* XXX */ break; case IPV6_2292RTHDR: OPTSET2292(IN6P_RTHDR); break; } break; case IPV6_PKTINFO: case IPV6_HOPOPTS: case IPV6_RTHDR: case IPV6_DSTOPTS: case IPV6_RTHDRDSTOPTS: case IPV6_NEXTHOP: { /* new advanced API (RFC3542) */ u_char *optbuf; u_char optbuf_storage[MCLBYTES]; int optlen; struct ip6_pktopts **optp; /* cannot mix with RFC2292 */ if (OPTBIT(IN6P_RFC2292)) { error = EINVAL; break; } /* * We only ensure valsize is not too large * here. Further validation will be done * later. */ error = sooptcopyin(sopt, optbuf_storage, sizeof(optbuf_storage), 0); if (error) break; optlen = sopt->sopt_valsize; optbuf = optbuf_storage; optp = &in6p->in6p_outputopts; error = ip6_pcbopt(optname, optbuf, optlen, optp, (td != NULL) ? td->td_ucred : NULL, uproto); break; } #undef OPTSET case IPV6_MULTICAST_IF: case IPV6_MULTICAST_HOPS: case IPV6_MULTICAST_LOOP: case IPV6_JOIN_GROUP: case IPV6_LEAVE_GROUP: case IPV6_MSFILTER: case MCAST_BLOCK_SOURCE: case MCAST_UNBLOCK_SOURCE: case MCAST_JOIN_GROUP: case MCAST_LEAVE_GROUP: case MCAST_JOIN_SOURCE_GROUP: case MCAST_LEAVE_SOURCE_GROUP: error = ip6_setmoptions(in6p, sopt); break; case IPV6_PORTRANGE: error = sooptcopyin(sopt, &optval, sizeof optval, sizeof optval); if (error) break; INP_WLOCK(in6p); switch (optval) { case IPV6_PORTRANGE_DEFAULT: in6p->inp_flags &= ~(INP_LOWPORT); in6p->inp_flags &= ~(INP_HIGHPORT); break; case IPV6_PORTRANGE_HIGH: in6p->inp_flags &= ~(INP_LOWPORT); in6p->inp_flags |= INP_HIGHPORT; break; case IPV6_PORTRANGE_LOW: in6p->inp_flags &= ~(INP_HIGHPORT); in6p->inp_flags |= INP_LOWPORT; break; default: error = EINVAL; break; } INP_WUNLOCK(in6p); break; #ifdef IPSEC case IPV6_IPSEC_POLICY: { caddr_t req; struct mbuf *m; if ((error = soopt_getm(sopt, &m)) != 0) /* XXX */ break; if ((error = soopt_mcopyin(sopt, m)) != 0) /* XXX */ break; req = mtod(m, caddr_t); error = ipsec_set_policy(in6p, optname, req, m->m_len, (sopt->sopt_td != NULL) ? sopt->sopt_td->td_ucred : NULL); m_freem(m); break; } #endif /* IPSEC */ default: error = ENOPROTOOPT; break; } break; case SOPT_GET: switch (optname) { case IPV6_2292PKTOPTIONS: #ifdef IPV6_PKTOPTIONS case IPV6_PKTOPTIONS: #endif /* * RFC3542 (effectively) deprecated the * semantics of the 2292-style pktoptions. * Since it was not reliable in nature (i.e., * applications had to expect the lack of some * information after all), it would make sense * to simplify this part by always returning * empty data. */ sopt->sopt_valsize = 0; break; case IPV6_RECVHOPOPTS: case IPV6_RECVDSTOPTS: case IPV6_RECVRTHDRDSTOPTS: case IPV6_UNICAST_HOPS: case IPV6_RECVPKTINFO: case IPV6_RECVHOPLIMIT: case IPV6_RECVRTHDR: case IPV6_RECVPATHMTU: case IPV6_V6ONLY: case IPV6_PORTRANGE: case IPV6_RECVTCLASS: case IPV6_AUTOFLOWLABEL: case IPV6_BINDANY: case IPV6_FLOWID: case IPV6_FLOWTYPE: #ifdef RSS case IPV6_RSSBUCKETID: #endif switch (optname) { case IPV6_RECVHOPOPTS: optval = OPTBIT(IN6P_HOPOPTS); break; case IPV6_RECVDSTOPTS: optval = OPTBIT(IN6P_DSTOPTS); break; case IPV6_RECVRTHDRDSTOPTS: optval = OPTBIT(IN6P_RTHDRDSTOPTS); break; case IPV6_UNICAST_HOPS: optval = in6p->in6p_hops; break; case IPV6_RECVPKTINFO: optval = OPTBIT(IN6P_PKTINFO); break; case IPV6_RECVHOPLIMIT: optval = OPTBIT(IN6P_HOPLIMIT); break; case IPV6_RECVRTHDR: optval = OPTBIT(IN6P_RTHDR); break; case IPV6_RECVPATHMTU: optval = OPTBIT(IN6P_MTU); break; case IPV6_V6ONLY: optval = OPTBIT(IN6P_IPV6_V6ONLY); break; case IPV6_PORTRANGE: { int flags; flags = in6p->inp_flags; if (flags & INP_HIGHPORT) optval = IPV6_PORTRANGE_HIGH; else if (flags & INP_LOWPORT) optval = IPV6_PORTRANGE_LOW; else optval = 0; break; } case IPV6_RECVTCLASS: optval = OPTBIT(IN6P_TCLASS); break; case IPV6_AUTOFLOWLABEL: optval = OPTBIT(IN6P_AUTOFLOWLABEL); break; case IPV6_BINDANY: optval = OPTBIT(INP_BINDANY); break; case IPV6_FLOWID: optval = in6p->inp_flowid; break; case IPV6_FLOWTYPE: optval = in6p->inp_flowtype; break; #ifdef RSS case IPV6_RSSBUCKETID: retval = rss_hash2bucket(in6p->inp_flowid, in6p->inp_flowtype, &rss_bucket); if (retval == 0) optval = rss_bucket; else error = EINVAL; break; #endif case IPV6_BINDMULTI: optval = OPTBIT2(INP_BINDMULTI); break; } if (error) break; error = sooptcopyout(sopt, &optval, sizeof optval); break; case IPV6_PATHMTU: { u_long pmtu = 0; struct ip6_mtuinfo mtuinfo; struct route_in6 sro; bzero(&sro, sizeof(sro)); if (!(so->so_state & SS_ISCONNECTED)) return (ENOTCONN); /* * XXX: we dot not consider the case of source * routing, or optional information to specify * the outgoing interface. */ error = ip6_getpmtu(&sro, NULL, NULL, &in6p->in6p_faddr, &pmtu, NULL, so->so_fibnum); if (sro.ro_rt) RTFREE(sro.ro_rt); if (error) break; if (pmtu > IPV6_MAXPACKET) pmtu = IPV6_MAXPACKET; bzero(&mtuinfo, sizeof(mtuinfo)); mtuinfo.ip6m_mtu = (u_int32_t)pmtu; optdata = (void *)&mtuinfo; optdatalen = sizeof(mtuinfo); error = sooptcopyout(sopt, optdata, optdatalen); break; } case IPV6_2292PKTINFO: case IPV6_2292HOPLIMIT: case IPV6_2292HOPOPTS: case IPV6_2292RTHDR: case IPV6_2292DSTOPTS: switch (optname) { case IPV6_2292PKTINFO: optval = OPTBIT(IN6P_PKTINFO); break; case IPV6_2292HOPLIMIT: optval = OPTBIT(IN6P_HOPLIMIT); break; case IPV6_2292HOPOPTS: optval = OPTBIT(IN6P_HOPOPTS); break; case IPV6_2292RTHDR: optval = OPTBIT(IN6P_RTHDR); break; case IPV6_2292DSTOPTS: optval = OPTBIT(IN6P_DSTOPTS|IN6P_RTHDRDSTOPTS); break; } error = sooptcopyout(sopt, &optval, sizeof optval); break; case IPV6_PKTINFO: case IPV6_HOPOPTS: case IPV6_RTHDR: case IPV6_DSTOPTS: case IPV6_RTHDRDSTOPTS: case IPV6_NEXTHOP: case IPV6_TCLASS: case IPV6_DONTFRAG: case IPV6_USE_MIN_MTU: case IPV6_PREFER_TEMPADDR: error = ip6_getpcbopt(in6p->in6p_outputopts, optname, sopt); break; case IPV6_MULTICAST_IF: case IPV6_MULTICAST_HOPS: case IPV6_MULTICAST_LOOP: case IPV6_MSFILTER: error = ip6_getmoptions(in6p, sopt); break; #ifdef IPSEC case IPV6_IPSEC_POLICY: { caddr_t req = NULL; size_t len = 0; struct mbuf *m = NULL; struct mbuf **mp = &m; size_t ovalsize = sopt->sopt_valsize; caddr_t oval = (caddr_t)sopt->sopt_val; error = soopt_getm(sopt, &m); /* XXX */ if (error != 0) break; error = soopt_mcopyin(sopt, m); /* XXX */ if (error != 0) break; sopt->sopt_valsize = ovalsize; sopt->sopt_val = oval; if (m) { req = mtod(m, caddr_t); len = m->m_len; } error = ipsec_get_policy(in6p, req, len, mp); if (error == 0) error = soopt_mcopyout(sopt, m); /* XXX */ if (error == 0 && m) m_freem(m); break; } #endif /* IPSEC */ default: error = ENOPROTOOPT; break; } break; } } return (error); } int ip6_raw_ctloutput(struct socket *so, struct sockopt *sopt) { int error = 0, optval, optlen; const int icmp6off = offsetof(struct icmp6_hdr, icmp6_cksum); struct inpcb *in6p = sotoinpcb(so); int level, op, optname; level = sopt->sopt_level; op = sopt->sopt_dir; optname = sopt->sopt_name; optlen = sopt->sopt_valsize; if (level != IPPROTO_IPV6) { return (EINVAL); } switch (optname) { case IPV6_CHECKSUM: /* * For ICMPv6 sockets, no modification allowed for checksum * offset, permit "no change" values to help existing apps. * * RFC3542 says: "An attempt to set IPV6_CHECKSUM * for an ICMPv6 socket will fail." * The current behavior does not meet RFC3542. */ switch (op) { case SOPT_SET: if (optlen != sizeof(int)) { error = EINVAL; break; } error = sooptcopyin(sopt, &optval, sizeof(optval), sizeof(optval)); if (error) break; if ((optval % 2) != 0) { /* the API assumes even offset values */ error = EINVAL; } else if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) { if (optval != icmp6off) error = EINVAL; } else in6p->in6p_cksum = optval; break; case SOPT_GET: if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) optval = icmp6off; else optval = in6p->in6p_cksum; error = sooptcopyout(sopt, &optval, sizeof(optval)); break; default: error = EINVAL; break; } break; default: error = ENOPROTOOPT; break; } return (error); } /* * Set up IP6 options in pcb for insertion in output packets or * specifying behavior of outgoing packets. */ static int ip6_pcbopts(struct ip6_pktopts **pktopt, struct mbuf *m, struct socket *so, struct sockopt *sopt) { struct ip6_pktopts *opt = *pktopt; int error = 0; struct thread *td = sopt->sopt_td; /* turn off any old options. */ if (opt) { #ifdef DIAGNOSTIC if (opt->ip6po_pktinfo || opt->ip6po_nexthop || opt->ip6po_hbh || opt->ip6po_dest1 || opt->ip6po_dest2 || opt->ip6po_rhinfo.ip6po_rhi_rthdr) printf("ip6_pcbopts: all specified options are cleared.\n"); #endif ip6_clearpktopts(opt, -1); } else opt = malloc(sizeof(*opt), M_IP6OPT, M_WAITOK); *pktopt = NULL; if (!m || m->m_len == 0) { /* * Only turning off any previous options, regardless of * whether the opt is just created or given. */ free(opt, M_IP6OPT); return (0); } /* set options specified by user. */ if ((error = ip6_setpktopts(m, opt, NULL, (td != NULL) ? td->td_ucred : NULL, so->so_proto->pr_protocol)) != 0) { ip6_clearpktopts(opt, -1); /* XXX: discard all options */ free(opt, M_IP6OPT); return (error); } *pktopt = opt; return (0); } /* * initialize ip6_pktopts. beware that there are non-zero default values in * the struct. */ void ip6_initpktopts(struct ip6_pktopts *opt) { bzero(opt, sizeof(*opt)); opt->ip6po_hlim = -1; /* -1 means default hop limit */ opt->ip6po_tclass = -1; /* -1 means default traffic class */ opt->ip6po_minmtu = IP6PO_MINMTU_MCASTONLY; opt->ip6po_prefer_tempaddr = IP6PO_TEMPADDR_SYSTEM; } static int ip6_pcbopt(int optname, u_char *buf, int len, struct ip6_pktopts **pktopt, struct ucred *cred, int uproto) { struct ip6_pktopts *opt; if (*pktopt == NULL) { *pktopt = malloc(sizeof(struct ip6_pktopts), M_IP6OPT, M_WAITOK); ip6_initpktopts(*pktopt); } opt = *pktopt; return (ip6_setpktopt(optname, buf, len, opt, cred, 1, 0, uproto)); } static int ip6_getpcbopt(struct ip6_pktopts *pktopt, int optname, struct sockopt *sopt) { void *optdata = NULL; int optdatalen = 0; struct ip6_ext *ip6e; int error = 0; struct in6_pktinfo null_pktinfo; int deftclass = 0, on; int defminmtu = IP6PO_MINMTU_MCASTONLY; int defpreftemp = IP6PO_TEMPADDR_SYSTEM; switch (optname) { case IPV6_PKTINFO: if (pktopt && pktopt->ip6po_pktinfo) optdata = (void *)pktopt->ip6po_pktinfo; else { /* XXX: we don't have to do this every time... */ bzero(&null_pktinfo, sizeof(null_pktinfo)); optdata = (void *)&null_pktinfo; } optdatalen = sizeof(struct in6_pktinfo); break; case IPV6_TCLASS: if (pktopt && pktopt->ip6po_tclass >= 0) optdata = (void *)&pktopt->ip6po_tclass; else optdata = (void *)&deftclass; optdatalen = sizeof(int); break; case IPV6_HOPOPTS: if (pktopt && pktopt->ip6po_hbh) { optdata = (void *)pktopt->ip6po_hbh; ip6e = (struct ip6_ext *)pktopt->ip6po_hbh; optdatalen = (ip6e->ip6e_len + 1) << 3; } break; case IPV6_RTHDR: if (pktopt && pktopt->ip6po_rthdr) { optdata = (void *)pktopt->ip6po_rthdr; ip6e = (struct ip6_ext *)pktopt->ip6po_rthdr; optdatalen = (ip6e->ip6e_len + 1) << 3; } break; case IPV6_RTHDRDSTOPTS: if (pktopt && pktopt->ip6po_dest1) { optdata = (void *)pktopt->ip6po_dest1; ip6e = (struct ip6_ext *)pktopt->ip6po_dest1; optdatalen = (ip6e->ip6e_len + 1) << 3; } break; case IPV6_DSTOPTS: if (pktopt && pktopt->ip6po_dest2) { optdata = (void *)pktopt->ip6po_dest2; ip6e = (struct ip6_ext *)pktopt->ip6po_dest2; optdatalen = (ip6e->ip6e_len + 1) << 3; } break; case IPV6_NEXTHOP: if (pktopt && pktopt->ip6po_nexthop) { optdata = (void *)pktopt->ip6po_nexthop; optdatalen = pktopt->ip6po_nexthop->sa_len; } break; case IPV6_USE_MIN_MTU: if (pktopt) optdata = (void *)&pktopt->ip6po_minmtu; else optdata = (void *)&defminmtu; optdatalen = sizeof(int); break; case IPV6_DONTFRAG: if (pktopt && ((pktopt->ip6po_flags) & IP6PO_DONTFRAG)) on = 1; else on = 0; optdata = (void *)&on; optdatalen = sizeof(on); break; case IPV6_PREFER_TEMPADDR: if (pktopt) optdata = (void *)&pktopt->ip6po_prefer_tempaddr; else optdata = (void *)&defpreftemp; optdatalen = sizeof(int); break; default: /* should not happen */ #ifdef DIAGNOSTIC panic("ip6_getpcbopt: unexpected option\n"); #endif return (ENOPROTOOPT); } error = sooptcopyout(sopt, optdata, optdatalen); return (error); } void ip6_clearpktopts(struct ip6_pktopts *pktopt, int optname) { if (pktopt == NULL) return; if (optname == -1 || optname == IPV6_PKTINFO) { if (pktopt->ip6po_pktinfo) free(pktopt->ip6po_pktinfo, M_IP6OPT); pktopt->ip6po_pktinfo = NULL; } if (optname == -1 || optname == IPV6_HOPLIMIT) pktopt->ip6po_hlim = -1; if (optname == -1 || optname == IPV6_TCLASS) pktopt->ip6po_tclass = -1; if (optname == -1 || optname == IPV6_NEXTHOP) { if (pktopt->ip6po_nextroute.ro_rt) { RTFREE(pktopt->ip6po_nextroute.ro_rt); pktopt->ip6po_nextroute.ro_rt = NULL; } if (pktopt->ip6po_nexthop) free(pktopt->ip6po_nexthop, M_IP6OPT); pktopt->ip6po_nexthop = NULL; } if (optname == -1 || optname == IPV6_HOPOPTS) { if (pktopt->ip6po_hbh) free(pktopt->ip6po_hbh, M_IP6OPT); pktopt->ip6po_hbh = NULL; } if (optname == -1 || optname == IPV6_RTHDRDSTOPTS) { if (pktopt->ip6po_dest1) free(pktopt->ip6po_dest1, M_IP6OPT); pktopt->ip6po_dest1 = NULL; } if (optname == -1 || optname == IPV6_RTHDR) { if (pktopt->ip6po_rhinfo.ip6po_rhi_rthdr) free(pktopt->ip6po_rhinfo.ip6po_rhi_rthdr, M_IP6OPT); pktopt->ip6po_rhinfo.ip6po_rhi_rthdr = NULL; if (pktopt->ip6po_route.ro_rt) { RTFREE(pktopt->ip6po_route.ro_rt); pktopt->ip6po_route.ro_rt = NULL; } } if (optname == -1 || optname == IPV6_DSTOPTS) { if (pktopt->ip6po_dest2) free(pktopt->ip6po_dest2, M_IP6OPT); pktopt->ip6po_dest2 = NULL; } } #define PKTOPT_EXTHDRCPY(type) \ do {\ if (src->type) {\ int hlen = (((struct ip6_ext *)src->type)->ip6e_len + 1) << 3;\ dst->type = malloc(hlen, M_IP6OPT, canwait);\ if (dst->type == NULL && canwait == M_NOWAIT)\ goto bad;\ bcopy(src->type, dst->type, hlen);\ }\ } while (/*CONSTCOND*/ 0) static int copypktopts(struct ip6_pktopts *dst, struct ip6_pktopts *src, int canwait) { if (dst == NULL || src == NULL) { printf("ip6_clearpktopts: invalid argument\n"); return (EINVAL); } dst->ip6po_hlim = src->ip6po_hlim; dst->ip6po_tclass = src->ip6po_tclass; dst->ip6po_flags = src->ip6po_flags; dst->ip6po_minmtu = src->ip6po_minmtu; dst->ip6po_prefer_tempaddr = src->ip6po_prefer_tempaddr; if (src->ip6po_pktinfo) { dst->ip6po_pktinfo = malloc(sizeof(*dst->ip6po_pktinfo), M_IP6OPT, canwait); if (dst->ip6po_pktinfo == NULL) goto bad; *dst->ip6po_pktinfo = *src->ip6po_pktinfo; } if (src->ip6po_nexthop) { dst->ip6po_nexthop = malloc(src->ip6po_nexthop->sa_len, M_IP6OPT, canwait); if (dst->ip6po_nexthop == NULL) goto bad; bcopy(src->ip6po_nexthop, dst->ip6po_nexthop, src->ip6po_nexthop->sa_len); } PKTOPT_EXTHDRCPY(ip6po_hbh); PKTOPT_EXTHDRCPY(ip6po_dest1); PKTOPT_EXTHDRCPY(ip6po_dest2); PKTOPT_EXTHDRCPY(ip6po_rthdr); /* not copy the cached route */ return (0); bad: ip6_clearpktopts(dst, -1); return (ENOBUFS); } #undef PKTOPT_EXTHDRCPY struct ip6_pktopts * ip6_copypktopts(struct ip6_pktopts *src, int canwait) { int error; struct ip6_pktopts *dst; dst = malloc(sizeof(*dst), M_IP6OPT, canwait); if (dst == NULL) return (NULL); ip6_initpktopts(dst); if ((error = copypktopts(dst, src, canwait)) != 0) { free(dst, M_IP6OPT); return (NULL); } return (dst); } void ip6_freepcbopts(struct ip6_pktopts *pktopt) { if (pktopt == NULL) return; ip6_clearpktopts(pktopt, -1); free(pktopt, M_IP6OPT); } /* * Set IPv6 outgoing packet options based on advanced API. */ int ip6_setpktopts(struct mbuf *control, struct ip6_pktopts *opt, struct ip6_pktopts *stickyopt, struct ucred *cred, int uproto) { struct cmsghdr *cm = 0; if (control == NULL || opt == NULL) return (EINVAL); ip6_initpktopts(opt); if (stickyopt) { int error; /* * If stickyopt is provided, make a local copy of the options * for this particular packet, then override them by ancillary * objects. * XXX: copypktopts() does not copy the cached route to a next * hop (if any). This is not very good in terms of efficiency, * but we can allow this since this option should be rarely * used. */ if ((error = copypktopts(opt, stickyopt, M_NOWAIT)) != 0) return (error); } /* * XXX: Currently, we assume all the optional information is stored * in a single mbuf. */ if (control->m_next) return (EINVAL); for (; control->m_len > 0; control->m_data += CMSG_ALIGN(cm->cmsg_len), control->m_len -= CMSG_ALIGN(cm->cmsg_len)) { int error; if (control->m_len < CMSG_LEN(0)) return (EINVAL); cm = mtod(control, struct cmsghdr *); if (cm->cmsg_len == 0 || cm->cmsg_len > control->m_len) return (EINVAL); if (cm->cmsg_level != IPPROTO_IPV6) continue; error = ip6_setpktopt(cm->cmsg_type, CMSG_DATA(cm), cm->cmsg_len - CMSG_LEN(0), opt, cred, 0, 1, uproto); if (error) return (error); } return (0); } /* * Set a particular packet option, as a sticky option or an ancillary data * item. "len" can be 0 only when it's a sticky option. * We have 4 cases of combination of "sticky" and "cmsg": * "sticky=0, cmsg=0": impossible * "sticky=0, cmsg=1": RFC2292 or RFC3542 ancillary data * "sticky=1, cmsg=0": RFC3542 socket option * "sticky=1, cmsg=1": RFC2292 socket option */ static int ip6_setpktopt(int optname, u_char *buf, int len, struct ip6_pktopts *opt, struct ucred *cred, int sticky, int cmsg, int uproto) { int minmtupolicy, preftemp; int error; if (!sticky && !cmsg) { #ifdef DIAGNOSTIC printf("ip6_setpktopt: impossible case\n"); #endif return (EINVAL); } /* * IPV6_2292xxx is for backward compatibility to RFC2292, and should * not be specified in the context of RFC3542. Conversely, * RFC3542 types should not be specified in the context of RFC2292. */ if (!cmsg) { switch (optname) { case IPV6_2292PKTINFO: case IPV6_2292HOPLIMIT: case IPV6_2292NEXTHOP: case IPV6_2292HOPOPTS: case IPV6_2292DSTOPTS: case IPV6_2292RTHDR: case IPV6_2292PKTOPTIONS: return (ENOPROTOOPT); } } if (sticky && cmsg) { switch (optname) { case IPV6_PKTINFO: case IPV6_HOPLIMIT: case IPV6_NEXTHOP: case IPV6_HOPOPTS: case IPV6_DSTOPTS: case IPV6_RTHDRDSTOPTS: case IPV6_RTHDR: case IPV6_USE_MIN_MTU: case IPV6_DONTFRAG: case IPV6_TCLASS: case IPV6_PREFER_TEMPADDR: /* XXX: not an RFC3542 option */ return (ENOPROTOOPT); } } switch (optname) { case IPV6_2292PKTINFO: case IPV6_PKTINFO: { struct ifnet *ifp = NULL; struct in6_pktinfo *pktinfo; if (len != sizeof(struct in6_pktinfo)) return (EINVAL); pktinfo = (struct in6_pktinfo *)buf; /* * An application can clear any sticky IPV6_PKTINFO option by * doing a "regular" setsockopt with ipi6_addr being * in6addr_any and ipi6_ifindex being zero. * [RFC 3542, Section 6] */ if (optname == IPV6_PKTINFO && opt->ip6po_pktinfo && pktinfo->ipi6_ifindex == 0 && IN6_IS_ADDR_UNSPECIFIED(&pktinfo->ipi6_addr)) { ip6_clearpktopts(opt, optname); break; } if (uproto == IPPROTO_TCP && optname == IPV6_PKTINFO && sticky && !IN6_IS_ADDR_UNSPECIFIED(&pktinfo->ipi6_addr)) { return (EINVAL); } if (IN6_IS_ADDR_MULTICAST(&pktinfo->ipi6_addr)) return (EINVAL); /* validate the interface index if specified. */ if (pktinfo->ipi6_ifindex > V_if_index) return (ENXIO); if (pktinfo->ipi6_ifindex) { ifp = ifnet_byindex(pktinfo->ipi6_ifindex); if (ifp == NULL) return (ENXIO); } if (ifp != NULL && ( ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)) return (ENETDOWN); if (ifp != NULL && !IN6_IS_ADDR_UNSPECIFIED(&pktinfo->ipi6_addr)) { struct in6_ifaddr *ia; ia = in6ifa_ifpwithaddr(ifp, &pktinfo->ipi6_addr); if (ia == NULL) return (EADDRNOTAVAIL); ifa_free(&ia->ia_ifa); } /* * We store the address anyway, and let in6_selectsrc() * validate the specified address. This is because ipi6_addr * may not have enough information about its scope zone, and * we may need additional information (such as outgoing * interface or the scope zone of a destination address) to * disambiguate the scope. * XXX: the delay of the validation may confuse the * application when it is used as a sticky option. */ if (opt->ip6po_pktinfo == NULL) { opt->ip6po_pktinfo = malloc(sizeof(*pktinfo), M_IP6OPT, M_NOWAIT); if (opt->ip6po_pktinfo == NULL) return (ENOBUFS); } bcopy(pktinfo, opt->ip6po_pktinfo, sizeof(*pktinfo)); break; } case IPV6_2292HOPLIMIT: case IPV6_HOPLIMIT: { int *hlimp; /* * RFC 3542 deprecated the usage of sticky IPV6_HOPLIMIT * to simplify the ordering among hoplimit options. */ if (optname == IPV6_HOPLIMIT && sticky) return (ENOPROTOOPT); if (len != sizeof(int)) return (EINVAL); hlimp = (int *)buf; if (*hlimp < -1 || *hlimp > 255) return (EINVAL); opt->ip6po_hlim = *hlimp; break; } case IPV6_TCLASS: { int tclass; if (len != sizeof(int)) return (EINVAL); tclass = *(int *)buf; if (tclass < -1 || tclass > 255) return (EINVAL); opt->ip6po_tclass = tclass; break; } case IPV6_2292NEXTHOP: case IPV6_NEXTHOP: if (cred != NULL) { error = priv_check_cred(cred, PRIV_NETINET_SETHDROPTS, 0); if (error) return (error); } if (len == 0) { /* just remove the option */ ip6_clearpktopts(opt, IPV6_NEXTHOP); break; } /* check if cmsg_len is large enough for sa_len */ if (len < sizeof(struct sockaddr) || len < *buf) return (EINVAL); switch (((struct sockaddr *)buf)->sa_family) { case AF_INET6: { struct sockaddr_in6 *sa6 = (struct sockaddr_in6 *)buf; int error; if (sa6->sin6_len != sizeof(struct sockaddr_in6)) return (EINVAL); if (IN6_IS_ADDR_UNSPECIFIED(&sa6->sin6_addr) || IN6_IS_ADDR_MULTICAST(&sa6->sin6_addr)) { return (EINVAL); } if ((error = sa6_embedscope(sa6, V_ip6_use_defzone)) != 0) { return (error); } break; } case AF_LINK: /* should eventually be supported */ default: return (EAFNOSUPPORT); } /* turn off the previous option, then set the new option. */ ip6_clearpktopts(opt, IPV6_NEXTHOP); opt->ip6po_nexthop = malloc(*buf, M_IP6OPT, M_NOWAIT); if (opt->ip6po_nexthop == NULL) return (ENOBUFS); bcopy(buf, opt->ip6po_nexthop, *buf); break; case IPV6_2292HOPOPTS: case IPV6_HOPOPTS: { struct ip6_hbh *hbh; int hbhlen; /* * XXX: We don't allow a non-privileged user to set ANY HbH * options, since per-option restriction has too much * overhead. */ if (cred != NULL) { error = priv_check_cred(cred, PRIV_NETINET_SETHDROPTS, 0); if (error) return (error); } if (len == 0) { ip6_clearpktopts(opt, IPV6_HOPOPTS); break; /* just remove the option */ } /* message length validation */ if (len < sizeof(struct ip6_hbh)) return (EINVAL); hbh = (struct ip6_hbh *)buf; hbhlen = (hbh->ip6h_len + 1) << 3; if (len != hbhlen) return (EINVAL); /* turn off the previous option, then set the new option. */ ip6_clearpktopts(opt, IPV6_HOPOPTS); opt->ip6po_hbh = malloc(hbhlen, M_IP6OPT, M_NOWAIT); if (opt->ip6po_hbh == NULL) return (ENOBUFS); bcopy(hbh, opt->ip6po_hbh, hbhlen); break; } case IPV6_2292DSTOPTS: case IPV6_DSTOPTS: case IPV6_RTHDRDSTOPTS: { struct ip6_dest *dest, **newdest = NULL; int destlen; if (cred != NULL) { /* XXX: see the comment for IPV6_HOPOPTS */ error = priv_check_cred(cred, PRIV_NETINET_SETHDROPTS, 0); if (error) return (error); } if (len == 0) { ip6_clearpktopts(opt, optname); break; /* just remove the option */ } /* message length validation */ if (len < sizeof(struct ip6_dest)) return (EINVAL); dest = (struct ip6_dest *)buf; destlen = (dest->ip6d_len + 1) << 3; if (len != destlen) return (EINVAL); /* * Determine the position that the destination options header * should be inserted; before or after the routing header. */ switch (optname) { case IPV6_2292DSTOPTS: /* * The old advacned API is ambiguous on this point. * Our approach is to determine the position based * according to the existence of a routing header. * Note, however, that this depends on the order of the * extension headers in the ancillary data; the 1st * part of the destination options header must appear * before the routing header in the ancillary data, * too. * RFC3542 solved the ambiguity by introducing * separate ancillary data or option types. */ if (opt->ip6po_rthdr == NULL) newdest = &opt->ip6po_dest1; else newdest = &opt->ip6po_dest2; break; case IPV6_RTHDRDSTOPTS: newdest = &opt->ip6po_dest1; break; case IPV6_DSTOPTS: newdest = &opt->ip6po_dest2; break; } /* turn off the previous option, then set the new option. */ ip6_clearpktopts(opt, optname); *newdest = malloc(destlen, M_IP6OPT, M_NOWAIT); if (*newdest == NULL) return (ENOBUFS); bcopy(dest, *newdest, destlen); break; } case IPV6_2292RTHDR: case IPV6_RTHDR: { struct ip6_rthdr *rth; int rthlen; if (len == 0) { ip6_clearpktopts(opt, IPV6_RTHDR); break; /* just remove the option */ } /* message length validation */ if (len < sizeof(struct ip6_rthdr)) return (EINVAL); rth = (struct ip6_rthdr *)buf; rthlen = (rth->ip6r_len + 1) << 3; if (len != rthlen) return (EINVAL); switch (rth->ip6r_type) { case IPV6_RTHDR_TYPE_0: if (rth->ip6r_len == 0) /* must contain one addr */ return (EINVAL); if (rth->ip6r_len % 2) /* length must be even */ return (EINVAL); if (rth->ip6r_len / 2 != rth->ip6r_segleft) return (EINVAL); break; default: return (EINVAL); /* not supported */ } /* turn off the previous option */ ip6_clearpktopts(opt, IPV6_RTHDR); opt->ip6po_rthdr = malloc(rthlen, M_IP6OPT, M_NOWAIT); if (opt->ip6po_rthdr == NULL) return (ENOBUFS); bcopy(rth, opt->ip6po_rthdr, rthlen); break; } case IPV6_USE_MIN_MTU: if (len != sizeof(int)) return (EINVAL); minmtupolicy = *(int *)buf; if (minmtupolicy != IP6PO_MINMTU_MCASTONLY && minmtupolicy != IP6PO_MINMTU_DISABLE && minmtupolicy != IP6PO_MINMTU_ALL) { return (EINVAL); } opt->ip6po_minmtu = minmtupolicy; break; case IPV6_DONTFRAG: if (len != sizeof(int)) return (EINVAL); if (uproto == IPPROTO_TCP || *(int *)buf == 0) { /* * we ignore this option for TCP sockets. * (RFC3542 leaves this case unspecified.) */ opt->ip6po_flags &= ~IP6PO_DONTFRAG; } else opt->ip6po_flags |= IP6PO_DONTFRAG; break; case IPV6_PREFER_TEMPADDR: if (len != sizeof(int)) return (EINVAL); preftemp = *(int *)buf; if (preftemp != IP6PO_TEMPADDR_SYSTEM && preftemp != IP6PO_TEMPADDR_NOTPREFER && preftemp != IP6PO_TEMPADDR_PREFER) { return (EINVAL); } opt->ip6po_prefer_tempaddr = preftemp; break; default: return (ENOPROTOOPT); } /* end of switch */ return (0); } /* * Routine called from ip6_output() to loop back a copy of an IP6 multicast * packet to the input queue of a specified interface. Note that this * calls the output routine of the loopback "driver", but with an interface * pointer that might NOT be &loif -- easier than replicating that code here. */ void ip6_mloopback(struct ifnet *ifp, struct mbuf *m, struct sockaddr_in6 *dst) { struct mbuf *copym; struct ip6_hdr *ip6; copym = m_copy(m, 0, M_COPYALL); if (copym == NULL) return; /* * Make sure to deep-copy IPv6 header portion in case the data * is in an mbuf cluster, so that we can safely override the IPv6 * header portion later. */ if (!M_WRITABLE(copym) || copym->m_len < sizeof(struct ip6_hdr)) { copym = m_pullup(copym, sizeof(struct ip6_hdr)); if (copym == NULL) return; } #ifdef DIAGNOSTIC if (copym->m_len < sizeof(*ip6)) { m_freem(copym); return; } #endif ip6 = mtod(copym, struct ip6_hdr *); /* * clear embedded scope identifiers if necessary. * in6_clearscope will touch the addresses only when necessary. */ in6_clearscope(&ip6->ip6_src); in6_clearscope(&ip6->ip6_dst); (void)if_simloop(ifp, copym, dst->sin6_family, 0); } /* * Chop IPv6 header off from the payload. */ static int ip6_splithdr(struct mbuf *m, struct ip6_exthdrs *exthdrs) { struct mbuf *mh; struct ip6_hdr *ip6; ip6 = mtod(m, struct ip6_hdr *); if (m->m_len > sizeof(*ip6)) { mh = m_gethdr(M_NOWAIT, MT_DATA); if (mh == NULL) { m_freem(m); return ENOBUFS; } m_move_pkthdr(mh, m); M_ALIGN(mh, sizeof(*ip6)); m->m_len -= sizeof(*ip6); m->m_data += sizeof(*ip6); mh->m_next = m; m = mh; m->m_len = sizeof(*ip6); bcopy((caddr_t)ip6, mtod(m, caddr_t), sizeof(*ip6)); } exthdrs->ip6e_ip6 = m; return 0; } /* * Compute IPv6 extension header length. */ int ip6_optlen(struct inpcb *in6p) { int len; if (!in6p->in6p_outputopts) return 0; len = 0; #define elen(x) \ (((struct ip6_ext *)(x)) ? (((struct ip6_ext *)(x))->ip6e_len + 1) << 3 : 0) len += elen(in6p->in6p_outputopts->ip6po_hbh); if (in6p->in6p_outputopts->ip6po_rthdr) /* dest1 is valid with rthdr only */ len += elen(in6p->in6p_outputopts->ip6po_dest1); len += elen(in6p->in6p_outputopts->ip6po_rthdr); len += elen(in6p->in6p_outputopts->ip6po_dest2); return len; #undef elen } Index: projects/cxl_iscsi/sys/netinet6/ip6_var.h =================================================================== --- projects/cxl_iscsi/sys/netinet6/ip6_var.h (revision 279716) +++ projects/cxl_iscsi/sys/netinet6/ip6_var.h (revision 279717) @@ -1,426 +1,425 @@ /*- * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: ip6_var.h,v 1.62 2001/05/03 14:51:48 itojun Exp $ */ /*- * Copyright (c) 1982, 1986, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)ip_var.h 8.1 (Berkeley) 6/10/93 * $FreeBSD$ */ #ifndef _NETINET6_IP6_VAR_H_ #define _NETINET6_IP6_VAR_H_ /* * IP6 reassembly queue structure. Each fragment * being reassembled is attached to one of these structures. */ struct ip6q { struct ip6asfrag *ip6q_down; struct ip6asfrag *ip6q_up; u_int32_t ip6q_ident; u_int8_t ip6q_nxt; u_int8_t ip6q_ecn; u_int8_t ip6q_ttl; struct in6_addr ip6q_src, ip6q_dst; struct ip6q *ip6q_next; struct ip6q *ip6q_prev; int ip6q_unfrglen; /* len of unfragmentable part */ #ifdef notyet u_char *ip6q_nxtp; #endif int ip6q_nfrag; /* # of fragments */ struct label *ip6q_label; }; struct ip6asfrag { struct ip6asfrag *ip6af_down; struct ip6asfrag *ip6af_up; struct mbuf *ip6af_m; int ip6af_offset; /* offset in ip6af_m to next header */ int ip6af_frglen; /* fragmentable part length */ int ip6af_off; /* fragment offset */ u_int16_t ip6af_mff; /* more fragment bit in frag off */ }; #define IP6_REASS_MBUF(ip6af) (*(struct mbuf **)&((ip6af)->ip6af_m)) /* * Structure attached to inpcb.in6p_moptions and * passed to ip6_output when IPv6 multicast options are in use. * This structure is lazy-allocated. */ struct ip6_moptions { struct ifnet *im6o_multicast_ifp; /* ifp for outgoing multicasts */ u_char im6o_multicast_hlim; /* hoplimit for outgoing multicasts */ u_char im6o_multicast_loop; /* 1 >= hear sends if a member */ u_short im6o_num_memberships; /* no. memberships this socket */ u_short im6o_max_memberships; /* max memberships this socket */ struct in6_multi **im6o_membership; /* group memberships */ struct in6_mfilter *im6o_mfilters; /* source filters */ }; /* * Control options for outgoing packets */ /* Routing header related info */ struct ip6po_rhinfo { struct ip6_rthdr *ip6po_rhi_rthdr; /* Routing header */ struct route_in6 ip6po_rhi_route; /* Route to the 1st hop */ }; #define ip6po_rthdr ip6po_rhinfo.ip6po_rhi_rthdr #define ip6po_route ip6po_rhinfo.ip6po_rhi_route /* Nexthop related info */ struct ip6po_nhinfo { struct sockaddr *ip6po_nhi_nexthop; struct route_in6 ip6po_nhi_route; /* Route to the nexthop */ }; #define ip6po_nexthop ip6po_nhinfo.ip6po_nhi_nexthop #define ip6po_nextroute ip6po_nhinfo.ip6po_nhi_route struct ip6_pktopts { struct mbuf *ip6po_m; /* Pointer to mbuf storing the data */ int ip6po_hlim; /* Hoplimit for outgoing packets */ /* Outgoing IF/address information */ struct in6_pktinfo *ip6po_pktinfo; /* Next-hop address information */ struct ip6po_nhinfo ip6po_nhinfo; struct ip6_hbh *ip6po_hbh; /* Hop-by-Hop options header */ /* Destination options header (before a routing header) */ struct ip6_dest *ip6po_dest1; /* Routing header related info. */ struct ip6po_rhinfo ip6po_rhinfo; /* Destination options header (after a routing header) */ struct ip6_dest *ip6po_dest2; int ip6po_tclass; /* traffic class */ int ip6po_minmtu; /* fragment vs PMTU discovery policy */ #define IP6PO_MINMTU_MCASTONLY -1 /* default; send at min MTU for multicast*/ #define IP6PO_MINMTU_DISABLE 0 /* always perform pmtu disc */ #define IP6PO_MINMTU_ALL 1 /* always send at min MTU */ int ip6po_prefer_tempaddr; /* whether temporary addresses are preferred as source address */ #define IP6PO_TEMPADDR_SYSTEM -1 /* follow the system default */ #define IP6PO_TEMPADDR_NOTPREFER 0 /* not prefer temporary address */ #define IP6PO_TEMPADDR_PREFER 1 /* prefer temporary address */ int ip6po_flags; #if 0 /* parameters in this block is obsolete. do not reuse the values. */ #define IP6PO_REACHCONF 0x01 /* upper-layer reachability confirmation. */ #define IP6PO_MINMTU 0x02 /* use minimum MTU (IPV6_USE_MIN_MTU) */ #endif #define IP6PO_DONTFRAG 0x04 /* disable fragmentation (IPV6_DONTFRAG) */ #define IP6PO_USECOA 0x08 /* use care of address */ }; /* * Control options for incoming packets */ struct ip6stat { uint64_t ip6s_total; /* total packets received */ uint64_t ip6s_tooshort; /* packet too short */ uint64_t ip6s_toosmall; /* not enough data */ uint64_t ip6s_fragments; /* fragments received */ uint64_t ip6s_fragdropped; /* frags dropped(dups, out of space) */ uint64_t ip6s_fragtimeout; /* fragments timed out */ uint64_t ip6s_fragoverflow; /* fragments that exceeded limit */ uint64_t ip6s_forward; /* packets forwarded */ uint64_t ip6s_cantforward; /* packets rcvd for unreachable dest */ uint64_t ip6s_redirectsent; /* packets forwarded on same net */ uint64_t ip6s_delivered; /* datagrams delivered to upper level*/ uint64_t ip6s_localout; /* total ip packets generated here */ uint64_t ip6s_odropped; /* lost packets due to nobufs, etc. */ uint64_t ip6s_reassembled; /* total packets reassembled ok */ uint64_t ip6s_fragmented; /* datagrams successfully fragmented */ uint64_t ip6s_ofragments; /* output fragments created */ uint64_t ip6s_cantfrag; /* don't fragment flag was set, etc. */ uint64_t ip6s_badoptions; /* error in option processing */ uint64_t ip6s_noroute; /* packets discarded due to no route */ uint64_t ip6s_badvers; /* ip6 version != 6 */ uint64_t ip6s_rawout; /* total raw ip packets generated */ uint64_t ip6s_badscope; /* scope error */ uint64_t ip6s_notmember; /* don't join this multicast group */ #define IP6S_HDRCNT 256 /* headers count */ uint64_t ip6s_nxthist[IP6S_HDRCNT]; /* next header history */ uint64_t ip6s_m1; /* one mbuf */ #define IP6S_M2MMAX 32 uint64_t ip6s_m2m[IP6S_M2MMAX]; /* two or more mbuf */ uint64_t ip6s_mext1; /* one ext mbuf */ uint64_t ip6s_mext2m; /* two or more ext mbuf */ uint64_t ip6s_exthdrtoolong; /* ext hdr are not contiguous */ uint64_t ip6s_nogif; /* no match gif found */ uint64_t ip6s_toomanyhdr; /* discarded due to too many headers */ /* * statistics for improvement of the source address selection * algorithm: * XXX: hardcoded 16 = # of ip6 multicast scope types + 1 */ #define IP6S_RULESMAX 16 #define IP6S_SCOPECNT 16 /* number of times that address selection fails */ uint64_t ip6s_sources_none; /* number of times that an address on the outgoing I/F is chosen */ uint64_t ip6s_sources_sameif[IP6S_SCOPECNT]; /* number of times that an address on a non-outgoing I/F is chosen */ uint64_t ip6s_sources_otherif[IP6S_SCOPECNT]; /* * number of times that an address that has the same scope * from the destination is chosen. */ uint64_t ip6s_sources_samescope[IP6S_SCOPECNT]; /* * number of times that an address that has a different scope * from the destination is chosen. */ uint64_t ip6s_sources_otherscope[IP6S_SCOPECNT]; /* number of times that a deprecated address is chosen */ uint64_t ip6s_sources_deprecated[IP6S_SCOPECNT]; /* number of times that each rule of source selection is applied. */ uint64_t ip6s_sources_rule[IP6S_RULESMAX]; }; #ifdef _KERNEL #include VNET_PCPUSTAT_DECLARE(struct ip6stat, ip6stat); #define IP6STAT_ADD(name, val) \ VNET_PCPUSTAT_ADD(struct ip6stat, ip6stat, name, (val)) #define IP6STAT_SUB(name, val) IP6STAT_ADD(name, -(val)) #define IP6STAT_INC(name) IP6STAT_ADD(name, 1) #define IP6STAT_DEC(name) IP6STAT_SUB(name, 1) #endif #ifdef _KERNEL /* flags passed to ip6_output as last parameter */ #define IPV6_UNSPECSRC 0x01 /* allow :: as the source address */ #define IPV6_FORWARDING 0x02 /* most of IPv6 header exists */ #define IPV6_MINMTU 0x04 /* use minimum MTU (IPV6_USE_MIN_MTU) */ #ifdef __NO_STRICT_ALIGNMENT #define IP6_HDR_ALIGNED_P(ip) 1 #else #define IP6_HDR_ALIGNED_P(ip) ((((intptr_t) (ip)) & 3) == 0) #endif VNET_DECLARE(int, ip6_defhlim); /* default hop limit */ VNET_DECLARE(int, ip6_defmcasthlim); /* default multicast hop limit */ VNET_DECLARE(int, ip6_forwarding); /* act as router? */ VNET_DECLARE(int, ip6_use_deprecated); /* allow deprecated addr as source */ VNET_DECLARE(int, ip6_rr_prune); /* router renumbering prefix * walk list every 5 sec. */ VNET_DECLARE(int, ip6_mcast_pmtu); /* enable pMTU discovery for multicast? */ VNET_DECLARE(int, ip6_v6only); #define V_ip6_defhlim VNET(ip6_defhlim) #define V_ip6_defmcasthlim VNET(ip6_defmcasthlim) #define V_ip6_forwarding VNET(ip6_forwarding) #define V_ip6_use_deprecated VNET(ip6_use_deprecated) #define V_ip6_rr_prune VNET(ip6_rr_prune) #define V_ip6_mcast_pmtu VNET(ip6_mcast_pmtu) #define V_ip6_v6only VNET(ip6_v6only) VNET_DECLARE(struct socket *, ip6_mrouter); /* multicast routing daemon */ VNET_DECLARE(int, ip6_sendredirects); /* send IP redirects when forwarding? */ VNET_DECLARE(int, ip6_maxfragpackets); /* Maximum packets in reassembly * queue */ VNET_DECLARE(int, ip6_maxfrags); /* Maximum fragments in reassembly * queue */ VNET_DECLARE(int, ip6_accept_rtadv); /* Acts as a host not a router */ VNET_DECLARE(int, ip6_no_radr); /* No defroute from RA */ VNET_DECLARE(int, ip6_norbit_raif); /* Disable R-bit in NA on RA * receiving IF. */ VNET_DECLARE(int, ip6_rfc6204w3); /* Accept defroute from RA even when forwarding enabled */ VNET_DECLARE(int, ip6_log_interval); VNET_DECLARE(time_t, ip6_log_time); VNET_DECLARE(int, ip6_hdrnestlimit); /* upper limit of # of extension * headers */ VNET_DECLARE(int, ip6_dad_count); /* DupAddrDetectionTransmits */ #define V_ip6_mrouter VNET(ip6_mrouter) #define V_ip6_sendredirects VNET(ip6_sendredirects) #define V_ip6_maxfragpackets VNET(ip6_maxfragpackets) #define V_ip6_maxfrags VNET(ip6_maxfrags) #define V_ip6_accept_rtadv VNET(ip6_accept_rtadv) #define V_ip6_no_radr VNET(ip6_no_radr) #define V_ip6_norbit_raif VNET(ip6_norbit_raif) #define V_ip6_rfc6204w3 VNET(ip6_rfc6204w3) #define V_ip6_log_interval VNET(ip6_log_interval) #define V_ip6_log_time VNET(ip6_log_time) #define V_ip6_hdrnestlimit VNET(ip6_hdrnestlimit) #define V_ip6_dad_count VNET(ip6_dad_count) VNET_DECLARE(int, ip6_auto_flowlabel); VNET_DECLARE(int, ip6_auto_linklocal); #define V_ip6_auto_flowlabel VNET(ip6_auto_flowlabel) #define V_ip6_auto_linklocal VNET(ip6_auto_linklocal) VNET_DECLARE(int, ip6_use_tempaddr); /* Whether to use temporary addresses */ VNET_DECLARE(int, ip6_prefer_tempaddr); /* Whether to prefer temporary * addresses in the source address * selection */ #define V_ip6_use_tempaddr VNET(ip6_use_tempaddr) #define V_ip6_prefer_tempaddr VNET(ip6_prefer_tempaddr) VNET_DECLARE(int, ip6_use_defzone); /* Whether to use the default scope * zone when unspecified */ #define V_ip6_use_defzone VNET(ip6_use_defzone) VNET_DECLARE (struct pfil_head, inet6_pfil_hook); /* packet filter hooks */ #define V_inet6_pfil_hook VNET(inet6_pfil_hook) #ifdef IPSTEALTH VNET_DECLARE(int, ip6stealth); #define V_ip6stealth VNET(ip6stealth) #endif extern struct pr_usrreqs rip6_usrreqs; struct sockopt; struct inpcb; int icmp6_ctloutput(struct socket *, struct sockopt *sopt); struct in6_ifaddr; void ip6_init(void); #ifdef VIMAGE void ip6_destroy(void); #endif int ip6proto_register(short); int ip6proto_unregister(short); void ip6_input(struct mbuf *); void ip6_freepcbopts(struct ip6_pktopts *); int ip6_unknown_opt(u_int8_t *, struct mbuf *, int); char * ip6_get_prevhdr(struct mbuf *, int); int ip6_nexthdr(struct mbuf *, int, int, int *); int ip6_lasthdr(struct mbuf *, int, int, int *); extern int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *); int ip6_process_hopopts(struct mbuf *, u_int8_t *, int, u_int32_t *, u_int32_t *); struct mbuf **ip6_savecontrol_v4(struct inpcb *, struct mbuf *, struct mbuf **, int *); void ip6_savecontrol(struct inpcb *, struct mbuf *, struct mbuf **); -void ip6_notify_pmtu(struct inpcb *, struct sockaddr_in6 *, - u_int32_t *); +void ip6_notify_pmtu(struct inpcb *, struct sockaddr_in6 *, u_int32_t); int ip6_sysctl(int *, u_int, void *, size_t *, void *, size_t); void ip6_forward(struct mbuf *, int); void ip6_mloopback(struct ifnet *, struct mbuf *, struct sockaddr_in6 *); int ip6_output(struct mbuf *, struct ip6_pktopts *, struct route_in6 *, int, struct ip6_moptions *, struct ifnet **, struct inpcb *); int ip6_ctloutput(struct socket *, struct sockopt *); int ip6_raw_ctloutput(struct socket *, struct sockopt *); void ip6_initpktopts(struct ip6_pktopts *); int ip6_setpktopts(struct mbuf *, struct ip6_pktopts *, struct ip6_pktopts *, struct ucred *, int); void ip6_clearpktopts(struct ip6_pktopts *, int); struct ip6_pktopts *ip6_copypktopts(struct ip6_pktopts *, int); int ip6_optlen(struct inpcb *); int ip6_deletefraghdr(struct mbuf *, int, int); int ip6_fragment(struct ifnet *, struct mbuf *, int, u_char, int); int route6_input(struct mbuf **, int *, int); void frag6_init(void); int frag6_input(struct mbuf **, int *, int); void frag6_slowtimo(void); void frag6_drain(void); void rip6_init(void); int rip6_input(struct mbuf **, int *, int); void rip6_ctlinput(int, struct sockaddr *, void *); int rip6_ctloutput(struct socket *, struct sockopt *); int rip6_output(struct mbuf *, struct socket *, ...); int rip6_usrreq(struct socket *, int, struct mbuf *, struct mbuf *, struct mbuf *, struct thread *); int dest6_input(struct mbuf **, int *, int); int none_input(struct mbuf **, int *, int); int in6_selectsrc(struct sockaddr_in6 *, struct ip6_pktopts *, struct inpcb *inp, struct route_in6 *, struct ucred *cred, struct ifnet **, struct in6_addr *); int in6_selectroute(struct sockaddr_in6 *, struct ip6_pktopts *, struct ip6_moptions *, struct route_in6 *, struct ifnet **, struct rtentry **); int in6_selectroute_fib(struct sockaddr_in6 *, struct ip6_pktopts *, struct ip6_moptions *, struct route_in6 *, struct ifnet **, struct rtentry **, u_int); u_int32_t ip6_randomid(void); u_int32_t ip6_randomflowlabel(void); void in6_delayed_cksum(struct mbuf *m, uint32_t plen, u_short offset); #endif /* _KERNEL */ #endif /* !_NETINET6_IP6_VAR_H_ */ Index: projects/cxl_iscsi/sys/netinet6/nd6.h =================================================================== --- projects/cxl_iscsi/sys/netinet6/nd6.h (revision 279716) +++ projects/cxl_iscsi/sys/netinet6/nd6.h (revision 279717) @@ -1,464 +1,466 @@ /*- * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: nd6.h,v 1.76 2001/12/18 02:10:31 itojun Exp $ * $FreeBSD$ */ #ifndef _NETINET6_ND6_H_ #define _NETINET6_ND6_H_ /* see net/route.h, or net/if_inarp.h */ #ifndef RTF_ANNOUNCE #define RTF_ANNOUNCE RTF_PROTO2 #endif #include #include struct llentry; #define ND6_LLINFO_NOSTATE -2 /* * We don't need the WAITDELETE state any more, but we keep the definition * in a comment line instead of removing it. This is necessary to avoid * unintentionally reusing the value for another purpose, which might * affect backward compatibility with old applications. * (20000711 jinmei@kame.net) */ /* #define ND6_LLINFO_WAITDELETE -1 */ #define ND6_LLINFO_INCOMPLETE 0 #define ND6_LLINFO_REACHABLE 1 #define ND6_LLINFO_STALE 2 #define ND6_LLINFO_DELAY 3 #define ND6_LLINFO_PROBE 4 #define ND6_IS_LLINFO_PROBREACH(n) ((n)->ln_state > ND6_LLINFO_INCOMPLETE) #define ND6_LLINFO_PERMANENT(n) (((n)->la_expire == 0) && ((n)->ln_state > ND6_LLINFO_INCOMPLETE)) struct nd_ifinfo { u_int32_t linkmtu; /* LinkMTU */ u_int32_t maxmtu; /* Upper bound of LinkMTU */ u_int32_t basereachable; /* BaseReachableTime */ u_int32_t reachable; /* Reachable Time */ u_int32_t retrans; /* Retrans Timer */ u_int32_t flags; /* Flags */ int recalctm; /* BaseReacable re-calculation timer */ u_int8_t chlim; /* CurHopLimit */ u_int8_t initialized; /* Flag to see the entry is initialized */ /* the following 3 members are for privacy extension for addrconf */ u_int8_t randomseed0[8]; /* upper 64 bits of MD5 digest */ u_int8_t randomseed1[8]; /* lower 64 bits (usually the EUI64 IFID) */ u_int8_t randomid[8]; /* current random ID */ }; #define ND6_IFF_PERFORMNUD 0x1 #define ND6_IFF_ACCEPT_RTADV 0x2 #define ND6_IFF_PREFER_SOURCE 0x4 /* Not used in FreeBSD. */ #define ND6_IFF_IFDISABLED 0x8 /* IPv6 operation is disabled due to * DAD failure. (XXX: not ND-specific) */ #define ND6_IFF_DONT_SET_IFROUTE 0x10 #define ND6_IFF_AUTO_LINKLOCAL 0x20 #define ND6_IFF_NO_RADR 0x40 #define ND6_IFF_NO_PREFER_IFACE 0x80 /* XXX: not related to ND. */ +#define ND6_IFF_IGNORELOOP 0x100 +#define ND6_IFF_NO_DAD 0x200 #define ND6_CREATE LLE_CREATE #define ND6_EXCLUSIVE LLE_EXCLUSIVE #ifdef _KERNEL #define ND_IFINFO(ifp) \ (((struct in6_ifextra *)(ifp)->if_afdata[AF_INET6])->nd_ifinfo) #define IN6_LINKMTU(ifp) \ ((ND_IFINFO(ifp)->linkmtu && ND_IFINFO(ifp)->linkmtu < (ifp)->if_mtu) \ ? ND_IFINFO(ifp)->linkmtu \ : ((ND_IFINFO(ifp)->maxmtu && ND_IFINFO(ifp)->maxmtu < (ifp)->if_mtu) \ ? ND_IFINFO(ifp)->maxmtu : (ifp)->if_mtu)) #endif struct in6_nbrinfo { char ifname[IFNAMSIZ]; /* if name, e.g. "en0" */ struct in6_addr addr; /* IPv6 address of the neighbor */ long asked; /* number of queries already sent for this addr */ int isrouter; /* if it acts as a router */ int state; /* reachability state */ int expire; /* lifetime for NDP state transition */ }; #define DRLSTSIZ 10 #define PRLSTSIZ 10 struct in6_drlist { char ifname[IFNAMSIZ]; struct { struct in6_addr rtaddr; u_char flags; u_short rtlifetime; u_long expire; u_short if_index; } defrouter[DRLSTSIZ]; }; struct in6_defrouter { struct sockaddr_in6 rtaddr; u_char flags; u_short rtlifetime; u_long expire; u_short if_index; }; #ifdef _KERNEL struct in6_oprlist { char ifname[IFNAMSIZ]; struct { struct in6_addr prefix; struct prf_ra raflags; u_char prefixlen; u_char origin; u_long vltime; u_long pltime; u_long expire; u_short if_index; u_short advrtrs; /* number of advertisement routers */ struct in6_addr advrtr[DRLSTSIZ]; /* XXX: explicit limit */ } prefix[PRLSTSIZ]; }; #endif struct in6_prlist { char ifname[IFNAMSIZ]; struct { struct in6_addr prefix; struct prf_ra raflags; u_char prefixlen; u_char origin; u_int32_t vltime; u_int32_t pltime; time_t expire; u_short if_index; u_short advrtrs; /* number of advertisement routers */ struct in6_addr advrtr[DRLSTSIZ]; /* XXX: explicit limit */ } prefix[PRLSTSIZ]; }; struct in6_prefix { struct sockaddr_in6 prefix; struct prf_ra raflags; u_char prefixlen; u_char origin; u_int32_t vltime; u_int32_t pltime; time_t expire; u_int32_t flags; int refcnt; u_short if_index; u_short advrtrs; /* number of advertisement routers */ /* struct sockaddr_in6 advrtr[] */ }; #ifdef _KERNEL struct in6_ondireq { char ifname[IFNAMSIZ]; struct { u_int32_t linkmtu; /* LinkMTU */ u_int32_t maxmtu; /* Upper bound of LinkMTU */ u_int32_t basereachable; /* BaseReachableTime */ u_int32_t reachable; /* Reachable Time */ u_int32_t retrans; /* Retrans Timer */ u_int32_t flags; /* Flags */ int recalctm; /* BaseReacable re-calculation timer */ u_int8_t chlim; /* CurHopLimit */ u_int8_t receivedra; } ndi; }; #endif struct in6_ndireq { char ifname[IFNAMSIZ]; struct nd_ifinfo ndi; }; struct in6_ndifreq { char ifname[IFNAMSIZ]; u_long ifindex; }; /* Prefix status */ #define NDPRF_ONLINK 0x1 #define NDPRF_DETACHED 0x2 /* protocol constants */ #define MAX_RTR_SOLICITATION_DELAY 1 /* 1sec */ #define RTR_SOLICITATION_INTERVAL 4 /* 4sec */ #define MAX_RTR_SOLICITATIONS 3 #define ND6_INFINITE_LIFETIME 0xffffffff #ifdef _KERNEL /* node constants */ #define MAX_REACHABLE_TIME 3600000 /* msec */ #define REACHABLE_TIME 30000 /* msec */ #define RETRANS_TIMER 1000 /* msec */ #define MIN_RANDOM_FACTOR 512 /* 1024 * 0.5 */ #define MAX_RANDOM_FACTOR 1536 /* 1024 * 1.5 */ #define DEF_TEMP_VALID_LIFETIME 604800 /* 1 week */ #define DEF_TEMP_PREFERRED_LIFETIME 86400 /* 1 day */ #define TEMPADDR_REGEN_ADVANCE 5 /* sec */ #define MAX_TEMP_DESYNC_FACTOR 600 /* 10 min */ #define ND_COMPUTE_RTIME(x) \ (((MIN_RANDOM_FACTOR * (x >> 10)) + (arc4random() & \ ((MAX_RANDOM_FACTOR - MIN_RANDOM_FACTOR) * (x >> 10)))) /1000) TAILQ_HEAD(nd_drhead, nd_defrouter); struct nd_defrouter { TAILQ_ENTRY(nd_defrouter) dr_entry; struct in6_addr rtaddr; u_char flags; /* flags on RA message */ u_short rtlifetime; u_long expire; struct ifnet *ifp; int installed; /* is installed into kernel routing table */ }; struct nd_prefixctl { struct ifnet *ndpr_ifp; /* prefix */ struct sockaddr_in6 ndpr_prefix; u_char ndpr_plen; u_int32_t ndpr_vltime; /* advertised valid lifetime */ u_int32_t ndpr_pltime; /* advertised preferred lifetime */ struct prf_ra ndpr_flags; }; struct nd_prefix { struct ifnet *ndpr_ifp; LIST_ENTRY(nd_prefix) ndpr_entry; struct sockaddr_in6 ndpr_prefix; /* prefix */ struct in6_addr ndpr_mask; /* netmask derived from the prefix */ u_int32_t ndpr_vltime; /* advertised valid lifetime */ u_int32_t ndpr_pltime; /* advertised preferred lifetime */ time_t ndpr_expire; /* expiration time of the prefix */ time_t ndpr_preferred; /* preferred time of the prefix */ time_t ndpr_lastupdate; /* reception time of last advertisement */ struct prf_ra ndpr_flags; u_int32_t ndpr_stateflags; /* actual state flags */ /* list of routers that advertise the prefix: */ LIST_HEAD(pr_rtrhead, nd_pfxrouter) ndpr_advrtrs; u_char ndpr_plen; int ndpr_refcnt; /* reference couter from addresses */ }; #define ndpr_raf ndpr_flags #define ndpr_raf_onlink ndpr_flags.onlink #define ndpr_raf_auto ndpr_flags.autonomous #define ndpr_raf_router ndpr_flags.router /* * Message format for use in obtaining information about prefixes * from inet6 sysctl function */ struct inet6_ndpr_msghdr { u_short inpm_msglen; /* to skip over non-understood messages */ u_char inpm_version; /* future binary compatibility */ u_char inpm_type; /* message type */ struct in6_addr inpm_prefix; u_long prm_vltim; u_long prm_pltime; u_long prm_expire; u_long prm_preferred; struct in6_prflags prm_flags; u_short prm_index; /* index for associated ifp */ u_char prm_plen; /* length of prefix in bits */ }; #define prm_raf_onlink prm_flags.prf_ra.onlink #define prm_raf_auto prm_flags.prf_ra.autonomous #define prm_statef_onlink prm_flags.prf_state.onlink #define prm_rrf_decrvalid prm_flags.prf_rr.decrvalid #define prm_rrf_decrprefd prm_flags.prf_rr.decrprefd struct nd_pfxrouter { LIST_ENTRY(nd_pfxrouter) pfr_entry; struct nd_defrouter *router; }; LIST_HEAD(nd_prhead, nd_prefix); /* nd6.c */ VNET_DECLARE(int, nd6_prune); VNET_DECLARE(int, nd6_delay); VNET_DECLARE(int, nd6_umaxtries); VNET_DECLARE(int, nd6_mmaxtries); VNET_DECLARE(int, nd6_useloopback); VNET_DECLARE(int, nd6_maxnudhint); VNET_DECLARE(int, nd6_gctimer); VNET_DECLARE(struct nd_drhead, nd_defrouter); VNET_DECLARE(struct nd_prhead, nd_prefix); VNET_DECLARE(int, nd6_debug); VNET_DECLARE(int, nd6_onlink_ns_rfc4861); #define V_nd6_prune VNET(nd6_prune) #define V_nd6_delay VNET(nd6_delay) #define V_nd6_umaxtries VNET(nd6_umaxtries) #define V_nd6_mmaxtries VNET(nd6_mmaxtries) #define V_nd6_useloopback VNET(nd6_useloopback) #define V_nd6_maxnudhint VNET(nd6_maxnudhint) #define V_nd6_gctimer VNET(nd6_gctimer) #define V_nd_defrouter VNET(nd_defrouter) #define V_nd_prefix VNET(nd_prefix) #define V_nd6_debug VNET(nd6_debug) #define V_nd6_onlink_ns_rfc4861 VNET(nd6_onlink_ns_rfc4861) #define nd6log(x) do { if (V_nd6_debug) log x; } while (/*CONSTCOND*/ 0) VNET_DECLARE(struct callout, nd6_timer_ch); #define V_nd6_timer_ch VNET(nd6_timer_ch) /* nd6_rtr.c */ VNET_DECLARE(int, nd6_defifindex); VNET_DECLARE(int, ip6_desync_factor); /* seconds */ VNET_DECLARE(u_int32_t, ip6_temp_preferred_lifetime); /* seconds */ VNET_DECLARE(u_int32_t, ip6_temp_valid_lifetime); /* seconds */ VNET_DECLARE(int, ip6_temp_regen_advance); /* seconds */ #define V_nd6_defifindex VNET(nd6_defifindex) #define V_ip6_desync_factor VNET(ip6_desync_factor) #define V_ip6_temp_preferred_lifetime VNET(ip6_temp_preferred_lifetime) #define V_ip6_temp_valid_lifetime VNET(ip6_temp_valid_lifetime) #define V_ip6_temp_regen_advance VNET(ip6_temp_regen_advance) union nd_opts { struct nd_opt_hdr *nd_opt_array[16]; /* max = ND_OPT_NONCE */ struct { struct nd_opt_hdr *zero; struct nd_opt_hdr *src_lladdr; struct nd_opt_hdr *tgt_lladdr; struct nd_opt_prefix_info *pi_beg; /* multiple opts, start */ struct nd_opt_rd_hdr *rh; struct nd_opt_mtu *mtu; struct nd_opt_hdr *__res6; struct nd_opt_hdr *__res7; struct nd_opt_hdr *__res8; struct nd_opt_hdr *__res9; struct nd_opt_hdr *__res10; struct nd_opt_hdr *__res11; struct nd_opt_hdr *__res12; struct nd_opt_hdr *__res13; struct nd_opt_nonce *nonce; struct nd_opt_hdr *__res15; struct nd_opt_hdr *search; /* multiple opts */ struct nd_opt_hdr *last; /* multiple opts */ int done; struct nd_opt_prefix_info *pi_end;/* multiple opts, end */ } nd_opt_each; }; #define nd_opts_src_lladdr nd_opt_each.src_lladdr #define nd_opts_tgt_lladdr nd_opt_each.tgt_lladdr #define nd_opts_pi nd_opt_each.pi_beg #define nd_opts_pi_end nd_opt_each.pi_end #define nd_opts_rh nd_opt_each.rh #define nd_opts_mtu nd_opt_each.mtu #define nd_opts_nonce nd_opt_each.nonce #define nd_opts_search nd_opt_each.search #define nd_opts_last nd_opt_each.last #define nd_opts_done nd_opt_each.done /* XXX: need nd6_var.h?? */ /* nd6.c */ void nd6_init(void); #ifdef VIMAGE void nd6_destroy(void); #endif struct nd_ifinfo *nd6_ifattach(struct ifnet *); void nd6_ifdetach(struct nd_ifinfo *); int nd6_is_addr_neighbor(struct sockaddr_in6 *, struct ifnet *); void nd6_option_init(void *, int, union nd_opts *); struct nd_opt_hdr *nd6_option(union nd_opts *); int nd6_options(union nd_opts *); struct llentry *nd6_lookup(struct in6_addr *, int, struct ifnet *); void nd6_setmtu(struct ifnet *); void nd6_llinfo_settimer(struct llentry *, long); void nd6_llinfo_settimer_locked(struct llentry *, long); void nd6_timer(void *); void nd6_purge(struct ifnet *); void nd6_nud_hint(struct rtentry *, struct in6_addr *, int); int nd6_resolve(struct ifnet *, struct rtentry *, struct mbuf *, struct sockaddr *, u_char *); int nd6_ioctl(u_long, caddr_t, struct ifnet *); struct llentry *nd6_cache_lladdr(struct ifnet *, struct in6_addr *, char *, int, int, int); int nd6_output(struct ifnet *, struct ifnet *, struct mbuf *, struct sockaddr_in6 *, struct rtentry *); void nd6_grab_holdchain(struct llentry *, struct mbuf **, struct sockaddr_in6 *); int nd6_flush_holdchain(struct ifnet *, struct ifnet *, struct mbuf *, struct sockaddr_in6 *); int nd6_need_cache(struct ifnet *); int nd6_add_ifa_lle(struct in6_ifaddr *); void nd6_rem_ifa_lle(struct in6_ifaddr *); int nd6_storelladdr(struct ifnet *, struct mbuf *, const struct sockaddr *, u_char *, uint32_t *); /* nd6_nbr.c */ void nd6_na_input(struct mbuf *, int, int); void nd6_na_output(struct ifnet *, const struct in6_addr *, const struct in6_addr *, u_long, int, struct sockaddr *); void nd6_ns_input(struct mbuf *, int, int); void nd6_ns_output(struct ifnet *, const struct in6_addr *, const struct in6_addr *, struct llentry *, uint8_t *); caddr_t nd6_ifptomac(struct ifnet *); void nd6_dad_init(void); void nd6_dad_start(struct ifaddr *, int); void nd6_dad_stop(struct ifaddr *); /* nd6_rtr.c */ void nd6_rs_input(struct mbuf *, int, int); void nd6_ra_input(struct mbuf *, int, int); void prelist_del(struct nd_prefix *); void defrouter_reset(void); void defrouter_select(void); void defrtrlist_del(struct nd_defrouter *); void prelist_remove(struct nd_prefix *); int nd6_prelist_add(struct nd_prefixctl *, struct nd_defrouter *, struct nd_prefix **); void pfxlist_onlink_check(void); struct nd_defrouter *defrouter_lookup(struct in6_addr *, struct ifnet *); struct nd_prefix *nd6_prefix_lookup(struct nd_prefixctl *); void rt6_flush(struct in6_addr *, struct ifnet *); int nd6_setdefaultiface(int); int in6_tmpifadd(const struct in6_ifaddr *, int, int); #endif /* _KERNEL */ #endif /* _NETINET6_ND6_H_ */ Index: projects/cxl_iscsi/sys/netinet6/nd6_nbr.c =================================================================== --- projects/cxl_iscsi/sys/netinet6/nd6_nbr.c (revision 279716) +++ projects/cxl_iscsi/sys/netinet6/nd6_nbr.c (revision 279717) @@ -1,1598 +1,1648 @@ /*- * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: nd6_nbr.c,v 1.86 2002/01/21 02:33:04 jinmei Exp $ */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_mpath.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef RADIX_MPATH #include #endif #include #include #include #include #define L3_ADDR_SIN6(le) ((struct sockaddr_in6 *) L3_ADDR(le)) #include #include #include #include #include #include #include #include #include #define SDL(s) ((struct sockaddr_dl *)s) struct dadq; static struct dadq *nd6_dad_find(struct ifaddr *, struct nd_opt_nonce *); static void nd6_dad_add(struct dadq *dp); static void nd6_dad_del(struct dadq *dp); static void nd6_dad_rele(struct dadq *); static void nd6_dad_starttimer(struct dadq *, int); static void nd6_dad_stoptimer(struct dadq *); static void nd6_dad_timer(struct dadq *); static void nd6_dad_duplicated(struct ifaddr *, struct dadq *); static void nd6_dad_ns_output(struct dadq *, struct ifaddr *); static void nd6_dad_ns_input(struct ifaddr *, struct nd_opt_nonce *); static void nd6_dad_na_input(struct ifaddr *); static void nd6_na_output_fib(struct ifnet *, const struct in6_addr *, const struct in6_addr *, u_long, int, struct sockaddr *, u_int); +static void nd6_ns_output_fib(struct ifnet *, const struct in6_addr *, + const struct in6_addr *, struct llentry *, uint8_t *, u_int); static VNET_DEFINE(int, dad_enhanced) = 1; #define V_dad_enhanced VNET(dad_enhanced) SYSCTL_DECL(_net_inet6_ip6); SYSCTL_INT(_net_inet6_ip6, OID_AUTO, dad_enhanced, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(dad_enhanced), 0, "Enable Enhanced DAD, which adds a random nonce to NS messages for DAD."); static VNET_DEFINE(int, dad_maxtry) = 15; /* max # of *tries* to transmit DAD packet */ #define V_dad_maxtry VNET(dad_maxtry) /* * Input a Neighbor Solicitation Message. * * Based on RFC 2461 * Based on RFC 2462 (duplicate address detection) */ void nd6_ns_input(struct mbuf *m, int off, int icmp6len) { struct ifnet *ifp = m->m_pkthdr.rcvif; struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); struct nd_neighbor_solicit *nd_ns; struct in6_addr saddr6 = ip6->ip6_src; struct in6_addr daddr6 = ip6->ip6_dst; struct in6_addr taddr6; struct in6_addr myaddr6; char *lladdr = NULL; struct ifaddr *ifa = NULL; int lladdrlen = 0; int anycast = 0, proxy = 0, tentative = 0; int tlladdr; int rflag; union nd_opts ndopts; struct sockaddr_dl proxydl; char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; rflag = (V_ip6_forwarding) ? ND_NA_FLAG_ROUTER : 0; if (ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV && V_ip6_norbit_raif) rflag = 0; #ifndef PULLDOWN_TEST IP6_EXTHDR_CHECK(m, off, icmp6len,); nd_ns = (struct nd_neighbor_solicit *)((caddr_t)ip6 + off); #else IP6_EXTHDR_GET(nd_ns, struct nd_neighbor_solicit *, m, off, icmp6len); if (nd_ns == NULL) { ICMP6STAT_INC(icp6s_tooshort); return; } #endif ip6 = mtod(m, struct ip6_hdr *); /* adjust pointer for safety */ taddr6 = nd_ns->nd_ns_target; if (in6_setscope(&taddr6, ifp, NULL) != 0) goto bad; if (ip6->ip6_hlim != 255) { nd6log((LOG_ERR, "nd6_ns_input: invalid hlim (%d) from %s to %s on %s\n", ip6->ip6_hlim, ip6_sprintf(ip6bufs, &ip6->ip6_src), ip6_sprintf(ip6bufd, &ip6->ip6_dst), if_name(ifp))); goto bad; } if (IN6_IS_ADDR_UNSPECIFIED(&saddr6)) { /* dst has to be a solicited node multicast address. */ if (daddr6.s6_addr16[0] == IPV6_ADDR_INT16_MLL && /* don't check ifindex portion */ daddr6.s6_addr32[1] == 0 && daddr6.s6_addr32[2] == IPV6_ADDR_INT32_ONE && daddr6.s6_addr8[12] == 0xff) { ; /* good */ } else { nd6log((LOG_INFO, "nd6_ns_input: bad DAD packet " "(wrong ip6 dst)\n")); goto bad; } } else if (!V_nd6_onlink_ns_rfc4861) { struct sockaddr_in6 src_sa6; /* * According to recent IETF discussions, it is not a good idea * to accept a NS from an address which would not be deemed * to be a neighbor otherwise. This point is expected to be * clarified in future revisions of the specification. */ bzero(&src_sa6, sizeof(src_sa6)); src_sa6.sin6_family = AF_INET6; src_sa6.sin6_len = sizeof(src_sa6); src_sa6.sin6_addr = saddr6; if (nd6_is_addr_neighbor(&src_sa6, ifp) == 0) { nd6log((LOG_INFO, "nd6_ns_input: " "NS packet from non-neighbor\n")); goto bad; } } if (IN6_IS_ADDR_MULTICAST(&taddr6)) { nd6log((LOG_INFO, "nd6_ns_input: bad NS target (multicast)\n")); goto bad; } icmp6len -= sizeof(*nd_ns); nd6_option_init(nd_ns + 1, icmp6len, &ndopts); if (nd6_options(&ndopts) < 0) { nd6log((LOG_INFO, "nd6_ns_input: invalid ND option, ignored\n")); /* nd6_options have incremented stats */ goto freeit; } if (ndopts.nd_opts_src_lladdr) { lladdr = (char *)(ndopts.nd_opts_src_lladdr + 1); lladdrlen = ndopts.nd_opts_src_lladdr->nd_opt_len << 3; } if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src) && lladdr) { nd6log((LOG_INFO, "nd6_ns_input: bad DAD packet " "(link-layer address option)\n")); goto bad; } /* * Attaching target link-layer address to the NA? * (RFC 2461 7.2.4) * * NS IP dst is unicast/anycast MUST NOT add * NS IP dst is solicited-node multicast MUST add * * In implementation, we add target link-layer address by default. * We do not add one in MUST NOT cases. */ if (!IN6_IS_ADDR_MULTICAST(&daddr6)) tlladdr = 0; else tlladdr = 1; /* * Target address (taddr6) must be either: * (1) Valid unicast/anycast address for my receiving interface, * (2) Unicast address for which I'm offering proxy service, or * (3) "tentative" address on which DAD is being performed. */ /* (1) and (3) check. */ if (ifp->if_carp) ifa = (*carp_iamatch6_p)(ifp, &taddr6); else ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp, &taddr6); /* (2) check. */ if (ifa == NULL) { struct route_in6 ro; int need_proxy; bzero(&ro, sizeof(ro)); ro.ro_dst.sin6_len = sizeof(struct sockaddr_in6); ro.ro_dst.sin6_family = AF_INET6; ro.ro_dst.sin6_addr = taddr6; /* Always use the default FIB. */ #ifdef RADIX_MPATH rtalloc_mpath_fib((struct route *)&ro, ntohl(taddr6.s6_addr32[3]), RT_DEFAULT_FIB); #else in6_rtalloc(&ro, RT_DEFAULT_FIB); #endif need_proxy = (ro.ro_rt && (ro.ro_rt->rt_flags & RTF_ANNOUNCE) != 0 && ro.ro_rt->rt_gateway->sa_family == AF_LINK); if (ro.ro_rt != NULL) { if (need_proxy) proxydl = *SDL(ro.ro_rt->rt_gateway); RTFREE(ro.ro_rt); } if (need_proxy) { /* * proxy NDP for single entry */ ifa = (struct ifaddr *)in6ifa_ifpforlinklocal(ifp, IN6_IFF_NOTREADY|IN6_IFF_ANYCAST); if (ifa) proxy = 1; } } if (ifa == NULL) { /* * We've got an NS packet, and we don't have that adddress * assigned for us. We MUST silently ignore it. * See RFC2461 7.2.3. */ goto freeit; } myaddr6 = *IFA_IN6(ifa); anycast = ((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST; tentative = ((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_TENTATIVE; if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DUPLICATED) goto freeit; if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) { nd6log((LOG_INFO, "nd6_ns_input: lladdrlen mismatch for %s " "(if %d, NS packet %d)\n", ip6_sprintf(ip6bufs, &taddr6), ifp->if_addrlen, lladdrlen - 2)); goto bad; } if (IN6_ARE_ADDR_EQUAL(&myaddr6, &saddr6)) { nd6log((LOG_INFO, "nd6_ns_input: duplicate IP6 address %s\n", ip6_sprintf(ip6bufs, &saddr6))); goto freeit; } /* * We have neighbor solicitation packet, with target address equals to * one of my tentative address. * * src addr how to process? * --- --- * multicast of course, invalid (rejected in ip6_input) * unicast somebody is doing address resolution -> ignore * unspec dup address detection * * The processing is defined in RFC 2462. */ if (tentative) { /* * If source address is unspecified address, it is for * duplicate address detection. * * If not, the packet is for addess resolution; * silently ignore it. */ if (IN6_IS_ADDR_UNSPECIFIED(&saddr6)) nd6_dad_ns_input(ifa, ndopts.nd_opts_nonce); goto freeit; } /* * If the source address is unspecified address, entries must not * be created or updated. * It looks that sender is performing DAD. Output NA toward * all-node multicast address, to tell the sender that I'm using * the address. * S bit ("solicited") must be zero. */ if (IN6_IS_ADDR_UNSPECIFIED(&saddr6)) { struct in6_addr in6_all; in6_all = in6addr_linklocal_allnodes; if (in6_setscope(&in6_all, ifp, NULL) != 0) goto bad; nd6_na_output_fib(ifp, &in6_all, &taddr6, ((anycast || proxy || !tlladdr) ? 0 : ND_NA_FLAG_OVERRIDE) | rflag, tlladdr, proxy ? (struct sockaddr *)&proxydl : NULL, M_GETFIB(m)); goto freeit; } nd6_cache_lladdr(ifp, &saddr6, lladdr, lladdrlen, ND_NEIGHBOR_SOLICIT, 0); nd6_na_output_fib(ifp, &saddr6, &taddr6, ((anycast || proxy || !tlladdr) ? 0 : ND_NA_FLAG_OVERRIDE) | rflag | ND_NA_FLAG_SOLICITED, tlladdr, proxy ? (struct sockaddr *)&proxydl : NULL, M_GETFIB(m)); freeit: if (ifa != NULL) ifa_free(ifa); m_freem(m); return; bad: nd6log((LOG_ERR, "nd6_ns_input: src=%s\n", ip6_sprintf(ip6bufs, &saddr6))); nd6log((LOG_ERR, "nd6_ns_input: dst=%s\n", ip6_sprintf(ip6bufs, &daddr6))); nd6log((LOG_ERR, "nd6_ns_input: tgt=%s\n", ip6_sprintf(ip6bufs, &taddr6))); ICMP6STAT_INC(icp6s_badns); if (ifa != NULL) ifa_free(ifa); m_freem(m); } /* * Output a Neighbor Solicitation Message. Caller specifies: * - ICMP6 header source IP6 address * - ND6 header target IP6 address * - ND6 header source datalink address * * Based on RFC 2461 * Based on RFC 2462 (duplicate address detection) * * ln - for source address determination * nonce - If non-NULL, NS is used for duplicate address detection and * the value (length is ND_OPT_NONCE_LEN) is used as a random nonce. */ -void -nd6_ns_output(struct ifnet *ifp, const struct in6_addr *daddr6, - const struct in6_addr *taddr6, struct llentry *ln, uint8_t *nonce) +static void +nd6_ns_output_fib(struct ifnet *ifp, const struct in6_addr *daddr6, + const struct in6_addr *taddr6, struct llentry *ln, uint8_t *nonce, + u_int fibnum) { struct mbuf *m; struct m_tag *mtag; struct ip6_hdr *ip6; struct nd_neighbor_solicit *nd_ns; struct ip6_moptions im6o; int icmp6len; int maxlen; caddr_t mac; struct route_in6 ro; if (IN6_IS_ADDR_MULTICAST(taddr6)) return; /* estimate the size of message */ maxlen = sizeof(*ip6) + sizeof(*nd_ns); maxlen += (sizeof(struct nd_opt_hdr) + ifp->if_addrlen + 7) & ~7; if (max_linkhdr + maxlen >= MCLBYTES) { #ifdef DIAGNOSTIC - printf("nd6_ns_output: max_linkhdr + maxlen >= MCLBYTES " - "(%d + %d > %d)\n", max_linkhdr, maxlen, MCLBYTES); + printf("%s: max_linkhdr + maxlen >= MCLBYTES " + "(%d + %d > %d)\n", __func__, max_linkhdr, maxlen, + MCLBYTES); #endif return; } if (max_linkhdr + maxlen > MHLEN) m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR); else m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) return; + M_SETFIB(m, fibnum); bzero(&ro, sizeof(ro)); if (daddr6 == NULL || IN6_IS_ADDR_MULTICAST(daddr6)) { m->m_flags |= M_MCAST; im6o.im6o_multicast_ifp = ifp; im6o.im6o_multicast_hlim = 255; im6o.im6o_multicast_loop = 0; } icmp6len = sizeof(*nd_ns); m->m_pkthdr.len = m->m_len = sizeof(*ip6) + icmp6len; m->m_data += max_linkhdr; /* or M_ALIGN() equivalent? */ /* fill neighbor solicitation packet */ ip6 = mtod(m, struct ip6_hdr *); ip6->ip6_flow = 0; ip6->ip6_vfc &= ~IPV6_VERSION_MASK; ip6->ip6_vfc |= IPV6_VERSION; /* ip6->ip6_plen will be set later */ ip6->ip6_nxt = IPPROTO_ICMPV6; ip6->ip6_hlim = 255; if (daddr6) ip6->ip6_dst = *daddr6; else { ip6->ip6_dst.s6_addr16[0] = IPV6_ADDR_INT16_MLL; ip6->ip6_dst.s6_addr16[1] = 0; ip6->ip6_dst.s6_addr32[1] = 0; ip6->ip6_dst.s6_addr32[2] = IPV6_ADDR_INT32_ONE; ip6->ip6_dst.s6_addr32[3] = taddr6->s6_addr32[3]; ip6->ip6_dst.s6_addr8[12] = 0xff; if (in6_setscope(&ip6->ip6_dst, ifp, NULL) != 0) goto bad; } if (nonce == NULL) { struct ifaddr *ifa; /* * RFC2461 7.2.2: * "If the source address of the packet prompting the * solicitation is the same as one of the addresses assigned * to the outgoing interface, that address SHOULD be placed * in the IP Source Address of the outgoing solicitation. * Otherwise, any one of the addresses assigned to the * interface should be used." * * We use the source address for the prompting packet * (saddr6), if: * - saddr6 is given from the caller (by giving "ln"), and * - saddr6 belongs to the outgoing interface. * Otherwise, we perform the source address selection as usual. */ struct in6_addr *hsrc; hsrc = NULL; if (ln != NULL) { LLE_RLOCK(ln); if (ln->la_hold != NULL) { struct ip6_hdr *hip6; /* hold ip6 */ /* * assuming every packet in la_hold has the same IP * header */ hip6 = mtod(ln->la_hold, struct ip6_hdr *); /* XXX pullup? */ if (sizeof(*hip6) < ln->la_hold->m_len) { ip6->ip6_src = hip6->ip6_src; hsrc = &hip6->ip6_src; } } LLE_RUNLOCK(ln); } if (hsrc && (ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp, hsrc)) != NULL) { /* ip6_src set already. */ ifa_free(ifa); } else { int error; struct sockaddr_in6 dst_sa; struct in6_addr src_in; struct ifnet *oifp; bzero(&dst_sa, sizeof(dst_sa)); dst_sa.sin6_family = AF_INET6; dst_sa.sin6_len = sizeof(dst_sa); dst_sa.sin6_addr = ip6->ip6_dst; oifp = ifp; error = in6_selectsrc(&dst_sa, NULL, NULL, &ro, NULL, &oifp, &src_in); if (error) { char ip6buf[INET6_ADDRSTRLEN]; - nd6log((LOG_DEBUG, - "nd6_ns_output: source can't be " - "determined: dst=%s, error=%d\n", + nd6log((LOG_DEBUG, "%s: source can't be " + "determined: dst=%s, error=%d\n", __func__, ip6_sprintf(ip6buf, &dst_sa.sin6_addr), error)); goto bad; } ip6->ip6_src = src_in; } } else { /* * Source address for DAD packet must always be IPv6 * unspecified address. (0::0) * We actually don't have to 0-clear the address (we did it * above), but we do so here explicitly to make the intention * clearer. */ bzero(&ip6->ip6_src, sizeof(ip6->ip6_src)); } nd_ns = (struct nd_neighbor_solicit *)(ip6 + 1); nd_ns->nd_ns_type = ND_NEIGHBOR_SOLICIT; nd_ns->nd_ns_code = 0; nd_ns->nd_ns_reserved = 0; nd_ns->nd_ns_target = *taddr6; in6_clearscope(&nd_ns->nd_ns_target); /* XXX */ /* * Add source link-layer address option. * * spec implementation * --- --- * DAD packet MUST NOT do not add the option * there's no link layer address: * impossible do not add the option * there's link layer address: * Multicast NS MUST add one add the option * Unicast NS SHOULD add one add the option */ if (nonce == NULL && (mac = nd6_ifptomac(ifp))) { int optlen = sizeof(struct nd_opt_hdr) + ifp->if_addrlen; struct nd_opt_hdr *nd_opt = (struct nd_opt_hdr *)(nd_ns + 1); /* 8 byte alignments... */ optlen = (optlen + 7) & ~7; m->m_pkthdr.len += optlen; m->m_len += optlen; icmp6len += optlen; bzero((caddr_t)nd_opt, optlen); nd_opt->nd_opt_type = ND_OPT_SOURCE_LINKADDR; nd_opt->nd_opt_len = optlen >> 3; bcopy(mac, (caddr_t)(nd_opt + 1), ifp->if_addrlen); } /* * Add a Nonce option (RFC 3971) to detect looped back NS messages. * This behavior is documented as Enhanced Duplicate Address * Detection in draft-ietf-6man-enhanced-dad-13. * net.inet6.ip6.dad_enhanced=0 disables this. */ if (V_dad_enhanced != 0 && nonce != NULL) { int optlen = sizeof(struct nd_opt_hdr) + ND_OPT_NONCE_LEN; struct nd_opt_hdr *nd_opt = (struct nd_opt_hdr *)(nd_ns + 1); /* 8-byte alignment is required. */ optlen = (optlen + 7) & ~7; m->m_pkthdr.len += optlen; m->m_len += optlen; icmp6len += optlen; bzero((caddr_t)nd_opt, optlen); nd_opt->nd_opt_type = ND_OPT_NONCE; nd_opt->nd_opt_len = optlen >> 3; bcopy(nonce, (caddr_t)(nd_opt + 1), ND_OPT_NONCE_LEN); } ip6->ip6_plen = htons((u_short)icmp6len); nd_ns->nd_ns_cksum = 0; nd_ns->nd_ns_cksum = in6_cksum(m, IPPROTO_ICMPV6, sizeof(*ip6), icmp6len); if (send_sendso_input_hook != NULL) { mtag = m_tag_get(PACKET_TAG_ND_OUTGOING, sizeof(unsigned short), M_NOWAIT); if (mtag == NULL) goto bad; *(unsigned short *)(mtag + 1) = nd_ns->nd_ns_type; m_tag_prepend(m, mtag); } ip6_output(m, NULL, &ro, (nonce != NULL) ? IPV6_UNSPECSRC : 0, &im6o, NULL, NULL); icmp6_ifstat_inc(ifp, ifs6_out_msg); icmp6_ifstat_inc(ifp, ifs6_out_neighborsolicit); ICMP6STAT_INC(icp6s_outhist[ND_NEIGHBOR_SOLICIT]); /* We don't cache this route. */ RO_RTFREE(&ro); return; bad: if (ro.ro_rt) { RTFREE(ro.ro_rt); } m_freem(m); return; } +#ifndef BURN_BRIDGES +void +nd6_ns_output(struct ifnet *ifp, const struct in6_addr *daddr6, + const struct in6_addr *taddr6, struct llentry *ln, uint8_t *nonce) +{ + + nd6_ns_output_fib(ifp, daddr6, taddr6, ln, nonce, RT_DEFAULT_FIB); +} +#endif /* * Neighbor advertisement input handling. * * Based on RFC 2461 * Based on RFC 2462 (duplicate address detection) * * the following items are not implemented yet: * - proxy advertisement delay rule (RFC2461 7.2.8, last paragraph, SHOULD) * - anycast advertisement delay rule (RFC2461 7.2.7, SHOULD) */ void nd6_na_input(struct mbuf *m, int off, int icmp6len) { struct ifnet *ifp = m->m_pkthdr.rcvif; struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); struct nd_neighbor_advert *nd_na; struct in6_addr daddr6 = ip6->ip6_dst; struct in6_addr taddr6; int flags; int is_router; int is_solicited; int is_override; char *lladdr = NULL; int lladdrlen = 0; int checklink = 0; struct ifaddr *ifa; struct llentry *ln = NULL; union nd_opts ndopts; struct mbuf *chain = NULL; struct sockaddr_in6 sin6; char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; if (ip6->ip6_hlim != 255) { nd6log((LOG_ERR, "nd6_na_input: invalid hlim (%d) from %s to %s on %s\n", ip6->ip6_hlim, ip6_sprintf(ip6bufs, &ip6->ip6_src), ip6_sprintf(ip6bufd, &ip6->ip6_dst), if_name(ifp))); goto bad; } #ifndef PULLDOWN_TEST IP6_EXTHDR_CHECK(m, off, icmp6len,); nd_na = (struct nd_neighbor_advert *)((caddr_t)ip6 + off); #else IP6_EXTHDR_GET(nd_na, struct nd_neighbor_advert *, m, off, icmp6len); if (nd_na == NULL) { ICMP6STAT_INC(icp6s_tooshort); return; } #endif flags = nd_na->nd_na_flags_reserved; is_router = ((flags & ND_NA_FLAG_ROUTER) != 0); is_solicited = ((flags & ND_NA_FLAG_SOLICITED) != 0); is_override = ((flags & ND_NA_FLAG_OVERRIDE) != 0); memset(&sin6, 0, sizeof(sin6)); taddr6 = nd_na->nd_na_target; if (in6_setscope(&taddr6, ifp, NULL)) goto bad; /* XXX: impossible */ if (IN6_IS_ADDR_MULTICAST(&taddr6)) { nd6log((LOG_ERR, "nd6_na_input: invalid target address %s\n", ip6_sprintf(ip6bufs, &taddr6))); goto bad; } if (IN6_IS_ADDR_MULTICAST(&daddr6)) if (is_solicited) { nd6log((LOG_ERR, "nd6_na_input: a solicited adv is multicasted\n")); goto bad; } icmp6len -= sizeof(*nd_na); nd6_option_init(nd_na + 1, icmp6len, &ndopts); if (nd6_options(&ndopts) < 0) { nd6log((LOG_INFO, "nd6_na_input: invalid ND option, ignored\n")); /* nd6_options have incremented stats */ goto freeit; } if (ndopts.nd_opts_tgt_lladdr) { lladdr = (char *)(ndopts.nd_opts_tgt_lladdr + 1); lladdrlen = ndopts.nd_opts_tgt_lladdr->nd_opt_len << 3; } /* * This effectively disables the DAD check on a non-master CARP * address. */ if (ifp->if_carp) ifa = (*carp_iamatch6_p)(ifp, &taddr6); else ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp, &taddr6); /* * Target address matches one of my interface address. * * If my address is tentative, this means that there's somebody * already using the same address as mine. This indicates DAD failure. * This is defined in RFC 2462. * * Otherwise, process as defined in RFC 2461. */ if (ifa && (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_TENTATIVE)) { ifa_free(ifa); nd6_dad_na_input(ifa); goto freeit; } /* Just for safety, maybe unnecessary. */ if (ifa) { ifa_free(ifa); log(LOG_ERR, "nd6_na_input: duplicate IP6 address %s\n", ip6_sprintf(ip6bufs, &taddr6)); goto freeit; } if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) { nd6log((LOG_INFO, "nd6_na_input: lladdrlen mismatch for %s " "(if %d, NA packet %d)\n", ip6_sprintf(ip6bufs, &taddr6), ifp->if_addrlen, lladdrlen - 2)); goto bad; } /* * If no neighbor cache entry is found, NA SHOULD silently be * discarded. */ IF_AFDATA_RLOCK(ifp); ln = nd6_lookup(&taddr6, LLE_EXCLUSIVE, ifp); IF_AFDATA_RUNLOCK(ifp); if (ln == NULL) { goto freeit; } if (ln->ln_state == ND6_LLINFO_INCOMPLETE) { /* * If the link-layer has address, and no lladdr option came, * discard the packet. */ if (ifp->if_addrlen && lladdr == NULL) { goto freeit; } /* * Record link-layer address, and update the state. */ bcopy(lladdr, &ln->ll_addr, ifp->if_addrlen); ln->la_flags |= LLE_VALID; EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_RESOLVED); if (is_solicited) { ln->ln_state = ND6_LLINFO_REACHABLE; ln->ln_byhint = 0; if (!ND6_LLINFO_PERMANENT(ln)) { nd6_llinfo_settimer_locked(ln, (long)ND_IFINFO(ln->lle_tbl->llt_ifp)->reachable * hz); } } else { ln->ln_state = ND6_LLINFO_STALE; nd6_llinfo_settimer_locked(ln, (long)V_nd6_gctimer * hz); } if ((ln->ln_router = is_router) != 0) { /* * This means a router's state has changed from * non-reachable to probably reachable, and might * affect the status of associated prefixes.. */ checklink = 1; } } else { int llchange; /* * Check if the link-layer address has changed or not. */ if (lladdr == NULL) llchange = 0; else { if (ln->la_flags & LLE_VALID) { if (bcmp(lladdr, &ln->ll_addr, ifp->if_addrlen)) llchange = 1; else llchange = 0; } else llchange = 1; } /* * This is VERY complex. Look at it with care. * * override solicit lladdr llchange action * (L: record lladdr) * * 0 0 n -- (2c) * 0 0 y n (2b) L * 0 0 y y (1) REACHABLE->STALE * 0 1 n -- (2c) *->REACHABLE * 0 1 y n (2b) L *->REACHABLE * 0 1 y y (1) REACHABLE->STALE * 1 0 n -- (2a) * 1 0 y n (2a) L * 1 0 y y (2a) L *->STALE * 1 1 n -- (2a) *->REACHABLE * 1 1 y n (2a) L *->REACHABLE * 1 1 y y (2a) L *->REACHABLE */ if (!is_override && (lladdr != NULL && llchange)) { /* (1) */ /* * If state is REACHABLE, make it STALE. * no other updates should be done. */ if (ln->ln_state == ND6_LLINFO_REACHABLE) { ln->ln_state = ND6_LLINFO_STALE; nd6_llinfo_settimer_locked(ln, (long)V_nd6_gctimer * hz); } goto freeit; } else if (is_override /* (2a) */ || (!is_override && (lladdr != NULL && !llchange)) /* (2b) */ || lladdr == NULL) { /* (2c) */ /* * Update link-local address, if any. */ if (lladdr != NULL) { bcopy(lladdr, &ln->ll_addr, ifp->if_addrlen); ln->la_flags |= LLE_VALID; EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_RESOLVED); } /* * If solicited, make the state REACHABLE. * If not solicited and the link-layer address was * changed, make it STALE. */ if (is_solicited) { ln->ln_state = ND6_LLINFO_REACHABLE; ln->ln_byhint = 0; if (!ND6_LLINFO_PERMANENT(ln)) { nd6_llinfo_settimer_locked(ln, (long)ND_IFINFO(ifp)->reachable * hz); } } else { if (lladdr != NULL && llchange) { ln->ln_state = ND6_LLINFO_STALE; nd6_llinfo_settimer_locked(ln, (long)V_nd6_gctimer * hz); } } } if (ln->ln_router && !is_router) { /* * The peer dropped the router flag. * Remove the sender from the Default Router List and * update the Destination Cache entries. */ struct nd_defrouter *dr; struct in6_addr *in6; in6 = &L3_ADDR_SIN6(ln)->sin6_addr; /* * Lock to protect the default router list. * XXX: this might be unnecessary, since this function * is only called under the network software interrupt * context. However, we keep it just for safety. */ dr = defrouter_lookup(in6, ln->lle_tbl->llt_ifp); if (dr) defrtrlist_del(dr); else if (ND_IFINFO(ln->lle_tbl->llt_ifp)->flags & ND6_IFF_ACCEPT_RTADV) { /* * Even if the neighbor is not in the default * router list, the neighbor may be used * as a next hop for some destinations * (e.g. redirect case). So we must * call rt6_flush explicitly. */ rt6_flush(&ip6->ip6_src, ifp); } } ln->ln_router = is_router; } /* XXX - QL * Does this matter? * rt->rt_flags &= ~RTF_REJECT; */ ln->la_asked = 0; if (ln->la_hold != NULL) nd6_grab_holdchain(ln, &chain, &sin6); freeit: if (ln != NULL) LLE_WUNLOCK(ln); if (chain != NULL) nd6_flush_holdchain(ifp, ifp, chain, &sin6); if (checklink) pfxlist_onlink_check(); m_freem(m); return; bad: if (ln != NULL) LLE_WUNLOCK(ln); ICMP6STAT_INC(icp6s_badna); m_freem(m); } /* * Neighbor advertisement output handling. * * Based on RFC 2461 * * the following items are not implemented yet: * - proxy advertisement delay rule (RFC2461 7.2.8, last paragraph, SHOULD) * - anycast advertisement delay rule (RFC2461 7.2.7, SHOULD) * * tlladdr - 1 if include target link-layer address * sdl0 - sockaddr_dl (= proxy NA) or NULL */ static void nd6_na_output_fib(struct ifnet *ifp, const struct in6_addr *daddr6_0, const struct in6_addr *taddr6, u_long flags, int tlladdr, struct sockaddr *sdl0, u_int fibnum) { struct mbuf *m; struct m_tag *mtag; struct ifnet *oifp; struct ip6_hdr *ip6; struct nd_neighbor_advert *nd_na; struct ip6_moptions im6o; struct in6_addr src, daddr6; struct sockaddr_in6 dst_sa; int icmp6len, maxlen, error; caddr_t mac = NULL; struct route_in6 ro; bzero(&ro, sizeof(ro)); daddr6 = *daddr6_0; /* make a local copy for modification */ /* estimate the size of message */ maxlen = sizeof(*ip6) + sizeof(*nd_na); maxlen += (sizeof(struct nd_opt_hdr) + ifp->if_addrlen + 7) & ~7; if (max_linkhdr + maxlen >= MCLBYTES) { #ifdef DIAGNOSTIC printf("nd6_na_output: max_linkhdr + maxlen >= MCLBYTES " "(%d + %d > %d)\n", max_linkhdr, maxlen, MCLBYTES); #endif return; } if (max_linkhdr + maxlen > MHLEN) m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR); else m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) return; M_SETFIB(m, fibnum); if (IN6_IS_ADDR_MULTICAST(&daddr6)) { m->m_flags |= M_MCAST; im6o.im6o_multicast_ifp = ifp; im6o.im6o_multicast_hlim = 255; im6o.im6o_multicast_loop = 0; } icmp6len = sizeof(*nd_na); m->m_pkthdr.len = m->m_len = sizeof(struct ip6_hdr) + icmp6len; m->m_data += max_linkhdr; /* or M_ALIGN() equivalent? */ /* fill neighbor advertisement packet */ ip6 = mtod(m, struct ip6_hdr *); ip6->ip6_flow = 0; ip6->ip6_vfc &= ~IPV6_VERSION_MASK; ip6->ip6_vfc |= IPV6_VERSION; ip6->ip6_nxt = IPPROTO_ICMPV6; ip6->ip6_hlim = 255; if (IN6_IS_ADDR_UNSPECIFIED(&daddr6)) { /* reply to DAD */ daddr6.s6_addr16[0] = IPV6_ADDR_INT16_MLL; daddr6.s6_addr16[1] = 0; daddr6.s6_addr32[1] = 0; daddr6.s6_addr32[2] = 0; daddr6.s6_addr32[3] = IPV6_ADDR_INT32_ONE; if (in6_setscope(&daddr6, ifp, NULL)) goto bad; flags &= ~ND_NA_FLAG_SOLICITED; } ip6->ip6_dst = daddr6; bzero(&dst_sa, sizeof(struct sockaddr_in6)); dst_sa.sin6_family = AF_INET6; dst_sa.sin6_len = sizeof(struct sockaddr_in6); dst_sa.sin6_addr = daddr6; /* * Select a source whose scope is the same as that of the dest. */ bcopy(&dst_sa, &ro.ro_dst, sizeof(dst_sa)); oifp = ifp; error = in6_selectsrc(&dst_sa, NULL, NULL, &ro, NULL, &oifp, &src); if (error) { char ip6buf[INET6_ADDRSTRLEN]; nd6log((LOG_DEBUG, "nd6_na_output: source can't be " "determined: dst=%s, error=%d\n", ip6_sprintf(ip6buf, &dst_sa.sin6_addr), error)); goto bad; } ip6->ip6_src = src; nd_na = (struct nd_neighbor_advert *)(ip6 + 1); nd_na->nd_na_type = ND_NEIGHBOR_ADVERT; nd_na->nd_na_code = 0; nd_na->nd_na_target = *taddr6; in6_clearscope(&nd_na->nd_na_target); /* XXX */ /* * "tlladdr" indicates NS's condition for adding tlladdr or not. * see nd6_ns_input() for details. * Basically, if NS packet is sent to unicast/anycast addr, * target lladdr option SHOULD NOT be included. */ if (tlladdr) { /* * sdl0 != NULL indicates proxy NA. If we do proxy, use * lladdr in sdl0. If we are not proxying (sending NA for * my address) use lladdr configured for the interface. */ if (sdl0 == NULL) { if (ifp->if_carp) mac = (*carp_macmatch6_p)(ifp, m, taddr6); if (mac == NULL) mac = nd6_ifptomac(ifp); } else if (sdl0->sa_family == AF_LINK) { struct sockaddr_dl *sdl; sdl = (struct sockaddr_dl *)sdl0; if (sdl->sdl_alen == ifp->if_addrlen) mac = LLADDR(sdl); } } if (tlladdr && mac) { int optlen = sizeof(struct nd_opt_hdr) + ifp->if_addrlen; struct nd_opt_hdr *nd_opt = (struct nd_opt_hdr *)(nd_na + 1); /* roundup to 8 bytes alignment! */ optlen = (optlen + 7) & ~7; m->m_pkthdr.len += optlen; m->m_len += optlen; icmp6len += optlen; bzero((caddr_t)nd_opt, optlen); nd_opt->nd_opt_type = ND_OPT_TARGET_LINKADDR; nd_opt->nd_opt_len = optlen >> 3; bcopy(mac, (caddr_t)(nd_opt + 1), ifp->if_addrlen); } else flags &= ~ND_NA_FLAG_OVERRIDE; ip6->ip6_plen = htons((u_short)icmp6len); nd_na->nd_na_flags_reserved = flags; nd_na->nd_na_cksum = 0; nd_na->nd_na_cksum = in6_cksum(m, IPPROTO_ICMPV6, sizeof(struct ip6_hdr), icmp6len); if (send_sendso_input_hook != NULL) { mtag = m_tag_get(PACKET_TAG_ND_OUTGOING, sizeof(unsigned short), M_NOWAIT); if (mtag == NULL) goto bad; *(unsigned short *)(mtag + 1) = nd_na->nd_na_type; m_tag_prepend(m, mtag); } ip6_output(m, NULL, &ro, 0, &im6o, NULL, NULL); icmp6_ifstat_inc(ifp, ifs6_out_msg); icmp6_ifstat_inc(ifp, ifs6_out_neighboradvert); ICMP6STAT_INC(icp6s_outhist[ND_NEIGHBOR_ADVERT]); /* We don't cache this route. */ RO_RTFREE(&ro); return; bad: if (ro.ro_rt) { RTFREE(ro.ro_rt); } m_freem(m); return; } #ifndef BURN_BRIDGES void nd6_na_output(struct ifnet *ifp, const struct in6_addr *daddr6_0, const struct in6_addr *taddr6, u_long flags, int tlladdr, struct sockaddr *sdl0) { nd6_na_output_fib(ifp, daddr6_0, taddr6, flags, tlladdr, sdl0, RT_DEFAULT_FIB); } #endif caddr_t nd6_ifptomac(struct ifnet *ifp) { switch (ifp->if_type) { case IFT_ARCNET: case IFT_ETHER: case IFT_FDDI: case IFT_IEEE1394: #ifdef IFT_L2VLAN case IFT_L2VLAN: #endif #ifdef IFT_IEEE80211 case IFT_IEEE80211: #endif case IFT_INFINIBAND: case IFT_BRIDGE: case IFT_ISO88025: return IF_LLADDR(ifp); default: return NULL; } } struct dadq { TAILQ_ENTRY(dadq) dad_list; struct ifaddr *dad_ifa; int dad_count; /* max NS to send */ int dad_ns_tcount; /* # of trials to send NS */ int dad_ns_ocount; /* NS sent so far */ int dad_ns_icount; int dad_na_icount; int dad_ns_lcount; /* looped back NS */ + int dad_loopbackprobe; /* probing state for loopback detection */ struct callout dad_timer_ch; struct vnet *dad_vnet; u_int dad_refcnt; #define ND_OPT_NONCE_LEN32 \ ((ND_OPT_NONCE_LEN + sizeof(uint32_t) - 1)/sizeof(uint32_t)) uint32_t dad_nonce[ND_OPT_NONCE_LEN32]; }; static VNET_DEFINE(TAILQ_HEAD(, dadq), dadq); static VNET_DEFINE(struct rwlock, dad_rwlock); #define V_dadq VNET(dadq) #define V_dad_rwlock VNET(dad_rwlock) #define DADQ_RLOCK() rw_rlock(&V_dad_rwlock) #define DADQ_RUNLOCK() rw_runlock(&V_dad_rwlock) #define DADQ_WLOCK() rw_wlock(&V_dad_rwlock) #define DADQ_WUNLOCK() rw_wunlock(&V_dad_rwlock) static void nd6_dad_add(struct dadq *dp) { DADQ_WLOCK(); TAILQ_INSERT_TAIL(&V_dadq, dp, dad_list); DADQ_WUNLOCK(); } static void nd6_dad_del(struct dadq *dp) { DADQ_WLOCK(); TAILQ_REMOVE(&V_dadq, dp, dad_list); DADQ_WUNLOCK(); nd6_dad_rele(dp); } static struct dadq * nd6_dad_find(struct ifaddr *ifa, struct nd_opt_nonce *n) { struct dadq *dp; - char ip6buf[INET6_ADDRSTRLEN]; DADQ_RLOCK(); TAILQ_FOREACH(dp, &V_dadq, dad_list) { if (dp->dad_ifa != ifa) continue; /* * Skip if the nonce matches the received one. * +2 in the length is required because of type and * length fields are included in a header. */ if (n != NULL && n->nd_opt_nonce_len == (ND_OPT_NONCE_LEN + 2) / 8 && memcmp(&n->nd_opt_nonce[0], &dp->dad_nonce[0], ND_OPT_NONCE_LEN) == 0) { - log(LOG_ERR, "%s: a looped back NS message is " - "detected during DAD for %s.\n", - if_name(ifa->ifa_ifp), - ip6_sprintf(ip6buf, IFA_IN6(ifa))); dp->dad_ns_lcount++; continue; } refcount_acquire(&dp->dad_refcnt); break; } DADQ_RUNLOCK(); return (dp); } static void nd6_dad_starttimer(struct dadq *dp, int ticks) { callout_reset(&dp->dad_timer_ch, ticks, (void (*)(void *))nd6_dad_timer, (void *)dp); } static void nd6_dad_stoptimer(struct dadq *dp) { callout_drain(&dp->dad_timer_ch); } static void nd6_dad_rele(struct dadq *dp) { if (refcount_release(&dp->dad_refcnt)) { ifa_free(dp->dad_ifa); free(dp, M_IP6NDP); } } void nd6_dad_init(void) { rw_init(&V_dad_rwlock, "nd6 DAD queue"); TAILQ_INIT(&V_dadq); } /* * Start Duplicate Address Detection (DAD) for specified interface address. */ void nd6_dad_start(struct ifaddr *ifa, int delay) { struct in6_ifaddr *ia = (struct in6_ifaddr *)ifa; struct dadq *dp; char ip6buf[INET6_ADDRSTRLEN]; /* * If we don't need DAD, don't do it. * There are several cases: * - DAD is disabled (ip6_dad_count == 0) * - the interface address is anycast */ if (!(ia->ia6_flags & IN6_IFF_TENTATIVE)) { log(LOG_DEBUG, "nd6_dad_start: called with non-tentative address " "%s(%s)\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), ifa->ifa_ifp ? if_name(ifa->ifa_ifp) : "???"); return; } if (ia->ia6_flags & IN6_IFF_ANYCAST) { ia->ia6_flags &= ~IN6_IFF_TENTATIVE; return; } if (!V_ip6_dad_count) { ia->ia6_flags &= ~IN6_IFF_TENTATIVE; return; } if (ifa->ifa_ifp == NULL) panic("nd6_dad_start: ifa->ifa_ifp == NULL"); if (!(ifa->ifa_ifp->if_flags & IFF_UP)) { return; } if (ND_IFINFO(ifa->ifa_ifp)->flags & ND6_IFF_IFDISABLED) return; if ((dp = nd6_dad_find(ifa, NULL)) != NULL) { /* DAD already in progress */ nd6_dad_rele(dp); return; } dp = malloc(sizeof(*dp), M_IP6NDP, M_NOWAIT | M_ZERO); if (dp == NULL) { log(LOG_ERR, "nd6_dad_start: memory allocation failed for " "%s(%s)\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), ifa->ifa_ifp ? if_name(ifa->ifa_ifp) : "???"); return; } callout_init(&dp->dad_timer_ch, 0); #ifdef VIMAGE dp->dad_vnet = curvnet; #endif nd6log((LOG_DEBUG, "%s: starting DAD for %s\n", if_name(ifa->ifa_ifp), ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr))); /* * Send NS packet for DAD, ip6_dad_count times. * Note that we must delay the first transmission, if this is the * first packet to be sent from the interface after interface * (re)initialization. */ dp->dad_ifa = ifa; ifa_ref(dp->dad_ifa); dp->dad_count = V_ip6_dad_count; dp->dad_ns_icount = dp->dad_na_icount = 0; dp->dad_ns_ocount = dp->dad_ns_tcount = 0; - dp->dad_ns_lcount = 0; + dp->dad_ns_lcount = dp->dad_loopbackprobe = 0; refcount_init(&dp->dad_refcnt, 1); nd6_dad_add(dp); if (delay == 0) { nd6_dad_ns_output(dp, ifa); nd6_dad_starttimer(dp, (long)ND_IFINFO(ifa->ifa_ifp)->retrans * hz / 1000); } else { nd6_dad_starttimer(dp, delay); } } /* * terminate DAD unconditionally. used for address removals. */ void nd6_dad_stop(struct ifaddr *ifa) { struct dadq *dp; dp = nd6_dad_find(ifa, NULL); if (!dp) { /* DAD wasn't started yet */ return; } nd6_dad_stoptimer(dp); /* * The DAD queue entry may have been removed by nd6_dad_timer() while * we were waiting for it to stop, so re-do the lookup. */ nd6_dad_rele(dp); if (nd6_dad_find(ifa, NULL) == NULL) return; nd6_dad_del(dp); nd6_dad_rele(dp); } static void nd6_dad_timer(struct dadq *dp) { CURVNET_SET(dp->dad_vnet); struct ifaddr *ifa = dp->dad_ifa; struct ifnet *ifp = dp->dad_ifa->ifa_ifp; struct in6_ifaddr *ia = (struct in6_ifaddr *)ifa; char ip6buf[INET6_ADDRSTRLEN]; /* Sanity check */ if (ia == NULL) { log(LOG_ERR, "nd6_dad_timer: called with null parameter\n"); goto err; } if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) { /* Do not need DAD for ifdisabled interface. */ log(LOG_ERR, "nd6_dad_timer: cancel DAD on %s because of " "ND6_IFF_IFDISABLED.\n", ifp->if_xname); goto err; } if (ia->ia6_flags & IN6_IFF_DUPLICATED) { log(LOG_ERR, "nd6_dad_timer: called with duplicated address " "%s(%s)\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), ifa->ifa_ifp ? if_name(ifa->ifa_ifp) : "???"); goto err; } if ((ia->ia6_flags & IN6_IFF_TENTATIVE) == 0) { log(LOG_ERR, "nd6_dad_timer: called with non-tentative address " "%s(%s)\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), ifa->ifa_ifp ? if_name(ifa->ifa_ifp) : "???"); goto err; } - /* timeouted with IFF_{RUNNING,UP} check */ - if (dp->dad_ns_tcount > V_dad_maxtry) { + /* Stop DAD if the interface is down even after dad_maxtry attempts. */ + if ((dp->dad_ns_tcount > V_dad_maxtry) && + (((ifp->if_flags & IFF_UP) == 0) || + ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0))) { nd6log((LOG_INFO, "%s: could not run DAD, driver problem?\n", if_name(ifa->ifa_ifp))); goto err; } /* Need more checks? */ if (dp->dad_ns_ocount < dp->dad_count) { /* * We have more NS to go. Send NS packet for DAD. */ nd6_dad_ns_output(dp, ifa); nd6_dad_starttimer(dp, (long)ND_IFINFO(ifa->ifa_ifp)->retrans * hz / 1000); goto done; } else { /* * We have transmitted sufficient number of DAD packets. * See what we've got. */ if (dp->dad_ns_icount > 0 || dp->dad_na_icount > 0) /* We've seen NS or NA, means DAD has failed. */ nd6_dad_duplicated(ifa, dp); - else { + else if (V_dad_enhanced != 0 && + dp->dad_ns_lcount > 0 && + dp->dad_ns_lcount > dp->dad_loopbackprobe) { /* + * A looped back probe is detected, + * Sec. 4.1 in draft-ietf-6man-enhanced-dad-13 + * requires transmission of additional probes until + * the loopback condition becomes clear. + */ + log(LOG_ERR, "%s: a looped back NS message is " + "detected during DAD for %s. " + "Another DAD probes are being sent.\n", + if_name(ifa->ifa_ifp), + ip6_sprintf(ip6buf, IFA_IN6(ifa))); + dp->dad_loopbackprobe = dp->dad_ns_lcount; + /* + * An interface with IGNORELOOP is one which a + * loopback is permanently expected while regular + * traffic works. In that case, stop DAD after + * MAX_MULTICAST_SOLICIT number of NS messages + * regardless of the number of received loopback NS + * by increasing dad_loopbackprobe in advance. + */ + if (ND_IFINFO(ifa->ifa_ifp)->flags & ND6_IFF_IGNORELOOP) + dp->dad_loopbackprobe += V_nd6_mmaxtries; + /* + * Send an NS immediately and increase dad_count by + * V_nd6_mmaxtries - 1. + */ + nd6_dad_ns_output(dp, ifa); + dp->dad_count = + dp->dad_ns_ocount + V_nd6_mmaxtries - 1; + nd6_dad_starttimer(dp, + (long)ND_IFINFO(ifa->ifa_ifp)->retrans * hz / 1000); + goto done; + } else { + /* * We are done with DAD. No NA came, no NS came. * No duplicate address found. Check IFDISABLED flag * again in case that it is changed between the * beginning of this function and here. */ if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) == 0) ia->ia6_flags &= ~IN6_IFF_TENTATIVE; nd6log((LOG_DEBUG, "%s: DAD complete for %s - no duplicates found\n", if_name(ifa->ifa_ifp), ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr))); + if (dp->dad_ns_lcount > 0) + log(LOG_ERR, "%s: DAD completed while " + "a looped back NS message is detected " + "during DAD for %s.\n", + if_name(ifa->ifa_ifp), + ip6_sprintf(ip6buf, IFA_IN6(ifa))); } } err: nd6_dad_del(dp); done: CURVNET_RESTORE(); } static void nd6_dad_duplicated(struct ifaddr *ifa, struct dadq *dp) { struct in6_ifaddr *ia = (struct in6_ifaddr *)ifa; struct ifnet *ifp; char ip6buf[INET6_ADDRSTRLEN]; log(LOG_ERR, "%s: DAD detected duplicate IPv6 address %s: " "NS in/out/loopback=%d/%d/%d, NA in=%d\n", if_name(ifa->ifa_ifp), ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), dp->dad_ns_icount, dp->dad_ns_ocount, dp->dad_ns_lcount, dp->dad_na_icount); ia->ia6_flags &= ~IN6_IFF_TENTATIVE; ia->ia6_flags |= IN6_IFF_DUPLICATED; ifp = ifa->ifa_ifp; log(LOG_ERR, "%s: DAD complete for %s - duplicate found\n", if_name(ifp), ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr)); log(LOG_ERR, "%s: manual intervention required\n", if_name(ifp)); /* * If the address is a link-local address formed from an interface * identifier based on the hardware address which is supposed to be * uniquely assigned (e.g., EUI-64 for an Ethernet interface), IP * operation on the interface SHOULD be disabled. * [RFC 4862, Section 5.4.5] */ if (IN6_IS_ADDR_LINKLOCAL(&ia->ia_addr.sin6_addr)) { struct in6_addr in6; /* * To avoid over-reaction, we only apply this logic when we are * very sure that hardware addresses are supposed to be unique. */ switch (ifp->if_type) { case IFT_ETHER: case IFT_FDDI: case IFT_ATM: case IFT_IEEE1394: #ifdef IFT_IEEE80211 case IFT_IEEE80211: #endif case IFT_INFINIBAND: in6 = ia->ia_addr.sin6_addr; if (in6_get_hw_ifid(ifp, &in6) == 0 && IN6_ARE_ADDR_EQUAL(&ia->ia_addr.sin6_addr, &in6)) { ND_IFINFO(ifp)->flags |= ND6_IFF_IFDISABLED; log(LOG_ERR, "%s: possible hardware address " "duplication detected, disable IPv6\n", if_name(ifp)); } break; } } } static void nd6_dad_ns_output(struct dadq *dp, struct ifaddr *ifa) { struct in6_ifaddr *ia = (struct in6_ifaddr *)ifa; struct ifnet *ifp = ifa->ifa_ifp; - uint8_t *nonce; int i; dp->dad_ns_tcount++; if ((ifp->if_flags & IFF_UP) == 0) { return; } if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { return; } dp->dad_ns_ocount++; if (V_dad_enhanced != 0) { for (i = 0; i < ND_OPT_NONCE_LEN32; i++) dp->dad_nonce[i] = arc4random(); - nonce = (uint8_t *)&dp->dad_nonce[0]; /* * XXXHRS: Note that in the case that * DupAddrDetectTransmits > 1, multiple NS messages with * different nonces can be looped back in an unexpected * order. The current implementation recognizes only * the latest nonce on the sender side. Practically it * should work well in almost all cases. */ - } else - nonce = NULL; - nd6_ns_output(ifp, NULL, &ia->ia_addr.sin6_addr, NULL, nonce); + } + nd6_ns_output(ifp, NULL, &ia->ia_addr.sin6_addr, NULL, + (uint8_t *)&dp->dad_nonce[0]); } static void nd6_dad_ns_input(struct ifaddr *ifa, struct nd_opt_nonce *ndopt_nonce) { struct in6_ifaddr *ia; struct ifnet *ifp; const struct in6_addr *taddr6; struct dadq *dp; if (ifa == NULL) panic("ifa == NULL in nd6_dad_ns_input"); ia = (struct in6_ifaddr *)ifa; ifp = ifa->ifa_ifp; taddr6 = &ia->ia_addr.sin6_addr; /* Ignore Nonce option when Enhanced DAD is disabled. */ if (V_dad_enhanced == 0) ndopt_nonce = NULL; dp = nd6_dad_find(ifa, ndopt_nonce); if (dp == NULL) return; dp->dad_ns_icount++; nd6_dad_rele(dp); } static void nd6_dad_na_input(struct ifaddr *ifa) { struct dadq *dp; if (ifa == NULL) panic("ifa == NULL in nd6_dad_na_input"); dp = nd6_dad_find(ifa, NULL); if (dp != NULL) { dp->dad_na_icount++; nd6_dad_rele(dp); } } Index: projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/cmd.c =================================================================== --- projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/cmd.c (revision 279716) +++ projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/cmd.c (revision 279717) @@ -1,2592 +1,2605 @@ /* * Copyright (c) 2004, 2005 Topspin Communications. All rights reserved. * Copyright (c) 2005, 2006, 2007, 2008, 2014 Mellanox Technologies. All rights reserved. * Copyright (c) 2005, 2006, 2007 Cisco Systems, Inc. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU * General Public License (GPL) Version 2, available from the file * COPYING in the main directory of this source tree, or the * OpenIB.org BSD license below: * * Redistribution and use in source and binary forms, with or * without modification, are permitted provided that the following * conditions are met: * * - Redistributions of source code must retain the above * copyright notice, this list of conditions and the following * disclaimer. * * - Redistributions in binary form must reproduce the above * copyright notice, this list of conditions and the following * disclaimer in the documentation and/or other materials * provided with the distribution. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ #include #include #include #include #include #include #include #include #include #include #include #include "mlx4.h" #include "fw.h" #define CMD_POLL_TOKEN 0xffff #define INBOX_MASK 0xffffffffffffff00ULL #define CMD_CHAN_VER 1 #define CMD_CHAN_IF_REV 1 enum { /* command completed successfully: */ CMD_STAT_OK = 0x00, /* Internal error (such as a bus error) occurred while processing command: */ CMD_STAT_INTERNAL_ERR = 0x01, /* Operation/command not supported or opcode modifier not supported: */ CMD_STAT_BAD_OP = 0x02, /* Parameter not supported or parameter out of range: */ CMD_STAT_BAD_PARAM = 0x03, /* System not enabled or bad system state: */ CMD_STAT_BAD_SYS_STATE = 0x04, /* Attempt to access reserved or unallocaterd resource: */ CMD_STAT_BAD_RESOURCE = 0x05, /* Requested resource is currently executing a command, or is otherwise busy: */ CMD_STAT_RESOURCE_BUSY = 0x06, /* Required capability exceeds device limits: */ CMD_STAT_EXCEED_LIM = 0x08, /* Resource is not in the appropriate state or ownership: */ CMD_STAT_BAD_RES_STATE = 0x09, /* Index out of range: */ CMD_STAT_BAD_INDEX = 0x0a, /* FW image corrupted: */ CMD_STAT_BAD_NVMEM = 0x0b, /* Error in ICM mapping (e.g. not enough auxiliary ICM pages to execute command): */ CMD_STAT_ICM_ERROR = 0x0c, /* Attempt to modify a QP/EE which is not in the presumed state: */ CMD_STAT_BAD_QP_STATE = 0x10, /* Bad segment parameters (Address/Size): */ CMD_STAT_BAD_SEG_PARAM = 0x20, /* Memory Region has Memory Windows bound to: */ CMD_STAT_REG_BOUND = 0x21, /* HCA local attached memory not present: */ CMD_STAT_LAM_NOT_PRE = 0x22, /* Bad management packet (silently discarded): */ CMD_STAT_BAD_PKT = 0x30, /* More outstanding CQEs in CQ than new CQ size: */ CMD_STAT_BAD_SIZE = 0x40, /* Multi Function device support required: */ CMD_STAT_MULTI_FUNC_REQ = 0x50, }; enum { HCR_IN_PARAM_OFFSET = 0x00, HCR_IN_MODIFIER_OFFSET = 0x08, HCR_OUT_PARAM_OFFSET = 0x0c, HCR_TOKEN_OFFSET = 0x14, HCR_STATUS_OFFSET = 0x18, HCR_OPMOD_SHIFT = 12, HCR_T_BIT = 21, HCR_E_BIT = 22, HCR_GO_BIT = 23 }; enum { GO_BIT_TIMEOUT_MSECS = 10000 }; enum mlx4_vlan_transition { MLX4_VLAN_TRANSITION_VST_VST = 0, MLX4_VLAN_TRANSITION_VST_VGT = 1, MLX4_VLAN_TRANSITION_VGT_VST = 2, MLX4_VLAN_TRANSITION_VGT_VGT = 3, }; struct mlx4_cmd_context { struct completion done; int result; int next; u64 out_param; u16 token; u8 fw_status; }; static int mlx4_master_process_vhcr(struct mlx4_dev *dev, int slave, struct mlx4_vhcr_cmd *in_vhcr); static int mlx4_status_to_errno(u8 status) { static const int trans_table[] = { [CMD_STAT_INTERNAL_ERR] = -EIO, [CMD_STAT_BAD_OP] = -EPERM, [CMD_STAT_BAD_PARAM] = -EINVAL, [CMD_STAT_BAD_SYS_STATE] = -ENXIO, [CMD_STAT_BAD_RESOURCE] = -EBADF, [CMD_STAT_RESOURCE_BUSY] = -EBUSY, [CMD_STAT_EXCEED_LIM] = -ENOMEM, [CMD_STAT_BAD_RES_STATE] = -EBADF, [CMD_STAT_BAD_INDEX] = -EBADF, [CMD_STAT_BAD_NVMEM] = -EFAULT, [CMD_STAT_ICM_ERROR] = -ENFILE, [CMD_STAT_BAD_QP_STATE] = -EINVAL, [CMD_STAT_BAD_SEG_PARAM] = -EFAULT, [CMD_STAT_REG_BOUND] = -EBUSY, [CMD_STAT_LAM_NOT_PRE] = -EAGAIN, [CMD_STAT_BAD_PKT] = -EINVAL, [CMD_STAT_BAD_SIZE] = -ENOMEM, [CMD_STAT_MULTI_FUNC_REQ] = -EACCES, }; if (status >= ARRAY_SIZE(trans_table) || (status != CMD_STAT_OK && trans_table[status] == 0)) return -EIO; return trans_table[status]; } static const char *cmd_to_str(u16 cmd) { switch (cmd) { case MLX4_CMD_SYS_EN: return "SYS_EN"; case MLX4_CMD_SYS_DIS: return "SYS_DIS"; case MLX4_CMD_MAP_FA: return "MAP_FA"; case MLX4_CMD_UNMAP_FA: return "UNMAP_FA"; case MLX4_CMD_RUN_FW: return "RUN_FW"; case MLX4_CMD_MOD_STAT_CFG: return "MOD_STAT_CFG"; case MLX4_CMD_QUERY_DEV_CAP: return "QUERY_DEV_CAP"; case MLX4_CMD_QUERY_FW: return "QUERY_FW"; case MLX4_CMD_ENABLE_LAM: return "ENABLE_LAM"; case MLX4_CMD_DISABLE_LAM: return "DISABLE_LAM"; case MLX4_CMD_QUERY_DDR: return "QUERY_DDR"; case MLX4_CMD_QUERY_ADAPTER: return "QUERY_ADAPTER"; case MLX4_CMD_INIT_HCA: return "INIT_HCA"; case MLX4_CMD_CLOSE_HCA: return "CLOSE_HCA"; case MLX4_CMD_INIT_PORT: return "INIT_PORT"; case MLX4_CMD_CLOSE_PORT: return "CLOSE_PORT"; case MLX4_CMD_QUERY_HCA: return "QUERY_HCA"; case MLX4_CMD_QUERY_PORT: return "QUERY_PORT"; case MLX4_CMD_SENSE_PORT: return "SENSE_PORT"; case MLX4_CMD_HW_HEALTH_CHECK: return "HW_HEALTH_CHECK"; case MLX4_CMD_SET_PORT: return "SET_PORT"; case MLX4_CMD_SET_NODE: return "SET_NODE"; case MLX4_CMD_QUERY_FUNC: return "QUERY_FUNC"; case MLX4_CMD_MAP_ICM: return "MAP_ICM"; case MLX4_CMD_UNMAP_ICM: return "UNMAP_ICM"; case MLX4_CMD_MAP_ICM_AUX: return "MAP_ICM_AUX"; case MLX4_CMD_UNMAP_ICM_AUX: return "UNMAP_ICM_AUX"; case MLX4_CMD_SET_ICM_SIZE: return "SET_ICM_SIZE"; /*master notify fw on finish for slave's flr*/ case MLX4_CMD_INFORM_FLR_DONE: return "INFORM_FLR_DONE"; case MLX4_CMD_GET_OP_REQ: return "GET_OP_REQ"; /* TPT commands */ case MLX4_CMD_SW2HW_MPT: return "SW2HW_MPT"; case MLX4_CMD_QUERY_MPT: return "QUERY_MPT"; case MLX4_CMD_HW2SW_MPT: return "HW2SW_MPT"; case MLX4_CMD_READ_MTT: return "READ_MTT"; case MLX4_CMD_WRITE_MTT: return "WRITE_MTT"; case MLX4_CMD_SYNC_TPT: return "SYNC_TPT"; /* EQ commands */ case MLX4_CMD_MAP_EQ: return "MAP_EQ"; case MLX4_CMD_SW2HW_EQ: return "SW2HW_EQ"; case MLX4_CMD_HW2SW_EQ: return "HW2SW_EQ"; case MLX4_CMD_QUERY_EQ: return "QUERY_EQ"; /* CQ commands */ case MLX4_CMD_SW2HW_CQ: return "SW2HW_CQ"; case MLX4_CMD_HW2SW_CQ: return "HW2SW_CQ"; case MLX4_CMD_QUERY_CQ: return "QUERY_CQ:"; case MLX4_CMD_MODIFY_CQ: return "MODIFY_CQ:"; /* SRQ commands */ case MLX4_CMD_SW2HW_SRQ: return "SW2HW_SRQ"; case MLX4_CMD_HW2SW_SRQ: return "HW2SW_SRQ"; case MLX4_CMD_QUERY_SRQ: return "QUERY_SRQ"; case MLX4_CMD_ARM_SRQ: return "ARM_SRQ"; /* QP/EE commands */ case MLX4_CMD_RST2INIT_QP: return "RST2INIT_QP"; case MLX4_CMD_INIT2RTR_QP: return "INIT2RTR_QP"; case MLX4_CMD_RTR2RTS_QP: return "RTR2RTS_QP"; case MLX4_CMD_RTS2RTS_QP: return "RTS2RTS_QP"; case MLX4_CMD_SQERR2RTS_QP: return "SQERR2RTS_QP"; case MLX4_CMD_2ERR_QP: return "2ERR_QP"; case MLX4_CMD_RTS2SQD_QP: return "RTS2SQD_QP"; case MLX4_CMD_SQD2SQD_QP: return "SQD2SQD_QP"; case MLX4_CMD_SQD2RTS_QP: return "SQD2RTS_QP"; case MLX4_CMD_2RST_QP: return "2RST_QP"; case MLX4_CMD_QUERY_QP: return "QUERY_QP"; case MLX4_CMD_INIT2INIT_QP: return "INIT2INIT_QP"; case MLX4_CMD_SUSPEND_QP: return "SUSPEND_QP"; case MLX4_CMD_UNSUSPEND_QP: return "UNSUSPEND_QP"; /* special QP and management commands */ case MLX4_CMD_CONF_SPECIAL_QP: return "CONF_SPECIAL_QP"; case MLX4_CMD_MAD_IFC: return "MAD_IFC"; /* multicast commands */ case MLX4_CMD_READ_MCG: return "READ_MCG"; case MLX4_CMD_WRITE_MCG: return "WRITE_MCG"; case MLX4_CMD_MGID_HASH: return "MGID_HASH"; /* miscellaneous commands */ case MLX4_CMD_DIAG_RPRT: return "DIAG_RPRT"; case MLX4_CMD_NOP: return "NOP"; case MLX4_CMD_ACCESS_MEM: return "ACCESS_MEM"; case MLX4_CMD_SET_VEP: return "SET_VEP"; /* Ethernet specific commands */ case MLX4_CMD_SET_VLAN_FLTR: return "SET_VLAN_FLTR"; case MLX4_CMD_SET_MCAST_FLTR: return "SET_MCAST_FLTR"; case MLX4_CMD_DUMP_ETH_STATS: return "DUMP_ETH_STATS"; /* Communication channel commands */ case MLX4_CMD_ARM_COMM_CHANNEL: return "ARM_COMM_CHANNEL"; case MLX4_CMD_GEN_EQE: return "GEN_EQE"; /* virtual commands */ case MLX4_CMD_ALLOC_RES: return "ALLOC_RES"; case MLX4_CMD_FREE_RES: return "FREE_RES"; case MLX4_CMD_MCAST_ATTACH: return "MCAST_ATTACH"; case MLX4_CMD_UCAST_ATTACH: return "UCAST_ATTACH"; case MLX4_CMD_PROMISC: return "PROMISC"; case MLX4_CMD_QUERY_FUNC_CAP: return "QUERY_FUNC_CAP"; case MLX4_CMD_QP_ATTACH: return "QP_ATTACH"; /* debug commands */ case MLX4_CMD_QUERY_DEBUG_MSG: return "QUERY_DEBUG_MSG"; case MLX4_CMD_SET_DEBUG_MSG: return "SET_DEBUG_MSG"; /* statistics commands */ case MLX4_CMD_QUERY_IF_STAT: return "QUERY_IF_STAT"; case MLX4_CMD_SET_IF_STAT: return "SET_IF_STAT"; /* register/delete flow steering network rules */ case MLX4_QP_FLOW_STEERING_ATTACH: return "QP_FLOW_STEERING_ATTACH"; case MLX4_QP_FLOW_STEERING_DETACH: return "QP_FLOW_STEERING_DETACH"; case MLX4_FLOW_STEERING_IB_UC_QP_RANGE: return "FLOW_STEERING_IB_UC_QP_RANGE"; default: return "OTHER"; } } static u8 mlx4_errno_to_status(int errno) { switch (errno) { case -EPERM: return CMD_STAT_BAD_OP; case -EINVAL: return CMD_STAT_BAD_PARAM; case -ENXIO: return CMD_STAT_BAD_SYS_STATE; case -EBUSY: return CMD_STAT_RESOURCE_BUSY; case -ENOMEM: return CMD_STAT_EXCEED_LIM; case -ENFILE: return CMD_STAT_ICM_ERROR; default: return CMD_STAT_INTERNAL_ERR; } } static int comm_pending(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); u32 status = readl(&priv->mfunc.comm->slave_read); return (swab32(status) >> 31) != priv->cmd.comm_toggle; } static void mlx4_comm_cmd_post(struct mlx4_dev *dev, u8 cmd, u16 param) { struct mlx4_priv *priv = mlx4_priv(dev); u32 val; priv->cmd.comm_toggle ^= 1; val = param | (cmd << 16) | (priv->cmd.comm_toggle << 31); __raw_writel((__force u32) cpu_to_be32(val), &priv->mfunc.comm->slave_write); mmiowb(); } static int mlx4_comm_cmd_poll(struct mlx4_dev *dev, u8 cmd, u16 param, unsigned long timeout) { struct mlx4_priv *priv = mlx4_priv(dev); unsigned long end; int err = 0; int ret_from_pending = 0; /* First, verify that the master reports correct status */ if (comm_pending(dev)) { mlx4_warn(dev, "Communication channel is not idle." "my toggle is %d (cmd:0x%x)\n", priv->cmd.comm_toggle, cmd); return -EAGAIN; } /* Write command */ down(&priv->cmd.poll_sem); mlx4_comm_cmd_post(dev, cmd, param); end = msecs_to_jiffies(timeout) + jiffies; while (comm_pending(dev) && time_before(jiffies, end)) cond_resched(); ret_from_pending = comm_pending(dev); if (ret_from_pending) { /* check if the slave is trying to boot in the middle of * FLR process. The only non-zero result in the RESET command * is MLX4_DELAY_RESET_SLAVE*/ if ((MLX4_COMM_CMD_RESET == cmd)) { mlx4_warn(dev, "Got slave FLRed from Communication" " channel (ret:0x%x)\n", ret_from_pending); err = MLX4_DELAY_RESET_SLAVE; } else { mlx4_warn(dev, "Communication channel timed out\n"); err = -ETIMEDOUT; } } up(&priv->cmd.poll_sem); return err; } static int mlx4_comm_cmd_wait(struct mlx4_dev *dev, u8 op, u16 param, unsigned long timeout) { struct mlx4_cmd *cmd = &mlx4_priv(dev)->cmd; struct mlx4_cmd_context *context; unsigned long end; int err = 0; down(&cmd->event_sem); end = msecs_to_jiffies(timeout) + jiffies; while (comm_pending(dev) && time_before(jiffies, end)) cond_resched(); if (comm_pending(dev)) { mlx4_warn(dev, "mlx4_comm_cmd_wait: Comm channel " "is not idle. My toggle is %d (op: 0x%x)\n", mlx4_priv(dev)->cmd.comm_toggle, op); up(&cmd->event_sem); return -EAGAIN; } spin_lock(&cmd->context_lock); BUG_ON(cmd->free_head < 0); context = &cmd->context[cmd->free_head]; context->token += cmd->token_mask + 1; cmd->free_head = context->next; spin_unlock(&cmd->context_lock); init_completion(&context->done); mlx4_comm_cmd_post(dev, op, param); /* In slave, wait unconditionally for completion */ wait_for_completion(&context->done); err = context->result; if (err && context->fw_status != CMD_STAT_MULTI_FUNC_REQ) { mlx4_err(dev, "command 0x%x failed: fw status = 0x%x\n", op, context->fw_status); goto out; } out: /* wait for comm channel ready * this is necessary for prevention the race * when switching between event to polling mode */ end = msecs_to_jiffies(timeout) + jiffies; while (comm_pending(dev) && time_before(jiffies, end)) cond_resched(); spin_lock(&cmd->context_lock); context->next = cmd->free_head; cmd->free_head = context - cmd->context; spin_unlock(&cmd->context_lock); up(&cmd->event_sem); return err; } int mlx4_comm_cmd(struct mlx4_dev *dev, u8 cmd, u16 param, unsigned long timeout) { if (mlx4_priv(dev)->cmd.use_events) return mlx4_comm_cmd_wait(dev, cmd, param, timeout); return mlx4_comm_cmd_poll(dev, cmd, param, timeout); } static int cmd_pending(struct mlx4_dev *dev) { u32 status; if (pci_channel_offline(dev->pdev)) return -EIO; status = readl(mlx4_priv(dev)->cmd.hcr + HCR_STATUS_OFFSET); return (status & swab32(1 << HCR_GO_BIT)) || (mlx4_priv(dev)->cmd.toggle == !!(status & swab32(1 << HCR_T_BIT))); } static int get_status(struct mlx4_dev *dev, u32 *status, int *go_bit, int *t_bit) { if (pci_channel_offline(dev->pdev)) return -EIO; *status = readl(mlx4_priv(dev)->cmd.hcr + HCR_STATUS_OFFSET); *t_bit = !!(*status & swab32(1 << HCR_T_BIT)); *go_bit = !!(*status & swab32(1 << HCR_GO_BIT)); return 0; } static int mlx4_cmd_post(struct mlx4_dev *dev, struct timespec *ts1, u64 in_param, u64 out_param, u32 in_modifier, u8 op_modifier, u16 op, u16 token, int event) { struct mlx4_cmd *cmd = &mlx4_priv(dev)->cmd; u32 __iomem *hcr = cmd->hcr; int ret = -EAGAIN; unsigned long end; int err, go_bit = 0, t_bit = 0; u32 status = 0; mutex_lock(&cmd->hcr_mutex); if (pci_channel_offline(dev->pdev)) { /* * Device is going through error recovery * and cannot accept commands. */ ret = -EIO; goto out; } end = jiffies; if (event) end += msecs_to_jiffies(GO_BIT_TIMEOUT_MSECS); while (cmd_pending(dev)) { if (pci_channel_offline(dev->pdev)) { /* * Device is going through error recovery * and cannot accept commands. */ ret = -EIO; goto out; } if (time_after_eq(jiffies, end)) { mlx4_err(dev, "%s:cmd_pending failed\n", __func__); goto out; } cond_resched(); } /* * We use writel (instead of something like memcpy_toio) * because writes of less than 32 bits to the HCR don't work * (and some architectures such as ia64 implement memcpy_toio * in terms of writeb). */ __raw_writel((__force u32) cpu_to_be32(in_param >> 32), hcr + 0); __raw_writel((__force u32) cpu_to_be32(in_param & 0xfffffffful), hcr + 1); __raw_writel((__force u32) cpu_to_be32(in_modifier), hcr + 2); __raw_writel((__force u32) cpu_to_be32(out_param >> 32), hcr + 3); __raw_writel((__force u32) cpu_to_be32(out_param & 0xfffffffful), hcr + 4); __raw_writel((__force u32) cpu_to_be32(token << 16), hcr + 5); if (ts1) ktime_get_ts(ts1); /* __raw_writel may not order writes. */ wmb(); __raw_writel((__force u32) cpu_to_be32((1 << HCR_GO_BIT) | (cmd->toggle << HCR_T_BIT) | (event ? (1 << HCR_E_BIT) : 0) | (op_modifier << HCR_OPMOD_SHIFT) | op), hcr + 6); /* * Make sure that our HCR writes don't get mixed in with * writes from another CPU starting a FW command. */ mmiowb(); cmd->toggle = cmd->toggle ^ 1; ret = 0; out: if (ret) { err = get_status(dev, &status, &go_bit, &t_bit); mlx4_warn(dev, "Could not post command %s (0x%x): ret=%d, " "in_param=0x%llx, in_mod=0x%x, op_mod=0x%x, " "get_status err=%d, status_reg=0x%x, go_bit=%d, " "t_bit=%d, toggle=0x%x\n", cmd_to_str(op), op, ret, (unsigned long long) in_param, in_modifier, op_modifier, err, status, go_bit, t_bit, cmd->toggle); } mutex_unlock(&cmd->hcr_mutex); return ret; } static int mlx4_slave_cmd(struct mlx4_dev *dev, u64 in_param, u64 *out_param, int out_is_imm, u32 in_modifier, u8 op_modifier, u16 op, unsigned long timeout) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_vhcr_cmd *vhcr = priv->mfunc.vhcr; int ret; mutex_lock(&priv->cmd.slave_cmd_mutex); vhcr->in_param = cpu_to_be64(in_param); vhcr->out_param = out_param ? cpu_to_be64(*out_param) : 0; vhcr->in_modifier = cpu_to_be32(in_modifier); vhcr->opcode = cpu_to_be16((((u16) op_modifier) << 12) | (op & 0xfff)); vhcr->token = cpu_to_be16(CMD_POLL_TOKEN); vhcr->status = 0; vhcr->flags = !!(priv->cmd.use_events) << 6; if (mlx4_is_master(dev)) { ret = mlx4_master_process_vhcr(dev, dev->caps.function, vhcr); if (!ret) { if (out_is_imm) { if (out_param) *out_param = be64_to_cpu(vhcr->out_param); else { mlx4_err(dev, "response expected while" "output mailbox is NULL for " "command 0x%x\n", op); vhcr->status = CMD_STAT_BAD_PARAM; } } ret = mlx4_status_to_errno(vhcr->status); } } else { ret = mlx4_comm_cmd(dev, MLX4_COMM_CMD_VHCR_POST, 0, MLX4_COMM_TIME + timeout); if (!ret) { if (out_is_imm) { if (out_param) *out_param = be64_to_cpu(vhcr->out_param); else { mlx4_err(dev, "response expected while" "output mailbox is NULL for " "command 0x%x\n", op); vhcr->status = CMD_STAT_BAD_PARAM; } } ret = mlx4_status_to_errno(vhcr->status); } else mlx4_err(dev, "failed execution of VHCR_POST command" "opcode %s (0x%x)\n", cmd_to_str(op), op); } mutex_unlock(&priv->cmd.slave_cmd_mutex); return ret; } static int mlx4_cmd_poll(struct mlx4_dev *dev, u64 in_param, u64 *out_param, int out_is_imm, u32 in_modifier, u8 op_modifier, u16 op, unsigned long timeout) { struct mlx4_priv *priv = mlx4_priv(dev); void __iomem *hcr = priv->cmd.hcr; int err = 0; unsigned long end; u32 stat; down(&priv->cmd.poll_sem); if (pci_channel_offline(dev->pdev)) { /* * Device is going through error recovery * and cannot accept commands. */ err = -EIO; goto out; } err = mlx4_cmd_post(dev, NULL, in_param, out_param ? *out_param : 0, in_modifier, op_modifier, op, CMD_POLL_TOKEN, 0); if (err) goto out; end = msecs_to_jiffies(timeout) + jiffies; while (cmd_pending(dev) && time_before(jiffies, end)) { if (pci_channel_offline(dev->pdev)) { /* * Device is going through error recovery * and cannot accept commands. */ err = -EIO; goto out; } cond_resched(); } if (cmd_pending(dev)) { mlx4_warn(dev, "command %s (0x%x) timed out (go bit not cleared)\n", cmd_to_str(op), op); err = -ETIMEDOUT; goto out; } if (out_is_imm) *out_param = (u64) be32_to_cpu((__force __be32) __raw_readl(hcr + HCR_OUT_PARAM_OFFSET)) << 32 | (u64) be32_to_cpu((__force __be32) __raw_readl(hcr + HCR_OUT_PARAM_OFFSET + 4)); stat = be32_to_cpu((__force __be32) __raw_readl(hcr + HCR_STATUS_OFFSET)) >> 24; err = mlx4_status_to_errno(stat); if (err) mlx4_err(dev, "command %s (0x%x) failed: fw status = 0x%x\n", cmd_to_str(op), op, stat); out: up(&priv->cmd.poll_sem); return err; } void mlx4_cmd_event(struct mlx4_dev *dev, u16 token, u8 status, u64 out_param) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_cmd_context *context = &priv->cmd.context[token & priv->cmd.token_mask]; /* previously timed out command completing at long last */ if (token != context->token) return; context->fw_status = status; context->result = mlx4_status_to_errno(status); context->out_param = out_param; complete(&context->done); } static int mlx4_cmd_wait(struct mlx4_dev *dev, u64 in_param, u64 *out_param, int out_is_imm, u32 in_modifier, u8 op_modifier, u16 op, unsigned long timeout) { struct mlx4_cmd *cmd = &mlx4_priv(dev)->cmd; struct mlx4_cmd_context *context; int err = 0; int go_bit = 0, t_bit = 0, stat_err; u32 status = 0; struct timespec ts1, ts2; ktime_t t1, t2, delta; s64 ds; if (out_is_imm && !out_param) return -EINVAL; down(&cmd->event_sem); spin_lock(&cmd->context_lock); BUG_ON(cmd->free_head < 0); context = &cmd->context[cmd->free_head]; context->token += cmd->token_mask + 1; cmd->free_head = context->next; spin_unlock(&cmd->context_lock); init_completion(&context->done); err = mlx4_cmd_post(dev, &ts1, in_param, out_param ? *out_param : 0, in_modifier, op_modifier, op, context->token, 1); if (err) goto out; if (!wait_for_completion_timeout(&context->done, msecs_to_jiffies(timeout))) { stat_err = get_status(dev, &status, &go_bit, &t_bit); mlx4_warn(dev, "command %s (0x%x) timed out: in_param=0x%llx, " "in_mod=0x%x, op_mod=0x%x, get_status err=%d, " "status_reg=0x%x, go_bit=%d, t_bit=%d, toggle=0x%x\n" , cmd_to_str(op), op, (unsigned long long) in_param, in_modifier, op_modifier, stat_err, status, go_bit, t_bit, mlx4_priv(dev)->cmd.toggle); err = -EBUSY; goto out; } if (mlx4_debug_level & MLX4_DEBUG_MASK_CMD_TIME) { ktime_get_ts(&ts2); t1 = timespec_to_ktime(ts1); t2 = timespec_to_ktime(ts2); delta = ktime_sub(t2, t1); ds = ktime_to_ns(delta); pr_info("mlx4: fw exec time for %s is %lld nsec\n", cmd_to_str(op), (long long) ds); } err = context->result; if (err) { mlx4_err(dev, "command %s (0x%x) failed: in_param=0x%llx, " "in_mod=0x%x, op_mod=0x%x, fw status = 0x%x\n", cmd_to_str(op), op, (unsigned long long) in_param, in_modifier, op_modifier, context->fw_status); + + switch(context->fw_status) { + case CMD_STAT_BAD_PARAM: + mlx4_err(dev, "Parameter is not supported, " + "parameter is out of range\n"); + break; + case CMD_STAT_EXCEED_LIM: + mlx4_err(dev, "Required capability exceeded " + "device limits\n"); + break; + default: + break; + } goto out; } if (out_is_imm) *out_param = context->out_param; out: spin_lock(&cmd->context_lock); context->next = cmd->free_head; cmd->free_head = context - cmd->context; spin_unlock(&cmd->context_lock); up(&cmd->event_sem); return err; } int __mlx4_cmd(struct mlx4_dev *dev, u64 in_param, u64 *out_param, int out_is_imm, u32 in_modifier, u8 op_modifier, u16 op, unsigned long timeout, int native) { if (pci_channel_offline(dev->pdev)) return -EIO; if (!mlx4_is_mfunc(dev) || (native && mlx4_is_master(dev))) { if (mlx4_priv(dev)->cmd.use_events) return mlx4_cmd_wait(dev, in_param, out_param, out_is_imm, in_modifier, op_modifier, op, timeout); else return mlx4_cmd_poll(dev, in_param, out_param, out_is_imm, in_modifier, op_modifier, op, timeout); } return mlx4_slave_cmd(dev, in_param, out_param, out_is_imm, in_modifier, op_modifier, op, timeout); } EXPORT_SYMBOL_GPL(__mlx4_cmd); static int mlx4_ARM_COMM_CHANNEL(struct mlx4_dev *dev) { return mlx4_cmd(dev, 0, 0, 0, MLX4_CMD_ARM_COMM_CHANNEL, MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE); } static int mlx4_ACCESS_MEM(struct mlx4_dev *dev, u64 master_addr, int slave, u64 slave_addr, int size, int is_read) { u64 in_param; u64 out_param; if ((slave_addr & 0xfff) | (master_addr & 0xfff) | (slave & ~0x7f) | (size & 0xff)) { mlx4_err(dev, "Bad access mem params - slave_addr:0x%llx " "master_addr:0x%llx slave_id:%d size:%d\n", (unsigned long long) slave_addr, (unsigned long long) master_addr, slave, size); return -EINVAL; } if (is_read) { in_param = (u64) slave | slave_addr; out_param = (u64) dev->caps.function | master_addr; } else { in_param = (u64) dev->caps.function | master_addr; out_param = (u64) slave | slave_addr; } return mlx4_cmd_imm(dev, in_param, &out_param, size, 0, MLX4_CMD_ACCESS_MEM, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE); } static int query_pkey_block(struct mlx4_dev *dev, u8 port, u16 index, u16 *pkey, struct mlx4_cmd_mailbox *inbox, struct mlx4_cmd_mailbox *outbox) { struct ib_smp *in_mad = (struct ib_smp *)(inbox->buf); struct ib_smp *out_mad = (struct ib_smp *)(outbox->buf); int err; int i; if (index & 0x1f) return -EINVAL; in_mad->attr_mod = cpu_to_be32(index / 32); err = mlx4_cmd_box(dev, inbox->dma, outbox->dma, port, 3, MLX4_CMD_MAD_IFC, MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE); if (err) return err; for (i = 0; i < 32; ++i) pkey[i] = be16_to_cpu(((__be16 *) out_mad->data)[i]); return err; } static int get_full_pkey_table(struct mlx4_dev *dev, u8 port, u16 *table, struct mlx4_cmd_mailbox *inbox, struct mlx4_cmd_mailbox *outbox) { int i; int err; for (i = 0; i < dev->caps.pkey_table_len[port]; i += 32) { err = query_pkey_block(dev, port, i, table + i, inbox, outbox); if (err) return err; } return 0; } #define PORT_CAPABILITY_LOCATION_IN_SMP 20 #define PORT_STATE_OFFSET 32 static enum ib_port_state vf_port_state(struct mlx4_dev *dev, int port, int vf) { if (mlx4_get_slave_port_state(dev, vf, port) == SLAVE_PORT_UP) return IB_PORT_ACTIVE; else return IB_PORT_DOWN; } static int mlx4_MAD_IFC_wrapper(struct mlx4_dev *dev, int slave, struct mlx4_vhcr *vhcr, struct mlx4_cmd_mailbox *inbox, struct mlx4_cmd_mailbox *outbox, struct mlx4_cmd_info *cmd) { struct ib_smp *smp = inbox->buf; u32 index; u8 port; u16 *table; int err; int vidx, pidx; struct mlx4_priv *priv = mlx4_priv(dev); struct ib_smp *outsmp = outbox->buf; __be16 *outtab = (__be16 *)(outsmp->data); __be32 slave_cap_mask; __be64 slave_node_guid; port = vhcr->in_modifier; if (smp->base_version == 1 && smp->mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED && smp->class_version == 1) { if (smp->method == IB_MGMT_METHOD_GET) { if (smp->attr_id == IB_SMP_ATTR_PKEY_TABLE) { index = be32_to_cpu(smp->attr_mod); if (port < 1 || port > dev->caps.num_ports) return -EINVAL; table = kcalloc(dev->caps.pkey_table_len[port], sizeof *table, GFP_KERNEL); if (!table) return -ENOMEM; /* need to get the full pkey table because the paravirtualized * pkeys may be scattered among several pkey blocks. */ err = get_full_pkey_table(dev, port, table, inbox, outbox); if (!err) { for (vidx = index * 32; vidx < (index + 1) * 32; ++vidx) { pidx = priv->virt2phys_pkey[slave][port - 1][vidx]; outtab[vidx % 32] = cpu_to_be16(table[pidx]); } } kfree(table); return err; } if (smp->attr_id == IB_SMP_ATTR_PORT_INFO) { /*get the slave specific caps:*/ /*do the command */ err = mlx4_cmd_box(dev, inbox->dma, outbox->dma, vhcr->in_modifier, vhcr->op_modifier, vhcr->op, MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE); /* modify the response for slaves */ if (!err && slave != mlx4_master_func_num(dev)) { u8 *state = outsmp->data + PORT_STATE_OFFSET; *state = (*state & 0xf0) | vf_port_state(dev, port, slave); slave_cap_mask = priv->mfunc.master.slave_state[slave].ib_cap_mask[port]; memcpy(outsmp->data + PORT_CAPABILITY_LOCATION_IN_SMP, &slave_cap_mask, 4); } return err; } if (smp->attr_id == IB_SMP_ATTR_GUID_INFO) { /* compute slave's gid block */ smp->attr_mod = cpu_to_be32(slave / 8); /* execute cmd */ err = mlx4_cmd_box(dev, inbox->dma, outbox->dma, vhcr->in_modifier, vhcr->op_modifier, vhcr->op, MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE); if (!err) { /* if needed, move slave gid to index 0 */ if (slave % 8) memcpy(outsmp->data, outsmp->data + (slave % 8) * 8, 8); /* delete all other gids */ memset(outsmp->data + 8, 0, 56); } return err; } if (smp->attr_id == IB_SMP_ATTR_NODE_INFO) { err = mlx4_cmd_box(dev, inbox->dma, outbox->dma, vhcr->in_modifier, vhcr->op_modifier, vhcr->op, MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE); if (!err) { slave_node_guid = mlx4_get_slave_node_guid(dev, slave); memcpy(outsmp->data + 12, &slave_node_guid, 8); } return err; } } } if (slave != mlx4_master_func_num(dev) && ((smp->mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) || (smp->mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED && smp->method == IB_MGMT_METHOD_SET))) { mlx4_err(dev, "slave %d is trying to execute a Subnet MGMT MAD, " "class 0x%x, method 0x%x for attr 0x%x. Rejecting\n", slave, smp->method, smp->mgmt_class, be16_to_cpu(smp->attr_id)); return -EPERM; } /*default:*/ return mlx4_cmd_box(dev, inbox->dma, outbox->dma, vhcr->in_modifier, vhcr->op_modifier, vhcr->op, MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE); } static int MLX4_CMD_DIAG_RPRT_wrapper(struct mlx4_dev *dev, int slave, struct mlx4_vhcr *vhcr, struct mlx4_cmd_mailbox *inbox, struct mlx4_cmd_mailbox *outbox, struct mlx4_cmd_info *cmd) { return -EPERM; } static int MLX4_CMD_UPDATE_QP_wrapper(struct mlx4_dev *dev, int slave, struct mlx4_vhcr *vhcr, struct mlx4_cmd_mailbox *inbox, struct mlx4_cmd_mailbox *outbox, struct mlx4_cmd_info *cmd) { return -EPERM; } int mlx4_DMA_wrapper(struct mlx4_dev *dev, int slave, struct mlx4_vhcr *vhcr, struct mlx4_cmd_mailbox *inbox, struct mlx4_cmd_mailbox *outbox, struct mlx4_cmd_info *cmd) { u64 in_param; u64 out_param; int err; in_param = cmd->has_inbox ? (u64) inbox->dma : vhcr->in_param; out_param = cmd->has_outbox ? (u64) outbox->dma : vhcr->out_param; if (cmd->encode_slave_id) { in_param &= 0xffffffffffffff00ll; in_param |= slave; } err = __mlx4_cmd(dev, in_param, &out_param, cmd->out_is_imm, vhcr->in_modifier, vhcr->op_modifier, vhcr->op, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE); if (cmd->out_is_imm) vhcr->out_param = out_param; return err; } static struct mlx4_cmd_info cmd_info[] = { { .opcode = MLX4_CMD_QUERY_FW, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_QUERY_FW_wrapper }, { .opcode = MLX4_CMD_QUERY_HCA, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = NULL }, { .opcode = MLX4_CMD_QUERY_DEV_CAP, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_QUERY_DEV_CAP_wrapper }, { .opcode = MLX4_CMD_QUERY_FUNC_CAP, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_QUERY_FUNC_CAP_wrapper }, { .opcode = MLX4_CMD_QUERY_ADAPTER, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = NULL }, { .opcode = MLX4_CMD_INIT_PORT, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_INIT_PORT_wrapper }, { .opcode = MLX4_CMD_CLOSE_PORT, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_CLOSE_PORT_wrapper }, { .opcode = MLX4_CMD_QUERY_PORT, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_QUERY_PORT_wrapper }, { .opcode = MLX4_CMD_SET_PORT, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_SET_PORT_wrapper }, { .opcode = MLX4_CMD_MAP_EQ, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_MAP_EQ_wrapper }, { .opcode = MLX4_CMD_SW2HW_EQ, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = true, .verify = NULL, .wrapper = mlx4_SW2HW_EQ_wrapper }, { .opcode = MLX4_CMD_HW_HEALTH_CHECK, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = NULL }, { .opcode = MLX4_CMD_DIAG_RPRT, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .skip_err_print = true, .verify = NULL, .wrapper = MLX4_CMD_DIAG_RPRT_wrapper }, { .opcode = MLX4_CMD_NOP, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = NULL }, { .opcode = MLX4_CMD_ALLOC_RES, .has_inbox = false, .has_outbox = false, .out_is_imm = true, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_ALLOC_RES_wrapper }, { .opcode = MLX4_CMD_FREE_RES, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_FREE_RES_wrapper }, { .opcode = MLX4_CMD_SW2HW_MPT, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = true, .verify = NULL, .wrapper = mlx4_SW2HW_MPT_wrapper }, { .opcode = MLX4_CMD_QUERY_MPT, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_QUERY_MPT_wrapper }, { .opcode = MLX4_CMD_HW2SW_MPT, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_HW2SW_MPT_wrapper }, { .opcode = MLX4_CMD_READ_MTT, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = NULL }, { .opcode = MLX4_CMD_WRITE_MTT, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_WRITE_MTT_wrapper }, { .opcode = MLX4_CMD_SYNC_TPT, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = NULL }, { .opcode = MLX4_CMD_HW2SW_EQ, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = true, .verify = NULL, .wrapper = mlx4_HW2SW_EQ_wrapper }, { .opcode = MLX4_CMD_QUERY_EQ, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = true, .verify = NULL, .wrapper = mlx4_QUERY_EQ_wrapper }, { .opcode = MLX4_CMD_SW2HW_CQ, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = true, .verify = NULL, .wrapper = mlx4_SW2HW_CQ_wrapper }, { .opcode = MLX4_CMD_HW2SW_CQ, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_HW2SW_CQ_wrapper }, { .opcode = MLX4_CMD_QUERY_CQ, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_QUERY_CQ_wrapper }, { .opcode = MLX4_CMD_MODIFY_CQ, .has_inbox = true, .has_outbox = false, .out_is_imm = true, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_MODIFY_CQ_wrapper }, { .opcode = MLX4_CMD_SW2HW_SRQ, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = true, .verify = NULL, .wrapper = mlx4_SW2HW_SRQ_wrapper }, { .opcode = MLX4_CMD_HW2SW_SRQ, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_HW2SW_SRQ_wrapper }, { .opcode = MLX4_CMD_QUERY_SRQ, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_QUERY_SRQ_wrapper }, { .opcode = MLX4_CMD_ARM_SRQ, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_ARM_SRQ_wrapper }, { .opcode = MLX4_CMD_RST2INIT_QP, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = true, .verify = NULL, .wrapper = mlx4_RST2INIT_QP_wrapper }, { .opcode = MLX4_CMD_INIT2INIT_QP, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_INIT2INIT_QP_wrapper }, { .opcode = MLX4_CMD_INIT2RTR_QP, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_INIT2RTR_QP_wrapper }, { .opcode = MLX4_CMD_RTR2RTS_QP, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_RTR2RTS_QP_wrapper }, { .opcode = MLX4_CMD_RTS2RTS_QP, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_RTS2RTS_QP_wrapper }, { .opcode = MLX4_CMD_SQERR2RTS_QP, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_SQERR2RTS_QP_wrapper }, { .opcode = MLX4_CMD_2ERR_QP, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_GEN_QP_wrapper }, { .opcode = MLX4_CMD_RTS2SQD_QP, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_GEN_QP_wrapper }, { .opcode = MLX4_CMD_SQD2SQD_QP, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_SQD2SQD_QP_wrapper }, { .opcode = MLX4_CMD_SQD2RTS_QP, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_SQD2RTS_QP_wrapper }, { .opcode = MLX4_CMD_2RST_QP, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_2RST_QP_wrapper }, { .opcode = MLX4_CMD_QUERY_QP, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_GEN_QP_wrapper }, { .opcode = MLX4_CMD_SUSPEND_QP, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_GEN_QP_wrapper }, { .opcode = MLX4_CMD_UNSUSPEND_QP, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_GEN_QP_wrapper }, { .opcode = MLX4_CMD_UPDATE_QP, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .skip_err_print = true, .verify = NULL, .wrapper = MLX4_CMD_UPDATE_QP_wrapper }, { .opcode = MLX4_CMD_CONF_SPECIAL_QP, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, /* XXX verify: only demux can do this */ .wrapper = NULL }, { .opcode = MLX4_CMD_MAD_IFC, .has_inbox = true, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_MAD_IFC_wrapper }, { .opcode = MLX4_CMD_QUERY_IF_STAT, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_QUERY_IF_STAT_wrapper }, /* Native multicast commands are not available for guests */ { .opcode = MLX4_CMD_QP_ATTACH, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_QP_ATTACH_wrapper }, { .opcode = MLX4_CMD_PROMISC, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_PROMISC_wrapper }, /* Ethernet specific commands */ { .opcode = MLX4_CMD_SET_VLAN_FLTR, .has_inbox = true, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_SET_VLAN_FLTR_wrapper }, { .opcode = MLX4_CMD_SET_MCAST_FLTR, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_SET_MCAST_FLTR_wrapper }, { .opcode = MLX4_CMD_DUMP_ETH_STATS, .has_inbox = false, .has_outbox = true, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_DUMP_ETH_STATS_wrapper }, { .opcode = MLX4_CMD_INFORM_FLR_DONE, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = NULL }, /* flow steering commands */ { .opcode = MLX4_QP_FLOW_STEERING_ATTACH, .has_inbox = true, .has_outbox = false, .out_is_imm = true, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_QP_FLOW_STEERING_ATTACH_wrapper }, { .opcode = MLX4_QP_FLOW_STEERING_DETACH, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .verify = NULL, .wrapper = mlx4_QP_FLOW_STEERING_DETACH_wrapper }, /* wol commands */ { .opcode = MLX4_CMD_MOD_STAT_CFG, .has_inbox = false, .has_outbox = false, .out_is_imm = false, .encode_slave_id = false, .skip_err_print = true, .verify = NULL, .wrapper = mlx4_MOD_STAT_CFG_wrapper }, }; static int mlx4_master_process_vhcr(struct mlx4_dev *dev, int slave, struct mlx4_vhcr_cmd *in_vhcr) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_cmd_info *cmd = NULL; struct mlx4_vhcr_cmd *vhcr_cmd = in_vhcr ? in_vhcr : priv->mfunc.vhcr; struct mlx4_vhcr *vhcr; struct mlx4_cmd_mailbox *inbox = NULL; struct mlx4_cmd_mailbox *outbox = NULL; u64 in_param; u64 out_param; int ret = 0; int i; int err = 0; /* Create sw representation of Virtual HCR */ vhcr = kzalloc(sizeof(struct mlx4_vhcr), GFP_KERNEL); if (!vhcr) return -ENOMEM; /* DMA in the vHCR */ if (!in_vhcr) { ret = mlx4_ACCESS_MEM(dev, priv->mfunc.vhcr_dma, slave, priv->mfunc.master.slave_state[slave].vhcr_dma, ALIGN(sizeof(struct mlx4_vhcr_cmd), MLX4_ACCESS_MEM_ALIGN), 1); if (ret) { mlx4_err(dev, "%s:Failed reading vhcr" "ret: 0x%x\n", __func__, ret); kfree(vhcr); return ret; } } /* Fill SW VHCR fields */ vhcr->in_param = be64_to_cpu(vhcr_cmd->in_param); vhcr->out_param = be64_to_cpu(vhcr_cmd->out_param); vhcr->in_modifier = be32_to_cpu(vhcr_cmd->in_modifier); vhcr->token = be16_to_cpu(vhcr_cmd->token); vhcr->op = be16_to_cpu(vhcr_cmd->opcode) & 0xfff; vhcr->op_modifier = (u8) (be16_to_cpu(vhcr_cmd->opcode) >> 12); vhcr->e_bit = vhcr_cmd->flags & (1 << 6); /* Lookup command */ for (i = 0; i < ARRAY_SIZE(cmd_info); ++i) { if (vhcr->op == cmd_info[i].opcode) { cmd = &cmd_info[i]; break; } } if (!cmd) { mlx4_err(dev, "unparavirt command: %s (0x%x) accepted from slave:%d\n", cmd_to_str(vhcr->op), vhcr->op, slave); vhcr_cmd->status = CMD_STAT_BAD_PARAM; goto out_status; } /* Read inbox */ if (cmd->has_inbox) { vhcr->in_param &= INBOX_MASK; inbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(inbox)) { vhcr_cmd->status = CMD_STAT_BAD_SIZE; inbox = NULL; goto out_status; } if (mlx4_ACCESS_MEM(dev, inbox->dma, slave, vhcr->in_param, MLX4_MAILBOX_SIZE, 1)) { mlx4_err(dev, "%s: Failed reading inbox for cmd %s (0x%x)\n", __func__, cmd_to_str(cmd->opcode), cmd->opcode); vhcr_cmd->status = CMD_STAT_INTERNAL_ERR; goto out_status; } } /* Apply permission and bound checks if applicable */ if (cmd->verify && cmd->verify(dev, slave, vhcr, inbox)) { mlx4_warn(dev, "Command %s (0x%x) from slave: %d failed protection " "checks for resource_id: %d\n", cmd_to_str(vhcr->op), vhcr->op, slave, vhcr->in_modifier); vhcr_cmd->status = CMD_STAT_BAD_OP; goto out_status; } /* Allocate outbox */ if (cmd->has_outbox) { outbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(outbox)) { vhcr_cmd->status = CMD_STAT_BAD_SIZE; outbox = NULL; goto out_status; } } /* Execute the command! */ if (cmd->wrapper) { err = cmd->wrapper(dev, slave, vhcr, inbox, outbox, cmd); if (cmd->out_is_imm) vhcr_cmd->out_param = cpu_to_be64(vhcr->out_param); } else { in_param = cmd->has_inbox ? (u64) inbox->dma : vhcr->in_param; out_param = cmd->has_outbox ? (u64) outbox->dma : vhcr->out_param; err = __mlx4_cmd(dev, in_param, &out_param, cmd->out_is_imm, vhcr->in_modifier, vhcr->op_modifier, vhcr->op, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE); if (cmd->out_is_imm) { vhcr->out_param = out_param; vhcr_cmd->out_param = cpu_to_be64(vhcr->out_param); } } if (err) { if (!cmd->skip_err_print) mlx4_warn(dev, "vhcr command %s (0x%x) slave:%d " "in_param 0x%llx in_mod=0x%x, op_mod=0x%x " "failed with error:%d, status %d\n", cmd_to_str(vhcr->op), vhcr->op, slave, (unsigned long long) vhcr->in_param, vhcr->in_modifier, vhcr->op_modifier, vhcr->errno, err); vhcr_cmd->status = mlx4_errno_to_status(err); goto out_status; } /* Write outbox if command completed successfully */ if (cmd->has_outbox && !vhcr_cmd->status) { ret = mlx4_ACCESS_MEM(dev, outbox->dma, slave, vhcr->out_param, MLX4_MAILBOX_SIZE, MLX4_CMD_WRAPPED); if (ret) { /* If we failed to write back the outbox after the *command was successfully executed, we must fail this * slave, as it is now in undefined state */ mlx4_err(dev, "%s: Failed writing outbox\n", __func__); goto out; } } out_status: /* DMA back vhcr result */ if (!in_vhcr) { ret = mlx4_ACCESS_MEM(dev, priv->mfunc.vhcr_dma, slave, priv->mfunc.master.slave_state[slave].vhcr_dma, ALIGN(sizeof(struct mlx4_vhcr), MLX4_ACCESS_MEM_ALIGN), MLX4_CMD_WRAPPED); if (ret) mlx4_err(dev, "%s:Failed writing vhcr result\n", __func__); else if (vhcr->e_bit && mlx4_GEN_EQE(dev, slave, &priv->mfunc.master.cmd_eqe)) mlx4_warn(dev, "Failed to generate command completion " "eqe for slave %d\n", slave); } out: kfree(vhcr); mlx4_free_cmd_mailbox(dev, inbox); mlx4_free_cmd_mailbox(dev, outbox); return ret; } static int mlx4_master_immediate_activate_vlan_qos(struct mlx4_priv *priv, int slave, int port) { struct mlx4_vport_oper_state *vp_oper; struct mlx4_vport_state *vp_admin; struct mlx4_vf_immed_vlan_work *work; int err; int admin_vlan_ix = NO_INDX; vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port]; vp_admin = &priv->mfunc.master.vf_admin[slave].vport[port]; if (vp_oper->state.default_vlan == vp_admin->default_vlan && vp_oper->state.default_qos == vp_admin->default_qos) return 0; work = kzalloc(sizeof(*work), GFP_KERNEL); if (!work) return -ENOMEM; if (vp_oper->state.default_vlan != vp_admin->default_vlan) { if (MLX4_VGT != vp_admin->default_vlan) { err = __mlx4_register_vlan(&priv->dev, port, vp_admin->default_vlan, &admin_vlan_ix); if (err) { mlx4_warn((&priv->dev), "No vlan resources slave %d, port %d\n", slave, port); return err; } } else { admin_vlan_ix = NO_INDX; } work->flags |= MLX4_VF_IMMED_VLAN_FLAG_VLAN; mlx4_dbg((&(priv->dev)), "alloc vlan %d idx %d slave %d port %d\n", (int)(vp_admin->default_vlan), admin_vlan_ix, slave, port); } /* save original vlan ix and vlan id */ work->orig_vlan_id = vp_oper->state.default_vlan; work->orig_vlan_ix = vp_oper->vlan_idx; /* handle new qos */ if (vp_oper->state.default_qos != vp_admin->default_qos) work->flags |= MLX4_VF_IMMED_VLAN_FLAG_QOS; if (work->flags & MLX4_VF_IMMED_VLAN_FLAG_VLAN) vp_oper->vlan_idx = admin_vlan_ix; vp_oper->state.default_vlan = vp_admin->default_vlan; vp_oper->state.default_qos = vp_admin->default_qos; /* iterate over QPs owned by this slave, using UPDATE_QP */ work->port = port; work->slave = slave; work->qos = vp_oper->state.default_qos; work->vlan_id = vp_oper->state.default_vlan; work->vlan_ix = vp_oper->vlan_idx; work->priv = priv; INIT_WORK(&work->work, mlx4_vf_immed_vlan_work_handler); queue_work(priv->mfunc.master.comm_wq, &work->work); return 0; } static int mlx4_master_activate_admin_state(struct mlx4_priv *priv, int slave) { int port, err; struct mlx4_vport_state *vp_admin; struct mlx4_vport_oper_state *vp_oper; for (port = 1; port <= MLX4_MAX_PORTS; port++) { vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port]; vp_admin = &priv->mfunc.master.vf_admin[slave].vport[port]; vp_oper->state = *vp_admin; if (MLX4_VGT != vp_admin->default_vlan) { err = __mlx4_register_vlan(&priv->dev, port, vp_admin->default_vlan, &(vp_oper->vlan_idx)); if (err) { vp_oper->vlan_idx = NO_INDX; mlx4_warn((&priv->dev), "No vlan resorces slave %d, port %d\n", slave, port); return err; } mlx4_dbg((&(priv->dev)), "alloc vlan %d idx %d slave %d port %d\n", (int)(vp_oper->state.default_vlan), vp_oper->vlan_idx, slave, port); } if (vp_admin->spoofchk) { vp_oper->mac_idx = __mlx4_register_mac(&priv->dev, port, vp_admin->mac); if (0 > vp_oper->mac_idx) { err = vp_oper->mac_idx; vp_oper->mac_idx = NO_INDX; mlx4_warn((&priv->dev), "No mac resources slave %d, port %d\n", slave, port); return err; } mlx4_dbg((&(priv->dev)), "alloc mac %llx idx %d slave %d port %d\n", (unsigned long long) vp_oper->state.mac, vp_oper->mac_idx, slave, port); } } return 0; } static void mlx4_master_deactivate_admin_state(struct mlx4_priv *priv, int slave) { int port; struct mlx4_vport_oper_state *vp_oper; for (port = 1; port <= MLX4_MAX_PORTS; port++) { vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port]; if (NO_INDX != vp_oper->vlan_idx) { __mlx4_unregister_vlan(&priv->dev, port, vp_oper->state.default_vlan); vp_oper->vlan_idx = NO_INDX; } if (NO_INDX != vp_oper->mac_idx) { __mlx4_unregister_mac(&priv->dev, port, vp_oper->state.mac); vp_oper->mac_idx = NO_INDX; } } return; } static void mlx4_master_do_cmd(struct mlx4_dev *dev, int slave, u8 cmd, u16 param, u8 toggle) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_slave_state *slave_state = priv->mfunc.master.slave_state; u32 reply; u8 is_going_down = 0; int i; unsigned long flags; slave_state[slave].comm_toggle ^= 1; reply = (u32) slave_state[slave].comm_toggle << 31; if (toggle != slave_state[slave].comm_toggle) { mlx4_warn(dev, "Incorrect toggle %d from slave %d. *** MASTER" "STATE COMPROMISIED ***\n", toggle, slave); goto reset_slave; } if (cmd == MLX4_COMM_CMD_RESET) { mlx4_warn(dev, "Received reset from slave:%d\n", slave); slave_state[slave].active = false; slave_state[slave].old_vlan_api = false; mlx4_master_deactivate_admin_state(priv, slave); for (i = 0; i < MLX4_EVENT_TYPES_NUM; ++i) { slave_state[slave].event_eq[i].eqn = -1; slave_state[slave].event_eq[i].token = 0; } /*check if we are in the middle of FLR process, if so return "retry" status to the slave*/ if (MLX4_COMM_CMD_FLR == slave_state[slave].last_cmd) goto inform_slave_state; mlx4_dispatch_event(dev, MLX4_DEV_EVENT_SLAVE_SHUTDOWN, slave); /* write the version in the event field */ reply |= mlx4_comm_get_version(); goto reset_slave; } /*command from slave in the middle of FLR*/ if (cmd != MLX4_COMM_CMD_RESET && MLX4_COMM_CMD_FLR == slave_state[slave].last_cmd) { mlx4_warn(dev, "slave:%d is Trying to run cmd (0x%x) " "in the middle of FLR\n", slave, cmd); return; } switch (cmd) { case MLX4_COMM_CMD_VHCR0: if (slave_state[slave].last_cmd != MLX4_COMM_CMD_RESET) goto reset_slave; slave_state[slave].vhcr_dma = ((u64) param) << 48; priv->mfunc.master.slave_state[slave].cookie = 0; break; case MLX4_COMM_CMD_VHCR1: if (slave_state[slave].last_cmd != MLX4_COMM_CMD_VHCR0) goto reset_slave; slave_state[slave].vhcr_dma |= ((u64) param) << 32; break; case MLX4_COMM_CMD_VHCR2: if (slave_state[slave].last_cmd != MLX4_COMM_CMD_VHCR1) goto reset_slave; slave_state[slave].vhcr_dma |= ((u64) param) << 16; break; case MLX4_COMM_CMD_VHCR_EN: if (slave_state[slave].last_cmd != MLX4_COMM_CMD_VHCR2) goto reset_slave; slave_state[slave].vhcr_dma |= param; if (mlx4_master_activate_admin_state(priv, slave)) goto reset_slave; slave_state[slave].active = true; mlx4_dispatch_event(dev, MLX4_DEV_EVENT_SLAVE_INIT, slave); break; case MLX4_COMM_CMD_VHCR_POST: if ((slave_state[slave].last_cmd != MLX4_COMM_CMD_VHCR_EN) && (slave_state[slave].last_cmd != MLX4_COMM_CMD_VHCR_POST)) goto reset_slave; mutex_lock(&priv->cmd.slave_cmd_mutex); if (mlx4_master_process_vhcr(dev, slave, NULL)) { mlx4_err(dev, "Failed processing vhcr for slave: %d," " resetting slave.\n", slave); mutex_unlock(&priv->cmd.slave_cmd_mutex); goto reset_slave; } mutex_unlock(&priv->cmd.slave_cmd_mutex); break; default: mlx4_warn(dev, "Bad comm cmd: %d from slave: %d\n", cmd, slave); goto reset_slave; } spin_lock_irqsave(&priv->mfunc.master.slave_state_lock, flags); if (!slave_state[slave].is_slave_going_down) slave_state[slave].last_cmd = cmd; else is_going_down = 1; spin_unlock_irqrestore(&priv->mfunc.master.slave_state_lock, flags); if (is_going_down) { mlx4_warn(dev, "Slave is going down aborting command (%d)" " executing from slave: %d\n", cmd, slave); return; } __raw_writel((__force u32) cpu_to_be32(reply), &priv->mfunc.comm[slave].slave_read); mmiowb(); return; reset_slave: /* cleanup any slave resources */ mlx4_delete_all_resources_for_slave(dev, slave); spin_lock_irqsave(&priv->mfunc.master.slave_state_lock, flags); if (!slave_state[slave].is_slave_going_down) slave_state[slave].last_cmd = MLX4_COMM_CMD_RESET; spin_unlock_irqrestore(&priv->mfunc.master.slave_state_lock, flags); /*with slave in the middle of flr, no need to clean resources again.*/ inform_slave_state: __raw_writel((__force u32) cpu_to_be32(reply), &priv->mfunc.comm[slave].slave_read); wmb(); } /* master command processing */ void mlx4_master_comm_channel(struct work_struct *work) { struct mlx4_mfunc_master_ctx *master = container_of(work, struct mlx4_mfunc_master_ctx, comm_work); struct mlx4_mfunc *mfunc = container_of(master, struct mlx4_mfunc, master); struct mlx4_priv *priv = container_of(mfunc, struct mlx4_priv, mfunc); struct mlx4_dev *dev = &priv->dev; __be32 *bit_vec; u32 comm_cmd; u32 vec; int i, j, slave; int toggle; int served = 0; int reported = 0; u32 slt; bit_vec = master->comm_arm_bit_vector; for (i = 0; i < COMM_CHANNEL_BIT_ARRAY_SIZE; i++) { vec = be32_to_cpu(bit_vec[i]); for (j = 0; j < 32; j++) { if (!(vec & (1 << j))) continue; ++reported; slave = (i * 32) + j; comm_cmd = swab32(readl( &mfunc->comm[slave].slave_write)); slt = swab32(readl(&mfunc->comm[slave].slave_read)) >> 31; toggle = comm_cmd >> 31; if (toggle != slt) { if (master->slave_state[slave].comm_toggle != slt) { mlx4_info(dev, "slave %d out of sync." " read toggle %d, state toggle %d. " "Resynching.\n", slave, slt, master->slave_state[slave].comm_toggle); master->slave_state[slave].comm_toggle = slt; } mlx4_master_do_cmd(dev, slave, comm_cmd >> 16 & 0xff, comm_cmd & 0xffff, toggle); ++served; } else mlx4_err(dev, "slave %d out of sync." " read toggle %d, write toggle %d.\n", slave, slt, toggle); } } if (reported && reported != served) mlx4_warn(dev, "Got command event with bitmask from %d slaves" " but %d were served\n", reported, served); } /* master command processing */ void mlx4_master_arm_comm_channel(struct work_struct *work) { struct mlx4_mfunc_master_ctx *master = container_of(work, struct mlx4_mfunc_master_ctx, arm_comm_work); struct mlx4_mfunc *mfunc = container_of(master, struct mlx4_mfunc, master); struct mlx4_priv *priv = container_of(mfunc, struct mlx4_priv, mfunc); struct mlx4_dev *dev = &priv->dev; if (mlx4_ARM_COMM_CHANNEL(dev)) mlx4_warn(dev, "Failed to arm comm channel events\n"); } static int sync_toggles(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); int wr_toggle; int rd_toggle; unsigned long end; wr_toggle = swab32(readl(&priv->mfunc.comm->slave_write)) >> 31; end = jiffies + msecs_to_jiffies(5000); while (time_before(jiffies, end)) { rd_toggle = swab32(readl(&priv->mfunc.comm->slave_read)) >> 31; if (rd_toggle == wr_toggle) { priv->cmd.comm_toggle = rd_toggle; return 0; } cond_resched(); } /* * we could reach here if for example the previous VM using this * function misbehaved and left the channel with unsynced state. We * should fix this here and give this VM a chance to use a properly * synced channel */ mlx4_warn(dev, "recovering from previously mis-behaved VM\n"); __raw_writel((__force u32) 0, &priv->mfunc.comm->slave_read); __raw_writel((__force u32) 0, &priv->mfunc.comm->slave_write); priv->cmd.comm_toggle = 0; return 0; } int mlx4_multi_func_init(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_slave_state *s_state; int i, j, err, port; if (mlx4_is_master(dev)) priv->mfunc.comm = ioremap(pci_resource_start(dev->pdev, priv->fw.comm_bar) + priv->fw.comm_base, MLX4_COMM_PAGESIZE); else priv->mfunc.comm = ioremap(pci_resource_start(dev->pdev, 2) + MLX4_SLAVE_COMM_BASE, MLX4_COMM_PAGESIZE); if (!priv->mfunc.comm) { mlx4_err(dev, "Couldn't map communication vector.\n"); goto err_vhcr; } if (mlx4_is_master(dev)) { priv->mfunc.master.slave_state = kzalloc(dev->num_slaves * sizeof(struct mlx4_slave_state), GFP_KERNEL); if (!priv->mfunc.master.slave_state) goto err_comm; priv->mfunc.master.vf_admin = kzalloc(dev->num_slaves * sizeof(struct mlx4_vf_admin_state), GFP_KERNEL); if (!priv->mfunc.master.vf_admin) goto err_comm_admin; priv->mfunc.master.vf_oper = kzalloc(dev->num_slaves * sizeof(struct mlx4_vf_oper_state), GFP_KERNEL); if (!priv->mfunc.master.vf_oper) goto err_comm_oper; for (i = 0; i < dev->num_slaves; ++i) { s_state = &priv->mfunc.master.slave_state[i]; s_state->last_cmd = MLX4_COMM_CMD_RESET; mutex_init(&priv->mfunc.master.gen_eqe_mutex[i]); for (j = 0; j < MLX4_EVENT_TYPES_NUM; ++j) s_state->event_eq[j].eqn = -1; __raw_writel((__force u32) 0, &priv->mfunc.comm[i].slave_write); __raw_writel((__force u32) 0, &priv->mfunc.comm[i].slave_read); mmiowb(); for (port = 1; port <= MLX4_MAX_PORTS; port++) { s_state->vlan_filter[port] = kzalloc(sizeof(struct mlx4_vlan_fltr), GFP_KERNEL); if (!s_state->vlan_filter[port]) { if (--port) kfree(s_state->vlan_filter[port]); goto err_slaves; } INIT_LIST_HEAD(&s_state->mcast_filters[port]); priv->mfunc.master.vf_admin[i].vport[port].default_vlan = MLX4_VGT; priv->mfunc.master.vf_oper[i].vport[port].state.default_vlan = MLX4_VGT; priv->mfunc.master.vf_oper[i].vport[port].vlan_idx = NO_INDX; priv->mfunc.master.vf_oper[i].vport[port].mac_idx = NO_INDX; } spin_lock_init(&s_state->lock); } memset(&priv->mfunc.master.cmd_eqe, 0, dev->caps.eqe_size); priv->mfunc.master.cmd_eqe.type = MLX4_EVENT_TYPE_CMD; INIT_WORK(&priv->mfunc.master.comm_work, mlx4_master_comm_channel); INIT_WORK(&priv->mfunc.master.arm_comm_work, mlx4_master_arm_comm_channel); INIT_WORK(&priv->mfunc.master.slave_event_work, mlx4_gen_slave_eqe); INIT_WORK(&priv->mfunc.master.slave_flr_event_work, mlx4_master_handle_slave_flr); spin_lock_init(&priv->mfunc.master.slave_state_lock); spin_lock_init(&priv->mfunc.master.slave_eq.event_lock); priv->mfunc.master.comm_wq = create_singlethread_workqueue("mlx4_comm"); if (!priv->mfunc.master.comm_wq) goto err_slaves; if (mlx4_init_resource_tracker(dev)) goto err_thread; err = mlx4_ARM_COMM_CHANNEL(dev); if (err) { mlx4_err(dev, " Failed to arm comm channel eq: %x\n", err); goto err_resource; } } else { err = sync_toggles(dev); if (err) { mlx4_err(dev, "Couldn't sync toggles\n"); goto err_comm; } } return 0; err_resource: mlx4_free_resource_tracker(dev, RES_TR_FREE_ALL); err_thread: flush_workqueue(priv->mfunc.master.comm_wq); destroy_workqueue(priv->mfunc.master.comm_wq); err_slaves: while (--i) { for (port = 1; port <= MLX4_MAX_PORTS; port++) kfree(priv->mfunc.master.slave_state[i].vlan_filter[port]); } kfree(priv->mfunc.master.vf_oper); err_comm_oper: kfree(priv->mfunc.master.vf_admin); err_comm_admin: kfree(priv->mfunc.master.slave_state); err_comm: iounmap(priv->mfunc.comm); err_vhcr: dma_free_coherent(&(dev->pdev->dev), PAGE_SIZE, priv->mfunc.vhcr, priv->mfunc.vhcr_dma); priv->mfunc.vhcr = NULL; return -ENOMEM; } int mlx4_cmd_init(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); mutex_init(&priv->cmd.hcr_mutex); mutex_init(&priv->cmd.slave_cmd_mutex); sema_init(&priv->cmd.poll_sem, 1); priv->cmd.use_events = 0; priv->cmd.toggle = 1; priv->cmd.hcr = NULL; priv->mfunc.vhcr = NULL; if (!mlx4_is_slave(dev)) { priv->cmd.hcr = ioremap(pci_resource_start(dev->pdev, 0) + MLX4_HCR_BASE, MLX4_HCR_SIZE); if (!priv->cmd.hcr) { mlx4_err(dev, "Couldn't map command register.\n"); return -ENOMEM; } } if (mlx4_is_mfunc(dev)) { priv->mfunc.vhcr = dma_alloc_coherent(&(dev->pdev->dev), PAGE_SIZE, &priv->mfunc.vhcr_dma, GFP_KERNEL); if (!priv->mfunc.vhcr) { mlx4_err(dev, "Couldn't allocate VHCR.\n"); goto err_hcr; } } priv->cmd.pool = pci_pool_create("mlx4_cmd", dev->pdev, MLX4_MAILBOX_SIZE, MLX4_MAILBOX_SIZE, 0); if (!priv->cmd.pool) goto err_vhcr; return 0; err_vhcr: if (mlx4_is_mfunc(dev)) dma_free_coherent(&(dev->pdev->dev), PAGE_SIZE, priv->mfunc.vhcr, priv->mfunc.vhcr_dma); priv->mfunc.vhcr = NULL; err_hcr: if (!mlx4_is_slave(dev)) iounmap(priv->cmd.hcr); return -ENOMEM; } void mlx4_multi_func_cleanup(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); int i, port; if (mlx4_is_master(dev)) { flush_workqueue(priv->mfunc.master.comm_wq); destroy_workqueue(priv->mfunc.master.comm_wq); for (i = 0; i < dev->num_slaves; i++) { for (port = 1; port <= MLX4_MAX_PORTS; port++) kfree(priv->mfunc.master.slave_state[i].vlan_filter[port]); } kfree(priv->mfunc.master.slave_state); kfree(priv->mfunc.master.vf_admin); kfree(priv->mfunc.master.vf_oper); } iounmap(priv->mfunc.comm); } void mlx4_cmd_cleanup(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); pci_pool_destroy(priv->cmd.pool); if (!mlx4_is_slave(dev)) iounmap(priv->cmd.hcr); if (mlx4_is_mfunc(dev)) dma_free_coherent(&(dev->pdev->dev), PAGE_SIZE, priv->mfunc.vhcr, priv->mfunc.vhcr_dma); priv->mfunc.vhcr = NULL; } /* * Switch to using events to issue FW commands (can only be called * after event queue for command events has been initialized). */ int mlx4_cmd_use_events(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); int i; int err = 0; priv->cmd.context = kmalloc(priv->cmd.max_cmds * sizeof (struct mlx4_cmd_context), GFP_KERNEL); if (!priv->cmd.context) return -ENOMEM; for (i = 0; i < priv->cmd.max_cmds; ++i) { priv->cmd.context[i].token = i; priv->cmd.context[i].next = i + 1; } priv->cmd.context[priv->cmd.max_cmds - 1].next = -1; priv->cmd.free_head = 0; sema_init(&priv->cmd.event_sem, priv->cmd.max_cmds); spin_lock_init(&priv->cmd.context_lock); for (priv->cmd.token_mask = 1; priv->cmd.token_mask < priv->cmd.max_cmds; priv->cmd.token_mask <<= 1) ; /* nothing */ --priv->cmd.token_mask; down(&priv->cmd.poll_sem); priv->cmd.use_events = 1; return err; } /* * Switch back to polling (used when shutting down the device) */ void mlx4_cmd_use_polling(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); int i; priv->cmd.use_events = 0; for (i = 0; i < priv->cmd.max_cmds; ++i) down(&priv->cmd.event_sem); kfree(priv->cmd.context); up(&priv->cmd.poll_sem); } struct mlx4_cmd_mailbox *mlx4_alloc_cmd_mailbox(struct mlx4_dev *dev) { struct mlx4_cmd_mailbox *mailbox; mailbox = kmalloc(sizeof *mailbox, GFP_KERNEL); if (!mailbox) return ERR_PTR(-ENOMEM); mailbox->buf = pci_pool_alloc(mlx4_priv(dev)->cmd.pool, GFP_KERNEL, &mailbox->dma); if (!mailbox->buf) { kfree(mailbox); return ERR_PTR(-ENOMEM); } memset(mailbox->buf, 0, MLX4_MAILBOX_SIZE); return mailbox; } EXPORT_SYMBOL_GPL(mlx4_alloc_cmd_mailbox); void mlx4_free_cmd_mailbox(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox) { if (!mailbox) return; pci_pool_free(mlx4_priv(dev)->cmd.pool, mailbox->buf, mailbox->dma); kfree(mailbox); } EXPORT_SYMBOL_GPL(mlx4_free_cmd_mailbox); u32 mlx4_comm_get_version(void) { return ((u32) CMD_CHAN_IF_REV << 8) | (u32) CMD_CHAN_VER; } static int mlx4_get_slave_indx(struct mlx4_dev *dev, int vf) { if ((vf < 0) || (vf >= dev->num_vfs)) { mlx4_err(dev, "Bad vf number:%d (number of activated vf: %d)\n", vf, dev->num_vfs); return -EINVAL; } return (vf+1); } int mlx4_set_vf_mac(struct mlx4_dev *dev, int port, int vf, u8 *mac) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_vport_state *s_info; int slave; if (!mlx4_is_master(dev)) return -EPROTONOSUPPORT; slave = mlx4_get_slave_indx(dev, vf); if (slave < 0) return -EINVAL; s_info = &priv->mfunc.master.vf_admin[slave].vport[port]; s_info->mac = mlx4_mac_to_u64(mac); mlx4_info(dev, "default mac on vf %d port %d to %llX will take afect only after vf restart\n", vf, port, (unsigned long long) s_info->mac); return 0; } EXPORT_SYMBOL_GPL(mlx4_set_vf_mac); int mlx4_set_vf_vlan(struct mlx4_dev *dev, int port, int vf, u16 vlan, u8 qos) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_vport_oper_state *vf_oper; struct mlx4_vport_state *vf_admin; int slave; if ((!mlx4_is_master(dev)) || !(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_VLAN_CONTROL)) return -EPROTONOSUPPORT; if ((vlan > 4095) || (qos > 7)) return -EINVAL; slave = mlx4_get_slave_indx(dev, vf); if (slave < 0) return -EINVAL; vf_admin = &priv->mfunc.master.vf_admin[slave].vport[port]; vf_oper = &priv->mfunc.master.vf_oper[slave].vport[port]; if ((0 == vlan) && (0 == qos)) vf_admin->default_vlan = MLX4_VGT; else vf_admin->default_vlan = vlan; vf_admin->default_qos = qos; if (priv->mfunc.master.slave_state[slave].active && dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_UPDATE_QP) { mlx4_info(dev, "updating vf %d port %d config params immediately\n", vf, port); mlx4_master_immediate_activate_vlan_qos(priv, slave, port); } return 0; } EXPORT_SYMBOL_GPL(mlx4_set_vf_vlan); /* mlx4_get_slave_default_vlan - * retrun true if VST ( default vlan) * if VST will fill vlan & qos (if not NULL) */ bool mlx4_get_slave_default_vlan(struct mlx4_dev *dev, int port, int slave, u16 *vlan, u8 *qos) { struct mlx4_vport_oper_state *vp_oper; struct mlx4_priv *priv; priv = mlx4_priv(dev); vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port]; if (MLX4_VGT != vp_oper->state.default_vlan) { if (vlan) *vlan = vp_oper->state.default_vlan; if (qos) *qos = vp_oper->state.default_qos; return true; } return false; } EXPORT_SYMBOL_GPL(mlx4_get_slave_default_vlan); int mlx4_set_vf_spoofchk(struct mlx4_dev *dev, int port, int vf, bool setting) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_vport_state *s_info; int slave; if ((!mlx4_is_master(dev)) || !(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_FSM)) return -EPROTONOSUPPORT; slave = mlx4_get_slave_indx(dev, vf); if (slave < 0) return -EINVAL; s_info = &priv->mfunc.master.vf_admin[slave].vport[port]; s_info->spoofchk = setting; return 0; } EXPORT_SYMBOL_GPL(mlx4_set_vf_spoofchk); int mlx4_set_vf_link_state(struct mlx4_dev *dev, int port, int vf, int link_state) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_vport_state *s_info; struct mlx4_vport_oper_state *vp_oper; int slave; u8 link_stat_event; slave = mlx4_get_slave_indx(dev, vf); if (slave < 0) return -EINVAL; switch (link_state) { case IFLA_VF_LINK_STATE_AUTO: /* get link curent state */ if (!priv->sense.do_sense_port[port]) link_stat_event = MLX4_PORT_CHANGE_SUBTYPE_ACTIVE; else link_stat_event = MLX4_PORT_CHANGE_SUBTYPE_DOWN; break; case IFLA_VF_LINK_STATE_ENABLE: link_stat_event = MLX4_PORT_CHANGE_SUBTYPE_ACTIVE; break; case IFLA_VF_LINK_STATE_DISABLE: link_stat_event = MLX4_PORT_CHANGE_SUBTYPE_DOWN; break; default: mlx4_warn(dev, "unknown value for link_state %02x on slave %d port %d\n", link_state, slave, port); return -EINVAL; }; /* update the admin & oper state on the link state */ s_info = &priv->mfunc.master.vf_admin[slave].vport[port]; vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port]; s_info->link_state = link_state; vp_oper->state.link_state = link_state; /* send event */ mlx4_gen_port_state_change_eqe(dev, slave, port, link_stat_event); return 0; } EXPORT_SYMBOL_GPL(mlx4_set_vf_link_state); int mlx4_get_vf_link_state(struct mlx4_dev *dev, int port, int vf) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_vport_state *s_info; int slave; if (!mlx4_is_master(dev)) return -EPROTONOSUPPORT; slave = mlx4_get_slave_indx(dev, vf); if (slave < 0) return -EINVAL; s_info = &priv->mfunc.master.vf_admin[slave].vport[port]; return s_info->link_state; } EXPORT_SYMBOL_GPL(mlx4_get_vf_link_state); Index: projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/en_main.c =================================================================== --- projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/en_main.c (revision 279716) +++ projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/en_main.c (revision 279717) @@ -1,352 +1,352 @@ /* * Copyright (c) 2007, 2014 Mellanox Technologies. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU * General Public License (GPL) Version 2, available from the file * COPYING in the main directory of this source tree, or the * OpenIB.org BSD license below: * * Redistribution and use in source and binary forms, with or * without modification, are permitted provided that the following * conditions are met: * * - Redistributions of source code must retain the above * copyright notice, this list of conditions and the following * disclaimer. * * - Redistributions in binary form must reproduce the above * copyright notice, this list of conditions and the following * disclaimer in the documentation and/or other materials * provided with the distribution. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. * */ #include #include #include #include #include #include #include #include "mlx4_en.h" MODULE_AUTHOR("Liran Liss, Yevgeny Petrilin"); MODULE_DESCRIPTION("Mellanox ConnectX HCA Ethernet driver"); MODULE_LICENSE("Dual BSD/GPL"); #ifdef __linux__ MODULE_VERSION(DRV_VERSION " ("DRV_RELDATE")"); #endif static const char mlx4_en_version[] = DRV_NAME ": Mellanox ConnectX HCA Ethernet driver v" DRV_VERSION " (" DRV_RELDATE ")\n"; #define MLX4_EN_PARM_INT(X, def_val, desc) \ static unsigned int X = def_val;\ module_param(X , uint, 0444); \ MODULE_PARM_DESC(X, desc); /* * Device scope module parameters */ /* Enable RSS UDP traffic */ MLX4_EN_PARM_INT(udp_rss, 1, "Enable RSS for incoming UDP traffic"); /* Priority pausing */ MLX4_EN_PARM_INT(pfctx, 0, "Priority based Flow Control policy on TX[7:0]." " Per priority bit mask"); MLX4_EN_PARM_INT(pfcrx, 0, "Priority based Flow Control policy on RX[7:0]." " Per priority bit mask"); #define MAX_PFC_TX 0xff #define MAX_PFC_RX 0xff static int mlx4_en_get_profile(struct mlx4_en_dev *mdev) { struct mlx4_en_profile *params = &mdev->profile; int i; params->udp_rss = udp_rss; params->num_tx_rings_p_up = min_t(int, mp_ncpus, MLX4_EN_MAX_TX_RING_P_UP); if (params->udp_rss && !(mdev->dev->caps.flags & MLX4_DEV_CAP_FLAG_UDP_RSS)) { mlx4_warn(mdev, "UDP RSS is not supported on this device.\n"); params->udp_rss = 0; } for (i = 1; i <= MLX4_MAX_PORTS; i++) { params->prof[i].rx_pause = 1; params->prof[i].rx_ppp = pfcrx; params->prof[i].tx_pause = 1; params->prof[i].tx_ppp = pfctx; params->prof[i].tx_ring_size = MLX4_EN_DEF_TX_RING_SIZE; params->prof[i].rx_ring_size = MLX4_EN_DEF_RX_RING_SIZE; params->prof[i].tx_ring_num = params->num_tx_rings_p_up * MLX4_EN_NUM_UP; params->prof[i].rss_rings = 0; } return 0; } static void *mlx4_en_get_netdev(struct mlx4_dev *dev, void *ctx, u8 port) { struct mlx4_en_dev *endev = ctx; return endev->pndev[port]; } static void mlx4_en_event(struct mlx4_dev *dev, void *endev_ptr, enum mlx4_dev_event event, unsigned long port) { struct mlx4_en_dev *mdev = (struct mlx4_en_dev *) endev_ptr; struct mlx4_en_priv *priv; switch (event) { case MLX4_DEV_EVENT_PORT_UP: case MLX4_DEV_EVENT_PORT_DOWN: if (!mdev->pndev[port]) return; priv = netdev_priv(mdev->pndev[port]); /* To prevent races, we poll the link state in a separate task rather than changing it here */ priv->link_state = event; queue_work(mdev->workqueue, &priv->linkstate_task); break; case MLX4_DEV_EVENT_CATASTROPHIC_ERROR: mlx4_err(mdev, "Internal error detected, restarting device\n"); break; case MLX4_DEV_EVENT_SLAVE_INIT: case MLX4_DEV_EVENT_SLAVE_SHUTDOWN: break; default: if (port < 1 || port > dev->caps.num_ports || !mdev->pndev[port]) return; mlx4_warn(mdev, "Unhandled event %d for port %d\n", event, (int) port); } } static void mlx4_en_remove(struct mlx4_dev *dev, void *endev_ptr) { struct mlx4_en_dev *mdev = endev_ptr; int i, ret; mutex_lock(&mdev->state_lock); mdev->device_up = false; mutex_unlock(&mdev->state_lock); mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH) if (mdev->pndev[i]) mlx4_en_destroy_netdev(mdev->pndev[i]); flush_workqueue(mdev->workqueue); destroy_workqueue(mdev->workqueue); ret = mlx4_mr_free(dev, &mdev->mr); if (ret) mlx4_err(mdev, "Error deregistering MR. The system may have become unstable."); iounmap(mdev->uar_map); mlx4_uar_free(dev, &mdev->priv_uar); mlx4_pd_free(dev, mdev->priv_pdn); kfree(mdev); } static void *mlx4_en_add(struct mlx4_dev *dev) { struct mlx4_en_dev *mdev; int i; int err; printk_once(KERN_INFO "%s", mlx4_en_version); mdev = kzalloc(sizeof *mdev, GFP_KERNEL); if (!mdev) { dev_err(&dev->pdev->dev, "Device struct alloc failed, " "aborting.\n"); err = -ENOMEM; goto err_free_res; } if (mlx4_pd_alloc(dev, &mdev->priv_pdn)) goto err_free_dev; if (mlx4_uar_alloc(dev, &mdev->priv_uar)) goto err_pd; mdev->uar_map = ioremap((phys_addr_t) mdev->priv_uar.pfn << PAGE_SHIFT, PAGE_SIZE); if (!mdev->uar_map) goto err_uar; spin_lock_init(&mdev->uar_lock); mdev->dev = dev; mdev->dma_device = &(dev->pdev->dev); mdev->pdev = dev->pdev; mdev->device_up = false; mdev->LSO_support = !!(dev->caps.flags & (1 << 15)); if (!mdev->LSO_support) mlx4_warn(mdev, "LSO not supported, please upgrade to later " "FW version to enable LSO\n"); if (mlx4_mr_alloc(mdev->dev, mdev->priv_pdn, 0, ~0ull, MLX4_PERM_LOCAL_WRITE | MLX4_PERM_LOCAL_READ, 0, 0, &mdev->mr)) { mlx4_err(mdev, "Failed allocating memory region\n"); goto err_map; } if (mlx4_mr_enable(mdev->dev, &mdev->mr)) { mlx4_err(mdev, "Failed enabling memory region\n"); goto err_mr; } /* Build device profile according to supplied module parameters */ err = mlx4_en_get_profile(mdev); if (err) { mlx4_err(mdev, "Bad module parameters, aborting.\n"); goto err_mr; } /* Configure which ports to start according to module parameters */ mdev->port_cnt = 0; mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH) mdev->port_cnt++; mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH) { if (!dev->caps.comp_pool) { mdev->profile.prof[i].rx_ring_num = rounddown_pow_of_two(max_t(int, MIN_RX_RINGS, min_t(int, dev->caps.num_comp_vectors, DEF_RX_RINGS))); } else { mdev->profile.prof[i].rx_ring_num = rounddown_pow_of_two( - min_t(int, dev->caps.comp_pool/ - dev->caps.num_ports - 1 , MAX_MSIX_P_PORT - 1)); + min_t(int, dev->caps.comp_pool / + dev->caps.num_ports, MAX_MSIX_P_PORT)); } } /* Create our own workqueue for reset/multicast tasks * Note: we cannot use the shared workqueue because of deadlocks caused * by the rtnl lock */ mdev->workqueue = create_singlethread_workqueue("mlx4_en"); if (!mdev->workqueue) { err = -ENOMEM; goto err_mr; } /* At this stage all non-port specific tasks are complete: * mark the card state as up */ mutex_init(&mdev->state_lock); mdev->device_up = true; /* Setup ports */ /* Create a netdev for each port */ mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH) { mlx4_info(mdev, "Activating port:%d\n", i); if (mlx4_en_init_netdev(mdev, i, &mdev->profile.prof[i])) mdev->pndev[i] = NULL; } return mdev; err_mr: err = mlx4_mr_free(dev, &mdev->mr); if (err) mlx4_err(mdev, "Error deregistering MR. The system may have become unstable."); err_map: if (mdev->uar_map) iounmap(mdev->uar_map); err_uar: mlx4_uar_free(dev, &mdev->priv_uar); err_pd: mlx4_pd_free(dev, mdev->priv_pdn); err_free_dev: kfree(mdev); err_free_res: return NULL; } static struct mlx4_interface mlx4_en_interface = { .add = mlx4_en_add, .remove = mlx4_en_remove, .event = mlx4_en_event, .get_dev = mlx4_en_get_netdev, .protocol = MLX4_PROT_ETH, }; static void mlx4_en_verify_params(void) { if (pfctx > MAX_PFC_TX) { pr_warn("mlx4_en: WARNING: illegal module parameter pfctx 0x%x - " "should be in range 0-0x%x, will be changed to default (0)\n", pfctx, MAX_PFC_TX); pfctx = 0; } if (pfcrx > MAX_PFC_RX) { pr_warn("mlx4_en: WARNING: illegal module parameter pfcrx 0x%x - " "should be in range 0-0x%x, will be changed to default (0)\n", pfcrx, MAX_PFC_RX); pfcrx = 0; } } static int __init mlx4_en_init(void) { mlx4_en_verify_params(); #ifdef CONFIG_DEBUG_FS int err = 0; err = mlx4_en_register_debugfs(); if (err) pr_err(KERN_ERR "Failed to register debugfs\n"); #endif return mlx4_register_interface(&mlx4_en_interface); } static void __exit mlx4_en_cleanup(void) { mlx4_unregister_interface(&mlx4_en_interface); #ifdef CONFIG_DEBUG_FS mlx4_en_unregister_debugfs(); #endif } module_init(mlx4_en_init); module_exit(mlx4_en_cleanup); static int mlxen_evhand(module_t mod, int event, void *arg) { return (0); } static moduledata_t mlxen_mod = { .name = "mlxen", .evhand = mlxen_evhand, }; DECLARE_MODULE(mlxen, mlxen_mod, SI_SUB_OFED_PREINIT, SI_ORDER_ANY); MODULE_DEPEND(mlxen, mlx4, 1, 1, 1); MODULE_DEPEND(mlxen, linuxapi, 1, 1, 1); Index: projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/en_netdev.c =================================================================== --- projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/en_netdev.c (revision 279716) +++ projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/en_netdev.c (revision 279717) @@ -1,2594 +1,2594 @@ /* * Copyright (c) 2007, 2014 Mellanox Technologies. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU * General Public License (GPL) Version 2, available from the file * COPYING in the main directory of this source tree, or the * OpenIB.org BSD license below: * * Redistribution and use in source and binary forms, with or * without modification, are permitted provided that the following * conditions are met: * * - Redistributions of source code must retain the above * copyright notice, this list of conditions and the following * disclaimer. * * - Redistributions in binary form must reproduce the above * copyright notice, this list of conditions and the following * disclaimer in the documentation and/or other materials * provided with the distribution. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. * */ #include #include #include #ifdef CONFIG_NET_RX_BUSY_POLL #include #endif #include #include #include #include #include #include #include #include #include "mlx4_en.h" #include "en_port.h" static void mlx4_en_sysctl_stat(struct mlx4_en_priv *priv); static void mlx4_en_sysctl_conf(struct mlx4_en_priv *priv); static int mlx4_en_unit; #ifdef CONFIG_NET_RX_BUSY_POLL /* must be called with local_bh_disable()d */ static int mlx4_en_low_latency_recv(struct napi_struct *napi) { struct mlx4_en_cq *cq = container_of(napi, struct mlx4_en_cq, napi); struct net_device *dev = cq->dev; struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_rx_ring *rx_ring = priv->rx_ring[cq->ring]; int done; if (!priv->port_up) return LL_FLUSH_FAILED; if (!mlx4_en_cq_lock_poll(cq)) return LL_FLUSH_BUSY; done = mlx4_en_process_rx_cq(dev, cq, 4); #ifdef LL_EXTENDED_STATS if (done) rx_ring->cleaned += done; else rx_ring->misses++; #endif mlx4_en_cq_unlock_poll(cq); return done; } #endif /* CONFIG_NET_RX_BUSY_POLL */ #ifdef CONFIG_RFS_ACCEL struct mlx4_en_filter { struct list_head next; struct work_struct work; u8 ip_proto; __be32 src_ip; __be32 dst_ip; __be16 src_port; __be16 dst_port; int rxq_index; struct mlx4_en_priv *priv; u32 flow_id; /* RFS infrastructure id */ int id; /* mlx4_en driver id */ u64 reg_id; /* Flow steering API id */ u8 activated; /* Used to prevent expiry before filter * is attached */ struct hlist_node filter_chain; }; static void mlx4_en_filter_rfs_expire(struct mlx4_en_priv *priv); static enum mlx4_net_trans_rule_id mlx4_ip_proto_to_trans_rule_id(u8 ip_proto) { switch (ip_proto) { case IPPROTO_UDP: return MLX4_NET_TRANS_RULE_ID_UDP; case IPPROTO_TCP: return MLX4_NET_TRANS_RULE_ID_TCP; default: return -EPROTONOSUPPORT; } }; static void mlx4_en_filter_work(struct work_struct *work) { struct mlx4_en_filter *filter = container_of(work, struct mlx4_en_filter, work); struct mlx4_en_priv *priv = filter->priv; struct mlx4_spec_list spec_tcp_udp = { .id = mlx4_ip_proto_to_trans_rule_id(filter->ip_proto), { .tcp_udp = { .dst_port = filter->dst_port, .dst_port_msk = (__force __be16)-1, .src_port = filter->src_port, .src_port_msk = (__force __be16)-1, }, }, }; struct mlx4_spec_list spec_ip = { .id = MLX4_NET_TRANS_RULE_ID_IPV4, { .ipv4 = { .dst_ip = filter->dst_ip, .dst_ip_msk = (__force __be32)-1, .src_ip = filter->src_ip, .src_ip_msk = (__force __be32)-1, }, }, }; struct mlx4_spec_list spec_eth = { .id = MLX4_NET_TRANS_RULE_ID_ETH, }; struct mlx4_net_trans_rule rule = { .list = LIST_HEAD_INIT(rule.list), .queue_mode = MLX4_NET_TRANS_Q_LIFO, .exclusive = 1, .allow_loopback = 1, .promisc_mode = MLX4_FS_REGULAR, .port = priv->port, .priority = MLX4_DOMAIN_RFS, }; int rc; __be64 mac_mask = cpu_to_be64(MLX4_MAC_MASK << 16); if (spec_tcp_udp.id < 0) { en_warn(priv, "RFS: ignoring unsupported ip protocol (%d)\n", filter->ip_proto); goto ignore; } list_add_tail(&spec_eth.list, &rule.list); list_add_tail(&spec_ip.list, &rule.list); list_add_tail(&spec_tcp_udp.list, &rule.list); rule.qpn = priv->rss_map.qps[filter->rxq_index].qpn; memcpy(spec_eth.eth.dst_mac, priv->dev->dev_addr, ETH_ALEN); memcpy(spec_eth.eth.dst_mac_msk, &mac_mask, ETH_ALEN); filter->activated = 0; if (filter->reg_id) { rc = mlx4_flow_detach(priv->mdev->dev, filter->reg_id); if (rc && rc != -ENOENT) en_err(priv, "Error detaching flow. rc = %d\n", rc); } rc = mlx4_flow_attach(priv->mdev->dev, &rule, &filter->reg_id); if (rc) en_err(priv, "Error attaching flow. err = %d\n", rc); ignore: mlx4_en_filter_rfs_expire(priv); filter->activated = 1; } static inline struct hlist_head * filter_hash_bucket(struct mlx4_en_priv *priv, __be32 src_ip, __be32 dst_ip, __be16 src_port, __be16 dst_port) { unsigned long l; int bucket_idx; l = (__force unsigned long)src_port | ((__force unsigned long)dst_port << 2); l ^= (__force unsigned long)(src_ip ^ dst_ip); bucket_idx = hash_long(l, MLX4_EN_FILTER_HASH_SHIFT); return &priv->filter_hash[bucket_idx]; } static struct mlx4_en_filter * mlx4_en_filter_alloc(struct mlx4_en_priv *priv, int rxq_index, __be32 src_ip, __be32 dst_ip, u8 ip_proto, __be16 src_port, __be16 dst_port, u32 flow_id) { struct mlx4_en_filter *filter = NULL; filter = kzalloc(sizeof(struct mlx4_en_filter), GFP_ATOMIC); if (!filter) return NULL; filter->priv = priv; filter->rxq_index = rxq_index; INIT_WORK(&filter->work, mlx4_en_filter_work); filter->src_ip = src_ip; filter->dst_ip = dst_ip; filter->ip_proto = ip_proto; filter->src_port = src_port; filter->dst_port = dst_port; filter->flow_id = flow_id; filter->id = priv->last_filter_id++ % RPS_NO_FILTER; list_add_tail(&filter->next, &priv->filters); hlist_add_head(&filter->filter_chain, filter_hash_bucket(priv, src_ip, dst_ip, src_port, dst_port)); return filter; } static void mlx4_en_filter_free(struct mlx4_en_filter *filter) { struct mlx4_en_priv *priv = filter->priv; int rc; list_del(&filter->next); rc = mlx4_flow_detach(priv->mdev->dev, filter->reg_id); if (rc && rc != -ENOENT) en_err(priv, "Error detaching flow. rc = %d\n", rc); kfree(filter); } static inline struct mlx4_en_filter * mlx4_en_filter_find(struct mlx4_en_priv *priv, __be32 src_ip, __be32 dst_ip, u8 ip_proto, __be16 src_port, __be16 dst_port) { struct hlist_node *elem; struct mlx4_en_filter *filter; struct mlx4_en_filter *ret = NULL; hlist_for_each_entry(filter, elem, filter_hash_bucket(priv, src_ip, dst_ip, src_port, dst_port), filter_chain) { if (filter->src_ip == src_ip && filter->dst_ip == dst_ip && filter->ip_proto == ip_proto && filter->src_port == src_port && filter->dst_port == dst_port) { ret = filter; break; } } return ret; } static int mlx4_en_filter_rfs(struct net_device *net_dev, const struct sk_buff *skb, u16 rxq_index, u32 flow_id) { struct mlx4_en_priv *priv = netdev_priv(net_dev); struct mlx4_en_filter *filter; const struct iphdr *ip; const __be16 *ports; u8 ip_proto; __be32 src_ip; __be32 dst_ip; __be16 src_port; __be16 dst_port; int nhoff = skb_network_offset(skb); int ret = 0; if (skb->protocol != htons(ETH_P_IP)) return -EPROTONOSUPPORT; ip = (const struct iphdr *)(skb->data + nhoff); if (ip_is_fragment(ip)) return -EPROTONOSUPPORT; if ((ip->protocol != IPPROTO_TCP) && (ip->protocol != IPPROTO_UDP)) return -EPROTONOSUPPORT; ports = (const __be16 *)(skb->data + nhoff + 4 * ip->ihl); ip_proto = ip->protocol; src_ip = ip->saddr; dst_ip = ip->daddr; src_port = ports[0]; dst_port = ports[1]; spin_lock_bh(&priv->filters_lock); filter = mlx4_en_filter_find(priv, src_ip, dst_ip, ip_proto, src_port, dst_port); if (filter) { if (filter->rxq_index == rxq_index) goto out; filter->rxq_index = rxq_index; } else { filter = mlx4_en_filter_alloc(priv, rxq_index, src_ip, dst_ip, ip_proto, src_port, dst_port, flow_id); if (!filter) { ret = -ENOMEM; goto err; } } queue_work(priv->mdev->workqueue, &filter->work); out: ret = filter->id; err: spin_unlock_bh(&priv->filters_lock); return ret; } void mlx4_en_cleanup_filters(struct mlx4_en_priv *priv, struct mlx4_en_rx_ring *rx_ring) { struct mlx4_en_filter *filter, *tmp; LIST_HEAD(del_list); spin_lock_bh(&priv->filters_lock); list_for_each_entry_safe(filter, tmp, &priv->filters, next) { list_move(&filter->next, &del_list); hlist_del(&filter->filter_chain); } spin_unlock_bh(&priv->filters_lock); list_for_each_entry_safe(filter, tmp, &del_list, next) { cancel_work_sync(&filter->work); mlx4_en_filter_free(filter); } } static void mlx4_en_filter_rfs_expire(struct mlx4_en_priv *priv) { struct mlx4_en_filter *filter = NULL, *tmp, *last_filter = NULL; LIST_HEAD(del_list); int i = 0; spin_lock_bh(&priv->filters_lock); list_for_each_entry_safe(filter, tmp, &priv->filters, next) { if (i > MLX4_EN_FILTER_EXPIRY_QUOTA) break; if (filter->activated && !work_pending(&filter->work) && rps_may_expire_flow(priv->dev, filter->rxq_index, filter->flow_id, filter->id)) { list_move(&filter->next, &del_list); hlist_del(&filter->filter_chain); } else last_filter = filter; i++; } if (last_filter && (&last_filter->next != priv->filters.next)) list_move(&priv->filters, &last_filter->next); spin_unlock_bh(&priv->filters_lock); list_for_each_entry_safe(filter, tmp, &del_list, next) mlx4_en_filter_free(filter); } #endif static void mlx4_en_vlan_rx_add_vid(void *arg, struct net_device *dev, u16 vid) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_dev *mdev = priv->mdev; int err; int idx; if (arg != priv) return; en_dbg(HW, priv, "adding VLAN:%d\n", vid); set_bit(vid, priv->active_vlans); /* Add VID to port VLAN filter */ mutex_lock(&mdev->state_lock); if (mdev->device_up && priv->port_up) { err = mlx4_SET_VLAN_FLTR(mdev->dev, priv); if (err) en_err(priv, "Failed configuring VLAN filter\n"); } if (mlx4_register_vlan(mdev->dev, priv->port, vid, &idx)) en_dbg(HW, priv, "failed adding vlan %d\n", vid); mutex_unlock(&mdev->state_lock); } static void mlx4_en_vlan_rx_kill_vid(void *arg, struct net_device *dev, u16 vid) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_dev *mdev = priv->mdev; int err; if (arg != priv) return; en_dbg(HW, priv, "Killing VID:%d\n", vid); clear_bit(vid, priv->active_vlans); /* Remove VID from port VLAN filter */ mutex_lock(&mdev->state_lock); mlx4_unregister_vlan(mdev->dev, priv->port, vid); if (mdev->device_up && priv->port_up) { err = mlx4_SET_VLAN_FLTR(mdev->dev, priv); if (err) en_err(priv, "Failed configuring VLAN filter\n"); } mutex_unlock(&mdev->state_lock); } static int mlx4_en_uc_steer_add(struct mlx4_en_priv *priv, unsigned char *mac, int *qpn, u64 *reg_id) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_dev *dev = mdev->dev; int err; switch (dev->caps.steering_mode) { case MLX4_STEERING_MODE_B0: { struct mlx4_qp qp; u8 gid[16] = {0}; qp.qpn = *qpn; memcpy(&gid[10], mac, ETH_ALEN); gid[5] = priv->port; err = mlx4_unicast_attach(dev, &qp, gid, 0, MLX4_PROT_ETH); break; } case MLX4_STEERING_MODE_DEVICE_MANAGED: { struct mlx4_spec_list spec_eth = { {NULL} }; __be64 mac_mask = cpu_to_be64(MLX4_MAC_MASK << 16); struct mlx4_net_trans_rule rule = { .queue_mode = MLX4_NET_TRANS_Q_FIFO, .exclusive = 0, .allow_loopback = 1, .promisc_mode = MLX4_FS_REGULAR, .priority = MLX4_DOMAIN_NIC, }; rule.port = priv->port; rule.qpn = *qpn; INIT_LIST_HEAD(&rule.list); spec_eth.id = MLX4_NET_TRANS_RULE_ID_ETH; memcpy(spec_eth.eth.dst_mac, mac, ETH_ALEN); memcpy(spec_eth.eth.dst_mac_msk, &mac_mask, ETH_ALEN); list_add_tail(&spec_eth.list, &rule.list); err = mlx4_flow_attach(dev, &rule, reg_id); break; } default: return -EINVAL; } if (err) en_warn(priv, "Failed Attaching Unicast\n"); return err; } static void mlx4_en_uc_steer_release(struct mlx4_en_priv *priv, unsigned char *mac, int qpn, u64 reg_id) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_dev *dev = mdev->dev; switch (dev->caps.steering_mode) { case MLX4_STEERING_MODE_B0: { struct mlx4_qp qp; u8 gid[16] = {0}; qp.qpn = qpn; memcpy(&gid[10], mac, ETH_ALEN); gid[5] = priv->port; mlx4_unicast_detach(dev, &qp, gid, MLX4_PROT_ETH); break; } case MLX4_STEERING_MODE_DEVICE_MANAGED: { mlx4_flow_detach(dev, reg_id); break; } default: en_err(priv, "Invalid steering mode.\n"); } } static int mlx4_en_get_qp(struct mlx4_en_priv *priv) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_dev *dev = mdev->dev; struct mlx4_mac_entry *entry; int index = 0; int err = 0; u64 reg_id; int *qpn = &priv->base_qpn; u64 mac = mlx4_mac_to_u64(IF_LLADDR(priv->dev)); en_dbg(DRV, priv, "Registering MAC: %pM for adding\n", IF_LLADDR(priv->dev)); index = mlx4_register_mac(dev, priv->port, mac); if (index < 0) { err = index; en_err(priv, "Failed adding MAC: %pM\n", IF_LLADDR(priv->dev)); return err; } if (dev->caps.steering_mode == MLX4_STEERING_MODE_A0) { int base_qpn = mlx4_get_base_qpn(dev, priv->port); *qpn = base_qpn + index; return 0; } err = mlx4_qp_reserve_range(dev, 1, 1, qpn, 0); en_dbg(DRV, priv, "Reserved qp %d\n", *qpn); if (err) { en_err(priv, "Failed to reserve qp for mac registration\n"); goto qp_err; } err = mlx4_en_uc_steer_add(priv, IF_LLADDR(priv->dev), qpn, ®_id); if (err) goto steer_err; entry = kmalloc(sizeof(*entry), GFP_KERNEL); if (!entry) { err = -ENOMEM; goto alloc_err; } memcpy(entry->mac, IF_LLADDR(priv->dev), sizeof(entry->mac)); entry->reg_id = reg_id; hlist_add_head(&entry->hlist, &priv->mac_hash[entry->mac[MLX4_EN_MAC_HASH_IDX]]); return 0; alloc_err: mlx4_en_uc_steer_release(priv, IF_LLADDR(priv->dev), *qpn, reg_id); steer_err: mlx4_qp_release_range(dev, *qpn, 1); qp_err: mlx4_unregister_mac(dev, priv->port, mac); return err; } static void mlx4_en_put_qp(struct mlx4_en_priv *priv) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_dev *dev = mdev->dev; int qpn = priv->base_qpn; u64 mac; if (dev->caps.steering_mode == MLX4_STEERING_MODE_A0) { mac = mlx4_mac_to_u64(IF_LLADDR(priv->dev)); en_dbg(DRV, priv, "Registering MAC: %pM for deleting\n", IF_LLADDR(priv->dev)); mlx4_unregister_mac(dev, priv->port, mac); } else { struct mlx4_mac_entry *entry; struct hlist_node *n, *tmp; struct hlist_head *bucket; unsigned int i; for (i = 0; i < MLX4_EN_MAC_HASH_SIZE; ++i) { bucket = &priv->mac_hash[i]; hlist_for_each_entry_safe(entry, n, tmp, bucket, hlist) { mac = mlx4_mac_to_u64(entry->mac); en_dbg(DRV, priv, "Registering MAC: %pM for deleting\n", entry->mac); mlx4_en_uc_steer_release(priv, entry->mac, qpn, entry->reg_id); mlx4_unregister_mac(dev, priv->port, mac); hlist_del(&entry->hlist); kfree(entry); } } en_dbg(DRV, priv, "Releasing qp: port %d, qpn %d\n", priv->port, qpn); mlx4_qp_release_range(dev, qpn, 1); priv->flags &= ~MLX4_EN_FLAG_FORCE_PROMISC; } } static void mlx4_en_clear_list(struct net_device *dev) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_mc_list *tmp, *mc_to_del; list_for_each_entry_safe(mc_to_del, tmp, &priv->mc_list, list) { list_del(&mc_to_del->list); kfree(mc_to_del); } } static void mlx4_en_cache_mclist(struct net_device *dev) { struct ifmultiaddr *ifma; struct mlx4_en_mc_list *tmp; struct mlx4_en_priv *priv = netdev_priv(dev); TAILQ_FOREACH(ifma, &dev->if_multiaddrs, ifma_link) { if (ifma->ifma_addr->sa_family != AF_LINK) continue; if (((struct sockaddr_dl *)ifma->ifma_addr)->sdl_alen != ETHER_ADDR_LEN) continue; /* Make sure the list didn't grow. */ tmp = kzalloc(sizeof(struct mlx4_en_mc_list), GFP_ATOMIC); memcpy(tmp->addr, LLADDR((struct sockaddr_dl *)ifma->ifma_addr), ETH_ALEN); list_add_tail(&tmp->list, &priv->mc_list); } } static void update_mclist_flags(struct mlx4_en_priv *priv, struct list_head *dst, struct list_head *src) { struct mlx4_en_mc_list *dst_tmp, *src_tmp, *new_mc; bool found; /* Find all the entries that should be removed from dst, * These are the entries that are not found in src */ list_for_each_entry(dst_tmp, dst, list) { found = false; list_for_each_entry(src_tmp, src, list) { if (!memcmp(dst_tmp->addr, src_tmp->addr, ETH_ALEN)) { found = true; break; } } if (!found) dst_tmp->action = MCLIST_REM; } /* Add entries that exist in src but not in dst * mark them as need to add */ list_for_each_entry(src_tmp, src, list) { found = false; list_for_each_entry(dst_tmp, dst, list) { if (!memcmp(dst_tmp->addr, src_tmp->addr, ETH_ALEN)) { dst_tmp->action = MCLIST_NONE; found = true; break; } } if (!found) { new_mc = kmalloc(sizeof(struct mlx4_en_mc_list), GFP_KERNEL); if (!new_mc) { en_err(priv, "Failed to allocate current multicast list\n"); return; } memcpy(new_mc, src_tmp, sizeof(struct mlx4_en_mc_list)); new_mc->action = MCLIST_ADD; list_add_tail(&new_mc->list, dst); } } } static void mlx4_en_set_rx_mode(struct net_device *dev) { struct mlx4_en_priv *priv = netdev_priv(dev); if (!priv->port_up) return; queue_work(priv->mdev->workqueue, &priv->rx_mode_task); } static void mlx4_en_set_promisc_mode(struct mlx4_en_priv *priv, struct mlx4_en_dev *mdev) { int err = 0; if (!(priv->flags & MLX4_EN_FLAG_PROMISC)) { priv->flags |= MLX4_EN_FLAG_PROMISC; /* Enable promiscouos mode */ switch (mdev->dev->caps.steering_mode) { case MLX4_STEERING_MODE_DEVICE_MANAGED: err = mlx4_flow_steer_promisc_add(mdev->dev, priv->port, priv->base_qpn, MLX4_FS_ALL_DEFAULT); if (err) en_err(priv, "Failed enabling promiscuous mode\n"); priv->flags |= MLX4_EN_FLAG_MC_PROMISC; break; case MLX4_STEERING_MODE_B0: err = mlx4_unicast_promisc_add(mdev->dev, priv->base_qpn, priv->port); if (err) en_err(priv, "Failed enabling unicast promiscuous mode\n"); /* Add the default qp number as multicast * promisc */ if (!(priv->flags & MLX4_EN_FLAG_MC_PROMISC)) { err = mlx4_multicast_promisc_add(mdev->dev, priv->base_qpn, priv->port); if (err) en_err(priv, "Failed enabling multicast promiscuous mode\n"); priv->flags |= MLX4_EN_FLAG_MC_PROMISC; } break; case MLX4_STEERING_MODE_A0: err = mlx4_SET_PORT_qpn_calc(mdev->dev, priv->port, priv->base_qpn, 1); if (err) en_err(priv, "Failed enabling promiscuous mode\n"); break; } /* Disable port multicast filter (unconditionally) */ err = mlx4_SET_MCAST_FLTR(mdev->dev, priv->port, 0, 0, MLX4_MCAST_DISABLE); if (err) en_err(priv, "Failed disabling multicast filter\n"); } } static void mlx4_en_clear_promisc_mode(struct mlx4_en_priv *priv, struct mlx4_en_dev *mdev) { int err = 0; priv->flags &= ~MLX4_EN_FLAG_PROMISC; /* Disable promiscouos mode */ switch (mdev->dev->caps.steering_mode) { case MLX4_STEERING_MODE_DEVICE_MANAGED: err = mlx4_flow_steer_promisc_remove(mdev->dev, priv->port, MLX4_FS_ALL_DEFAULT); if (err) en_err(priv, "Failed disabling promiscuous mode\n"); priv->flags &= ~MLX4_EN_FLAG_MC_PROMISC; break; case MLX4_STEERING_MODE_B0: err = mlx4_unicast_promisc_remove(mdev->dev, priv->base_qpn, priv->port); if (err) en_err(priv, "Failed disabling unicast promiscuous mode\n"); /* Disable Multicast promisc */ if (priv->flags & MLX4_EN_FLAG_MC_PROMISC) { err = mlx4_multicast_promisc_remove(mdev->dev, priv->base_qpn, priv->port); if (err) en_err(priv, "Failed disabling multicast promiscuous mode\n"); priv->flags &= ~MLX4_EN_FLAG_MC_PROMISC; } break; case MLX4_STEERING_MODE_A0: err = mlx4_SET_PORT_qpn_calc(mdev->dev, priv->port, priv->base_qpn, 0); if (err) en_err(priv, "Failed disabling promiscuous mode\n"); break; } } static void mlx4_en_do_multicast(struct mlx4_en_priv *priv, struct net_device *dev, struct mlx4_en_dev *mdev) { struct mlx4_en_mc_list *mclist, *tmp; u8 mc_list[16] = {0}; int err = 0; u64 mcast_addr = 0; /* Enable/disable the multicast filter according to IFF_ALLMULTI */ if (dev->if_flags & IFF_ALLMULTI) { err = mlx4_SET_MCAST_FLTR(mdev->dev, priv->port, 0, 0, MLX4_MCAST_DISABLE); if (err) en_err(priv, "Failed disabling multicast filter\n"); /* Add the default qp number as multicast promisc */ if (!(priv->flags & MLX4_EN_FLAG_MC_PROMISC)) { switch (mdev->dev->caps.steering_mode) { case MLX4_STEERING_MODE_DEVICE_MANAGED: err = mlx4_flow_steer_promisc_add(mdev->dev, priv->port, priv->base_qpn, MLX4_FS_MC_DEFAULT); break; case MLX4_STEERING_MODE_B0: err = mlx4_multicast_promisc_add(mdev->dev, priv->base_qpn, priv->port); break; case MLX4_STEERING_MODE_A0: break; } if (err) en_err(priv, "Failed entering multicast promisc mode\n"); priv->flags |= MLX4_EN_FLAG_MC_PROMISC; } } else { /* Disable Multicast promisc */ if (priv->flags & MLX4_EN_FLAG_MC_PROMISC) { switch (mdev->dev->caps.steering_mode) { case MLX4_STEERING_MODE_DEVICE_MANAGED: err = mlx4_flow_steer_promisc_remove(mdev->dev, priv->port, MLX4_FS_MC_DEFAULT); break; case MLX4_STEERING_MODE_B0: err = mlx4_multicast_promisc_remove(mdev->dev, priv->base_qpn, priv->port); break; case MLX4_STEERING_MODE_A0: break; } if (err) en_err(priv, "Failed disabling multicast promiscuous mode\n"); priv->flags &= ~MLX4_EN_FLAG_MC_PROMISC; } err = mlx4_SET_MCAST_FLTR(mdev->dev, priv->port, 0, 0, MLX4_MCAST_DISABLE); if (err) en_err(priv, "Failed disabling multicast filter\n"); /* Flush mcast filter and init it with broadcast address */ mlx4_SET_MCAST_FLTR(mdev->dev, priv->port, ETH_BCAST, 1, MLX4_MCAST_CONFIG); /* Update multicast list - we cache all addresses so they won't * change while HW is updated holding the command semaphor */ mlx4_en_cache_mclist(dev); list_for_each_entry(mclist, &priv->mc_list, list) { mcast_addr = mlx4_mac_to_u64(mclist->addr); mlx4_SET_MCAST_FLTR(mdev->dev, priv->port, mcast_addr, 0, MLX4_MCAST_CONFIG); } err = mlx4_SET_MCAST_FLTR(mdev->dev, priv->port, 0, 0, MLX4_MCAST_ENABLE); if (err) en_err(priv, "Failed enabling multicast filter\n"); update_mclist_flags(priv, &priv->curr_list, &priv->mc_list); list_for_each_entry_safe(mclist, tmp, &priv->curr_list, list) { if (mclist->action == MCLIST_REM) { /* detach this address and delete from list */ memcpy(&mc_list[10], mclist->addr, ETH_ALEN); mc_list[5] = priv->port; err = mlx4_multicast_detach(mdev->dev, &priv->rss_map.indir_qp, mc_list, MLX4_PROT_ETH, mclist->reg_id); if (err) en_err(priv, "Fail to detach multicast address\n"); /* remove from list */ list_del(&mclist->list); kfree(mclist); } else if (mclist->action == MCLIST_ADD) { /* attach the address */ memcpy(&mc_list[10], mclist->addr, ETH_ALEN); /* needed for B0 steering support */ mc_list[5] = priv->port; err = mlx4_multicast_attach(mdev->dev, &priv->rss_map.indir_qp, mc_list, priv->port, 0, MLX4_PROT_ETH, &mclist->reg_id); if (err) en_err(priv, "Fail to attach multicast address\n"); } } } } static void mlx4_en_do_set_rx_mode(struct work_struct *work) { struct mlx4_en_priv *priv = container_of(work, struct mlx4_en_priv, rx_mode_task); struct mlx4_en_dev *mdev = priv->mdev; struct net_device *dev = priv->dev; mutex_lock(&mdev->state_lock); if (!mdev->device_up) { en_dbg(HW, priv, "Card is not up, ignoring rx mode change.\n"); goto out; } if (!priv->port_up) { en_dbg(HW, priv, "Port is down, ignoring rx mode change.\n"); goto out; } if (!mlx4_en_QUERY_PORT(mdev, priv->port)) { if (priv->port_state.link_state) { priv->last_link_state = MLX4_DEV_EVENT_PORT_UP; /* Important note: the following call for if_link_state_change * is needed for interface up scenario (start port, link state * change) */ /* update netif baudrate */ priv->dev->if_baudrate = IF_Mbps(priv->port_state.link_speed); if_link_state_change(priv->dev, LINK_STATE_UP); en_dbg(HW, priv, "Link Up\n"); } } /* Promsicuous mode: disable all filters */ if ((dev->if_flags & IFF_PROMISC) || (priv->flags & MLX4_EN_FLAG_FORCE_PROMISC)) { mlx4_en_set_promisc_mode(priv, mdev); goto out; } /* Not in promiscuous mode */ if (priv->flags & MLX4_EN_FLAG_PROMISC) mlx4_en_clear_promisc_mode(priv, mdev); mlx4_en_do_multicast(priv, dev, mdev); out: mutex_unlock(&mdev->state_lock); } #ifdef CONFIG_NET_POLL_CONTROLLER static void mlx4_en_netpoll(struct net_device *dev) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_cq *cq; unsigned long flags; int i; for (i = 0; i < priv->rx_ring_num; i++) { cq = priv->rx_cq[i]; spin_lock_irqsave(&cq->lock, flags); napi_synchronize(&cq->napi); mlx4_en_process_rx_cq(dev, cq, 0); spin_unlock_irqrestore(&cq->lock, flags); } } #endif static void mlx4_en_watchdog_timeout(void *arg) { struct mlx4_en_priv *priv = arg; struct mlx4_en_dev *mdev = priv->mdev; en_dbg(DRV, priv, "Scheduling watchdog\n"); queue_work(mdev->workqueue, &priv->watchdog_task); if (priv->port_up) callout_reset(&priv->watchdog_timer, MLX4_EN_WATCHDOG_TIMEOUT, mlx4_en_watchdog_timeout, priv); } static void mlx4_en_set_default_moderation(struct mlx4_en_priv *priv) { struct mlx4_en_cq *cq; int i; /* If we haven't received a specific coalescing setting * (module param), we set the moderation parameters as follows: * - moder_cnt is set to the number of mtu sized packets to * satisfy our coelsing target. * - moder_time is set to a fixed value. */ priv->rx_frames = MLX4_EN_RX_COAL_TARGET / priv->dev->if_mtu + 1; priv->rx_usecs = MLX4_EN_RX_COAL_TIME; priv->tx_frames = MLX4_EN_TX_COAL_PKTS; priv->tx_usecs = MLX4_EN_TX_COAL_TIME; en_dbg(INTR, priv, "Default coalesing params for mtu: %u - " "rx_frames:%d rx_usecs:%d\n", (unsigned)priv->dev->if_mtu, priv->rx_frames, priv->rx_usecs); /* Setup cq moderation params */ for (i = 0; i < priv->rx_ring_num; i++) { cq = priv->rx_cq[i]; cq->moder_cnt = priv->rx_frames; cq->moder_time = priv->rx_usecs; priv->last_moder_time[i] = MLX4_EN_AUTO_CONF; priv->last_moder_packets[i] = 0; priv->last_moder_bytes[i] = 0; } for (i = 0; i < priv->tx_ring_num; i++) { cq = priv->tx_cq[i]; cq->moder_cnt = priv->tx_frames; cq->moder_time = priv->tx_usecs; } /* Reset auto-moderation params */ priv->pkt_rate_low = MLX4_EN_RX_RATE_LOW; priv->rx_usecs_low = MLX4_EN_RX_COAL_TIME_LOW; priv->pkt_rate_high = MLX4_EN_RX_RATE_HIGH; priv->rx_usecs_high = MLX4_EN_RX_COAL_TIME_HIGH; priv->sample_interval = MLX4_EN_SAMPLE_INTERVAL; priv->adaptive_rx_coal = 1; priv->last_moder_jiffies = 0; priv->last_moder_tx_packets = 0; } static void mlx4_en_auto_moderation(struct mlx4_en_priv *priv) { unsigned long period = (unsigned long) (jiffies - priv->last_moder_jiffies); struct mlx4_en_cq *cq; unsigned long packets; unsigned long rate; unsigned long avg_pkt_size; unsigned long rx_packets; unsigned long rx_bytes; unsigned long rx_pkt_diff; int moder_time; int ring, err; if (!priv->adaptive_rx_coal || period < priv->sample_interval * HZ) return; for (ring = 0; ring < priv->rx_ring_num; ring++) { spin_lock(&priv->stats_lock); rx_packets = priv->rx_ring[ring]->packets; rx_bytes = priv->rx_ring[ring]->bytes; spin_unlock(&priv->stats_lock); rx_pkt_diff = ((unsigned long) (rx_packets - priv->last_moder_packets[ring])); packets = rx_pkt_diff; rate = packets * HZ / period; avg_pkt_size = packets ? ((unsigned long) (rx_bytes - priv->last_moder_bytes[ring])) / packets : 0; /* Apply auto-moderation only when packet rate * exceeds a rate that it matters */ if (rate > (MLX4_EN_RX_RATE_THRESH / priv->rx_ring_num) && avg_pkt_size > MLX4_EN_AVG_PKT_SMALL) { if (rate < priv->pkt_rate_low) moder_time = priv->rx_usecs_low; else if (rate > priv->pkt_rate_high) moder_time = priv->rx_usecs_high; else moder_time = (rate - priv->pkt_rate_low) * (priv->rx_usecs_high - priv->rx_usecs_low) / (priv->pkt_rate_high - priv->pkt_rate_low) + priv->rx_usecs_low; } else { moder_time = priv->rx_usecs_low; } if (moder_time != priv->last_moder_time[ring]) { priv->last_moder_time[ring] = moder_time; cq = priv->rx_cq[ring]; cq->moder_time = moder_time; err = mlx4_en_set_cq_moder(priv, cq); if (err) en_err(priv, "Failed modifying moderation for cq:%d\n", ring); } priv->last_moder_packets[ring] = rx_packets; priv->last_moder_bytes[ring] = rx_bytes; } priv->last_moder_jiffies = jiffies; } static void mlx4_en_do_get_stats(struct work_struct *work) { struct delayed_work *delay = to_delayed_work(work); struct mlx4_en_priv *priv = container_of(delay, struct mlx4_en_priv, stats_task); struct mlx4_en_dev *mdev = priv->mdev; int err; mutex_lock(&mdev->state_lock); if (mdev->device_up) { if (priv->port_up) { err = mlx4_en_DUMP_ETH_STATS(mdev, priv->port, 0); if (err) en_dbg(HW, priv, "Could not update stats\n"); mlx4_en_auto_moderation(priv); } queue_delayed_work(mdev->workqueue, &priv->stats_task, STATS_DELAY); } mutex_unlock(&mdev->state_lock); } /* mlx4_en_service_task - Run service task for tasks that needed to be done * periodically */ static void mlx4_en_service_task(struct work_struct *work) { struct delayed_work *delay = to_delayed_work(work); struct mlx4_en_priv *priv = container_of(delay, struct mlx4_en_priv, service_task); struct mlx4_en_dev *mdev = priv->mdev; mutex_lock(&mdev->state_lock); if (mdev->device_up) { queue_delayed_work(mdev->workqueue, &priv->service_task, SERVICE_TASK_DELAY); } mutex_unlock(&mdev->state_lock); } static void mlx4_en_linkstate(struct work_struct *work) { struct mlx4_en_priv *priv = container_of(work, struct mlx4_en_priv, linkstate_task); struct mlx4_en_dev *mdev = priv->mdev; int linkstate = priv->link_state; mutex_lock(&mdev->state_lock); /* If observable port state changed set carrier state and * report to system log */ if (priv->last_link_state != linkstate) { if (linkstate == MLX4_DEV_EVENT_PORT_DOWN) { en_info(priv, "Link Down\n"); if_link_state_change(priv->dev, LINK_STATE_DOWN); /* update netif baudrate */ priv->dev->if_baudrate = 0; /* make sure the port is up before notifying the OS. * This is tricky since we get here on INIT_PORT and * in such case we can't tell the OS the port is up. * To solve this there is a call to if_link_state_change * in set_rx_mode. * */ } else if (priv->port_up && (linkstate == MLX4_DEV_EVENT_PORT_UP)){ if (mlx4_en_QUERY_PORT(priv->mdev, priv->port)) en_info(priv, "Query port failed\n"); priv->dev->if_baudrate = IF_Mbps(priv->port_state.link_speed); en_info(priv, "Link Up\n"); if_link_state_change(priv->dev, LINK_STATE_UP); } } priv->last_link_state = linkstate; mutex_unlock(&mdev->state_lock); } int mlx4_en_start_port(struct net_device *dev) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_en_cq *cq; struct mlx4_en_tx_ring *tx_ring; int rx_index = 0; int tx_index = 0; int err = 0; int i; int j; u8 mc_list[16] = {0}; if (priv->port_up) { en_dbg(DRV, priv, "start port called while port already up\n"); return 0; } INIT_LIST_HEAD(&priv->mc_list); INIT_LIST_HEAD(&priv->curr_list); INIT_LIST_HEAD(&priv->ethtool_list); /* Calculate Rx buf size */ dev->if_mtu = min(dev->if_mtu, priv->max_mtu); mlx4_en_calc_rx_buf(dev); priv->rx_alloc_size = max_t(int, 2 * roundup_pow_of_two(priv->rx_mb_size), PAGE_SIZE); priv->rx_alloc_order = get_order(priv->rx_alloc_size); priv->rx_buf_size = roundup_pow_of_two(priv->rx_mb_size); priv->log_rx_info = ROUNDUP_LOG2(sizeof(struct mlx4_en_rx_buf)); en_dbg(DRV, priv, "Rx buf size:%d\n", priv->rx_mb_size); /* Configure rx cq's and rings */ err = mlx4_en_activate_rx_rings(priv); if (err) { en_err(priv, "Failed to activate RX rings\n"); return err; } for (i = 0; i < priv->rx_ring_num; i++) { cq = priv->rx_cq[i]; mlx4_en_cq_init_lock(cq); err = mlx4_en_activate_cq(priv, cq, i); if (err) { en_err(priv, "Failed activating Rx CQ\n"); goto cq_err; } for (j = 0; j < cq->size; j++) cq->buf[j].owner_sr_opcode = MLX4_CQE_OWNER_MASK; err = mlx4_en_set_cq_moder(priv, cq); if (err) { en_err(priv, "Failed setting cq moderation parameters"); mlx4_en_deactivate_cq(priv, cq); goto cq_err; } mlx4_en_arm_cq(priv, cq); priv->rx_ring[i]->cqn = cq->mcq.cqn; ++rx_index; } /* Set qp number */ en_dbg(DRV, priv, "Getting qp number for port %d\n", priv->port); err = mlx4_en_get_qp(priv); if (err) { en_err(priv, "Failed getting eth qp\n"); goto cq_err; } mdev->mac_removed[priv->port] = 0; /* gets default allocated counter index from func cap */ /* or sink counter index if no resources */ priv->counter_index = mdev->dev->caps.def_counter_index[priv->port - 1]; en_dbg(DRV, priv, "%s: default counter index %d for port %d\n", __func__, priv->counter_index, priv->port); err = mlx4_en_config_rss_steer(priv); if (err) { en_err(priv, "Failed configuring rss steering\n"); goto mac_err; } err = mlx4_en_create_drop_qp(priv); if (err) goto rss_err; /* Configure tx cq's and rings */ for (i = 0; i < priv->tx_ring_num; i++) { /* Configure cq */ cq = priv->tx_cq[i]; err = mlx4_en_activate_cq(priv, cq, i); if (err) { - en_err(priv, "Failed allocating Tx CQ\n"); + en_err(priv, "Failed activating Tx CQ\n"); goto tx_err; } err = mlx4_en_set_cq_moder(priv, cq); if (err) { en_err(priv, "Failed setting cq moderation parameters"); mlx4_en_deactivate_cq(priv, cq); goto tx_err; } en_dbg(DRV, priv, "Resetting index of collapsed CQ:%d to -1\n", i); cq->buf->wqe_index = cpu_to_be16(0xffff); /* Configure ring */ tx_ring = priv->tx_ring[i]; err = mlx4_en_activate_tx_ring(priv, tx_ring, cq->mcq.cqn, i / priv->num_tx_rings_p_up); if (err) { - en_err(priv, "Failed allocating Tx ring\n"); + en_err(priv, "Failed activating Tx ring %d\n", i); mlx4_en_deactivate_cq(priv, cq); goto tx_err; } /* Arm CQ for TX completions */ mlx4_en_arm_cq(priv, cq); /* Set initial ownership of all Tx TXBBs to SW (1) */ for (j = 0; j < tx_ring->buf_size; j += STAMP_STRIDE) *((u32 *) (tx_ring->buf + j)) = 0xffffffff; ++tx_index; } /* Configure port */ err = mlx4_SET_PORT_general(mdev->dev, priv->port, priv->rx_mb_size, priv->prof->tx_pause, priv->prof->tx_ppp, priv->prof->rx_pause, priv->prof->rx_ppp); if (err) { en_err(priv, "Failed setting port general configurations for port %d, with error %d\n", priv->port, err); goto tx_err; } /* Set default qp number */ err = mlx4_SET_PORT_qpn_calc(mdev->dev, priv->port, priv->base_qpn, 0); if (err) { en_err(priv, "Failed setting default qp numbers\n"); goto tx_err; } /* Init port */ en_dbg(HW, priv, "Initializing port\n"); err = mlx4_INIT_PORT(mdev->dev, priv->port); if (err) { en_err(priv, "Failed Initializing port\n"); goto tx_err; } /* Attach rx QP to bradcast address */ memset(&mc_list[10], 0xff, ETH_ALEN); mc_list[5] = priv->port; /* needed for B0 steering support */ if (mlx4_multicast_attach(mdev->dev, &priv->rss_map.indir_qp, mc_list, priv->port, 0, MLX4_PROT_ETH, &priv->broadcast_id)) mlx4_warn(mdev, "Failed Attaching Broadcast\n"); /* Must redo promiscuous mode setup. */ priv->flags &= ~(MLX4_EN_FLAG_PROMISC | MLX4_EN_FLAG_MC_PROMISC); /* Schedule multicast task to populate multicast list */ queue_work(mdev->workqueue, &priv->rx_mode_task); mlx4_set_stats_bitmap(mdev->dev, priv->stats_bitmap); priv->port_up = true; /* Enable the queues. */ dev->if_drv_flags &= ~IFF_DRV_OACTIVE; dev->if_drv_flags |= IFF_DRV_RUNNING; #ifdef CONFIG_DEBUG_FS mlx4_en_create_debug_files(priv); #endif callout_reset(&priv->watchdog_timer, MLX4_EN_WATCHDOG_TIMEOUT, mlx4_en_watchdog_timeout, priv); return 0; tx_err: while (tx_index--) { mlx4_en_deactivate_tx_ring(priv, priv->tx_ring[tx_index]); mlx4_en_deactivate_cq(priv, priv->tx_cq[tx_index]); } mlx4_en_destroy_drop_qp(priv); rss_err: mlx4_en_release_rss_steer(priv); mac_err: mlx4_en_put_qp(priv); cq_err: while (rx_index--) mlx4_en_deactivate_cq(priv, priv->rx_cq[rx_index]); for (i = 0; i < priv->rx_ring_num; i++) mlx4_en_deactivate_rx_ring(priv, priv->rx_ring[i]); return err; /* need to close devices */ } void mlx4_en_stop_port(struct net_device *dev) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_en_mc_list *mclist, *tmp; int i; u8 mc_list[16] = {0}; if (!priv->port_up) { en_dbg(DRV, priv, "stop port called while port already down\n"); return; } #ifdef CONFIG_DEBUG_FS mlx4_en_delete_debug_files(priv); #endif /* close port*/ mlx4_CLOSE_PORT(mdev->dev, priv->port); /* Set port as not active */ priv->port_up = false; if (priv->counter_index != 0xff) { mlx4_counter_free(mdev->dev, priv->port, priv->counter_index); priv->counter_index = 0xff; } /* Promsicuous mode */ if (mdev->dev->caps.steering_mode == MLX4_STEERING_MODE_DEVICE_MANAGED) { priv->flags &= ~(MLX4_EN_FLAG_PROMISC | MLX4_EN_FLAG_MC_PROMISC); mlx4_flow_steer_promisc_remove(mdev->dev, priv->port, MLX4_FS_ALL_DEFAULT); mlx4_flow_steer_promisc_remove(mdev->dev, priv->port, MLX4_FS_MC_DEFAULT); } else if (priv->flags & MLX4_EN_FLAG_PROMISC) { priv->flags &= ~MLX4_EN_FLAG_PROMISC; /* Disable promiscouos mode */ mlx4_unicast_promisc_remove(mdev->dev, priv->base_qpn, priv->port); /* Disable Multicast promisc */ if (priv->flags & MLX4_EN_FLAG_MC_PROMISC) { mlx4_multicast_promisc_remove(mdev->dev, priv->base_qpn, priv->port); priv->flags &= ~MLX4_EN_FLAG_MC_PROMISC; } } /* Detach All multicasts */ memset(&mc_list[10], 0xff, ETH_ALEN); mc_list[5] = priv->port; /* needed for B0 steering support */ mlx4_multicast_detach(mdev->dev, &priv->rss_map.indir_qp, mc_list, MLX4_PROT_ETH, priv->broadcast_id); list_for_each_entry(mclist, &priv->curr_list, list) { memcpy(&mc_list[10], mclist->addr, ETH_ALEN); mc_list[5] = priv->port; mlx4_multicast_detach(mdev->dev, &priv->rss_map.indir_qp, mc_list, MLX4_PROT_ETH, mclist->reg_id); } mlx4_en_clear_list(dev); list_for_each_entry_safe(mclist, tmp, &priv->curr_list, list) { list_del(&mclist->list); kfree(mclist); } /* Flush multicast filter */ mlx4_SET_MCAST_FLTR(mdev->dev, priv->port, 0, 1, MLX4_MCAST_CONFIG); mlx4_en_destroy_drop_qp(priv); /* Free TX Rings */ for (i = 0; i < priv->tx_ring_num; i++) { mlx4_en_deactivate_tx_ring(priv, priv->tx_ring[i]); mlx4_en_deactivate_cq(priv, priv->tx_cq[i]); } msleep(10); for (i = 0; i < priv->tx_ring_num; i++) mlx4_en_free_tx_buf(dev, priv->tx_ring[i]); /* Free RSS qps */ mlx4_en_release_rss_steer(priv); /* Unregister Mac address for the port */ mlx4_en_put_qp(priv); mdev->mac_removed[priv->port] = 1; /* Free RX Rings */ for (i = 0; i < priv->rx_ring_num; i++) { struct mlx4_en_cq *cq = priv->rx_cq[i]; mlx4_en_deactivate_rx_ring(priv, priv->rx_ring[i]); mlx4_en_deactivate_cq(priv, cq); } callout_stop(&priv->watchdog_timer); dev->if_drv_flags &= ~(IFF_DRV_RUNNING | IFF_DRV_OACTIVE); } static void mlx4_en_restart(struct work_struct *work) { struct mlx4_en_priv *priv = container_of(work, struct mlx4_en_priv, watchdog_task); struct mlx4_en_dev *mdev = priv->mdev; struct net_device *dev = priv->dev; struct mlx4_en_tx_ring *ring; int i; if (priv->blocked == 0 || priv->port_up == 0) return; for (i = 0; i < priv->tx_ring_num; i++) { ring = priv->tx_ring[i]; if (ring->blocked && ring->watchdog_time + MLX4_EN_WATCHDOG_TIMEOUT < ticks) goto reset; } return; reset: priv->port_stats.tx_timeout++; en_dbg(DRV, priv, "Watchdog task called for port %d\n", priv->port); mutex_lock(&mdev->state_lock); if (priv->port_up) { mlx4_en_stop_port(dev); //for (i = 0; i < priv->tx_ring_num; i++) // netdev_tx_reset_queue(priv->tx_ring[i]->tx_queue); if (mlx4_en_start_port(dev)) en_err(priv, "Failed restarting port %d\n", priv->port); } mutex_unlock(&mdev->state_lock); } static void mlx4_en_clear_stats(struct net_device *dev) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_dev *mdev = priv->mdev; int i; if (!mlx4_is_slave(mdev->dev)) if (mlx4_en_DUMP_ETH_STATS(mdev, priv->port, 1)) en_dbg(HW, priv, "Failed dumping statistics\n"); memset(&priv->pstats, 0, sizeof(priv->pstats)); memset(&priv->pkstats, 0, sizeof(priv->pkstats)); memset(&priv->port_stats, 0, sizeof(priv->port_stats)); memset(&priv->vport_stats, 0, sizeof(priv->vport_stats)); for (i = 0; i < priv->tx_ring_num; i++) { priv->tx_ring[i]->bytes = 0; priv->tx_ring[i]->packets = 0; priv->tx_ring[i]->tx_csum = 0; } for (i = 0; i < priv->rx_ring_num; i++) { priv->rx_ring[i]->bytes = 0; priv->rx_ring[i]->packets = 0; priv->rx_ring[i]->csum_ok = 0; priv->rx_ring[i]->csum_none = 0; } } static void mlx4_en_open(void* arg) { struct mlx4_en_priv *priv; struct mlx4_en_dev *mdev; struct net_device *dev; int err = 0; priv = arg; mdev = priv->mdev; dev = priv->dev; mutex_lock(&mdev->state_lock); if (!mdev->device_up) { en_err(priv, "Cannot open - device down/disabled\n"); goto out; } /* Reset HW statistics and SW counters */ mlx4_en_clear_stats(dev); err = mlx4_en_start_port(dev); if (err) en_err(priv, "Failed starting port:%d\n", priv->port); out: mutex_unlock(&mdev->state_lock); return; } void mlx4_en_free_resources(struct mlx4_en_priv *priv) { int i; #ifdef CONFIG_RFS_ACCEL if (priv->dev->rx_cpu_rmap) { free_irq_cpu_rmap(priv->dev->rx_cpu_rmap); priv->dev->rx_cpu_rmap = NULL; } #endif for (i = 0; i < priv->tx_ring_num; i++) { if (priv->tx_ring && priv->tx_ring[i]) mlx4_en_destroy_tx_ring(priv, &priv->tx_ring[i]); if (priv->tx_cq && priv->tx_cq[i]) mlx4_en_destroy_cq(priv, &priv->tx_cq[i]); } for (i = 0; i < priv->rx_ring_num; i++) { if (priv->rx_ring[i]) mlx4_en_destroy_rx_ring(priv, &priv->rx_ring[i], priv->prof->rx_ring_size, priv->stride); if (priv->rx_cq[i]) mlx4_en_destroy_cq(priv, &priv->rx_cq[i]); } if (priv->sysctl) sysctl_ctx_free(&priv->stat_ctx); } int mlx4_en_alloc_resources(struct mlx4_en_priv *priv) { struct mlx4_en_port_profile *prof = priv->prof; int i; int node = 0; /* Create rx Rings */ for (i = 0; i < priv->rx_ring_num; i++) { if (mlx4_en_create_cq(priv, &priv->rx_cq[i], prof->rx_ring_size, i, RX, node)) goto err; if (mlx4_en_create_rx_ring(priv, &priv->rx_ring[i], prof->rx_ring_size, node)) goto err; } /* Create tx Rings */ for (i = 0; i < priv->tx_ring_num; i++) { if (mlx4_en_create_cq(priv, &priv->tx_cq[i], prof->tx_ring_size, i, TX, node)) goto err; if (mlx4_en_create_tx_ring(priv, &priv->tx_ring[i], prof->tx_ring_size, TXBB_SIZE, node, i)) goto err; } #ifdef CONFIG_RFS_ACCEL priv->dev->rx_cpu_rmap = alloc_irq_cpu_rmap(priv->rx_ring_num); if (!priv->dev->rx_cpu_rmap) goto err; #endif /* Re-create stat sysctls in case the number of rings changed. */ mlx4_en_sysctl_stat(priv); return 0; err: en_err(priv, "Failed to allocate NIC resources\n"); for (i = 0; i < priv->rx_ring_num; i++) { if (priv->rx_ring[i]) mlx4_en_destroy_rx_ring(priv, &priv->rx_ring[i], prof->rx_ring_size, priv->stride); if (priv->rx_cq[i]) mlx4_en_destroy_cq(priv, &priv->rx_cq[i]); } for (i = 0; i < priv->tx_ring_num; i++) { if (priv->tx_ring[i]) mlx4_en_destroy_tx_ring(priv, &priv->tx_ring[i]); if (priv->tx_cq[i]) mlx4_en_destroy_cq(priv, &priv->tx_cq[i]); } priv->port_up = false; return -ENOMEM; } struct en_port_attribute { struct attribute attr; ssize_t (*show)(struct en_port *, struct en_port_attribute *, char *buf); ssize_t (*store)(struct en_port *, struct en_port_attribute *, char *buf, size_t count); }; #define PORT_ATTR_RO(_name) \ struct en_port_attribute en_port_attr_##_name = __ATTR_RO(_name) #define EN_PORT_ATTR(_name, _mode, _show, _store) \ struct en_port_attribute en_port_attr_##_name = __ATTR(_name, _mode, _show, _store) void mlx4_en_destroy_netdev(struct net_device *dev) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_dev *mdev = priv->mdev; en_dbg(DRV, priv, "Destroying netdev on port:%d\n", priv->port); if (priv->vlan_attach != NULL) EVENTHANDLER_DEREGISTER(vlan_config, priv->vlan_attach); if (priv->vlan_detach != NULL) EVENTHANDLER_DEREGISTER(vlan_unconfig, priv->vlan_detach); /* Unregister device - this will close the port if it was up */ if (priv->registered) ether_ifdetach(dev); if (priv->allocated) mlx4_free_hwq_res(mdev->dev, &priv->res, MLX4_EN_PAGE_SIZE); mutex_lock(&mdev->state_lock); mlx4_en_stop_port(dev); mutex_unlock(&mdev->state_lock); cancel_delayed_work(&priv->stats_task); cancel_delayed_work(&priv->service_task); /* flush any pending task for this netdev */ flush_workqueue(mdev->workqueue); callout_drain(&priv->watchdog_timer); /* Detach the netdev so tasks would not attempt to access it */ mutex_lock(&mdev->state_lock); mdev->pndev[priv->port] = NULL; mutex_unlock(&mdev->state_lock); mlx4_en_free_resources(priv); /* freeing the sysctl conf cannot be called from within mlx4_en_free_resources */ if (priv->sysctl) sysctl_ctx_free(&priv->conf_ctx); kfree(priv->tx_ring); kfree(priv->tx_cq); kfree(priv); if_free(dev); } static int mlx4_en_change_mtu(struct net_device *dev, int new_mtu) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_dev *mdev = priv->mdev; int err = 0; en_dbg(DRV, priv, "Change MTU called - current:%u new:%u\n", (unsigned)dev->if_mtu, (unsigned)new_mtu); if ((new_mtu < MLX4_EN_MIN_MTU) || (new_mtu > priv->max_mtu)) { en_err(priv, "Bad MTU size:%d.\n", new_mtu); return -EPERM; } mutex_lock(&mdev->state_lock); dev->if_mtu = new_mtu; if (dev->if_drv_flags & IFF_DRV_RUNNING) { if (!mdev->device_up) { /* NIC is probably restarting - let watchdog task reset * * the port */ en_dbg(DRV, priv, "Change MTU called with card down!?\n"); } else { mlx4_en_stop_port(dev); err = mlx4_en_start_port(dev); if (err) { en_err(priv, "Failed restarting port:%d\n", priv->port); queue_work(mdev->workqueue, &priv->watchdog_task); } } } mutex_unlock(&mdev->state_lock); return 0; } static int mlx4_en_calc_media(struct mlx4_en_priv *priv) { int trans_type; int active; active = IFM_ETHER; if (priv->last_link_state == MLX4_DEV_EVENT_PORT_DOWN) return (active); /* * [ShaharK] mlx4_en_QUERY_PORT sleeps and cannot be called under a * non-sleepable lock. * I moved it to the periodic mlx4_en_do_get_stats. if (mlx4_en_QUERY_PORT(priv->mdev, priv->port)) return (active); */ active |= IFM_FDX; trans_type = priv->port_state.transciver; /* XXX I don't know all of the transceiver values. */ switch (priv->port_state.link_speed) { case 1000: active |= IFM_1000_T; break; case 10000: if (trans_type > 0 && trans_type <= 0xC) active |= IFM_10G_SR; else if (trans_type == 0x80 || trans_type == 0) active |= IFM_10G_CX4; break; case 40000: active |= IFM_40G_CR4; break; } if (priv->prof->tx_pause) active |= IFM_ETH_TXPAUSE; if (priv->prof->rx_pause) active |= IFM_ETH_RXPAUSE; return (active); } static void mlx4_en_media_status(struct ifnet *dev, struct ifmediareq *ifmr) { struct mlx4_en_priv *priv; priv = dev->if_softc; ifmr->ifm_status = IFM_AVALID; if (priv->last_link_state != MLX4_DEV_EVENT_PORT_DOWN) ifmr->ifm_status |= IFM_ACTIVE; ifmr->ifm_active = mlx4_en_calc_media(priv); return; } static int mlx4_en_media_change(struct ifnet *dev) { struct mlx4_en_priv *priv; struct ifmedia *ifm; int rxpause; int txpause; int error; priv = dev->if_softc; ifm = &priv->media; rxpause = txpause = 0; error = 0; if (IFM_TYPE(ifm->ifm_media) != IFM_ETHER) return (EINVAL); switch (IFM_SUBTYPE(ifm->ifm_media)) { case IFM_AUTO: break; case IFM_10G_SR: case IFM_10G_CX4: case IFM_1000_T: case IFM_40G_CR4: if ((IFM_SUBTYPE(ifm->ifm_media) == IFM_SUBTYPE(mlx4_en_calc_media(priv))) && (ifm->ifm_media & IFM_FDX)) break; /* Fallthrough */ default: printf("%s: Only auto media type\n", if_name(dev)); return (EINVAL); } /* Allow user to set/clear pause */ if (IFM_OPTIONS(ifm->ifm_media) & IFM_ETH_RXPAUSE) rxpause = 1; if (IFM_OPTIONS(ifm->ifm_media) & IFM_ETH_TXPAUSE) txpause = 1; if (priv->prof->tx_pause != txpause || priv->prof->rx_pause != rxpause) { priv->prof->tx_pause = txpause; priv->prof->rx_pause = rxpause; error = -mlx4_SET_PORT_general(priv->mdev->dev, priv->port, priv->rx_mb_size + ETHER_CRC_LEN, priv->prof->tx_pause, priv->prof->tx_ppp, priv->prof->rx_pause, priv->prof->rx_ppp); } return (error); } static int mlx4_en_ioctl(struct ifnet *dev, u_long command, caddr_t data) { struct mlx4_en_priv *priv; struct mlx4_en_dev *mdev; struct ifreq *ifr; int error; int mask; error = 0; mask = 0; priv = dev->if_softc; mdev = priv->mdev; ifr = (struct ifreq *) data; switch (command) { case SIOCSIFMTU: error = -mlx4_en_change_mtu(dev, ifr->ifr_mtu); break; case SIOCSIFFLAGS: mutex_lock(&mdev->state_lock); if (dev->if_flags & IFF_UP) { if ((dev->if_drv_flags & IFF_DRV_RUNNING) == 0) mlx4_en_start_port(dev); else mlx4_en_set_rx_mode(dev); } else { if (dev->if_drv_flags & IFF_DRV_RUNNING) { mlx4_en_stop_port(dev); if_link_state_change(dev, LINK_STATE_DOWN); } } mutex_unlock(&mdev->state_lock); break; case SIOCADDMULTI: case SIOCDELMULTI: mlx4_en_set_rx_mode(dev); break; case SIOCSIFMEDIA: case SIOCGIFMEDIA: error = ifmedia_ioctl(dev, ifr, &priv->media, command); break; case SIOCSIFCAP: mutex_lock(&mdev->state_lock); mask = ifr->ifr_reqcap ^ dev->if_capenable; if (mask & IFCAP_HWCSUM) dev->if_capenable ^= IFCAP_HWCSUM; if (mask & IFCAP_TSO4) dev->if_capenable ^= IFCAP_TSO4; if (mask & IFCAP_TSO6) dev->if_capenable ^= IFCAP_TSO6; if (mask & IFCAP_LRO) dev->if_capenable ^= IFCAP_LRO; if (mask & IFCAP_VLAN_HWTAGGING) dev->if_capenable ^= IFCAP_VLAN_HWTAGGING; if (mask & IFCAP_VLAN_HWFILTER) dev->if_capenable ^= IFCAP_VLAN_HWFILTER; if (mask & IFCAP_WOL_MAGIC) dev->if_capenable ^= IFCAP_WOL_MAGIC; if (dev->if_drv_flags & IFF_DRV_RUNNING) mlx4_en_start_port(dev); mutex_unlock(&mdev->state_lock); VLAN_CAPABILITIES(dev); break; default: error = ether_ioctl(dev, command, data); break; } return (error); } int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port, struct mlx4_en_port_profile *prof) { struct net_device *dev; struct mlx4_en_priv *priv; uint8_t dev_addr[ETHER_ADDR_LEN]; int err; int i; priv = kzalloc(sizeof(*priv), GFP_KERNEL); dev = priv->dev = if_alloc(IFT_ETHER); if (dev == NULL) { en_err(priv, "Net device allocation failed\n"); kfree(priv); return -ENOMEM; } dev->if_softc = priv; if_initname(dev, "mlxen", atomic_fetchadd_int(&mlx4_en_unit, 1)); dev->if_mtu = ETHERMTU; dev->if_init = mlx4_en_open; dev->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST; dev->if_ioctl = mlx4_en_ioctl; dev->if_transmit = mlx4_en_transmit; dev->if_qflush = mlx4_en_qflush; dev->if_snd.ifq_maxlen = prof->tx_ring_size; /* * Initialize driver private data */ priv->counter_index = 0xff; spin_lock_init(&priv->stats_lock); INIT_WORK(&priv->rx_mode_task, mlx4_en_do_set_rx_mode); INIT_WORK(&priv->watchdog_task, mlx4_en_restart); INIT_WORK(&priv->linkstate_task, mlx4_en_linkstate); INIT_DELAYED_WORK(&priv->stats_task, mlx4_en_do_get_stats); INIT_DELAYED_WORK(&priv->service_task, mlx4_en_service_task); callout_init(&priv->watchdog_timer, 1); #ifdef CONFIG_RFS_ACCEL INIT_LIST_HEAD(&priv->filters); spin_lock_init(&priv->filters_lock); #endif priv->msg_enable = MLX4_EN_MSG_LEVEL; priv->dev = dev; priv->mdev = mdev; priv->ddev = &mdev->pdev->dev; priv->prof = prof; priv->port = port; priv->port_up = false; priv->flags = prof->flags; priv->ctrl_flags = cpu_to_be32(MLX4_WQE_CTRL_CQ_UPDATE | MLX4_WQE_CTRL_SOLICITED); priv->num_tx_rings_p_up = mdev->profile.num_tx_rings_p_up; priv->tx_ring_num = prof->tx_ring_num; priv->tx_ring = kcalloc(MAX_TX_RINGS, sizeof(struct mlx4_en_tx_ring *), GFP_KERNEL); if (!priv->tx_ring) { err = -ENOMEM; goto out; } priv->tx_cq = kcalloc(sizeof(struct mlx4_en_cq *), MAX_TX_RINGS, GFP_KERNEL); if (!priv->tx_cq) { err = -ENOMEM; goto out; } priv->rx_ring_num = prof->rx_ring_num; priv->cqe_factor = (mdev->dev->caps.cqe_size == 64) ? 1 : 0; priv->mac_index = -1; priv->last_ifq_jiffies = 0; priv->if_counters_rx_errors = 0; priv->if_counters_rx_no_buffer = 0; #ifdef CONFIG_MLX4_EN_DCB if (!mlx4_is_slave(priv->mdev->dev)) { priv->dcbx_cap = DCB_CAP_DCBX_HOST; priv->flags |= MLX4_EN_FLAG_DCB_ENABLED; if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_ETS_CFG) { dev->dcbnl_ops = &mlx4_en_dcbnl_ops; } else { en_info(priv, "QoS disabled - no HW support\n"); dev->dcbnl_ops = &mlx4_en_dcbnl_pfc_ops; } } #endif for (i = 0; i < MLX4_EN_MAC_HASH_SIZE; ++i) INIT_HLIST_HEAD(&priv->mac_hash[i]); /* Query for default mac and max mtu */ priv->max_mtu = mdev->dev->caps.eth_mtu_cap[priv->port]; priv->mac = mdev->dev->caps.def_mac[priv->port]; if (ILLEGAL_MAC(priv->mac)) { #if BITS_PER_LONG == 64 en_err(priv, "Port: %d, invalid mac burned: 0x%lx, quiting\n", priv->port, priv->mac); #elif BITS_PER_LONG == 32 en_err(priv, "Port: %d, invalid mac burned: 0x%llx, quiting\n", priv->port, priv->mac); #endif err = -EINVAL; goto out; } priv->stride = roundup_pow_of_two(sizeof(struct mlx4_en_rx_desc) + DS_SIZE); mlx4_en_sysctl_conf(priv); err = mlx4_en_alloc_resources(priv); if (err) goto out; /* Allocate page for receive rings */ err = mlx4_alloc_hwq_res(mdev->dev, &priv->res, MLX4_EN_PAGE_SIZE, MLX4_EN_PAGE_SIZE); if (err) { en_err(priv, "Failed to allocate page for rx qps\n"); goto out; } priv->allocated = 1; /* * Set driver features */ dev->if_capabilities |= IFCAP_RXCSUM | IFCAP_TXCSUM; dev->if_capabilities |= IFCAP_VLAN_MTU | IFCAP_VLAN_HWTAGGING; dev->if_capabilities |= IFCAP_VLAN_HWCSUM | IFCAP_VLAN_HWFILTER; dev->if_capabilities |= IFCAP_LINKSTATE | IFCAP_JUMBO_MTU; dev->if_capabilities |= IFCAP_LRO; if (mdev->LSO_support) dev->if_capabilities |= IFCAP_TSO4 | IFCAP_TSO6 | IFCAP_VLAN_HWTSO; /* set TSO limits so that we don't have to drop TX packets */ dev->if_hw_tsomax = 65536 - (ETHER_HDR_LEN + ETHER_VLAN_ENCAP_LEN); dev->if_hw_tsomaxsegcount = 16; dev->if_hw_tsomaxsegsize = 65536; /* XXX can do up to 4GByte */ dev->if_capenable = dev->if_capabilities; dev->if_hwassist = 0; if (dev->if_capenable & (IFCAP_TSO4 | IFCAP_TSO6)) dev->if_hwassist |= CSUM_TSO; if (dev->if_capenable & IFCAP_TXCSUM) dev->if_hwassist |= (CSUM_TCP | CSUM_UDP | CSUM_IP); /* Register for VLAN events */ priv->vlan_attach = EVENTHANDLER_REGISTER(vlan_config, mlx4_en_vlan_rx_add_vid, priv, EVENTHANDLER_PRI_FIRST); priv->vlan_detach = EVENTHANDLER_REGISTER(vlan_unconfig, mlx4_en_vlan_rx_kill_vid, priv, EVENTHANDLER_PRI_FIRST); mdev->pndev[priv->port] = dev; priv->last_link_state = MLX4_DEV_EVENT_PORT_DOWN; mlx4_en_set_default_moderation(priv); /* Set default MAC */ for (i = 0; i < ETHER_ADDR_LEN; i++) dev_addr[ETHER_ADDR_LEN - 1 - i] = (u8) (priv->mac >> (8 * i)); ether_ifattach(dev, dev_addr); if_link_state_change(dev, LINK_STATE_DOWN); ifmedia_init(&priv->media, IFM_IMASK | IFM_ETH_FMASK, mlx4_en_media_change, mlx4_en_media_status); ifmedia_add(&priv->media, IFM_ETHER | IFM_FDX | IFM_1000_T, 0, NULL); ifmedia_add(&priv->media, IFM_ETHER | IFM_FDX | IFM_10G_SR, 0, NULL); ifmedia_add(&priv->media, IFM_ETHER | IFM_FDX | IFM_10G_CX4, 0, NULL); ifmedia_add(&priv->media, IFM_ETHER | IFM_FDX | IFM_40G_CR4, 0, NULL); ifmedia_add(&priv->media, IFM_ETHER | IFM_AUTO, 0, NULL); ifmedia_set(&priv->media, IFM_ETHER | IFM_AUTO); en_warn(priv, "Using %d TX rings\n", prof->tx_ring_num); en_warn(priv, "Using %d RX rings\n", prof->rx_ring_num); priv->registered = 1; en_warn(priv, "Using %d TX rings\n", prof->tx_ring_num); en_warn(priv, "Using %d RX rings\n", prof->rx_ring_num); priv->rx_mb_size = dev->if_mtu + ETH_HLEN + VLAN_HLEN + ETH_FCS_LEN; err = mlx4_SET_PORT_general(mdev->dev, priv->port, priv->rx_mb_size, prof->tx_pause, prof->tx_ppp, prof->rx_pause, prof->rx_ppp); if (err) { en_err(priv, "Failed setting port general configurations " "for port %d, with error %d\n", priv->port, err); goto out; } /* Init port */ en_warn(priv, "Initializing port\n"); err = mlx4_INIT_PORT(mdev->dev, priv->port); if (err) { en_err(priv, "Failed Initializing port\n"); goto out; } queue_delayed_work(mdev->workqueue, &priv->stats_task, STATS_DELAY); if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_TS) queue_delayed_work(mdev->workqueue, &priv->service_task, SERVICE_TASK_DELAY); return 0; out: mlx4_en_destroy_netdev(dev); return err; } + static int mlx4_en_set_ring_size(struct net_device *dev, int rx_size, int tx_size) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_dev *mdev = priv->mdev; int port_up = 0; int err = 0; rx_size = roundup_pow_of_two(rx_size); rx_size = max_t(u32, rx_size, MLX4_EN_MIN_RX_SIZE); rx_size = min_t(u32, rx_size, MLX4_EN_MAX_RX_SIZE); tx_size = roundup_pow_of_two(tx_size); tx_size = max_t(u32, tx_size, MLX4_EN_MIN_TX_SIZE); tx_size = min_t(u32, tx_size, MLX4_EN_MAX_TX_SIZE); if (rx_size == (priv->port_up ? priv->rx_ring[0]->actual_size : priv->rx_ring[0]->size) && tx_size == priv->tx_ring[0]->size) return 0; mutex_lock(&mdev->state_lock); if (priv->port_up) { port_up = 1; mlx4_en_stop_port(dev); } mlx4_en_free_resources(priv); priv->prof->tx_ring_size = tx_size; priv->prof->rx_ring_size = rx_size; err = mlx4_en_alloc_resources(priv); if (err) { en_err(priv, "Failed reallocating port resources\n"); goto out; } if (port_up) { err = mlx4_en_start_port(dev); if (err) en_err(priv, "Failed starting port\n"); } out: mutex_unlock(&mdev->state_lock); return err; } static int mlx4_en_set_rx_ring_size(SYSCTL_HANDLER_ARGS) { struct mlx4_en_priv *priv; int size; int error; priv = arg1; size = priv->prof->rx_ring_size; error = sysctl_handle_int(oidp, &size, 0, req); if (error || !req->newptr) return (error); error = -mlx4_en_set_ring_size(priv->dev, size, priv->prof->tx_ring_size); return (error); } static int mlx4_en_set_tx_ring_size(SYSCTL_HANDLER_ARGS) { struct mlx4_en_priv *priv; int size; int error; priv = arg1; size = priv->prof->tx_ring_size; error = sysctl_handle_int(oidp, &size, 0, req); if (error || !req->newptr) return (error); error = -mlx4_en_set_ring_size(priv->dev, priv->prof->rx_ring_size, size); return (error); } static int mlx4_en_set_tx_ppp(SYSCTL_HANDLER_ARGS) { struct mlx4_en_priv *priv; int ppp; int error; priv = arg1; ppp = priv->prof->tx_ppp; error = sysctl_handle_int(oidp, &ppp, 0, req); if (error || !req->newptr) return (error); if (ppp > 0xff || ppp < 0) return (-EINVAL); priv->prof->tx_ppp = ppp; error = -mlx4_SET_PORT_general(priv->mdev->dev, priv->port, priv->rx_mb_size + ETHER_CRC_LEN, priv->prof->tx_pause, priv->prof->tx_ppp, priv->prof->rx_pause, priv->prof->rx_ppp); return (error); } static int mlx4_en_set_rx_ppp(SYSCTL_HANDLER_ARGS) { struct mlx4_en_priv *priv; struct mlx4_en_dev *mdev; int ppp; int error; int port_up; port_up = 0; priv = arg1; mdev = priv->mdev; ppp = priv->prof->rx_ppp; error = sysctl_handle_int(oidp, &ppp, 0, req); if (error || !req->newptr) return (error); if (ppp > 0xff || ppp < 0) return (-EINVAL); /* See if we have to change the number of tx queues. */ if (!ppp != !priv->prof->rx_ppp) { mutex_lock(&mdev->state_lock); if (priv->port_up) { port_up = 1; mlx4_en_stop_port(priv->dev); } mlx4_en_free_resources(priv); priv->prof->rx_ppp = ppp; error = -mlx4_en_alloc_resources(priv); if (error) en_err(priv, "Failed reallocating port resources\n"); if (error == 0 && port_up) { error = -mlx4_en_start_port(priv->dev); if (error) en_err(priv, "Failed starting port\n"); } mutex_unlock(&mdev->state_lock); return (error); } priv->prof->rx_ppp = ppp; error = -mlx4_SET_PORT_general(priv->mdev->dev, priv->port, priv->rx_mb_size + ETHER_CRC_LEN, priv->prof->tx_pause, priv->prof->tx_ppp, priv->prof->rx_pause, priv->prof->rx_ppp); return (error); } static void mlx4_en_sysctl_conf(struct mlx4_en_priv *priv) { struct net_device *dev; struct sysctl_ctx_list *ctx; struct sysctl_oid *node; struct sysctl_oid_list *node_list; struct sysctl_oid *coal; struct sysctl_oid_list *coal_list; const char *pnameunit; dev = priv->dev; ctx = &priv->conf_ctx; pnameunit = device_get_nameunit(priv->mdev->pdev->dev.bsddev); sysctl_ctx_init(ctx); priv->sysctl = SYSCTL_ADD_NODE(ctx, SYSCTL_STATIC_CHILDREN(_hw), OID_AUTO, dev->if_xname, CTLFLAG_RD, 0, "mlx4 10gig ethernet"); node = SYSCTL_ADD_NODE(ctx, SYSCTL_CHILDREN(priv->sysctl), OID_AUTO, "conf", CTLFLAG_RD, NULL, "Configuration"); node_list = SYSCTL_CHILDREN(node); SYSCTL_ADD_UINT(ctx, node_list, OID_AUTO, "msg_enable", CTLFLAG_RW, &priv->msg_enable, 0, "Driver message enable bitfield"); SYSCTL_ADD_UINT(ctx, node_list, OID_AUTO, "rx_rings", CTLFLAG_RD, &priv->rx_ring_num, 0, "Number of receive rings"); SYSCTL_ADD_UINT(ctx, node_list, OID_AUTO, "tx_rings", CTLFLAG_RD, &priv->tx_ring_num, 0, "Number of transmit rings"); SYSCTL_ADD_PROC(ctx, node_list, OID_AUTO, "rx_size", CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE, priv, 0, mlx4_en_set_rx_ring_size, "I", "Receive ring size"); SYSCTL_ADD_PROC(ctx, node_list, OID_AUTO, "tx_size", CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE, priv, 0, mlx4_en_set_tx_ring_size, "I", "Transmit ring size"); SYSCTL_ADD_PROC(ctx, node_list, OID_AUTO, "tx_ppp", CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE, priv, 0, mlx4_en_set_tx_ppp, "I", "TX Per-priority pause"); SYSCTL_ADD_PROC(ctx, node_list, OID_AUTO, "rx_ppp", CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE, priv, 0, mlx4_en_set_rx_ppp, "I", "RX Per-priority pause"); SYSCTL_ADD_UINT(ctx, node_list, OID_AUTO, "port_num", CTLFLAG_RD, &priv->port, 0, "Port Number"); SYSCTL_ADD_STRING(ctx, node_list, OID_AUTO, "device_name", CTLFLAG_RD, __DECONST(void *, pnameunit), 0, "PCI device name"); /* Add coalescer configuration. */ coal = SYSCTL_ADD_NODE(ctx, node_list, OID_AUTO, "coalesce", CTLFLAG_RD, NULL, "Interrupt coalesce configuration"); coal_list = SYSCTL_CHILDREN(node); SYSCTL_ADD_UINT(ctx, coal_list, OID_AUTO, "pkt_rate_low", CTLFLAG_RW, &priv->pkt_rate_low, 0, "Packets per-second for minimum delay"); SYSCTL_ADD_UINT(ctx, coal_list, OID_AUTO, "rx_usecs_low", CTLFLAG_RW, &priv->rx_usecs_low, 0, "Minimum RX delay in micro-seconds"); SYSCTL_ADD_UINT(ctx, coal_list, OID_AUTO, "pkt_rate_high", CTLFLAG_RW, &priv->pkt_rate_high, 0, "Packets per-second for maximum delay"); SYSCTL_ADD_UINT(ctx, coal_list, OID_AUTO, "rx_usecs_high", CTLFLAG_RW, &priv->rx_usecs_high, 0, "Maximum RX delay in micro-seconds"); SYSCTL_ADD_UINT(ctx, coal_list, OID_AUTO, "sample_interval", CTLFLAG_RW, &priv->sample_interval, 0, "adaptive frequency in units of HZ ticks"); SYSCTL_ADD_UINT(ctx, coal_list, OID_AUTO, "adaptive_rx_coal", CTLFLAG_RW, &priv->adaptive_rx_coal, 0, "Enable adaptive rx coalescing"); } - static void mlx4_en_sysctl_stat(struct mlx4_en_priv *priv) { struct net_device *dev; struct sysctl_ctx_list *ctx; struct sysctl_oid *node; struct sysctl_oid_list *node_list; struct sysctl_oid *ring_node; struct sysctl_oid_list *ring_list; struct mlx4_en_tx_ring *tx_ring; struct mlx4_en_rx_ring *rx_ring; char namebuf[128]; int i; dev = priv->dev; ctx = &priv->stat_ctx; sysctl_ctx_init(ctx); node = SYSCTL_ADD_NODE(ctx, SYSCTL_CHILDREN(priv->sysctl), OID_AUTO, "stat", CTLFLAG_RD, NULL, "Statistics"); node_list = SYSCTL_CHILDREN(node); #ifdef MLX4_EN_PERF_STAT SYSCTL_ADD_UINT(ctx, node_list, OID_AUTO, "tx_poll", CTLFLAG_RD, &priv->pstats.tx_poll, "TX Poll calls"); SYSCTL_ADD_QUAD(ctx, node_list, OID_AUTO, "tx_pktsz_avg", CTLFLAG_RD, &priv->pstats.tx_pktsz_avg, "TX average packet size"); SYSCTL_ADD_UINT(ctx, node_list, OID_AUTO, "inflight_avg", CTLFLAG_RD, &priv->pstats.inflight_avg, "TX average packets in-flight"); SYSCTL_ADD_UINT(ctx, node_list, OID_AUTO, "tx_coal_avg", CTLFLAG_RD, &priv->pstats.tx_coal_avg, "TX average coalesced completions"); SYSCTL_ADD_UINT(ctx, node_list, OID_AUTO, "rx_coal_avg", CTLFLAG_RD, &priv->pstats.rx_coal_avg, "RX average coalesced completions"); #endif SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tso_packets", CTLFLAG_RD, &priv->port_stats.tso_packets, "TSO packets sent"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "queue_stopped", CTLFLAG_RD, &priv->port_stats.queue_stopped, "Queue full"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "wake_queue", CTLFLAG_RD, &priv->port_stats.wake_queue, "Queue resumed after full"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_timeout", CTLFLAG_RD, &priv->port_stats.tx_timeout, "Transmit timeouts"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_alloc_failed", CTLFLAG_RD, &priv->port_stats.rx_alloc_failed, "RX failed to allocate mbuf"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_chksum_good", CTLFLAG_RD, &priv->port_stats.rx_chksum_good, "RX checksum offload success"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_chksum_none", CTLFLAG_RD, &priv->port_stats.rx_chksum_none, "RX without checksum offload"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_chksum_offload", CTLFLAG_RD, &priv->port_stats.tx_chksum_offload, "TX checksum offloads"); /* Could strdup the names and add in a loop. This is simpler. */ SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_bytes", CTLFLAG_RD, &priv->pkstats.rx_bytes, "RX Bytes"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_packets", CTLFLAG_RD, &priv->pkstats.rx_packets, "RX packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_multicast_packets", CTLFLAG_RD, &priv->pkstats.rx_multicast_packets, "RX Multicast Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_broadcast_packets", CTLFLAG_RD, &priv->pkstats.rx_broadcast_packets, "RX Broadcast Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_errors", CTLFLAG_RD, &priv->pkstats.rx_errors, "RX Errors"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_dropped", CTLFLAG_RD, &priv->pkstats.rx_dropped, "RX Dropped"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_length_errors", CTLFLAG_RD, &priv->pkstats.rx_length_errors, "RX Length Errors"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_over_errors", CTLFLAG_RD, &priv->pkstats.rx_over_errors, "RX Over Errors"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_crc_errors", CTLFLAG_RD, &priv->pkstats.rx_crc_errors, "RX CRC Errors"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_jabbers", CTLFLAG_RD, &priv->pkstats.rx_jabbers, "RX Jabbers"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_in_range_length_error", CTLFLAG_RD, &priv->pkstats.rx_in_range_length_error, "RX IN_Range Length Error"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_out_range_length_error", CTLFLAG_RD, &priv->pkstats.rx_out_range_length_error, "RX Out Range Length Error"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_lt_64_bytes_packets", CTLFLAG_RD, &priv->pkstats.rx_lt_64_bytes_packets, "RX Lt 64 Bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_127_bytes_packets", CTLFLAG_RD, &priv->pkstats.rx_127_bytes_packets, "RX 127 bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_255_bytes_packets", CTLFLAG_RD, &priv->pkstats.rx_255_bytes_packets, "RX 255 bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_511_bytes_packets", CTLFLAG_RD, &priv->pkstats.rx_511_bytes_packets, "RX 511 bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_1023_bytes_packets", CTLFLAG_RD, &priv->pkstats.rx_1023_bytes_packets, "RX 1023 bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_1518_bytes_packets", CTLFLAG_RD, &priv->pkstats.rx_1518_bytes_packets, "RX 1518 bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_1522_bytes_packets", CTLFLAG_RD, &priv->pkstats.rx_1522_bytes_packets, "RX 1522 bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_1548_bytes_packets", CTLFLAG_RD, &priv->pkstats.rx_1548_bytes_packets, "RX 1548 bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "rx_gt_1548_bytes_packets", CTLFLAG_RD, &priv->pkstats.rx_gt_1548_bytes_packets, "RX Greater Then 1548 bytes Packets"); struct mlx4_en_pkt_stats { unsigned long tx_packets; unsigned long tx_bytes; unsigned long tx_multicast_packets; unsigned long tx_broadcast_packets; unsigned long tx_errors; unsigned long tx_dropped; unsigned long tx_lt_64_bytes_packets; unsigned long tx_127_bytes_packets; unsigned long tx_255_bytes_packets; unsigned long tx_511_bytes_packets; unsigned long tx_1023_bytes_packets; unsigned long tx_1518_bytes_packets; unsigned long tx_1522_bytes_packets; unsigned long tx_1548_bytes_packets; unsigned long tx_gt_1548_bytes_packets; unsigned long rx_prio[NUM_PRIORITIES][NUM_PRIORITY_STATS]; unsigned long tx_prio[NUM_PRIORITIES][NUM_PRIORITY_STATS]; #define NUM_PKT_STATS 72 }; SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_packets", CTLFLAG_RD, &priv->pkstats.tx_packets, "TX packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_bytes", CTLFLAG_RD, &priv->pkstats.tx_packets, "TX Bytes"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_multicast_packets", CTLFLAG_RD, &priv->pkstats.tx_multicast_packets, "TX Multicast Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_broadcast_packets", CTLFLAG_RD, &priv->pkstats.tx_broadcast_packets, "TX Broadcast Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_errors", CTLFLAG_RD, &priv->pkstats.tx_errors, "TX Errors"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_dropped", CTLFLAG_RD, &priv->pkstats.tx_dropped, "TX Dropped"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_lt_64_bytes_packets", CTLFLAG_RD, &priv->pkstats.tx_lt_64_bytes_packets, "TX Less Then 64 Bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_127_bytes_packets", CTLFLAG_RD, &priv->pkstats.tx_127_bytes_packets, "TX 127 Bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_255_bytes_packets", CTLFLAG_RD, &priv->pkstats.tx_255_bytes_packets, "TX 255 Bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_511_bytes_packets", CTLFLAG_RD, &priv->pkstats.tx_511_bytes_packets, "TX 511 Bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_1023_bytes_packets", CTLFLAG_RD, &priv->pkstats.tx_1023_bytes_packets, "TX 1023 Bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_1518_bytes_packets", CTLFLAG_RD, &priv->pkstats.tx_1518_bytes_packets, "TX 1518 Bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_1522_bytes_packets", CTLFLAG_RD, &priv->pkstats.tx_1522_bytes_packets, "TX 1522 Bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_1548_bytes_packets", CTLFLAG_RD, &priv->pkstats.tx_1548_bytes_packets, "TX 1548 Bytes Packets"); SYSCTL_ADD_ULONG(ctx, node_list, OID_AUTO, "tx_gt_1548_bytes_packets", CTLFLAG_RD, &priv->pkstats.tx_gt_1548_bytes_packets, "TX Greater Then 1548 Bytes Packets"); for (i = 0; i < priv->tx_ring_num; i++) { tx_ring = priv->tx_ring[i]; snprintf(namebuf, sizeof(namebuf), "tx_ring%d", i); ring_node = SYSCTL_ADD_NODE(ctx, node_list, OID_AUTO, namebuf, CTLFLAG_RD, NULL, "TX Ring"); ring_list = SYSCTL_CHILDREN(ring_node); SYSCTL_ADD_ULONG(ctx, ring_list, OID_AUTO, "packets", CTLFLAG_RD, &tx_ring->packets, "TX packets"); SYSCTL_ADD_ULONG(ctx, ring_list, OID_AUTO, "bytes", CTLFLAG_RD, &tx_ring->bytes, "TX bytes"); } for (i = 0; i < priv->rx_ring_num; i++) { rx_ring = priv->rx_ring[i]; snprintf(namebuf, sizeof(namebuf), "rx_ring%d", i); ring_node = SYSCTL_ADD_NODE(ctx, node_list, OID_AUTO, namebuf, CTLFLAG_RD, NULL, "RX Ring"); ring_list = SYSCTL_CHILDREN(ring_node); SYSCTL_ADD_ULONG(ctx, ring_list, OID_AUTO, "packets", CTLFLAG_RD, &rx_ring->packets, "RX packets"); SYSCTL_ADD_ULONG(ctx, ring_list, OID_AUTO, "bytes", CTLFLAG_RD, &rx_ring->bytes, "RX bytes"); SYSCTL_ADD_ULONG(ctx, ring_list, OID_AUTO, "error", CTLFLAG_RD, &rx_ring->errors, "RX soft errors"); } } Index: projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/en_rx.c =================================================================== --- projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/en_rx.c (revision 279716) +++ projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/en_rx.c (revision 279717) @@ -1,901 +1,901 @@ /* * Copyright (c) 2007, 2014 Mellanox Technologies. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU * General Public License (GPL) Version 2, available from the file * COPYING in the main directory of this source tree, or the * OpenIB.org BSD license below: * * Redistribution and use in source and binary forms, with or * without modification, are permitted provided that the following * conditions are met: * * - Redistributions of source code must retain the above * copyright notice, this list of conditions and the following * disclaimer. * * - Redistributions in binary form must reproduce the above * copyright notice, this list of conditions and the following * disclaimer in the documentation and/or other materials * provided with the distribution. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. * */ #include "opt_inet.h" #include #include #include #include #include #include #include #ifdef CONFIG_NET_RX_BUSY_POLL #include #endif #include "mlx4_en.h" static void mlx4_en_init_rx_desc(struct mlx4_en_priv *priv, struct mlx4_en_rx_ring *ring, int index) { struct mlx4_en_rx_desc *rx_desc = ring->buf + ring->stride * index; int possible_frags; int i; /* Set size and memtype fields */ for (i = 0; i < priv->num_frags; i++) { rx_desc->data[i].byte_count = cpu_to_be32(priv->frag_info[i].frag_size); rx_desc->data[i].lkey = cpu_to_be32(priv->mdev->mr.key); } /* If the number of used fragments does not fill up the ring stride, * * remaining (unused) fragments must be padded with null address/size * * and a special memory key */ possible_frags = (ring->stride - sizeof(struct mlx4_en_rx_desc)) / DS_SIZE; for (i = priv->num_frags; i < possible_frags; i++) { rx_desc->data[i].byte_count = 0; rx_desc->data[i].lkey = cpu_to_be32(MLX4_EN_MEMTYPE_PAD); rx_desc->data[i].addr = 0; } } static int mlx4_en_alloc_buf(struct mlx4_en_priv *priv, struct mlx4_en_rx_desc *rx_desc, struct mbuf **mb_list, int i) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_en_frag_info *frag_info = &priv->frag_info[i]; struct mbuf *mb; dma_addr_t dma; if (i == 0) mb = m_getjcl(M_NOWAIT, MT_DATA, M_PKTHDR, frag_info->frag_size); else mb = m_getjcl(M_NOWAIT, MT_DATA, 0, frag_info->frag_size); if (mb == NULL) { priv->port_stats.rx_alloc_failed++; return -ENOMEM; } dma = pci_map_single(mdev->pdev, mb->m_data, frag_info->frag_size, PCI_DMA_FROMDEVICE); rx_desc->data[i].addr = cpu_to_be64(dma); mb_list[i] = mb; return 0; } static int mlx4_en_prepare_rx_desc(struct mlx4_en_priv *priv, struct mlx4_en_rx_ring *ring, int index) { struct mlx4_en_rx_desc *rx_desc = ring->buf + (index * ring->stride); struct mbuf **mb_list = ring->rx_info + (index << priv->log_rx_info); int i; for (i = 0; i < priv->num_frags; i++) if (mlx4_en_alloc_buf(priv, rx_desc, mb_list, i)) goto err; return 0; err: while (i--) m_free(mb_list[i]); return -ENOMEM; } static inline void mlx4_en_update_rx_prod_db(struct mlx4_en_rx_ring *ring) { *ring->wqres.db.db = cpu_to_be32(ring->prod & 0xffff); } static void mlx4_en_free_rx_desc(struct mlx4_en_priv *priv, struct mlx4_en_rx_ring *ring, int index) { struct mlx4_en_frag_info *frag_info; struct mlx4_en_dev *mdev = priv->mdev; struct mbuf **mb_list; struct mlx4_en_rx_desc *rx_desc = ring->buf + (index << ring->log_stride); dma_addr_t dma; int nr; mb_list = ring->rx_info + (index << priv->log_rx_info); for (nr = 0; nr < priv->num_frags; nr++) { en_dbg(DRV, priv, "Freeing fragment:%d\n", nr); frag_info = &priv->frag_info[nr]; dma = be64_to_cpu(rx_desc->data[nr].addr); #if BITS_PER_LONG == 64 en_dbg(DRV, priv, "Unmaping buffer at dma:0x%lx\n", (u64) dma); #elif BITS_PER_LONG == 32 en_dbg(DRV, priv, "Unmaping buffer at dma:0x%llx\n", (u64) dma); #endif pci_unmap_single(mdev->pdev, dma, frag_info->frag_size, PCI_DMA_FROMDEVICE); m_free(mb_list[nr]); } } static int mlx4_en_fill_rx_buffers(struct mlx4_en_priv *priv) { struct mlx4_en_rx_ring *ring; int ring_ind; int buf_ind; int new_size; int err; for (buf_ind = 0; buf_ind < priv->prof->rx_ring_size; buf_ind++) { for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) { ring = priv->rx_ring[ring_ind]; err = mlx4_en_prepare_rx_desc(priv, ring, ring->actual_size); if (err) { if (ring->actual_size == 0) { en_err(priv, "Failed to allocate " "enough rx buffers\n"); return -ENOMEM; } else { new_size = rounddown_pow_of_two(ring->actual_size); en_warn(priv, "Only %d buffers allocated " "reducing ring size to %d\n", ring->actual_size, new_size); goto reduce_rings; } } ring->actual_size++; ring->prod++; } } return 0; reduce_rings: for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) { ring = priv->rx_ring[ring_ind]; while (ring->actual_size > new_size) { ring->actual_size--; ring->prod--; mlx4_en_free_rx_desc(priv, ring, ring->actual_size); } } return 0; } static void mlx4_en_free_rx_buf(struct mlx4_en_priv *priv, struct mlx4_en_rx_ring *ring) { int index; en_dbg(DRV, priv, "Freeing Rx buf - cons:%d prod:%d\n", ring->cons, ring->prod); /* Unmap and free Rx buffers */ BUG_ON((u32) (ring->prod - ring->cons) > ring->actual_size); while (ring->cons != ring->prod) { index = ring->cons & ring->size_mask; en_dbg(DRV, priv, "Processing descriptor:%d\n", index); mlx4_en_free_rx_desc(priv, ring, index); ++ring->cons; } } #if MLX4_EN_MAX_RX_FRAGS == 3 static int frag_sizes[] = { FRAG_SZ0, FRAG_SZ1, FRAG_SZ2, }; #elif MLX4_EN_MAX_RX_FRAGS == 2 static int frag_sizes[] = { FRAG_SZ0, FRAG_SZ1, }; #else #error "Unknown MAX_RX_FRAGS" #endif void mlx4_en_calc_rx_buf(struct net_device *dev) { struct mlx4_en_priv *priv = netdev_priv(dev); int eff_mtu = dev->if_mtu + ETH_HLEN + VLAN_HLEN + ETH_FCS_LEN; int buf_size = 0; int i, frag; for (i = 0, frag = 0; buf_size < eff_mtu; frag++, i++) { /* * Allocate small to large but only as much as is needed for * the tail. */ while (i > 0 && eff_mtu - buf_size <= frag_sizes[i - 1]) i--; priv->frag_info[frag].frag_size = frag_sizes[i]; priv->frag_info[frag].frag_prefix_size = buf_size; buf_size += priv->frag_info[frag].frag_size; } priv->num_frags = frag; priv->rx_mb_size = eff_mtu; priv->log_rx_info = ROUNDUP_LOG2(priv->num_frags * sizeof(struct mbuf *)); en_dbg(DRV, priv, "Rx buffer scatter-list (effective-mtu:%d " "num_frags:%d):\n", eff_mtu, priv->num_frags); for (i = 0; i < priv->num_frags; i++) { en_dbg(DRV, priv, " frag:%d - size:%d prefix:%d\n", i, priv->frag_info[i].frag_size, priv->frag_info[i].frag_prefix_size); } } int mlx4_en_create_rx_ring(struct mlx4_en_priv *priv, struct mlx4_en_rx_ring **pring, u32 size, int node) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_en_rx_ring *ring; int err = -ENOMEM; int tmp; ring = kzalloc(sizeof(struct mlx4_en_rx_ring), GFP_KERNEL); if (!ring) { en_err(priv, "Failed to allocate RX ring structure\n"); return -ENOMEM; } ring->prod = 0; ring->cons = 0; ring->size = size; ring->size_mask = size - 1; ring->stride = roundup_pow_of_two(sizeof(struct mlx4_en_rx_desc) + DS_SIZE * MLX4_EN_MAX_RX_FRAGS); ring->log_stride = ffs(ring->stride) - 1; ring->buf_size = ring->size * ring->stride + TXBB_SIZE; tmp = size * roundup_pow_of_two(MLX4_EN_MAX_RX_FRAGS * sizeof(struct mbuf *)); ring->rx_info = kmalloc(tmp, GFP_KERNEL); if (!ring->rx_info) { err = -ENOMEM; goto err_ring; } en_dbg(DRV, priv, "Allocated rx_info ring at addr:%p size:%d\n", ring->rx_info, tmp); err = mlx4_alloc_hwq_res(mdev->dev, &ring->wqres, ring->buf_size, 2 * PAGE_SIZE); if (err) goto err_info; err = mlx4_en_map_buffer(&ring->wqres.buf); if (err) { en_err(priv, "Failed to map RX buffer\n"); goto err_hwq; } ring->buf = ring->wqres.buf.direct.buf; *pring = ring; return 0; err_hwq: mlx4_free_hwq_res(mdev->dev, &ring->wqres, ring->buf_size); err_info: vfree(ring->rx_info); err_ring: kfree(ring); return err; } int mlx4_en_activate_rx_rings(struct mlx4_en_priv *priv) { struct mlx4_en_rx_ring *ring; int i; int ring_ind; int err; int stride = roundup_pow_of_two(sizeof(struct mlx4_en_rx_desc) + DS_SIZE * priv->num_frags); for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) { ring = priv->rx_ring[ring_ind]; ring->prod = 0; ring->cons = 0; ring->actual_size = 0; ring->cqn = priv->rx_cq[ring_ind]->mcq.cqn; ring->rx_alloc_order = priv->rx_alloc_order; ring->rx_alloc_size = priv->rx_alloc_size; ring->rx_buf_size = priv->rx_buf_size; ring->rx_mb_size = priv->rx_mb_size; ring->stride = stride; if (ring->stride <= TXBB_SIZE) ring->buf += TXBB_SIZE; ring->log_stride = ffs(ring->stride) - 1; ring->buf_size = ring->size * ring->stride; memset(ring->buf, 0, ring->buf_size); mlx4_en_update_rx_prod_db(ring); /* Initialize all descriptors */ for (i = 0; i < ring->size; i++) mlx4_en_init_rx_desc(priv, ring, i); #ifdef INET /* Configure lro mngr */ if (priv->dev->if_capenable & IFCAP_LRO) { if (tcp_lro_init(&ring->lro)) priv->dev->if_capenable &= ~IFCAP_LRO; else ring->lro.ifp = priv->dev; } #endif } err = mlx4_en_fill_rx_buffers(priv); if (err) goto err_buffers; for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) { ring = priv->rx_ring[ring_ind]; ring->size_mask = ring->actual_size - 1; mlx4_en_update_rx_prod_db(ring); } return 0; err_buffers: for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) mlx4_en_free_rx_buf(priv, priv->rx_ring[ring_ind]); ring_ind = priv->rx_ring_num - 1; while (ring_ind >= 0) { ring = priv->rx_ring[ring_ind]; if (ring->stride <= TXBB_SIZE) ring->buf -= TXBB_SIZE; ring_ind--; } return err; } void mlx4_en_destroy_rx_ring(struct mlx4_en_priv *priv, struct mlx4_en_rx_ring **pring, u32 size, u16 stride) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_en_rx_ring *ring = *pring; mlx4_en_unmap_buffer(&ring->wqres.buf); mlx4_free_hwq_res(mdev->dev, &ring->wqres, size * stride + TXBB_SIZE); vfree(ring->rx_info); kfree(ring); *pring = NULL; #ifdef CONFIG_RFS_ACCEL mlx4_en_cleanup_filters(priv, ring); #endif } void mlx4_en_deactivate_rx_ring(struct mlx4_en_priv *priv, struct mlx4_en_rx_ring *ring) { #ifdef INET tcp_lro_free(&ring->lro); #endif mlx4_en_free_rx_buf(priv, ring); if (ring->stride <= TXBB_SIZE) ring->buf -= TXBB_SIZE; } static void validate_loopback(struct mlx4_en_priv *priv, struct mbuf *mb) { int i; int offset = ETHER_HDR_LEN; for (i = 0; i < MLX4_LOOPBACK_TEST_PAYLOAD; i++, offset++) { if (*(mb->m_data + offset) != (unsigned char) (i & 0xff)) goto out_loopback; } /* Loopback found */ priv->loopback_ok = 1; out_loopback: m_freem(mb); } static inline int invalid_cqe(struct mlx4_en_priv *priv, struct mlx4_cqe *cqe) { /* Drop packet on bad receive or bad checksum */ if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) == MLX4_CQE_OPCODE_ERROR)) { en_err(priv, "CQE completed in error - vendor syndrom:%d syndrom:%d\n", ((struct mlx4_err_cqe *)cqe)->vendor_err_syndrome, ((struct mlx4_err_cqe *)cqe)->syndrome); return 1; } if (unlikely(cqe->badfcs_enc & MLX4_CQE_BAD_FCS)) { en_dbg(RX_ERR, priv, "Accepted frame with bad FCS\n"); return 1; } return 0; } /* Unmap a completed descriptor and free unused pages */ static int mlx4_en_complete_rx_desc(struct mlx4_en_priv *priv, struct mlx4_en_rx_desc *rx_desc, struct mbuf **mb_list, int length) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_en_frag_info *frag_info; dma_addr_t dma; struct mbuf *mb; int nr; mb = mb_list[0]; mb->m_pkthdr.len = length; /* Collect used fragments while replacing them in the HW descirptors */ for (nr = 0; nr < priv->num_frags; nr++) { frag_info = &priv->frag_info[nr]; if (length <= frag_info->frag_prefix_size) break; if (nr) mb->m_next = mb_list[nr]; mb = mb_list[nr]; mb->m_len = frag_info->frag_size; dma = be64_to_cpu(rx_desc->data[nr].addr); /* Allocate a replacement page */ if (mlx4_en_alloc_buf(priv, rx_desc, mb_list, nr)) goto fail; /* Unmap buffer */ - pci_unmap_single(mdev->pdev, dma, frag_info[nr].frag_size, + pci_unmap_single(mdev->pdev, dma, frag_info->frag_size, PCI_DMA_FROMDEVICE); } /* Adjust size of last fragment to match actual length */ mb->m_len = length - priv->frag_info[nr - 1].frag_prefix_size; mb->m_next = NULL; return 0; fail: /* Drop all accumulated fragments (which have already been replaced in * the descriptor) of this packet; remaining fragments are reused... */ while (nr > 0) { nr--; m_free(mb_list[nr]); } return -ENOMEM; } static struct mbuf *mlx4_en_rx_mb(struct mlx4_en_priv *priv, struct mlx4_en_rx_desc *rx_desc, struct mbuf **mb_list, unsigned int length) { struct mbuf *mb; mb = mb_list[0]; /* Move relevant fragments to mb */ if (unlikely(mlx4_en_complete_rx_desc(priv, rx_desc, mb_list, length))) return NULL; return mb; } /* For cpu arch with cache line of 64B the performance is better when cqe size==64B * To enlarge cqe size from 32B to 64B --> 32B of garbage (i.e. 0xccccccc) * was added in the beginning of each cqe (the real data is in the corresponding 32B). * The following calc ensures that when factor==1, it means we are alligned to 64B * and we get the real cqe data*/ #define CQE_FACTOR_INDEX(index, factor) ((index << factor) + factor) int mlx4_en_process_rx_cq(struct net_device *dev, struct mlx4_en_cq *cq, int budget) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_cqe *cqe; struct mlx4_en_rx_ring *ring = priv->rx_ring[cq->ring]; struct mbuf **mb_list; struct mlx4_en_rx_desc *rx_desc; struct mbuf *mb; struct mlx4_cq *mcq = &cq->mcq; struct mlx4_cqe *buf = cq->buf; #ifdef INET struct lro_entry *queued; #endif int index; unsigned int length; int polled = 0; u32 cons_index = mcq->cons_index; u32 size_mask = ring->size_mask; int size = cq->size; int factor = priv->cqe_factor; if (!priv->port_up) return 0; /* We assume a 1:1 mapping between CQEs and Rx descriptors, so Rx * descriptor offset can be deducted from the CQE index instead of * reading 'cqe->index' */ index = cons_index & size_mask; cqe = &buf[CQE_FACTOR_INDEX(index, factor)]; /* Process all completed CQEs */ while (XNOR(cqe->owner_sr_opcode & MLX4_CQE_OWNER_MASK, cons_index & size)) { mb_list = ring->rx_info + (index << priv->log_rx_info); rx_desc = ring->buf + (index << ring->log_stride); /* * make sure we read the CQE after we read the ownership bit */ rmb(); if (invalid_cqe(priv, cqe)) { goto next; } /* * Packet is OK - process it. */ length = be32_to_cpu(cqe->byte_cnt); length -= ring->fcs_del; mb = mlx4_en_rx_mb(priv, rx_desc, mb_list, length); if (!mb) { ring->errors++; goto next; } ring->bytes += length; ring->packets++; if (unlikely(priv->validate_loopback)) { validate_loopback(priv, mb); goto next; } mb->m_pkthdr.flowid = cq->ring; M_HASHTYPE_SET(mb, M_HASHTYPE_OPAQUE); mb->m_pkthdr.rcvif = dev; if (be32_to_cpu(cqe->vlan_my_qpn) & MLX4_CQE_VLAN_PRESENT_MASK) { mb->m_pkthdr.ether_vtag = be16_to_cpu(cqe->sl_vid); mb->m_flags |= M_VLANTAG; } if (likely(dev->if_capabilities & IFCAP_RXCSUM) && (cqe->status & cpu_to_be16(MLX4_CQE_STATUS_IPOK)) && (cqe->checksum == cpu_to_be16(0xffff))) { priv->port_stats.rx_chksum_good++; mb->m_pkthdr.csum_flags = CSUM_IP_CHECKED | CSUM_IP_VALID | CSUM_DATA_VALID | CSUM_PSEUDO_HDR; mb->m_pkthdr.csum_data = htons(0xffff); /* This packet is eligible for LRO if it is: * - DIX Ethernet (type interpretation) * - TCP/IP (v4) * - without IP options * - not an IP fragment */ #ifdef INET if (mlx4_en_can_lro(cqe->status) && (dev->if_capenable & IFCAP_LRO)) { if (ring->lro.lro_cnt != 0 && tcp_lro_rx(&ring->lro, mb, 0) == 0) goto next; } #endif /* LRO not possible, complete processing here */ INC_PERF_COUNTER(priv->pstats.lro_misses); } else { mb->m_pkthdr.csum_flags = 0; priv->port_stats.rx_chksum_none++; } /* Push it up the stack */ dev->if_input(dev, mb); next: ++cons_index; index = cons_index & size_mask; cqe = &buf[CQE_FACTOR_INDEX(index, factor)]; if (++polled == budget) goto out; } /* Flush all pending IP reassembly sessions */ out: #ifdef INET while ((queued = SLIST_FIRST(&ring->lro.lro_active)) != NULL) { SLIST_REMOVE_HEAD(&ring->lro.lro_active, next); tcp_lro_flush(&ring->lro, queued); } #endif AVG_PERF_COUNTER(priv->pstats.rx_coal_avg, polled); mcq->cons_index = cons_index; mlx4_cq_set_ci(mcq); wmb(); /* ensure HW sees CQ consumer before we post new buffers */ ring->cons = mcq->cons_index; ring->prod += polled; /* Polled descriptors were realocated in place */ mlx4_en_update_rx_prod_db(ring); return polled; } /* Rx CQ polling - called by NAPI */ static int mlx4_en_poll_rx_cq(struct mlx4_en_cq *cq, int budget) { struct net_device *dev = cq->dev; int done; done = mlx4_en_process_rx_cq(dev, cq, budget); cq->tot_rx += done; return done; } void mlx4_en_rx_irq(struct mlx4_cq *mcq) { struct mlx4_en_cq *cq = container_of(mcq, struct mlx4_en_cq, mcq); struct mlx4_en_priv *priv = netdev_priv(cq->dev); int done; // Shoot one within the irq context // Because there is no NAPI in freeBSD done = mlx4_en_poll_rx_cq(cq, MLX4_EN_RX_BUDGET); if (priv->port_up && (done == MLX4_EN_RX_BUDGET) ) { taskqueue_enqueue(cq->tq, &cq->cq_task); } else { mlx4_en_arm_cq(priv, cq); } } void mlx4_en_rx_que(void *context, int pending) { struct mlx4_en_cq *cq; cq = context; while (mlx4_en_poll_rx_cq(cq, MLX4_EN_RX_BUDGET) == MLX4_EN_RX_BUDGET); mlx4_en_arm_cq(cq->dev->if_softc, cq); } /* RSS related functions */ static int mlx4_en_config_rss_qp(struct mlx4_en_priv *priv, int qpn, struct mlx4_en_rx_ring *ring, enum mlx4_qp_state *state, struct mlx4_qp *qp) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_qp_context *context; int err = 0; context = kmalloc(sizeof *context , GFP_KERNEL); if (!context) { en_err(priv, "Failed to allocate qp context\n"); return -ENOMEM; } err = mlx4_qp_alloc(mdev->dev, qpn, qp); if (err) { en_err(priv, "Failed to allocate qp #%x\n", qpn); goto out; } qp->event = mlx4_en_sqp_event; memset(context, 0, sizeof *context); mlx4_en_fill_qp_context(priv, ring->actual_size, ring->stride, 0, 0, qpn, ring->cqn, -1, context); context->db_rec_addr = cpu_to_be64(ring->wqres.db.dma); /* Cancel FCS removal if FW allows */ if (mdev->dev->caps.flags & MLX4_DEV_CAP_FLAG_FCS_KEEP) { context->param3 |= cpu_to_be32(1 << 29); ring->fcs_del = ETH_FCS_LEN; } else ring->fcs_del = 0; err = mlx4_qp_to_ready(mdev->dev, &ring->wqres.mtt, context, qp, state); if (err) { mlx4_qp_remove(mdev->dev, qp); mlx4_qp_free(mdev->dev, qp); } mlx4_en_update_rx_prod_db(ring); out: kfree(context); return err; } int mlx4_en_create_drop_qp(struct mlx4_en_priv *priv) { int err; u32 qpn; err = mlx4_qp_reserve_range(priv->mdev->dev, 1, 1, &qpn, 0); if (err) { en_err(priv, "Failed reserving drop qpn\n"); return err; } err = mlx4_qp_alloc(priv->mdev->dev, qpn, &priv->drop_qp); if (err) { en_err(priv, "Failed allocating drop qp\n"); mlx4_qp_release_range(priv->mdev->dev, qpn, 1); return err; } return 0; } void mlx4_en_destroy_drop_qp(struct mlx4_en_priv *priv) { u32 qpn; qpn = priv->drop_qp.qpn; mlx4_qp_remove(priv->mdev->dev, &priv->drop_qp); mlx4_qp_free(priv->mdev->dev, &priv->drop_qp); mlx4_qp_release_range(priv->mdev->dev, qpn, 1); } /* Allocate rx qp's and configure them according to rss map */ int mlx4_en_config_rss_steer(struct mlx4_en_priv *priv) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_en_rss_map *rss_map = &priv->rss_map; struct mlx4_qp_context context; struct mlx4_rss_context *rss_context; int rss_rings; void *ptr; u8 rss_mask = (MLX4_RSS_IPV4 | MLX4_RSS_TCP_IPV4 | MLX4_RSS_IPV6 | MLX4_RSS_TCP_IPV6); int i; int err = 0; int good_qps = 0; static const u32 rsskey[10] = { 0xD181C62C, 0xF7F4DB5B, 0x1983A2FC, 0x943E1ADB, 0xD9389E6B, 0xD1039C2C, 0xA74499AD, 0x593D56D9, 0xF3253C06, 0x2ADC1FFC}; en_dbg(DRV, priv, "Configuring rss steering\n"); err = mlx4_qp_reserve_range(mdev->dev, priv->rx_ring_num, priv->rx_ring_num, &rss_map->base_qpn, 0); if (err) { en_err(priv, "Failed reserving %d qps\n", priv->rx_ring_num); return err; } for (i = 0; i < priv->rx_ring_num; i++) { priv->rx_ring[i]->qpn = rss_map->base_qpn + i; err = mlx4_en_config_rss_qp(priv, priv->rx_ring[i]->qpn, priv->rx_ring[i], &rss_map->state[i], &rss_map->qps[i]); if (err) goto rss_err; ++good_qps; } /* Configure RSS indirection qp */ err = mlx4_qp_alloc(mdev->dev, priv->base_qpn, &rss_map->indir_qp); if (err) { en_err(priv, "Failed to allocate RSS indirection QP\n"); goto rss_err; } rss_map->indir_qp.event = mlx4_en_sqp_event; mlx4_en_fill_qp_context(priv, 0, 0, 0, 1, priv->base_qpn, priv->rx_ring[0]->cqn, -1, &context); if (!priv->prof->rss_rings || priv->prof->rss_rings > priv->rx_ring_num) rss_rings = priv->rx_ring_num; else rss_rings = priv->prof->rss_rings; ptr = ((void *) &context) + offsetof(struct mlx4_qp_context, pri_path) + MLX4_RSS_OFFSET_IN_QPC_PRI_PATH; rss_context = ptr; rss_context->base_qpn = cpu_to_be32(ilog2(rss_rings) << 24 | (rss_map->base_qpn)); rss_context->default_qpn = cpu_to_be32(rss_map->base_qpn); if (priv->mdev->profile.udp_rss) { rss_mask |= MLX4_RSS_UDP_IPV4 | MLX4_RSS_UDP_IPV6; rss_context->base_qpn_udp = rss_context->default_qpn; } rss_context->flags = rss_mask; rss_context->hash_fn = MLX4_RSS_HASH_TOP; for (i = 0; i < 10; i++) rss_context->rss_key[i] = cpu_to_be32(rsskey[i]); err = mlx4_qp_to_ready(mdev->dev, &priv->res.mtt, &context, &rss_map->indir_qp, &rss_map->indir_state); if (err) goto indir_err; return 0; indir_err: mlx4_qp_modify(mdev->dev, NULL, rss_map->indir_state, MLX4_QP_STATE_RST, NULL, 0, 0, &rss_map->indir_qp); mlx4_qp_remove(mdev->dev, &rss_map->indir_qp); mlx4_qp_free(mdev->dev, &rss_map->indir_qp); rss_err: for (i = 0; i < good_qps; i++) { mlx4_qp_modify(mdev->dev, NULL, rss_map->state[i], MLX4_QP_STATE_RST, NULL, 0, 0, &rss_map->qps[i]); mlx4_qp_remove(mdev->dev, &rss_map->qps[i]); mlx4_qp_free(mdev->dev, &rss_map->qps[i]); } mlx4_qp_release_range(mdev->dev, rss_map->base_qpn, priv->rx_ring_num); return err; } void mlx4_en_release_rss_steer(struct mlx4_en_priv *priv) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_en_rss_map *rss_map = &priv->rss_map; int i; mlx4_qp_modify(mdev->dev, NULL, rss_map->indir_state, MLX4_QP_STATE_RST, NULL, 0, 0, &rss_map->indir_qp); mlx4_qp_remove(mdev->dev, &rss_map->indir_qp); mlx4_qp_free(mdev->dev, &rss_map->indir_qp); for (i = 0; i < priv->rx_ring_num; i++) { mlx4_qp_modify(mdev->dev, NULL, rss_map->state[i], MLX4_QP_STATE_RST, NULL, 0, 0, &rss_map->qps[i]); mlx4_qp_remove(mdev->dev, &rss_map->qps[i]); mlx4_qp_free(mdev->dev, &rss_map->qps[i]); } mlx4_qp_release_range(mdev->dev, rss_map->base_qpn, priv->rx_ring_num); } Index: projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/en_tx.c =================================================================== --- projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/en_tx.c (revision 279716) +++ projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/en_tx.c (revision 279717) @@ -1,1111 +1,1111 @@ /* * Copyright (c) 2007, 2014 Mellanox Technologies. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU * General Public License (GPL) Version 2, available from the file * COPYING in the main directory of this source tree, or the * OpenIB.org BSD license below: * * Redistribution and use in source and binary forms, with or * without modification, are permitted provided that the following * conditions are met: * * - Redistributions of source code must retain the above * copyright notice, this list of conditions and the following * disclaimer. * * - Redistributions in binary form must reproduce the above * copyright notice, this list of conditions and the following * disclaimer in the documentation and/or other materials * provided with the distribution. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. * */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "mlx4_en.h" #include "utils.h" enum { MAX_INLINE = 104, /* 128 - 16 - 4 - 4 */ MAX_BF = 256, MIN_PKT_LEN = 17, }; static int inline_thold __read_mostly = MAX_INLINE; module_param_named(inline_thold, inline_thold, uint, 0444); MODULE_PARM_DESC(inline_thold, "threshold for using inline data"); int mlx4_en_create_tx_ring(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring **pring, u32 size, u16 stride, int node, int queue_idx) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_en_tx_ring *ring; int tmp; int err; ring = kzalloc_node(sizeof(struct mlx4_en_tx_ring), GFP_KERNEL, node); if (!ring) { ring = kzalloc(sizeof(struct mlx4_en_tx_ring), GFP_KERNEL); if (!ring) { en_err(priv, "Failed allocating TX ring\n"); return -ENOMEM; } } ring->size = size; ring->size_mask = size - 1; ring->stride = stride; ring->full_size = ring->size - HEADROOM - MAX_DESC_TXBBS; ring->inline_thold = min(inline_thold, MAX_INLINE); mtx_init(&ring->tx_lock.m, "mlx4 tx", NULL, MTX_DEF); mtx_init(&ring->comp_lock.m, "mlx4 comp", NULL, MTX_DEF); /* Allocate the buf ring */ ring->br = buf_ring_alloc(MLX4_EN_DEF_TX_QUEUE_SIZE, M_DEVBUF, M_WAITOK, &ring->tx_lock.m); if (ring->br == NULL) { en_err(priv, "Failed allocating tx_info ring\n"); return -ENOMEM; } tmp = size * sizeof(struct mlx4_en_tx_info); ring->tx_info = vmalloc_node(tmp, node); if (!ring->tx_info) { ring->tx_info = vmalloc(tmp); if (!ring->tx_info) { err = -ENOMEM; goto err_ring; } } en_dbg(DRV, priv, "Allocated tx_info ring at addr:%p size:%d\n", ring->tx_info, tmp); ring->bounce_buf = kmalloc_node(MAX_DESC_SIZE, GFP_KERNEL, node); if (!ring->bounce_buf) { ring->bounce_buf = kmalloc(MAX_DESC_SIZE, GFP_KERNEL); if (!ring->bounce_buf) { err = -ENOMEM; goto err_info; } } ring->buf_size = ALIGN(size * ring->stride, MLX4_EN_PAGE_SIZE); /* Allocate HW buffers on provided NUMA node */ err = mlx4_alloc_hwq_res(mdev->dev, &ring->wqres, ring->buf_size, 2 * PAGE_SIZE); if (err) { en_err(priv, "Failed allocating hwq resources\n"); goto err_bounce; } err = mlx4_en_map_buffer(&ring->wqres.buf); if (err) { en_err(priv, "Failed to map TX buffer\n"); goto err_hwq_res; } ring->buf = ring->wqres.buf.direct.buf; en_dbg(DRV, priv, "Allocated TX ring (addr:%p) - buf:%p size:%d " "buf_size:%d dma:%llx\n", ring, ring->buf, ring->size, ring->buf_size, (unsigned long long) ring->wqres.buf.direct.map); err = mlx4_qp_reserve_range(mdev->dev, 1, 1, &ring->qpn, MLX4_RESERVE_BF_QP); if (err) { en_err(priv, "failed reserving qp for TX ring\n"); goto err_map; } err = mlx4_qp_alloc(mdev->dev, ring->qpn, &ring->qp); if (err) { en_err(priv, "Failed allocating qp %d\n", ring->qpn); goto err_reserve; } ring->qp.event = mlx4_en_sqp_event; err = mlx4_bf_alloc(mdev->dev, &ring->bf, node); if (err) { en_dbg(DRV, priv, "working without blueflame (%d)", err); ring->bf.uar = &mdev->priv_uar; ring->bf.uar->map = mdev->uar_map; ring->bf_enabled = false; } else ring->bf_enabled = true; ring->queue_index = queue_idx; if (queue_idx < priv->num_tx_rings_p_up ) CPU_SET(queue_idx, &ring->affinity_mask); *pring = ring; return 0; err_reserve: mlx4_qp_release_range(mdev->dev, ring->qpn, 1); err_map: mlx4_en_unmap_buffer(&ring->wqres.buf); err_hwq_res: mlx4_free_hwq_res(mdev->dev, &ring->wqres, ring->buf_size); err_bounce: kfree(ring->bounce_buf); err_info: vfree(ring->tx_info); err_ring: buf_ring_free(ring->br, M_DEVBUF); kfree(ring); return err; } void mlx4_en_destroy_tx_ring(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring **pring) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_en_tx_ring *ring = *pring; en_dbg(DRV, priv, "Destroying tx ring, qpn: %d\n", ring->qpn); buf_ring_free(ring->br, M_DEVBUF); if (ring->bf_enabled) mlx4_bf_free(mdev->dev, &ring->bf); mlx4_qp_remove(mdev->dev, &ring->qp); mlx4_qp_free(mdev->dev, &ring->qp); mlx4_qp_release_range(priv->mdev->dev, ring->qpn, 1); mlx4_en_unmap_buffer(&ring->wqres.buf); mlx4_free_hwq_res(mdev->dev, &ring->wqres, ring->buf_size); kfree(ring->bounce_buf); vfree(ring->tx_info); mtx_destroy(&ring->tx_lock.m); mtx_destroy(&ring->comp_lock.m); kfree(ring); *pring = NULL; } int mlx4_en_activate_tx_ring(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring *ring, int cq, int user_prio) { struct mlx4_en_dev *mdev = priv->mdev; int err; ring->cqn = cq; ring->prod = 0; ring->cons = 0xffffffff; ring->last_nr_txbb = 1; ring->poll_cnt = 0; ring->blocked = 0; memset(ring->tx_info, 0, ring->size * sizeof(struct mlx4_en_tx_info)); memset(ring->buf, 0, ring->buf_size); ring->qp_state = MLX4_QP_STATE_RST; ring->doorbell_qpn = ring->qp.qpn << 8; mlx4_en_fill_qp_context(priv, ring->size, ring->stride, 1, 0, ring->qpn, ring->cqn, user_prio, &ring->context); if (ring->bf_enabled) ring->context.usr_page = cpu_to_be32(ring->bf.uar->index); err = mlx4_qp_to_ready(mdev->dev, &ring->wqres.mtt, &ring->context, &ring->qp, &ring->qp_state); return err; } void mlx4_en_deactivate_tx_ring(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring *ring) { struct mlx4_en_dev *mdev = priv->mdev; mlx4_qp_modify(mdev->dev, NULL, ring->qp_state, MLX4_QP_STATE_RST, NULL, 0, 0, &ring->qp); } static void mlx4_en_stamp_wqe(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring *ring, int index, u8 owner) { struct mlx4_en_tx_info *tx_info = &ring->tx_info[index]; struct mlx4_en_tx_desc *tx_desc = ring->buf + index * TXBB_SIZE; void *end = ring->buf + ring->buf_size; __be32 *ptr = (__be32 *)tx_desc; __be32 stamp = cpu_to_be32(STAMP_VAL | (!!owner << STAMP_SHIFT)); int i; /* Optimize the common case when there are no wraparounds */ if (likely((void *)tx_desc + tx_info->nr_txbb * TXBB_SIZE <= end)) /* Stamp the freed descriptor */ for (i = 0; i < tx_info->nr_txbb * TXBB_SIZE; i += STAMP_STRIDE) { *ptr = stamp; ptr += STAMP_DWORDS; } else /* Stamp the freed descriptor */ for (i = 0; i < tx_info->nr_txbb * TXBB_SIZE; i += STAMP_STRIDE) { *ptr = stamp; ptr += STAMP_DWORDS; if ((void *)ptr >= end) { ptr = ring->buf; stamp ^= cpu_to_be32(0x80000000); } } } static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring *ring, int index, u8 owner, u64 timestamp) { struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_en_tx_info *tx_info = &ring->tx_info[index]; struct mlx4_en_tx_desc *tx_desc = ring->buf + index * TXBB_SIZE; struct mlx4_wqe_data_seg *data = (void *) tx_desc + tx_info->data_offset; struct mbuf *mb = tx_info->mb; void *end = ring->buf + ring->buf_size; int frags = tx_info->nr_segs;; int i; /* Optimize the common case when there are no wraparounds */ if (likely((void *) tx_desc + tx_info->nr_txbb * TXBB_SIZE <= end)) { if (!tx_info->inl) { if (tx_info->linear) { dma_unmap_single(priv->ddev, (dma_addr_t) be64_to_cpu(data->addr), be32_to_cpu(data->byte_count), PCI_DMA_TODEVICE); ++data; } for (i = 0; i < frags; i++) { pci_unmap_single(mdev->pdev, (dma_addr_t) be64_to_cpu(data[i].addr), data[i].byte_count, PCI_DMA_TODEVICE); } } } else { if (!tx_info->inl) { if ((void *) data >= end) { data = ring->buf + ((void *)data - end); } if (tx_info->linear) { dma_unmap_single(priv->ddev, (dma_addr_t) be64_to_cpu(data->addr), be32_to_cpu(data->byte_count), PCI_DMA_TODEVICE); ++data; } for (i = 0; i < frags; i++) { /* Check for wraparound before unmapping */ if ((void *) data >= end) data = ring->buf; pci_unmap_single(mdev->pdev, (dma_addr_t) be64_to_cpu(data->addr), data->byte_count, PCI_DMA_TODEVICE); ++data; } } } /* Send a copy of the frame to the BPF listener */ if (priv->dev && priv->dev->if_bpf) ETHER_BPF_MTAP(priv->dev, mb); m_freem(mb); return tx_info->nr_txbb; } int mlx4_en_free_tx_buf(struct net_device *dev, struct mlx4_en_tx_ring *ring) { struct mlx4_en_priv *priv = netdev_priv(dev); int cnt = 0; /* Skip last polled descriptor */ ring->cons += ring->last_nr_txbb; en_dbg(DRV, priv, "Freeing Tx buf - cons:0x%x prod:0x%x\n", ring->cons, ring->prod); if ((u32) (ring->prod - ring->cons) > ring->size) { en_warn(priv, "Tx consumer passed producer!\n"); return 0; } while (ring->cons != ring->prod) { ring->last_nr_txbb = mlx4_en_free_tx_desc(priv, ring, ring->cons & ring->size_mask, !!(ring->cons & ring->size), 0); ring->cons += ring->last_nr_txbb; cnt++; } if (cnt) en_dbg(DRV, priv, "Freed %d uncompleted tx descriptors\n", cnt); return cnt; } static int mlx4_en_process_tx_cq(struct net_device *dev, struct mlx4_en_cq *cq) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_cq *mcq = &cq->mcq; struct mlx4_en_tx_ring *ring = priv->tx_ring[cq->ring]; struct mlx4_cqe *cqe; u16 index; u16 new_index, ring_index, stamp_index; u32 txbbs_skipped = 0; u32 txbbs_stamp = 0; u32 cons_index = mcq->cons_index; int size = cq->size; u32 size_mask = ring->size_mask; struct mlx4_cqe *buf = cq->buf; u32 packets = 0; u32 bytes = 0; int factor = priv->cqe_factor; u64 timestamp = 0; int done = 0; if (!priv->port_up) return 0; index = cons_index & size_mask; cqe = &buf[(index << factor) + factor]; ring_index = ring->cons & size_mask; stamp_index = ring_index; /* Process all completed CQEs */ while (XNOR(cqe->owner_sr_opcode & MLX4_CQE_OWNER_MASK, cons_index & size)) { /* * make sure we read the CQE after we read the * ownership bit */ rmb(); if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) == MLX4_CQE_OPCODE_ERROR)) { en_err(priv, "CQE completed in error - vendor syndrom: 0x%x syndrom: 0x%x\n", ((struct mlx4_err_cqe *)cqe)-> vendor_err_syndrome, ((struct mlx4_err_cqe *)cqe)->syndrome); } /* Skip over last polled CQE */ new_index = be16_to_cpu(cqe->wqe_index) & size_mask; do { txbbs_skipped += ring->last_nr_txbb; ring_index = (ring_index + ring->last_nr_txbb) & size_mask; /* free next descriptor */ ring->last_nr_txbb = mlx4_en_free_tx_desc( priv, ring, ring_index, !!((ring->cons + txbbs_skipped) & ring->size), timestamp); mlx4_en_stamp_wqe(priv, ring, stamp_index, !!((ring->cons + txbbs_stamp) & ring->size)); stamp_index = ring_index; txbbs_stamp = txbbs_skipped; packets++; bytes += ring->tx_info[ring_index].nr_bytes; } while (ring_index != new_index); ++cons_index; index = cons_index & size_mask; cqe = &buf[(index << factor) + factor]; } /* * To prevent CQ overflow we first update CQ consumer and only then * the ring consumer. */ mcq->cons_index = cons_index; mlx4_cq_set_ci(mcq); wmb(); ring->cons += txbbs_skipped; /* Wakeup Tx queue if it was stopped and ring is not full */ if (unlikely(ring->blocked) && (ring->prod - ring->cons) <= ring->full_size) { ring->blocked = 0; if (atomic_fetchadd_int(&priv->blocked, -1) == 1) atomic_clear_int(&dev->if_drv_flags ,IFF_DRV_OACTIVE); ring->wake_queue++; priv->port_stats.wake_queue++; } return done; } void mlx4_en_tx_irq(struct mlx4_cq *mcq) { struct mlx4_en_cq *cq = container_of(mcq, struct mlx4_en_cq, mcq); struct mlx4_en_priv *priv = netdev_priv(cq->dev); struct mlx4_en_tx_ring *ring = priv->tx_ring[cq->ring]; if (!spin_trylock(&ring->comp_lock)) return; mlx4_en_process_tx_cq(cq->dev, cq); mod_timer(&cq->timer, jiffies + 1); spin_unlock(&ring->comp_lock); } void mlx4_en_poll_tx_cq(unsigned long data) { struct mlx4_en_cq *cq = (struct mlx4_en_cq *) data; struct mlx4_en_priv *priv = netdev_priv(cq->dev); struct mlx4_en_tx_ring *ring = priv->tx_ring[cq->ring]; u32 inflight; INC_PERF_COUNTER(priv->pstats.tx_poll); if (!spin_trylock(&ring->comp_lock)) { mod_timer(&cq->timer, jiffies + MLX4_EN_TX_POLL_TIMEOUT); return; } mlx4_en_process_tx_cq(cq->dev, cq); inflight = (u32) (ring->prod - ring->cons - ring->last_nr_txbb); /* If there are still packets in flight and the timer has not already * been scheduled by the Tx routine then schedule it here to guarantee * completion processing of these packets */ if (inflight && priv->port_up) mod_timer(&cq->timer, jiffies + MLX4_EN_TX_POLL_TIMEOUT); spin_unlock(&ring->comp_lock); } static struct mlx4_en_tx_desc *mlx4_en_bounce_to_desc(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring *ring, u32 index, unsigned int desc_size) { u32 copy = (ring->size - index) * TXBB_SIZE; int i; for (i = desc_size - copy - 4; i >= 0; i -= 4) { if ((i & (TXBB_SIZE - 1)) == 0) wmb(); *((u32 *) (ring->buf + i)) = *((u32 *) (ring->bounce_buf + copy + i)); } for (i = copy - 4; i >= 4 ; i -= 4) { if ((i & (TXBB_SIZE - 1)) == 0) wmb(); *((u32 *) (ring->buf + index * TXBB_SIZE + i)) = *((u32 *) (ring->bounce_buf + i)); } /* Return real descriptor location */ return ring->buf + index * TXBB_SIZE; } static inline void mlx4_en_xmit_poll(struct mlx4_en_priv *priv, int tx_ind) { struct mlx4_en_cq *cq = priv->tx_cq[tx_ind]; struct mlx4_en_tx_ring *ring = priv->tx_ring[tx_ind]; /* If we don't have a pending timer, set one up to catch our recent post in case the interface becomes idle */ if (!timer_pending(&cq->timer)) mod_timer(&cq->timer, jiffies + MLX4_EN_TX_POLL_TIMEOUT); /* Poll the CQ every mlx4_en_TX_MODER_POLL packets */ if ((++ring->poll_cnt & (MLX4_EN_TX_POLL_MODER - 1)) == 0) if (spin_trylock(&ring->comp_lock)) { mlx4_en_process_tx_cq(priv->dev, cq); spin_unlock(&ring->comp_lock); } } static int is_inline(struct mbuf *mb, int thold) { if (thold && mb->m_pkthdr.len <= thold && (mb->m_pkthdr.csum_flags & CSUM_TSO) == 0) return 1; return 0; } static int inline_size(struct mbuf *mb) { int len; len = mb->m_pkthdr.len; if (len + CTRL_SIZE + sizeof(struct mlx4_wqe_inline_seg) <= MLX4_INLINE_ALIGN) return ALIGN(len + CTRL_SIZE + sizeof(struct mlx4_wqe_inline_seg), 16); else return ALIGN(len + CTRL_SIZE + 2 * sizeof(struct mlx4_wqe_inline_seg), 16); } static int get_head_size(struct mbuf *mb) { struct ether_vlan_header *eh; struct tcphdr *th; struct ip *ip; int ip_hlen, tcp_hlen; struct ip6_hdr *ip6; uint16_t eth_type; int eth_hdr_len; eh = mtod(mb, struct ether_vlan_header *); if (mb->m_len < ETHER_HDR_LEN) return (0); if (eh->evl_encap_proto == htons(ETHERTYPE_VLAN)) { eth_type = ntohs(eh->evl_proto); eth_hdr_len = ETHER_HDR_LEN + ETHER_VLAN_ENCAP_LEN; } else { eth_type = ntohs(eh->evl_encap_proto); eth_hdr_len = ETHER_HDR_LEN; } if (mb->m_len < eth_hdr_len) return (0); switch (eth_type) { case ETHERTYPE_IP: ip = (struct ip *)(mb->m_data + eth_hdr_len); if (mb->m_len < eth_hdr_len + sizeof(*ip)) return (0); if (ip->ip_p != IPPROTO_TCP) return (0); ip_hlen = ip->ip_hl << 2; eth_hdr_len += ip_hlen; break; case ETHERTYPE_IPV6: ip6 = (struct ip6_hdr *)(mb->m_data + eth_hdr_len); if (mb->m_len < eth_hdr_len + sizeof(*ip6)) return (0); if (ip6->ip6_nxt != IPPROTO_TCP) return (0); eth_hdr_len += sizeof(*ip6); break; default: return (0); } if (mb->m_len < eth_hdr_len + sizeof(*th)) return (0); th = (struct tcphdr *)(mb->m_data + eth_hdr_len); tcp_hlen = th->th_off << 2; eth_hdr_len += tcp_hlen; if (mb->m_len < eth_hdr_len) return (0); return (eth_hdr_len); } static int get_real_size(struct mbuf *mb, struct net_device *dev, int *p_n_segs, int *lso_header_size, int inl) { struct mbuf *m; int nr_segs = 0; for (m = mb; m != NULL; m = m->m_next) if (m->m_len) nr_segs++; if (mb->m_pkthdr.csum_flags & CSUM_TSO) { *lso_header_size = get_head_size(mb); if (*lso_header_size) { if (mb->m_len == *lso_header_size) nr_segs--; *p_n_segs = nr_segs; return CTRL_SIZE + nr_segs * DS_SIZE + ALIGN(*lso_header_size + 4, DS_SIZE); } } else *lso_header_size = 0; *p_n_segs = nr_segs; if (inl) return inline_size(mb); return (CTRL_SIZE + nr_segs * DS_SIZE); } static struct mbuf *mb_copy(struct mbuf *mb, int *offp, char *data, int len) { int bytes; int off; off = *offp; while (len) { bytes = min(mb->m_len - off, len); if (bytes) memcpy(data, mb->m_data + off, bytes); len -= bytes; data += bytes; off += bytes; if (off == mb->m_len) { off = 0; mb = mb->m_next; } } *offp = off; return (mb); } static void build_inline_wqe(struct mlx4_en_tx_desc *tx_desc, struct mbuf *mb, int real_size, u16 *vlan_tag, int tx_ind) { struct mlx4_wqe_inline_seg *inl = &tx_desc->inl; int spc = MLX4_INLINE_ALIGN - CTRL_SIZE - sizeof *inl; int len; int off; off = 0; len = mb->m_pkthdr.len; if (len <= spc) { inl->byte_count = cpu_to_be32(1 << 31 | (max_t(typeof(len), len, MIN_PKT_LEN))); mb_copy(mb, &off, (void *)(inl + 1), len); if (len < MIN_PKT_LEN) memset(((void *)(inl + 1)) + len, 0, MIN_PKT_LEN - len); } else { inl->byte_count = cpu_to_be32(1 << 31 | spc); mb = mb_copy(mb, &off, (void *)(inl + 1), spc); inl = (void *) (inl + 1) + spc; mb_copy(mb, &off, (void *)(inl + 1), len - spc); wmb(); inl->byte_count = cpu_to_be32(1 << 31 | (len - spc)); } tx_desc->ctrl.vlan_tag = cpu_to_be16(*vlan_tag); tx_desc->ctrl.ins_vlan = MLX4_WQE_CTRL_INS_VLAN * !!(*vlan_tag); tx_desc->ctrl.fence_size = (real_size / 16) & 0x3f; } static unsigned long hashrandom; static void hashrandom_init(void *arg) { hashrandom = random(); } SYSINIT(hashrandom_init, SI_SUB_KLD, SI_ORDER_SECOND, &hashrandom_init, NULL); u16 mlx4_en_select_queue(struct net_device *dev, struct mbuf *mb) { struct mlx4_en_priv *priv = netdev_priv(dev); u32 rings_p_up = priv->num_tx_rings_p_up; - u32 vlan_tag = 0; u32 up = 0; u32 queue_index; +#if (MLX4_EN_NUM_UP > 1) /* Obtain VLAN information if present */ if (mb->m_flags & M_VLANTAG) { - vlan_tag = mb->m_pkthdr.ether_vtag; - up = (vlan_tag >> 13); + u32 vlan_tag = mb->m_pkthdr.ether_vtag; + up = (vlan_tag >> 13) % MLX4_EN_NUM_UP; } - +#endif /* check if flowid is set */ if (M_HASHTYPE_GET(mb) != M_HASHTYPE_NONE) queue_index = mb->m_pkthdr.flowid; else queue_index = mlx4_en_hashmbuf(MLX4_F_HASHL3 | MLX4_F_HASHL4, mb, hashrandom); return ((queue_index % rings_p_up) + (up * rings_p_up)); } static void mlx4_bf_copy(void __iomem *dst, unsigned long *src, unsigned bytecnt) { __iowrite64_copy(dst, src, bytecnt / 8); } static u64 mlx4_en_mac_to_u64(u8 *addr) { u64 mac = 0; int i; for (i = 0; i < ETHER_ADDR_LEN; i++) { mac <<= 8; mac |= addr[i]; } return mac; } static int mlx4_en_xmit(struct net_device *dev, int tx_ind, struct mbuf **mbp) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_dev *mdev = priv->mdev; struct mlx4_en_tx_ring *ring; struct mlx4_en_cq *cq; struct mlx4_en_tx_desc *tx_desc; struct mlx4_wqe_data_seg *data; struct mlx4_en_tx_info *tx_info; struct mbuf *m; int nr_txbb; int nr_segs; int desc_size; int real_size; dma_addr_t dma; u32 index, bf_index, ring_size; __be32 op_own; u16 vlan_tag = 0; int i; int lso_header_size; bool bounce = false; bool inl = false; struct mbuf *mb; mb = *mbp; int defrag = 1; if (!priv->port_up) goto tx_drop; ring = priv->tx_ring[tx_ind]; ring_size = ring->size; inl = is_inline(mb, ring->inline_thold); retry: real_size = get_real_size(mb, dev, &nr_segs, &lso_header_size, inl); if (unlikely(!real_size)) goto tx_drop; /* Align descriptor to TXBB size */ desc_size = ALIGN(real_size, TXBB_SIZE); nr_txbb = desc_size / TXBB_SIZE; if (unlikely(nr_txbb > MAX_DESC_TXBBS)) { if (defrag) { mb = m_defrag(*mbp, M_NOWAIT); if (mb == NULL) { mb = *mbp; goto tx_drop; } *mbp = mb; defrag = 0; goto retry; } en_warn(priv, "Oversized header or SG list\n"); goto tx_drop; } /* Obtain VLAN information if present */ if (mb->m_flags & M_VLANTAG) { vlan_tag = mb->m_pkthdr.ether_vtag; } /* Check available TXBBs and 2K spare for prefetch * Even if netif_tx_stop_queue() will be called * driver will send current packet to ensure * that at least one completion will be issued after * stopping the queue */ if (unlikely((int)(ring->prod - ring->cons) > ring->full_size)) { /* every full Tx ring stops queue */ if (ring->blocked == 0) atomic_add_int(&priv->blocked, 1); /* Set HW-queue-is-full flag */ atomic_set_int(&dev->if_drv_flags, IFF_DRV_OACTIVE); ring->blocked = 1; priv->port_stats.queue_stopped++; ring->queue_stopped++; /* Use interrupts to find out when queue opened */ cq = priv->tx_cq[tx_ind]; mlx4_en_arm_cq(priv, cq); return EBUSY; } /* Track current inflight packets for performance analysis */ AVG_PERF_COUNTER(priv->pstats.inflight_avg, (u32) (ring->prod - ring->cons - 1)); /* Packet is good - grab an index and transmit it */ index = ring->prod & ring->size_mask; bf_index = ring->prod; /* See if we have enough space for whole descriptor TXBB for setting * SW ownership on next descriptor; if not, use a bounce buffer. */ if (likely(index + nr_txbb <= ring_size)) tx_desc = ring->buf + index * TXBB_SIZE; else { tx_desc = (struct mlx4_en_tx_desc *) ring->bounce_buf; bounce = true; } /* Save mb in tx_info ring */ tx_info = &ring->tx_info[index]; tx_info->mb = mb; tx_info->nr_txbb = nr_txbb; tx_info->nr_segs = nr_segs; if (lso_header_size) { memcpy(tx_desc->lso.header, mb->m_data, lso_header_size); data = ((void *)&tx_desc->lso + ALIGN(lso_header_size + 4, DS_SIZE)); /* lso header is part of m_data. * need to omit when mapping DMA */ mb->m_data += lso_header_size; mb->m_len -= lso_header_size; } else data = &tx_desc->data; /* valid only for none inline segments */ tx_info->data_offset = (void *)data - (void *)tx_desc; if (inl) { tx_info->inl = 1; } else { for (i = 0, m = mb; i < nr_segs; i++, m = m->m_next) { if (m->m_len == 0) { i--; continue; } dma = pci_map_single(mdev->dev->pdev, m->m_data, m->m_len, PCI_DMA_TODEVICE); data->addr = cpu_to_be64(dma); data->lkey = cpu_to_be32(mdev->mr.key); wmb(); data->byte_count = cpu_to_be32(m->m_len); data++; } if (lso_header_size) { mb->m_data -= lso_header_size; mb->m_len += lso_header_size; } tx_info->inl = 0; } /* Prepare ctrl segement apart opcode+ownership, which depends on * whether LSO is used */ tx_desc->ctrl.vlan_tag = cpu_to_be16(vlan_tag); tx_desc->ctrl.ins_vlan = MLX4_WQE_CTRL_INS_VLAN * !!vlan_tag; tx_desc->ctrl.fence_size = (real_size / 16) & 0x3f; tx_desc->ctrl.srcrb_flags = priv->ctrl_flags; if (mb->m_pkthdr.csum_flags & (CSUM_IP | CSUM_TSO | CSUM_TCP | CSUM_UDP | CSUM_TCP_IPV6 | CSUM_UDP_IPV6)) { if (mb->m_pkthdr.csum_flags & (CSUM_IP | CSUM_TSO)) tx_desc->ctrl.srcrb_flags |= cpu_to_be32(MLX4_WQE_CTRL_IP_CSUM); if (mb->m_pkthdr.csum_flags & (CSUM_TCP | CSUM_UDP | CSUM_UDP_IPV6 | CSUM_TCP_IPV6 | CSUM_TSO)) tx_desc->ctrl.srcrb_flags |= cpu_to_be32(MLX4_WQE_CTRL_TCP_UDP_CSUM); priv->port_stats.tx_chksum_offload++; ring->tx_csum++; } if (unlikely(priv->validate_loopback)) { /* Copy dst mac address to wqe */ struct ether_header *ethh; u64 mac; u32 mac_l, mac_h; ethh = mtod(mb, struct ether_header *); mac = mlx4_en_mac_to_u64(ethh->ether_dhost); if (mac) { mac_h = (u32) ((mac & 0xffff00000000ULL) >> 16); mac_l = (u32) (mac & 0xffffffff); tx_desc->ctrl.srcrb_flags |= cpu_to_be32(mac_h); tx_desc->ctrl.imm = cpu_to_be32(mac_l); } } /* Handle LSO (TSO) packets */ if (lso_header_size) { int segsz; /* Mark opcode as LSO */ op_own = cpu_to_be32(MLX4_OPCODE_LSO | (1 << 6)) | ((ring->prod & ring_size) ? cpu_to_be32(MLX4_EN_BIT_DESC_OWN) : 0); /* Fill in the LSO prefix */ tx_desc->lso.mss_hdr_size = cpu_to_be32( mb->m_pkthdr.tso_segsz << 16 | lso_header_size); priv->port_stats.tso_packets++; segsz = mb->m_pkthdr.tso_segsz; i = ((mb->m_pkthdr.len - lso_header_size + segsz - 1) / segsz); tx_info->nr_bytes= mb->m_pkthdr.len + (i - 1) * lso_header_size; ring->packets += i; } else { /* Normal (Non LSO) packet */ op_own = cpu_to_be32(MLX4_OPCODE_SEND) | ((ring->prod & ring_size) ? cpu_to_be32(MLX4_EN_BIT_DESC_OWN) : 0); tx_info->nr_bytes = max(mb->m_pkthdr.len, (unsigned int)ETHER_MIN_LEN - ETHER_CRC_LEN); ring->packets++; } ring->bytes += tx_info->nr_bytes; AVG_PERF_COUNTER(priv->pstats.tx_pktsz_avg, mb->m_pkthdr.len); if (tx_info->inl) { build_inline_wqe(tx_desc, mb, real_size, &vlan_tag, tx_ind); tx_info->inl = 1; } ring->prod += nr_txbb; /* If we used a bounce buffer then copy descriptor back into place */ if (unlikely(bounce)) tx_desc = mlx4_en_bounce_to_desc(priv, ring, index, desc_size); if (ring->bf_enabled && desc_size <= MAX_BF && !bounce && !vlan_tag) { *(__be32 *) (&tx_desc->ctrl.vlan_tag) |= cpu_to_be32(ring->doorbell_qpn); op_own |= htonl((bf_index & 0xffff) << 8); /* Ensure new descirptor hits memory * before setting ownership of this descriptor to HW */ wmb(); tx_desc->ctrl.owner_opcode = op_own; wmb(); mlx4_bf_copy(ring->bf.reg + ring->bf.offset, (unsigned long *) &tx_desc->ctrl, desc_size); wmb(); ring->bf.offset ^= ring->bf.buf_size; } else { /* Ensure new descirptor hits memory * before setting ownership of this descriptor to HW */ wmb(); tx_desc->ctrl.owner_opcode = op_own; wmb(); writel(cpu_to_be32(ring->doorbell_qpn), ring->bf.uar->map + MLX4_SEND_DOORBELL); } return 0; tx_drop: *mbp = NULL; m_freem(mb); return EINVAL; } static int mlx4_en_transmit_locked(struct ifnet *dev, int tx_ind, struct mbuf *m) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_tx_ring *ring; struct mbuf *next; int enqueued, err = 0; ring = priv->tx_ring[tx_ind]; if ((dev->if_drv_flags & (IFF_DRV_RUNNING | IFF_DRV_OACTIVE)) != IFF_DRV_RUNNING || priv->port_up == 0) { if (m != NULL) err = drbr_enqueue(dev, ring->br, m); return (err); } enqueued = 0; if (m != NULL) { if ((err = drbr_enqueue(dev, ring->br, m)) != 0) return (err); } /* Process the queue */ while ((next = drbr_peek(dev, ring->br)) != NULL) { if ((err = mlx4_en_xmit(dev, tx_ind, &next)) != 0) { if (next == NULL) { drbr_advance(dev, ring->br); } else { drbr_putback(dev, ring->br, next); } break; } drbr_advance(dev, ring->br); enqueued++; if ((dev->if_drv_flags & IFF_DRV_RUNNING) == 0) break; } if (enqueued > 0) ring->watchdog_time = ticks; return (err); } void mlx4_en_tx_que(void *context, int pending) { struct mlx4_en_tx_ring *ring; struct mlx4_en_priv *priv; struct net_device *dev; struct mlx4_en_cq *cq; int tx_ind; cq = context; dev = cq->dev; priv = dev->if_softc; tx_ind = cq->ring; ring = priv->tx_ring[tx_ind]; if (dev->if_drv_flags & IFF_DRV_RUNNING) { mlx4_en_xmit_poll(priv, tx_ind); spin_lock(&ring->tx_lock); if (!drbr_empty(dev, ring->br)) mlx4_en_transmit_locked(dev, tx_ind, NULL); spin_unlock(&ring->tx_lock); } } int mlx4_en_transmit(struct ifnet *dev, struct mbuf *m) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_tx_ring *ring; struct mlx4_en_cq *cq; int i, err = 0; /* Compute which queue to use */ i = mlx4_en_select_queue(dev, m); ring = priv->tx_ring[i]; if (spin_trylock(&ring->tx_lock)) { err = mlx4_en_transmit_locked(dev, i, m); spin_unlock(&ring->tx_lock); /* Poll CQ here */ mlx4_en_xmit_poll(priv, i); } else { err = drbr_enqueue(dev, ring->br, m); cq = priv->tx_cq[i]; taskqueue_enqueue(cq->tq, &cq->cq_task); } return (err); } /* * Flush ring buffers. */ void mlx4_en_qflush(struct ifnet *dev) { struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_tx_ring *ring; struct mbuf *m; for (int i = 0; i < priv->tx_ring_num; i++) { ring = priv->tx_ring[i]; spin_lock(&ring->tx_lock); while ((m = buf_ring_dequeue_sc(ring->br)) != NULL) m_freem(m); spin_unlock(&ring->tx_lock); } if_qflush(dev); } Index: projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/main.c =================================================================== --- projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/main.c (revision 279716) +++ projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/main.c (revision 279717) @@ -1,3795 +1,3813 @@ /* * Copyright (c) 2004, 2005 Topspin Communications. All rights reserved. * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved. * Copyright (c) 2005, 2006, 2007, 2008, 2014 Mellanox Technologies. All rights reserved. * Copyright (c) 2006, 2007 Cisco Systems, Inc. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU * General Public License (GPL) Version 2, available from the file * COPYING in the main directory of this source tree, or the * OpenIB.org BSD license below: * * Redistribution and use in source and binary forms, with or * without modification, are permitted provided that the following * conditions are met: * * - Redistributions of source code must retain the above * copyright notice, this list of conditions and the following * disclaimer. * * - Redistributions in binary form must reproduce the above * copyright notice, this list of conditions and the following * disclaimer in the documentation and/or other materials * provided with the distribution. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include "mlx4.h" #include "fw.h" #include "icm.h" #include "mlx4_stats.h" MODULE_AUTHOR("Roland Dreier"); MODULE_DESCRIPTION("Mellanox ConnectX HCA low-level driver"); MODULE_LICENSE("Dual BSD/GPL"); struct workqueue_struct *mlx4_wq; #ifdef CONFIG_MLX4_DEBUG int mlx4_debug_level = 0; module_param_named(debug_level, mlx4_debug_level, int, 0644); MODULE_PARM_DESC(debug_level, "Enable debug tracing if > 0"); #endif /* CONFIG_MLX4_DEBUG */ #ifdef CONFIG_PCI_MSI static int msi_x = 1; module_param(msi_x, int, 0444); MODULE_PARM_DESC(msi_x, "0 - don't use MSI-X, 1 - use MSI-X, >1 - limit number of MSI-X irqs to msi_x (non-SRIOV only)"); #else /* CONFIG_PCI_MSI */ #define msi_x (0) #endif /* CONFIG_PCI_MSI */ static int enable_sys_tune = 0; module_param(enable_sys_tune, int, 0444); MODULE_PARM_DESC(enable_sys_tune, "Tune the cpu's for better performance (default 0)"); int mlx4_blck_lb = 1; module_param_named(block_loopback, mlx4_blck_lb, int, 0644); MODULE_PARM_DESC(block_loopback, "Block multicast loopback packets if > 0 " "(default: 1)"); enum { DEFAULT_DOMAIN = 0, BDF_STR_SIZE = 8, /* bb:dd.f- */ DBDF_STR_SIZE = 13 /* mmmm:bb:dd.f- */ }; enum { NUM_VFS, PROBE_VF, PORT_TYPE_ARRAY }; enum { VALID_DATA, INVALID_DATA, INVALID_STR }; struct param_data { int id; struct mlx4_dbdf2val_lst dbdf2val; }; static struct param_data num_vfs = { .id = NUM_VFS, .dbdf2val = { .name = "num_vfs param", .num_vals = 1, .def_val = {0}, .range = {0, MLX4_MAX_NUM_VF} } }; module_param_string(num_vfs, num_vfs.dbdf2val.str, sizeof(num_vfs.dbdf2val.str), 0444); MODULE_PARM_DESC(num_vfs, "Either single value (e.g. '5') to define uniform num_vfs value for all devices functions\n" "\t\tor a string to map device function numbers to their num_vfs values (e.g. '0000:04:00.0-5,002b:1c:0b.a-15').\n" "\t\tHexadecimal digits for the device function (e.g. 002b:1c:0b.a) and decimal for num_vfs value (e.g. 15)."); static struct param_data probe_vf = { .id = PROBE_VF, .dbdf2val = { .name = "probe_vf param", .num_vals = 1, .def_val = {0}, .range = {0, MLX4_MAX_NUM_VF} } }; module_param_string(probe_vf, probe_vf.dbdf2val.str, sizeof(probe_vf.dbdf2val.str), 0444); MODULE_PARM_DESC(probe_vf, "Either single value (e.g. '3') to define uniform number of VFs to probe by the pf driver for all devices functions\n" "\t\tor a string to map device function numbers to their probe_vf values (e.g. '0000:04:00.0-3,002b:1c:0b.a-13').\n" "\t\tHexadecimal digits for the device function (e.g. 002b:1c:0b.a) and decimal for probe_vf value (e.g. 13)."); int mlx4_log_num_mgm_entry_size = MLX4_DEFAULT_MGM_LOG_ENTRY_SIZE; module_param_named(log_num_mgm_entry_size, mlx4_log_num_mgm_entry_size, int, 0444); MODULE_PARM_DESC(log_num_mgm_entry_size, "log mgm size, that defines the num" " of qp per mcg, for example:" " 10 gives 248.range: 7 <=" " log_num_mgm_entry_size <= 12." " To activate device managed" " flow steering when available, set to -1"); static int high_rate_steer; module_param(high_rate_steer, int, 0444); MODULE_PARM_DESC(high_rate_steer, "Enable steering mode for higher packet rate" " (default off)"); static int fast_drop; module_param_named(fast_drop, fast_drop, int, 0444); MODULE_PARM_DESC(fast_drop, "Enable fast packet drop when no recieve WQEs are posted"); int mlx4_enable_64b_cqe_eqe = 1; module_param_named(enable_64b_cqe_eqe, mlx4_enable_64b_cqe_eqe, int, 0644); MODULE_PARM_DESC(enable_64b_cqe_eqe, "Enable 64 byte CQEs/EQEs when the the FW supports this if non-zero (default: 1)"); #define HCA_GLOBAL_CAP_MASK 0 #define PF_CONTEXT_BEHAVIOUR_MASK MLX4_FUNC_CAP_64B_EQE_CQE static char mlx4_version[] __devinitdata = DRV_NAME ": Mellanox ConnectX core driver v" DRV_VERSION " (" DRV_RELDATE ")\n"; static int log_num_mac = 7; module_param_named(log_num_mac, log_num_mac, int, 0444); MODULE_PARM_DESC(log_num_mac, "Log2 max number of MACs per ETH port (1-7)"); static int log_num_vlan; module_param_named(log_num_vlan, log_num_vlan, int, 0444); MODULE_PARM_DESC(log_num_vlan, "(Obsolete) Log2 max number of VLANs per ETH port (0-7)"); /* Log2 max number of VLANs per ETH port (0-7) */ #define MLX4_LOG_NUM_VLANS 7 int log_mtts_per_seg = ilog2(1); module_param_named(log_mtts_per_seg, log_mtts_per_seg, int, 0444); MODULE_PARM_DESC(log_mtts_per_seg, "Log2 number of MTT entries per segment " "(0-7) (default: 0)"); static struct param_data port_type_array = { .id = PORT_TYPE_ARRAY, .dbdf2val = { .name = "port_type_array param", .num_vals = 2, .def_val = {MLX4_PORT_TYPE_ETH, MLX4_PORT_TYPE_ETH}, .range = {MLX4_PORT_TYPE_IB, MLX4_PORT_TYPE_NA} } }; module_param_string(port_type_array, port_type_array.dbdf2val.str, sizeof(port_type_array.dbdf2val.str), 0444); MODULE_PARM_DESC(port_type_array, "Either pair of values (e.g. '1,2') to define uniform port1/port2 types configuration for all devices functions\n" "\t\tor a string to map device function numbers to their pair of port types values (e.g. '0000:04:00.0-1;2,002b:1c:0b.a-1;1').\n" "\t\tValid port types: 1-ib, 2-eth, 3-auto, 4-N/A\n" "\t\tIn case that only one port is available use the N/A port type for port2 (e.g '1,4')."); struct mlx4_port_config { struct list_head list; enum mlx4_port_type port_type[MLX4_MAX_PORTS + 1]; struct pci_dev *pdev; }; #define MLX4_LOG_NUM_MTT 20 /* We limit to 30 as of a bit map issue which uses int and not uint. see mlx4_buddy_init -> bitmap_zero which gets int. */ #define MLX4_MAX_LOG_NUM_MTT 30 static struct mlx4_profile mod_param_profile = { .num_qp = 19, .num_srq = 16, .rdmarc_per_qp = 4, .num_cq = 16, .num_mcg = 13, .num_mpt = 19, .num_mtt_segs = 0, /* max(20, 2*MTTs for host memory)) */ }; module_param_named(log_num_qp, mod_param_profile.num_qp, int, 0444); MODULE_PARM_DESC(log_num_qp, "log maximum number of QPs per HCA (default: 19)"); module_param_named(log_num_srq, mod_param_profile.num_srq, int, 0444); MODULE_PARM_DESC(log_num_srq, "log maximum number of SRQs per HCA " "(default: 16)"); module_param_named(log_rdmarc_per_qp, mod_param_profile.rdmarc_per_qp, int, 0444); MODULE_PARM_DESC(log_rdmarc_per_qp, "log number of RDMARC buffers per QP " "(default: 4)"); module_param_named(log_num_cq, mod_param_profile.num_cq, int, 0444); MODULE_PARM_DESC(log_num_cq, "log maximum number of CQs per HCA (default: 16)"); module_param_named(log_num_mcg, mod_param_profile.num_mcg, int, 0444); MODULE_PARM_DESC(log_num_mcg, "log maximum number of multicast groups per HCA " "(default: 13)"); module_param_named(log_num_mpt, mod_param_profile.num_mpt, int, 0444); MODULE_PARM_DESC(log_num_mpt, "log maximum number of memory protection table entries per " "HCA (default: 19)"); module_param_named(log_num_mtt, mod_param_profile.num_mtt_segs, int, 0444); MODULE_PARM_DESC(log_num_mtt, "log maximum number of memory translation table segments per " "HCA (default: max(20, 2*MTTs for register all of the host memory limited to 30))"); enum { MLX4_IF_STATE_BASIC, MLX4_IF_STATE_EXTENDED }; static inline u64 dbdf_to_u64(int domain, int bus, int dev, int fn) { return (domain << 20) | (bus << 12) | (dev << 4) | fn; } static inline void pr_bdf_err(const char *dbdf, const char *pname) { pr_warn("mlx4_core: '%s' is not valid bdf in '%s'\n", dbdf, pname); } static inline void pr_val_err(const char *dbdf, const char *pname, const char *val) { pr_warn("mlx4_core: value '%s' of bdf '%s' in '%s' is not valid\n" , val, dbdf, pname); } static inline void pr_out_of_range_bdf(const char *dbdf, int val, struct mlx4_dbdf2val_lst *dbdf2val) { pr_warn("mlx4_core: value %d in bdf '%s' of '%s' is out of its valid range (%d,%d)\n" , val, dbdf, dbdf2val->name , dbdf2val->range.min, dbdf2val->range.max); } static inline void pr_out_of_range(struct mlx4_dbdf2val_lst *dbdf2val) { pr_warn("mlx4_core: value of '%s' is out of its valid range (%d,%d)\n" , dbdf2val->name , dbdf2val->range.min, dbdf2val->range.max); } static inline int is_in_range(int val, struct mlx4_range *r) { return (val >= r->min && val <= r->max); } static int update_defaults(struct param_data *pdata) { long int val[MLX4_MAX_BDF_VALS]; int ret; char *t, *p = pdata->dbdf2val.str; char sval[32]; int val_len; if (!strlen(p) || strchr(p, ':') || strchr(p, '.') || strchr(p, ';')) return INVALID_STR; switch (pdata->id) { case PORT_TYPE_ARRAY: t = strchr(p, ','); if (!t || t == p || (t - p) > sizeof(sval)) return INVALID_STR; val_len = t - p; strncpy(sval, p, val_len); sval[val_len] = 0; ret = kstrtol(sval, 0, &val[0]); if (ret == -EINVAL) return INVALID_STR; if (ret || !is_in_range(val[0], &pdata->dbdf2val.range)) { pr_out_of_range(&pdata->dbdf2val); return INVALID_DATA; } ret = kstrtol(t + 1, 0, &val[1]); if (ret == -EINVAL) return INVALID_STR; if (ret || !is_in_range(val[1], &pdata->dbdf2val.range)) { pr_out_of_range(&pdata->dbdf2val); return INVALID_DATA; } pdata->dbdf2val.tbl[0].val[0] = val[0]; pdata->dbdf2val.tbl[0].val[1] = val[1]; break; case NUM_VFS: case PROBE_VF: ret = kstrtol(p, 0, &val[0]); if (ret == -EINVAL) return INVALID_STR; if (ret || !is_in_range(val[0], &pdata->dbdf2val.range)) { pr_out_of_range(&pdata->dbdf2val); return INVALID_DATA; } pdata->dbdf2val.tbl[0].val[0] = val[0]; break; } pdata->dbdf2val.tbl[1].dbdf = MLX4_ENDOF_TBL; return VALID_DATA; } int mlx4_fill_dbdf2val_tbl(struct mlx4_dbdf2val_lst *dbdf2val_lst) { int domain, bus, dev, fn; u64 dbdf; char *p, *t, *v; char tmp[32]; char sbdf[32]; char sep = ','; int j, k, str_size, i = 1; int prfx_size; p = dbdf2val_lst->str; for (j = 0; j < dbdf2val_lst->num_vals; j++) dbdf2val_lst->tbl[0].val[j] = dbdf2val_lst->def_val[j]; dbdf2val_lst->tbl[1].dbdf = MLX4_ENDOF_TBL; str_size = strlen(dbdf2val_lst->str); if (str_size == 0) return 0; while (strlen(p)) { prfx_size = BDF_STR_SIZE; sbdf[prfx_size] = 0; strncpy(sbdf, p, prfx_size); domain = DEFAULT_DOMAIN; if (sscanf(sbdf, "%02x:%02x.%x-", &bus, &dev, &fn) != 3) { prfx_size = DBDF_STR_SIZE; sbdf[prfx_size] = 0; strncpy(sbdf, p, prfx_size); if (sscanf(sbdf, "%04x:%02x:%02x.%x-", &domain, &bus, &dev, &fn) != 4) { pr_bdf_err(sbdf, dbdf2val_lst->name); goto err; } sprintf(tmp, "%04x:%02x:%02x.%x-", domain, bus, dev, fn); } else { sprintf(tmp, "%02x:%02x.%x-", bus, dev, fn); } if (strnicmp(sbdf, tmp, sizeof(tmp))) { pr_bdf_err(sbdf, dbdf2val_lst->name); goto err; } dbdf = dbdf_to_u64(domain, bus, dev, fn); for (j = 1; j < i; j++) if (dbdf2val_lst->tbl[j].dbdf == dbdf) { pr_warn("mlx4_core: in '%s', %s appears multiple times\n" , dbdf2val_lst->name, sbdf); goto err; } if (i >= MLX4_DEVS_TBL_SIZE) { pr_warn("mlx4_core: Too many devices in '%s'\n" , dbdf2val_lst->name); goto err; } p += prfx_size; t = strchr(p, sep); t = t ? t : p + strlen(p); if (p >= t) { pr_val_err(sbdf, dbdf2val_lst->name, ""); goto err; } for (k = 0; k < dbdf2val_lst->num_vals; k++) { char sval[32]; long int val; int ret, val_len; char vsep = ';'; v = (k == dbdf2val_lst->num_vals - 1) ? t : strchr(p, vsep); if (!v || v > t || v == p || (v - p) > sizeof(sval)) { pr_val_err(sbdf, dbdf2val_lst->name, p); goto err; } val_len = v - p; strncpy(sval, p, val_len); sval[val_len] = 0; ret = kstrtol(sval, 0, &val); if (ret) { if (strchr(p, vsep)) pr_warn("mlx4_core: too many vals in bdf '%s' of '%s'\n" , sbdf, dbdf2val_lst->name); else pr_val_err(sbdf, dbdf2val_lst->name, sval); goto err; } if (!is_in_range(val, &dbdf2val_lst->range)) { pr_out_of_range_bdf(sbdf, val, dbdf2val_lst); goto err; } dbdf2val_lst->tbl[i].val[k] = val; p = v; if (p[0] == vsep) p++; } dbdf2val_lst->tbl[i].dbdf = dbdf; if (strlen(p)) { if (p[0] != sep) { pr_warn("mlx4_core: expect separator '%c' before '%s' in '%s'\n" , sep, p, dbdf2val_lst->name); goto err; } p++; } i++; if (i < MLX4_DEVS_TBL_SIZE) dbdf2val_lst->tbl[i].dbdf = MLX4_ENDOF_TBL; } return 0; err: dbdf2val_lst->tbl[1].dbdf = MLX4_ENDOF_TBL; pr_warn("mlx4_core: The value of '%s' is incorrect. The value is discarded!\n" , dbdf2val_lst->name); return -EINVAL; } EXPORT_SYMBOL(mlx4_fill_dbdf2val_tbl); int mlx4_get_val(struct mlx4_dbdf2val *tbl, struct pci_dev *pdev, int idx, int *val) { u64 dbdf; int i = 1; *val = tbl[0].val[idx]; if (!pdev) return -EINVAL; dbdf = dbdf_to_u64(pci_get_domain(pdev->dev.bsddev), pci_get_bus(pdev->dev.bsddev), PCI_SLOT(pdev->devfn), PCI_FUNC(pdev->devfn)); while ((i < MLX4_DEVS_TBL_SIZE) && (tbl[i].dbdf != MLX4_ENDOF_TBL)) { if (tbl[i].dbdf == dbdf) { *val = tbl[i].val[idx]; return 0; } i++; } return 0; } EXPORT_SYMBOL(mlx4_get_val); static void process_mod_param_profile(struct mlx4_profile *profile) { vm_size_t hwphyssz; hwphyssz = 0; TUNABLE_ULONG_FETCH("hw.realmem", (u_long *) &hwphyssz); profile->num_qp = 1 << mod_param_profile.num_qp; profile->num_srq = 1 << mod_param_profile.num_srq; profile->rdmarc_per_qp = 1 << mod_param_profile.rdmarc_per_qp; profile->num_cq = 1 << mod_param_profile.num_cq; profile->num_mcg = 1 << mod_param_profile.num_mcg; profile->num_mpt = 1 << mod_param_profile.num_mpt; /* * We want to scale the number of MTTs with the size of the * system memory, since it makes sense to register a lot of * memory on a system with a lot of memory. As a heuristic, * make sure we have enough MTTs to register twice the system * memory (with PAGE_SIZE entries). * * This number has to be a power of two and fit into 32 bits * due to device limitations. We cap this at 2^30 as of bit map * limitation to work with int instead of uint (mlx4_buddy_init -> bitmap_zero) * That limits us to 4TB of memory registration per HCA with * 4KB pages, which is probably OK for the next few months. */ if (mod_param_profile.num_mtt_segs) profile->num_mtt_segs = 1 << mod_param_profile.num_mtt_segs; else { profile->num_mtt_segs = roundup_pow_of_two(max_t(unsigned, 1 << (MLX4_LOG_NUM_MTT - log_mtts_per_seg), min(1UL << (MLX4_MAX_LOG_NUM_MTT - log_mtts_per_seg), (hwphyssz << 1) >> log_mtts_per_seg))); /* set the actual value, so it will be reflected to the user using the sysfs */ mod_param_profile.num_mtt_segs = ilog2(profile->num_mtt_segs); } } int mlx4_check_port_params(struct mlx4_dev *dev, enum mlx4_port_type *port_type) { int i; for (i = 0; i < dev->caps.num_ports - 1; i++) { if (port_type[i] != port_type[i + 1]) { if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP)) { mlx4_err(dev, "Only same port types supported " "on this HCA, aborting.\n"); return -EINVAL; } } } for (i = 0; i < dev->caps.num_ports; i++) { if (!(port_type[i] & dev->caps.supported_type[i+1])) { mlx4_err(dev, "Requested port type for port %d is not " "supported on this HCA\n", i + 1); return -EINVAL; } } return 0; } static void mlx4_set_port_mask(struct mlx4_dev *dev) { int i; for (i = 1; i <= dev->caps.num_ports; ++i) dev->caps.port_mask[i] = dev->caps.port_type[i]; } static int mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) { int err; int i; err = mlx4_QUERY_DEV_CAP(dev, dev_cap); if (err) { mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting.\n"); return err; } if (dev_cap->min_page_sz > PAGE_SIZE) { mlx4_err(dev, "HCA minimum page size of %d bigger than " "kernel PAGE_SIZE of %d, aborting.\n", dev_cap->min_page_sz, PAGE_SIZE); return -ENODEV; } if (dev_cap->num_ports > MLX4_MAX_PORTS) { mlx4_err(dev, "HCA has %d ports, but we only support %d, " "aborting.\n", dev_cap->num_ports, MLX4_MAX_PORTS); return -ENODEV; } if (dev_cap->uar_size > pci_resource_len(dev->pdev, 2)) { mlx4_err(dev, "HCA reported UAR size of 0x%x bigger than " "PCI resource 2 size of 0x%llx, aborting.\n", dev_cap->uar_size, (unsigned long long) pci_resource_len(dev->pdev, 2)); return -ENODEV; } dev->caps.num_ports = dev_cap->num_ports; dev->phys_caps.num_phys_eqs = MLX4_MAX_EQ_NUM; for (i = 1; i <= dev->caps.num_ports; ++i) { dev->caps.vl_cap[i] = dev_cap->max_vl[i]; dev->caps.ib_mtu_cap[i] = dev_cap->ib_mtu[i]; dev->phys_caps.gid_phys_table_len[i] = dev_cap->max_gids[i]; dev->phys_caps.pkey_phys_table_len[i] = dev_cap->max_pkeys[i]; /* set gid and pkey table operating lengths by default * to non-sriov values */ dev->caps.gid_table_len[i] = dev_cap->max_gids[i]; dev->caps.pkey_table_len[i] = dev_cap->max_pkeys[i]; dev->caps.port_width_cap[i] = dev_cap->max_port_width[i]; dev->caps.eth_mtu_cap[i] = dev_cap->eth_mtu[i]; dev->caps.def_mac[i] = dev_cap->def_mac[i]; dev->caps.supported_type[i] = dev_cap->supported_port_types[i]; dev->caps.suggested_type[i] = dev_cap->suggested_type[i]; dev->caps.default_sense[i] = dev_cap->default_sense[i]; dev->caps.trans_type[i] = dev_cap->trans_type[i]; dev->caps.vendor_oui[i] = dev_cap->vendor_oui[i]; dev->caps.wavelength[i] = dev_cap->wavelength[i]; dev->caps.trans_code[i] = dev_cap->trans_code[i]; } dev->caps.uar_page_size = PAGE_SIZE; dev->caps.num_uars = dev_cap->uar_size / PAGE_SIZE; dev->caps.local_ca_ack_delay = dev_cap->local_ca_ack_delay; dev->caps.bf_reg_size = dev_cap->bf_reg_size; dev->caps.bf_regs_per_page = dev_cap->bf_regs_per_page; dev->caps.max_sq_sg = dev_cap->max_sq_sg; dev->caps.max_rq_sg = dev_cap->max_rq_sg; dev->caps.max_wqes = dev_cap->max_qp_sz; dev->caps.max_qp_init_rdma = dev_cap->max_requester_per_qp; dev->caps.max_srq_wqes = dev_cap->max_srq_sz; dev->caps.max_srq_sge = dev_cap->max_rq_sg - 1; dev->caps.reserved_srqs = dev_cap->reserved_srqs; dev->caps.max_sq_desc_sz = dev_cap->max_sq_desc_sz; dev->caps.max_rq_desc_sz = dev_cap->max_rq_desc_sz; /* * Subtract 1 from the limit because we need to allocate a * spare CQE to enable resizing the CQ */ dev->caps.max_cqes = dev_cap->max_cq_sz - 1; dev->caps.reserved_cqs = dev_cap->reserved_cqs; dev->caps.reserved_eqs = dev_cap->reserved_eqs; dev->caps.reserved_mtts = dev_cap->reserved_mtts; dev->caps.reserved_mrws = dev_cap->reserved_mrws; /* The first 128 UARs are used for EQ doorbells */ dev->caps.reserved_uars = max_t(int, 128, dev_cap->reserved_uars); dev->caps.reserved_pds = dev_cap->reserved_pds; dev->caps.reserved_xrcds = (dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC) ? dev_cap->reserved_xrcds : 0; dev->caps.max_xrcds = (dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC) ? dev_cap->max_xrcds : 0; dev->caps.mtt_entry_sz = dev_cap->mtt_entry_sz; dev->caps.max_msg_sz = dev_cap->max_msg_sz; dev->caps.page_size_cap = ~(u32) (dev_cap->min_page_sz - 1); dev->caps.flags = dev_cap->flags; dev->caps.flags2 = dev_cap->flags2; dev->caps.bmme_flags = dev_cap->bmme_flags; dev->caps.reserved_lkey = dev_cap->reserved_lkey; dev->caps.stat_rate_support = dev_cap->stat_rate_support; dev->caps.cq_timestamp = dev_cap->timestamp_support; dev->caps.max_gso_sz = dev_cap->max_gso_sz; dev->caps.max_rss_tbl_sz = dev_cap->max_rss_tbl_sz; /* Sense port always allowed on supported devices for ConnectX-1 and -2 */ if (mlx4_priv(dev)->pci_dev_data & MLX4_PCI_DEV_FORCE_SENSE_PORT) dev->caps.flags |= MLX4_DEV_CAP_FLAG_SENSE_SUPPORT; /* Don't do sense port on multifunction devices (for now at least) */ if (mlx4_is_mfunc(dev)) dev->caps.flags &= ~MLX4_DEV_CAP_FLAG_SENSE_SUPPORT; dev->caps.log_num_macs = log_num_mac; dev->caps.log_num_vlans = MLX4_LOG_NUM_VLANS; dev->caps.fast_drop = fast_drop ? !!(dev->caps.flags & MLX4_DEV_CAP_FLAG_FAST_DROP) : 0; for (i = 1; i <= dev->caps.num_ports; ++i) { dev->caps.port_type[i] = MLX4_PORT_TYPE_NONE; if (dev->caps.supported_type[i]) { /* if only ETH is supported - assign ETH */ if (dev->caps.supported_type[i] == MLX4_PORT_TYPE_ETH) dev->caps.port_type[i] = MLX4_PORT_TYPE_ETH; /* if only IB is supported, assign IB */ else if (dev->caps.supported_type[i] == MLX4_PORT_TYPE_IB) dev->caps.port_type[i] = MLX4_PORT_TYPE_IB; else { /* * if IB and ETH are supported, we set the port * type according to user selection of port type; * if there is no user selection, take the FW hint */ int pta; mlx4_get_val(port_type_array.dbdf2val.tbl, pci_physfn(dev->pdev), i - 1, &pta); if (pta == MLX4_PORT_TYPE_NONE) { dev->caps.port_type[i] = dev->caps.suggested_type[i] ? MLX4_PORT_TYPE_ETH : MLX4_PORT_TYPE_IB; } else if (pta == MLX4_PORT_TYPE_NA) { mlx4_err(dev, "Port %d is valid port. " "It is not allowed to configure its type to N/A(%d)\n", i, MLX4_PORT_TYPE_NA); return -EINVAL; } else { dev->caps.port_type[i] = pta; } } } /* * Link sensing is allowed on the port if 3 conditions are true: * 1. Both protocols are supported on the port. * 2. Different types are supported on the port * 3. FW declared that it supports link sensing */ mlx4_priv(dev)->sense.sense_allowed[i] = ((dev->caps.supported_type[i] == MLX4_PORT_TYPE_AUTO) && (dev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP) && (dev->caps.flags & MLX4_DEV_CAP_FLAG_SENSE_SUPPORT)); /* Disablling auto sense for default Eth ports support */ mlx4_priv(dev)->sense.sense_allowed[i] = 0; /* * If "default_sense" bit is set, we move the port to "AUTO" mode * and perform sense_port FW command to try and set the correct * port type from beginning */ if (mlx4_priv(dev)->sense.sense_allowed[i] && dev->caps.default_sense[i]) { enum mlx4_port_type sensed_port = MLX4_PORT_TYPE_NONE; dev->caps.possible_type[i] = MLX4_PORT_TYPE_AUTO; mlx4_SENSE_PORT(dev, i, &sensed_port); if (sensed_port != MLX4_PORT_TYPE_NONE) dev->caps.port_type[i] = sensed_port; } else { dev->caps.possible_type[i] = dev->caps.port_type[i]; } if (dev->caps.log_num_macs > dev_cap->log_max_macs[i]) { dev->caps.log_num_macs = dev_cap->log_max_macs[i]; mlx4_warn(dev, "Requested number of MACs is too much " "for port %d, reducing to %d.\n", i, 1 << dev->caps.log_num_macs); } if (dev->caps.log_num_vlans > dev_cap->log_max_vlans[i]) { dev->caps.log_num_vlans = dev_cap->log_max_vlans[i]; mlx4_warn(dev, "Requested number of VLANs is too much " "for port %d, reducing to %d.\n", i, 1 << dev->caps.log_num_vlans); } } dev->caps.max_basic_counters = dev_cap->max_basic_counters; dev->caps.max_extended_counters = dev_cap->max_extended_counters; /* support extended counters if available */ if (dev->caps.flags & MLX4_DEV_CAP_FLAG_COUNTERS_EXT) dev->caps.max_counters = dev->caps.max_extended_counters; else dev->caps.max_counters = dev->caps.max_basic_counters; dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW] = dev_cap->reserved_qps; dev->caps.reserved_qps_cnt[MLX4_QP_REGION_ETH_ADDR] = dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FC_ADDR] = (1 << dev->caps.log_num_macs) * (1 << dev->caps.log_num_vlans) * dev->caps.num_ports; dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FC_EXCH] = MLX4_NUM_FEXCH; dev->caps.reserved_qps = dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW] + dev->caps.reserved_qps_cnt[MLX4_QP_REGION_ETH_ADDR] + dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FC_ADDR] + dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FC_EXCH]; dev->caps.sync_qp = dev_cap->sync_qp; if (dev->pdev->device == 0x1003) dev->caps.cq_flags |= MLX4_DEV_CAP_CQ_FLAG_IO; dev->caps.sqp_demux = (mlx4_is_master(dev)) ? MLX4_MAX_NUM_SLAVES : 0; if (!mlx4_enable_64b_cqe_eqe && !mlx4_is_slave(dev)) { if (dev_cap->flags & (MLX4_DEV_CAP_FLAG_64B_CQE | MLX4_DEV_CAP_FLAG_64B_EQE)) { mlx4_warn(dev, "64B EQEs/CQEs supported by the device but not enabled\n"); dev->caps.flags &= ~MLX4_DEV_CAP_FLAG_64B_CQE; dev->caps.flags &= ~MLX4_DEV_CAP_FLAG_64B_EQE; } } if ((dev->caps.flags & (MLX4_DEV_CAP_FLAG_64B_CQE | MLX4_DEV_CAP_FLAG_64B_EQE)) && mlx4_is_master(dev)) dev->caps.function_caps |= MLX4_FUNC_CAP_64B_EQE_CQE; if (!mlx4_is_slave(dev)) { for (i = 0; i < dev->caps.num_ports; ++i) dev->caps.def_counter_index[i] = i << 1; } return 0; } /*The function checks if there are live vf, return the num of them*/ static int mlx4_how_many_lives_vf(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_slave_state *s_state; int i; int ret = 0; for (i = 1/*the ppf is 0*/; i < dev->num_slaves; ++i) { s_state = &priv->mfunc.master.slave_state[i]; if (s_state->active && s_state->last_cmd != MLX4_COMM_CMD_RESET) { mlx4_warn(dev, "%s: slave: %d is still active\n", __func__, i); ret++; } } return ret; } int mlx4_get_parav_qkey(struct mlx4_dev *dev, u32 qpn, u32 *qkey) { u32 qk = MLX4_RESERVED_QKEY_BASE; if (qpn >= dev->phys_caps.base_tunnel_sqpn + 8 * MLX4_MFUNC_MAX || qpn < dev->phys_caps.base_proxy_sqpn) return -EINVAL; if (qpn >= dev->phys_caps.base_tunnel_sqpn) /* tunnel qp */ qk += qpn - dev->phys_caps.base_tunnel_sqpn; else qk += qpn - dev->phys_caps.base_proxy_sqpn; *qkey = qk; return 0; } EXPORT_SYMBOL(mlx4_get_parav_qkey); void mlx4_sync_pkey_table(struct mlx4_dev *dev, int slave, int port, int i, int val) { struct mlx4_priv *priv = container_of(dev, struct mlx4_priv, dev); if (!mlx4_is_master(dev)) return; priv->virt2phys_pkey[slave][port - 1][i] = val; } EXPORT_SYMBOL(mlx4_sync_pkey_table); void mlx4_put_slave_node_guid(struct mlx4_dev *dev, int slave, __be64 guid) { struct mlx4_priv *priv = container_of(dev, struct mlx4_priv, dev); if (!mlx4_is_master(dev)) return; priv->slave_node_guids[slave] = guid; } EXPORT_SYMBOL(mlx4_put_slave_node_guid); __be64 mlx4_get_slave_node_guid(struct mlx4_dev *dev, int slave) { struct mlx4_priv *priv = container_of(dev, struct mlx4_priv, dev); if (!mlx4_is_master(dev)) return 0; return priv->slave_node_guids[slave]; } EXPORT_SYMBOL(mlx4_get_slave_node_guid); int mlx4_is_slave_active(struct mlx4_dev *dev, int slave) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_slave_state *s_slave; if (!mlx4_is_master(dev)) return 0; s_slave = &priv->mfunc.master.slave_state[slave]; return !!s_slave->active; } EXPORT_SYMBOL(mlx4_is_slave_active); static void slave_adjust_steering_mode(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap, struct mlx4_init_hca_param *hca_param) { dev->caps.steering_mode = hca_param->steering_mode; if (dev->caps.steering_mode == MLX4_STEERING_MODE_DEVICE_MANAGED) dev->caps.num_qp_per_mgm = dev_cap->fs_max_num_qp_per_entry; else dev->caps.num_qp_per_mgm = 4 * ((1 << hca_param->log_mc_entry_sz)/16 - 2); mlx4_dbg(dev, "Steering mode is: %s\n", mlx4_steering_mode_str(dev->caps.steering_mode)); } static int mlx4_slave_cap(struct mlx4_dev *dev) { int err; u32 page_size; struct mlx4_dev_cap dev_cap; struct mlx4_func_cap func_cap; struct mlx4_init_hca_param hca_param; int i; memset(&hca_param, 0, sizeof(hca_param)); err = mlx4_QUERY_HCA(dev, &hca_param); if (err) { mlx4_err(dev, "QUERY_HCA command failed, aborting.\n"); return err; } /*fail if the hca has an unknown capability */ if ((hca_param.global_caps | HCA_GLOBAL_CAP_MASK) != HCA_GLOBAL_CAP_MASK) { mlx4_err(dev, "Unknown hca global capabilities\n"); return -ENOSYS; } mlx4_log_num_mgm_entry_size = hca_param.log_mc_entry_sz; dev->caps.hca_core_clock = hca_param.hca_core_clock; memset(&dev_cap, 0, sizeof(dev_cap)); dev->caps.max_qp_dest_rdma = 1 << hca_param.log_rd_per_qp; err = mlx4_dev_cap(dev, &dev_cap); if (err) { mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting.\n"); return err; } err = mlx4_QUERY_FW(dev); if (err) mlx4_err(dev, "QUERY_FW command failed: could not get FW version.\n"); if (!hca_param.mw_enable) { dev->caps.flags &= ~MLX4_DEV_CAP_FLAG_MEM_WINDOW; dev->caps.bmme_flags &= ~MLX4_BMME_FLAG_TYPE_2_WIN; } page_size = ~dev->caps.page_size_cap + 1; mlx4_warn(dev, "HCA minimum page size:%d\n", page_size); if (page_size > PAGE_SIZE) { mlx4_err(dev, "HCA minimum page size of %d bigger than " "kernel PAGE_SIZE of %d, aborting.\n", page_size, PAGE_SIZE); return -ENODEV; } /* slave gets uar page size from QUERY_HCA fw command */ dev->caps.uar_page_size = 1 << (hca_param.uar_page_sz + 12); /* TODO: relax this assumption */ if (dev->caps.uar_page_size != PAGE_SIZE) { mlx4_err(dev, "UAR size:%d != kernel PAGE_SIZE of %d\n", dev->caps.uar_page_size, PAGE_SIZE); return -ENODEV; } memset(&func_cap, 0, sizeof(func_cap)); err = mlx4_QUERY_FUNC_CAP(dev, 0, &func_cap); if (err) { mlx4_err(dev, "QUERY_FUNC_CAP general command failed, aborting (%d).\n", err); return err; } if ((func_cap.pf_context_behaviour | PF_CONTEXT_BEHAVIOUR_MASK) != PF_CONTEXT_BEHAVIOUR_MASK) { mlx4_err(dev, "Unknown pf context behaviour\n"); return -ENOSYS; } dev->caps.num_ports = func_cap.num_ports; dev->quotas.qp = func_cap.qp_quota; dev->quotas.srq = func_cap.srq_quota; dev->quotas.cq = func_cap.cq_quota; dev->quotas.mpt = func_cap.mpt_quota; dev->quotas.mtt = func_cap.mtt_quota; dev->caps.num_qps = 1 << hca_param.log_num_qps; dev->caps.num_srqs = 1 << hca_param.log_num_srqs; dev->caps.num_cqs = 1 << hca_param.log_num_cqs; dev->caps.num_mpts = 1 << hca_param.log_mpt_sz; dev->caps.num_eqs = func_cap.max_eq; dev->caps.reserved_eqs = func_cap.reserved_eq; dev->caps.num_pds = MLX4_NUM_PDS; dev->caps.num_mgms = 0; dev->caps.num_amgms = 0; if (dev->caps.num_ports > MLX4_MAX_PORTS) { mlx4_err(dev, "HCA has %d ports, but we only support %d, " "aborting.\n", dev->caps.num_ports, MLX4_MAX_PORTS); return -ENODEV; } dev->caps.qp0_tunnel = kcalloc(dev->caps.num_ports, sizeof (u32), GFP_KERNEL); dev->caps.qp0_proxy = kcalloc(dev->caps.num_ports, sizeof (u32), GFP_KERNEL); dev->caps.qp1_tunnel = kcalloc(dev->caps.num_ports, sizeof (u32), GFP_KERNEL); dev->caps.qp1_proxy = kcalloc(dev->caps.num_ports, sizeof (u32), GFP_KERNEL); if (!dev->caps.qp0_tunnel || !dev->caps.qp0_proxy || !dev->caps.qp1_tunnel || !dev->caps.qp1_proxy) { err = -ENOMEM; goto err_mem; } for (i = 1; i <= dev->caps.num_ports; ++i) { err = mlx4_QUERY_FUNC_CAP(dev, (u32) i, &func_cap); if (err) { mlx4_err(dev, "QUERY_FUNC_CAP port command failed for" " port %d, aborting (%d).\n", i, err); goto err_mem; } dev->caps.qp0_tunnel[i - 1] = func_cap.qp0_tunnel_qpn; dev->caps.qp0_proxy[i - 1] = func_cap.qp0_proxy_qpn; dev->caps.qp1_tunnel[i - 1] = func_cap.qp1_tunnel_qpn; dev->caps.qp1_proxy[i - 1] = func_cap.qp1_proxy_qpn; dev->caps.def_counter_index[i - 1] = func_cap.def_counter_index; dev->caps.port_mask[i] = dev->caps.port_type[i]; err = mlx4_get_slave_pkey_gid_tbl_len(dev, i, &dev->caps.gid_table_len[i], &dev->caps.pkey_table_len[i]); if (err) goto err_mem; } if (dev->caps.uar_page_size * (dev->caps.num_uars - dev->caps.reserved_uars) > pci_resource_len(dev->pdev, 2)) { mlx4_err(dev, "HCA reported UAR region size of 0x%x bigger than " "PCI resource 2 size of 0x%llx, aborting.\n", dev->caps.uar_page_size * dev->caps.num_uars, (unsigned long long) pci_resource_len(dev->pdev, 2)); err = -ENOMEM; goto err_mem; } if (hca_param.dev_cap_enabled & MLX4_DEV_CAP_64B_EQE_ENABLED) { dev->caps.eqe_size = 64; dev->caps.eqe_factor = 1; } else { dev->caps.eqe_size = 32; dev->caps.eqe_factor = 0; } if (hca_param.dev_cap_enabled & MLX4_DEV_CAP_64B_CQE_ENABLED) { dev->caps.cqe_size = 64; dev->caps.userspace_caps |= MLX4_USER_DEV_CAP_64B_CQE; } else { dev->caps.cqe_size = 32; } dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_TS; mlx4_warn(dev, "Timestamping is not supported in slave mode.\n"); slave_adjust_steering_mode(dev, &dev_cap, &hca_param); return 0; err_mem: kfree(dev->caps.qp0_tunnel); kfree(dev->caps.qp0_proxy); kfree(dev->caps.qp1_tunnel); kfree(dev->caps.qp1_proxy); dev->caps.qp0_tunnel = dev->caps.qp0_proxy = dev->caps.qp1_tunnel = dev->caps.qp1_proxy = NULL; return err; } static void mlx4_request_modules(struct mlx4_dev *dev) { int port; int has_ib_port = false; int has_eth_port = false; #define EN_DRV_NAME "mlx4_en" #define IB_DRV_NAME "mlx4_ib" for (port = 1; port <= dev->caps.num_ports; port++) { if (dev->caps.port_type[port] == MLX4_PORT_TYPE_IB) has_ib_port = true; else if (dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH) has_eth_port = true; } if (has_ib_port) request_module_nowait(IB_DRV_NAME); if (has_eth_port) request_module_nowait(EN_DRV_NAME); } /* * Change the port configuration of the device. * Every user of this function must hold the port mutex. */ int mlx4_change_port_types(struct mlx4_dev *dev, enum mlx4_port_type *port_types) { int err = 0; int change = 0; int port; for (port = 0; port < dev->caps.num_ports; port++) { /* Change the port type only if the new type is different * from the current, and not set to Auto */ if (port_types[port] != dev->caps.port_type[port + 1]) change = 1; } if (change) { mlx4_unregister_device(dev); for (port = 1; port <= dev->caps.num_ports; port++) { mlx4_CLOSE_PORT(dev, port); dev->caps.port_type[port] = port_types[port - 1]; err = mlx4_SET_PORT(dev, port, -1); if (err) { mlx4_err(dev, "Failed to set port %d, " "aborting\n", port); goto out; } } mlx4_set_port_mask(dev); err = mlx4_register_device(dev); if (err) { mlx4_err(dev, "Failed to register device\n"); goto out; } mlx4_request_modules(dev); } out: return err; } static ssize_t show_port_type(struct device *dev, struct device_attribute *attr, char *buf) { struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info, port_attr); struct mlx4_dev *mdev = info->dev; char type[8]; sprintf(type, "%s", (mdev->caps.port_type[info->port] == MLX4_PORT_TYPE_IB) ? "ib" : "eth"); if (mdev->caps.possible_type[info->port] == MLX4_PORT_TYPE_AUTO) sprintf(buf, "auto (%s)\n", type); else sprintf(buf, "%s\n", type); return strlen(buf); } static ssize_t set_port_type(struct device *dev, struct device_attribute *attr, const char *buf, size_t count) { struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info, port_attr); struct mlx4_dev *mdev = info->dev; struct mlx4_priv *priv = mlx4_priv(mdev); enum mlx4_port_type types[MLX4_MAX_PORTS]; enum mlx4_port_type new_types[MLX4_MAX_PORTS]; int i; int err = 0; if (!strcmp(buf, "ib\n")) info->tmp_type = MLX4_PORT_TYPE_IB; else if (!strcmp(buf, "eth\n")) info->tmp_type = MLX4_PORT_TYPE_ETH; else if (!strcmp(buf, "auto\n")) info->tmp_type = MLX4_PORT_TYPE_AUTO; else { mlx4_err(mdev, "%s is not supported port type\n", buf); return -EINVAL; } if ((info->tmp_type & mdev->caps.supported_type[info->port]) != info->tmp_type) { mlx4_err(mdev, "Requested port type for port %d is not supported on this HCA\n", info->port); return -EINVAL; } mlx4_stop_sense(mdev); mutex_lock(&priv->port_mutex); /* Possible type is always the one that was delivered */ mdev->caps.possible_type[info->port] = info->tmp_type; for (i = 0; i < mdev->caps.num_ports; i++) { types[i] = priv->port[i+1].tmp_type ? priv->port[i+1].tmp_type : mdev->caps.possible_type[i+1]; if (types[i] == MLX4_PORT_TYPE_AUTO) types[i] = mdev->caps.port_type[i+1]; } if (!(mdev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP) && !(mdev->caps.flags & MLX4_DEV_CAP_FLAG_SENSE_SUPPORT)) { for (i = 1; i <= mdev->caps.num_ports; i++) { if (mdev->caps.possible_type[i] == MLX4_PORT_TYPE_AUTO) { mdev->caps.possible_type[i] = mdev->caps.port_type[i]; err = -EINVAL; } } } if (err) { mlx4_err(mdev, "Auto sensing is not supported on this HCA. " "Set only 'eth' or 'ib' for both ports " "(should be the same)\n"); goto out; } mlx4_do_sense_ports(mdev, new_types, types); err = mlx4_check_port_params(mdev, new_types); if (err) goto out; /* We are about to apply the changes after the configuration * was verified, no need to remember the temporary types * any more */ for (i = 0; i < mdev->caps.num_ports; i++) priv->port[i + 1].tmp_type = 0; err = mlx4_change_port_types(mdev, new_types); out: mlx4_start_sense(mdev); mutex_unlock(&priv->port_mutex); return err ? err : count; } enum ibta_mtu { IB_MTU_256 = 1, IB_MTU_512 = 2, IB_MTU_1024 = 3, IB_MTU_2048 = 4, IB_MTU_4096 = 5 }; static inline int int_to_ibta_mtu(int mtu) { switch (mtu) { case 256: return IB_MTU_256; case 512: return IB_MTU_512; case 1024: return IB_MTU_1024; case 2048: return IB_MTU_2048; case 4096: return IB_MTU_4096; default: return -1; } } static inline int ibta_mtu_to_int(enum ibta_mtu mtu) { switch (mtu) { case IB_MTU_256: return 256; case IB_MTU_512: return 512; case IB_MTU_1024: return 1024; case IB_MTU_2048: return 2048; case IB_MTU_4096: return 4096; default: return -1; } } static ssize_t show_port_ib_mtu(struct device *dev, struct device_attribute *attr, char *buf) { struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info, port_mtu_attr); struct mlx4_dev *mdev = info->dev; + /* When port type is eth, port mtu value isn't used. */ if (mdev->caps.port_type[info->port] == MLX4_PORT_TYPE_ETH) - mlx4_warn(mdev, "port level mtu is only used for IB ports\n"); + return -EINVAL; sprintf(buf, "%d\n", ibta_mtu_to_int(mdev->caps.port_ib_mtu[info->port])); return strlen(buf); } static ssize_t set_port_ib_mtu(struct device *dev, struct device_attribute *attr, const char *buf, size_t count) { struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info, port_mtu_attr); struct mlx4_dev *mdev = info->dev; struct mlx4_priv *priv = mlx4_priv(mdev); int err, port, mtu, ibta_mtu = -1; if (mdev->caps.port_type[info->port] == MLX4_PORT_TYPE_ETH) { mlx4_warn(mdev, "port level mtu is only used for IB ports\n"); return -EINVAL; } mtu = (int) simple_strtol(buf, NULL, 0); ibta_mtu = int_to_ibta_mtu(mtu); if (ibta_mtu < 0) { mlx4_err(mdev, "%s is invalid IBTA mtu\n", buf); return -EINVAL; } mdev->caps.port_ib_mtu[info->port] = ibta_mtu; mlx4_stop_sense(mdev); mutex_lock(&priv->port_mutex); mlx4_unregister_device(mdev); for (port = 1; port <= mdev->caps.num_ports; port++) { mlx4_CLOSE_PORT(mdev, port); err = mlx4_SET_PORT(mdev, port, -1); if (err) { mlx4_err(mdev, "Failed to set port %d, " "aborting\n", port); goto err_set_port; } } err = mlx4_register_device(mdev); err_set_port: mutex_unlock(&priv->port_mutex); mlx4_start_sense(mdev); return err ? err : count; } static int mlx4_load_fw(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); int err, unmap_flag = 0; priv->fw.fw_icm = mlx4_alloc_icm(dev, priv->fw.fw_pages, GFP_HIGHUSER | __GFP_NOWARN, 0); if (!priv->fw.fw_icm) { mlx4_err(dev, "Couldn't allocate FW area, aborting.\n"); return -ENOMEM; } err = mlx4_MAP_FA(dev, priv->fw.fw_icm); if (err) { mlx4_err(dev, "MAP_FA command failed, aborting.\n"); goto err_free; } err = mlx4_RUN_FW(dev); if (err) { mlx4_err(dev, "RUN_FW command failed, aborting.\n"); goto err_unmap_fa; } return 0; err_unmap_fa: unmap_flag = mlx4_UNMAP_FA(dev); if (unmap_flag) pr_warn("mlx4_core: mlx4_UNMAP_FA failed.\n"); err_free: if (!unmap_flag) mlx4_free_icm(dev, priv->fw.fw_icm, 0); return err; } static int mlx4_init_cmpt_table(struct mlx4_dev *dev, u64 cmpt_base, int cmpt_entry_sz) { struct mlx4_priv *priv = mlx4_priv(dev); int err; int num_eqs; err = mlx4_init_icm_table(dev, &priv->qp_table.cmpt_table, cmpt_base + ((u64) (MLX4_CMPT_TYPE_QP * cmpt_entry_sz) << MLX4_CMPT_SHIFT), cmpt_entry_sz, dev->caps.num_qps, dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW], 0, 0); if (err) goto err; err = mlx4_init_icm_table(dev, &priv->srq_table.cmpt_table, cmpt_base + ((u64) (MLX4_CMPT_TYPE_SRQ * cmpt_entry_sz) << MLX4_CMPT_SHIFT), cmpt_entry_sz, dev->caps.num_srqs, dev->caps.reserved_srqs, 0, 0); if (err) goto err_qp; err = mlx4_init_icm_table(dev, &priv->cq_table.cmpt_table, cmpt_base + ((u64) (MLX4_CMPT_TYPE_CQ * cmpt_entry_sz) << MLX4_CMPT_SHIFT), cmpt_entry_sz, dev->caps.num_cqs, dev->caps.reserved_cqs, 0, 0); if (err) goto err_srq; num_eqs = (mlx4_is_master(dev)) ? dev->phys_caps.num_phys_eqs : dev->caps.num_eqs; err = mlx4_init_icm_table(dev, &priv->eq_table.cmpt_table, cmpt_base + ((u64) (MLX4_CMPT_TYPE_EQ * cmpt_entry_sz) << MLX4_CMPT_SHIFT), cmpt_entry_sz, num_eqs, num_eqs, 0, 0); if (err) goto err_cq; return 0; err_cq: mlx4_cleanup_icm_table(dev, &priv->cq_table.cmpt_table); err_srq: mlx4_cleanup_icm_table(dev, &priv->srq_table.cmpt_table); err_qp: mlx4_cleanup_icm_table(dev, &priv->qp_table.cmpt_table); err: return err; } static int mlx4_init_icm(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap, struct mlx4_init_hca_param *init_hca, u64 icm_size) { struct mlx4_priv *priv = mlx4_priv(dev); u64 aux_pages; int num_eqs; int err, unmap_flag = 0; err = mlx4_SET_ICM_SIZE(dev, icm_size, &aux_pages); if (err) { mlx4_err(dev, "SET_ICM_SIZE command failed, aborting.\n"); return err; } mlx4_dbg(dev, "%lld KB of HCA context requires %lld KB aux memory.\n", (unsigned long long) icm_size >> 10, (unsigned long long) aux_pages << 2); priv->fw.aux_icm = mlx4_alloc_icm(dev, aux_pages, GFP_HIGHUSER | __GFP_NOWARN, 0); if (!priv->fw.aux_icm) { mlx4_err(dev, "Couldn't allocate aux memory, aborting.\n"); return -ENOMEM; } err = mlx4_MAP_ICM_AUX(dev, priv->fw.aux_icm); if (err) { mlx4_err(dev, "MAP_ICM_AUX command failed, aborting.\n"); goto err_free_aux; } err = mlx4_init_cmpt_table(dev, init_hca->cmpt_base, dev_cap->cmpt_entry_sz); if (err) { mlx4_err(dev, "Failed to map cMPT context memory, aborting.\n"); goto err_unmap_aux; } num_eqs = (mlx4_is_master(dev)) ? dev->phys_caps.num_phys_eqs : dev->caps.num_eqs; err = mlx4_init_icm_table(dev, &priv->eq_table.table, init_hca->eqc_base, dev_cap->eqc_entry_sz, num_eqs, num_eqs, 0, 0); if (err) { mlx4_err(dev, "Failed to map EQ context memory, aborting.\n"); goto err_unmap_cmpt; } /* * Reserved MTT entries must be aligned up to a cacheline * boundary, since the FW will write to them, while the driver * writes to all other MTT entries. (The variable * dev->caps.mtt_entry_sz below is really the MTT segment * size, not the raw entry size) */ dev->caps.reserved_mtts = ALIGN(dev->caps.reserved_mtts * dev->caps.mtt_entry_sz, dma_get_cache_alignment()) / dev->caps.mtt_entry_sz; err = mlx4_init_icm_table(dev, &priv->mr_table.mtt_table, init_hca->mtt_base, dev->caps.mtt_entry_sz, dev->caps.num_mtts, dev->caps.reserved_mtts, 1, 0); if (err) { mlx4_err(dev, "Failed to map MTT context memory, aborting.\n"); goto err_unmap_eq; } err = mlx4_init_icm_table(dev, &priv->mr_table.dmpt_table, init_hca->dmpt_base, dev_cap->dmpt_entry_sz, dev->caps.num_mpts, dev->caps.reserved_mrws, 1, 1); if (err) { mlx4_err(dev, "Failed to map dMPT context memory, aborting.\n"); goto err_unmap_mtt; } err = mlx4_init_icm_table(dev, &priv->qp_table.qp_table, init_hca->qpc_base, dev_cap->qpc_entry_sz, dev->caps.num_qps, dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW], 0, 0); if (err) { mlx4_err(dev, "Failed to map QP context memory, aborting.\n"); goto err_unmap_dmpt; } err = mlx4_init_icm_table(dev, &priv->qp_table.auxc_table, init_hca->auxc_base, dev_cap->aux_entry_sz, dev->caps.num_qps, dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW], 0, 0); if (err) { mlx4_err(dev, "Failed to map AUXC context memory, aborting.\n"); goto err_unmap_qp; } err = mlx4_init_icm_table(dev, &priv->qp_table.altc_table, init_hca->altc_base, dev_cap->altc_entry_sz, dev->caps.num_qps, dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW], 0, 0); if (err) { mlx4_err(dev, "Failed to map ALTC context memory, aborting.\n"); goto err_unmap_auxc; } err = mlx4_init_icm_table(dev, &priv->qp_table.rdmarc_table, init_hca->rdmarc_base, dev_cap->rdmarc_entry_sz << priv->qp_table.rdmarc_shift, dev->caps.num_qps, dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW], 0, 0); if (err) { mlx4_err(dev, "Failed to map RDMARC context memory, aborting\n"); goto err_unmap_altc; } err = mlx4_init_icm_table(dev, &priv->cq_table.table, init_hca->cqc_base, dev_cap->cqc_entry_sz, dev->caps.num_cqs, dev->caps.reserved_cqs, 0, 0); if (err) { mlx4_err(dev, "Failed to map CQ context memory, aborting.\n"); goto err_unmap_rdmarc; } err = mlx4_init_icm_table(dev, &priv->srq_table.table, init_hca->srqc_base, dev_cap->srq_entry_sz, dev->caps.num_srqs, dev->caps.reserved_srqs, 0, 0); if (err) { mlx4_err(dev, "Failed to map SRQ context memory, aborting.\n"); goto err_unmap_cq; } /* * For flow steering device managed mode it is required to use * mlx4_init_icm_table. For B0 steering mode it's not strictly * required, but for simplicity just map the whole multicast * group table now. The table isn't very big and it's a lot * easier than trying to track ref counts. */ err = mlx4_init_icm_table(dev, &priv->mcg_table.table, init_hca->mc_base, mlx4_get_mgm_entry_size(dev), dev->caps.num_mgms + dev->caps.num_amgms, dev->caps.num_mgms + dev->caps.num_amgms, 0, 0); if (err) { mlx4_err(dev, "Failed to map MCG context memory, aborting.\n"); goto err_unmap_srq; } return 0; err_unmap_srq: mlx4_cleanup_icm_table(dev, &priv->srq_table.table); err_unmap_cq: mlx4_cleanup_icm_table(dev, &priv->cq_table.table); err_unmap_rdmarc: mlx4_cleanup_icm_table(dev, &priv->qp_table.rdmarc_table); err_unmap_altc: mlx4_cleanup_icm_table(dev, &priv->qp_table.altc_table); err_unmap_auxc: mlx4_cleanup_icm_table(dev, &priv->qp_table.auxc_table); err_unmap_qp: mlx4_cleanup_icm_table(dev, &priv->qp_table.qp_table); err_unmap_dmpt: mlx4_cleanup_icm_table(dev, &priv->mr_table.dmpt_table); err_unmap_mtt: mlx4_cleanup_icm_table(dev, &priv->mr_table.mtt_table); err_unmap_eq: mlx4_cleanup_icm_table(dev, &priv->eq_table.table); err_unmap_cmpt: mlx4_cleanup_icm_table(dev, &priv->eq_table.cmpt_table); mlx4_cleanup_icm_table(dev, &priv->cq_table.cmpt_table); mlx4_cleanup_icm_table(dev, &priv->srq_table.cmpt_table); mlx4_cleanup_icm_table(dev, &priv->qp_table.cmpt_table); err_unmap_aux: unmap_flag = mlx4_UNMAP_ICM_AUX(dev); if (unmap_flag) pr_warn("mlx4_core: mlx4_UNMAP_ICM_AUX failed.\n"); err_free_aux: if (!unmap_flag) mlx4_free_icm(dev, priv->fw.aux_icm, 0); return err; } static void mlx4_free_icms(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); mlx4_cleanup_icm_table(dev, &priv->mcg_table.table); mlx4_cleanup_icm_table(dev, &priv->srq_table.table); mlx4_cleanup_icm_table(dev, &priv->cq_table.table); mlx4_cleanup_icm_table(dev, &priv->qp_table.rdmarc_table); mlx4_cleanup_icm_table(dev, &priv->qp_table.altc_table); mlx4_cleanup_icm_table(dev, &priv->qp_table.auxc_table); mlx4_cleanup_icm_table(dev, &priv->qp_table.qp_table); mlx4_cleanup_icm_table(dev, &priv->mr_table.dmpt_table); mlx4_cleanup_icm_table(dev, &priv->mr_table.mtt_table); mlx4_cleanup_icm_table(dev, &priv->eq_table.table); mlx4_cleanup_icm_table(dev, &priv->eq_table.cmpt_table); mlx4_cleanup_icm_table(dev, &priv->cq_table.cmpt_table); mlx4_cleanup_icm_table(dev, &priv->srq_table.cmpt_table); mlx4_cleanup_icm_table(dev, &priv->qp_table.cmpt_table); if (!mlx4_UNMAP_ICM_AUX(dev)) mlx4_free_icm(dev, priv->fw.aux_icm, 0); else pr_warn("mlx4_core: mlx4_UNMAP_ICM_AUX failed.\n"); } static void mlx4_slave_exit(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); mutex_lock(&priv->cmd.slave_cmd_mutex); if (mlx4_comm_cmd(dev, MLX4_COMM_CMD_RESET, 0, MLX4_COMM_TIME)) mlx4_warn(dev, "Failed to close slave function.\n"); mutex_unlock(&priv->cmd.slave_cmd_mutex); } static int map_bf_area(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); resource_size_t bf_start; resource_size_t bf_len; int err = 0; if (!dev->caps.bf_reg_size) return -ENXIO; bf_start = pci_resource_start(dev->pdev, 2) + (dev->caps.num_uars << PAGE_SHIFT); bf_len = pci_resource_len(dev->pdev, 2) - (dev->caps.num_uars << PAGE_SHIFT); priv->bf_mapping = io_mapping_create_wc(bf_start, bf_len); if (!priv->bf_mapping) err = -ENOMEM; return err; } static void unmap_bf_area(struct mlx4_dev *dev) { if (mlx4_priv(dev)->bf_mapping) io_mapping_free(mlx4_priv(dev)->bf_mapping); } int mlx4_read_clock(struct mlx4_dev *dev) { u32 clockhi, clocklo, clockhi1; cycle_t cycles; int i; struct mlx4_priv *priv = mlx4_priv(dev); if (!priv->clock_mapping) return -ENOTSUPP; for (i = 0; i < 10; i++) { clockhi = swab32(readl(priv->clock_mapping)); clocklo = swab32(readl(priv->clock_mapping + 4)); clockhi1 = swab32(readl(priv->clock_mapping)); if (clockhi == clockhi1) break; } cycles = (u64) clockhi << 32 | (u64) clocklo; return cycles; } EXPORT_SYMBOL_GPL(mlx4_read_clock); static int map_internal_clock(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); priv->clock_mapping = ioremap(pci_resource_start(dev->pdev, priv->fw.clock_bar) + priv->fw.clock_offset, MLX4_CLOCK_SIZE); if (!priv->clock_mapping) return -ENOMEM; return 0; } int mlx4_get_internal_clock_params(struct mlx4_dev *dev, struct mlx4_clock_params *params) { struct mlx4_priv *priv = mlx4_priv(dev); if (mlx4_is_slave(dev)) return -ENOTSUPP; if (!params) return -EINVAL; params->bar = priv->fw.clock_bar; params->offset = priv->fw.clock_offset; params->size = MLX4_CLOCK_SIZE; return 0; } EXPORT_SYMBOL_GPL(mlx4_get_internal_clock_params); static void unmap_internal_clock(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); if (priv->clock_mapping) iounmap(priv->clock_mapping); } static void mlx4_close_hca(struct mlx4_dev *dev) { unmap_internal_clock(dev); unmap_bf_area(dev); if (mlx4_is_slave(dev)) { mlx4_slave_exit(dev); } else { mlx4_CLOSE_HCA(dev, 0); mlx4_free_icms(dev); if (!mlx4_UNMAP_FA(dev)) mlx4_free_icm(dev, mlx4_priv(dev)->fw.fw_icm, 0); else pr_warn("mlx4_core: mlx4_UNMAP_FA failed.\n"); } } static int mlx4_init_slave(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); u64 dma = (u64) priv->mfunc.vhcr_dma; int num_of_reset_retries = NUM_OF_RESET_RETRIES; int ret_from_reset = 0; u32 slave_read; u32 cmd_channel_ver; mutex_lock(&priv->cmd.slave_cmd_mutex); priv->cmd.max_cmds = 1; mlx4_warn(dev, "Sending reset\n"); ret_from_reset = mlx4_comm_cmd(dev, MLX4_COMM_CMD_RESET, 0, MLX4_COMM_TIME); /* if we are in the middle of flr the slave will try * NUM_OF_RESET_RETRIES times before leaving.*/ if (ret_from_reset) { if (MLX4_DELAY_RESET_SLAVE == ret_from_reset) { msleep(SLEEP_TIME_IN_RESET); while (ret_from_reset && num_of_reset_retries) { mlx4_warn(dev, "slave is currently in the" "middle of FLR. retrying..." "(try num:%d)\n", (NUM_OF_RESET_RETRIES - num_of_reset_retries + 1)); ret_from_reset = mlx4_comm_cmd(dev, MLX4_COMM_CMD_RESET, 0, MLX4_COMM_TIME); num_of_reset_retries = num_of_reset_retries - 1; } } else goto err; } /* check the driver version - the slave I/F revision * must match the master's */ slave_read = swab32(readl(&priv->mfunc.comm->slave_read)); cmd_channel_ver = mlx4_comm_get_version(); if (MLX4_COMM_GET_IF_REV(cmd_channel_ver) != MLX4_COMM_GET_IF_REV(slave_read)) { mlx4_err(dev, "slave driver version is not supported" " by the master\n"); goto err; } mlx4_warn(dev, "Sending vhcr0\n"); if (mlx4_comm_cmd(dev, MLX4_COMM_CMD_VHCR0, dma >> 48, MLX4_COMM_TIME)) goto err; if (mlx4_comm_cmd(dev, MLX4_COMM_CMD_VHCR1, dma >> 32, MLX4_COMM_TIME)) goto err; if (mlx4_comm_cmd(dev, MLX4_COMM_CMD_VHCR2, dma >> 16, MLX4_COMM_TIME)) goto err; if (mlx4_comm_cmd(dev, MLX4_COMM_CMD_VHCR_EN, dma, MLX4_COMM_TIME)) goto err; mutex_unlock(&priv->cmd.slave_cmd_mutex); return 0; err: mlx4_comm_cmd(dev, MLX4_COMM_CMD_RESET, 0, 0); mutex_unlock(&priv->cmd.slave_cmd_mutex); return -EIO; } static void mlx4_parav_master_pf_caps(struct mlx4_dev *dev) { int i; for (i = 1; i <= dev->caps.num_ports; i++) { if (dev->caps.port_type[i] == MLX4_PORT_TYPE_ETH) dev->caps.gid_table_len[i] = mlx4_get_slave_num_gids(dev, 0); else dev->caps.gid_table_len[i] = 1; dev->caps.pkey_table_len[i] = dev->phys_caps.pkey_phys_table_len[i] - 1; } } static int choose_log_fs_mgm_entry_size(int qp_per_entry) { int i = MLX4_MIN_MGM_LOG_ENTRY_SIZE; for (i = MLX4_MIN_MGM_LOG_ENTRY_SIZE; i <= MLX4_MAX_MGM_LOG_ENTRY_SIZE; i++) { if (qp_per_entry <= 4 * ((1 << i) / 16 - 2)) break; } return (i <= MLX4_MAX_MGM_LOG_ENTRY_SIZE) ? i : -1; } static void choose_steering_mode(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) { int nvfs; mlx4_get_val(num_vfs.dbdf2val.tbl, pci_physfn(dev->pdev), 0, &nvfs); if (high_rate_steer && !mlx4_is_mfunc(dev)) { dev->caps.flags &= ~(MLX4_DEV_CAP_FLAG_VEP_MC_STEER | MLX4_DEV_CAP_FLAG_VEP_UC_STEER); dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_FS_EN; } if (mlx4_log_num_mgm_entry_size == -1 && dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_FS_EN && (!mlx4_is_mfunc(dev) || (dev_cap->fs_max_num_qp_per_entry >= (nvfs + 1))) && choose_log_fs_mgm_entry_size(dev_cap->fs_max_num_qp_per_entry) >= MLX4_MIN_MGM_LOG_ENTRY_SIZE) { dev->oper_log_mgm_entry_size = choose_log_fs_mgm_entry_size(dev_cap->fs_max_num_qp_per_entry); dev->caps.steering_mode = MLX4_STEERING_MODE_DEVICE_MANAGED; dev->caps.num_qp_per_mgm = dev_cap->fs_max_num_qp_per_entry; } else { if (dev->caps.flags & MLX4_DEV_CAP_FLAG_VEP_UC_STEER && dev->caps.flags & MLX4_DEV_CAP_FLAG_VEP_MC_STEER) dev->caps.steering_mode = MLX4_STEERING_MODE_B0; else { dev->caps.steering_mode = MLX4_STEERING_MODE_A0; if (dev->caps.flags & MLX4_DEV_CAP_FLAG_VEP_UC_STEER || dev->caps.flags & MLX4_DEV_CAP_FLAG_VEP_MC_STEER) mlx4_warn(dev, "Must have both UC_STEER and MC_STEER flags " "set to use B0 steering. Falling back to A0 steering mode.\n"); } dev->oper_log_mgm_entry_size = mlx4_log_num_mgm_entry_size > 0 ? mlx4_log_num_mgm_entry_size : MLX4_DEFAULT_MGM_LOG_ENTRY_SIZE; dev->caps.num_qp_per_mgm = mlx4_get_qp_per_mgm(dev); } mlx4_dbg(dev, "Steering mode is: %s, oper_log_mgm_entry_size = %d, " "log_num_mgm_entry_size = %d\n", mlx4_steering_mode_str(dev->caps.steering_mode), dev->oper_log_mgm_entry_size, mlx4_log_num_mgm_entry_size); } static int mlx4_init_hca(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_dev_cap *dev_cap = NULL; struct mlx4_adapter adapter; struct mlx4_mod_stat_cfg mlx4_cfg; struct mlx4_profile profile; struct mlx4_init_hca_param init_hca; u64 icm_size; int err; if (!mlx4_is_slave(dev)) { err = mlx4_QUERY_FW(dev); if (err) { if (err == -EACCES) mlx4_info(dev, "non-primary physical function, skipping.\n"); else mlx4_err(dev, "QUERY_FW command failed, aborting.\n"); return err; } err = mlx4_load_fw(dev); if (err) { mlx4_err(dev, "Failed to start FW, aborting.\n"); return err; } mlx4_cfg.log_pg_sz_m = 1; mlx4_cfg.log_pg_sz = 0; err = mlx4_MOD_STAT_CFG(dev, &mlx4_cfg); if (err) mlx4_warn(dev, "Failed to override log_pg_sz parameter\n"); dev_cap = kzalloc(sizeof *dev_cap, GFP_KERNEL); if (!dev_cap) { mlx4_err(dev, "Failed to allocate memory for dev_cap\n"); err = -ENOMEM; goto err_stop_fw; } err = mlx4_dev_cap(dev, dev_cap); if (err) { mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting.\n"); goto err_stop_fw; } choose_steering_mode(dev, dev_cap); if (mlx4_is_master(dev)) mlx4_parav_master_pf_caps(dev); process_mod_param_profile(&profile); if (dev->caps.steering_mode == MLX4_STEERING_MODE_DEVICE_MANAGED) profile.num_mcg = MLX4_FS_NUM_MCG; icm_size = mlx4_make_profile(dev, &profile, dev_cap, &init_hca); if ((long long) icm_size < 0) { err = icm_size; goto err_stop_fw; } dev->caps.max_fmr_maps = (1 << (32 - ilog2(dev->caps.num_mpts))) - 1; init_hca.log_uar_sz = ilog2(dev->caps.num_uars); init_hca.uar_page_sz = PAGE_SHIFT - 12; err = mlx4_init_icm(dev, dev_cap, &init_hca, icm_size); if (err) goto err_stop_fw; init_hca.mw_enable = 1; err = mlx4_INIT_HCA(dev, &init_hca); if (err) { mlx4_err(dev, "INIT_HCA command failed, aborting.\n"); goto err_free_icm; } /* * Read HCA frequency by QUERY_HCA command */ if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_TS) { memset(&init_hca, 0, sizeof(init_hca)); err = mlx4_QUERY_HCA(dev, &init_hca); if (err) { mlx4_err(dev, "QUERY_HCA command failed, disable timestamp.\n"); dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_TS; } else { dev->caps.hca_core_clock = init_hca.hca_core_clock; } /* In case we got HCA frequency 0 - disable timestamping * to avoid dividing by zero */ if (!dev->caps.hca_core_clock) { dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_TS; mlx4_err(dev, "HCA frequency is 0. Timestamping is not supported."); } else if (map_internal_clock(dev)) { /* Map internal clock, * in case of failure disable timestamping */ dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_TS; mlx4_err(dev, "Failed to map internal clock. Timestamping is not supported.\n"); } } } else { err = mlx4_init_slave(dev); if (err) { mlx4_err(dev, "Failed to initialize slave\n"); return err; } err = mlx4_slave_cap(dev); if (err) { mlx4_err(dev, "Failed to obtain slave caps\n"); goto err_close; } } if (map_bf_area(dev)) mlx4_dbg(dev, "Failed to map blue flame area\n"); /* Only the master set the ports, all the rest got it from it.*/ if (!mlx4_is_slave(dev)) mlx4_set_port_mask(dev); err = mlx4_QUERY_ADAPTER(dev, &adapter); if (err) { mlx4_err(dev, "QUERY_ADAPTER command failed, aborting.\n"); goto unmap_bf; } priv->eq_table.inta_pin = adapter.inta_pin; memcpy(dev->board_id, adapter.board_id, sizeof dev->board_id); memcpy(dev->vsd, adapter.vsd, sizeof(dev->vsd)); dev->vsd_vendor_id = adapter.vsd_vendor_id; if (!mlx4_is_slave(dev)) kfree(dev_cap); return 0; unmap_bf: if (!mlx4_is_slave(dev)) unmap_internal_clock(dev); unmap_bf_area(dev); if (mlx4_is_slave(dev)) { kfree(dev->caps.qp0_tunnel); kfree(dev->caps.qp0_proxy); kfree(dev->caps.qp1_tunnel); kfree(dev->caps.qp1_proxy); } err_close: if (mlx4_is_slave(dev)) mlx4_slave_exit(dev); else mlx4_CLOSE_HCA(dev, 0); err_free_icm: if (!mlx4_is_slave(dev)) mlx4_free_icms(dev); err_stop_fw: if (!mlx4_is_slave(dev)) { if (!mlx4_UNMAP_FA(dev)) mlx4_free_icm(dev, priv->fw.fw_icm, 0); else pr_warn("mlx4_core: mlx4_UNMAP_FA failed.\n"); kfree(dev_cap); } return err; } static int mlx4_init_counters_table(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); int nent_pow2, port_indx, vf_index, num_counters; int res, index = 0; struct counter_index *new_counter_index; if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_COUNTERS)) return -ENOENT; if (!mlx4_is_slave(dev) && dev->caps.max_counters == dev->caps.max_extended_counters) { res = mlx4_cmd(dev, MLX4_IF_STATE_EXTENDED, 0, 0, MLX4_CMD_SET_IF_STAT, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE); if (res) { mlx4_err(dev, "Failed to set extended counters (err=%d)\n", res); return res; } } mutex_init(&priv->counters_table.mutex); if (mlx4_is_slave(dev)) { for (port_indx = 0; port_indx < dev->caps.num_ports; port_indx++) { INIT_LIST_HEAD(&priv->counters_table.global_port_list[port_indx]); if (dev->caps.def_counter_index[port_indx] != 0xFF) { new_counter_index = kmalloc(sizeof(struct counter_index), GFP_KERNEL); if (!new_counter_index) return -ENOMEM; new_counter_index->index = dev->caps.def_counter_index[port_indx]; list_add_tail(&new_counter_index->list, &priv->counters_table.global_port_list[port_indx]); } } mlx4_dbg(dev, "%s: slave allocated %d counters for %d ports\n", __func__, dev->caps.num_ports, dev->caps.num_ports); return 0; } nent_pow2 = roundup_pow_of_two(dev->caps.max_counters); for (port_indx = 0; port_indx < dev->caps.num_ports; port_indx++) { INIT_LIST_HEAD(&priv->counters_table.global_port_list[port_indx]); /* allocating 2 counters per port for PFs */ /* For the PF, the ETH default counters are 0,2; */ /* and the RoCE default counters are 1,3 */ for (num_counters = 0; num_counters < 2; num_counters++, index++) { new_counter_index = kmalloc(sizeof(struct counter_index), GFP_KERNEL); if (!new_counter_index) return -ENOMEM; new_counter_index->index = index; list_add_tail(&new_counter_index->list, &priv->counters_table.global_port_list[port_indx]); } } if (mlx4_is_master(dev)) { for (vf_index = 0; vf_index < dev->num_vfs; vf_index++) { for (port_indx = 0; port_indx < dev->caps.num_ports; port_indx++) { INIT_LIST_HEAD(&priv->counters_table.vf_list[vf_index][port_indx]); new_counter_index = kmalloc(sizeof(struct counter_index), GFP_KERNEL); if (!new_counter_index) return -ENOMEM; if (index < nent_pow2 - 2) { new_counter_index->index = index; index++; } else { new_counter_index->index = MLX4_SINK_COUNTER_INDEX; } list_add_tail(&new_counter_index->list, &priv->counters_table.vf_list[vf_index][port_indx]); } } res = mlx4_bitmap_init(&priv->counters_table.bitmap, nent_pow2, nent_pow2 - 1, index, 1); mlx4_dbg(dev, "%s: master allocated %d counters for %d VFs\n", __func__, index, dev->num_vfs); } else { res = mlx4_bitmap_init(&priv->counters_table.bitmap, nent_pow2, nent_pow2 - 1, index, 1); mlx4_dbg(dev, "%s: native allocated %d counters for %d ports\n", __func__, index, dev->caps.num_ports); } return 0; } static void mlx4_cleanup_counters_table(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); int i, j; struct counter_index *port, *tmp_port; struct counter_index *vf, *tmp_vf; mutex_lock(&priv->counters_table.mutex); if (dev->caps.flags & MLX4_DEV_CAP_FLAG_COUNTERS) { for (i = 0; i < dev->caps.num_ports; i++) { list_for_each_entry_safe(port, tmp_port, &priv->counters_table.global_port_list[i], list) { list_del(&port->list); kfree(port); } } if (!mlx4_is_slave(dev)) { for (i = 0; i < dev->num_vfs; i++) { for (j = 0; j < dev->caps.num_ports; j++) { list_for_each_entry_safe(vf, tmp_vf, &priv->counters_table.vf_list[i][j], list) { /* clear the counter statistic */ if (__mlx4_clear_if_stat(dev, vf->index)) mlx4_dbg(dev, "%s: reset counter %d failed\n", __func__, vf->index); list_del(&vf->list); kfree(vf); } } } mlx4_bitmap_cleanup(&priv->counters_table.bitmap); } } mutex_unlock(&priv->counters_table.mutex); } int __mlx4_slave_counters_free(struct mlx4_dev *dev, int slave) { struct mlx4_priv *priv = mlx4_priv(dev); int i, first; struct counter_index *vf, *tmp_vf; /* clean VF's counters for the next useg */ if (slave > 0 && slave <= dev->num_vfs) { mlx4_dbg(dev, "%s: free counters of slave(%d)\n" , __func__, slave); mutex_lock(&priv->counters_table.mutex); for (i = 0; i < dev->caps.num_ports; i++) { first = 0; list_for_each_entry_safe(vf, tmp_vf, &priv->counters_table.vf_list[slave - 1][i], list) { /* clear the counter statistic */ if (__mlx4_clear_if_stat(dev, vf->index)) mlx4_dbg(dev, "%s: reset counter %d failed\n", __func__, vf->index); if (first++ && vf->index != MLX4_SINK_COUNTER_INDEX) { mlx4_dbg(dev, "%s: delete counter index %d for slave %d and port %d\n" , __func__, vf->index, slave, i + 1); mlx4_bitmap_free(&priv->counters_table.bitmap, vf->index, MLX4_USE_RR); list_del(&vf->list); kfree(vf); } else { mlx4_dbg(dev, "%s: can't delete default counter index %d for slave %d and port %d\n" , __func__, vf->index, slave, i + 1); } } } mutex_unlock(&priv->counters_table.mutex); } return 0; } int __mlx4_counter_alloc(struct mlx4_dev *dev, int slave, int port, u32 *idx) { struct mlx4_priv *priv = mlx4_priv(dev); struct counter_index *new_counter_index; if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_COUNTERS)) return -ENOENT; if ((slave > MLX4_MAX_NUM_VF) || (slave < 0) || (port < 0) || (port > MLX4_MAX_PORTS)) { mlx4_dbg(dev, "%s: invalid slave(%d) or port(%d) index\n", __func__, slave, port); return -EINVAL; } /* handle old guest request does not support request by port index */ if (port == 0) { *idx = MLX4_SINK_COUNTER_INDEX; mlx4_dbg(dev, "%s: allocated default counter index %d for slave %d port %d\n" , __func__, *idx, slave, port); return 0; } mutex_lock(&priv->counters_table.mutex); *idx = mlx4_bitmap_alloc(&priv->counters_table.bitmap); /* if no resources return the default counter of the slave and port */ if (*idx == -1) { if (slave == 0) { /* its the ethernet counter ?????? */ new_counter_index = list_entry(priv->counters_table.global_port_list[port - 1].next, struct counter_index, list); } else { new_counter_index = list_entry(priv->counters_table.vf_list[slave - 1][port - 1].next, struct counter_index, list); } *idx = new_counter_index->index; mlx4_dbg(dev, "%s: allocated defualt counter index %d for slave %d port %d\n" , __func__, *idx, slave, port); goto out; } if (slave == 0) { /* native or master */ new_counter_index = kmalloc(sizeof(struct counter_index), GFP_KERNEL); if (!new_counter_index) goto no_mem; new_counter_index->index = *idx; list_add_tail(&new_counter_index->list, &priv->counters_table.global_port_list[port - 1]); } else { new_counter_index = kmalloc(sizeof(struct counter_index), GFP_KERNEL); if (!new_counter_index) goto no_mem; new_counter_index->index = *idx; list_add_tail(&new_counter_index->list, &priv->counters_table.vf_list[slave - 1][port - 1]); } mlx4_dbg(dev, "%s: allocated counter index %d for slave %d port %d\n" , __func__, *idx, slave, port); out: mutex_unlock(&priv->counters_table.mutex); return 0; no_mem: mlx4_bitmap_free(&priv->counters_table.bitmap, *idx, MLX4_USE_RR); mutex_unlock(&priv->counters_table.mutex); *idx = MLX4_SINK_COUNTER_INDEX; mlx4_dbg(dev, "%s: failed err (%d)\n" , __func__, -ENOMEM); return -ENOMEM; } int mlx4_counter_alloc(struct mlx4_dev *dev, u8 port, u32 *idx) { u64 out_param; int err; struct mlx4_priv *priv = mlx4_priv(dev); struct counter_index *new_counter_index, *c_index; if (mlx4_is_mfunc(dev)) { err = mlx4_cmd_imm(dev, 0, &out_param, ((u32) port) << 8 | (u32) RES_COUNTER, RES_OP_RESERVE, MLX4_CMD_ALLOC_RES, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED); if (!err) { *idx = get_param_l(&out_param); if (*idx == MLX4_SINK_COUNTER_INDEX) return -ENOSPC; mutex_lock(&priv->counters_table.mutex); c_index = list_entry(priv->counters_table.global_port_list[port - 1].next, struct counter_index, list); mutex_unlock(&priv->counters_table.mutex); if (c_index->index == *idx) return -EEXIST; if (mlx4_is_slave(dev)) { new_counter_index = kmalloc(sizeof(struct counter_index), GFP_KERNEL); if (!new_counter_index) { mlx4_counter_free(dev, port, *idx); return -ENOMEM; } new_counter_index->index = *idx; mutex_lock(&priv->counters_table.mutex); list_add_tail(&new_counter_index->list, &priv->counters_table.global_port_list[port - 1]); mutex_unlock(&priv->counters_table.mutex); mlx4_dbg(dev, "%s: allocated counter index %d for port %d\n" , __func__, *idx, port); } } return err; } return __mlx4_counter_alloc(dev, 0, port, idx); } EXPORT_SYMBOL_GPL(mlx4_counter_alloc); void __mlx4_counter_free(struct mlx4_dev *dev, int slave, int port, u32 idx) { /* check if native or slave and deletes acordingly */ struct mlx4_priv *priv = mlx4_priv(dev); struct counter_index *pf, *tmp_pf; struct counter_index *vf, *tmp_vf; int first; if (idx == MLX4_SINK_COUNTER_INDEX) { mlx4_dbg(dev, "%s: try to delete default counter index %d for port %d\n" , __func__, idx, port); return; } if ((slave > MLX4_MAX_NUM_VF) || (slave < 0) || (port < 0) || (port > MLX4_MAX_PORTS)) { mlx4_warn(dev, "%s: deletion failed due to invalid slave(%d) or port(%d) index\n" , __func__, slave, idx); return; } mutex_lock(&priv->counters_table.mutex); if (slave == 0) { first = 0; list_for_each_entry_safe(pf, tmp_pf, &priv->counters_table.global_port_list[port - 1], list) { /* the first 2 counters are reserved */ if (pf->index == idx) { /* clear the counter statistic */ if (__mlx4_clear_if_stat(dev, pf->index)) mlx4_dbg(dev, "%s: reset counter %d failed\n", __func__, pf->index); if (1 < first && idx != MLX4_SINK_COUNTER_INDEX) { list_del(&pf->list); kfree(pf); mlx4_dbg(dev, "%s: delete counter index %d for native device (%d) port %d\n" , __func__, idx, slave, port); mlx4_bitmap_free(&priv->counters_table.bitmap, idx, MLX4_USE_RR); goto out; } else { mlx4_dbg(dev, "%s: can't delete default counter index %d for native device (%d) port %d\n" , __func__, idx, slave, port); goto out; } } first++; } mlx4_dbg(dev, "%s: can't delete counter index %d for native device (%d) port %d\n" , __func__, idx, slave, port); } else { first = 0; list_for_each_entry_safe(vf, tmp_vf, &priv->counters_table.vf_list[slave - 1][port - 1], list) { /* the first element is reserved */ if (vf->index == idx) { /* clear the counter statistic */ if (__mlx4_clear_if_stat(dev, vf->index)) mlx4_dbg(dev, "%s: reset counter %d failed\n", __func__, vf->index); if (first) { list_del(&vf->list); kfree(vf); mlx4_dbg(dev, "%s: delete counter index %d for slave %d port %d\n", __func__, idx, slave, port); mlx4_bitmap_free(&priv->counters_table.bitmap, idx, MLX4_USE_RR); goto out; } else { mlx4_dbg(dev, "%s: can't delete default slave (%d) counter index %d for port %d\n" , __func__, slave, idx, port); goto out; } } first++; } mlx4_dbg(dev, "%s: can't delete slave (%d) counter index %d for port %d\n" , __func__, slave, idx, port); } out: mutex_unlock(&priv->counters_table.mutex); } void mlx4_counter_free(struct mlx4_dev *dev, u8 port, u32 idx) { u64 in_param = 0; struct mlx4_priv *priv = mlx4_priv(dev); struct counter_index *counter, *tmp_counter; int first = 0; if (mlx4_is_mfunc(dev)) { set_param_l(&in_param, idx); mlx4_cmd(dev, in_param, ((u32) port) << 8 | (u32) RES_COUNTER, RES_OP_RESERVE, MLX4_CMD_FREE_RES, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED); if (mlx4_is_slave(dev) && idx != MLX4_SINK_COUNTER_INDEX) { mutex_lock(&priv->counters_table.mutex); list_for_each_entry_safe(counter, tmp_counter, &priv->counters_table.global_port_list[port - 1], list) { if (counter->index == idx && first++) { list_del(&counter->list); kfree(counter); mlx4_dbg(dev, "%s: delete counter index %d for port %d\n" , __func__, idx, port); mutex_unlock(&priv->counters_table.mutex); return; } } mutex_unlock(&priv->counters_table.mutex); } return; } __mlx4_counter_free(dev, 0, port, idx); } EXPORT_SYMBOL_GPL(mlx4_counter_free); int __mlx4_clear_if_stat(struct mlx4_dev *dev, u8 counter_index) { struct mlx4_cmd_mailbox *if_stat_mailbox = NULL; int err = 0; u32 if_stat_in_mod = (counter_index & 0xff) | (1 << 31); if (counter_index == MLX4_SINK_COUNTER_INDEX) return -EINVAL; if (mlx4_is_slave(dev)) return 0; if_stat_mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(if_stat_mailbox)) { err = PTR_ERR(if_stat_mailbox); return err; } err = mlx4_cmd_box(dev, 0, if_stat_mailbox->dma, if_stat_in_mod, 0, MLX4_CMD_QUERY_IF_STAT, MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE); mlx4_free_cmd_mailbox(dev, if_stat_mailbox); return err; } u8 mlx4_get_default_counter_index(struct mlx4_dev *dev, int slave, int port) { struct mlx4_priv *priv = mlx4_priv(dev); struct counter_index *new_counter_index; if (dev->caps.port_type[port] == MLX4_PORT_TYPE_IB) { mlx4_dbg(dev, "%s: return counter index %d for slave %d port (MLX4_PORT_TYPE_IB) %d\n", __func__, MLX4_SINK_COUNTER_INDEX, slave, port); return (u8)MLX4_SINK_COUNTER_INDEX; } mutex_lock(&priv->counters_table.mutex); if (slave == 0) { new_counter_index = list_entry(priv->counters_table.global_port_list[port - 1].next, struct counter_index, list); } else { new_counter_index = list_entry(priv->counters_table.vf_list[slave - 1][port - 1].next, struct counter_index, list); } mutex_unlock(&priv->counters_table.mutex); mlx4_dbg(dev, "%s: return counter index %d for slave %d port %d\n", __func__, new_counter_index->index, slave, port); return (u8)new_counter_index->index; } int mlx4_get_vport_ethtool_stats(struct mlx4_dev *dev, int port, struct mlx4_en_vport_stats *vport_stats, int reset) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_cmd_mailbox *if_stat_mailbox = NULL; union mlx4_counter *counter; int err = 0; u32 if_stat_in_mod; struct counter_index *vport, *tmp_vport; if (!vport_stats) return -EINVAL; if_stat_mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(if_stat_mailbox)) { err = PTR_ERR(if_stat_mailbox); return err; } mutex_lock(&priv->counters_table.mutex); list_for_each_entry_safe(vport, tmp_vport, &priv->counters_table.global_port_list[port - 1], list) { if (vport->index == MLX4_SINK_COUNTER_INDEX) continue; memset(if_stat_mailbox->buf, 0, sizeof(union mlx4_counter)); if_stat_in_mod = (vport->index & 0xff) | ((reset & 1) << 31); err = mlx4_cmd_box(dev, 0, if_stat_mailbox->dma, if_stat_in_mod, 0, MLX4_CMD_QUERY_IF_STAT, MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE); if (err) { mlx4_dbg(dev, "%s: failed to read statistics for counter index %d\n", __func__, vport->index); goto if_stat_out; } counter = (union mlx4_counter *)if_stat_mailbox->buf; if ((counter->control.cnt_mode & 0xf) == 1) { vport_stats->rx_broadcast_packets += be64_to_cpu(counter->ext.counters[0].IfRxBroadcastFrames); vport_stats->rx_unicast_packets += be64_to_cpu(counter->ext.counters[0].IfRxUnicastFrames); vport_stats->rx_multicast_packets += be64_to_cpu(counter->ext.counters[0].IfRxMulticastFrames); vport_stats->tx_broadcast_packets += be64_to_cpu(counter->ext.counters[0].IfTxBroadcastFrames); vport_stats->tx_unicast_packets += be64_to_cpu(counter->ext.counters[0].IfTxUnicastFrames); vport_stats->tx_multicast_packets += be64_to_cpu(counter->ext.counters[0].IfTxMulticastFrames); vport_stats->rx_broadcast_bytes += be64_to_cpu(counter->ext.counters[0].IfRxBroadcastOctets); vport_stats->rx_unicast_bytes += be64_to_cpu(counter->ext.counters[0].IfRxUnicastOctets); vport_stats->rx_multicast_bytes += be64_to_cpu(counter->ext.counters[0].IfRxMulticastOctets); vport_stats->tx_broadcast_bytes += be64_to_cpu(counter->ext.counters[0].IfTxBroadcastOctets); vport_stats->tx_unicast_bytes += be64_to_cpu(counter->ext.counters[0].IfTxUnicastOctets); vport_stats->tx_multicast_bytes += be64_to_cpu(counter->ext.counters[0].IfTxMulticastOctets); vport_stats->rx_errors += be64_to_cpu(counter->ext.counters[0].IfRxErrorFrames); vport_stats->rx_dropped += be64_to_cpu(counter->ext.counters[0].IfRxNoBufferFrames); vport_stats->tx_errors += be64_to_cpu(counter->ext.counters[0].IfTxDroppedFrames); } } if_stat_out: mutex_unlock(&priv->counters_table.mutex); mlx4_free_cmd_mailbox(dev, if_stat_mailbox); return err; } EXPORT_SYMBOL_GPL(mlx4_get_vport_ethtool_stats); static int mlx4_setup_hca(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); int err; int port; __be32 ib_port_default_caps; err = mlx4_init_uar_table(dev); if (err) { mlx4_err(dev, "Failed to initialize " "user access region table (err=%d), aborting.\n", err); return err; } err = mlx4_uar_alloc(dev, &priv->driver_uar); if (err) { mlx4_err(dev, "Failed to allocate driver access region " "(err=%d), aborting.\n", err); goto err_uar_table_free; } priv->kar = ioremap((phys_addr_t) priv->driver_uar.pfn << PAGE_SHIFT, PAGE_SIZE); if (!priv->kar) { mlx4_err(dev, "Couldn't map kernel access region, " "aborting.\n"); err = -ENOMEM; goto err_uar_free; } err = mlx4_init_pd_table(dev); if (err) { mlx4_err(dev, "Failed to initialize " "protection domain table (err=%d), aborting.\n", err); goto err_kar_unmap; } err = mlx4_init_xrcd_table(dev); if (err) { mlx4_err(dev, "Failed to initialize " "reliable connection domain table (err=%d), " "aborting.\n", err); goto err_pd_table_free; } err = mlx4_init_mr_table(dev); if (err) { mlx4_err(dev, "Failed to initialize " "memory region table (err=%d), aborting.\n", err); goto err_xrcd_table_free; } if (!mlx4_is_slave(dev)) { err = mlx4_init_mcg_table(dev); if (err) { mlx4_err(dev, "Failed to initialize " "multicast group table (err=%d), aborting.\n", err); goto err_mr_table_free; } } err = mlx4_init_eq_table(dev); if (err) { mlx4_err(dev, "Failed to initialize " "event queue table (err=%d), aborting.\n", err); goto err_mcg_table_free; } err = mlx4_cmd_use_events(dev); if (err) { mlx4_err(dev, "Failed to switch to event-driven " "firmware commands (err=%d), aborting.\n", err); goto err_eq_table_free; } err = mlx4_NOP(dev); if (err) { if (dev->flags & MLX4_FLAG_MSI_X) { mlx4_warn(dev, "NOP command failed to generate MSI-X " "interrupt IRQ %d).\n", priv->eq_table.eq[dev->caps.num_comp_vectors].irq); mlx4_warn(dev, "Trying again without MSI-X.\n"); } else { mlx4_err(dev, "NOP command failed to generate interrupt " "(IRQ %d), aborting.\n", priv->eq_table.eq[dev->caps.num_comp_vectors].irq); mlx4_err(dev, "BIOS or ACPI interrupt routing problem?\n"); } goto err_cmd_poll; } mlx4_dbg(dev, "NOP command IRQ test passed\n"); err = mlx4_init_cq_table(dev); if (err) { mlx4_err(dev, "Failed to initialize " "completion queue table (err=%d), aborting.\n", err); goto err_cmd_poll; } err = mlx4_init_srq_table(dev); if (err) { mlx4_err(dev, "Failed to initialize " "shared receive queue table (err=%d), aborting.\n", err); goto err_cq_table_free; } err = mlx4_init_qp_table(dev); if (err) { mlx4_err(dev, "Failed to initialize " "queue pair table (err=%d), aborting.\n", err); goto err_srq_table_free; } err = mlx4_init_counters_table(dev); if (err && err != -ENOENT) { mlx4_err(dev, "Failed to initialize counters table (err=%d), " "aborting.\n", err); goto err_qp_table_free; } if (!mlx4_is_slave(dev)) { for (port = 1; port <= dev->caps.num_ports; port++) { ib_port_default_caps = 0; err = mlx4_get_port_ib_caps(dev, port, &ib_port_default_caps); if (err) mlx4_warn(dev, "failed to get port %d default " "ib capabilities (%d). Continuing " "with caps = 0\n", port, err); dev->caps.ib_port_def_cap[port] = ib_port_default_caps; /* initialize per-slave default ib port capabilities */ if (mlx4_is_master(dev)) { int i; for (i = 0; i < dev->num_slaves; i++) { if (i == mlx4_master_func_num(dev)) continue; priv->mfunc.master.slave_state[i].ib_cap_mask[port] = ib_port_default_caps; } } dev->caps.port_ib_mtu[port] = IB_MTU_4096; err = mlx4_SET_PORT(dev, port, mlx4_is_master(dev) ? dev->caps.pkey_table_len[port] : -1); if (err) { mlx4_err(dev, "Failed to set port %d (err=%d), " "aborting\n", port, err); goto err_counters_table_free; } } } return 0; err_counters_table_free: mlx4_cleanup_counters_table(dev); err_qp_table_free: mlx4_cleanup_qp_table(dev); err_srq_table_free: mlx4_cleanup_srq_table(dev); err_cq_table_free: mlx4_cleanup_cq_table(dev); err_cmd_poll: mlx4_cmd_use_polling(dev); err_eq_table_free: mlx4_cleanup_eq_table(dev); err_mcg_table_free: if (!mlx4_is_slave(dev)) mlx4_cleanup_mcg_table(dev); err_mr_table_free: mlx4_cleanup_mr_table(dev); err_xrcd_table_free: mlx4_cleanup_xrcd_table(dev); err_pd_table_free: mlx4_cleanup_pd_table(dev); err_kar_unmap: iounmap(priv->kar); err_uar_free: mlx4_uar_free(dev, &priv->driver_uar); err_uar_table_free: mlx4_cleanup_uar_table(dev); return err; } static void mlx4_enable_msi_x(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); struct msix_entry *entries; int nreq = min_t(int, dev->caps.num_ports * min_t(int, num_possible_cpus() + 1, MAX_MSIX_P_PORT) + MSIX_LEGACY_SZ, MAX_MSIX); int err; int i; if (msi_x) { nreq = min_t(int, dev->caps.num_eqs - dev->caps.reserved_eqs, nreq); if (msi_x > 1 && !mlx4_is_mfunc(dev)) nreq = min_t(int, nreq, msi_x); entries = kcalloc(nreq, sizeof *entries, GFP_KERNEL); if (!entries) goto no_msi; for (i = 0; i < nreq; ++i) entries[i].entry = i; retry: err = pci_enable_msix(dev->pdev, entries, nreq); if (err) { /* Try again if at least 2 vectors are available */ if (err > 1) { mlx4_info(dev, "Requested %d vectors, " "but only %d MSI-X vectors available, " "trying again\n", nreq, err); nreq = err; goto retry; } kfree(entries); + /* if error, or can't alloc even 1 IRQ */ + if (err < 0) { + mlx4_err(dev, "No IRQs left, device can't " + "be started.\n"); + goto no_irq; + } goto no_msi; } if (nreq < MSIX_LEGACY_SZ + dev->caps.num_ports * MIN_MSIX_P_PORT) { /*Working in legacy mode , all EQ's shared*/ dev->caps.comp_pool = 0; dev->caps.num_comp_vectors = nreq - 1; } else { dev->caps.comp_pool = nreq - MSIX_LEGACY_SZ; dev->caps.num_comp_vectors = MSIX_LEGACY_SZ - 1; } for (i = 0; i < nreq; ++i) priv->eq_table.eq[i].irq = entries[i].vector; dev->flags |= MLX4_FLAG_MSI_X; kfree(entries); return; } no_msi: dev->caps.num_comp_vectors = 1; dev->caps.comp_pool = 0; for (i = 0; i < 2; ++i) priv->eq_table.eq[i].irq = dev->pdev->irq; + return; +no_irq: + dev->caps.num_comp_vectors = 0; + dev->caps.comp_pool = 0; } static int mlx4_init_port_info(struct mlx4_dev *dev, int port) { struct mlx4_port_info *info = &mlx4_priv(dev)->port[port]; int err = 0; info->dev = dev; info->port = port; if (!mlx4_is_slave(dev)) { mlx4_init_mac_table(dev, &info->mac_table); mlx4_init_vlan_table(dev, &info->vlan_table); info->base_qpn = mlx4_get_base_qpn(dev, port); } sprintf(info->dev_name, "mlx4_port%d", port); info->port_attr.attr.name = info->dev_name; if (mlx4_is_mfunc(dev)) info->port_attr.attr.mode = S_IRUGO; else { info->port_attr.attr.mode = S_IRUGO | S_IWUSR; info->port_attr.store = set_port_type; } info->port_attr.show = show_port_type; sysfs_attr_init(&info->port_attr.attr); err = device_create_file(&dev->pdev->dev, &info->port_attr); if (err) { mlx4_err(dev, "Failed to create file for port %d\n", port); info->port = -1; } sprintf(info->dev_mtu_name, "mlx4_port%d_mtu", port); info->port_mtu_attr.attr.name = info->dev_mtu_name; if (mlx4_is_mfunc(dev)) info->port_mtu_attr.attr.mode = S_IRUGO; else { info->port_mtu_attr.attr.mode = S_IRUGO | S_IWUSR; info->port_mtu_attr.store = set_port_ib_mtu; } info->port_mtu_attr.show = show_port_ib_mtu; sysfs_attr_init(&info->port_mtu_attr.attr); err = device_create_file(&dev->pdev->dev, &info->port_mtu_attr); if (err) { mlx4_err(dev, "Failed to create mtu file for port %d\n", port); device_remove_file(&info->dev->pdev->dev, &info->port_attr); info->port = -1; } return err; } static void mlx4_cleanup_port_info(struct mlx4_port_info *info) { if (info->port < 0) return; device_remove_file(&info->dev->pdev->dev, &info->port_attr); device_remove_file(&info->dev->pdev->dev, &info->port_mtu_attr); } static int mlx4_init_steering(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); int num_entries = dev->caps.num_ports; int i, j; priv->steer = kzalloc(sizeof(struct mlx4_steer) * num_entries, GFP_KERNEL); if (!priv->steer) return -ENOMEM; for (i = 0; i < num_entries; i++) for (j = 0; j < MLX4_NUM_STEERS; j++) { INIT_LIST_HEAD(&priv->steer[i].promisc_qps[j]); INIT_LIST_HEAD(&priv->steer[i].steer_entries[j]); } return 0; } static void mlx4_clear_steering(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_steer_index *entry, *tmp_entry; struct mlx4_promisc_qp *pqp, *tmp_pqp; int num_entries = dev->caps.num_ports; int i, j; for (i = 0; i < num_entries; i++) { for (j = 0; j < MLX4_NUM_STEERS; j++) { list_for_each_entry_safe(pqp, tmp_pqp, &priv->steer[i].promisc_qps[j], list) { list_del(&pqp->list); kfree(pqp); } list_for_each_entry_safe(entry, tmp_entry, &priv->steer[i].steer_entries[j], list) { list_del(&entry->list); list_for_each_entry_safe(pqp, tmp_pqp, &entry->duplicates, list) { list_del(&pqp->list); kfree(pqp); } kfree(entry); } } } kfree(priv->steer); } static int extended_func_num(struct pci_dev *pdev) { return PCI_SLOT(pdev->devfn) * 8 + PCI_FUNC(pdev->devfn); } #define MLX4_OWNER_BASE 0x8069c #define MLX4_OWNER_SIZE 4 static int mlx4_get_ownership(struct mlx4_dev *dev) { void __iomem *owner; u32 ret; if (pci_channel_offline(dev->pdev)) return -EIO; owner = ioremap(pci_resource_start(dev->pdev, 0) + MLX4_OWNER_BASE, MLX4_OWNER_SIZE); if (!owner) { mlx4_err(dev, "Failed to obtain ownership bit\n"); return -ENOMEM; } ret = readl(owner); iounmap(owner); return (int) !!ret; } static void mlx4_free_ownership(struct mlx4_dev *dev) { void __iomem *owner; if (pci_channel_offline(dev->pdev)) return; owner = ioremap(pci_resource_start(dev->pdev, 0) + MLX4_OWNER_BASE, MLX4_OWNER_SIZE); if (!owner) { mlx4_err(dev, "Failed to obtain ownership bit\n"); return; } writel(0, owner); msleep(1000); iounmap(owner); } static int __mlx4_init_one(struct pci_dev *pdev, int pci_dev_data) { struct mlx4_priv *priv; struct mlx4_dev *dev; int err; int port; int nvfs, prb_vf; pr_info(DRV_NAME ": Initializing %s\n", pci_name(pdev)); err = pci_enable_device(pdev); if (err) { dev_err(&pdev->dev, "Cannot enable PCI device, " "aborting.\n"); return err; } mlx4_get_val(num_vfs.dbdf2val.tbl, pci_physfn(pdev), 0, &nvfs); mlx4_get_val(probe_vf.dbdf2val.tbl, pci_physfn(pdev), 0, &prb_vf); if (nvfs > MLX4_MAX_NUM_VF) { dev_err(&pdev->dev, "There are more VF's (%d) than allowed(%d)\n", nvfs, MLX4_MAX_NUM_VF); return -EINVAL; } if (nvfs < 0) { dev_err(&pdev->dev, "num_vfs module parameter cannot be negative\n"); return -EINVAL; } /* * Check for BARs. */ if (!(pci_dev_data & MLX4_PCI_DEV_IS_VF) && !(pci_resource_flags(pdev, 0) & IORESOURCE_MEM)) { dev_err(&pdev->dev, "Missing DCS, aborting." "(driver_data: 0x%x, pci_resource_flags(pdev, 0):0x%x)\n", pci_dev_data, pci_resource_flags(pdev, 0)); err = -ENODEV; goto err_disable_pdev; } if (!(pci_resource_flags(pdev, 2) & IORESOURCE_MEM)) { dev_err(&pdev->dev, "Missing UAR, aborting.\n"); err = -ENODEV; goto err_disable_pdev; } err = pci_request_regions(pdev, DRV_NAME); if (err) { dev_err(&pdev->dev, "Couldn't get PCI resources, aborting\n"); goto err_disable_pdev; } pci_set_master(pdev); err = pci_set_dma_mask(pdev, DMA_BIT_MASK(64)); if (err) { dev_warn(&pdev->dev, "Warning: couldn't set 64-bit PCI DMA mask.\n"); err = pci_set_dma_mask(pdev, DMA_BIT_MASK(32)); if (err) { dev_err(&pdev->dev, "Can't set PCI DMA mask, aborting.\n"); goto err_release_regions; } } err = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64)); if (err) { dev_warn(&pdev->dev, "Warning: couldn't set 64-bit " "consistent PCI DMA mask.\n"); err = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(32)); if (err) { dev_err(&pdev->dev, "Can't set consistent PCI DMA mask, " "aborting.\n"); goto err_release_regions; } } /* Allow large DMA segments, up to the firmware limit of 1 GB */ dma_set_max_seg_size(&pdev->dev, 1024 * 1024 * 1024); priv = kzalloc(sizeof *priv, GFP_KERNEL); if (!priv) { dev_err(&pdev->dev, "Device struct alloc failed, " "aborting.\n"); err = -ENOMEM; goto err_release_regions; } dev = &priv->dev; dev->pdev = pdev; INIT_LIST_HEAD(&priv->dev_list); INIT_LIST_HEAD(&priv->ctx_list); spin_lock_init(&priv->ctx_lock); mutex_init(&priv->port_mutex); INIT_LIST_HEAD(&priv->pgdir_list); mutex_init(&priv->pgdir_mutex); INIT_LIST_HEAD(&priv->bf_list); mutex_init(&priv->bf_mutex); dev->rev_id = pdev->revision; dev->numa_node = dev_to_node(&pdev->dev); /* Detect if this device is a virtual function */ if (pci_dev_data & MLX4_PCI_DEV_IS_VF) { /* When acting as pf, we normally skip vfs unless explicitly * requested to probe them. */ if (nvfs && extended_func_num(pdev) > prb_vf) { mlx4_warn(dev, "Skipping virtual function:%d\n", extended_func_num(pdev)); err = -ENODEV; goto err_free_dev; } mlx4_warn(dev, "Detected virtual function - running in slave mode\n"); dev->flags |= MLX4_FLAG_SLAVE; } else { /* We reset the device and enable SRIOV only for physical * devices. Try to claim ownership on the device; * if already taken, skip -- do not allow multiple PFs */ err = mlx4_get_ownership(dev); if (err) { if (err < 0) goto err_free_dev; else { mlx4_warn(dev, "Multiple PFs not yet supported." " Skipping PF.\n"); err = -EINVAL; goto err_free_dev; } } if (nvfs) { mlx4_warn(dev, "Enabling SR-IOV with %d VFs\n", nvfs); err = pci_enable_sriov(pdev, nvfs); if (err) { mlx4_err(dev, "Failed to enable SR-IOV, continuing without SR-IOV (err = %d).\n", err); err = 0; } else { mlx4_warn(dev, "Running in master mode\n"); dev->flags |= MLX4_FLAG_SRIOV | MLX4_FLAG_MASTER; dev->num_vfs = nvfs; } } atomic_set(&priv->opreq_count, 0); INIT_WORK(&priv->opreq_task, mlx4_opreq_action); /* * Now reset the HCA before we touch the PCI capabilities or * attempt a firmware command, since a boot ROM may have left * the HCA in an undefined state. */ err = mlx4_reset(dev); if (err) { mlx4_err(dev, "Failed to reset HCA, aborting.\n"); goto err_sriov; } } slave_start: err = mlx4_cmd_init(dev); if (err) { mlx4_err(dev, "Failed to init command interface, aborting.\n"); goto err_sriov; } /* In slave functions, the communication channel must be initialized * before posting commands. Also, init num_slaves before calling * mlx4_init_hca */ if (mlx4_is_mfunc(dev)) { if (mlx4_is_master(dev)) dev->num_slaves = MLX4_MAX_NUM_SLAVES; else { dev->num_slaves = 0; err = mlx4_multi_func_init(dev); if (err) { mlx4_err(dev, "Failed to init slave mfunc" " interface, aborting.\n"); goto err_cmd; } } } err = mlx4_init_hca(dev); if (err) { if (err == -EACCES) { /* Not primary Physical function * Running in slave mode */ mlx4_cmd_cleanup(dev); dev->flags |= MLX4_FLAG_SLAVE; dev->flags &= ~MLX4_FLAG_MASTER; goto slave_start; } else goto err_mfunc; } /* In master functions, the communication channel must be initialized * after obtaining its address from fw */ if (mlx4_is_master(dev)) { err = mlx4_multi_func_init(dev); if (err) { mlx4_err(dev, "Failed to init master mfunc" "interface, aborting.\n"); goto err_close; } } err = mlx4_alloc_eq_table(dev); if (err) goto err_master_mfunc; priv->msix_ctl.pool_bm = 0; mutex_init(&priv->msix_ctl.pool_lock); mlx4_enable_msi_x(dev); + + /* no MSIX and no shared IRQ */ + if (!dev->caps.num_comp_vectors && !dev->caps.comp_pool) { + err = -ENOSPC; + goto err_free_eq; + } + if ((mlx4_is_mfunc(dev)) && !(dev->flags & MLX4_FLAG_MSI_X)) { err = -ENOSYS; mlx4_err(dev, "INTx is not supported in multi-function mode." " aborting.\n"); goto err_free_eq; } if (!mlx4_is_slave(dev)) { err = mlx4_init_steering(dev); if (err) goto err_free_eq; } err = mlx4_setup_hca(dev); if (err == -EBUSY && (dev->flags & MLX4_FLAG_MSI_X) && !mlx4_is_mfunc(dev)) { dev->flags &= ~MLX4_FLAG_MSI_X; dev->caps.num_comp_vectors = 1; dev->caps.comp_pool = 0; pci_disable_msix(pdev); err = mlx4_setup_hca(dev); } if (err) goto err_steer; mlx4_init_quotas(dev); for (port = 1; port <= dev->caps.num_ports; port++) { err = mlx4_init_port_info(dev, port); if (err) goto err_port; } err = mlx4_register_device(dev); if (err) goto err_port; mlx4_request_modules(dev); mlx4_sense_init(dev); mlx4_start_sense(dev); priv->pci_dev_data = pci_dev_data; pci_set_drvdata(pdev, dev); return 0; err_port: for (--port; port >= 1; --port) mlx4_cleanup_port_info(&priv->port[port]); mlx4_cleanup_counters_table(dev); mlx4_cleanup_qp_table(dev); mlx4_cleanup_srq_table(dev); mlx4_cleanup_cq_table(dev); mlx4_cmd_use_polling(dev); mlx4_cleanup_eq_table(dev); mlx4_cleanup_mcg_table(dev); mlx4_cleanup_mr_table(dev); mlx4_cleanup_xrcd_table(dev); mlx4_cleanup_pd_table(dev); mlx4_cleanup_uar_table(dev); err_steer: if (!mlx4_is_slave(dev)) mlx4_clear_steering(dev); err_free_eq: mlx4_free_eq_table(dev); err_master_mfunc: if (mlx4_is_master(dev)) { mlx4_free_resource_tracker(dev, RES_TR_FREE_STRUCTS_ONLY); mlx4_multi_func_cleanup(dev); } if (mlx4_is_slave(dev)) { kfree(dev->caps.qp0_tunnel); kfree(dev->caps.qp0_proxy); kfree(dev->caps.qp1_tunnel); kfree(dev->caps.qp1_proxy); } err_close: if (dev->flags & MLX4_FLAG_MSI_X) pci_disable_msix(pdev); mlx4_close_hca(dev); err_mfunc: if (mlx4_is_slave(dev)) mlx4_multi_func_cleanup(dev); err_cmd: mlx4_cmd_cleanup(dev); err_sriov: if (dev->flags & MLX4_FLAG_SRIOV) pci_disable_sriov(pdev); if (!mlx4_is_slave(dev)) mlx4_free_ownership(dev); err_free_dev: kfree(priv); err_release_regions: pci_release_regions(pdev); err_disable_pdev: pci_disable_device(pdev); pci_set_drvdata(pdev, NULL); return err; } static int __devinit mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id) { printk_once(KERN_INFO "%s", mlx4_version); return __mlx4_init_one(pdev, id->driver_data); } static void mlx4_remove_one(struct pci_dev *pdev) { struct mlx4_dev *dev = pci_get_drvdata(pdev); struct mlx4_priv *priv = mlx4_priv(dev); int p; if (dev) { /* in SRIOV it is not allowed to unload the pf's * driver while there are alive vf's */ if (mlx4_is_master(dev)) { if (mlx4_how_many_lives_vf(dev)) mlx4_err(dev, "Removing PF when there are assigned VF's !!!\n"); } mlx4_stop_sense(dev); mlx4_unregister_device(dev); for (p = 1; p <= dev->caps.num_ports; p++) { mlx4_cleanup_port_info(&priv->port[p]); mlx4_CLOSE_PORT(dev, p); } if (mlx4_is_master(dev)) mlx4_free_resource_tracker(dev, RES_TR_FREE_SLAVES_ONLY); mlx4_cleanup_counters_table(dev); mlx4_cleanup_qp_table(dev); mlx4_cleanup_srq_table(dev); mlx4_cleanup_cq_table(dev); mlx4_cmd_use_polling(dev); mlx4_cleanup_eq_table(dev); mlx4_cleanup_mcg_table(dev); mlx4_cleanup_mr_table(dev); mlx4_cleanup_xrcd_table(dev); mlx4_cleanup_pd_table(dev); if (mlx4_is_master(dev)) mlx4_free_resource_tracker(dev, RES_TR_FREE_STRUCTS_ONLY); iounmap(priv->kar); mlx4_uar_free(dev, &priv->driver_uar); mlx4_cleanup_uar_table(dev); if (!mlx4_is_slave(dev)) mlx4_clear_steering(dev); mlx4_free_eq_table(dev); if (mlx4_is_master(dev)) mlx4_multi_func_cleanup(dev); mlx4_close_hca(dev); if (mlx4_is_slave(dev)) mlx4_multi_func_cleanup(dev); mlx4_cmd_cleanup(dev); if (dev->flags & MLX4_FLAG_MSI_X) pci_disable_msix(pdev); if (dev->flags & MLX4_FLAG_SRIOV) { mlx4_warn(dev, "Disabling SR-IOV\n"); pci_disable_sriov(pdev); } if (!mlx4_is_slave(dev)) mlx4_free_ownership(dev); kfree(dev->caps.qp0_tunnel); kfree(dev->caps.qp0_proxy); kfree(dev->caps.qp1_tunnel); kfree(dev->caps.qp1_proxy); kfree(priv); pci_release_regions(pdev); pci_disable_device(pdev); pci_set_drvdata(pdev, NULL); } } static int restore_current_port_types(struct mlx4_dev *dev, enum mlx4_port_type *types, enum mlx4_port_type *poss_types) { struct mlx4_priv *priv = mlx4_priv(dev); int err, i; mlx4_stop_sense(dev); mutex_lock(&priv->port_mutex); for (i = 0; i < dev->caps.num_ports; i++) dev->caps.possible_type[i + 1] = poss_types[i]; err = mlx4_change_port_types(dev, types); mlx4_start_sense(dev); mutex_unlock(&priv->port_mutex); return err; } int mlx4_restart_one(struct pci_dev *pdev) { struct mlx4_dev *dev = pci_get_drvdata(pdev); struct mlx4_priv *priv = mlx4_priv(dev); enum mlx4_port_type curr_type[MLX4_MAX_PORTS]; enum mlx4_port_type poss_type[MLX4_MAX_PORTS]; int pci_dev_data, err, i; pci_dev_data = priv->pci_dev_data; for (i = 0; i < dev->caps.num_ports; i++) { curr_type[i] = dev->caps.port_type[i + 1]; poss_type[i] = dev->caps.possible_type[i + 1]; } mlx4_remove_one(pdev); err = __mlx4_init_one(pdev, pci_dev_data); if (err) return err; dev = pci_get_drvdata(pdev); err = restore_current_port_types(dev, curr_type, poss_type); if (err) mlx4_err(dev, "mlx4_restart_one: could not restore original port types (%d)\n", err); return 0; } static DEFINE_PCI_DEVICE_TABLE(mlx4_pci_table) = { /* MT25408 "Hermon" SDR */ { PCI_VDEVICE(MELLANOX, 0x6340), MLX4_PCI_DEV_FORCE_SENSE_PORT }, /* MT25408 "Hermon" DDR */ { PCI_VDEVICE(MELLANOX, 0x634a), MLX4_PCI_DEV_FORCE_SENSE_PORT }, /* MT25408 "Hermon" QDR */ { PCI_VDEVICE(MELLANOX, 0x6354), MLX4_PCI_DEV_FORCE_SENSE_PORT }, /* MT25408 "Hermon" DDR PCIe gen2 */ { PCI_VDEVICE(MELLANOX, 0x6732), MLX4_PCI_DEV_FORCE_SENSE_PORT }, /* MT25408 "Hermon" QDR PCIe gen2 */ { PCI_VDEVICE(MELLANOX, 0x673c), MLX4_PCI_DEV_FORCE_SENSE_PORT }, /* MT25408 "Hermon" EN 10GigE */ { PCI_VDEVICE(MELLANOX, 0x6368), MLX4_PCI_DEV_FORCE_SENSE_PORT }, /* MT25408 "Hermon" EN 10GigE PCIe gen2 */ { PCI_VDEVICE(MELLANOX, 0x6750), MLX4_PCI_DEV_FORCE_SENSE_PORT }, /* MT25458 ConnectX EN 10GBASE-T 10GigE */ { PCI_VDEVICE(MELLANOX, 0x6372), MLX4_PCI_DEV_FORCE_SENSE_PORT }, /* MT25458 ConnectX EN 10GBASE-T+Gen2 10GigE */ { PCI_VDEVICE(MELLANOX, 0x675a), MLX4_PCI_DEV_FORCE_SENSE_PORT }, /* MT26468 ConnectX EN 10GigE PCIe gen2*/ { PCI_VDEVICE(MELLANOX, 0x6764), MLX4_PCI_DEV_FORCE_SENSE_PORT }, /* MT26438 ConnectX EN 40GigE PCIe gen2 5GT/s */ { PCI_VDEVICE(MELLANOX, 0x6746), MLX4_PCI_DEV_FORCE_SENSE_PORT }, /* MT26478 ConnectX2 40GigE PCIe gen2 */ { PCI_VDEVICE(MELLANOX, 0x676e), MLX4_PCI_DEV_FORCE_SENSE_PORT }, /* MT25400 Family [ConnectX-2 Virtual Function] */ { PCI_VDEVICE(MELLANOX, 0x1002), MLX4_PCI_DEV_IS_VF }, /* MT27500 Family [ConnectX-3] */ { PCI_VDEVICE(MELLANOX, 0x1003), 0 }, /* MT27500 Family [ConnectX-3 Virtual Function] */ { PCI_VDEVICE(MELLANOX, 0x1004), MLX4_PCI_DEV_IS_VF }, { PCI_VDEVICE(MELLANOX, 0x1005), 0 }, /* MT27510 Family */ { PCI_VDEVICE(MELLANOX, 0x1006), 0 }, /* MT27511 Family */ { PCI_VDEVICE(MELLANOX, 0x1007), 0 }, /* MT27520 Family */ { PCI_VDEVICE(MELLANOX, 0x1008), 0 }, /* MT27521 Family */ { PCI_VDEVICE(MELLANOX, 0x1009), 0 }, /* MT27530 Family */ { PCI_VDEVICE(MELLANOX, 0x100a), 0 }, /* MT27531 Family */ { PCI_VDEVICE(MELLANOX, 0x100b), 0 }, /* MT27540 Family */ { PCI_VDEVICE(MELLANOX, 0x100c), 0 }, /* MT27541 Family */ { PCI_VDEVICE(MELLANOX, 0x100d), 0 }, /* MT27550 Family */ { PCI_VDEVICE(MELLANOX, 0x100e), 0 }, /* MT27551 Family */ { PCI_VDEVICE(MELLANOX, 0x100f), 0 }, /* MT27560 Family */ { PCI_VDEVICE(MELLANOX, 0x1010), 0 }, /* MT27561 Family */ { 0, } }; MODULE_DEVICE_TABLE(pci, mlx4_pci_table); static pci_ers_result_t mlx4_pci_err_detected(struct pci_dev *pdev, pci_channel_state_t state) { mlx4_remove_one(pdev); return state == pci_channel_io_perm_failure ? PCI_ERS_RESULT_DISCONNECT : PCI_ERS_RESULT_NEED_RESET; } static pci_ers_result_t mlx4_pci_slot_reset(struct pci_dev *pdev) { int ret = __mlx4_init_one(pdev, 0); return ret ? PCI_ERS_RESULT_DISCONNECT : PCI_ERS_RESULT_RECOVERED; } static const struct pci_error_handlers mlx4_err_handler = { .error_detected = mlx4_pci_err_detected, .slot_reset = mlx4_pci_slot_reset, }; static int suspend(struct pci_dev *pdev, pm_message_t state) { mlx4_remove_one(pdev); return 0; } static int resume(struct pci_dev *pdev) { return __mlx4_init_one(pdev, 0); } static struct pci_driver mlx4_driver = { .name = DRV_NAME, .id_table = mlx4_pci_table, .probe = mlx4_init_one, .remove = __devexit_p(mlx4_remove_one), .suspend = suspend, .resume = resume, .err_handler = &mlx4_err_handler, }; static int __init mlx4_verify_params(void) { int status; status = update_defaults(&port_type_array); if (status == INVALID_STR) { if (mlx4_fill_dbdf2val_tbl(&port_type_array.dbdf2val)) return -1; } else if (status == INVALID_DATA) { return -1; } status = update_defaults(&num_vfs); if (status == INVALID_STR) { if (mlx4_fill_dbdf2val_tbl(&num_vfs.dbdf2val)) return -1; } else if (status == INVALID_DATA) { return -1; } status = update_defaults(&probe_vf); if (status == INVALID_STR) { if (mlx4_fill_dbdf2val_tbl(&probe_vf.dbdf2val)) return -1; } else if (status == INVALID_DATA) { return -1; } if (msi_x < 0) { pr_warn("mlx4_core: bad msi_x: %d\n", msi_x); return -1; } if ((log_num_mac < 0) || (log_num_mac > 7)) { pr_warning("mlx4_core: bad num_mac: %d\n", log_num_mac); return -1; } if (log_num_vlan != 0) pr_warning("mlx4_core: log_num_vlan - obsolete module param, using %d\n", MLX4_LOG_NUM_VLANS); if (mlx4_set_4k_mtu != -1) pr_warning("mlx4_core: set_4k_mtu - obsolete module param\n"); if ((log_mtts_per_seg < 0) || (log_mtts_per_seg > 7)) { pr_warning("mlx4_core: bad log_mtts_per_seg: %d\n", log_mtts_per_seg); return -1; } if (mlx4_log_num_mgm_entry_size != -1 && (mlx4_log_num_mgm_entry_size < MLX4_MIN_MGM_LOG_ENTRY_SIZE || mlx4_log_num_mgm_entry_size > MLX4_MAX_MGM_LOG_ENTRY_SIZE)) { pr_warning("mlx4_core: mlx4_log_num_mgm_entry_size (%d) not " "in legal range (-1 or %d..%d)\n", mlx4_log_num_mgm_entry_size, MLX4_MIN_MGM_LOG_ENTRY_SIZE, MLX4_MAX_MGM_LOG_ENTRY_SIZE); return -1; } if (mod_param_profile.num_qp < 18 || mod_param_profile.num_qp > 23) { pr_warning("mlx4_core: bad log_num_qp: %d\n", mod_param_profile.num_qp); return -1; } if (mod_param_profile.num_srq < 10) { pr_warning("mlx4_core: too low log_num_srq: %d\n", mod_param_profile.num_srq); return -1; } if (mod_param_profile.num_cq < 10) { pr_warning("mlx4_core: too low log_num_cq: %d\n", mod_param_profile.num_cq); return -1; } if (mod_param_profile.num_mpt < 10) { pr_warning("mlx4_core: too low log_num_mpt: %d\n", mod_param_profile.num_mpt); return -1; } if (mod_param_profile.num_mtt_segs && mod_param_profile.num_mtt_segs < 15) { pr_warning("mlx4_core: too low log_num_mtt: %d\n", mod_param_profile.num_mtt_segs); return -1; } if (mod_param_profile.num_mtt_segs > MLX4_MAX_LOG_NUM_MTT) { pr_warning("mlx4_core: too high log_num_mtt: %d\n", mod_param_profile.num_mtt_segs); return -1; } return 0; } static int __init mlx4_init(void) { int ret; if (mlx4_verify_params()) return -EINVAL; mlx4_catas_init(); mlx4_wq = create_singlethread_workqueue("mlx4"); if (!mlx4_wq) return -ENOMEM; if (enable_sys_tune) sys_tune_init(); ret = pci_register_driver(&mlx4_driver); if (ret < 0) goto err; return 0; err: if (enable_sys_tune) sys_tune_fini(); destroy_workqueue(mlx4_wq); return ret; } static void __exit mlx4_cleanup(void) { if (enable_sys_tune) sys_tune_fini(); pci_unregister_driver(&mlx4_driver); destroy_workqueue(mlx4_wq); } module_init_order(mlx4_init, SI_ORDER_MIDDLE); module_exit(mlx4_cleanup); static int mlx4_evhand(module_t mod, int event, void *arg) { return (0); } static moduledata_t mlx4_mod = { .name = "mlx4", .evhand = mlx4_evhand, }; MODULE_VERSION(mlx4, 1); DECLARE_MODULE(mlx4, mlx4_mod, SI_SUB_OFED_PREINIT, SI_ORDER_ANY); MODULE_DEPEND(mlx4, linuxapi, 1, 1, 1); Index: projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/mcg.c =================================================================== --- projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/mcg.c (revision 279716) +++ projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/mcg.c (revision 279717) @@ -1,1538 +1,1543 @@ /* * Copyright (c) 2006, 2007 Cisco Systems, Inc. All rights reserved. * Copyright (c) 2007, 2008, 2014 Mellanox Technologies. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU * General Public License (GPL) Version 2, available from the file * COPYING in the main directory of this source tree, or the * OpenIB.org BSD license below: * * Redistribution and use in source and binary forms, with or * without modification, are permitted provided that the following * conditions are met: * * - Redistributions of source code must retain the above * copyright notice, this list of conditions and the following * disclaimer. * * - Redistributions in binary form must reproduce the above * copyright notice, this list of conditions and the following * disclaimer in the documentation and/or other materials * provided with the distribution. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ #include #include #include #include +#include #include "mlx4.h" int mlx4_get_mgm_entry_size(struct mlx4_dev *dev) { return 1 << dev->oper_log_mgm_entry_size; } int mlx4_get_qp_per_mgm(struct mlx4_dev *dev) { return 4 * (mlx4_get_mgm_entry_size(dev) / 16 - 2); } static int mlx4_QP_FLOW_STEERING_ATTACH(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox, u32 size, u64 *reg_id) { u64 imm; int err = 0; err = mlx4_cmd_imm(dev, mailbox->dma, &imm, size, 0, MLX4_QP_FLOW_STEERING_ATTACH, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE); if (err) return err; *reg_id = imm; return err; } static int mlx4_QP_FLOW_STEERING_DETACH(struct mlx4_dev *dev, u64 regid) { int err = 0; err = mlx4_cmd(dev, regid, 0, 0, MLX4_QP_FLOW_STEERING_DETACH, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE); return err; } static int mlx4_READ_ENTRY(struct mlx4_dev *dev, int index, struct mlx4_cmd_mailbox *mailbox) { return mlx4_cmd_box(dev, 0, mailbox->dma, index, 0, MLX4_CMD_READ_MCG, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE); } static int mlx4_WRITE_ENTRY(struct mlx4_dev *dev, int index, struct mlx4_cmd_mailbox *mailbox) { return mlx4_cmd(dev, mailbox->dma, index, 0, MLX4_CMD_WRITE_MCG, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE); } static int mlx4_WRITE_PROMISC(struct mlx4_dev *dev, u8 port, u8 steer, struct mlx4_cmd_mailbox *mailbox) { u32 in_mod; in_mod = (u32) port << 16 | steer << 1; return mlx4_cmd(dev, mailbox->dma, in_mod, 0x1, MLX4_CMD_WRITE_MCG, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE); } static int mlx4_GID_HASH(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox, u16 *hash, u8 op_mod) { u64 imm; int err; err = mlx4_cmd_imm(dev, mailbox->dma, &imm, 0, op_mod, MLX4_CMD_MGID_HASH, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE); if (!err) *hash = imm; return err; } static struct mlx4_promisc_qp *get_promisc_qp(struct mlx4_dev *dev, u8 port, enum mlx4_steer_type steer, u32 qpn) { struct mlx4_steer *s_steer; struct mlx4_promisc_qp *pqp; if (port < 1 || port > dev->caps.num_ports) return NULL; s_steer = &mlx4_priv(dev)->steer[port - 1]; list_for_each_entry(pqp, &s_steer->promisc_qps[steer], list) { if (pqp->qpn == qpn) return pqp; } /* not found */ return NULL; } /* * Add new entry to steering data structure. * All promisc QPs should be added as well */ static int new_steering_entry(struct mlx4_dev *dev, u8 port, enum mlx4_steer_type steer, unsigned int index, u32 qpn) { struct mlx4_steer *s_steer; struct mlx4_cmd_mailbox *mailbox; struct mlx4_mgm *mgm; u32 members_count; struct mlx4_steer_index *new_entry; struct mlx4_promisc_qp *pqp; struct mlx4_promisc_qp *dqp = NULL; u32 prot; int err; if (port < 1 || port > dev->caps.num_ports) return -EINVAL; s_steer = &mlx4_priv(dev)->steer[port - 1]; new_entry = kzalloc(sizeof *new_entry, GFP_KERNEL); if (!new_entry) return -ENOMEM; INIT_LIST_HEAD(&new_entry->duplicates); new_entry->index = index; list_add_tail(&new_entry->list, &s_steer->steer_entries[steer]); /* If the given qpn is also a promisc qp, * it should be inserted to duplicates list */ pqp = get_promisc_qp(dev, port, steer, qpn); if (pqp) { dqp = kmalloc(sizeof *dqp, GFP_KERNEL); if (!dqp) { err = -ENOMEM; goto out_alloc; } dqp->qpn = qpn; list_add_tail(&dqp->list, &new_entry->duplicates); } /* if no promisc qps for this vep, we are done */ if (list_empty(&s_steer->promisc_qps[steer])) return 0; /* now need to add all the promisc qps to the new * steering entry, as they should also receive the packets * destined to this address */ mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(mailbox)) { err = -ENOMEM; goto out_alloc; } mgm = mailbox->buf; err = mlx4_READ_ENTRY(dev, index, mailbox); if (err) goto out_mailbox; members_count = be32_to_cpu(mgm->members_count) & 0xffffff; prot = be32_to_cpu(mgm->members_count) >> 30; list_for_each_entry(pqp, &s_steer->promisc_qps[steer], list) { /* don't add already existing qpn */ if (pqp->qpn == qpn) continue; if (members_count == dev->caps.num_qp_per_mgm) { /* out of space */ err = -ENOMEM; goto out_mailbox; } /* add the qpn */ mgm->qp[members_count++] = cpu_to_be32(pqp->qpn & MGM_QPN_MASK); } /* update the qps count and update the entry with all the promisc qps*/ mgm->members_count = cpu_to_be32(members_count | (prot << 30)); err = mlx4_WRITE_ENTRY(dev, index, mailbox); out_mailbox: mlx4_free_cmd_mailbox(dev, mailbox); if (!err) return 0; out_alloc: if (dqp) { list_del(&dqp->list); kfree(dqp); } list_del(&new_entry->list); kfree(new_entry); return err; } /* update the data structures with existing steering entry */ static int existing_steering_entry(struct mlx4_dev *dev, u8 port, enum mlx4_steer_type steer, unsigned int index, u32 qpn) { struct mlx4_steer *s_steer; struct mlx4_steer_index *tmp_entry, *entry = NULL; struct mlx4_promisc_qp *pqp; struct mlx4_promisc_qp *dqp; if (port < 1 || port > dev->caps.num_ports) return -EINVAL; s_steer = &mlx4_priv(dev)->steer[port - 1]; pqp = get_promisc_qp(dev, port, steer, qpn); if (!pqp) return 0; /* nothing to do */ list_for_each_entry(tmp_entry, &s_steer->steer_entries[steer], list) { if (tmp_entry->index == index) { entry = tmp_entry; break; } } if (unlikely(!entry)) { mlx4_warn(dev, "Steering entry at index %x is not registered\n", index); return -EINVAL; } /* the given qpn is listed as a promisc qpn * we need to add it as a duplicate to this entry * for future references */ list_for_each_entry(dqp, &entry->duplicates, list) { if (qpn == dqp->qpn) return 0; /* qp is already duplicated */ } /* add the qp as a duplicate on this index */ dqp = kmalloc(sizeof *dqp, GFP_KERNEL); if (!dqp) return -ENOMEM; dqp->qpn = qpn; list_add_tail(&dqp->list, &entry->duplicates); return 0; } /* Check whether a qpn is a duplicate on steering entry * If so, it should not be removed from mgm */ static bool check_duplicate_entry(struct mlx4_dev *dev, u8 port, enum mlx4_steer_type steer, unsigned int index, u32 qpn) { struct mlx4_steer *s_steer; struct mlx4_steer_index *tmp_entry, *entry = NULL; struct mlx4_promisc_qp *dqp, *tmp_dqp; if (port < 1 || port > dev->caps.num_ports) return NULL; s_steer = &mlx4_priv(dev)->steer[port - 1]; /* if qp is not promisc, it cannot be duplicated */ if (!get_promisc_qp(dev, port, steer, qpn)) return false; /* The qp is promisc qp so it is a duplicate on this index * Find the index entry, and remove the duplicate */ list_for_each_entry(tmp_entry, &s_steer->steer_entries[steer], list) { if (tmp_entry->index == index) { entry = tmp_entry; break; } } if (unlikely(!entry)) { mlx4_warn(dev, "Steering entry for index %x is not registered\n", index); return false; } list_for_each_entry_safe(dqp, tmp_dqp, &entry->duplicates, list) { if (dqp->qpn == qpn) { list_del(&dqp->list); kfree(dqp); } } return true; } /* * returns true if all the QPs != tqpn contained in this entry * are Promisc QPs. return false otherwise. */ static bool promisc_steering_entry(struct mlx4_dev *dev, u8 port, enum mlx4_steer_type steer, unsigned int index, u32 tqpn, u32 *members_count) { struct mlx4_steer *s_steer; struct mlx4_cmd_mailbox *mailbox; struct mlx4_mgm *mgm; u32 m_count; bool ret = false; int i; if (port < 1 || port > dev->caps.num_ports) return false; s_steer = &mlx4_priv(dev)->steer[port - 1]; mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(mailbox)) return false; mgm = mailbox->buf; if (mlx4_READ_ENTRY(dev, index, mailbox)) goto out; m_count = be32_to_cpu(mgm->members_count) & 0xffffff; if (members_count) *members_count = m_count; for (i = 0; i < m_count; i++) { u32 qpn = be32_to_cpu(mgm->qp[i]) & MGM_QPN_MASK; if (!get_promisc_qp(dev, port, steer, qpn) && qpn != tqpn) { /* the qp is not promisc, the entry can't be removed */ goto out; } } ret = true; out: mlx4_free_cmd_mailbox(dev, mailbox); return ret; } /* IF a steering entry contains only promisc QPs, it can be removed. */ static bool can_remove_steering_entry(struct mlx4_dev *dev, u8 port, enum mlx4_steer_type steer, unsigned int index, u32 tqpn) { struct mlx4_steer *s_steer; struct mlx4_steer_index *entry = NULL, *tmp_entry; u32 members_count; bool ret = false; if (port < 1 || port > dev->caps.num_ports) return NULL; s_steer = &mlx4_priv(dev)->steer[port - 1]; if (!promisc_steering_entry(dev, port, steer, index, tqpn, &members_count)) goto out; /* All the qps currently registered for this entry are promiscuous, * Checking for duplicates */ ret = true; list_for_each_entry_safe(entry, tmp_entry, &s_steer->steer_entries[steer], list) { if (entry->index == index) { if (list_empty(&entry->duplicates) || members_count == 1) { struct mlx4_promisc_qp *pqp, *tmp_pqp; /* * If there is only 1 entry in duplicates than * this is the QP we want to delete, going over * the list and deleting the entry. */ list_del(&entry->list); list_for_each_entry_safe(pqp, tmp_pqp, &entry->duplicates, list) { list_del(&pqp->list); kfree(pqp); } kfree(entry); } else { /* This entry contains duplicates so it shouldn't be removed */ ret = false; goto out; } } } out: return ret; } static int add_promisc_qp(struct mlx4_dev *dev, u8 port, enum mlx4_steer_type steer, u32 qpn) { struct mlx4_steer *s_steer; struct mlx4_cmd_mailbox *mailbox; struct mlx4_mgm *mgm; struct mlx4_steer_index *entry; struct mlx4_promisc_qp *pqp; struct mlx4_promisc_qp *dqp; u32 members_count; u32 prot; int i; bool found; int err; struct mlx4_priv *priv = mlx4_priv(dev); if (port < 1 || port > dev->caps.num_ports) return -EINVAL; s_steer = &mlx4_priv(dev)->steer[port - 1]; mutex_lock(&priv->mcg_table.mutex); if (get_promisc_qp(dev, port, steer, qpn)) { err = 0; /* Noting to do, already exists */ goto out_mutex; } pqp = kmalloc(sizeof *pqp, GFP_KERNEL); if (!pqp) { err = -ENOMEM; goto out_mutex; } pqp->qpn = qpn; mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(mailbox)) { err = -ENOMEM; goto out_alloc; } mgm = mailbox->buf; if (!(mlx4_is_mfunc(dev) && steer == MLX4_UC_STEER)) { /* the promisc qp needs to be added for each one of the steering * entries, if it already exists, needs to be added as a duplicate * for this entry */ list_for_each_entry(entry, &s_steer->steer_entries[steer], list) { err = mlx4_READ_ENTRY(dev, entry->index, mailbox); if (err) goto out_mailbox; members_count = be32_to_cpu(mgm->members_count) & 0xffffff; prot = be32_to_cpu(mgm->members_count) >> 30; found = false; for (i = 0; i < members_count; i++) { if ((be32_to_cpu(mgm->qp[i]) & MGM_QPN_MASK) == qpn) { /* Entry already exists, add to duplicates */ dqp = kmalloc(sizeof *dqp, GFP_KERNEL); if (!dqp) { err = -ENOMEM; goto out_mailbox; } dqp->qpn = qpn; list_add_tail(&dqp->list, &entry->duplicates); found = true; } } if (!found) { /* Need to add the qpn to mgm */ if (members_count == dev->caps.num_qp_per_mgm) { /* entry is full */ err = -ENOMEM; goto out_mailbox; } mgm->qp[members_count++] = cpu_to_be32(qpn & MGM_QPN_MASK); mgm->members_count = cpu_to_be32(members_count | (prot << 30)); err = mlx4_WRITE_ENTRY(dev, entry->index, mailbox); if (err) goto out_mailbox; } } } /* add the new qpn to list of promisc qps */ list_add_tail(&pqp->list, &s_steer->promisc_qps[steer]); /* now need to add all the promisc qps to default entry */ memset(mgm, 0, sizeof *mgm); members_count = 0; list_for_each_entry(dqp, &s_steer->promisc_qps[steer], list) { if (members_count == dev->caps.num_qp_per_mgm) { /* entry is full */ err = -ENOMEM; goto out_list; } mgm->qp[members_count++] = cpu_to_be32(dqp->qpn & MGM_QPN_MASK); } mgm->members_count = cpu_to_be32(members_count | MLX4_PROT_ETH << 30); err = mlx4_WRITE_PROMISC(dev, port, steer, mailbox); if (err) goto out_list; mlx4_free_cmd_mailbox(dev, mailbox); mutex_unlock(&priv->mcg_table.mutex); return 0; out_list: list_del(&pqp->list); out_mailbox: mlx4_free_cmd_mailbox(dev, mailbox); out_alloc: kfree(pqp); out_mutex: mutex_unlock(&priv->mcg_table.mutex); return err; } static int remove_promisc_qp(struct mlx4_dev *dev, u8 port, enum mlx4_steer_type steer, u32 qpn) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_steer *s_steer; struct mlx4_cmd_mailbox *mailbox; struct mlx4_mgm *mgm; struct mlx4_steer_index *entry, *tmp_entry; struct mlx4_promisc_qp *pqp; struct mlx4_promisc_qp *dqp; u32 members_count; bool found; bool back_to_list = false; int i, loc = -1; int err; if (port < 1 || port > dev->caps.num_ports) return -EINVAL; s_steer = &mlx4_priv(dev)->steer[port - 1]; mutex_lock(&priv->mcg_table.mutex); pqp = get_promisc_qp(dev, port, steer, qpn); if (unlikely(!pqp)) { mlx4_warn(dev, "QP %x is not promiscuous QP\n", qpn); /* nothing to do */ err = 0; goto out_mutex; } /*remove from list of promisc qps */ list_del(&pqp->list); /* set the default entry not to include the removed one */ mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(mailbox)) { err = -ENOMEM; back_to_list = true; goto out_list; } mgm = mailbox->buf; memset(mgm, 0, sizeof *mgm); members_count = 0; list_for_each_entry(dqp, &s_steer->promisc_qps[steer], list) mgm->qp[members_count++] = cpu_to_be32(dqp->qpn & MGM_QPN_MASK); mgm->members_count = cpu_to_be32(members_count | MLX4_PROT_ETH << 30); err = mlx4_WRITE_PROMISC(dev, port, steer, mailbox); if (err) goto out_mailbox; if (!(mlx4_is_mfunc(dev) && steer == MLX4_UC_STEER)) { /* remove the qp from all the steering entries*/ list_for_each_entry_safe(entry, tmp_entry, &s_steer->steer_entries[steer], list) { found = false; list_for_each_entry(dqp, &entry->duplicates, list) { if (dqp->qpn == qpn) { found = true; break; } } if (found) { /* a duplicate, no need to change the mgm, * only update the duplicates list */ list_del(&dqp->list); kfree(dqp); } else { err = mlx4_READ_ENTRY(dev, entry->index, mailbox); if (err) goto out_mailbox; members_count = be32_to_cpu(mgm->members_count) & 0xffffff; if (!members_count) { mlx4_warn(dev, "QP %06x wasn't found in entry %x mcount=0." " deleting entry...\n", qpn, entry->index); list_del(&entry->list); kfree(entry); continue; } for (i = 0; i < members_count; ++i) if ((be32_to_cpu(mgm->qp[i]) & MGM_QPN_MASK) == qpn) { loc = i; break; } if (loc < 0) { mlx4_err(dev, "QP %06x wasn't found in entry %d\n", qpn, entry->index); err = -EINVAL; goto out_mailbox; } /* copy the last QP in this MGM over removed QP */ mgm->qp[loc] = mgm->qp[members_count - 1]; mgm->qp[members_count - 1] = 0; mgm->members_count = cpu_to_be32(--members_count | (MLX4_PROT_ETH << 30)); err = mlx4_WRITE_ENTRY(dev, entry->index, mailbox); if (err) goto out_mailbox; } } } out_mailbox: mlx4_free_cmd_mailbox(dev, mailbox); out_list: if (back_to_list) list_add_tail(&pqp->list, &s_steer->promisc_qps[steer]); else kfree(pqp); out_mutex: mutex_unlock(&priv->mcg_table.mutex); return err; } /* * Caller must hold MCG table semaphore. gid and mgm parameters must * be properly aligned for command interface. * * Returns 0 unless a firmware command error occurs. * * If GID is found in MGM or MGM is empty, *index = *hash, *prev = -1 * and *mgm holds MGM entry. * * if GID is found in AMGM, *index = index in AMGM, *prev = index of * previous entry in hash chain and *mgm holds AMGM entry. * * If no AMGM exists for given gid, *index = -1, *prev = index of last * entry in hash chain and *mgm holds end of hash chain. */ static int find_entry(struct mlx4_dev *dev, u8 port, u8 *gid, enum mlx4_protocol prot, struct mlx4_cmd_mailbox *mgm_mailbox, int *prev, int *index) { struct mlx4_cmd_mailbox *mailbox; struct mlx4_mgm *mgm = mgm_mailbox->buf; u8 *mgid; int err; u16 hash; u8 op_mod = (prot == MLX4_PROT_ETH) ? !!(dev->caps.flags & MLX4_DEV_CAP_FLAG_VEP_MC_STEER) : 0; mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(mailbox)) return -ENOMEM; mgid = mailbox->buf; memcpy(mgid, gid, 16); err = mlx4_GID_HASH(dev, mailbox, &hash, op_mod); mlx4_free_cmd_mailbox(dev, mailbox); if (err) return err; - if (0) - mlx4_dbg(dev, "Hash for %pI6 is %04x\n", gid, hash); + if (0) { + mlx4_dbg(dev, "Hash for "GID_PRINT_FMT" is %04x\n", + GID_PRINT_ARGS(gid), hash); + } *index = hash; *prev = -1; do { err = mlx4_READ_ENTRY(dev, *index, mgm_mailbox); if (err) return err; if (!(be32_to_cpu(mgm->members_count) & 0xffffff)) { if (*index != hash) { mlx4_err(dev, "Found zero MGID in AMGM.\n"); err = -EINVAL; } return err; } if (!memcmp(mgm->gid, gid, 16) && be32_to_cpu(mgm->members_count) >> 30 == prot) return err; *prev = *index; *index = be32_to_cpu(mgm->next_gid_index) >> 6; } while (*index); *index = -1; return err; } static const u8 __promisc_mode[] = { [MLX4_FS_REGULAR] = 0x0, [MLX4_FS_ALL_DEFAULT] = 0x1, [MLX4_FS_MC_DEFAULT] = 0x3, [MLX4_FS_UC_SNIFFER] = 0x4, [MLX4_FS_MC_SNIFFER] = 0x5, }; int map_sw_to_hw_steering_mode(struct mlx4_dev *dev, enum mlx4_net_trans_promisc_mode flow_type) { if (flow_type >= MLX4_FS_MODE_NUM || flow_type < 0) { mlx4_err(dev, "Invalid flow type. type = %d\n", flow_type); return -EINVAL; } return __promisc_mode[flow_type]; } EXPORT_SYMBOL_GPL(map_sw_to_hw_steering_mode); static void trans_rule_ctrl_to_hw(struct mlx4_net_trans_rule *ctrl, struct mlx4_net_trans_rule_hw_ctrl *hw) { u8 flags = 0; flags = ctrl->queue_mode == MLX4_NET_TRANS_Q_LIFO ? 1 : 0; flags |= ctrl->exclusive ? (1 << 2) : 0; flags |= ctrl->allow_loopback ? (1 << 3) : 0; hw->flags = flags; hw->type = __promisc_mode[ctrl->promisc_mode]; hw->prio = cpu_to_be16(ctrl->priority); hw->port = ctrl->port; hw->qpn = cpu_to_be32(ctrl->qpn); } const u16 __sw_id_hw[] = { [MLX4_NET_TRANS_RULE_ID_ETH] = 0xE001, [MLX4_NET_TRANS_RULE_ID_IB] = 0xE005, [MLX4_NET_TRANS_RULE_ID_IPV6] = 0xE003, [MLX4_NET_TRANS_RULE_ID_IPV4] = 0xE002, [MLX4_NET_TRANS_RULE_ID_TCP] = 0xE004, [MLX4_NET_TRANS_RULE_ID_UDP] = 0xE006 }; int map_sw_to_hw_steering_id(struct mlx4_dev *dev, enum mlx4_net_trans_rule_id id) { if (id >= MLX4_NET_TRANS_RULE_NUM || id < 0) { mlx4_err(dev, "Invalid network rule id. id = %d\n", id); return -EINVAL; } return __sw_id_hw[id]; } EXPORT_SYMBOL_GPL(map_sw_to_hw_steering_id); static const int __rule_hw_sz[] = { [MLX4_NET_TRANS_RULE_ID_ETH] = sizeof(struct mlx4_net_trans_rule_hw_eth), [MLX4_NET_TRANS_RULE_ID_IB] = sizeof(struct mlx4_net_trans_rule_hw_ib), [MLX4_NET_TRANS_RULE_ID_IPV6] = 0, [MLX4_NET_TRANS_RULE_ID_IPV4] = sizeof(struct mlx4_net_trans_rule_hw_ipv4), [MLX4_NET_TRANS_RULE_ID_TCP] = sizeof(struct mlx4_net_trans_rule_hw_tcp_udp), [MLX4_NET_TRANS_RULE_ID_UDP] = sizeof(struct mlx4_net_trans_rule_hw_tcp_udp) }; int hw_rule_sz(struct mlx4_dev *dev, enum mlx4_net_trans_rule_id id) { if (id >= MLX4_NET_TRANS_RULE_NUM || id < 0) { mlx4_err(dev, "Invalid network rule id. id = %d\n", id); return -EINVAL; } return __rule_hw_sz[id]; } EXPORT_SYMBOL_GPL(hw_rule_sz); static int parse_trans_rule(struct mlx4_dev *dev, struct mlx4_spec_list *spec, struct _rule_hw *rule_hw) { if (hw_rule_sz(dev, spec->id) < 0) return -EINVAL; memset(rule_hw, 0, hw_rule_sz(dev, spec->id)); rule_hw->id = cpu_to_be16(__sw_id_hw[spec->id]); rule_hw->size = hw_rule_sz(dev, spec->id) >> 2; switch (spec->id) { case MLX4_NET_TRANS_RULE_ID_ETH: memcpy(rule_hw->eth.dst_mac, spec->eth.dst_mac, ETH_ALEN); memcpy(rule_hw->eth.dst_mac_msk, spec->eth.dst_mac_msk, ETH_ALEN); memcpy(rule_hw->eth.src_mac, spec->eth.src_mac, ETH_ALEN); memcpy(rule_hw->eth.src_mac_msk, spec->eth.src_mac_msk, ETH_ALEN); if (spec->eth.ether_type_enable) { rule_hw->eth.ether_type_enable = 1; rule_hw->eth.ether_type = spec->eth.ether_type; } rule_hw->eth.vlan_tag = spec->eth.vlan_id; rule_hw->eth.vlan_tag_msk = spec->eth.vlan_id_msk; break; case MLX4_NET_TRANS_RULE_ID_IB: rule_hw->ib.l3_qpn = spec->ib.l3_qpn; rule_hw->ib.qpn_mask = spec->ib.qpn_msk; memcpy(&rule_hw->ib.dst_gid, &spec->ib.dst_gid, 16); memcpy(&rule_hw->ib.dst_gid_msk, &spec->ib.dst_gid_msk, 16); break; case MLX4_NET_TRANS_RULE_ID_IPV6: return -EOPNOTSUPP; case MLX4_NET_TRANS_RULE_ID_IPV4: rule_hw->ipv4.src_ip = spec->ipv4.src_ip; rule_hw->ipv4.src_ip_msk = spec->ipv4.src_ip_msk; rule_hw->ipv4.dst_ip = spec->ipv4.dst_ip; rule_hw->ipv4.dst_ip_msk = spec->ipv4.dst_ip_msk; break; case MLX4_NET_TRANS_RULE_ID_TCP: case MLX4_NET_TRANS_RULE_ID_UDP: rule_hw->tcp_udp.dst_port = spec->tcp_udp.dst_port; rule_hw->tcp_udp.dst_port_msk = spec->tcp_udp.dst_port_msk; rule_hw->tcp_udp.src_port = spec->tcp_udp.src_port; rule_hw->tcp_udp.src_port_msk = spec->tcp_udp.src_port_msk; break; default: return -EINVAL; } return __rule_hw_sz[spec->id]; } static void mlx4_err_rule(struct mlx4_dev *dev, char *str, struct mlx4_net_trans_rule *rule) { #define BUF_SIZE 256 struct mlx4_spec_list *cur; char buf[BUF_SIZE]; int len = 0; mlx4_err(dev, "%s", str); len += snprintf(buf + len, BUF_SIZE - len, "port = %d prio = 0x%x qp = 0x%x ", rule->port, rule->priority, rule->qpn); list_for_each_entry(cur, &rule->list, list) { switch (cur->id) { case MLX4_NET_TRANS_RULE_ID_ETH: len += snprintf(buf + len, BUF_SIZE - len, "dmac = %pM ", &cur->eth.dst_mac); if (cur->eth.ether_type) len += snprintf(buf + len, BUF_SIZE - len, "ethertype = 0x%x ", be16_to_cpu(cur->eth.ether_type)); if (cur->eth.vlan_id) len += snprintf(buf + len, BUF_SIZE - len, "vlan-id = %d ", be16_to_cpu(cur->eth.vlan_id)); break; case MLX4_NET_TRANS_RULE_ID_IPV4: if (cur->ipv4.src_ip) len += snprintf(buf + len, BUF_SIZE - len, "src-ip = %pI4 ", &cur->ipv4.src_ip); if (cur->ipv4.dst_ip) len += snprintf(buf + len, BUF_SIZE - len, "dst-ip = %pI4 ", &cur->ipv4.dst_ip); break; case MLX4_NET_TRANS_RULE_ID_TCP: case MLX4_NET_TRANS_RULE_ID_UDP: if (cur->tcp_udp.src_port) len += snprintf(buf + len, BUF_SIZE - len, "src-port = %d ", be16_to_cpu(cur->tcp_udp.src_port)); if (cur->tcp_udp.dst_port) len += snprintf(buf + len, BUF_SIZE - len, "dst-port = %d ", be16_to_cpu(cur->tcp_udp.dst_port)); break; case MLX4_NET_TRANS_RULE_ID_IB: len += snprintf(buf + len, BUF_SIZE - len, - "dst-gid = %pI6\n", cur->ib.dst_gid); + "dst-gid = "GID_PRINT_FMT"\n", + GID_PRINT_ARGS(cur->ib.dst_gid)); len += snprintf(buf + len, BUF_SIZE - len, - "dst-gid-mask = %pI6\n", - cur->ib.dst_gid_msk); + "dst-gid-mask = "GID_PRINT_FMT"\n", + GID_PRINT_ARGS(cur->ib.dst_gid_msk)); break; case MLX4_NET_TRANS_RULE_ID_IPV6: break; default: break; } } len += snprintf(buf + len, BUF_SIZE - len, "\n"); mlx4_err(dev, "%s", buf); if (len >= BUF_SIZE) mlx4_err(dev, "Network rule error message was truncated, print buffer is too small.\n"); } int mlx4_flow_attach(struct mlx4_dev *dev, struct mlx4_net_trans_rule *rule, u64 *reg_id) { struct mlx4_cmd_mailbox *mailbox; struct mlx4_spec_list *cur; u32 size = 0; int ret; mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(mailbox)) return PTR_ERR(mailbox); memset(mailbox->buf, 0, sizeof(struct mlx4_net_trans_rule_hw_ctrl)); trans_rule_ctrl_to_hw(rule, mailbox->buf); size += sizeof(struct mlx4_net_trans_rule_hw_ctrl); list_for_each_entry(cur, &rule->list, list) { ret = parse_trans_rule(dev, cur, mailbox->buf + size); if (ret < 0) { mlx4_free_cmd_mailbox(dev, mailbox); return -EINVAL; } size += ret; } ret = mlx4_QP_FLOW_STEERING_ATTACH(dev, mailbox, size >> 2, reg_id); if (ret == -ENOMEM) mlx4_err_rule(dev, "mcg table is full. Fail to register network rule.\n", rule); else if (ret) mlx4_err_rule(dev, "Fail to register network rule.\n", rule); mlx4_free_cmd_mailbox(dev, mailbox); return ret; } EXPORT_SYMBOL_GPL(mlx4_flow_attach); int mlx4_flow_detach(struct mlx4_dev *dev, u64 reg_id) { int err; err = mlx4_QP_FLOW_STEERING_DETACH(dev, reg_id); if (err) mlx4_err(dev, "Fail to detach network rule. registration id = 0x%llx\n", (unsigned long long)reg_id); return err; } EXPORT_SYMBOL_GPL(mlx4_flow_detach); int mlx4_FLOW_STEERING_IB_UC_QP_RANGE(struct mlx4_dev *dev, u32 min_range_qpn, u32 max_range_qpn) { int err; u64 in_param; in_param = ((u64) min_range_qpn) << 32; in_param |= ((u64) max_range_qpn) & 0xFFFFFFFF; err = mlx4_cmd(dev, in_param, 0, 0, MLX4_FLOW_STEERING_IB_UC_QP_RANGE, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE); return err; } EXPORT_SYMBOL_GPL(mlx4_FLOW_STEERING_IB_UC_QP_RANGE); int mlx4_qp_attach_common(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16], int block_mcast_loopback, enum mlx4_protocol prot, enum mlx4_steer_type steer) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_cmd_mailbox *mailbox; struct mlx4_mgm *mgm; u32 members_count; int index, prev; int link = 0; int i; int err; u8 port = gid[5]; u8 new_entry = 0; mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(mailbox)) return PTR_ERR(mailbox); mgm = mailbox->buf; mutex_lock(&priv->mcg_table.mutex); err = find_entry(dev, port, gid, prot, mailbox, &prev, &index); if (err) goto out; if (index != -1) { if (!(be32_to_cpu(mgm->members_count) & 0xffffff)) { new_entry = 1; memcpy(mgm->gid, gid, 16); } } else { link = 1; index = mlx4_bitmap_alloc(&priv->mcg_table.bitmap); if (index == -1) { mlx4_err(dev, "No AMGM entries left\n"); err = -ENOMEM; goto out; } index += dev->caps.num_mgms; new_entry = 1; memset(mgm, 0, sizeof *mgm); memcpy(mgm->gid, gid, 16); } members_count = be32_to_cpu(mgm->members_count) & 0xffffff; if (members_count == dev->caps.num_qp_per_mgm) { mlx4_err(dev, "MGM at index %x is full.\n", index); err = -ENOMEM; goto out; } for (i = 0; i < members_count; ++i) if ((be32_to_cpu(mgm->qp[i]) & MGM_QPN_MASK) == qp->qpn) { mlx4_dbg(dev, "QP %06x already a member of MGM\n", qp->qpn); err = 0; goto out; } mgm->qp[members_count++] = cpu_to_be32((qp->qpn & MGM_QPN_MASK) | (!!mlx4_blck_lb << MGM_BLCK_LB_BIT)); mgm->members_count = cpu_to_be32(members_count | (u32) prot << 30); err = mlx4_WRITE_ENTRY(dev, index, mailbox); if (err) goto out; /* if !link, still add the new entry. */ if (!link) goto skip_link; err = mlx4_READ_ENTRY(dev, prev, mailbox); if (err) goto out; mgm->next_gid_index = cpu_to_be32(index << 6); err = mlx4_WRITE_ENTRY(dev, prev, mailbox); if (err) goto out; skip_link: if (prot == MLX4_PROT_ETH) { /* manage the steering entry for promisc mode */ if (new_entry) new_steering_entry(dev, port, steer, index, qp->qpn); else existing_steering_entry(dev, port, steer, index, qp->qpn); } out: if (err && link && index != -1) { if (index < dev->caps.num_mgms) mlx4_warn(dev, "Got AMGM index %d < %d", index, dev->caps.num_mgms); else mlx4_bitmap_free(&priv->mcg_table.bitmap, index - dev->caps.num_mgms, MLX4_USE_RR); } mutex_unlock(&priv->mcg_table.mutex); mlx4_free_cmd_mailbox(dev, mailbox); return err; } int mlx4_qp_detach_common(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16], enum mlx4_protocol prot, enum mlx4_steer_type steer) { struct mlx4_priv *priv = mlx4_priv(dev); struct mlx4_cmd_mailbox *mailbox; struct mlx4_mgm *mgm; u32 members_count; int prev, index; int i, loc = -1; int err; u8 port = gid[5]; bool removed_entry = false; mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(mailbox)) return PTR_ERR(mailbox); mgm = mailbox->buf; mutex_lock(&priv->mcg_table.mutex); err = find_entry(dev, port, gid, prot, mailbox, &prev, &index); if (err) goto out; if (index == -1) { - mlx4_err(dev, "MGID %pI6 not found\n", gid); + mlx4_err(dev, "MGID "GID_PRINT_FMT" not found\n", + GID_PRINT_ARGS(gid)); err = -EINVAL; goto out; } /* if this QP is also a promisc QP, it shouldn't be removed only if at least one none promisc QP is also attached to this MCG */ if (prot == MLX4_PROT_ETH && check_duplicate_entry(dev, port, steer, index, qp->qpn) && !promisc_steering_entry(dev, port, steer, index, qp->qpn, NULL)) goto out; members_count = be32_to_cpu(mgm->members_count) & 0xffffff; for (i = 0; i < members_count; ++i) if ((be32_to_cpu(mgm->qp[i]) & MGM_QPN_MASK) == qp->qpn) { loc = i; break; } if (loc == -1) { mlx4_err(dev, "QP %06x not found in MGM\n", qp->qpn); err = -EINVAL; goto out; } /* copy the last QP in this MGM over removed QP */ mgm->qp[loc] = mgm->qp[members_count - 1]; mgm->qp[members_count - 1] = 0; mgm->members_count = cpu_to_be32(--members_count | (u32) prot << 30); if (prot == MLX4_PROT_ETH) removed_entry = can_remove_steering_entry(dev, port, steer, index, qp->qpn); if (members_count && (prot != MLX4_PROT_ETH || !removed_entry)) { err = mlx4_WRITE_ENTRY(dev, index, mailbox); goto out; } /* We are going to delete the entry, members count should be 0 */ mgm->members_count = cpu_to_be32((u32) prot << 30); if (prev == -1) { /* Remove entry from MGM */ int amgm_index = be32_to_cpu(mgm->next_gid_index) >> 6; if (amgm_index) { err = mlx4_READ_ENTRY(dev, amgm_index, mailbox); if (err) goto out; } else memset(mgm->gid, 0, 16); err = mlx4_WRITE_ENTRY(dev, index, mailbox); if (err) goto out; if (amgm_index) { if (amgm_index < dev->caps.num_mgms) mlx4_warn(dev, "MGM entry %d had AMGM index %d < %d", index, amgm_index, dev->caps.num_mgms); else mlx4_bitmap_free(&priv->mcg_table.bitmap, amgm_index - dev->caps.num_mgms, MLX4_USE_RR); } } else { /* Remove entry from AMGM */ int cur_next_index = be32_to_cpu(mgm->next_gid_index) >> 6; err = mlx4_READ_ENTRY(dev, prev, mailbox); if (err) goto out; mgm->next_gid_index = cpu_to_be32(cur_next_index << 6); err = mlx4_WRITE_ENTRY(dev, prev, mailbox); if (err) goto out; if (index < dev->caps.num_mgms) mlx4_warn(dev, "entry %d had next AMGM index %d < %d", prev, index, dev->caps.num_mgms); else mlx4_bitmap_free(&priv->mcg_table.bitmap, index - dev->caps.num_mgms, MLX4_USE_RR); } out: mutex_unlock(&priv->mcg_table.mutex); mlx4_free_cmd_mailbox(dev, mailbox); return err; } static int mlx4_QP_ATTACH(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16], u8 attach, u8 block_loopback, enum mlx4_protocol prot) { struct mlx4_cmd_mailbox *mailbox; int err = 0; int qpn; if (!mlx4_is_mfunc(dev)) return -EBADF; mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(mailbox)) return PTR_ERR(mailbox); memcpy(mailbox->buf, gid, 16); qpn = qp->qpn; qpn |= (prot << 28); if (attach && block_loopback) qpn |= (1 << 31); err = mlx4_cmd(dev, mailbox->dma, qpn, attach, MLX4_CMD_QP_ATTACH, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED); mlx4_free_cmd_mailbox(dev, mailbox); return err; } int mlx4_trans_to_dmfs_attach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16], u8 port, int block_mcast_loopback, enum mlx4_protocol prot, u64 *reg_id) { struct mlx4_spec_list spec = { {NULL} }; __be64 mac_mask = cpu_to_be64(MLX4_MAC_MASK << 16); struct mlx4_net_trans_rule rule = { .queue_mode = MLX4_NET_TRANS_Q_FIFO, .exclusive = 0, .promisc_mode = MLX4_FS_REGULAR, .priority = MLX4_DOMAIN_NIC, }; rule.allow_loopback = !block_mcast_loopback; rule.port = port; rule.qpn = qp->qpn; INIT_LIST_HEAD(&rule.list); switch (prot) { case MLX4_PROT_ETH: spec.id = MLX4_NET_TRANS_RULE_ID_ETH; memcpy(spec.eth.dst_mac, &gid[10], ETH_ALEN); memcpy(spec.eth.dst_mac_msk, &mac_mask, ETH_ALEN); break; case MLX4_PROT_IB_IPV6: spec.id = MLX4_NET_TRANS_RULE_ID_IB; memcpy(spec.ib.dst_gid, gid, 16); memset(&spec.ib.dst_gid_msk, 0xff, 16); break; default: return -EINVAL; } list_add_tail(&spec.list, &rule.list); return mlx4_flow_attach(dev, &rule, reg_id); } int mlx4_multicast_attach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16], u8 port, int block_mcast_loopback, enum mlx4_protocol prot, u64 *reg_id) { enum mlx4_steer_type steer; steer = (is_valid_ether_addr(&gid[10])) ? MLX4_UC_STEER : MLX4_MC_STEER; switch (dev->caps.steering_mode) { case MLX4_STEERING_MODE_A0: if (prot == MLX4_PROT_ETH) return 0; case MLX4_STEERING_MODE_B0: if (prot == MLX4_PROT_ETH) gid[7] |= (steer << 1); if (mlx4_is_mfunc(dev)) return mlx4_QP_ATTACH(dev, qp, gid, 1, block_mcast_loopback, prot); return mlx4_qp_attach_common(dev, qp, gid, block_mcast_loopback, prot, MLX4_MC_STEER); case MLX4_STEERING_MODE_DEVICE_MANAGED: return mlx4_trans_to_dmfs_attach(dev, qp, gid, port, block_mcast_loopback, prot, reg_id); default: return -EINVAL; } } EXPORT_SYMBOL_GPL(mlx4_multicast_attach); int mlx4_multicast_detach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16], enum mlx4_protocol prot, u64 reg_id) { enum mlx4_steer_type steer; steer = (is_valid_ether_addr(&gid[10])) ? MLX4_UC_STEER : MLX4_MC_STEER; switch (dev->caps.steering_mode) { case MLX4_STEERING_MODE_A0: if (prot == MLX4_PROT_ETH) return 0; case MLX4_STEERING_MODE_B0: if (prot == MLX4_PROT_ETH) gid[7] |= (steer << 1); if (mlx4_is_mfunc(dev)) return mlx4_QP_ATTACH(dev, qp, gid, 0, 0, prot); return mlx4_qp_detach_common(dev, qp, gid, prot, MLX4_MC_STEER); case MLX4_STEERING_MODE_DEVICE_MANAGED: return mlx4_flow_detach(dev, reg_id); default: return -EINVAL; } } EXPORT_SYMBOL_GPL(mlx4_multicast_detach); int mlx4_flow_steer_promisc_add(struct mlx4_dev *dev, u8 port, u32 qpn, enum mlx4_net_trans_promisc_mode mode) { struct mlx4_net_trans_rule rule; u64 *regid_p; switch (mode) { case MLX4_FS_ALL_DEFAULT: regid_p = &dev->regid_promisc_array[port]; break; case MLX4_FS_MC_DEFAULT: regid_p = &dev->regid_allmulti_array[port]; break; default: return -1; } if (*regid_p != 0) return -1; rule.promisc_mode = mode; rule.port = port; rule.qpn = qpn; INIT_LIST_HEAD(&rule.list); mlx4_err(dev, "going promisc on %x\n", port); return mlx4_flow_attach(dev, &rule, regid_p); } EXPORT_SYMBOL_GPL(mlx4_flow_steer_promisc_add); int mlx4_flow_steer_promisc_remove(struct mlx4_dev *dev, u8 port, enum mlx4_net_trans_promisc_mode mode) { int ret; u64 *regid_p; switch (mode) { case MLX4_FS_ALL_DEFAULT: regid_p = &dev->regid_promisc_array[port]; break; case MLX4_FS_MC_DEFAULT: regid_p = &dev->regid_allmulti_array[port]; break; default: return -1; } if (*regid_p == 0) return -1; ret = mlx4_flow_detach(dev, *regid_p); if (ret == 0) *regid_p = 0; return ret; } EXPORT_SYMBOL_GPL(mlx4_flow_steer_promisc_remove); int mlx4_unicast_attach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16], int block_mcast_loopback, enum mlx4_protocol prot) { if (prot == MLX4_PROT_ETH) gid[7] |= (MLX4_UC_STEER << 1); if (mlx4_is_mfunc(dev)) return mlx4_QP_ATTACH(dev, qp, gid, 1, block_mcast_loopback, prot); return mlx4_qp_attach_common(dev, qp, gid, block_mcast_loopback, prot, MLX4_UC_STEER); } EXPORT_SYMBOL_GPL(mlx4_unicast_attach); int mlx4_unicast_detach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16], enum mlx4_protocol prot) { if (prot == MLX4_PROT_ETH) gid[7] |= (MLX4_UC_STEER << 1); if (mlx4_is_mfunc(dev)) return mlx4_QP_ATTACH(dev, qp, gid, 0, 0, prot); return mlx4_qp_detach_common(dev, qp, gid, prot, MLX4_UC_STEER); } EXPORT_SYMBOL_GPL(mlx4_unicast_detach); int mlx4_PROMISC_wrapper(struct mlx4_dev *dev, int slave, struct mlx4_vhcr *vhcr, struct mlx4_cmd_mailbox *inbox, struct mlx4_cmd_mailbox *outbox, struct mlx4_cmd_info *cmd) { u32 qpn = (u32) vhcr->in_param & 0xffffffff; u8 port = vhcr->in_param >> 62; enum mlx4_steer_type steer = vhcr->in_modifier; /* Promiscuous unicast is not allowed in mfunc for VFs */ if ((slave != dev->caps.function) && (steer == MLX4_UC_STEER)) return 0; if (vhcr->op_modifier) return add_promisc_qp(dev, port, steer, qpn); else return remove_promisc_qp(dev, port, steer, qpn); } static int mlx4_PROMISC(struct mlx4_dev *dev, u32 qpn, enum mlx4_steer_type steer, u8 add, u8 port) { return mlx4_cmd(dev, (u64) qpn | (u64) port << 62, (u32) steer, add, MLX4_CMD_PROMISC, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED); } int mlx4_multicast_promisc_add(struct mlx4_dev *dev, u32 qpn, u8 port) { if (mlx4_is_mfunc(dev)) return mlx4_PROMISC(dev, qpn, MLX4_MC_STEER, 1, port); return add_promisc_qp(dev, port, MLX4_MC_STEER, qpn); } EXPORT_SYMBOL_GPL(mlx4_multicast_promisc_add); int mlx4_multicast_promisc_remove(struct mlx4_dev *dev, u32 qpn, u8 port) { if (mlx4_is_mfunc(dev)) return mlx4_PROMISC(dev, qpn, MLX4_MC_STEER, 0, port); return remove_promisc_qp(dev, port, MLX4_MC_STEER, qpn); } EXPORT_SYMBOL_GPL(mlx4_multicast_promisc_remove); int mlx4_unicast_promisc_add(struct mlx4_dev *dev, u32 qpn, u8 port) { if (mlx4_is_mfunc(dev)) return mlx4_PROMISC(dev, qpn, MLX4_UC_STEER, 1, port); return add_promisc_qp(dev, port, MLX4_UC_STEER, qpn); } EXPORT_SYMBOL_GPL(mlx4_unicast_promisc_add); int mlx4_unicast_promisc_remove(struct mlx4_dev *dev, u32 qpn, u8 port) { if (mlx4_is_mfunc(dev)) return mlx4_PROMISC(dev, qpn, MLX4_UC_STEER, 0, port); return remove_promisc_qp(dev, port, MLX4_UC_STEER, qpn); } EXPORT_SYMBOL_GPL(mlx4_unicast_promisc_remove); int mlx4_init_mcg_table(struct mlx4_dev *dev) { struct mlx4_priv *priv = mlx4_priv(dev); int err; /* No need for mcg_table when fw managed the mcg table*/ if (dev->caps.steering_mode == MLX4_STEERING_MODE_DEVICE_MANAGED) return 0; err = mlx4_bitmap_init(&priv->mcg_table.bitmap, dev->caps.num_amgms, dev->caps.num_amgms - 1, 0, 0); if (err) return err; mutex_init(&priv->mcg_table.mutex); return 0; } void mlx4_cleanup_mcg_table(struct mlx4_dev *dev) { if (dev->caps.steering_mode != MLX4_STEERING_MODE_DEVICE_MANAGED) mlx4_bitmap_cleanup(&mlx4_priv(dev)->mcg_table.bitmap); } Index: projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/mlx4_en.h =================================================================== --- projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/mlx4_en.h (revision 279716) +++ projects/cxl_iscsi/sys/ofed/drivers/net/mlx4/mlx4_en.h (revision 279717) @@ -1,951 +1,946 @@ /* * Copyright (c) 2007, 2014 Mellanox Technologies. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU * General Public License (GPL) Version 2, available from the file * COPYING in the main directory of this source tree, or the * OpenIB.org BSD license below: * * Redistribution and use in source and binary forms, with or * without modification, are permitted provided that the following * conditions are met: * * - Redistributions of source code must retain the above * copyright notice, this list of conditions and the following * disclaimer. * * - Redistributions in binary form must reproduce the above * copyright notice, this list of conditions and the following * disclaimer in the documentation and/or other materials * provided with the distribution. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. * */ #ifndef _MLX4_EN_H_ #define _MLX4_EN_H_ #include #include #include #include #include #include #include #include #ifdef CONFIG_MLX4_EN_DCB #include #endif #include #include #include #include #include #include #include #include "en_port.h" #include "mlx4_stats.h" #define DRV_NAME "mlx4_en" #define DRV_VERSION "2.1" #define DRV_RELDATE __DATE__ #define MLX4_EN_MSG_LEVEL (NETIF_MSG_LINK | NETIF_MSG_IFDOWN) /* * Device constants */ #define MLX4_EN_PAGE_SHIFT 12 #define MLX4_EN_PAGE_SIZE (1 << MLX4_EN_PAGE_SHIFT) #define DEF_RX_RINGS 16 #define MAX_RX_RINGS 128 #define MIN_RX_RINGS 4 #define TXBB_SIZE 64 #define HEADROOM (2048 / TXBB_SIZE + 1) #define STAMP_STRIDE 64 #define STAMP_DWORDS (STAMP_STRIDE / 4) #define STAMP_SHIFT 31 #define STAMP_VAL 0x7fffffff #define STATS_DELAY (HZ / 4) #define SERVICE_TASK_DELAY (HZ / 4) #define MAX_NUM_OF_FS_RULES 256 #define MLX4_EN_FILTER_HASH_SHIFT 4 #define MLX4_EN_FILTER_EXPIRY_QUOTA 60 #ifdef CONFIG_NET_RX_BUSY_POLL #define LL_EXTENDED_STATS #endif /* vlan valid range */ #define VLAN_MIN_VALUE 1 #define VLAN_MAX_VALUE 4094 /* Typical TSO descriptor with 16 gather entries is 352 bytes... */ #define MAX_DESC_SIZE 512 #define MAX_DESC_TXBBS (MAX_DESC_SIZE / TXBB_SIZE) /* * OS related constants and tunables */ #define MLX4_EN_WATCHDOG_TIMEOUT (15 * HZ) #define MLX4_EN_ALLOC_SIZE PAGE_ALIGN(PAGE_SIZE) #define MLX4_EN_ALLOC_ORDER get_order(MLX4_EN_ALLOC_SIZE) enum mlx4_en_alloc_type { MLX4_EN_ALLOC_NEW = 0, MLX4_EN_ALLOC_REPLACEMENT = 1, }; /* Receive fragment sizes; we use at most 3 fragments (for 9600 byte MTU * and 4K allocations) */ #if MJUMPAGESIZE == 4096 enum { FRAG_SZ0 = MCLBYTES, FRAG_SZ1 = MJUMPAGESIZE, FRAG_SZ2 = MJUMPAGESIZE, }; #define MLX4_EN_MAX_RX_FRAGS 3 #elif MJUMPAGESIZE == 8192 enum { FRAG_SZ0 = MCLBYTES, FRAG_SZ1 = MJUMPAGESIZE, }; #define MLX4_EN_MAX_RX_FRAGS 2 #elif MJUMPAGESIZE == 8192 #else #error "Unknown PAGE_SIZE" #endif /* Maximum ring sizes */ #define MLX4_EN_DEF_TX_QUEUE_SIZE 4096 /* Minimum packet number till arming the CQ */ #define MLX4_EN_MIN_RX_ARM 2048 #define MLX4_EN_MIN_TX_ARM 2048 /* Maximum ring sizes */ #define MLX4_EN_MAX_TX_SIZE 8192 #define MLX4_EN_MAX_RX_SIZE 8192 /* Minimum ring sizes */ #define MLX4_EN_MIN_RX_SIZE (4096 / TXBB_SIZE) #define MLX4_EN_MIN_TX_SIZE (4096 / TXBB_SIZE) #define MLX4_EN_SMALL_PKT_SIZE 64 #define MLX4_EN_MAX_TX_RING_P_UP 32 #define MLX4_EN_NUM_UP 1 #define MAX_TX_RINGS (MLX4_EN_MAX_TX_RING_P_UP * \ - (MLX4_EN_NUM_UP + 1)) + MLX4_EN_NUM_UP) #define MLX4_EN_DEF_TX_RING_SIZE 1024 #define MLX4_EN_DEF_RX_RING_SIZE 1024 /* Target number of bytes to coalesce with interrupt moderation */ #define MLX4_EN_RX_COAL_TARGET 0x20000 #define MLX4_EN_RX_COAL_TIME 0x10 #define MLX4_EN_TX_COAL_PKTS 64 #define MLX4_EN_TX_COAL_TIME 64 #define MLX4_EN_RX_RATE_LOW 400000 #define MLX4_EN_RX_COAL_TIME_LOW 0 #define MLX4_EN_RX_RATE_HIGH 450000 #define MLX4_EN_RX_COAL_TIME_HIGH 128 #define MLX4_EN_RX_SIZE_THRESH 1024 #define MLX4_EN_RX_RATE_THRESH (1000000 / MLX4_EN_RX_COAL_TIME_HIGH) #define MLX4_EN_SAMPLE_INTERVAL 0 #define MLX4_EN_AVG_PKT_SMALL 256 #define MLX4_EN_AUTO_CONF 0xffff #define MLX4_EN_DEF_RX_PAUSE 1 #define MLX4_EN_DEF_TX_PAUSE 1 /* Interval between successive polls in the Tx routine when polling is used instead of interrupts (in per-core Tx rings) - should be power of 2 */ #define MLX4_EN_TX_POLL_MODER 16 #define MLX4_EN_TX_POLL_TIMEOUT (HZ / 4) #define MLX4_EN_64_ALIGN (64 - NET_SKB_PAD) #define SMALL_PACKET_SIZE (256 - NET_IP_ALIGN) #define HEADER_COPY_SIZE (128) #define MLX4_LOOPBACK_TEST_PAYLOAD (HEADER_COPY_SIZE - ETHER_HDR_LEN) #define MLX4_EN_MIN_MTU 46 #define ETH_BCAST 0xffffffffffffULL #define MLX4_EN_LOOPBACK_RETRIES 5 #define MLX4_EN_LOOPBACK_TIMEOUT 100 #ifdef MLX4_EN_PERF_STAT /* Number of samples to 'average' */ #define AVG_SIZE 128 #define AVG_FACTOR 1024 #define INC_PERF_COUNTER(cnt) (++(cnt)) #define ADD_PERF_COUNTER(cnt, add) ((cnt) += (add)) #define AVG_PERF_COUNTER(cnt, sample) \ ((cnt) = ((cnt) * (AVG_SIZE - 1) + (sample) * AVG_FACTOR) / AVG_SIZE) #define GET_PERF_COUNTER(cnt) (cnt) #define GET_AVG_PERF_COUNTER(cnt) ((cnt) / AVG_FACTOR) #else #define INC_PERF_COUNTER(cnt) do {} while (0) #define ADD_PERF_COUNTER(cnt, add) do {} while (0) #define AVG_PERF_COUNTER(cnt, sample) do {} while (0) #define GET_PERF_COUNTER(cnt) (0) #define GET_AVG_PERF_COUNTER(cnt) (0) #endif /* MLX4_EN_PERF_STAT */ /* * Configurables */ enum cq_type { RX = 0, TX = 1, }; /* * Useful macros */ #define ROUNDUP_LOG2(x) ilog2(roundup_pow_of_two(x)) #define XNOR(x, y) (!(x) == !(y)) #define ILLEGAL_MAC(addr) (addr == 0xffffffffffffULL || addr == 0x0) struct mlx4_en_tx_info { struct mbuf *mb; u32 nr_txbb; u32 nr_bytes; u8 linear; u8 nr_segs; u8 data_offset; u8 inl; #if 0 u8 ts_requested; #endif }; #define MLX4_EN_BIT_DESC_OWN 0x80000000 #define CTRL_SIZE sizeof(struct mlx4_wqe_ctrl_seg) #define MLX4_EN_MEMTYPE_PAD 0x100 #define DS_SIZE sizeof(struct mlx4_wqe_data_seg) struct mlx4_en_tx_desc { struct mlx4_wqe_ctrl_seg ctrl; union { struct mlx4_wqe_data_seg data; /* at least one data segment */ struct mlx4_wqe_lso_seg lso; struct mlx4_wqe_inline_seg inl; }; }; #define MLX4_EN_USE_SRQ 0x01000000 #define MLX4_EN_TX_BUDGET 64*4 //Compensate for no NAPI in freeBSD - might need some fine tunning in the future. #define MLX4_EN_RX_BUDGET 64 #define MLX4_EN_CX3_LOW_ID 0x1000 #define MLX4_EN_CX3_HIGH_ID 0x1005 struct mlx4_en_tx_ring { spinlock_t tx_lock; struct mlx4_hwq_resources wqres; u32 size ; /* number of TXBBs */ u32 size_mask; u16 stride; u16 cqn; /* index of port CQ associated with this ring */ u32 prod; u32 cons; u32 buf_size; u32 doorbell_qpn; void *buf; u16 poll_cnt; int blocked; struct mlx4_en_tx_info *tx_info; u8 *bounce_buf; u8 queue_index; cpuset_t affinity_mask; struct buf_ring *br; u32 last_nr_txbb; struct mlx4_qp qp; struct mlx4_qp_context context; int qpn; enum mlx4_qp_state qp_state; struct mlx4_srq dummy; unsigned long bytes; unsigned long packets; unsigned long tx_csum; unsigned long queue_stopped; unsigned long wake_queue; struct mlx4_bf bf; bool bf_enabled; struct netdev_queue *tx_queue; int hwtstamp_tx_type; spinlock_t comp_lock; int full_size; int inline_thold; u64 watchdog_time; }; struct mlx4_en_rx_desc { /* actual number of entries depends on rx ring stride */ struct mlx4_wqe_data_seg data[0]; }; struct mlx4_en_rx_buf { dma_addr_t dma; struct page *page; unsigned int page_offset; }; struct mlx4_en_rx_ring { struct mlx4_hwq_resources wqres; u32 size ; /* number of Rx descs*/ u32 actual_size; u32 size_mask; u16 stride; u16 log_stride; u16 cqn; /* index of port CQ associated with this ring */ u32 prod; u32 cons; u32 buf_size; u8 fcs_del; u16 rx_alloc_order; u32 rx_alloc_size; u32 rx_buf_size; u32 rx_mb_size; int qpn; void *buf; void *rx_info; unsigned long errors; unsigned long bytes; unsigned long packets; #ifdef LL_EXTENDED_STATS unsigned long yields; unsigned long misses; unsigned long cleaned; #endif unsigned long csum_ok; unsigned long csum_none; int hwtstamp_rx_filter; int numa_node; struct lro_ctrl lro; }; static inline int mlx4_en_can_lro(__be16 status) { - static __be16 status_all; - static __be16 status_ipv4_ipok_tcp; - static __be16 status_ipv6_ipok_tcp; - - status_all = cpu_to_be16( + const __be16 status_all = cpu_to_be16( MLX4_CQE_STATUS_IPV4 | MLX4_CQE_STATUS_IPV4F | MLX4_CQE_STATUS_IPV6 | MLX4_CQE_STATUS_IPV4OPT | MLX4_CQE_STATUS_TCP | MLX4_CQE_STATUS_UDP | MLX4_CQE_STATUS_IPOK); - status_ipv4_ipok_tcp = cpu_to_be16( + const __be16 status_ipv4_ipok_tcp = cpu_to_be16( MLX4_CQE_STATUS_IPV4 | MLX4_CQE_STATUS_IPOK | MLX4_CQE_STATUS_TCP); - status_ipv6_ipok_tcp = cpu_to_be16( + const __be16 status_ipv6_ipok_tcp = cpu_to_be16( MLX4_CQE_STATUS_IPV6 | MLX4_CQE_STATUS_IPOK | MLX4_CQE_STATUS_TCP); status &= status_all; return (status == status_ipv4_ipok_tcp || status == status_ipv6_ipok_tcp); } - struct mlx4_en_cq { struct mlx4_cq mcq; struct mlx4_hwq_resources wqres; int ring; spinlock_t lock; struct net_device *dev; /* Per-core Tx cq processing support */ struct timer_list timer; int size; int buf_size; unsigned vector; enum cq_type is_tx; u16 moder_time; u16 moder_cnt; struct mlx4_cqe *buf; struct task cq_task; struct taskqueue *tq; #define MLX4_EN_OPCODE_ERROR 0x1e u32 tot_rx; u32 tot_tx; #ifdef CONFIG_NET_RX_BUSY_POLL unsigned int state; #define MLX4_EN_CQ_STATEIDLE 0 #define MLX4_EN_CQ_STATENAPI 1 /* NAPI owns this CQ */ #define MLX4_EN_CQ_STATEPOLL 2 /* poll owns this CQ */ #define MLX4_CQ_LOCKED (MLX4_EN_CQ_STATENAPI | MLX4_EN_CQ_STATEPOLL) #define MLX4_EN_CQ_STATENAPI_YIELD 4 /* NAPI yielded this CQ */ #define MLX4_EN_CQ_STATEPOLL_YIELD 8 /* poll yielded this CQ */ #define CQ_YIELD (MLX4_EN_CQ_STATENAPI_YIELD | MLX4_EN_CQ_STATEPOLL_YIELD) #define CQ_USER_PEND (MLX4_EN_CQ_STATEPOLL | MLX4_EN_CQ_STATEPOLL_YIELD) spinlock_t poll_lock; /* protects from LLS/napi conflicts */ #endif /* CONFIG_NET_RX_BUSY_POLL */ }; struct mlx4_en_port_profile { u32 flags; u32 tx_ring_num; u32 rx_ring_num; u32 tx_ring_size; u32 rx_ring_size; u8 rx_pause; u8 rx_ppp; u8 tx_pause; u8 tx_ppp; int rss_rings; }; struct mlx4_en_profile { int rss_xor; int udp_rss; u8 rss_mask; u32 active_ports; u32 small_pkt_int; u8 no_reset; u8 num_tx_rings_p_up; struct mlx4_en_port_profile prof[MLX4_MAX_PORTS + 1]; }; struct mlx4_en_dev { struct mlx4_dev *dev; struct pci_dev *pdev; struct mutex state_lock; struct net_device *pndev[MLX4_MAX_PORTS + 1]; u32 port_cnt; bool device_up; struct mlx4_en_profile profile; u32 LSO_support; struct workqueue_struct *workqueue; struct device *dma_device; void __iomem *uar_map; struct mlx4_uar priv_uar; struct mlx4_mr mr; u32 priv_pdn; spinlock_t uar_lock; u8 mac_removed[MLX4_MAX_PORTS + 1]; unsigned long last_overflow_check; unsigned long overflow_period; }; struct mlx4_en_rss_map { int base_qpn; struct mlx4_qp qps[MAX_RX_RINGS]; enum mlx4_qp_state state[MAX_RX_RINGS]; struct mlx4_qp indir_qp; enum mlx4_qp_state indir_state; }; struct mlx4_en_port_state { int link_state; int link_speed; int transciver; int autoneg; }; enum mlx4_en_mclist_act { MCLIST_NONE, MCLIST_REM, MCLIST_ADD, }; struct mlx4_en_mc_list { struct list_head list; enum mlx4_en_mclist_act action; u8 addr[ETH_ALEN]; u64 reg_id; }; #ifdef CONFIG_MLX4_EN_DCB /* Minimal TC BW - setting to 0 will block traffic */ #define MLX4_EN_BW_MIN 1 #define MLX4_EN_BW_MAX 100 /* Utilize 100% of the line */ #define MLX4_EN_TC_ETS 7 #endif enum { MLX4_EN_FLAG_PROMISC = (1 << 0), MLX4_EN_FLAG_MC_PROMISC = (1 << 1), /* whether we need to enable hardware loopback by putting dmac * in Tx WQE */ MLX4_EN_FLAG_ENABLE_HW_LOOPBACK = (1 << 2), /* whether we need to drop packets that hardware loopback-ed */ MLX4_EN_FLAG_RX_FILTER_NEEDED = (1 << 3), MLX4_EN_FLAG_FORCE_PROMISC = (1 << 4), #ifdef CONFIG_MLX4_EN_DCB MLX4_EN_FLAG_DCB_ENABLED = (1 << 5) #endif }; #define MLX4_EN_MAC_HASH_SIZE (1 << BITS_PER_BYTE) #define MLX4_EN_MAC_HASH_IDX 5 struct en_port { struct kobject kobj; struct mlx4_dev *dev; u8 port_num; u8 vport_num; }; struct mlx4_en_frag_info { u16 frag_size; u16 frag_prefix_size; }; struct mlx4_en_priv { struct mlx4_en_dev *mdev; struct mlx4_en_port_profile *prof; struct net_device *dev; unsigned long active_vlans[BITS_TO_LONGS(VLAN_N_VID)]; struct mlx4_en_port_state port_state; spinlock_t stats_lock; /* To allow rules removal while port is going down */ struct list_head ethtool_list; unsigned long last_moder_packets[MAX_RX_RINGS]; unsigned long last_moder_tx_packets; unsigned long last_moder_bytes[MAX_RX_RINGS]; unsigned long last_moder_jiffies; int last_moder_time[MAX_RX_RINGS]; u16 rx_usecs; u16 rx_frames; u16 tx_usecs; u16 tx_frames; u32 pkt_rate_low; u32 rx_usecs_low; u32 pkt_rate_high; u32 rx_usecs_high; u32 sample_interval; u32 adaptive_rx_coal; u32 msg_enable; u32 loopback_ok; u32 validate_loopback; struct mlx4_hwq_resources res; int link_state; int last_link_state; bool port_up; int port; int registered; int allocated; int stride; unsigned char current_mac[ETH_ALEN + 2]; u64 mac; int mac_index; unsigned max_mtu; int base_qpn; int cqe_factor; struct mlx4_en_rss_map rss_map; __be32 ctrl_flags; u32 flags; u8 num_tx_rings_p_up; u32 tx_ring_num; u32 rx_ring_num; u32 rx_mb_size; struct mlx4_en_frag_info frag_info[MLX4_EN_MAX_RX_FRAGS]; u16 rx_alloc_order; u32 rx_alloc_size; u32 rx_buf_size; u16 num_frags; u16 log_rx_info; struct mlx4_en_tx_ring **tx_ring; struct mlx4_en_rx_ring *rx_ring[MAX_RX_RINGS]; struct mlx4_en_cq **tx_cq; struct mlx4_en_cq *rx_cq[MAX_RX_RINGS]; struct mlx4_qp drop_qp; struct work_struct rx_mode_task; struct work_struct watchdog_task; struct work_struct linkstate_task; struct delayed_work stats_task; struct delayed_work service_task; struct mlx4_en_perf_stats pstats; struct mlx4_en_pkt_stats pkstats; struct mlx4_en_pkt_stats pkstats_last; struct mlx4_en_flow_stats flowstats[MLX4_NUM_PRIORITIES]; struct mlx4_en_port_stats port_stats; struct mlx4_en_vport_stats vport_stats; struct mlx4_en_vf_stats vf_stats; DECLARE_BITMAP(stats_bitmap, NUM_ALL_STATS); struct list_head mc_list; struct list_head curr_list; u64 broadcast_id; struct mlx4_en_stat_out_mbox hw_stats; int vids[128]; bool wol; struct device *ddev; struct dentry *dev_root; u32 counter_index; eventhandler_tag vlan_attach; eventhandler_tag vlan_detach; struct callout watchdog_timer; struct ifmedia media; volatile int blocked; struct sysctl_oid *sysctl; struct sysctl_ctx_list conf_ctx; struct sysctl_ctx_list stat_ctx; #define MLX4_EN_MAC_HASH_IDX 5 struct hlist_head mac_hash[MLX4_EN_MAC_HASH_SIZE]; #ifdef CONFIG_MLX4_EN_DCB struct ieee_ets ets; u16 maxrate[IEEE_8021QAZ_MAX_TCS]; u8 dcbx_cap; #endif #ifdef CONFIG_RFS_ACCEL spinlock_t filters_lock; int last_filter_id; struct list_head filters; struct hlist_head filter_hash[1 << MLX4_EN_FILTER_HASH_SHIFT]; #endif struct en_port *vf_ports[MLX4_MAX_NUM_VF]; unsigned long last_ifq_jiffies; u64 if_counters_rx_errors; u64 if_counters_rx_no_buffer; }; enum mlx4_en_wol { MLX4_EN_WOL_MAGIC = (1ULL << 61), MLX4_EN_WOL_ENABLED = (1ULL << 62), }; struct mlx4_mac_entry { struct hlist_node hlist; unsigned char mac[ETH_ALEN + 2]; u64 reg_id; }; #ifdef CONFIG_NET_RX_BUSY_POLL static inline void mlx4_en_cq_init_lock(struct mlx4_en_cq *cq) { spin_lock_init(&cq->poll_lock); cq->state = MLX4_EN_CQ_STATEIDLE; } /* called from the device poll rutine to get ownership of a cq */ static inline bool mlx4_en_cq_lock_napi(struct mlx4_en_cq *cq) { int rc = true; spin_lock(&cq->poll_lock); if (cq->state & MLX4_CQ_LOCKED) { WARN_ON(cq->state & MLX4_EN_CQ_STATENAPI); cq->state |= MLX4_EN_CQ_STATENAPI_YIELD; rc = false; } else /* we don't care if someone yielded */ cq->state = MLX4_EN_CQ_STATENAPI; spin_unlock(&cq->poll_lock); return rc; } /* returns true is someone tried to get the cq while napi had it */ static inline bool mlx4_en_cq_unlock_napi(struct mlx4_en_cq *cq) { int rc = false; spin_lock(&cq->poll_lock); WARN_ON(cq->state & (MLX4_EN_CQ_STATEPOLL | MLX4_EN_CQ_STATENAPI_YIELD)); if (cq->state & MLX4_EN_CQ_STATEPOLL_YIELD) rc = true; cq->state = MLX4_EN_CQ_STATEIDLE; spin_unlock(&cq->poll_lock); return rc; } /* called from mlx4_en_low_latency_poll() */ static inline bool mlx4_en_cq_lock_poll(struct mlx4_en_cq *cq) { int rc = true; spin_lock_bh(&cq->poll_lock); if ((cq->state & MLX4_CQ_LOCKED)) { struct net_device *dev = cq->dev; struct mlx4_en_priv *priv = netdev_priv(dev); struct mlx4_en_rx_ring *rx_ring = priv->rx_ring[cq->ring]; cq->state |= MLX4_EN_CQ_STATEPOLL_YIELD; rc = false; #ifdef LL_EXTENDED_STATS rx_ring->yields++; #endif } else /* preserve yield marks */ cq->state |= MLX4_EN_CQ_STATEPOLL; spin_unlock_bh(&cq->poll_lock); return rc; } /* returns true if someone tried to get the cq while it was locked */ static inline bool mlx4_en_cq_unlock_poll(struct mlx4_en_cq *cq) { int rc = false; spin_lock_bh(&cq->poll_lock); WARN_ON(cq->state & (MLX4_EN_CQ_STATENAPI)); if (cq->state & MLX4_EN_CQ_STATEPOLL_YIELD) rc = true; cq->state = MLX4_EN_CQ_STATEIDLE; spin_unlock_bh(&cq->poll_lock); return rc; } /* true if a socket is polling, even if it did not get the lock */ static inline bool mlx4_en_cq_ll_polling(struct mlx4_en_cq *cq) { WARN_ON(!(cq->state & MLX4_CQ_LOCKED)); return cq->state & CQ_USER_PEND; } #else static inline void mlx4_en_cq_init_lock(struct mlx4_en_cq *cq) { } static inline bool mlx4_en_cq_lock_napi(struct mlx4_en_cq *cq) { return true; } static inline bool mlx4_en_cq_unlock_napi(struct mlx4_en_cq *cq) { return false; } static inline bool mlx4_en_cq_lock_poll(struct mlx4_en_cq *cq) { return false; } static inline bool mlx4_en_cq_unlock_poll(struct mlx4_en_cq *cq) { return false; } static inline bool mlx4_en_cq_ll_polling(struct mlx4_en_cq *cq) { return false; } #endif /* CONFIG_NET_RX_BUSY_POLL */ #define MLX4_EN_WOL_DO_MODIFY (1ULL << 63) void mlx4_en_destroy_netdev(struct net_device *dev); int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port, struct mlx4_en_port_profile *prof); int mlx4_en_start_port(struct net_device *dev); void mlx4_en_stop_port(struct net_device *dev); void mlx4_en_free_resources(struct mlx4_en_priv *priv); int mlx4_en_alloc_resources(struct mlx4_en_priv *priv); int mlx4_en_pre_config(struct mlx4_en_priv *priv); int mlx4_en_create_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq **pcq, int entries, int ring, enum cq_type mode, int node); void mlx4_en_destroy_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq **pcq); int mlx4_en_activate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq, int cq_idx); void mlx4_en_deactivate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq); int mlx4_en_set_cq_moder(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq); int mlx4_en_arm_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq); void mlx4_en_tx_irq(struct mlx4_cq *mcq); u16 mlx4_en_select_queue(struct net_device *dev, struct mbuf *mb); int mlx4_en_transmit(struct ifnet *dev, struct mbuf *m); int mlx4_en_create_tx_ring(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring **pring, u32 size, u16 stride, int node, int queue_idx); void mlx4_en_destroy_tx_ring(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring **pring); int mlx4_en_activate_tx_ring(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring *ring, int cq, int user_prio); void mlx4_en_deactivate_tx_ring(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring *ring); void mlx4_en_qflush(struct ifnet *dev); int mlx4_en_create_rx_ring(struct mlx4_en_priv *priv, struct mlx4_en_rx_ring **pring, u32 size, int node); void mlx4_en_destroy_rx_ring(struct mlx4_en_priv *priv, struct mlx4_en_rx_ring **pring, u32 size, u16 stride); void mlx4_en_tx_que(void *context, int pending); void mlx4_en_rx_que(void *context, int pending); int mlx4_en_activate_rx_rings(struct mlx4_en_priv *priv); void mlx4_en_deactivate_rx_ring(struct mlx4_en_priv *priv, struct mlx4_en_rx_ring *ring); int mlx4_en_process_rx_cq(struct net_device *dev, struct mlx4_en_cq *cq, int budget); void mlx4_en_poll_tx_cq(unsigned long data); void mlx4_en_fill_qp_context(struct mlx4_en_priv *priv, int size, int stride, int is_tx, int rss, int qpn, int cqn, int user_prio, struct mlx4_qp_context *context); void mlx4_en_sqp_event(struct mlx4_qp *qp, enum mlx4_event event); int mlx4_en_map_buffer(struct mlx4_buf *buf); void mlx4_en_unmap_buffer(struct mlx4_buf *buf); void mlx4_en_calc_rx_buf(struct net_device *dev); int mlx4_en_config_rss_steer(struct mlx4_en_priv *priv); void mlx4_en_release_rss_steer(struct mlx4_en_priv *priv); int mlx4_en_create_drop_qp(struct mlx4_en_priv *priv); void mlx4_en_destroy_drop_qp(struct mlx4_en_priv *priv); int mlx4_en_free_tx_buf(struct net_device *dev, struct mlx4_en_tx_ring *ring); void mlx4_en_rx_irq(struct mlx4_cq *mcq); int mlx4_SET_MCAST_FLTR(struct mlx4_dev *dev, u8 port, u64 mac, u64 clear, u8 mode); int mlx4_SET_VLAN_FLTR(struct mlx4_dev *dev, struct mlx4_en_priv *priv); int mlx4_en_DUMP_ETH_STATS(struct mlx4_en_dev *mdev, u8 port, u8 reset); int mlx4_en_QUERY_PORT(struct mlx4_en_dev *mdev, u8 port); int mlx4_en_get_vport_stats(struct mlx4_en_dev *mdev, u8 port); void mlx4_en_create_debug_files(struct mlx4_en_priv *priv); void mlx4_en_delete_debug_files(struct mlx4_en_priv *priv); int mlx4_en_register_debugfs(void); void mlx4_en_unregister_debugfs(void); #ifdef CONFIG_MLX4_EN_DCB extern const struct dcbnl_rtnl_ops mlx4_en_dcbnl_ops; extern const struct dcbnl_rtnl_ops mlx4_en_dcbnl_pfc_ops; #endif int mlx4_en_setup_tc(struct net_device *dev, u8 up); #ifdef CONFIG_RFS_ACCEL void mlx4_en_cleanup_filters(struct mlx4_en_priv *priv, struct mlx4_en_rx_ring *rx_ring); #endif #define MLX4_EN_NUM_SELF_TEST 5 void mlx4_en_ex_selftest(struct net_device *dev, u32 *flags, u64 *buf); void mlx4_en_ptp_overflow_check(struct mlx4_en_dev *mdev); /* * Functions for time stamping */ #define SKBTX_HW_TSTAMP (1 << 0) #define SKBTX_IN_PROGRESS (1 << 2) u64 mlx4_en_get_cqe_ts(struct mlx4_cqe *cqe); /* Functions for caching and restoring statistics */ int mlx4_en_get_sset_count(struct net_device *dev, int sset); void mlx4_en_restore_ethtool_stats(struct mlx4_en_priv *priv, u64 *data); /* * Globals */ extern const struct ethtool_ops mlx4_en_ethtool_ops; /* * Defines for link speed - needed by selftest */ #define MLX4_EN_LINK_SPEED_1G 1000 #define MLX4_EN_LINK_SPEED_10G 10000 #define MLX4_EN_LINK_SPEED_40G 40000 enum { NETIF_MSG_DRV = 0x0001, NETIF_MSG_PROBE = 0x0002, NETIF_MSG_LINK = 0x0004, NETIF_MSG_TIMER = 0x0008, NETIF_MSG_IFDOWN = 0x0010, NETIF_MSG_IFUP = 0x0020, NETIF_MSG_RX_ERR = 0x0040, NETIF_MSG_TX_ERR = 0x0080, NETIF_MSG_TX_QUEUED = 0x0100, NETIF_MSG_INTR = 0x0200, NETIF_MSG_TX_DONE = 0x0400, NETIF_MSG_RX_STATUS = 0x0800, NETIF_MSG_PKTDATA = 0x1000, NETIF_MSG_HW = 0x2000, NETIF_MSG_WOL = 0x4000, }; /* * printk / logging functions */ #define en_print(level, priv, format, arg...) \ { \ if ((priv)->registered) \ printk(level "%s: %s: " format, DRV_NAME, \ (priv->dev)->if_xname, ## arg); \ else \ printk(level "%s: %s: Port %d: " format, \ DRV_NAME, dev_name(&priv->mdev->pdev->dev), \ (priv)->port, ## arg); \ } #define en_dbg(mlevel, priv, format, arg...) \ do { \ if (NETIF_MSG_##mlevel & priv->msg_enable) \ en_print(KERN_DEBUG, priv, format, ##arg); \ } while (0) #define en_warn(priv, format, arg...) \ en_print(KERN_WARNING, priv, format, ##arg) #define en_err(priv, format, arg...) \ en_print(KERN_ERR, priv, format, ##arg) #define en_info(priv, format, arg...) \ en_print(KERN_INFO, priv, format, ## arg) #define mlx4_err(mdev, format, arg...) \ pr_err("%s %s: " format, DRV_NAME, \ dev_name(&mdev->pdev->dev), ##arg) #define mlx4_info(mdev, format, arg...) \ pr_info("%s %s: " format, DRV_NAME, \ dev_name(&mdev->pdev->dev), ##arg) #define mlx4_warn(mdev, format, arg...) \ pr_warning("%s %s: " format, DRV_NAME, \ dev_name(&mdev->pdev->dev), ##arg) #endif Index: projects/cxl_iscsi/sys/ofed/include/linux/kernel.h =================================================================== --- projects/cxl_iscsi/sys/ofed/include/linux/kernel.h (revision 279716) +++ projects/cxl_iscsi/sys/ofed/include/linux/kernel.h (revision 279717) @@ -1,185 +1,187 @@ /*- * Copyright (c) 2010 Isilon Systems, Inc. * Copyright (c) 2010 iX Systems, Inc. * Copyright (c) 2010 Panasas, Inc. * Copyright (c) 2013, 2014 Mellanox Technologies, Ltd. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice unmodified, this list of conditions, and the following * disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #ifndef _LINUX_KERNEL_H_ #define _LINUX_KERNEL_H_ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define KERN_CONT "" #define KERN_EMERG "<0>" #define KERN_ALERT "<1>" #define KERN_CRIT "<2>" #define KERN_ERR "<3>" #define KERN_WARNING "<4>" #define KERN_NOTICE "<5>" #define KERN_INFO "<6>" #define KERN_DEBUG "<7>" #define BUILD_BUG_ON(x) CTASSERT(!(x)) #define BUG() panic("BUG") #define BUG_ON(condition) do { if (condition) BUG(); } while(0) #define WARN_ON BUG_ON #undef ALIGN #define ALIGN(x, y) roundup2((x), (y)) +#undef PTR_ALIGN +#define PTR_ALIGN(p, a) ((__typeof(p))ALIGN((uintptr_t)(p), (a))) #define DIV_ROUND_UP howmany #define FIELD_SIZEOF(t, f) sizeof(((t *)0)->f) #define printk(X...) printf(X) /* * The "pr_debug()" and "pr_devel()" macros should produce zero code * unless DEBUG is defined: */ #ifdef DEBUG #define pr_debug(fmt, ...) \ log(LOG_DEBUG, fmt, ##__VA_ARGS__) #define pr_devel(fmt, ...) \ log(LOG_DEBUG, pr_fmt(fmt), ##__VA_ARGS__) #else #define pr_debug(fmt, ...) \ ({ if (0) log(LOG_DEBUG, fmt, ##__VA_ARGS__); 0; }) #define pr_devel(fmt, ...) \ ({ if (0) log(LOG_DEBUG, pr_fmt(fmt), ##__VA_ARGS__); 0; }) #endif #define udelay(t) DELAY(t) #define usleep_range(min,max) DELAY(min) #ifndef pr_fmt #define pr_fmt(fmt) fmt #endif /* * Print a one-time message (analogous to WARN_ONCE() et al): */ #define printk_once(...) do { \ static bool __print_once; \ \ if (!__print_once) { \ __print_once = true; \ printk(__VA_ARGS__); \ } \ } while (0) /* * Log a one-time message (analogous to WARN_ONCE() et al): */ #define log_once(level,...) do { \ static bool __log_once; \ \ if (!__log_once) { \ __log_once = true; \ log(level, __VA_ARGS__); \ } \ } while (0) #define pr_emerg(fmt, ...) \ log(LOG_EMERG, pr_fmt(fmt), ##__VA_ARGS__) #define pr_alert(fmt, ...) \ log(LOG_ALERT, pr_fmt(fmt), ##__VA_ARGS__) #define pr_crit(fmt, ...) \ log(LOG_CRIT, pr_fmt(fmt), ##__VA_ARGS__) #define pr_err(fmt, ...) \ log(LOG_ERR, pr_fmt(fmt), ##__VA_ARGS__) #define pr_warning(fmt, ...) \ log(LOG_WARNING, pr_fmt(fmt), ##__VA_ARGS__) #define pr_warn pr_warning #define pr_notice(fmt, ...) \ log(LOG_NOTICE, pr_fmt(fmt), ##__VA_ARGS__) #define pr_info(fmt, ...) \ log(LOG_INFO, pr_fmt(fmt), ##__VA_ARGS__) #define pr_info_once(fmt, ...) \ log_once(LOG_INFO, pr_fmt(fmt), ##__VA_ARGS__) #define pr_cont(fmt, ...) \ printk(KERN_CONT fmt, ##__VA_ARGS__) #ifndef WARN #define WARN(condition, format...) ({ \ int __ret_warn_on = !!(condition); \ if (unlikely(__ret_warn_on)) \ pr_warning(format); \ unlikely(__ret_warn_on); \ }) #endif #define container_of(ptr, type, member) \ ({ \ __typeof(((type *)0)->member) *_p = (ptr); \ (type *)((char *)_p - offsetof(type, member)); \ }) #define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0])) #define simple_strtoul strtoul #define simple_strtol strtol #define kstrtol(a,b,c) ({*(c) = strtol(a,0,b);}) #define min(x, y) ((x) < (y) ? (x) : (y)) #define max(x, y) ((x) > (y) ? (x) : (y)) #define min_t(type, _x, _y) ((type)(_x) < (type)(_y) ? (type)(_x) : (type)(_y)) #define max_t(type, _x, _y) ((type)(_x) > (type)(_y) ? (type)(_x) : (type)(_y)) /* * This looks more complex than it should be. But we need to * get the type for the ~ right in round_down (it needs to be * as wide as the result!), and we want to evaluate the macro * arguments just once each. */ #define __round_mask(x, y) ((__typeof__(x))((y)-1)) #define round_up(x, y) ((((x)-1) | __round_mask(x, y))+1) #define round_down(x, y) ((x) & ~__round_mask(x, y)) #define num_possible_cpus() mp_ncpus #define num_online_cpus() mp_ncpus typedef struct pm_message { int event; } pm_message_t; #endif /* _LINUX_KERNEL_H_ */ Index: projects/cxl_iscsi/sys/ofed/include/linux/mlx4/device.h =================================================================== --- projects/cxl_iscsi/sys/ofed/include/linux/mlx4/device.h (revision 279716) +++ projects/cxl_iscsi/sys/ofed/include/linux/mlx4/device.h (revision 279717) @@ -1,1286 +1,1297 @@ /* * Copyright (c) 2006, 2007 Cisco Systems, Inc. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU * General Public License (GPL) Version 2, available from the file * COPYING in the main directory of this source tree, or the * OpenIB.org BSD license below: * * Redistribution and use in source and binary forms, with or * without modification, are permitted provided that the following * conditions are met: * * - Redistributions of source code must retain the above * copyright notice, this list of conditions and the following * disclaimer. * * - Redistributions in binary form must reproduce the above * copyright notice, this list of conditions and the following * disclaimer in the documentation and/or other materials * provided with the distribution. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ #ifndef MLX4_DEVICE_H #define MLX4_DEVICE_H #include #include #include #include #include #include #include #include #define MAX_MSIX_P_PORT 17 #define MAX_MSIX 64 #define MSIX_LEGACY_SZ 4 #define MIN_MSIX_P_PORT 5 #define MLX4_ROCE_MAX_GIDS 128 #define MLX4_ROCE_PF_GIDS 16 #define MLX4_NUM_UP 8 #define MLX4_NUM_TC 8 #define MLX4_MAX_100M_UNITS_VAL 255 /* * work around: can't set values * greater then this value when * using 100 Mbps units. */ #define MLX4_RATELIMIT_100M_UNITS 3 /* 100 Mbps */ #define MLX4_RATELIMIT_1G_UNITS 4 /* 1 Gbps */ #define MLX4_RATELIMIT_DEFAULT 0x00ff #define CORE_CLOCK_MASK 0xffffffffffffULL enum { MLX4_FLAG_MSI_X = 1 << 0, MLX4_FLAG_OLD_PORT_CMDS = 1 << 1, MLX4_FLAG_MASTER = 1 << 2, MLX4_FLAG_SLAVE = 1 << 3, MLX4_FLAG_SRIOV = 1 << 4, MLX4_FLAG_DEV_NUM_STR = 1 << 5, MLX4_FLAG_OLD_REG_MAC = 1 << 6, }; enum { MLX4_PORT_CAP_IS_SM = 1 << 1, MLX4_PORT_CAP_DEV_MGMT_SUP = 1 << 19, }; enum { MLX4_MAX_PORTS = 2, MLX4_MAX_PORT_PKEYS = 128 }; /* base qkey for use in sriov tunnel-qp/proxy-qp communication. * These qkeys must not be allowed for general use. This is a 64k range, * and to test for violation, we use the mask (protect against future chg). */ #define MLX4_RESERVED_QKEY_BASE (0xFFFF0000) #define MLX4_RESERVED_QKEY_MASK (0xFFFF0000) enum { MLX4_BOARD_ID_LEN = 64, MLX4_VSD_LEN = 208 }; enum { MLX4_MAX_NUM_PF = 16, MLX4_MAX_NUM_VF = 64, MLX4_MFUNC_MAX = 80, MLX4_MAX_EQ_NUM = 1024, MLX4_MFUNC_EQ_NUM = 4, MLX4_MFUNC_MAX_EQES = 8, MLX4_MFUNC_EQE_MASK = (MLX4_MFUNC_MAX_EQES - 1) }; /* Driver supports 3 diffrent device methods to manage traffic steering: * -device managed - High level API for ib and eth flow steering. FW is * managing flow steering tables. * - B0 steering mode - Common low level API for ib and (if supported) eth. * - A0 steering mode - Limited low level API for eth. In case of IB, * B0 mode is in use. */ enum { MLX4_STEERING_MODE_A0, MLX4_STEERING_MODE_B0, MLX4_STEERING_MODE_DEVICE_MANAGED }; static inline const char *mlx4_steering_mode_str(int steering_mode) { switch (steering_mode) { case MLX4_STEERING_MODE_A0: return "A0 steering"; case MLX4_STEERING_MODE_B0: return "B0 steering"; case MLX4_STEERING_MODE_DEVICE_MANAGED: return "Device managed flow steering"; default: return "Unrecognize steering mode"; } } enum { MLX4_DEV_CAP_FLAG_RC = 1LL << 0, MLX4_DEV_CAP_FLAG_UC = 1LL << 1, MLX4_DEV_CAP_FLAG_UD = 1LL << 2, MLX4_DEV_CAP_FLAG_XRC = 1LL << 3, MLX4_DEV_CAP_FLAG_SRQ = 1LL << 6, MLX4_DEV_CAP_FLAG_IPOIB_CSUM = 1LL << 7, MLX4_DEV_CAP_FLAG_BAD_PKEY_CNTR = 1LL << 8, MLX4_DEV_CAP_FLAG_BAD_QKEY_CNTR = 1LL << 9, MLX4_DEV_CAP_FLAG_DPDP = 1LL << 12, MLX4_DEV_CAP_FLAG_BLH = 1LL << 15, MLX4_DEV_CAP_FLAG_MEM_WINDOW = 1LL << 16, MLX4_DEV_CAP_FLAG_APM = 1LL << 17, MLX4_DEV_CAP_FLAG_ATOMIC = 1LL << 18, MLX4_DEV_CAP_FLAG_RAW_MCAST = 1LL << 19, MLX4_DEV_CAP_FLAG_UD_AV_PORT = 1LL << 20, MLX4_DEV_CAP_FLAG_UD_MCAST = 1LL << 21, MLX4_DEV_CAP_FLAG_IBOE = 1LL << 30, MLX4_DEV_CAP_FLAG_UC_LOOPBACK = 1LL << 32, MLX4_DEV_CAP_FLAG_FCS_KEEP = 1LL << 34, MLX4_DEV_CAP_FLAG_WOL_PORT1 = 1LL << 37, MLX4_DEV_CAP_FLAG_WOL_PORT2 = 1LL << 38, MLX4_DEV_CAP_FLAG_UDP_RSS = 1LL << 40, MLX4_DEV_CAP_FLAG_VEP_UC_STEER = 1LL << 41, MLX4_DEV_CAP_FLAG_VEP_MC_STEER = 1LL << 42, MLX4_DEV_CAP_FLAG_CROSS_CHANNEL = 1LL << 44, MLX4_DEV_CAP_FLAG_COUNTERS = 1LL << 48, MLX4_DEV_CAP_FLAG_COUNTERS_EXT = 1LL << 49, MLX4_DEV_CAP_FLAG_SET_PORT_ETH_SCHED = 1LL << 53, MLX4_DEV_CAP_FLAG_SENSE_SUPPORT = 1LL << 55, MLX4_DEV_CAP_FLAG_FAST_DROP = 1LL << 57, MLX4_DEV_CAP_FLAG_PORT_MNG_CHG_EV = 1LL << 59, MLX4_DEV_CAP_FLAG_64B_EQE = 1LL << 61, MLX4_DEV_CAP_FLAG_64B_CQE = 1LL << 62 }; enum { MLX4_DEV_CAP_FLAG2_RSS = 1LL << 0, MLX4_DEV_CAP_FLAG2_RSS_TOP = 1LL << 1, MLX4_DEV_CAP_FLAG2_RSS_XOR = 1LL << 2, MLX4_DEV_CAP_FLAG2_FS_EN = 1LL << 3, MLX4_DEV_CAP_FLAG2_FSM = 1LL << 4, MLX4_DEV_CAP_FLAG2_VLAN_CONTROL = 1LL << 5, MLX4_DEV_CAP_FLAG2_UPDATE_QP = 1LL << 6, MLX4_DEV_CAP_FLAG2_LB_SRC_CHK = 1LL << 7, MLX4_DEV_CAP_FLAG2_DMFS_IPOIB = 1LL << 8, MLX4_DEV_CAP_FLAG2_ETS_CFG = 1LL << 9, MLX4_DEV_CAP_FLAG2_ETH_BACKPL_AN_REP = 1LL << 10, MLX4_DEV_CAP_FLAG2_FLOWSTATS_EN = 1LL << 11, MLX4_DEV_CAP_FLAG2_RECOVERABLE_ERROR_EVENT = 1LL << 12, - MLX4_DEV_CAP_FLAG2_TS = 1LL << 13, - MLX4_DEV_CAP_FLAG2_DRIVER_VERSION_TO_FW = 1LL << 14 + MLX4_DEV_CAP_FLAG2_TS = 1LL << 13, + MLX4_DEV_CAP_FLAG2_DRIVER_VERSION_TO_FW = 1LL << 14, + MLX4_DEV_CAP_FLAG2_REASSIGN_MAC_EN = 1LL << 15, + MLX4_DEV_CAP_FLAG2_VXLAN_OFFLOADS = 1LL << 16, + MLX4_DEV_CAP_FLAG2_FS_EN_NCSI = 1LL << 17, + MLX4_DEV_CAP_FLAG2_80_VFS = 1LL << 18, + MLX4_DEV_CAP_FLAG2_DMFS_TAG_MODE = 1LL << 19, + MLX4_DEV_CAP_FLAG2_ROCEV2 = 1LL << 20, + MLX4_DEV_CAP_FLAG2_ETH_PROT_CTRL = 1LL << 21, + MLX4_DEV_CAP_FLAG2_CQE_STRIDE = 1LL << 22, + MLX4_DEV_CAP_FLAG2_EQE_STRIDE = 1LL << 23, + MLX4_DEV_CAP_FLAG2_UPDATE_QP_SRC_CHECK_LB = 1LL << 24, + MLX4_DEV_CAP_FLAG2_RX_CSUM_MODE = 1LL << 25, }; /* bit enums for an 8-bit flags field indicating special use * QPs which require special handling in qp_reserve_range. * Currently, this only includes QPs used by the ETH interface, * where we expect to use blueflame. These QPs must not have * bits 6 and 7 set in their qp number. * * This enum may use only bits 0..7. */ enum { MLX4_RESERVE_BF_QP = 1 << 7, }; enum { MLX4_DEV_CAP_CQ_FLAG_IO = 1 << 0 }; enum { MLX4_DEV_CAP_64B_EQE_ENABLED = 1LL << 0, MLX4_DEV_CAP_64B_CQE_ENABLED = 1LL << 1 }; enum { MLX4_USER_DEV_CAP_64B_CQE = 1L << 0 }; enum { MLX4_FUNC_CAP_64B_EQE_CQE = 1L << 0 }; #define MLX4_ATTR_EXTENDED_PORT_INFO cpu_to_be16(0xff90) enum { MLX4_BMME_FLAG_WIN_TYPE_2B = 1 << 1, MLX4_BMME_FLAG_LOCAL_INV = 1 << 6, MLX4_BMME_FLAG_REMOTE_INV = 1 << 7, MLX4_BMME_FLAG_TYPE_2_WIN = 1 << 9, MLX4_BMME_FLAG_RESERVED_LKEY = 1 << 10, MLX4_BMME_FLAG_FAST_REG_WR = 1 << 11, }; enum mlx4_event { MLX4_EVENT_TYPE_COMP = 0x00, MLX4_EVENT_TYPE_PATH_MIG = 0x01, MLX4_EVENT_TYPE_COMM_EST = 0x02, MLX4_EVENT_TYPE_SQ_DRAINED = 0x03, MLX4_EVENT_TYPE_SRQ_QP_LAST_WQE = 0x13, MLX4_EVENT_TYPE_SRQ_LIMIT = 0x14, MLX4_EVENT_TYPE_CQ_ERROR = 0x04, MLX4_EVENT_TYPE_WQ_CATAS_ERROR = 0x05, MLX4_EVENT_TYPE_EEC_CATAS_ERROR = 0x06, MLX4_EVENT_TYPE_PATH_MIG_FAILED = 0x07, MLX4_EVENT_TYPE_WQ_INVAL_REQ_ERROR = 0x10, MLX4_EVENT_TYPE_WQ_ACCESS_ERROR = 0x11, MLX4_EVENT_TYPE_SRQ_CATAS_ERROR = 0x12, MLX4_EVENT_TYPE_LOCAL_CATAS_ERROR = 0x08, MLX4_EVENT_TYPE_PORT_CHANGE = 0x09, MLX4_EVENT_TYPE_EQ_OVERFLOW = 0x0f, MLX4_EVENT_TYPE_ECC_DETECT = 0x0e, MLX4_EVENT_TYPE_CMD = 0x0a, MLX4_EVENT_TYPE_VEP_UPDATE = 0x19, MLX4_EVENT_TYPE_COMM_CHANNEL = 0x18, MLX4_EVENT_TYPE_OP_REQUIRED = 0x1a, MLX4_EVENT_TYPE_FATAL_WARNING = 0x1b, MLX4_EVENT_TYPE_FLR_EVENT = 0x1c, MLX4_EVENT_TYPE_PORT_MNG_CHG_EVENT = 0x1d, MLX4_EVENT_TYPE_RECOVERABLE_ERROR_EVENT = 0x3e, MLX4_EVENT_TYPE_NONE = 0xff, }; enum { MLX4_PORT_CHANGE_SUBTYPE_DOWN = 1, MLX4_PORT_CHANGE_SUBTYPE_ACTIVE = 4 }; enum { MLX4_RECOVERABLE_ERROR_EVENT_SUBTYPE_BAD_CABLE = 1, MLX4_RECOVERABLE_ERROR_EVENT_SUBTYPE_UNSUPPORTED_CABLE = 2, }; enum { MLX4_FATAL_WARNING_SUBTYPE_WARMING = 0, }; enum slave_port_state { SLAVE_PORT_DOWN = 0, SLAVE_PENDING_UP, SLAVE_PORT_UP, }; enum slave_port_gen_event { SLAVE_PORT_GEN_EVENT_DOWN = 0, SLAVE_PORT_GEN_EVENT_UP, SLAVE_PORT_GEN_EVENT_NONE, }; enum slave_port_state_event { MLX4_PORT_STATE_DEV_EVENT_PORT_DOWN, MLX4_PORT_STATE_DEV_EVENT_PORT_UP, MLX4_PORT_STATE_IB_PORT_STATE_EVENT_GID_VALID, MLX4_PORT_STATE_IB_EVENT_GID_INVALID, }; enum { MLX4_PERM_LOCAL_READ = 1 << 10, MLX4_PERM_LOCAL_WRITE = 1 << 11, MLX4_PERM_REMOTE_READ = 1 << 12, MLX4_PERM_REMOTE_WRITE = 1 << 13, MLX4_PERM_ATOMIC = 1 << 14, MLX4_PERM_BIND_MW = 1 << 15, }; enum { MLX4_OPCODE_NOP = 0x00, MLX4_OPCODE_SEND_INVAL = 0x01, MLX4_OPCODE_RDMA_WRITE = 0x08, MLX4_OPCODE_RDMA_WRITE_IMM = 0x09, MLX4_OPCODE_SEND = 0x0a, MLX4_OPCODE_SEND_IMM = 0x0b, MLX4_OPCODE_LSO = 0x0e, MLX4_OPCODE_RDMA_READ = 0x10, MLX4_OPCODE_ATOMIC_CS = 0x11, MLX4_OPCODE_ATOMIC_FA = 0x12, MLX4_OPCODE_MASKED_ATOMIC_CS = 0x14, MLX4_OPCODE_MASKED_ATOMIC_FA = 0x15, MLX4_OPCODE_BIND_MW = 0x18, MLX4_OPCODE_FMR = 0x19, MLX4_OPCODE_LOCAL_INVAL = 0x1b, MLX4_OPCODE_CONFIG_CMD = 0x1f, MLX4_RECV_OPCODE_RDMA_WRITE_IMM = 0x00, MLX4_RECV_OPCODE_SEND = 0x01, MLX4_RECV_OPCODE_SEND_IMM = 0x02, MLX4_RECV_OPCODE_SEND_INVAL = 0x03, MLX4_CQE_OPCODE_ERROR = 0x1e, MLX4_CQE_OPCODE_RESIZE = 0x16, }; enum { MLX4_STAT_RATE_OFFSET = 5 }; enum mlx4_protocol { MLX4_PROT_IB_IPV6 = 0, MLX4_PROT_ETH, MLX4_PROT_IB_IPV4, MLX4_PROT_FCOE }; enum { MLX4_MTT_FLAG_PRESENT = 1 }; enum { MLX4_MAX_MTT_SHIFT = 31 }; enum mlx4_qp_region { MLX4_QP_REGION_FW = 0, MLX4_QP_REGION_ETH_ADDR, MLX4_QP_REGION_FC_ADDR, MLX4_QP_REGION_FC_EXCH, MLX4_NUM_QP_REGION }; enum mlx4_port_type { MLX4_PORT_TYPE_NONE = 0, MLX4_PORT_TYPE_IB = 1, MLX4_PORT_TYPE_ETH = 2, MLX4_PORT_TYPE_AUTO = 3, MLX4_PORT_TYPE_NA = 4 }; enum mlx4_special_vlan_idx { MLX4_NO_VLAN_IDX = 0, MLX4_VLAN_MISS_IDX, MLX4_VLAN_REGULAR }; enum mlx4_steer_type { MLX4_MC_STEER = 0, MLX4_UC_STEER, MLX4_NUM_STEERS }; enum { MLX4_NUM_FEXCH = 64 * 1024, }; enum { MLX4_MAX_FAST_REG_PAGES = 511, }; enum { MLX4_DEV_PMC_SUBTYPE_GUID_INFO = 0x14, MLX4_DEV_PMC_SUBTYPE_PORT_INFO = 0x15, MLX4_DEV_PMC_SUBTYPE_PKEY_TABLE = 0x16, }; /* Port mgmt change event handling */ enum { MLX4_EQ_PORT_INFO_MSTR_SM_LID_CHANGE_MASK = 1 << 0, MLX4_EQ_PORT_INFO_GID_PFX_CHANGE_MASK = 1 << 1, MLX4_EQ_PORT_INFO_LID_CHANGE_MASK = 1 << 2, MLX4_EQ_PORT_INFO_CLIENT_REREG_MASK = 1 << 3, MLX4_EQ_PORT_INFO_MSTR_SM_SL_CHANGE_MASK = 1 << 4, }; #define MSTR_SM_CHANGE_MASK (MLX4_EQ_PORT_INFO_MSTR_SM_SL_CHANGE_MASK | \ MLX4_EQ_PORT_INFO_MSTR_SM_LID_CHANGE_MASK) static inline u64 mlx4_fw_ver(u64 major, u64 minor, u64 subminor) { return (major << 32) | (minor << 16) | subminor; } struct mlx4_phys_caps { u32 gid_phys_table_len[MLX4_MAX_PORTS + 1]; u32 pkey_phys_table_len[MLX4_MAX_PORTS + 1]; u32 num_phys_eqs; u32 base_sqpn; u32 base_proxy_sqpn; u32 base_tunnel_sqpn; }; struct mlx4_caps { u64 fw_ver; u32 function; int num_ports; int vl_cap[MLX4_MAX_PORTS + 1]; int ib_mtu_cap[MLX4_MAX_PORTS + 1]; __be32 ib_port_def_cap[MLX4_MAX_PORTS + 1]; u64 def_mac[MLX4_MAX_PORTS + 1]; int eth_mtu_cap[MLX4_MAX_PORTS + 1]; int gid_table_len[MLX4_MAX_PORTS + 1]; int pkey_table_len[MLX4_MAX_PORTS + 1]; int trans_type[MLX4_MAX_PORTS + 1]; int vendor_oui[MLX4_MAX_PORTS + 1]; int wavelength[MLX4_MAX_PORTS + 1]; u64 trans_code[MLX4_MAX_PORTS + 1]; int local_ca_ack_delay; int num_uars; u32 uar_page_size; int bf_reg_size; int bf_regs_per_page; int max_sq_sg; int max_rq_sg; int num_qps; int max_wqes; int max_sq_desc_sz; int max_rq_desc_sz; int max_qp_init_rdma; int max_qp_dest_rdma; u32 *qp0_proxy; u32 *qp1_proxy; u32 *qp0_tunnel; u32 *qp1_tunnel; int num_srqs; int max_srq_wqes; int max_srq_sge; int reserved_srqs; int num_cqs; int max_cqes; int reserved_cqs; int num_eqs; int reserved_eqs; int num_comp_vectors; int comp_pool; int num_mpts; int max_fmr_maps; u64 num_mtts; int fmr_reserved_mtts; int reserved_mtts; int reserved_mrws; int reserved_uars; int num_mgms; int num_amgms; int reserved_mcgs; int num_qp_per_mgm; int steering_mode; int num_pds; int reserved_pds; int max_xrcds; int reserved_xrcds; int mtt_entry_sz; u32 max_msg_sz; u32 page_size_cap; u64 flags; u64 flags2; u32 bmme_flags; u32 reserved_lkey; u16 stat_rate_support; u8 cq_timestamp; u8 port_width_cap[MLX4_MAX_PORTS + 1]; int max_gso_sz; int max_rss_tbl_sz; int reserved_qps_cnt[MLX4_NUM_QP_REGION]; int reserved_qps; int reserved_qps_base[MLX4_NUM_QP_REGION]; int log_num_macs; int log_num_vlans; enum mlx4_port_type port_type[MLX4_MAX_PORTS + 1]; u8 supported_type[MLX4_MAX_PORTS + 1]; u8 suggested_type[MLX4_MAX_PORTS + 1]; u8 default_sense[MLX4_MAX_PORTS + 1]; u32 port_mask[MLX4_MAX_PORTS + 1]; enum mlx4_port_type possible_type[MLX4_MAX_PORTS + 1]; u32 max_counters; u8 port_ib_mtu[MLX4_MAX_PORTS + 1]; u16 sqp_demux; u32 sync_qp; u32 cq_flags; u32 eqe_size; u32 cqe_size; u8 eqe_factor; u32 userspace_caps; /* userspace must be aware to */ u32 function_caps; /* functions must be aware to */ u8 fast_drop; u16 hca_core_clock; u32 max_basic_counters; u32 max_extended_counters; u8 def_counter_index[MLX4_MAX_PORTS + 1]; }; struct mlx4_buf_list { void *buf; dma_addr_t map; }; struct mlx4_buf { struct mlx4_buf_list direct; struct mlx4_buf_list *page_list; int nbufs; int npages; int page_shift; }; struct mlx4_mtt { u32 offset; int order; int page_shift; }; enum { MLX4_DB_PER_PAGE = PAGE_SIZE / 4 }; struct mlx4_db_pgdir { struct list_head list; DECLARE_BITMAP(order0, MLX4_DB_PER_PAGE); DECLARE_BITMAP(order1, MLX4_DB_PER_PAGE / 2); unsigned long *bits[2]; __be32 *db_page; dma_addr_t db_dma; }; struct mlx4_ib_user_db_page; struct mlx4_db { __be32 *db; union { struct mlx4_db_pgdir *pgdir; struct mlx4_ib_user_db_page *user_page; } u; dma_addr_t dma; int index; int order; }; struct mlx4_hwq_resources { struct mlx4_db db; struct mlx4_mtt mtt; struct mlx4_buf buf; }; struct mlx4_mr { struct mlx4_mtt mtt; u64 iova; u64 size; u32 key; u32 pd; u32 access; int enabled; }; enum mlx4_mw_type { MLX4_MW_TYPE_1 = 1, MLX4_MW_TYPE_2 = 2, }; struct mlx4_mw { u32 key; u32 pd; enum mlx4_mw_type type; int enabled; }; struct mlx4_fmr { struct mlx4_mr mr; struct mlx4_mpt_entry *mpt; __be64 *mtts; dma_addr_t dma_handle; int max_pages; int max_maps; int maps; u8 page_shift; }; struct mlx4_uar { unsigned long pfn; int index; struct list_head bf_list; unsigned free_bf_bmap; void __iomem *map; void __iomem *bf_map; }; struct mlx4_bf { unsigned long offset; int buf_size; struct mlx4_uar *uar; void __iomem *reg; }; struct mlx4_cq { void (*comp) (struct mlx4_cq *); void (*event) (struct mlx4_cq *, enum mlx4_event); struct mlx4_uar *uar; u32 cons_index; __be32 *set_ci_db; __be32 *arm_db; int arm_sn; int cqn; unsigned vector; atomic_t refcount; struct completion free; int eqn; u16 irq; }; struct mlx4_qp { void (*event) (struct mlx4_qp *, enum mlx4_event); int qpn; atomic_t refcount; struct completion free; }; struct mlx4_srq { void (*event) (struct mlx4_srq *, enum mlx4_event); int srqn; int max; int max_gs; int wqe_shift; atomic_t refcount; struct completion free; }; struct mlx4_av { __be32 port_pd; u8 reserved1; u8 g_slid; __be16 dlid; u8 reserved2; u8 gid_index; u8 stat_rate; u8 hop_limit; __be32 sl_tclass_flowlabel; u8 dgid[16]; }; struct mlx4_eth_av { __be32 port_pd; u8 reserved1; u8 smac_idx; u16 reserved2; u8 reserved3; u8 gid_index; u8 stat_rate; u8 hop_limit; __be32 sl_tclass_flowlabel; u8 dgid[16]; u8 s_mac[6]; u8 reserved4[2]; __be16 vlan; u8 mac[6]; }; union mlx4_ext_av { struct mlx4_av ib; struct mlx4_eth_av eth; }; struct mlx4_if_stat_control { u8 reserved1[3]; /* Extended counters enabled */ u8 cnt_mode; /* Number of interfaces */ __be32 num_of_if; __be32 reserved[2]; }; struct mlx4_if_stat_basic { struct mlx4_if_stat_control control; struct { __be64 IfRxFrames; __be64 IfRxOctets; __be64 IfTxFrames; __be64 IfTxOctets; } counters[]; }; #define MLX4_IF_STAT_BSC_SZ(ports)(sizeof(struct mlx4_if_stat_extended) +\ sizeof(((struct mlx4_if_stat_extended *)0)->\ counters[0]) * ports) struct mlx4_if_stat_extended { struct mlx4_if_stat_control control; struct { __be64 IfRxUnicastFrames; __be64 IfRxUnicastOctets; __be64 IfRxMulticastFrames; __be64 IfRxMulticastOctets; __be64 IfRxBroadcastFrames; __be64 IfRxBroadcastOctets; __be64 IfRxNoBufferFrames; __be64 IfRxNoBufferOctets; __be64 IfRxErrorFrames; __be64 IfRxErrorOctets; __be32 reserved[39]; __be64 IfTxUnicastFrames; __be64 IfTxUnicastOctets; __be64 IfTxMulticastFrames; __be64 IfTxMulticastOctets; __be64 IfTxBroadcastFrames; __be64 IfTxBroadcastOctets; __be64 IfTxDroppedFrames; __be64 IfTxDroppedOctets; __be64 IfTxRequestedFramesSent; __be64 IfTxGeneratedFramesSent; __be64 IfTxTsoOctets; } __packed counters[]; }; #define MLX4_IF_STAT_EXT_SZ(ports) (sizeof(struct mlx4_if_stat_extended) +\ sizeof(((struct mlx4_if_stat_extended *)\ 0)->counters[0]) * ports) union mlx4_counter { struct mlx4_if_stat_control control; struct mlx4_if_stat_basic basic; struct mlx4_if_stat_extended ext; }; #define MLX4_IF_STAT_SZ(ports) MLX4_IF_STAT_EXT_SZ(ports) struct mlx4_quotas { int qp; int cq; int srq; int mpt; int mtt; int counter; int xrcd; }; struct mlx4_dev { struct pci_dev *pdev; unsigned long flags; unsigned long num_slaves; struct mlx4_caps caps; struct mlx4_phys_caps phys_caps; struct mlx4_quotas quotas; struct radix_tree_root qp_table_tree; u8 rev_id; char board_id[MLX4_BOARD_ID_LEN]; u16 vsd_vendor_id; char vsd[MLX4_VSD_LEN]; int num_vfs; int numa_node; int oper_log_mgm_entry_size; u64 regid_promisc_array[MLX4_MAX_PORTS + 1]; u64 regid_allmulti_array[MLX4_MAX_PORTS + 1]; }; struct mlx4_clock_params { u64 offset; u8 bar; u8 size; }; struct mlx4_eqe { u8 reserved1; u8 type; u8 reserved2; u8 subtype; union { u32 raw[6]; struct { __be32 cqn; } __packed comp; struct { u16 reserved1; __be16 token; u32 reserved2; u8 reserved3[3]; u8 status; __be64 out_param; } __packed cmd; struct { __be32 qpn; } __packed qp; struct { __be32 srqn; } __packed srq; struct { __be32 cqn; u32 reserved1; u8 reserved2[3]; u8 syndrome; } __packed cq_err; struct { u32 reserved1[2]; __be32 port; } __packed port_change; struct { #define COMM_CHANNEL_BIT_ARRAY_SIZE 4 u32 reserved; u32 bit_vec[COMM_CHANNEL_BIT_ARRAY_SIZE]; } __packed comm_channel_arm; struct { u8 port; u8 reserved[3]; __be64 mac; } __packed mac_update; struct { __be32 slave_id; } __packed flr_event; struct { __be16 current_temperature; __be16 warning_threshold; } __packed warming; struct { u8 reserved[3]; u8 port; union { struct { __be16 mstr_sm_lid; __be16 port_lid; __be32 changed_attr; u8 reserved[3]; u8 mstr_sm_sl; __be64 gid_prefix; } __packed port_info; struct { __be32 block_ptr; __be32 tbl_entries_mask; } __packed tbl_change_info; } params; } __packed port_mgmt_change; struct { u8 reserved[3]; u8 port; u32 reserved1[5]; } __packed bad_cable; } event; u8 slave_id; u8 reserved3[2]; u8 owner; } __packed; struct mlx4_init_port_param { int set_guid0; int set_node_guid; int set_si_guid; u16 mtu; int port_width_cap; u16 vl_cap; u16 max_gid; u16 max_pkey; u64 guid0; u64 node_guid; u64 si_guid; }; #define mlx4_foreach_port(port, dev, type) \ for ((port) = 1; (port) <= (dev)->caps.num_ports; (port)++) \ if ((type) == (dev)->caps.port_mask[(port)]) #define mlx4_foreach_non_ib_transport_port(port, dev) \ for ((port) = 1; (port) <= (dev)->caps.num_ports; (port)++) \ if (((dev)->caps.port_mask[port] != MLX4_PORT_TYPE_IB)) #define mlx4_foreach_ib_transport_port(port, dev) \ for ((port) = 1; (port) <= (dev)->caps.num_ports; (port)++) \ if (((dev)->caps.port_mask[port] == MLX4_PORT_TYPE_IB) || \ ((dev)->caps.flags & MLX4_DEV_CAP_FLAG_IBOE)) #define MLX4_INVALID_SLAVE_ID 0xFF #define MLX4_SINK_COUNTER_INDEX 0xff void handle_port_mgmt_change_event(struct work_struct *work); static inline int mlx4_master_func_num(struct mlx4_dev *dev) { return dev->caps.function; } static inline int mlx4_is_master(struct mlx4_dev *dev) { return dev->flags & MLX4_FLAG_MASTER; } static inline int mlx4_num_reserved_sqps(struct mlx4_dev *dev) { return dev->phys_caps.base_sqpn + 8 + 16 * MLX4_MFUNC_MAX * !!mlx4_is_master(dev); } static inline int mlx4_is_qp_reserved(struct mlx4_dev *dev, u32 qpn) { return (qpn < dev->phys_caps.base_sqpn + 8 + 16 * MLX4_MFUNC_MAX * !!mlx4_is_master(dev)); } static inline int mlx4_is_guest_proxy(struct mlx4_dev *dev, int slave, u32 qpn) { int guest_proxy_base = dev->phys_caps.base_proxy_sqpn + slave * 8; if (qpn >= guest_proxy_base && qpn < guest_proxy_base + 8) return 1; return 0; } static inline int mlx4_is_mfunc(struct mlx4_dev *dev) { return dev->flags & (MLX4_FLAG_SLAVE | MLX4_FLAG_MASTER); } static inline int mlx4_is_slave(struct mlx4_dev *dev) { return dev->flags & MLX4_FLAG_SLAVE; } int mlx4_buf_alloc(struct mlx4_dev *dev, int size, int max_direct, struct mlx4_buf *buf); void mlx4_buf_free(struct mlx4_dev *dev, int size, struct mlx4_buf *buf); static inline void *mlx4_buf_offset(struct mlx4_buf *buf, int offset) { if (BITS_PER_LONG == 64 || buf->nbufs == 1) - return buf->direct.buf + offset; + return (u8 *)buf->direct.buf + offset; else - return buf->page_list[offset >> PAGE_SHIFT].buf + + return (u8 *)buf->page_list[offset >> PAGE_SHIFT].buf + (offset & (PAGE_SIZE - 1)); } int mlx4_pd_alloc(struct mlx4_dev *dev, u32 *pdn); void mlx4_pd_free(struct mlx4_dev *dev, u32 pdn); int mlx4_xrcd_alloc(struct mlx4_dev *dev, u32 *xrcdn); void mlx4_xrcd_free(struct mlx4_dev *dev, u32 xrcdn); int mlx4_uar_alloc(struct mlx4_dev *dev, struct mlx4_uar *uar); void mlx4_uar_free(struct mlx4_dev *dev, struct mlx4_uar *uar); int mlx4_bf_alloc(struct mlx4_dev *dev, struct mlx4_bf *bf, int node); void mlx4_bf_free(struct mlx4_dev *dev, struct mlx4_bf *bf); int mlx4_mtt_init(struct mlx4_dev *dev, int npages, int page_shift, struct mlx4_mtt *mtt); void mlx4_mtt_cleanup(struct mlx4_dev *dev, struct mlx4_mtt *mtt); u64 mlx4_mtt_addr(struct mlx4_dev *dev, struct mlx4_mtt *mtt); int mlx4_mr_alloc(struct mlx4_dev *dev, u32 pd, u64 iova, u64 size, u32 access, int npages, int page_shift, struct mlx4_mr *mr); int mlx4_mr_free(struct mlx4_dev *dev, struct mlx4_mr *mr); int mlx4_mr_enable(struct mlx4_dev *dev, struct mlx4_mr *mr); int mlx4_mw_alloc(struct mlx4_dev *dev, u32 pd, enum mlx4_mw_type type, struct mlx4_mw *mw); void mlx4_mw_free(struct mlx4_dev *dev, struct mlx4_mw *mw); int mlx4_mw_enable(struct mlx4_dev *dev, struct mlx4_mw *mw); int mlx4_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt, int start_index, int npages, u64 *page_list); int mlx4_buf_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt, struct mlx4_buf *buf); int mlx4_db_alloc(struct mlx4_dev *dev, struct mlx4_db *db, int order); void mlx4_db_free(struct mlx4_dev *dev, struct mlx4_db *db); int mlx4_alloc_hwq_res(struct mlx4_dev *dev, struct mlx4_hwq_resources *wqres, int size, int max_direct); void mlx4_free_hwq_res(struct mlx4_dev *mdev, struct mlx4_hwq_resources *wqres, int size); int mlx4_cq_alloc(struct mlx4_dev *dev, int nent, struct mlx4_mtt *mtt, struct mlx4_uar *uar, u64 db_rec, struct mlx4_cq *cq, unsigned vector, int collapsed, int timestamp_en); void mlx4_cq_free(struct mlx4_dev *dev, struct mlx4_cq *cq); int mlx4_qp_reserve_range(struct mlx4_dev *dev, int cnt, int align, int *base, u8 flags); void mlx4_qp_release_range(struct mlx4_dev *dev, int base_qpn, int cnt); int mlx4_qp_alloc(struct mlx4_dev *dev, int qpn, struct mlx4_qp *qp); void mlx4_qp_free(struct mlx4_dev *dev, struct mlx4_qp *qp); int mlx4_srq_alloc(struct mlx4_dev *dev, u32 pdn, u32 cqn, u16 xrcdn, struct mlx4_mtt *mtt, u64 db_rec, struct mlx4_srq *srq); void mlx4_srq_free(struct mlx4_dev *dev, struct mlx4_srq *srq); int mlx4_srq_arm(struct mlx4_dev *dev, struct mlx4_srq *srq, int limit_watermark); int mlx4_srq_query(struct mlx4_dev *dev, struct mlx4_srq *srq, int *limit_watermark); int mlx4_INIT_PORT(struct mlx4_dev *dev, int port); int mlx4_CLOSE_PORT(struct mlx4_dev *dev, int port); int mlx4_unicast_attach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16], int block_mcast_loopback, enum mlx4_protocol prot); int mlx4_unicast_detach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16], enum mlx4_protocol prot); int mlx4_multicast_attach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16], u8 port, int block_mcast_loopback, enum mlx4_protocol protocol, u64 *reg_id); int mlx4_multicast_detach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16], enum mlx4_protocol protocol, u64 reg_id); enum { MLX4_DOMAIN_UVERBS = 0x1000, MLX4_DOMAIN_ETHTOOL = 0x2000, MLX4_DOMAIN_RFS = 0x3000, MLX4_DOMAIN_NIC = 0x5000, }; enum mlx4_net_trans_rule_id { MLX4_NET_TRANS_RULE_ID_ETH = 0, MLX4_NET_TRANS_RULE_ID_IB, MLX4_NET_TRANS_RULE_ID_IPV6, MLX4_NET_TRANS_RULE_ID_IPV4, MLX4_NET_TRANS_RULE_ID_TCP, MLX4_NET_TRANS_RULE_ID_UDP, MLX4_NET_TRANS_RULE_NUM, /* should be last */ MLX4_NET_TRANS_RULE_DUMMY = -1, /* force enum to be signed */ }; extern const u16 __sw_id_hw[]; static inline int map_hw_to_sw_id(u16 header_id) { int i; for (i = 0; i < MLX4_NET_TRANS_RULE_NUM; i++) { if (header_id == __sw_id_hw[i]) return i; } return -EINVAL; } enum mlx4_net_trans_promisc_mode { MLX4_FS_REGULAR = 1, MLX4_FS_ALL_DEFAULT, MLX4_FS_MC_DEFAULT, MLX4_FS_UC_SNIFFER, MLX4_FS_MC_SNIFFER, MLX4_FS_MODE_NUM, /* should be last */ MLX4_FS_MODE_DUMMY = -1, /* force enum to be signed */ }; struct mlx4_spec_eth { u8 dst_mac[6]; u8 dst_mac_msk[6]; u8 src_mac[6]; u8 src_mac_msk[6]; u8 ether_type_enable; __be16 ether_type; __be16 vlan_id_msk; __be16 vlan_id; }; struct mlx4_spec_tcp_udp { __be16 dst_port; __be16 dst_port_msk; __be16 src_port; __be16 src_port_msk; }; struct mlx4_spec_ipv4 { __be32 dst_ip; __be32 dst_ip_msk; __be32 src_ip; __be32 src_ip_msk; }; struct mlx4_spec_ib { __be32 l3_qpn; __be32 qpn_msk; u8 dst_gid[16]; u8 dst_gid_msk[16]; }; struct mlx4_spec_list { struct list_head list; enum mlx4_net_trans_rule_id id; union { struct mlx4_spec_eth eth; struct mlx4_spec_ib ib; struct mlx4_spec_ipv4 ipv4; struct mlx4_spec_tcp_udp tcp_udp; }; }; enum mlx4_net_trans_hw_rule_queue { MLX4_NET_TRANS_Q_FIFO, MLX4_NET_TRANS_Q_LIFO, }; struct mlx4_net_trans_rule { struct list_head list; enum mlx4_net_trans_hw_rule_queue queue_mode; bool exclusive; bool allow_loopback; enum mlx4_net_trans_promisc_mode promisc_mode; u8 port; u16 priority; u32 qpn; }; struct mlx4_net_trans_rule_hw_ctrl { __be16 prio; u8 type; u8 flags; u8 rsvd1; u8 funcid; u8 vep; u8 port; __be32 qpn; __be32 rsvd2; }; struct mlx4_net_trans_rule_hw_ib { u8 size; u8 rsvd1; __be16 id; u32 rsvd2; __be32 l3_qpn; __be32 qpn_mask; u8 dst_gid[16]; u8 dst_gid_msk[16]; } __packed; struct mlx4_net_trans_rule_hw_eth { u8 size; u8 rsvd; __be16 id; u8 rsvd1[6]; u8 dst_mac[6]; u16 rsvd2; u8 dst_mac_msk[6]; u16 rsvd3; u8 src_mac[6]; u16 rsvd4; u8 src_mac_msk[6]; u8 rsvd5; u8 ether_type_enable; __be16 ether_type; __be16 vlan_tag_msk; __be16 vlan_tag; } __packed; struct mlx4_net_trans_rule_hw_tcp_udp { u8 size; u8 rsvd; __be16 id; __be16 rsvd1[3]; __be16 dst_port; __be16 rsvd2; __be16 dst_port_msk; __be16 rsvd3; __be16 src_port; __be16 rsvd4; __be16 src_port_msk; } __packed; struct mlx4_net_trans_rule_hw_ipv4 { u8 size; u8 rsvd; __be16 id; __be32 rsvd1; __be32 dst_ip; __be32 dst_ip_msk; __be32 src_ip; __be32 src_ip_msk; } __packed; struct _rule_hw { union { struct { u8 size; u8 rsvd; __be16 id; }; struct mlx4_net_trans_rule_hw_eth eth; struct mlx4_net_trans_rule_hw_ib ib; struct mlx4_net_trans_rule_hw_ipv4 ipv4; struct mlx4_net_trans_rule_hw_tcp_udp tcp_udp; }; }; int mlx4_flow_steer_promisc_add(struct mlx4_dev *dev, u8 port, u32 qpn, enum mlx4_net_trans_promisc_mode mode); int mlx4_flow_steer_promisc_remove(struct mlx4_dev *dev, u8 port, enum mlx4_net_trans_promisc_mode mode); int mlx4_multicast_promisc_add(struct mlx4_dev *dev, u32 qpn, u8 port); int mlx4_multicast_promisc_remove(struct mlx4_dev *dev, u32 qpn, u8 port); int mlx4_unicast_promisc_add(struct mlx4_dev *dev, u32 qpn, u8 port); int mlx4_unicast_promisc_remove(struct mlx4_dev *dev, u32 qpn, u8 port); int mlx4_register_mac(struct mlx4_dev *dev, u8 port, u64 mac); void mlx4_unregister_mac(struct mlx4_dev *dev, u8 port, u64 mac); int mlx4_get_base_qpn(struct mlx4_dev *dev, u8 port); int __mlx4_replace_mac(struct mlx4_dev *dev, u8 port, int qpn, u64 new_mac); void mlx4_set_stats_bitmap(struct mlx4_dev *dev, unsigned long *stats_bitmap); int mlx4_SET_PORT_general(struct mlx4_dev *dev, u8 port, int mtu, u8 pptx, u8 pfctx, u8 pprx, u8 pfcrx); int mlx4_SET_PORT_qpn_calc(struct mlx4_dev *dev, u8 port, u32 base_qpn, u8 promisc); int mlx4_SET_PORT_PRIO2TC(struct mlx4_dev *dev, u8 port, u8 *prio2tc); int mlx4_SET_PORT_SCHEDULER(struct mlx4_dev *dev, u8 port, u8 *tc_tx_bw, u8 *pg, u16 *ratelimit); int mlx4_find_cached_vlan(struct mlx4_dev *dev, u8 port, u16 vid, int *idx); int mlx4_register_vlan(struct mlx4_dev *dev, u8 port, u16 vlan, int *index); void mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, u16 vlan); int mlx4_map_phys_fmr(struct mlx4_dev *dev, struct mlx4_fmr *fmr, u64 *page_list, int npages, u64 iova, u32 *lkey, u32 *rkey); int mlx4_fmr_alloc(struct mlx4_dev *dev, u32 pd, u32 access, int max_pages, int max_maps, u8 page_shift, struct mlx4_fmr *fmr); int mlx4_fmr_enable(struct mlx4_dev *dev, struct mlx4_fmr *fmr); void mlx4_fmr_unmap(struct mlx4_dev *dev, struct mlx4_fmr *fmr, u32 *lkey, u32 *rkey); int mlx4_fmr_free(struct mlx4_dev *dev, struct mlx4_fmr *fmr); int mlx4_SYNC_TPT(struct mlx4_dev *dev); int mlx4_query_diag_counters(struct mlx4_dev *mlx4_dev, int array_length, u8 op_modifier, u32 in_offset[], u32 counter_out[]); int mlx4_test_interrupts(struct mlx4_dev *dev); int mlx4_assign_eq(struct mlx4_dev *dev, char* name, int * vector); void mlx4_release_eq(struct mlx4_dev *dev, int vec); int mlx4_wol_read(struct mlx4_dev *dev, u64 *config, int port); int mlx4_wol_write(struct mlx4_dev *dev, u64 config, int port); int mlx4_counter_alloc(struct mlx4_dev *dev, u8 port, u32 *idx); void mlx4_counter_free(struct mlx4_dev *dev, u8 port, u32 idx); int mlx4_flow_attach(struct mlx4_dev *dev, struct mlx4_net_trans_rule *rule, u64 *reg_id); int mlx4_flow_detach(struct mlx4_dev *dev, u64 reg_id); int map_sw_to_hw_steering_mode(struct mlx4_dev *dev, enum mlx4_net_trans_promisc_mode flow_type); int map_sw_to_hw_steering_id(struct mlx4_dev *dev, enum mlx4_net_trans_rule_id id); int hw_rule_sz(struct mlx4_dev *dev, enum mlx4_net_trans_rule_id id); void mlx4_sync_pkey_table(struct mlx4_dev *dev, int slave, int port, int i, int val); int mlx4_get_parav_qkey(struct mlx4_dev *dev, u32 qpn, u32 *qkey); int mlx4_is_slave_active(struct mlx4_dev *dev, int slave); int mlx4_gen_pkey_eqe(struct mlx4_dev *dev, int slave, u8 port); int mlx4_gen_guid_change_eqe(struct mlx4_dev *dev, int slave, u8 port); int mlx4_gen_slaves_port_mgt_ev(struct mlx4_dev *dev, u8 port, int attr, u16 lid, u8 sl); int mlx4_gen_port_state_change_eqe(struct mlx4_dev *dev, int slave, u8 port, u8 port_subtype_change); enum slave_port_state mlx4_get_slave_port_state(struct mlx4_dev *dev, int slave, u8 port); int set_and_calc_slave_port_state(struct mlx4_dev *dev, int slave, u8 port, int event, enum slave_port_gen_event *gen_event); void mlx4_put_slave_node_guid(struct mlx4_dev *dev, int slave, __be64 guid); __be64 mlx4_get_slave_node_guid(struct mlx4_dev *dev, int slave); int mlx4_get_slave_from_roce_gid(struct mlx4_dev *dev, int port, u8 *gid, int *slave_id); int mlx4_get_roce_gid_from_slave(struct mlx4_dev *dev, int port, int slave_id, u8 *gid); int mlx4_FLOW_STEERING_IB_UC_QP_RANGE(struct mlx4_dev *dev, u32 min_range_qpn, u32 max_range_qpn); int mlx4_read_clock(struct mlx4_dev *dev); int mlx4_get_internal_clock_params(struct mlx4_dev *dev, struct mlx4_clock_params *params); #endif /* MLX4_DEVICE_H */ Index: projects/cxl_iscsi/sys/ofed/include/linux/mlx4/qp.h =================================================================== --- projects/cxl_iscsi/sys/ofed/include/linux/mlx4/qp.h (revision 279716) +++ projects/cxl_iscsi/sys/ofed/include/linux/mlx4/qp.h (revision 279717) @@ -1,445 +1,447 @@ /* * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU * General Public License (GPL) Version 2, available from the file * COPYING in the main directory of this source tree, or the * OpenIB.org BSD license below: * * Redistribution and use in source and binary forms, with or * without modification, are permitted provided that the following * conditions are met: * * - Redistributions of source code must retain the above * copyright notice, this list of conditions and the following * disclaimer. * * - Redistributions in binary form must reproduce the above * copyright notice, this list of conditions and the following * disclaimer in the documentation and/or other materials * provided with the distribution. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ #ifndef MLX4_QP_H #define MLX4_QP_H #include #include #define MLX4_INVALID_LKEY 0x100 enum ib_m_qp_attr_mask { IB_M_EXT_CLASS_1 = 1 << 28, IB_M_EXT_CLASS_2 = 1 << 29, IB_M_EXT_CLASS_3 = 1 << 30, IB_M_QP_MOD_VEND_MASK = (IB_M_EXT_CLASS_1 | IB_M_EXT_CLASS_2 | IB_M_EXT_CLASS_3) }; enum mlx4_qp_optpar { MLX4_QP_OPTPAR_ALT_ADDR_PATH = 1 << 0, MLX4_QP_OPTPAR_RRE = 1 << 1, MLX4_QP_OPTPAR_RAE = 1 << 2, MLX4_QP_OPTPAR_RWE = 1 << 3, MLX4_QP_OPTPAR_PKEY_INDEX = 1 << 4, MLX4_QP_OPTPAR_Q_KEY = 1 << 5, MLX4_QP_OPTPAR_RNR_TIMEOUT = 1 << 6, MLX4_QP_OPTPAR_PRIMARY_ADDR_PATH = 1 << 7, MLX4_QP_OPTPAR_SRA_MAX = 1 << 8, MLX4_QP_OPTPAR_RRA_MAX = 1 << 9, MLX4_QP_OPTPAR_PM_STATE = 1 << 10, MLX4_QP_OPTPAR_RETRY_COUNT = 1 << 12, MLX4_QP_OPTPAR_RNR_RETRY = 1 << 13, MLX4_QP_OPTPAR_ACK_TIMEOUT = 1 << 14, MLX4_QP_OPTPAR_SCHED_QUEUE = 1 << 16, MLX4_QP_OPTPAR_COUNTER_INDEX = 1 << 20 }; enum mlx4_qp_state { MLX4_QP_STATE_RST = 0, MLX4_QP_STATE_INIT = 1, MLX4_QP_STATE_RTR = 2, MLX4_QP_STATE_RTS = 3, MLX4_QP_STATE_SQER = 4, MLX4_QP_STATE_SQD = 5, MLX4_QP_STATE_ERR = 6, MLX4_QP_STATE_SQ_DRAINING = 7, MLX4_QP_NUM_STATE }; enum { MLX4_QP_ST_RC = 0x0, MLX4_QP_ST_UC = 0x1, MLX4_QP_ST_RD = 0x2, MLX4_QP_ST_UD = 0x3, MLX4_QP_ST_XRC = 0x6, MLX4_QP_ST_MLX = 0x7 }; enum { MLX4_QP_PM_MIGRATED = 0x3, MLX4_QP_PM_ARMED = 0x0, MLX4_QP_PM_REARM = 0x1 }; enum { /* params1 */ MLX4_QP_BIT_SRE = 1 << 15, MLX4_QP_BIT_SWE = 1 << 14, MLX4_QP_BIT_SAE = 1 << 13, /* params2 */ MLX4_QP_BIT_RRE = 1 << 15, MLX4_QP_BIT_RWE = 1 << 14, MLX4_QP_BIT_RAE = 1 << 13, MLX4_QP_BIT_RIC = 1 << 4, MLX4_QP_BIT_COLL_SYNC_RQ = 1 << 2, MLX4_QP_BIT_COLL_SYNC_SQ = 1 << 1, MLX4_QP_BIT_COLL_MASTER = 1 << 0 }; enum { MLX4_RSS_HASH_XOR = 0, MLX4_RSS_HASH_TOP = 1, MLX4_RSS_UDP_IPV6 = 1 << 0, MLX4_RSS_UDP_IPV4 = 1 << 1, MLX4_RSS_TCP_IPV6 = 1 << 2, MLX4_RSS_IPV6 = 1 << 3, MLX4_RSS_TCP_IPV4 = 1 << 4, MLX4_RSS_IPV4 = 1 << 5, /* offset of mlx4_rss_context within mlx4_qp_context.pri_path */ MLX4_RSS_OFFSET_IN_QPC_PRI_PATH = 0x24, /* offset of being RSS indirection QP within mlx4_qp_context.flags */ MLX4_RSS_QPC_FLAG_OFFSET = 13, }; struct mlx4_rss_context { __be32 base_qpn; __be32 default_qpn; u16 reserved; u8 hash_fn; u8 flags; __be32 rss_key[10]; __be32 base_qpn_udp; }; struct mlx4_qp_path { u8 fl; u8 vlan_control; u8 disable_pkey_check; u8 pkey_index; u8 counter_index; u8 grh_mylmc; __be16 rlid; u8 ackto; u8 mgid_index; u8 static_rate; u8 hop_limit; __be32 tclass_flowlabel; u8 rgid[16]; u8 sched_queue; u8 vlan_index; u8 feup; u8 fvl_rx; u8 reserved4[2]; u8 dmac[6]; }; enum { /* fl */ MLX4_FL_CV = 1 << 6, MLX4_FL_ETH_HIDE_CQE_VLAN = 1 << 2, MLX4_FL_ETH_SRC_CHECK_MC_LB = 1 << 1, MLX4_FL_ETH_SRC_CHECK_UC_LB = 1 << 0, }; enum { /* vlan_control */ MLX4_VLAN_CTRL_ETH_SRC_CHECK_IF_COUNTER = 1 << 7, MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED = 1 << 6, MLX4_VLAN_CTRL_ETH_RX_BLOCK_TAGGED = 1 << 2, MLX4_VLAN_CTRL_ETH_RX_BLOCK_PRIO_TAGGED = 1 << 1,/* 802.1p priorty tag*/ MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED = 1 << 0 }; enum { /* feup */ MLX4_FEUP_FORCE_ETH_UP = 1 << 6, /* force Eth UP */ MLX4_FSM_FORCE_ETH_SRC_MAC = 1 << 5, /* force Source MAC */ MLX4_FVL_FORCE_ETH_VLAN = 1 << 3 /* force Eth vlan */ }; enum { /* fvl_rx */ MLX4_FVL_RX_FORCE_ETH_VLAN = 1 << 0 /* enforce Eth rx vlan */ }; struct mlx4_qp_context { __be32 flags; __be32 pd; u8 mtu_msgmax; u8 rq_size_stride; u8 sq_size_stride; u8 rlkey; __be32 usr_page; __be32 local_qpn; __be32 remote_qpn; struct mlx4_qp_path pri_path; struct mlx4_qp_path alt_path; __be32 params1; u32 reserved1; __be32 next_send_psn; __be32 cqn_send; u32 reserved2[2]; __be32 last_acked_psn; __be32 ssn; __be32 params2; __be32 rnr_nextrecvpsn; __be32 xrcd; __be32 cqn_recv; __be64 db_rec_addr; __be32 qkey; __be32 srqn; __be32 msn; __be16 rq_wqe_counter; __be16 sq_wqe_counter; u32 reserved3[2]; __be32 param3; __be32 nummmcpeers_basemkey; u8 log_page_size; u8 reserved4[2]; u8 mtt_base_addr_h; __be32 mtt_base_addr_l; u32 reserved5[10]; }; struct mlx4_update_qp_context { __be64 qp_mask; __be64 primary_addr_path_mask; __be64 secondary_addr_path_mask; u64 reserved1; struct mlx4_qp_context qp_context; u64 reserved2[58]; }; enum { MLX4_UPD_QP_MASK_PM_STATE = 32, MLX4_UPD_QP_MASK_VSD = 33, }; enum { MLX4_UPD_QP_PATH_MASK_PKEY_INDEX = 0 + 32, MLX4_UPD_QP_PATH_MASK_FSM = 1 + 32, MLX4_UPD_QP_PATH_MASK_MAC_INDEX = 2 + 32, MLX4_UPD_QP_PATH_MASK_FVL = 3 + 32, MLX4_UPD_QP_PATH_MASK_CV = 4 + 32, MLX4_UPD_QP_PATH_MASK_VLAN_INDEX = 5 + 32, MLX4_UPD_QP_PATH_MASK_ETH_HIDE_CQE_VLAN = 6 + 32, MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_UNTAGGED = 7 + 32, MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_1P = 8 + 32, MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_TAGGED = 9 + 32, MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_UNTAGGED = 10 + 32, MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_1P = 11 + 32, MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_TAGGED = 12 + 32, MLX4_UPD_QP_PATH_MASK_FEUP = 13 + 32, MLX4_UPD_QP_PATH_MASK_SCHED_QUEUE = 14 + 32, MLX4_UPD_QP_PATH_MASK_IF_COUNTER_INDEX = 15 + 32, MLX4_UPD_QP_PATH_MASK_FVL_RX = 16 + 32, + MLX4_UPD_QP_PATH_MASK_ETH_SRC_CHECK_UC_LB = 18 + 32, + MLX4_UPD_QP_PATH_MASK_ETH_SRC_CHECK_MC_LB = 19 + 32, }; enum { /* param3 */ MLX4_STRIP_VLAN = 1 << 30 }; /* Which firmware version adds support for NEC (NoErrorCompletion) bit */ #define MLX4_FW_VER_WQE_CTRL_NEC mlx4_fw_ver(2, 2, 232) enum { MLX4_WQE_CTRL_NEC = 1 << 29, MLX4_WQE_CTRL_FENCE = 1 << 6, MLX4_WQE_CTRL_CQ_UPDATE = 3 << 2, MLX4_WQE_CTRL_SOLICITED = 1 << 1, MLX4_WQE_CTRL_IP_CSUM = 1 << 4, MLX4_WQE_CTRL_TCP_UDP_CSUM = 1 << 5, MLX4_WQE_CTRL_INS_VLAN = 1 << 6, MLX4_WQE_CTRL_STRONG_ORDER = 1 << 7, MLX4_WQE_CTRL_FORCE_LOOPBACK = 1 << 0, }; struct mlx4_wqe_ctrl_seg { __be32 owner_opcode; __be16 vlan_tag; u8 ins_vlan; u8 fence_size; /* * High 24 bits are SRC remote buffer; low 8 bits are flags: * [7] SO (strong ordering) * [5] TCP/UDP checksum * [4] IP checksum * [3:2] C (generate completion queue entry) * [1] SE (solicited event) * [0] FL (force loopback) */ union { __be32 srcrb_flags; __be16 srcrb_flags16[2]; }; /* * imm is immediate data for send/RDMA write w/ immediate; * also invalidation key for send with invalidate; input * modifier for WQEs on CCQs. */ __be32 imm; }; enum { MLX4_WQE_MLX_VL15 = 1 << 17, MLX4_WQE_MLX_SLR = 1 << 16 }; struct mlx4_wqe_mlx_seg { u8 owner; u8 reserved1[2]; u8 opcode; __be16 sched_prio; u8 reserved2; u8 size; /* * [17] VL15 * [16] SLR * [15:12] static rate * [11:8] SL * [4] ICRC * [3:2] C * [0] FL (force loopback) */ __be32 flags; __be16 rlid; u16 reserved3; }; struct mlx4_wqe_datagram_seg { __be32 av[8]; __be32 dqpn; __be32 qkey; __be16 vlan; u8 mac[6]; }; struct mlx4_wqe_lso_seg { __be32 mss_hdr_size; __be32 header[0]; }; enum mlx4_wqe_bind_seg_flags2 { MLX4_WQE_BIND_TYPE_2 = (1<<31), MLX4_WQE_BIND_ZERO_BASED = (1<<30), }; struct mlx4_wqe_bind_seg { __be32 flags1; __be32 flags2; __be32 new_rkey; __be32 lkey; __be64 addr; __be64 length; }; enum { MLX4_WQE_FMR_PERM_LOCAL_READ = 1 << 27, MLX4_WQE_FMR_PERM_LOCAL_WRITE = 1 << 28, MLX4_WQE_FMR_AND_BIND_PERM_REMOTE_READ = 1 << 29, MLX4_WQE_FMR_AND_BIND_PERM_REMOTE_WRITE = 1 << 30, MLX4_WQE_FMR_AND_BIND_PERM_ATOMIC = 1 << 31 }; struct mlx4_wqe_fmr_seg { __be32 flags; __be32 mem_key; __be64 buf_list; __be64 start_addr; __be64 reg_len; __be32 offset; __be32 page_size; u32 reserved[2]; }; struct mlx4_wqe_fmr_ext_seg { u8 flags; u8 reserved; __be16 app_mask; __be16 wire_app_tag; __be16 mem_app_tag; __be32 wire_ref_tag_base; __be32 mem_ref_tag_base; }; struct mlx4_wqe_local_inval_seg { u64 reserved1; __be32 mem_key; u32 reserved2; u64 reserved3[2]; }; struct mlx4_wqe_raddr_seg { __be64 raddr; __be32 rkey; u32 reserved; }; struct mlx4_wqe_atomic_seg { __be64 swap_add; __be64 compare; }; struct mlx4_wqe_masked_atomic_seg { __be64 swap_add; __be64 compare; __be64 swap_add_mask; __be64 compare_mask; }; struct mlx4_wqe_data_seg { __be32 byte_count; __be32 lkey; __be64 addr; }; enum { MLX4_INLINE_ALIGN = 64, MLX4_INLINE_SEG = 1 << 31, }; struct mlx4_wqe_inline_seg { __be32 byte_count; }; int mlx4_qp_modify(struct mlx4_dev *dev, struct mlx4_mtt *mtt, enum mlx4_qp_state cur_state, enum mlx4_qp_state new_state, struct mlx4_qp_context *context, enum mlx4_qp_optpar optpar, int sqd_event, struct mlx4_qp *qp); int mlx4_qp_query(struct mlx4_dev *dev, struct mlx4_qp *qp, struct mlx4_qp_context *context); int mlx4_qp_to_ready(struct mlx4_dev *dev, struct mlx4_mtt *mtt, struct mlx4_qp_context *context, struct mlx4_qp *qp, enum mlx4_qp_state *qp_state); static inline struct mlx4_qp *__mlx4_qp_lookup(struct mlx4_dev *dev, u32 qpn) { return radix_tree_lookup(&dev->qp_table_tree, qpn & (dev->caps.num_qps - 1)); } void mlx4_qp_remove(struct mlx4_dev *dev, struct mlx4_qp *qp); #endif /* MLX4_QP_H */ Index: projects/cxl_iscsi/sys/powerpc/aim/mmu_oea.c =================================================================== --- projects/cxl_iscsi/sys/powerpc/aim/mmu_oea.c (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/aim/mmu_oea.c (revision 279717) @@ -1,2685 +1,2687 @@ /*- * Copyright (c) 2001 The NetBSD Foundation, Inc. * All rights reserved. * * This code is derived from software contributed to The NetBSD Foundation * by Matt Thomas of Allegro Networks, Inc. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. */ /*- * Copyright (C) 1995, 1996 Wolfgang Solfrank. * Copyright (C) 1995, 1996 TooLs GmbH. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by TooLs GmbH. * 4. The name of TooLs GmbH may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * $NetBSD: pmap.c,v 1.28 2000/03/26 20:42:36 kleink Exp $ */ /*- * Copyright (C) 2001 Benno Rice. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY Benno Rice ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); /* * Manages physical address maps. * * Since the information managed by this module is also stored by the * logical address mapping module, this module may throw away valid virtual * to physical mappings at almost any time. However, invalidations of * mappings must be done as requested. * * In order to cope with hardware architectures which make virtual to * physical map invalidates expensive, this module may delay invalidate * reduced protection operations until such time as they are actually * necessary. This module is given full information as to which processors * are currently using which maps, and to when physical maps must be made * correct. */ #include "opt_kstack_pages.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "mmu_if.h" #define MOEA_DEBUG #define TODO panic("%s: not implemented", __func__); #define VSID_MAKE(sr, hash) ((sr) | (((hash) & 0xfffff) << 4)) #define VSID_TO_SR(vsid) ((vsid) & 0xf) #define VSID_TO_HASH(vsid) (((vsid) >> 4) & 0xfffff) struct ofw_map { vm_offset_t om_va; vm_size_t om_len; vm_offset_t om_pa; u_int om_mode; }; extern unsigned char _etext[]; extern unsigned char _end[]; /* * Map of physical memory regions. */ static struct mem_region *regions; static struct mem_region *pregions; static u_int phys_avail_count; static int regions_sz, pregions_sz; static struct ofw_map *translations; /* * Lock for the pteg and pvo tables. */ struct mtx moea_table_mutex; struct mtx moea_vsid_mutex; /* tlbie instruction synchronization */ static struct mtx tlbie_mtx; /* * PTEG data. */ static struct pteg *moea_pteg_table; u_int moea_pteg_count; u_int moea_pteg_mask; /* * PVO data. */ struct pvo_head *moea_pvo_table; /* pvo entries by pteg index */ struct pvo_head moea_pvo_kunmanaged = LIST_HEAD_INITIALIZER(moea_pvo_kunmanaged); /* list of unmanaged pages */ static struct rwlock_padalign pvh_global_lock; uma_zone_t moea_upvo_zone; /* zone for pvo entries for unmanaged pages */ uma_zone_t moea_mpvo_zone; /* zone for pvo entries for managed pages */ #define BPVO_POOL_SIZE 32768 static struct pvo_entry *moea_bpvo_pool; static int moea_bpvo_pool_index = 0; #define VSID_NBPW (sizeof(u_int32_t) * 8) static u_int moea_vsid_bitmap[NPMAPS / VSID_NBPW]; static boolean_t moea_initialized = FALSE; /* * Statistics. */ u_int moea_pte_valid = 0; u_int moea_pte_overflow = 0; u_int moea_pte_replacements = 0; u_int moea_pvo_entries = 0; u_int moea_pvo_enter_calls = 0; u_int moea_pvo_remove_calls = 0; u_int moea_pte_spills = 0; SYSCTL_INT(_machdep, OID_AUTO, moea_pte_valid, CTLFLAG_RD, &moea_pte_valid, 0, ""); SYSCTL_INT(_machdep, OID_AUTO, moea_pte_overflow, CTLFLAG_RD, &moea_pte_overflow, 0, ""); SYSCTL_INT(_machdep, OID_AUTO, moea_pte_replacements, CTLFLAG_RD, &moea_pte_replacements, 0, ""); SYSCTL_INT(_machdep, OID_AUTO, moea_pvo_entries, CTLFLAG_RD, &moea_pvo_entries, 0, ""); SYSCTL_INT(_machdep, OID_AUTO, moea_pvo_enter_calls, CTLFLAG_RD, &moea_pvo_enter_calls, 0, ""); SYSCTL_INT(_machdep, OID_AUTO, moea_pvo_remove_calls, CTLFLAG_RD, &moea_pvo_remove_calls, 0, ""); SYSCTL_INT(_machdep, OID_AUTO, moea_pte_spills, CTLFLAG_RD, &moea_pte_spills, 0, ""); /* * Allocate physical memory for use in moea_bootstrap. */ static vm_offset_t moea_bootstrap_alloc(vm_size_t, u_int); /* * PTE calls. */ static int moea_pte_insert(u_int, struct pte *); /* * PVO calls. */ static int moea_pvo_enter(pmap_t, uma_zone_t, struct pvo_head *, vm_offset_t, vm_offset_t, u_int, int); static void moea_pvo_remove(struct pvo_entry *, int); static struct pvo_entry *moea_pvo_find_va(pmap_t, vm_offset_t, int *); static struct pte *moea_pvo_to_pte(const struct pvo_entry *, int); /* * Utility routines. */ static int moea_enter_locked(pmap_t, vm_offset_t, vm_page_t, vm_prot_t, u_int, int8_t); static void moea_syncicache(vm_offset_t, vm_size_t); static boolean_t moea_query_bit(vm_page_t, int); static u_int moea_clear_bit(vm_page_t, int); static void moea_kremove(mmu_t, vm_offset_t); int moea_pte_spill(vm_offset_t); /* * Kernel MMU interface */ void moea_clear_modify(mmu_t, vm_page_t); void moea_copy_page(mmu_t, vm_page_t, vm_page_t); void moea_copy_pages(mmu_t mmu, vm_page_t *ma, vm_offset_t a_offset, vm_page_t *mb, vm_offset_t b_offset, int xfersize); int moea_enter(mmu_t, pmap_t, vm_offset_t, vm_page_t, vm_prot_t, u_int, int8_t); void moea_enter_object(mmu_t, pmap_t, vm_offset_t, vm_offset_t, vm_page_t, vm_prot_t); void moea_enter_quick(mmu_t, pmap_t, vm_offset_t, vm_page_t, vm_prot_t); vm_paddr_t moea_extract(mmu_t, pmap_t, vm_offset_t); vm_page_t moea_extract_and_hold(mmu_t, pmap_t, vm_offset_t, vm_prot_t); void moea_init(mmu_t); boolean_t moea_is_modified(mmu_t, vm_page_t); boolean_t moea_is_prefaultable(mmu_t, pmap_t, vm_offset_t); boolean_t moea_is_referenced(mmu_t, vm_page_t); int moea_ts_referenced(mmu_t, vm_page_t); vm_offset_t moea_map(mmu_t, vm_offset_t *, vm_paddr_t, vm_paddr_t, int); boolean_t moea_page_exists_quick(mmu_t, pmap_t, vm_page_t); int moea_page_wired_mappings(mmu_t, vm_page_t); void moea_pinit(mmu_t, pmap_t); void moea_pinit0(mmu_t, pmap_t); void moea_protect(mmu_t, pmap_t, vm_offset_t, vm_offset_t, vm_prot_t); void moea_qenter(mmu_t, vm_offset_t, vm_page_t *, int); void moea_qremove(mmu_t, vm_offset_t, int); void moea_release(mmu_t, pmap_t); void moea_remove(mmu_t, pmap_t, vm_offset_t, vm_offset_t); void moea_remove_all(mmu_t, vm_page_t); void moea_remove_write(mmu_t, vm_page_t); void moea_unwire(mmu_t, pmap_t, vm_offset_t, vm_offset_t); void moea_zero_page(mmu_t, vm_page_t); void moea_zero_page_area(mmu_t, vm_page_t, int, int); void moea_zero_page_idle(mmu_t, vm_page_t); void moea_activate(mmu_t, struct thread *); void moea_deactivate(mmu_t, struct thread *); void moea_cpu_bootstrap(mmu_t, int); void moea_bootstrap(mmu_t, vm_offset_t, vm_offset_t); void *moea_mapdev(mmu_t, vm_paddr_t, vm_size_t); void *moea_mapdev_attr(mmu_t, vm_offset_t, vm_size_t, vm_memattr_t); void moea_unmapdev(mmu_t, vm_offset_t, vm_size_t); vm_paddr_t moea_kextract(mmu_t, vm_offset_t); void moea_kenter_attr(mmu_t, vm_offset_t, vm_offset_t, vm_memattr_t); void moea_kenter(mmu_t, vm_offset_t, vm_paddr_t); void moea_page_set_memattr(mmu_t mmu, vm_page_t m, vm_memattr_t ma); boolean_t moea_dev_direct_mapped(mmu_t, vm_paddr_t, vm_size_t); static void moea_sync_icache(mmu_t, pmap_t, vm_offset_t, vm_size_t); void moea_dumpsys_map(mmu_t mmu, vm_paddr_t pa, size_t sz, void **va); void moea_scan_init(mmu_t mmu); static mmu_method_t moea_methods[] = { MMUMETHOD(mmu_clear_modify, moea_clear_modify), MMUMETHOD(mmu_copy_page, moea_copy_page), MMUMETHOD(mmu_copy_pages, moea_copy_pages), MMUMETHOD(mmu_enter, moea_enter), MMUMETHOD(mmu_enter_object, moea_enter_object), MMUMETHOD(mmu_enter_quick, moea_enter_quick), MMUMETHOD(mmu_extract, moea_extract), MMUMETHOD(mmu_extract_and_hold, moea_extract_and_hold), MMUMETHOD(mmu_init, moea_init), MMUMETHOD(mmu_is_modified, moea_is_modified), MMUMETHOD(mmu_is_prefaultable, moea_is_prefaultable), MMUMETHOD(mmu_is_referenced, moea_is_referenced), MMUMETHOD(mmu_ts_referenced, moea_ts_referenced), MMUMETHOD(mmu_map, moea_map), MMUMETHOD(mmu_page_exists_quick,moea_page_exists_quick), MMUMETHOD(mmu_page_wired_mappings,moea_page_wired_mappings), MMUMETHOD(mmu_pinit, moea_pinit), MMUMETHOD(mmu_pinit0, moea_pinit0), MMUMETHOD(mmu_protect, moea_protect), MMUMETHOD(mmu_qenter, moea_qenter), MMUMETHOD(mmu_qremove, moea_qremove), MMUMETHOD(mmu_release, moea_release), MMUMETHOD(mmu_remove, moea_remove), MMUMETHOD(mmu_remove_all, moea_remove_all), MMUMETHOD(mmu_remove_write, moea_remove_write), MMUMETHOD(mmu_sync_icache, moea_sync_icache), MMUMETHOD(mmu_unwire, moea_unwire), MMUMETHOD(mmu_zero_page, moea_zero_page), MMUMETHOD(mmu_zero_page_area, moea_zero_page_area), MMUMETHOD(mmu_zero_page_idle, moea_zero_page_idle), MMUMETHOD(mmu_activate, moea_activate), MMUMETHOD(mmu_deactivate, moea_deactivate), MMUMETHOD(mmu_page_set_memattr, moea_page_set_memattr), /* Internal interfaces */ MMUMETHOD(mmu_bootstrap, moea_bootstrap), MMUMETHOD(mmu_cpu_bootstrap, moea_cpu_bootstrap), MMUMETHOD(mmu_mapdev_attr, moea_mapdev_attr), MMUMETHOD(mmu_mapdev, moea_mapdev), MMUMETHOD(mmu_unmapdev, moea_unmapdev), MMUMETHOD(mmu_kextract, moea_kextract), MMUMETHOD(mmu_kenter, moea_kenter), MMUMETHOD(mmu_kenter_attr, moea_kenter_attr), MMUMETHOD(mmu_dev_direct_mapped,moea_dev_direct_mapped), MMUMETHOD(mmu_scan_init, moea_scan_init), MMUMETHOD(mmu_dumpsys_map, moea_dumpsys_map), { 0, 0 } }; MMU_DEF(oea_mmu, MMU_TYPE_OEA, moea_methods, 0); static __inline uint32_t moea_calc_wimg(vm_offset_t pa, vm_memattr_t ma) { uint32_t pte_lo; int i; if (ma != VM_MEMATTR_DEFAULT) { switch (ma) { case VM_MEMATTR_UNCACHEABLE: return (PTE_I | PTE_G); case VM_MEMATTR_WRITE_COMBINING: case VM_MEMATTR_WRITE_BACK: case VM_MEMATTR_PREFETCHABLE: return (PTE_I); case VM_MEMATTR_WRITE_THROUGH: return (PTE_W | PTE_M); } } /* * Assume the page is cache inhibited and access is guarded unless * it's in our available memory array. */ pte_lo = PTE_I | PTE_G; for (i = 0; i < pregions_sz; i++) { if ((pa >= pregions[i].mr_start) && (pa < (pregions[i].mr_start + pregions[i].mr_size))) { pte_lo = PTE_M; break; } } return pte_lo; } static void tlbie(vm_offset_t va) { mtx_lock_spin(&tlbie_mtx); __asm __volatile("ptesync"); __asm __volatile("tlbie %0" :: "r"(va)); __asm __volatile("eieio; tlbsync; ptesync"); mtx_unlock_spin(&tlbie_mtx); } static void tlbia(void) { vm_offset_t va; for (va = 0; va < 0x00040000; va += 0x00001000) { __asm __volatile("tlbie %0" :: "r"(va)); powerpc_sync(); } __asm __volatile("tlbsync"); powerpc_sync(); } static __inline int va_to_sr(u_int *sr, vm_offset_t va) { return (sr[(uintptr_t)va >> ADDR_SR_SHFT]); } static __inline u_int va_to_pteg(u_int sr, vm_offset_t addr) { u_int hash; hash = (sr & SR_VSID_MASK) ^ (((u_int)addr & ADDR_PIDX) >> ADDR_PIDX_SHFT); return (hash & moea_pteg_mask); } static __inline struct pvo_head * vm_page_to_pvoh(vm_page_t m) { return (&m->md.mdpg_pvoh); } static __inline void moea_attr_clear(vm_page_t m, int ptebit) { rw_assert(&pvh_global_lock, RA_WLOCKED); m->md.mdpg_attrs &= ~ptebit; } static __inline int moea_attr_fetch(vm_page_t m) { return (m->md.mdpg_attrs); } static __inline void moea_attr_save(vm_page_t m, int ptebit) { rw_assert(&pvh_global_lock, RA_WLOCKED); m->md.mdpg_attrs |= ptebit; } static __inline int moea_pte_compare(const struct pte *pt, const struct pte *pvo_pt) { if (pt->pte_hi == pvo_pt->pte_hi) return (1); return (0); } static __inline int moea_pte_match(struct pte *pt, u_int sr, vm_offset_t va, int which) { return (pt->pte_hi & ~PTE_VALID) == (((sr & SR_VSID_MASK) << PTE_VSID_SHFT) | ((va >> ADDR_API_SHFT) & PTE_API) | which); } static __inline void moea_pte_create(struct pte *pt, u_int sr, vm_offset_t va, u_int pte_lo) { mtx_assert(&moea_table_mutex, MA_OWNED); /* * Construct a PTE. Default to IMB initially. Valid bit only gets * set when the real pte is set in memory. * * Note: Don't set the valid bit for correct operation of tlb update. */ pt->pte_hi = ((sr & SR_VSID_MASK) << PTE_VSID_SHFT) | (((va & ADDR_PIDX) >> ADDR_API_SHFT) & PTE_API); pt->pte_lo = pte_lo; } static __inline void moea_pte_synch(struct pte *pt, struct pte *pvo_pt) { mtx_assert(&moea_table_mutex, MA_OWNED); pvo_pt->pte_lo |= pt->pte_lo & (PTE_REF | PTE_CHG); } static __inline void moea_pte_clear(struct pte *pt, vm_offset_t va, int ptebit) { mtx_assert(&moea_table_mutex, MA_OWNED); /* * As shown in Section 7.6.3.2.3 */ pt->pte_lo &= ~ptebit; tlbie(va); } static __inline void moea_pte_set(struct pte *pt, struct pte *pvo_pt) { mtx_assert(&moea_table_mutex, MA_OWNED); pvo_pt->pte_hi |= PTE_VALID; /* * Update the PTE as defined in section 7.6.3.1. * Note that the REF/CHG bits are from pvo_pt and thus should have * been saved so this routine can restore them (if desired). */ pt->pte_lo = pvo_pt->pte_lo; powerpc_sync(); pt->pte_hi = pvo_pt->pte_hi; powerpc_sync(); moea_pte_valid++; } static __inline void moea_pte_unset(struct pte *pt, struct pte *pvo_pt, vm_offset_t va) { mtx_assert(&moea_table_mutex, MA_OWNED); pvo_pt->pte_hi &= ~PTE_VALID; /* * Force the reg & chg bits back into the PTEs. */ powerpc_sync(); /* * Invalidate the pte. */ pt->pte_hi &= ~PTE_VALID; tlbie(va); /* * Save the reg & chg bits. */ moea_pte_synch(pt, pvo_pt); moea_pte_valid--; } static __inline void moea_pte_change(struct pte *pt, struct pte *pvo_pt, vm_offset_t va) { /* * Invalidate the PTE */ moea_pte_unset(pt, pvo_pt, va); moea_pte_set(pt, pvo_pt); } /* * Quick sort callout for comparing memory regions. */ static int om_cmp(const void *a, const void *b); static int om_cmp(const void *a, const void *b) { const struct ofw_map *mapa; const struct ofw_map *mapb; mapa = a; mapb = b; if (mapa->om_pa < mapb->om_pa) return (-1); else if (mapa->om_pa > mapb->om_pa) return (1); else return (0); } void moea_cpu_bootstrap(mmu_t mmup, int ap) { u_int sdr; int i; if (ap) { powerpc_sync(); __asm __volatile("mtdbatu 0,%0" :: "r"(battable[0].batu)); __asm __volatile("mtdbatl 0,%0" :: "r"(battable[0].batl)); isync(); __asm __volatile("mtibatu 0,%0" :: "r"(battable[0].batu)); __asm __volatile("mtibatl 0,%0" :: "r"(battable[0].batl)); isync(); } __asm __volatile("mtdbatu 1,%0" :: "r"(battable[8].batu)); __asm __volatile("mtdbatl 1,%0" :: "r"(battable[8].batl)); isync(); __asm __volatile("mtibatu 1,%0" :: "r"(0)); __asm __volatile("mtdbatu 2,%0" :: "r"(0)); __asm __volatile("mtibatu 2,%0" :: "r"(0)); __asm __volatile("mtdbatu 3,%0" :: "r"(0)); __asm __volatile("mtibatu 3,%0" :: "r"(0)); isync(); for (i = 0; i < 16; i++) mtsrin(i << ADDR_SR_SHFT, kernel_pmap->pm_sr[i]); powerpc_sync(); sdr = (u_int)moea_pteg_table | (moea_pteg_mask >> 10); __asm __volatile("mtsdr1 %0" :: "r"(sdr)); isync(); tlbia(); } void moea_bootstrap(mmu_t mmup, vm_offset_t kernelstart, vm_offset_t kernelend) { ihandle_t mmui; phandle_t chosen, mmu; int sz; int i, j; vm_size_t size, physsz, hwphyssz; vm_offset_t pa, va, off; void *dpcpu; register_t msr; /* * Set up BAT0 to map the lowest 256 MB area */ battable[0x0].batl = BATL(0x00000000, BAT_M, BAT_PP_RW); battable[0x0].batu = BATU(0x00000000, BAT_BL_256M, BAT_Vs); /* * Map PCI memory space. */ battable[0x8].batl = BATL(0x80000000, BAT_I|BAT_G, BAT_PP_RW); battable[0x8].batu = BATU(0x80000000, BAT_BL_256M, BAT_Vs); battable[0x9].batl = BATL(0x90000000, BAT_I|BAT_G, BAT_PP_RW); battable[0x9].batu = BATU(0x90000000, BAT_BL_256M, BAT_Vs); battable[0xa].batl = BATL(0xa0000000, BAT_I|BAT_G, BAT_PP_RW); battable[0xa].batu = BATU(0xa0000000, BAT_BL_256M, BAT_Vs); battable[0xb].batl = BATL(0xb0000000, BAT_I|BAT_G, BAT_PP_RW); battable[0xb].batu = BATU(0xb0000000, BAT_BL_256M, BAT_Vs); /* * Map obio devices. */ battable[0xf].batl = BATL(0xf0000000, BAT_I|BAT_G, BAT_PP_RW); battable[0xf].batu = BATU(0xf0000000, BAT_BL_256M, BAT_Vs); /* * Use an IBAT and a DBAT to map the bottom segment of memory * where we are. Turn off instruction relocation temporarily * to prevent faults while reprogramming the IBAT. */ msr = mfmsr(); mtmsr(msr & ~PSL_IR); __asm (".balign 32; \n" "mtibatu 0,%0; mtibatl 0,%1; isync; \n" "mtdbatu 0,%0; mtdbatl 0,%1; isync" :: "r"(battable[0].batu), "r"(battable[0].batl)); mtmsr(msr); /* map pci space */ __asm __volatile("mtdbatu 1,%0" :: "r"(battable[8].batu)); __asm __volatile("mtdbatl 1,%0" :: "r"(battable[8].batl)); isync(); /* set global direct map flag */ hw_direct_map = 1; mem_regions(&pregions, &pregions_sz, ®ions, ®ions_sz); CTR0(KTR_PMAP, "moea_bootstrap: physical memory"); for (i = 0; i < pregions_sz; i++) { vm_offset_t pa; vm_offset_t end; CTR3(KTR_PMAP, "physregion: %#x - %#x (%#x)", pregions[i].mr_start, pregions[i].mr_start + pregions[i].mr_size, pregions[i].mr_size); /* * Install entries into the BAT table to allow all * of physmem to be convered by on-demand BAT entries. * The loop will sometimes set the same battable element * twice, but that's fine since they won't be used for * a while yet. */ pa = pregions[i].mr_start & 0xf0000000; end = pregions[i].mr_start + pregions[i].mr_size; do { u_int n = pa >> ADDR_SR_SHFT; battable[n].batl = BATL(pa, BAT_M, BAT_PP_RW); battable[n].batu = BATU(pa, BAT_BL_256M, BAT_Vs); pa += SEGMENT_LENGTH; } while (pa < end); } if (sizeof(phys_avail)/sizeof(phys_avail[0]) < regions_sz) panic("moea_bootstrap: phys_avail too small"); phys_avail_count = 0; physsz = 0; hwphyssz = 0; TUNABLE_ULONG_FETCH("hw.physmem", (u_long *) &hwphyssz); for (i = 0, j = 0; i < regions_sz; i++, j += 2) { CTR3(KTR_PMAP, "region: %#x - %#x (%#x)", regions[i].mr_start, regions[i].mr_start + regions[i].mr_size, regions[i].mr_size); if (hwphyssz != 0 && (physsz + regions[i].mr_size) >= hwphyssz) { if (physsz < hwphyssz) { phys_avail[j] = regions[i].mr_start; phys_avail[j + 1] = regions[i].mr_start + hwphyssz - physsz; physsz = hwphyssz; phys_avail_count++; } break; } phys_avail[j] = regions[i].mr_start; phys_avail[j + 1] = regions[i].mr_start + regions[i].mr_size; phys_avail_count++; physsz += regions[i].mr_size; } /* Check for overlap with the kernel and exception vectors */ for (j = 0; j < 2*phys_avail_count; j+=2) { if (phys_avail[j] < EXC_LAST) phys_avail[j] += EXC_LAST; if (kernelstart >= phys_avail[j] && kernelstart < phys_avail[j+1]) { if (kernelend < phys_avail[j+1]) { phys_avail[2*phys_avail_count] = (kernelend & ~PAGE_MASK) + PAGE_SIZE; phys_avail[2*phys_avail_count + 1] = phys_avail[j+1]; phys_avail_count++; } phys_avail[j+1] = kernelstart & ~PAGE_MASK; } if (kernelend >= phys_avail[j] && kernelend < phys_avail[j+1]) { if (kernelstart > phys_avail[j]) { phys_avail[2*phys_avail_count] = phys_avail[j]; phys_avail[2*phys_avail_count + 1] = kernelstart & ~PAGE_MASK; phys_avail_count++; } phys_avail[j] = (kernelend & ~PAGE_MASK) + PAGE_SIZE; } } physmem = btoc(physsz); /* * Allocate PTEG table. */ #ifdef PTEGCOUNT moea_pteg_count = PTEGCOUNT; #else moea_pteg_count = 0x1000; while (moea_pteg_count < physmem) moea_pteg_count <<= 1; moea_pteg_count >>= 1; #endif /* PTEGCOUNT */ size = moea_pteg_count * sizeof(struct pteg); CTR2(KTR_PMAP, "moea_bootstrap: %d PTEGs, %d bytes", moea_pteg_count, size); moea_pteg_table = (struct pteg *)moea_bootstrap_alloc(size, size); CTR1(KTR_PMAP, "moea_bootstrap: PTEG table at %p", moea_pteg_table); bzero((void *)moea_pteg_table, moea_pteg_count * sizeof(struct pteg)); moea_pteg_mask = moea_pteg_count - 1; /* * Allocate pv/overflow lists. */ size = sizeof(struct pvo_head) * moea_pteg_count; moea_pvo_table = (struct pvo_head *)moea_bootstrap_alloc(size, PAGE_SIZE); CTR1(KTR_PMAP, "moea_bootstrap: PVO table at %p", moea_pvo_table); for (i = 0; i < moea_pteg_count; i++) LIST_INIT(&moea_pvo_table[i]); /* * Initialize the lock that synchronizes access to the pteg and pvo * tables. */ mtx_init(&moea_table_mutex, "pmap table", NULL, MTX_DEF | MTX_RECURSE); mtx_init(&moea_vsid_mutex, "VSID table", NULL, MTX_DEF); mtx_init(&tlbie_mtx, "tlbie", NULL, MTX_SPIN); /* * Initialise the unmanaged pvo pool. */ moea_bpvo_pool = (struct pvo_entry *)moea_bootstrap_alloc( BPVO_POOL_SIZE*sizeof(struct pvo_entry), 0); moea_bpvo_pool_index = 0; /* * Make sure kernel vsid is allocated as well as VSID 0. */ moea_vsid_bitmap[(KERNEL_VSIDBITS & (NPMAPS - 1)) / VSID_NBPW] |= 1 << (KERNEL_VSIDBITS % VSID_NBPW); moea_vsid_bitmap[0] |= 1; /* * Initialize the kernel pmap (which is statically allocated). */ PMAP_LOCK_INIT(kernel_pmap); for (i = 0; i < 16; i++) kernel_pmap->pm_sr[i] = EMPTY_SEGMENT + i; CPU_FILL(&kernel_pmap->pm_active); RB_INIT(&kernel_pmap->pmap_pvo); /* * Initialize the global pv list lock. */ rw_init(&pvh_global_lock, "pmap pv global"); /* * Set up the Open Firmware mappings */ chosen = OF_finddevice("/chosen"); if (chosen != -1 && OF_getprop(chosen, "mmu", &mmui, 4) != -1 && (mmu = OF_instance_to_package(mmui)) != -1 && (sz = OF_getproplen(mmu, "translations")) != -1) { translations = NULL; for (i = 0; phys_avail[i] != 0; i += 2) { if (phys_avail[i + 1] >= sz) { translations = (struct ofw_map *)phys_avail[i]; break; } } if (translations == NULL) panic("moea_bootstrap: no space to copy translations"); bzero(translations, sz); if (OF_getprop(mmu, "translations", translations, sz) == -1) panic("moea_bootstrap: can't get ofw translations"); CTR0(KTR_PMAP, "moea_bootstrap: translations"); sz /= sizeof(*translations); qsort(translations, sz, sizeof (*translations), om_cmp); for (i = 0; i < sz; i++) { CTR3(KTR_PMAP, "translation: pa=%#x va=%#x len=%#x", translations[i].om_pa, translations[i].om_va, translations[i].om_len); /* * If the mapping is 1:1, let the RAM and device * on-demand BAT tables take care of the translation. */ if (translations[i].om_va == translations[i].om_pa) continue; /* Enter the pages */ for (off = 0; off < translations[i].om_len; off += PAGE_SIZE) moea_kenter(mmup, translations[i].om_va + off, translations[i].om_pa + off); } } /* * Calculate the last available physical address. */ for (i = 0; phys_avail[i + 2] != 0; i += 2) ; Maxmem = powerpc_btop(phys_avail[i + 1]); moea_cpu_bootstrap(mmup,0); pmap_bootstrapped++; /* * Set the start and end of kva. */ virtual_avail = VM_MIN_KERNEL_ADDRESS; virtual_end = VM_MAX_SAFE_KERNEL_ADDRESS; /* * Allocate a kernel stack with a guard page for thread0 and map it * into the kernel page map. */ pa = moea_bootstrap_alloc(KSTACK_PAGES * PAGE_SIZE, PAGE_SIZE); va = virtual_avail + KSTACK_GUARD_PAGES * PAGE_SIZE; virtual_avail = va + KSTACK_PAGES * PAGE_SIZE; CTR2(KTR_PMAP, "moea_bootstrap: kstack0 at %#x (%#x)", pa, va); thread0.td_kstack = va; thread0.td_kstack_pages = KSTACK_PAGES; for (i = 0; i < KSTACK_PAGES; i++) { moea_kenter(mmup, va, pa); pa += PAGE_SIZE; va += PAGE_SIZE; } /* * Allocate virtual address space for the message buffer. */ pa = msgbuf_phys = moea_bootstrap_alloc(msgbufsize, PAGE_SIZE); msgbufp = (struct msgbuf *)virtual_avail; va = virtual_avail; virtual_avail += round_page(msgbufsize); while (va < virtual_avail) { moea_kenter(mmup, va, pa); pa += PAGE_SIZE; va += PAGE_SIZE; } /* * Allocate virtual address space for the dynamic percpu area. */ pa = moea_bootstrap_alloc(DPCPU_SIZE, PAGE_SIZE); dpcpu = (void *)virtual_avail; va = virtual_avail; virtual_avail += DPCPU_SIZE; while (va < virtual_avail) { moea_kenter(mmup, va, pa); pa += PAGE_SIZE; va += PAGE_SIZE; } dpcpu_init(dpcpu, 0); } /* * Activate a user pmap. The pmap must be activated before it's address * space can be accessed in any way. */ void moea_activate(mmu_t mmu, struct thread *td) { pmap_t pm, pmr; /* * Load all the data we need up front to encourage the compiler to * not issue any loads while we have interrupts disabled below. */ pm = &td->td_proc->p_vmspace->vm_pmap; pmr = pm->pmap_phys; CPU_SET(PCPU_GET(cpuid), &pm->pm_active); PCPU_SET(curpmap, pmr); + + mtsrin(USER_SR << ADDR_SR_SHFT, td->td_pcb->pcb_cpu.aim.usr_vsid); } void moea_deactivate(mmu_t mmu, struct thread *td) { pmap_t pm; pm = &td->td_proc->p_vmspace->vm_pmap; CPU_CLR(PCPU_GET(cpuid), &pm->pm_active); PCPU_SET(curpmap, NULL); } void moea_unwire(mmu_t mmu, pmap_t pm, vm_offset_t sva, vm_offset_t eva) { struct pvo_entry key, *pvo; PMAP_LOCK(pm); key.pvo_vaddr = sva; for (pvo = RB_NFIND(pvo_tree, &pm->pmap_pvo, &key); pvo != NULL && PVO_VADDR(pvo) < eva; pvo = RB_NEXT(pvo_tree, &pm->pmap_pvo, pvo)) { if ((pvo->pvo_vaddr & PVO_WIRED) == 0) panic("moea_unwire: pvo %p is missing PVO_WIRED", pvo); pvo->pvo_vaddr &= ~PVO_WIRED; pm->pm_stats.wired_count--; } PMAP_UNLOCK(pm); } void moea_copy_page(mmu_t mmu, vm_page_t msrc, vm_page_t mdst) { vm_offset_t dst; vm_offset_t src; dst = VM_PAGE_TO_PHYS(mdst); src = VM_PAGE_TO_PHYS(msrc); bcopy((void *)src, (void *)dst, PAGE_SIZE); } void moea_copy_pages(mmu_t mmu, vm_page_t *ma, vm_offset_t a_offset, vm_page_t *mb, vm_offset_t b_offset, int xfersize) { void *a_cp, *b_cp; vm_offset_t a_pg_offset, b_pg_offset; int cnt; while (xfersize > 0) { a_pg_offset = a_offset & PAGE_MASK; cnt = min(xfersize, PAGE_SIZE - a_pg_offset); a_cp = (char *)VM_PAGE_TO_PHYS(ma[a_offset >> PAGE_SHIFT]) + a_pg_offset; b_pg_offset = b_offset & PAGE_MASK; cnt = min(cnt, PAGE_SIZE - b_pg_offset); b_cp = (char *)VM_PAGE_TO_PHYS(mb[b_offset >> PAGE_SHIFT]) + b_pg_offset; bcopy(a_cp, b_cp, cnt); a_offset += cnt; b_offset += cnt; xfersize -= cnt; } } /* * Zero a page of physical memory by temporarily mapping it into the tlb. */ void moea_zero_page(mmu_t mmu, vm_page_t m) { vm_offset_t off, pa = VM_PAGE_TO_PHYS(m); for (off = 0; off < PAGE_SIZE; off += cacheline_size) __asm __volatile("dcbz 0,%0" :: "r"(pa + off)); } void moea_zero_page_area(mmu_t mmu, vm_page_t m, int off, int size) { vm_offset_t pa = VM_PAGE_TO_PHYS(m); void *va = (void *)(pa + off); bzero(va, size); } void moea_zero_page_idle(mmu_t mmu, vm_page_t m) { moea_zero_page(mmu, m); } /* * Map the given physical page at the specified virtual address in the * target pmap with the protection requested. If specified the page * will be wired down. */ int moea_enter(mmu_t mmu, pmap_t pmap, vm_offset_t va, vm_page_t m, vm_prot_t prot, u_int flags, int8_t psind) { int error; for (;;) { rw_wlock(&pvh_global_lock); PMAP_LOCK(pmap); error = moea_enter_locked(pmap, va, m, prot, flags, psind); rw_wunlock(&pvh_global_lock); PMAP_UNLOCK(pmap); if (error != ENOMEM) return (KERN_SUCCESS); if ((flags & PMAP_ENTER_NOSLEEP) != 0) return (KERN_RESOURCE_SHORTAGE); VM_OBJECT_ASSERT_UNLOCKED(m->object); VM_WAIT; } } /* * Map the given physical page at the specified virtual address in the * target pmap with the protection requested. If specified the page * will be wired down. * * The global pvh and pmap must be locked. */ static int moea_enter_locked(pmap_t pmap, vm_offset_t va, vm_page_t m, vm_prot_t prot, u_int flags, int8_t psind __unused) { struct pvo_head *pvo_head; uma_zone_t zone; u_int pte_lo, pvo_flags; int error; if (pmap_bootstrapped) rw_assert(&pvh_global_lock, RA_WLOCKED); PMAP_LOCK_ASSERT(pmap, MA_OWNED); if ((m->oflags & VPO_UNMANAGED) == 0 && !vm_page_xbusied(m)) VM_OBJECT_ASSERT_LOCKED(m->object); if ((m->oflags & VPO_UNMANAGED) != 0 || !moea_initialized) { pvo_head = &moea_pvo_kunmanaged; zone = moea_upvo_zone; pvo_flags = 0; } else { pvo_head = vm_page_to_pvoh(m); zone = moea_mpvo_zone; pvo_flags = PVO_MANAGED; } pte_lo = moea_calc_wimg(VM_PAGE_TO_PHYS(m), pmap_page_get_memattr(m)); if (prot & VM_PROT_WRITE) { pte_lo |= PTE_BW; if (pmap_bootstrapped && (m->oflags & VPO_UNMANAGED) == 0) vm_page_aflag_set(m, PGA_WRITEABLE); } else pte_lo |= PTE_BR; if ((flags & PMAP_ENTER_WIRED) != 0) pvo_flags |= PVO_WIRED; error = moea_pvo_enter(pmap, zone, pvo_head, va, VM_PAGE_TO_PHYS(m), pte_lo, pvo_flags); /* * Flush the real page from the instruction cache. This has be done * for all user mappings to prevent information leakage via the * instruction cache. moea_pvo_enter() returns ENOENT for the first * mapping for a page. */ if (pmap != kernel_pmap && error == ENOENT && (pte_lo & (PTE_I | PTE_G)) == 0) moea_syncicache(VM_PAGE_TO_PHYS(m), PAGE_SIZE); return (error); } /* * Maps a sequence of resident pages belonging to the same object. * The sequence begins with the given page m_start. This page is * mapped at the given virtual address start. Each subsequent page is * mapped at a virtual address that is offset from start by the same * amount as the page is offset from m_start within the object. The * last page in the sequence is the page with the largest offset from * m_start that can be mapped at a virtual address less than the given * virtual address end. Not every virtual page between start and end * is mapped; only those for which a resident page exists with the * corresponding offset from m_start are mapped. */ void moea_enter_object(mmu_t mmu, pmap_t pm, vm_offset_t start, vm_offset_t end, vm_page_t m_start, vm_prot_t prot) { vm_page_t m; vm_pindex_t diff, psize; VM_OBJECT_ASSERT_LOCKED(m_start->object); psize = atop(end - start); m = m_start; rw_wlock(&pvh_global_lock); PMAP_LOCK(pm); while (m != NULL && (diff = m->pindex - m_start->pindex) < psize) { moea_enter_locked(pm, start + ptoa(diff), m, prot & (VM_PROT_READ | VM_PROT_EXECUTE), 0, 0); m = TAILQ_NEXT(m, listq); } rw_wunlock(&pvh_global_lock); PMAP_UNLOCK(pm); } void moea_enter_quick(mmu_t mmu, pmap_t pm, vm_offset_t va, vm_page_t m, vm_prot_t prot) { rw_wlock(&pvh_global_lock); PMAP_LOCK(pm); moea_enter_locked(pm, va, m, prot & (VM_PROT_READ | VM_PROT_EXECUTE), 0, 0); rw_wunlock(&pvh_global_lock); PMAP_UNLOCK(pm); } vm_paddr_t moea_extract(mmu_t mmu, pmap_t pm, vm_offset_t va) { struct pvo_entry *pvo; vm_paddr_t pa; PMAP_LOCK(pm); pvo = moea_pvo_find_va(pm, va & ~ADDR_POFF, NULL); if (pvo == NULL) pa = 0; else pa = (pvo->pvo_pte.pte.pte_lo & PTE_RPGN) | (va & ADDR_POFF); PMAP_UNLOCK(pm); return (pa); } /* * Atomically extract and hold the physical page with the given * pmap and virtual address pair if that mapping permits the given * protection. */ vm_page_t moea_extract_and_hold(mmu_t mmu, pmap_t pmap, vm_offset_t va, vm_prot_t prot) { struct pvo_entry *pvo; vm_page_t m; vm_paddr_t pa; m = NULL; pa = 0; PMAP_LOCK(pmap); retry: pvo = moea_pvo_find_va(pmap, va & ~ADDR_POFF, NULL); if (pvo != NULL && (pvo->pvo_pte.pte.pte_hi & PTE_VALID) && ((pvo->pvo_pte.pte.pte_lo & PTE_PP) == PTE_RW || (prot & VM_PROT_WRITE) == 0)) { if (vm_page_pa_tryrelock(pmap, pvo->pvo_pte.pte.pte_lo & PTE_RPGN, &pa)) goto retry; m = PHYS_TO_VM_PAGE(pvo->pvo_pte.pte.pte_lo & PTE_RPGN); vm_page_hold(m); } PA_UNLOCK_COND(pa); PMAP_UNLOCK(pmap); return (m); } void moea_init(mmu_t mmu) { moea_upvo_zone = uma_zcreate("UPVO entry", sizeof (struct pvo_entry), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_VM | UMA_ZONE_NOFREE); moea_mpvo_zone = uma_zcreate("MPVO entry", sizeof(struct pvo_entry), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_VM | UMA_ZONE_NOFREE); moea_initialized = TRUE; } boolean_t moea_is_referenced(mmu_t mmu, vm_page_t m) { boolean_t rv; KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("moea_is_referenced: page %p is not managed", m)); rw_wlock(&pvh_global_lock); rv = moea_query_bit(m, PTE_REF); rw_wunlock(&pvh_global_lock); return (rv); } boolean_t moea_is_modified(mmu_t mmu, vm_page_t m) { boolean_t rv; KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("moea_is_modified: page %p is not managed", m)); /* * If the page is not exclusive busied, then PGA_WRITEABLE cannot be * concurrently set while the object is locked. Thus, if PGA_WRITEABLE * is clear, no PTEs can have PTE_CHG set. */ VM_OBJECT_ASSERT_WLOCKED(m->object); if (!vm_page_xbusied(m) && (m->aflags & PGA_WRITEABLE) == 0) return (FALSE); rw_wlock(&pvh_global_lock); rv = moea_query_bit(m, PTE_CHG); rw_wunlock(&pvh_global_lock); return (rv); } boolean_t moea_is_prefaultable(mmu_t mmu, pmap_t pmap, vm_offset_t va) { struct pvo_entry *pvo; boolean_t rv; PMAP_LOCK(pmap); pvo = moea_pvo_find_va(pmap, va & ~ADDR_POFF, NULL); rv = pvo == NULL || (pvo->pvo_pte.pte.pte_hi & PTE_VALID) == 0; PMAP_UNLOCK(pmap); return (rv); } void moea_clear_modify(mmu_t mmu, vm_page_t m) { KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("moea_clear_modify: page %p is not managed", m)); VM_OBJECT_ASSERT_WLOCKED(m->object); KASSERT(!vm_page_xbusied(m), ("moea_clear_modify: page %p is exclusive busy", m)); /* * If the page is not PGA_WRITEABLE, then no PTEs can have PTE_CHG * set. If the object containing the page is locked and the page is * not exclusive busied, then PGA_WRITEABLE cannot be concurrently set. */ if ((m->aflags & PGA_WRITEABLE) == 0) return; rw_wlock(&pvh_global_lock); moea_clear_bit(m, PTE_CHG); rw_wunlock(&pvh_global_lock); } /* * Clear the write and modified bits in each of the given page's mappings. */ void moea_remove_write(mmu_t mmu, vm_page_t m) { struct pvo_entry *pvo; struct pte *pt; pmap_t pmap; u_int lo; KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("moea_remove_write: page %p is not managed", m)); /* * If the page is not exclusive busied, then PGA_WRITEABLE cannot be * set by another thread while the object is locked. Thus, * if PGA_WRITEABLE is clear, no page table entries need updating. */ VM_OBJECT_ASSERT_WLOCKED(m->object); if (!vm_page_xbusied(m) && (m->aflags & PGA_WRITEABLE) == 0) return; rw_wlock(&pvh_global_lock); lo = moea_attr_fetch(m); powerpc_sync(); LIST_FOREACH(pvo, vm_page_to_pvoh(m), pvo_vlink) { pmap = pvo->pvo_pmap; PMAP_LOCK(pmap); if ((pvo->pvo_pte.pte.pte_lo & PTE_PP) != PTE_BR) { pt = moea_pvo_to_pte(pvo, -1); pvo->pvo_pte.pte.pte_lo &= ~PTE_PP; pvo->pvo_pte.pte.pte_lo |= PTE_BR; if (pt != NULL) { moea_pte_synch(pt, &pvo->pvo_pte.pte); lo |= pvo->pvo_pte.pte.pte_lo; pvo->pvo_pte.pte.pte_lo &= ~PTE_CHG; moea_pte_change(pt, &pvo->pvo_pte.pte, pvo->pvo_vaddr); mtx_unlock(&moea_table_mutex); } } PMAP_UNLOCK(pmap); } if ((lo & PTE_CHG) != 0) { moea_attr_clear(m, PTE_CHG); vm_page_dirty(m); } vm_page_aflag_clear(m, PGA_WRITEABLE); rw_wunlock(&pvh_global_lock); } /* * moea_ts_referenced: * * Return a count of reference bits for a page, clearing those bits. * It is not necessary for every reference bit to be cleared, but it * is necessary that 0 only be returned when there are truly no * reference bits set. * * XXX: The exact number of bits to check and clear is a matter that * should be tested and standardized at some point in the future for * optimal aging of shared pages. */ int moea_ts_referenced(mmu_t mmu, vm_page_t m) { int count; KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("moea_ts_referenced: page %p is not managed", m)); rw_wlock(&pvh_global_lock); count = moea_clear_bit(m, PTE_REF); rw_wunlock(&pvh_global_lock); return (count); } /* * Modify the WIMG settings of all mappings for a page. */ void moea_page_set_memattr(mmu_t mmu, vm_page_t m, vm_memattr_t ma) { struct pvo_entry *pvo; struct pvo_head *pvo_head; struct pte *pt; pmap_t pmap; u_int lo; if ((m->oflags & VPO_UNMANAGED) != 0) { m->md.mdpg_cache_attrs = ma; return; } rw_wlock(&pvh_global_lock); pvo_head = vm_page_to_pvoh(m); lo = moea_calc_wimg(VM_PAGE_TO_PHYS(m), ma); LIST_FOREACH(pvo, pvo_head, pvo_vlink) { pmap = pvo->pvo_pmap; PMAP_LOCK(pmap); pt = moea_pvo_to_pte(pvo, -1); pvo->pvo_pte.pte.pte_lo &= ~PTE_WIMG; pvo->pvo_pte.pte.pte_lo |= lo; if (pt != NULL) { moea_pte_change(pt, &pvo->pvo_pte.pte, pvo->pvo_vaddr); if (pvo->pvo_pmap == kernel_pmap) isync(); } mtx_unlock(&moea_table_mutex); PMAP_UNLOCK(pmap); } m->md.mdpg_cache_attrs = ma; rw_wunlock(&pvh_global_lock); } /* * Map a wired page into kernel virtual address space. */ void moea_kenter(mmu_t mmu, vm_offset_t va, vm_paddr_t pa) { moea_kenter_attr(mmu, va, pa, VM_MEMATTR_DEFAULT); } void moea_kenter_attr(mmu_t mmu, vm_offset_t va, vm_offset_t pa, vm_memattr_t ma) { u_int pte_lo; int error; #if 0 if (va < VM_MIN_KERNEL_ADDRESS) panic("moea_kenter: attempt to enter non-kernel address %#x", va); #endif pte_lo = moea_calc_wimg(pa, ma); PMAP_LOCK(kernel_pmap); error = moea_pvo_enter(kernel_pmap, moea_upvo_zone, &moea_pvo_kunmanaged, va, pa, pte_lo, PVO_WIRED); if (error != 0 && error != ENOENT) panic("moea_kenter: failed to enter va %#x pa %#x: %d", va, pa, error); PMAP_UNLOCK(kernel_pmap); } /* * Extract the physical page address associated with the given kernel virtual * address. */ vm_paddr_t moea_kextract(mmu_t mmu, vm_offset_t va) { struct pvo_entry *pvo; vm_paddr_t pa; /* * Allow direct mappings on 32-bit OEA */ if (va < VM_MIN_KERNEL_ADDRESS) { return (va); } PMAP_LOCK(kernel_pmap); pvo = moea_pvo_find_va(kernel_pmap, va & ~ADDR_POFF, NULL); KASSERT(pvo != NULL, ("moea_kextract: no addr found")); pa = (pvo->pvo_pte.pte.pte_lo & PTE_RPGN) | (va & ADDR_POFF); PMAP_UNLOCK(kernel_pmap); return (pa); } /* * Remove a wired page from kernel virtual address space. */ void moea_kremove(mmu_t mmu, vm_offset_t va) { moea_remove(mmu, kernel_pmap, va, va + PAGE_SIZE); } /* * Map a range of physical addresses into kernel virtual address space. * * The value passed in *virt is a suggested virtual address for the mapping. * Architectures which can support a direct-mapped physical to virtual region * can return the appropriate address within that region, leaving '*virt' * unchanged. We cannot and therefore do not; *virt is updated with the * first usable address after the mapped region. */ vm_offset_t moea_map(mmu_t mmu, vm_offset_t *virt, vm_paddr_t pa_start, vm_paddr_t pa_end, int prot) { vm_offset_t sva, va; sva = *virt; va = sva; for (; pa_start < pa_end; pa_start += PAGE_SIZE, va += PAGE_SIZE) moea_kenter(mmu, va, pa_start); *virt = va; return (sva); } /* * Returns true if the pmap's pv is one of the first * 16 pvs linked to from this page. This count may * be changed upwards or downwards in the future; it * is only necessary that true be returned for a small * subset of pmaps for proper page aging. */ boolean_t moea_page_exists_quick(mmu_t mmu, pmap_t pmap, vm_page_t m) { int loops; struct pvo_entry *pvo; boolean_t rv; KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("moea_page_exists_quick: page %p is not managed", m)); loops = 0; rv = FALSE; rw_wlock(&pvh_global_lock); LIST_FOREACH(pvo, vm_page_to_pvoh(m), pvo_vlink) { if (pvo->pvo_pmap == pmap) { rv = TRUE; break; } if (++loops >= 16) break; } rw_wunlock(&pvh_global_lock); return (rv); } /* * Return the number of managed mappings to the given physical page * that are wired. */ int moea_page_wired_mappings(mmu_t mmu, vm_page_t m) { struct pvo_entry *pvo; int count; count = 0; if ((m->oflags & VPO_UNMANAGED) != 0) return (count); rw_wlock(&pvh_global_lock); LIST_FOREACH(pvo, vm_page_to_pvoh(m), pvo_vlink) if ((pvo->pvo_vaddr & PVO_WIRED) != 0) count++; rw_wunlock(&pvh_global_lock); return (count); } static u_int moea_vsidcontext; void moea_pinit(mmu_t mmu, pmap_t pmap) { int i, mask; u_int entropy; KASSERT((int)pmap < VM_MIN_KERNEL_ADDRESS, ("moea_pinit: virt pmap")); RB_INIT(&pmap->pmap_pvo); entropy = 0; __asm __volatile("mftb %0" : "=r"(entropy)); if ((pmap->pmap_phys = (pmap_t)moea_kextract(mmu, (vm_offset_t)pmap)) == NULL) { pmap->pmap_phys = pmap; } mtx_lock(&moea_vsid_mutex); /* * Allocate some segment registers for this pmap. */ for (i = 0; i < NPMAPS; i += VSID_NBPW) { u_int hash, n; /* * Create a new value by mutiplying by a prime and adding in * entropy from the timebase register. This is to make the * VSID more random so that the PT hash function collides * less often. (Note that the prime casues gcc to do shifts * instead of a multiply.) */ moea_vsidcontext = (moea_vsidcontext * 0x1105) + entropy; hash = moea_vsidcontext & (NPMAPS - 1); if (hash == 0) /* 0 is special, avoid it */ continue; n = hash >> 5; mask = 1 << (hash & (VSID_NBPW - 1)); hash = (moea_vsidcontext & 0xfffff); if (moea_vsid_bitmap[n] & mask) { /* collision? */ /* anything free in this bucket? */ if (moea_vsid_bitmap[n] == 0xffffffff) { entropy = (moea_vsidcontext >> 20); continue; } i = ffs(~moea_vsid_bitmap[n]) - 1; mask = 1 << i; hash &= 0xfffff & ~(VSID_NBPW - 1); hash |= i; } KASSERT(!(moea_vsid_bitmap[n] & mask), ("Allocating in-use VSID group %#x\n", hash)); moea_vsid_bitmap[n] |= mask; for (i = 0; i < 16; i++) pmap->pm_sr[i] = VSID_MAKE(i, hash); mtx_unlock(&moea_vsid_mutex); return; } mtx_unlock(&moea_vsid_mutex); panic("moea_pinit: out of segments"); } /* * Initialize the pmap associated with process 0. */ void moea_pinit0(mmu_t mmu, pmap_t pm) { PMAP_LOCK_INIT(pm); moea_pinit(mmu, pm); bzero(&pm->pm_stats, sizeof(pm->pm_stats)); } /* * Set the physical protection on the specified range of this map as requested. */ void moea_protect(mmu_t mmu, pmap_t pm, vm_offset_t sva, vm_offset_t eva, vm_prot_t prot) { struct pvo_entry *pvo, *tpvo, key; struct pte *pt; KASSERT(pm == &curproc->p_vmspace->vm_pmap || pm == kernel_pmap, ("moea_protect: non current pmap")); if ((prot & VM_PROT_READ) == VM_PROT_NONE) { moea_remove(mmu, pm, sva, eva); return; } rw_wlock(&pvh_global_lock); PMAP_LOCK(pm); key.pvo_vaddr = sva; for (pvo = RB_NFIND(pvo_tree, &pm->pmap_pvo, &key); pvo != NULL && PVO_VADDR(pvo) < eva; pvo = tpvo) { tpvo = RB_NEXT(pvo_tree, &pm->pmap_pvo, pvo); /* * Grab the PTE pointer before we diddle with the cached PTE * copy. */ pt = moea_pvo_to_pte(pvo, -1); /* * Change the protection of the page. */ pvo->pvo_pte.pte.pte_lo &= ~PTE_PP; pvo->pvo_pte.pte.pte_lo |= PTE_BR; /* * If the PVO is in the page table, update that pte as well. */ if (pt != NULL) { moea_pte_change(pt, &pvo->pvo_pte.pte, pvo->pvo_vaddr); mtx_unlock(&moea_table_mutex); } } rw_wunlock(&pvh_global_lock); PMAP_UNLOCK(pm); } /* * Map a list of wired pages into kernel virtual address space. This is * intended for temporary mappings which do not need page modification or * references recorded. Existing mappings in the region are overwritten. */ void moea_qenter(mmu_t mmu, vm_offset_t sva, vm_page_t *m, int count) { vm_offset_t va; va = sva; while (count-- > 0) { moea_kenter(mmu, va, VM_PAGE_TO_PHYS(*m)); va += PAGE_SIZE; m++; } } /* * Remove page mappings from kernel virtual address space. Intended for * temporary mappings entered by moea_qenter. */ void moea_qremove(mmu_t mmu, vm_offset_t sva, int count) { vm_offset_t va; va = sva; while (count-- > 0) { moea_kremove(mmu, va); va += PAGE_SIZE; } } void moea_release(mmu_t mmu, pmap_t pmap) { int idx, mask; /* * Free segment register's VSID */ if (pmap->pm_sr[0] == 0) panic("moea_release"); mtx_lock(&moea_vsid_mutex); idx = VSID_TO_HASH(pmap->pm_sr[0]) & (NPMAPS-1); mask = 1 << (idx % VSID_NBPW); idx /= VSID_NBPW; moea_vsid_bitmap[idx] &= ~mask; mtx_unlock(&moea_vsid_mutex); } /* * Remove the given range of addresses from the specified map. */ void moea_remove(mmu_t mmu, pmap_t pm, vm_offset_t sva, vm_offset_t eva) { struct pvo_entry *pvo, *tpvo, key; rw_wlock(&pvh_global_lock); PMAP_LOCK(pm); key.pvo_vaddr = sva; for (pvo = RB_NFIND(pvo_tree, &pm->pmap_pvo, &key); pvo != NULL && PVO_VADDR(pvo) < eva; pvo = tpvo) { tpvo = RB_NEXT(pvo_tree, &pm->pmap_pvo, pvo); moea_pvo_remove(pvo, -1); } PMAP_UNLOCK(pm); rw_wunlock(&pvh_global_lock); } /* * Remove physical page from all pmaps in which it resides. moea_pvo_remove() * will reflect changes in pte's back to the vm_page. */ void moea_remove_all(mmu_t mmu, vm_page_t m) { struct pvo_head *pvo_head; struct pvo_entry *pvo, *next_pvo; pmap_t pmap; rw_wlock(&pvh_global_lock); pvo_head = vm_page_to_pvoh(m); for (pvo = LIST_FIRST(pvo_head); pvo != NULL; pvo = next_pvo) { next_pvo = LIST_NEXT(pvo, pvo_vlink); pmap = pvo->pvo_pmap; PMAP_LOCK(pmap); moea_pvo_remove(pvo, -1); PMAP_UNLOCK(pmap); } if ((m->aflags & PGA_WRITEABLE) && moea_query_bit(m, PTE_CHG)) { moea_attr_clear(m, PTE_CHG); vm_page_dirty(m); } vm_page_aflag_clear(m, PGA_WRITEABLE); rw_wunlock(&pvh_global_lock); } /* * Allocate a physical page of memory directly from the phys_avail map. * Can only be called from moea_bootstrap before avail start and end are * calculated. */ static vm_offset_t moea_bootstrap_alloc(vm_size_t size, u_int align) { vm_offset_t s, e; int i, j; size = round_page(size); for (i = 0; phys_avail[i + 1] != 0; i += 2) { if (align != 0) s = (phys_avail[i] + align - 1) & ~(align - 1); else s = phys_avail[i]; e = s + size; if (s < phys_avail[i] || e > phys_avail[i + 1]) continue; if (s == phys_avail[i]) { phys_avail[i] += size; } else if (e == phys_avail[i + 1]) { phys_avail[i + 1] -= size; } else { for (j = phys_avail_count * 2; j > i; j -= 2) { phys_avail[j] = phys_avail[j - 2]; phys_avail[j + 1] = phys_avail[j - 1]; } phys_avail[i + 3] = phys_avail[i + 1]; phys_avail[i + 1] = s; phys_avail[i + 2] = e; phys_avail_count++; } return (s); } panic("moea_bootstrap_alloc: could not allocate memory"); } static void moea_syncicache(vm_offset_t pa, vm_size_t len) { __syncicache((void *)pa, len); } static int moea_pvo_enter(pmap_t pm, uma_zone_t zone, struct pvo_head *pvo_head, vm_offset_t va, vm_offset_t pa, u_int pte_lo, int flags) { struct pvo_entry *pvo; u_int sr; int first; u_int ptegidx; int i; int bootstrap; moea_pvo_enter_calls++; first = 0; bootstrap = 0; /* * Compute the PTE Group index. */ va &= ~ADDR_POFF; sr = va_to_sr(pm->pm_sr, va); ptegidx = va_to_pteg(sr, va); /* * Remove any existing mapping for this page. Reuse the pvo entry if * there is a mapping. */ mtx_lock(&moea_table_mutex); LIST_FOREACH(pvo, &moea_pvo_table[ptegidx], pvo_olink) { if (pvo->pvo_pmap == pm && PVO_VADDR(pvo) == va) { if ((pvo->pvo_pte.pte.pte_lo & PTE_RPGN) == pa && (pvo->pvo_pte.pte.pte_lo & PTE_PP) == (pte_lo & PTE_PP)) { /* * The PTE is not changing. Instead, this may * be a request to change the mapping's wired * attribute. */ mtx_unlock(&moea_table_mutex); if ((flags & PVO_WIRED) != 0 && (pvo->pvo_vaddr & PVO_WIRED) == 0) { pvo->pvo_vaddr |= PVO_WIRED; pm->pm_stats.wired_count++; } else if ((flags & PVO_WIRED) == 0 && (pvo->pvo_vaddr & PVO_WIRED) != 0) { pvo->pvo_vaddr &= ~PVO_WIRED; pm->pm_stats.wired_count--; } return (0); } moea_pvo_remove(pvo, -1); break; } } /* * If we aren't overwriting a mapping, try to allocate. */ if (moea_initialized) { pvo = uma_zalloc(zone, M_NOWAIT); } else { if (moea_bpvo_pool_index >= BPVO_POOL_SIZE) { panic("moea_enter: bpvo pool exhausted, %d, %d, %d", moea_bpvo_pool_index, BPVO_POOL_SIZE, BPVO_POOL_SIZE * sizeof(struct pvo_entry)); } pvo = &moea_bpvo_pool[moea_bpvo_pool_index]; moea_bpvo_pool_index++; bootstrap = 1; } if (pvo == NULL) { mtx_unlock(&moea_table_mutex); return (ENOMEM); } moea_pvo_entries++; pvo->pvo_vaddr = va; pvo->pvo_pmap = pm; LIST_INSERT_HEAD(&moea_pvo_table[ptegidx], pvo, pvo_olink); pvo->pvo_vaddr &= ~ADDR_POFF; if (flags & PVO_WIRED) pvo->pvo_vaddr |= PVO_WIRED; if (pvo_head != &moea_pvo_kunmanaged) pvo->pvo_vaddr |= PVO_MANAGED; if (bootstrap) pvo->pvo_vaddr |= PVO_BOOTSTRAP; moea_pte_create(&pvo->pvo_pte.pte, sr, va, pa | pte_lo); /* * Add to pmap list */ RB_INSERT(pvo_tree, &pm->pmap_pvo, pvo); /* * Remember if the list was empty and therefore will be the first * item. */ if (LIST_FIRST(pvo_head) == NULL) first = 1; LIST_INSERT_HEAD(pvo_head, pvo, pvo_vlink); if (pvo->pvo_vaddr & PVO_WIRED) pm->pm_stats.wired_count++; pm->pm_stats.resident_count++; i = moea_pte_insert(ptegidx, &pvo->pvo_pte.pte); KASSERT(i < 8, ("Invalid PTE index")); if (i >= 0) { PVO_PTEGIDX_SET(pvo, i); } else { panic("moea_pvo_enter: overflow"); moea_pte_overflow++; } mtx_unlock(&moea_table_mutex); return (first ? ENOENT : 0); } static void moea_pvo_remove(struct pvo_entry *pvo, int pteidx) { struct pte *pt; /* * If there is an active pte entry, we need to deactivate it (and * save the ref & cfg bits). */ pt = moea_pvo_to_pte(pvo, pteidx); if (pt != NULL) { moea_pte_unset(pt, &pvo->pvo_pte.pte, pvo->pvo_vaddr); mtx_unlock(&moea_table_mutex); PVO_PTEGIDX_CLR(pvo); } else { moea_pte_overflow--; } /* * Update our statistics. */ pvo->pvo_pmap->pm_stats.resident_count--; if (pvo->pvo_vaddr & PVO_WIRED) pvo->pvo_pmap->pm_stats.wired_count--; /* * Save the REF/CHG bits into their cache if the page is managed. */ if ((pvo->pvo_vaddr & PVO_MANAGED) == PVO_MANAGED) { struct vm_page *pg; pg = PHYS_TO_VM_PAGE(pvo->pvo_pte.pte.pte_lo & PTE_RPGN); if (pg != NULL) { moea_attr_save(pg, pvo->pvo_pte.pte.pte_lo & (PTE_REF | PTE_CHG)); } } /* * Remove this PVO from the PV and pmap lists. */ LIST_REMOVE(pvo, pvo_vlink); RB_REMOVE(pvo_tree, &pvo->pvo_pmap->pmap_pvo, pvo); /* * Remove this from the overflow list and return it to the pool * if we aren't going to reuse it. */ LIST_REMOVE(pvo, pvo_olink); if (!(pvo->pvo_vaddr & PVO_BOOTSTRAP)) uma_zfree(pvo->pvo_vaddr & PVO_MANAGED ? moea_mpvo_zone : moea_upvo_zone, pvo); moea_pvo_entries--; moea_pvo_remove_calls++; } static __inline int moea_pvo_pte_index(const struct pvo_entry *pvo, int ptegidx) { int pteidx; /* * We can find the actual pte entry without searching by grabbing * the PTEG index from 3 unused bits in pte_lo[11:9] and by * noticing the HID bit. */ pteidx = ptegidx * 8 + PVO_PTEGIDX_GET(pvo); if (pvo->pvo_pte.pte.pte_hi & PTE_HID) pteidx ^= moea_pteg_mask * 8; return (pteidx); } static struct pvo_entry * moea_pvo_find_va(pmap_t pm, vm_offset_t va, int *pteidx_p) { struct pvo_entry *pvo; int ptegidx; u_int sr; va &= ~ADDR_POFF; sr = va_to_sr(pm->pm_sr, va); ptegidx = va_to_pteg(sr, va); mtx_lock(&moea_table_mutex); LIST_FOREACH(pvo, &moea_pvo_table[ptegidx], pvo_olink) { if (pvo->pvo_pmap == pm && PVO_VADDR(pvo) == va) { if (pteidx_p) *pteidx_p = moea_pvo_pte_index(pvo, ptegidx); break; } } mtx_unlock(&moea_table_mutex); return (pvo); } static struct pte * moea_pvo_to_pte(const struct pvo_entry *pvo, int pteidx) { struct pte *pt; /* * If we haven't been supplied the ptegidx, calculate it. */ if (pteidx == -1) { int ptegidx; u_int sr; sr = va_to_sr(pvo->pvo_pmap->pm_sr, pvo->pvo_vaddr); ptegidx = va_to_pteg(sr, pvo->pvo_vaddr); pteidx = moea_pvo_pte_index(pvo, ptegidx); } pt = &moea_pteg_table[pteidx >> 3].pt[pteidx & 7]; mtx_lock(&moea_table_mutex); if ((pvo->pvo_pte.pte.pte_hi & PTE_VALID) && !PVO_PTEGIDX_ISSET(pvo)) { panic("moea_pvo_to_pte: pvo %p has valid pte in pvo but no " "valid pte index", pvo); } if ((pvo->pvo_pte.pte.pte_hi & PTE_VALID) == 0 && PVO_PTEGIDX_ISSET(pvo)) { panic("moea_pvo_to_pte: pvo %p has valid pte index in pvo " "pvo but no valid pte", pvo); } if ((pt->pte_hi ^ (pvo->pvo_pte.pte.pte_hi & ~PTE_VALID)) == PTE_VALID) { if ((pvo->pvo_pte.pte.pte_hi & PTE_VALID) == 0) { panic("moea_pvo_to_pte: pvo %p has valid pte in " "moea_pteg_table %p but invalid in pvo", pvo, pt); } if (((pt->pte_lo ^ pvo->pvo_pte.pte.pte_lo) & ~(PTE_CHG|PTE_REF)) != 0) { panic("moea_pvo_to_pte: pvo %p pte does not match " "pte %p in moea_pteg_table", pvo, pt); } mtx_assert(&moea_table_mutex, MA_OWNED); return (pt); } if (pvo->pvo_pte.pte.pte_hi & PTE_VALID) { panic("moea_pvo_to_pte: pvo %p has invalid pte %p in " "moea_pteg_table but valid in pvo: %8x, %8x", pvo, pt, pvo->pvo_pte.pte.pte_hi, pt->pte_hi); } mtx_unlock(&moea_table_mutex); return (NULL); } /* * XXX: THIS STUFF SHOULD BE IN pte.c? */ int moea_pte_spill(vm_offset_t addr) { struct pvo_entry *source_pvo, *victim_pvo; struct pvo_entry *pvo; int ptegidx, i, j; u_int sr; struct pteg *pteg; struct pte *pt; moea_pte_spills++; sr = mfsrin(addr); ptegidx = va_to_pteg(sr, addr); /* * Have to substitute some entry. Use the primary hash for this. * Use low bits of timebase as random generator. */ pteg = &moea_pteg_table[ptegidx]; mtx_lock(&moea_table_mutex); __asm __volatile("mftb %0" : "=r"(i)); i &= 7; pt = &pteg->pt[i]; source_pvo = NULL; victim_pvo = NULL; LIST_FOREACH(pvo, &moea_pvo_table[ptegidx], pvo_olink) { /* * We need to find a pvo entry for this address. */ if (source_pvo == NULL && moea_pte_match(&pvo->pvo_pte.pte, sr, addr, pvo->pvo_pte.pte.pte_hi & PTE_HID)) { /* * Now found an entry to be spilled into the pteg. * The PTE is now valid, so we know it's active. */ j = moea_pte_insert(ptegidx, &pvo->pvo_pte.pte); if (j >= 0) { PVO_PTEGIDX_SET(pvo, j); moea_pte_overflow--; mtx_unlock(&moea_table_mutex); return (1); } source_pvo = pvo; if (victim_pvo != NULL) break; } /* * We also need the pvo entry of the victim we are replacing * so save the R & C bits of the PTE. */ if ((pt->pte_hi & PTE_HID) == 0 && victim_pvo == NULL && moea_pte_compare(pt, &pvo->pvo_pte.pte)) { victim_pvo = pvo; if (source_pvo != NULL) break; } } if (source_pvo == NULL) { mtx_unlock(&moea_table_mutex); return (0); } if (victim_pvo == NULL) { if ((pt->pte_hi & PTE_HID) == 0) panic("moea_pte_spill: victim p-pte (%p) has no pvo" "entry", pt); /* * If this is a secondary PTE, we need to search it's primary * pvo bucket for the matching PVO. */ LIST_FOREACH(pvo, &moea_pvo_table[ptegidx ^ moea_pteg_mask], pvo_olink) { /* * We also need the pvo entry of the victim we are * replacing so save the R & C bits of the PTE. */ if (moea_pte_compare(pt, &pvo->pvo_pte.pte)) { victim_pvo = pvo; break; } } if (victim_pvo == NULL) panic("moea_pte_spill: victim s-pte (%p) has no pvo" "entry", pt); } /* * We are invalidating the TLB entry for the EA we are replacing even * though it's valid. If we don't, we lose any ref/chg bit changes * contained in the TLB entry. */ source_pvo->pvo_pte.pte.pte_hi &= ~PTE_HID; moea_pte_unset(pt, &victim_pvo->pvo_pte.pte, victim_pvo->pvo_vaddr); moea_pte_set(pt, &source_pvo->pvo_pte.pte); PVO_PTEGIDX_CLR(victim_pvo); PVO_PTEGIDX_SET(source_pvo, i); moea_pte_replacements++; mtx_unlock(&moea_table_mutex); return (1); } static __inline struct pvo_entry * moea_pte_spillable_ident(u_int ptegidx) { struct pte *pt; struct pvo_entry *pvo_walk, *pvo = NULL; LIST_FOREACH(pvo_walk, &moea_pvo_table[ptegidx], pvo_olink) { if (pvo_walk->pvo_vaddr & PVO_WIRED) continue; if (!(pvo_walk->pvo_pte.pte.pte_hi & PTE_VALID)) continue; pt = moea_pvo_to_pte(pvo_walk, -1); if (pt == NULL) continue; pvo = pvo_walk; mtx_unlock(&moea_table_mutex); if (!(pt->pte_lo & PTE_REF)) return (pvo_walk); } return (pvo); } static int moea_pte_insert(u_int ptegidx, struct pte *pvo_pt) { struct pte *pt; struct pvo_entry *victim_pvo; int i; int victim_idx; u_int pteg_bkpidx = ptegidx; mtx_assert(&moea_table_mutex, MA_OWNED); /* * First try primary hash. */ for (pt = moea_pteg_table[ptegidx].pt, i = 0; i < 8; i++, pt++) { if ((pt->pte_hi & PTE_VALID) == 0) { pvo_pt->pte_hi &= ~PTE_HID; moea_pte_set(pt, pvo_pt); return (i); } } /* * Now try secondary hash. */ ptegidx ^= moea_pteg_mask; for (pt = moea_pteg_table[ptegidx].pt, i = 0; i < 8; i++, pt++) { if ((pt->pte_hi & PTE_VALID) == 0) { pvo_pt->pte_hi |= PTE_HID; moea_pte_set(pt, pvo_pt); return (i); } } /* Try again, but this time try to force a PTE out. */ ptegidx = pteg_bkpidx; victim_pvo = moea_pte_spillable_ident(ptegidx); if (victim_pvo == NULL) { ptegidx ^= moea_pteg_mask; victim_pvo = moea_pte_spillable_ident(ptegidx); } if (victim_pvo == NULL) { panic("moea_pte_insert: overflow"); return (-1); } victim_idx = moea_pvo_pte_index(victim_pvo, ptegidx); if (pteg_bkpidx == ptegidx) pvo_pt->pte_hi &= ~PTE_HID; else pvo_pt->pte_hi |= PTE_HID; /* * Synchronize the sacrifice PTE with its PVO, then mark both * invalid. The PVO will be reused when/if the VM system comes * here after a fault. */ pt = &moea_pteg_table[victim_idx >> 3].pt[victim_idx & 7]; if (pt->pte_hi != victim_pvo->pvo_pte.pte.pte_hi) panic("Victim PVO doesn't match PTE! PVO: %8x, PTE: %8x", victim_pvo->pvo_pte.pte.pte_hi, pt->pte_hi); /* * Set the new PTE. */ moea_pte_unset(pt, &victim_pvo->pvo_pte.pte, victim_pvo->pvo_vaddr); PVO_PTEGIDX_CLR(victim_pvo); moea_pte_overflow++; moea_pte_set(pt, pvo_pt); return (victim_idx & 7); } static boolean_t moea_query_bit(vm_page_t m, int ptebit) { struct pvo_entry *pvo; struct pte *pt; rw_assert(&pvh_global_lock, RA_WLOCKED); if (moea_attr_fetch(m) & ptebit) return (TRUE); LIST_FOREACH(pvo, vm_page_to_pvoh(m), pvo_vlink) { /* * See if we saved the bit off. If so, cache it and return * success. */ if (pvo->pvo_pte.pte.pte_lo & ptebit) { moea_attr_save(m, ptebit); return (TRUE); } } /* * No luck, now go through the hard part of looking at the PTEs * themselves. Sync so that any pending REF/CHG bits are flushed to * the PTEs. */ powerpc_sync(); LIST_FOREACH(pvo, vm_page_to_pvoh(m), pvo_vlink) { /* * See if this pvo has a valid PTE. if so, fetch the * REF/CHG bits from the valid PTE. If the appropriate * ptebit is set, cache it and return success. */ pt = moea_pvo_to_pte(pvo, -1); if (pt != NULL) { moea_pte_synch(pt, &pvo->pvo_pte.pte); mtx_unlock(&moea_table_mutex); if (pvo->pvo_pte.pte.pte_lo & ptebit) { moea_attr_save(m, ptebit); return (TRUE); } } } return (FALSE); } static u_int moea_clear_bit(vm_page_t m, int ptebit) { u_int count; struct pvo_entry *pvo; struct pte *pt; rw_assert(&pvh_global_lock, RA_WLOCKED); /* * Clear the cached value. */ moea_attr_clear(m, ptebit); /* * Sync so that any pending REF/CHG bits are flushed to the PTEs (so * we can reset the right ones). note that since the pvo entries and * list heads are accessed via BAT0 and are never placed in the page * table, we don't have to worry about further accesses setting the * REF/CHG bits. */ powerpc_sync(); /* * For each pvo entry, clear the pvo's ptebit. If this pvo has a * valid pte clear the ptebit from the valid pte. */ count = 0; LIST_FOREACH(pvo, vm_page_to_pvoh(m), pvo_vlink) { pt = moea_pvo_to_pte(pvo, -1); if (pt != NULL) { moea_pte_synch(pt, &pvo->pvo_pte.pte); if (pvo->pvo_pte.pte.pte_lo & ptebit) { count++; moea_pte_clear(pt, PVO_VADDR(pvo), ptebit); } mtx_unlock(&moea_table_mutex); } pvo->pvo_pte.pte.pte_lo &= ~ptebit; } return (count); } /* * Return true if the physical range is encompassed by the battable[idx] */ static int moea_bat_mapped(int idx, vm_offset_t pa, vm_size_t size) { u_int prot; u_int32_t start; u_int32_t end; u_int32_t bat_ble; /* * Return immediately if not a valid mapping */ if (!(battable[idx].batu & BAT_Vs)) return (EINVAL); /* * The BAT entry must be cache-inhibited, guarded, and r/w * so it can function as an i/o page */ prot = battable[idx].batl & (BAT_I|BAT_G|BAT_PP_RW); if (prot != (BAT_I|BAT_G|BAT_PP_RW)) return (EPERM); /* * The address should be within the BAT range. Assume that the * start address in the BAT has the correct alignment (thus * not requiring masking) */ start = battable[idx].batl & BAT_PBS; bat_ble = (battable[idx].batu & ~(BAT_EBS)) | 0x03; end = start | (bat_ble << 15) | 0x7fff; if ((pa < start) || ((pa + size) > end)) return (ERANGE); return (0); } boolean_t moea_dev_direct_mapped(mmu_t mmu, vm_paddr_t pa, vm_size_t size) { int i; /* * This currently does not work for entries that * overlap 256M BAT segments. */ for(i = 0; i < 16; i++) if (moea_bat_mapped(i, pa, size) == 0) return (0); return (EFAULT); } /* * Map a set of physical memory pages into the kernel virtual * address space. Return a pointer to where it is mapped. This * routine is intended to be used for mapping device memory, * NOT real memory. */ void * moea_mapdev(mmu_t mmu, vm_paddr_t pa, vm_size_t size) { return (moea_mapdev_attr(mmu, pa, size, VM_MEMATTR_DEFAULT)); } void * moea_mapdev_attr(mmu_t mmu, vm_offset_t pa, vm_size_t size, vm_memattr_t ma) { vm_offset_t va, tmpva, ppa, offset; int i; ppa = trunc_page(pa); offset = pa & PAGE_MASK; size = roundup(offset + size, PAGE_SIZE); /* * If the physical address lies within a valid BAT table entry, * return the 1:1 mapping. This currently doesn't work * for regions that overlap 256M BAT segments. */ for (i = 0; i < 16; i++) { if (moea_bat_mapped(i, pa, size) == 0) return ((void *) pa); } va = kva_alloc(size); if (!va) panic("moea_mapdev: Couldn't alloc kernel virtual memory"); for (tmpva = va; size > 0;) { moea_kenter_attr(mmu, tmpva, ppa, ma); tlbie(tmpva); size -= PAGE_SIZE; tmpva += PAGE_SIZE; ppa += PAGE_SIZE; } return ((void *)(va + offset)); } void moea_unmapdev(mmu_t mmu, vm_offset_t va, vm_size_t size) { vm_offset_t base, offset; /* * If this is outside kernel virtual space, then it's a * battable entry and doesn't require unmapping */ if ((va >= VM_MIN_KERNEL_ADDRESS) && (va <= virtual_end)) { base = trunc_page(va); offset = va & PAGE_MASK; size = roundup(offset + size, PAGE_SIZE); kva_free(base, size); } } static void moea_sync_icache(mmu_t mmu, pmap_t pm, vm_offset_t va, vm_size_t sz) { struct pvo_entry *pvo; vm_offset_t lim; vm_paddr_t pa; vm_size_t len; PMAP_LOCK(pm); while (sz > 0) { lim = round_page(va); len = MIN(lim - va, sz); pvo = moea_pvo_find_va(pm, va & ~ADDR_POFF, NULL); if (pvo != NULL) { pa = (pvo->pvo_pte.pte.pte_lo & PTE_RPGN) | (va & ADDR_POFF); moea_syncicache(pa, len); } va += len; sz -= len; } PMAP_UNLOCK(pm); } void moea_dumpsys_map(mmu_t mmu, vm_paddr_t pa, size_t sz, void **va) { *va = (void *)pa; } extern struct dump_pa dump_map[PHYS_AVAIL_SZ + 1]; void moea_scan_init(mmu_t mmu) { struct pvo_entry *pvo; vm_offset_t va; int i; if (!do_minidump) { /* Initialize phys. segments for dumpsys(). */ memset(&dump_map, 0, sizeof(dump_map)); mem_regions(&pregions, &pregions_sz, ®ions, ®ions_sz); for (i = 0; i < pregions_sz; i++) { dump_map[i].pa_start = pregions[i].mr_start; dump_map[i].pa_size = pregions[i].mr_size; } return; } /* Virtual segments for minidumps: */ memset(&dump_map, 0, sizeof(dump_map)); /* 1st: kernel .data and .bss. */ dump_map[0].pa_start = trunc_page((uintptr_t)_etext); dump_map[0].pa_size = round_page((uintptr_t)_end) - dump_map[0].pa_start; /* 2nd: msgbuf and tables (see pmap_bootstrap()). */ dump_map[1].pa_start = (vm_paddr_t)msgbufp->msg_ptr; dump_map[1].pa_size = round_page(msgbufp->msg_size); /* 3rd: kernel VM. */ va = dump_map[1].pa_start + dump_map[1].pa_size; /* Find start of next chunk (from va). */ while (va < virtual_end) { /* Don't dump the buffer cache. */ if (va >= kmi.buffer_sva && va < kmi.buffer_eva) { va = kmi.buffer_eva; continue; } pvo = moea_pvo_find_va(kernel_pmap, va & ~ADDR_POFF, NULL); if (pvo != NULL && (pvo->pvo_pte.pte.pte_hi & PTE_VALID)) break; va += PAGE_SIZE; } if (va < virtual_end) { dump_map[2].pa_start = va; va += PAGE_SIZE; /* Find last page in chunk. */ while (va < virtual_end) { /* Don't run into the buffer cache. */ if (va == kmi.buffer_sva) break; pvo = moea_pvo_find_va(kernel_pmap, va & ~ADDR_POFF, NULL); if (pvo == NULL || !(pvo->pvo_pte.pte.pte_hi & PTE_VALID)) break; va += PAGE_SIZE; } dump_map[2].pa_size = va - dump_map[2].pa_start; } } Index: projects/cxl_iscsi/sys/powerpc/aim/mmu_oea64.c =================================================================== --- projects/cxl_iscsi/sys/powerpc/aim/mmu_oea64.c (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/aim/mmu_oea64.c (revision 279717) @@ -1,2649 +1,2654 @@ /*- * Copyright (c) 2008-2015 Nathan Whitehorn * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); /* * Manages physical address maps. * * Since the information managed by this module is also stored by the * logical address mapping module, this module may throw away valid virtual * to physical mappings at almost any time. However, invalidations of * mappings must be done as requested. * * In order to cope with hardware architectures which make virtual to * physical map invalidates expensive, this module may delay invalidate * reduced protection operations until such time as they are actually * necessary. This module is given full information as to which processors * are currently using which maps, and to when physical maps must be made * correct. */ #include "opt_compat.h" #include "opt_kstack_pages.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "mmu_oea64.h" #include "mmu_if.h" #include "moea64_if.h" void moea64_release_vsid(uint64_t vsid); uintptr_t moea64_get_unique_vsid(void); #define DISABLE_TRANS(msr) msr = mfmsr(); mtmsr(msr & ~PSL_DR) #define ENABLE_TRANS(msr) mtmsr(msr) #define VSID_MAKE(sr, hash) ((sr) | (((hash) & 0xfffff) << 4)) #define VSID_TO_HASH(vsid) (((vsid) >> 4) & 0xfffff) #define VSID_HASH_MASK 0x0000007fffffffffULL /* * Locking semantics: * * There are two locks of interest: the page locks and the pmap locks, which * protect their individual PVO lists and are locked in that order. The contents * of all PVO entries are protected by the locks of their respective pmaps. * The pmap of any PVO is guaranteed not to change so long as the PVO is linked * into any list. * */ #define PV_LOCK_COUNT PA_LOCK_COUNT*3 static struct mtx_padalign pv_lock[PV_LOCK_COUNT]; #define PV_LOCKPTR(pa) ((struct mtx *)(&pv_lock[pa_index(pa) % PV_LOCK_COUNT])) #define PV_LOCK(pa) mtx_lock(PV_LOCKPTR(pa)) #define PV_UNLOCK(pa) mtx_unlock(PV_LOCKPTR(pa)) #define PV_LOCKASSERT(pa) mtx_assert(PV_LOCKPTR(pa), MA_OWNED) #define PV_PAGE_LOCK(m) PV_LOCK(VM_PAGE_TO_PHYS(m)) #define PV_PAGE_UNLOCK(m) PV_UNLOCK(VM_PAGE_TO_PHYS(m)) #define PV_PAGE_LOCKASSERT(m) PV_LOCKASSERT(VM_PAGE_TO_PHYS(m)) struct ofw_map { cell_t om_va; cell_t om_len; uint64_t om_pa; cell_t om_mode; }; extern unsigned char _etext[]; extern unsigned char _end[]; extern int ofw_real_mode; /* * Map of physical memory regions. */ static struct mem_region *regions; static struct mem_region *pregions; static u_int phys_avail_count; static int regions_sz, pregions_sz; extern void bs_remap_earlyboot(void); /* * Lock for the SLB tables. */ struct mtx moea64_slb_mutex; /* * PTEG data. */ u_int moea64_pteg_count; u_int moea64_pteg_mask; /* * PVO data. */ uma_zone_t moea64_pvo_zone; /* zone for pvo entries */ static struct pvo_entry *moea64_bpvo_pool; static int moea64_bpvo_pool_index = 0; static int moea64_bpvo_pool_size = 327680; TUNABLE_INT("machdep.moea64_bpvo_pool_size", &moea64_bpvo_pool_size); SYSCTL_INT(_machdep, OID_AUTO, moea64_allocated_bpvo_entries, CTLFLAG_RD, &moea64_bpvo_pool_index, 0, ""); #define VSID_NBPW (sizeof(u_int32_t) * 8) #ifdef __powerpc64__ #define NVSIDS (NPMAPS * 16) #define VSID_HASHMASK 0xffffffffUL #else #define NVSIDS NPMAPS #define VSID_HASHMASK 0xfffffUL #endif static u_int moea64_vsid_bitmap[NVSIDS / VSID_NBPW]; static boolean_t moea64_initialized = FALSE; /* * Statistics. */ u_int moea64_pte_valid = 0; u_int moea64_pte_overflow = 0; u_int moea64_pvo_entries = 0; u_int moea64_pvo_enter_calls = 0; u_int moea64_pvo_remove_calls = 0; SYSCTL_INT(_machdep, OID_AUTO, moea64_pte_valid, CTLFLAG_RD, &moea64_pte_valid, 0, ""); SYSCTL_INT(_machdep, OID_AUTO, moea64_pte_overflow, CTLFLAG_RD, &moea64_pte_overflow, 0, ""); SYSCTL_INT(_machdep, OID_AUTO, moea64_pvo_entries, CTLFLAG_RD, &moea64_pvo_entries, 0, ""); SYSCTL_INT(_machdep, OID_AUTO, moea64_pvo_enter_calls, CTLFLAG_RD, &moea64_pvo_enter_calls, 0, ""); SYSCTL_INT(_machdep, OID_AUTO, moea64_pvo_remove_calls, CTLFLAG_RD, &moea64_pvo_remove_calls, 0, ""); vm_offset_t moea64_scratchpage_va[2]; struct pvo_entry *moea64_scratchpage_pvo[2]; struct mtx moea64_scratchpage_mtx; uint64_t moea64_large_page_mask = 0; uint64_t moea64_large_page_size = 0; int moea64_large_page_shift = 0; /* * PVO calls. */ static int moea64_pvo_enter(mmu_t mmu, struct pvo_entry *pvo, struct pvo_head *pvo_head); static void moea64_pvo_remove_from_pmap(mmu_t mmu, struct pvo_entry *pvo); static void moea64_pvo_remove_from_page(mmu_t mmu, struct pvo_entry *pvo); static struct pvo_entry *moea64_pvo_find_va(pmap_t, vm_offset_t); /* * Utility routines. */ static boolean_t moea64_query_bit(mmu_t, vm_page_t, uint64_t); static u_int moea64_clear_bit(mmu_t, vm_page_t, uint64_t); static void moea64_kremove(mmu_t, vm_offset_t); static void moea64_syncicache(mmu_t, pmap_t pmap, vm_offset_t va, vm_offset_t pa, vm_size_t sz); /* * Kernel MMU interface */ void moea64_clear_modify(mmu_t, vm_page_t); void moea64_copy_page(mmu_t, vm_page_t, vm_page_t); void moea64_copy_pages(mmu_t mmu, vm_page_t *ma, vm_offset_t a_offset, vm_page_t *mb, vm_offset_t b_offset, int xfersize); int moea64_enter(mmu_t, pmap_t, vm_offset_t, vm_page_t, vm_prot_t, u_int flags, int8_t psind); void moea64_enter_object(mmu_t, pmap_t, vm_offset_t, vm_offset_t, vm_page_t, vm_prot_t); void moea64_enter_quick(mmu_t, pmap_t, vm_offset_t, vm_page_t, vm_prot_t); vm_paddr_t moea64_extract(mmu_t, pmap_t, vm_offset_t); vm_page_t moea64_extract_and_hold(mmu_t, pmap_t, vm_offset_t, vm_prot_t); void moea64_init(mmu_t); boolean_t moea64_is_modified(mmu_t, vm_page_t); boolean_t moea64_is_prefaultable(mmu_t, pmap_t, vm_offset_t); boolean_t moea64_is_referenced(mmu_t, vm_page_t); int moea64_ts_referenced(mmu_t, vm_page_t); vm_offset_t moea64_map(mmu_t, vm_offset_t *, vm_paddr_t, vm_paddr_t, int); boolean_t moea64_page_exists_quick(mmu_t, pmap_t, vm_page_t); int moea64_page_wired_mappings(mmu_t, vm_page_t); void moea64_pinit(mmu_t, pmap_t); void moea64_pinit0(mmu_t, pmap_t); void moea64_protect(mmu_t, pmap_t, vm_offset_t, vm_offset_t, vm_prot_t); void moea64_qenter(mmu_t, vm_offset_t, vm_page_t *, int); void moea64_qremove(mmu_t, vm_offset_t, int); void moea64_release(mmu_t, pmap_t); void moea64_remove(mmu_t, pmap_t, vm_offset_t, vm_offset_t); void moea64_remove_pages(mmu_t, pmap_t); void moea64_remove_all(mmu_t, vm_page_t); void moea64_remove_write(mmu_t, vm_page_t); void moea64_unwire(mmu_t, pmap_t, vm_offset_t, vm_offset_t); void moea64_zero_page(mmu_t, vm_page_t); void moea64_zero_page_area(mmu_t, vm_page_t, int, int); void moea64_zero_page_idle(mmu_t, vm_page_t); void moea64_activate(mmu_t, struct thread *); void moea64_deactivate(mmu_t, struct thread *); void *moea64_mapdev(mmu_t, vm_paddr_t, vm_size_t); void *moea64_mapdev_attr(mmu_t, vm_offset_t, vm_size_t, vm_memattr_t); void moea64_unmapdev(mmu_t, vm_offset_t, vm_size_t); vm_paddr_t moea64_kextract(mmu_t, vm_offset_t); void moea64_page_set_memattr(mmu_t, vm_page_t m, vm_memattr_t ma); void moea64_kenter_attr(mmu_t, vm_offset_t, vm_offset_t, vm_memattr_t ma); void moea64_kenter(mmu_t, vm_offset_t, vm_paddr_t); boolean_t moea64_dev_direct_mapped(mmu_t, vm_paddr_t, vm_size_t); static void moea64_sync_icache(mmu_t, pmap_t, vm_offset_t, vm_size_t); void moea64_dumpsys_map(mmu_t mmu, vm_paddr_t pa, size_t sz, void **va); void moea64_scan_init(mmu_t mmu); static mmu_method_t moea64_methods[] = { MMUMETHOD(mmu_clear_modify, moea64_clear_modify), MMUMETHOD(mmu_copy_page, moea64_copy_page), MMUMETHOD(mmu_copy_pages, moea64_copy_pages), MMUMETHOD(mmu_enter, moea64_enter), MMUMETHOD(mmu_enter_object, moea64_enter_object), MMUMETHOD(mmu_enter_quick, moea64_enter_quick), MMUMETHOD(mmu_extract, moea64_extract), MMUMETHOD(mmu_extract_and_hold, moea64_extract_and_hold), MMUMETHOD(mmu_init, moea64_init), MMUMETHOD(mmu_is_modified, moea64_is_modified), MMUMETHOD(mmu_is_prefaultable, moea64_is_prefaultable), MMUMETHOD(mmu_is_referenced, moea64_is_referenced), MMUMETHOD(mmu_ts_referenced, moea64_ts_referenced), MMUMETHOD(mmu_map, moea64_map), MMUMETHOD(mmu_page_exists_quick,moea64_page_exists_quick), MMUMETHOD(mmu_page_wired_mappings,moea64_page_wired_mappings), MMUMETHOD(mmu_pinit, moea64_pinit), MMUMETHOD(mmu_pinit0, moea64_pinit0), MMUMETHOD(mmu_protect, moea64_protect), MMUMETHOD(mmu_qenter, moea64_qenter), MMUMETHOD(mmu_qremove, moea64_qremove), MMUMETHOD(mmu_release, moea64_release), MMUMETHOD(mmu_remove, moea64_remove), MMUMETHOD(mmu_remove_pages, moea64_remove_pages), MMUMETHOD(mmu_remove_all, moea64_remove_all), MMUMETHOD(mmu_remove_write, moea64_remove_write), MMUMETHOD(mmu_sync_icache, moea64_sync_icache), MMUMETHOD(mmu_unwire, moea64_unwire), MMUMETHOD(mmu_zero_page, moea64_zero_page), MMUMETHOD(mmu_zero_page_area, moea64_zero_page_area), MMUMETHOD(mmu_zero_page_idle, moea64_zero_page_idle), MMUMETHOD(mmu_activate, moea64_activate), MMUMETHOD(mmu_deactivate, moea64_deactivate), MMUMETHOD(mmu_page_set_memattr, moea64_page_set_memattr), /* Internal interfaces */ MMUMETHOD(mmu_mapdev, moea64_mapdev), MMUMETHOD(mmu_mapdev_attr, moea64_mapdev_attr), MMUMETHOD(mmu_unmapdev, moea64_unmapdev), MMUMETHOD(mmu_kextract, moea64_kextract), MMUMETHOD(mmu_kenter, moea64_kenter), MMUMETHOD(mmu_kenter_attr, moea64_kenter_attr), MMUMETHOD(mmu_dev_direct_mapped,moea64_dev_direct_mapped), MMUMETHOD(mmu_scan_init, moea64_scan_init), MMUMETHOD(mmu_dumpsys_map, moea64_dumpsys_map), { 0, 0 } }; MMU_DEF(oea64_mmu, "mmu_oea64_base", moea64_methods, 0); static struct pvo_head * vm_page_to_pvoh(vm_page_t m) { mtx_assert(PV_LOCKPTR(VM_PAGE_TO_PHYS(m)), MA_OWNED); return (&m->md.mdpg_pvoh); } static struct pvo_entry * alloc_pvo_entry(int bootstrap) { struct pvo_entry *pvo; if (!moea64_initialized || bootstrap) { if (moea64_bpvo_pool_index >= moea64_bpvo_pool_size) { panic("moea64_enter: bpvo pool exhausted, %d, %d, %zd", moea64_bpvo_pool_index, moea64_bpvo_pool_size, moea64_bpvo_pool_size * sizeof(struct pvo_entry)); } pvo = &moea64_bpvo_pool[ atomic_fetchadd_int(&moea64_bpvo_pool_index, 1)]; bzero(pvo, sizeof(*pvo)); pvo->pvo_vaddr = PVO_BOOTSTRAP; } else { pvo = uma_zalloc(moea64_pvo_zone, M_NOWAIT); bzero(pvo, sizeof(*pvo)); } return (pvo); } static void init_pvo_entry(struct pvo_entry *pvo, pmap_t pmap, vm_offset_t va) { uint64_t vsid; uint64_t hash; int shift; PMAP_LOCK_ASSERT(pmap, MA_OWNED); pvo->pvo_pmap = pmap; va &= ~ADDR_POFF; pvo->pvo_vaddr |= va; vsid = va_to_vsid(pmap, va); pvo->pvo_vpn = (uint64_t)((va & ADDR_PIDX) >> ADDR_PIDX_SHFT) | (vsid << 16); shift = (pvo->pvo_vaddr & PVO_LARGE) ? moea64_large_page_shift : ADDR_PIDX_SHFT; hash = (vsid & VSID_HASH_MASK) ^ (((uint64_t)va & ADDR_PIDX) >> shift); pvo->pvo_pte.slot = (hash & moea64_pteg_mask) << 3; } static void free_pvo_entry(struct pvo_entry *pvo) { if (!(pvo->pvo_vaddr & PVO_BOOTSTRAP)) uma_zfree(moea64_pvo_zone, pvo); } void moea64_pte_from_pvo(const struct pvo_entry *pvo, struct lpte *lpte) { lpte->pte_hi = (pvo->pvo_vpn >> (ADDR_API_SHFT64 - ADDR_PIDX_SHFT)) & LPTE_AVPN_MASK; lpte->pte_hi |= LPTE_VALID; if (pvo->pvo_vaddr & PVO_LARGE) lpte->pte_hi |= LPTE_BIG; if (pvo->pvo_vaddr & PVO_WIRED) lpte->pte_hi |= LPTE_WIRED; if (pvo->pvo_vaddr & PVO_HID) lpte->pte_hi |= LPTE_HID; lpte->pte_lo = pvo->pvo_pte.pa; /* Includes WIMG bits */ if (pvo->pvo_pte.prot & VM_PROT_WRITE) lpte->pte_lo |= LPTE_BW; else lpte->pte_lo |= LPTE_BR; if (!(pvo->pvo_pte.prot & VM_PROT_EXECUTE)) lpte->pte_lo |= LPTE_NOEXEC; } static __inline uint64_t moea64_calc_wimg(vm_offset_t pa, vm_memattr_t ma) { uint64_t pte_lo; int i; if (ma != VM_MEMATTR_DEFAULT) { switch (ma) { case VM_MEMATTR_UNCACHEABLE: return (LPTE_I | LPTE_G); case VM_MEMATTR_WRITE_COMBINING: case VM_MEMATTR_WRITE_BACK: case VM_MEMATTR_PREFETCHABLE: return (LPTE_I); case VM_MEMATTR_WRITE_THROUGH: return (LPTE_W | LPTE_M); } } /* * Assume the page is cache inhibited and access is guarded unless * it's in our available memory array. */ pte_lo = LPTE_I | LPTE_G; for (i = 0; i < pregions_sz; i++) { if ((pa >= pregions[i].mr_start) && (pa < (pregions[i].mr_start + pregions[i].mr_size))) { pte_lo &= ~(LPTE_I | LPTE_G); pte_lo |= LPTE_M; break; } } return pte_lo; } /* * Quick sort callout for comparing memory regions. */ static int om_cmp(const void *a, const void *b); static int om_cmp(const void *a, const void *b) { const struct ofw_map *mapa; const struct ofw_map *mapb; mapa = a; mapb = b; if (mapa->om_pa < mapb->om_pa) return (-1); else if (mapa->om_pa > mapb->om_pa) return (1); else return (0); } static void moea64_add_ofw_mappings(mmu_t mmup, phandle_t mmu, size_t sz) { struct ofw_map translations[sz/(4*sizeof(cell_t))]; /*>= 4 cells per */ pcell_t acells, trans_cells[sz/sizeof(cell_t)]; struct pvo_entry *pvo; register_t msr; vm_offset_t off; vm_paddr_t pa_base; int i, j; bzero(translations, sz); OF_getprop(OF_finddevice("/"), "#address-cells", &acells, sizeof(acells)); if (OF_getprop(mmu, "translations", trans_cells, sz) == -1) panic("moea64_bootstrap: can't get ofw translations"); CTR0(KTR_PMAP, "moea64_add_ofw_mappings: translations"); sz /= sizeof(cell_t); for (i = 0, j = 0; i < sz; j++) { translations[j].om_va = trans_cells[i++]; translations[j].om_len = trans_cells[i++]; translations[j].om_pa = trans_cells[i++]; if (acells == 2) { translations[j].om_pa <<= 32; translations[j].om_pa |= trans_cells[i++]; } translations[j].om_mode = trans_cells[i++]; } KASSERT(i == sz, ("Translations map has incorrect cell count (%d/%zd)", i, sz)); sz = j; qsort(translations, sz, sizeof (*translations), om_cmp); for (i = 0; i < sz; i++) { pa_base = translations[i].om_pa; #ifndef __powerpc64__ if ((translations[i].om_pa >> 32) != 0) panic("OFW translations above 32-bit boundary!"); #endif if (pa_base % PAGE_SIZE) panic("OFW translation not page-aligned (phys)!"); if (translations[i].om_va % PAGE_SIZE) panic("OFW translation not page-aligned (virt)!"); CTR3(KTR_PMAP, "translation: pa=%#zx va=%#x len=%#x", pa_base, translations[i].om_va, translations[i].om_len); /* Now enter the pages for this mapping */ DISABLE_TRANS(msr); for (off = 0; off < translations[i].om_len; off += PAGE_SIZE) { /* If this address is direct-mapped, skip remapping */ if (hw_direct_map && translations[i].om_va == pa_base && moea64_calc_wimg(pa_base + off, VM_MEMATTR_DEFAULT) == LPTE_M) continue; PMAP_LOCK(kernel_pmap); pvo = moea64_pvo_find_va(kernel_pmap, translations[i].om_va + off); PMAP_UNLOCK(kernel_pmap); if (pvo != NULL) continue; moea64_kenter(mmup, translations[i].om_va + off, pa_base + off); } ENABLE_TRANS(msr); } } #ifdef __powerpc64__ static void moea64_probe_large_page(void) { uint16_t pvr = mfpvr() >> 16; switch (pvr) { case IBM970: case IBM970FX: case IBM970MP: powerpc_sync(); isync(); mtspr(SPR_HID4, mfspr(SPR_HID4) & ~HID4_970_DISABLE_LG_PG); powerpc_sync(); isync(); /* FALLTHROUGH */ default: moea64_large_page_size = 0x1000000; /* 16 MB */ moea64_large_page_shift = 24; } moea64_large_page_mask = moea64_large_page_size - 1; } static void moea64_bootstrap_slb_prefault(vm_offset_t va, int large) { struct slb *cache; struct slb entry; uint64_t esid, slbe; uint64_t i; cache = PCPU_GET(slb); esid = va >> ADDR_SR_SHFT; slbe = (esid << SLBE_ESID_SHIFT) | SLBE_VALID; for (i = 0; i < 64; i++) { if (cache[i].slbe == (slbe | i)) return; } entry.slbe = slbe; entry.slbv = KERNEL_VSID(esid) << SLBV_VSID_SHIFT; if (large) entry.slbv |= SLBV_L; slb_insert_kernel(entry.slbe, entry.slbv); } #endif static void moea64_setup_direct_map(mmu_t mmup, vm_offset_t kernelstart, vm_offset_t kernelend) { struct pvo_entry *pvo; register_t msr; vm_paddr_t pa; vm_offset_t size, off; uint64_t pte_lo; int i; if (moea64_large_page_size == 0) hw_direct_map = 0; DISABLE_TRANS(msr); if (hw_direct_map) { PMAP_LOCK(kernel_pmap); for (i = 0; i < pregions_sz; i++) { for (pa = pregions[i].mr_start; pa < pregions[i].mr_start + pregions[i].mr_size; pa += moea64_large_page_size) { pte_lo = LPTE_M; pvo = alloc_pvo_entry(1 /* bootstrap */); pvo->pvo_vaddr |= PVO_WIRED | PVO_LARGE; init_pvo_entry(pvo, kernel_pmap, pa); /* * Set memory access as guarded if prefetch within * the page could exit the available physmem area. */ if (pa & moea64_large_page_mask) { pa &= moea64_large_page_mask; pte_lo |= LPTE_G; } if (pa + moea64_large_page_size > pregions[i].mr_start + pregions[i].mr_size) pte_lo |= LPTE_G; pvo->pvo_pte.prot = VM_PROT_READ | VM_PROT_WRITE | VM_PROT_EXECUTE; pvo->pvo_pte.pa = pa | pte_lo; moea64_pvo_enter(mmup, pvo, NULL); } } PMAP_UNLOCK(kernel_pmap); } else { size = moea64_bpvo_pool_size*sizeof(struct pvo_entry); off = (vm_offset_t)(moea64_bpvo_pool); for (pa = off; pa < off + size; pa += PAGE_SIZE) moea64_kenter(mmup, pa, pa); /* * Map certain important things, like ourselves. * * NOTE: We do not map the exception vector space. That code is * used only in real mode, and leaving it unmapped allows us to * catch NULL pointer deferences, instead of making NULL a valid * address. */ for (pa = kernelstart & ~PAGE_MASK; pa < kernelend; pa += PAGE_SIZE) moea64_kenter(mmup, pa, pa); } ENABLE_TRANS(msr); /* * Allow user to override unmapped_buf_allowed for testing. * XXXKIB Only direct map implementation was tested. */ if (!TUNABLE_INT_FETCH("vfs.unmapped_buf_allowed", &unmapped_buf_allowed)) unmapped_buf_allowed = hw_direct_map; } void moea64_early_bootstrap(mmu_t mmup, vm_offset_t kernelstart, vm_offset_t kernelend) { int i, j; vm_size_t physsz, hwphyssz; #ifndef __powerpc64__ /* We don't have a direct map since there is no BAT */ hw_direct_map = 0; /* Make sure battable is zero, since we have no BAT */ for (i = 0; i < 16; i++) { battable[i].batu = 0; battable[i].batl = 0; } #else moea64_probe_large_page(); /* Use a direct map if we have large page support */ if (moea64_large_page_size > 0) hw_direct_map = 1; else hw_direct_map = 0; #endif /* Get physical memory regions from firmware */ mem_regions(&pregions, &pregions_sz, ®ions, ®ions_sz); CTR0(KTR_PMAP, "moea64_bootstrap: physical memory"); if (sizeof(phys_avail)/sizeof(phys_avail[0]) < regions_sz) panic("moea64_bootstrap: phys_avail too small"); phys_avail_count = 0; physsz = 0; hwphyssz = 0; TUNABLE_ULONG_FETCH("hw.physmem", (u_long *) &hwphyssz); for (i = 0, j = 0; i < regions_sz; i++, j += 2) { CTR3(KTR_PMAP, "region: %#zx - %#zx (%#zx)", regions[i].mr_start, regions[i].mr_start + regions[i].mr_size, regions[i].mr_size); if (hwphyssz != 0 && (physsz + regions[i].mr_size) >= hwphyssz) { if (physsz < hwphyssz) { phys_avail[j] = regions[i].mr_start; phys_avail[j + 1] = regions[i].mr_start + hwphyssz - physsz; physsz = hwphyssz; phys_avail_count++; } break; } phys_avail[j] = regions[i].mr_start; phys_avail[j + 1] = regions[i].mr_start + regions[i].mr_size; phys_avail_count++; physsz += regions[i].mr_size; } /* Check for overlap with the kernel and exception vectors */ for (j = 0; j < 2*phys_avail_count; j+=2) { if (phys_avail[j] < EXC_LAST) phys_avail[j] += EXC_LAST; if (kernelstart >= phys_avail[j] && kernelstart < phys_avail[j+1]) { if (kernelend < phys_avail[j+1]) { phys_avail[2*phys_avail_count] = (kernelend & ~PAGE_MASK) + PAGE_SIZE; phys_avail[2*phys_avail_count + 1] = phys_avail[j+1]; phys_avail_count++; } phys_avail[j+1] = kernelstart & ~PAGE_MASK; } if (kernelend >= phys_avail[j] && kernelend < phys_avail[j+1]) { if (kernelstart > phys_avail[j]) { phys_avail[2*phys_avail_count] = phys_avail[j]; phys_avail[2*phys_avail_count + 1] = kernelstart & ~PAGE_MASK; phys_avail_count++; } phys_avail[j] = (kernelend & ~PAGE_MASK) + PAGE_SIZE; } } physmem = btoc(physsz); #ifdef PTEGCOUNT moea64_pteg_count = PTEGCOUNT; #else moea64_pteg_count = 0x1000; while (moea64_pteg_count < physmem) moea64_pteg_count <<= 1; moea64_pteg_count >>= 1; #endif /* PTEGCOUNT */ } void moea64_mid_bootstrap(mmu_t mmup, vm_offset_t kernelstart, vm_offset_t kernelend) { int i; /* * Set PTEG mask */ moea64_pteg_mask = moea64_pteg_count - 1; /* * Initialize SLB table lock and page locks */ mtx_init(&moea64_slb_mutex, "SLB table", NULL, MTX_DEF); for (i = 0; i < PV_LOCK_COUNT; i++) mtx_init(&pv_lock[i], "page pv", NULL, MTX_DEF); /* * Initialise the bootstrap pvo pool. */ moea64_bpvo_pool = (struct pvo_entry *)moea64_bootstrap_alloc( moea64_bpvo_pool_size*sizeof(struct pvo_entry), 0); moea64_bpvo_pool_index = 0; /* * Make sure kernel vsid is allocated as well as VSID 0. */ #ifndef __powerpc64__ moea64_vsid_bitmap[(KERNEL_VSIDBITS & (NVSIDS - 1)) / VSID_NBPW] |= 1 << (KERNEL_VSIDBITS % VSID_NBPW); moea64_vsid_bitmap[0] |= 1; #endif /* * Initialize the kernel pmap (which is statically allocated). */ #ifdef __powerpc64__ for (i = 0; i < 64; i++) { pcpup->pc_slb[i].slbv = 0; pcpup->pc_slb[i].slbe = 0; } #else for (i = 0; i < 16; i++) kernel_pmap->pm_sr[i] = EMPTY_SEGMENT + i; #endif kernel_pmap->pmap_phys = kernel_pmap; CPU_FILL(&kernel_pmap->pm_active); RB_INIT(&kernel_pmap->pmap_pvo); PMAP_LOCK_INIT(kernel_pmap); /* * Now map in all the other buffers we allocated earlier */ moea64_setup_direct_map(mmup, kernelstart, kernelend); } void moea64_late_bootstrap(mmu_t mmup, vm_offset_t kernelstart, vm_offset_t kernelend) { ihandle_t mmui; phandle_t chosen; phandle_t mmu; ssize_t sz; int i; vm_offset_t pa, va; void *dpcpu; /* * Set up the Open Firmware pmap and add its mappings if not in real * mode. */ chosen = OF_finddevice("/chosen"); if (!ofw_real_mode && chosen != -1 && OF_getprop(chosen, "mmu", &mmui, 4) != -1) { mmu = OF_instance_to_package(mmui); if (mmu == -1 || (sz = OF_getproplen(mmu, "translations")) == -1) sz = 0; if (sz > 6144 /* tmpstksz - 2 KB headroom */) panic("moea64_bootstrap: too many ofw translations"); if (sz > 0) moea64_add_ofw_mappings(mmup, mmu, sz); } /* * Calculate the last available physical address. */ for (i = 0; phys_avail[i + 2] != 0; i += 2) ; Maxmem = powerpc_btop(phys_avail[i + 1]); /* * Initialize MMU and remap early physical mappings */ MMU_CPU_BOOTSTRAP(mmup,0); mtmsr(mfmsr() | PSL_DR | PSL_IR); pmap_bootstrapped++; bs_remap_earlyboot(); /* * Set the start and end of kva. */ virtual_avail = VM_MIN_KERNEL_ADDRESS; virtual_end = VM_MAX_SAFE_KERNEL_ADDRESS; /* * Map the entire KVA range into the SLB. We must not fault there. */ #ifdef __powerpc64__ for (va = virtual_avail; va < virtual_end; va += SEGMENT_LENGTH) moea64_bootstrap_slb_prefault(va, 0); #endif /* * Figure out how far we can extend virtual_end into segment 16 * without running into existing mappings. Segment 16 is guaranteed * to contain neither RAM nor devices (at least on Apple hardware), * but will generally contain some OFW mappings we should not * step on. */ #ifndef __powerpc64__ /* KVA is in high memory on PPC64 */ PMAP_LOCK(kernel_pmap); while (virtual_end < VM_MAX_KERNEL_ADDRESS && moea64_pvo_find_va(kernel_pmap, virtual_end+1) == NULL) virtual_end += PAGE_SIZE; PMAP_UNLOCK(kernel_pmap); #endif /* * Allocate a kernel stack with a guard page for thread0 and map it * into the kernel page map. */ pa = moea64_bootstrap_alloc(KSTACK_PAGES * PAGE_SIZE, PAGE_SIZE); va = virtual_avail + KSTACK_GUARD_PAGES * PAGE_SIZE; virtual_avail = va + KSTACK_PAGES * PAGE_SIZE; CTR2(KTR_PMAP, "moea64_bootstrap: kstack0 at %#x (%#x)", pa, va); thread0.td_kstack = va; thread0.td_kstack_pages = KSTACK_PAGES; for (i = 0; i < KSTACK_PAGES; i++) { moea64_kenter(mmup, va, pa); pa += PAGE_SIZE; va += PAGE_SIZE; } /* * Allocate virtual address space for the message buffer. */ pa = msgbuf_phys = moea64_bootstrap_alloc(msgbufsize, PAGE_SIZE); msgbufp = (struct msgbuf *)virtual_avail; va = virtual_avail; virtual_avail += round_page(msgbufsize); while (va < virtual_avail) { moea64_kenter(mmup, va, pa); pa += PAGE_SIZE; va += PAGE_SIZE; } /* * Allocate virtual address space for the dynamic percpu area. */ pa = moea64_bootstrap_alloc(DPCPU_SIZE, PAGE_SIZE); dpcpu = (void *)virtual_avail; va = virtual_avail; virtual_avail += DPCPU_SIZE; while (va < virtual_avail) { moea64_kenter(mmup, va, pa); pa += PAGE_SIZE; va += PAGE_SIZE; } dpcpu_init(dpcpu, 0); /* * Allocate some things for page zeroing. We put this directly * in the page table and use MOEA64_PTE_REPLACE to avoid any * of the PVO book-keeping or other parts of the VM system * from even knowing that this hack exists. */ if (!hw_direct_map) { mtx_init(&moea64_scratchpage_mtx, "pvo zero page", NULL, MTX_DEF); for (i = 0; i < 2; i++) { moea64_scratchpage_va[i] = (virtual_end+1) - PAGE_SIZE; virtual_end -= PAGE_SIZE; moea64_kenter(mmup, moea64_scratchpage_va[i], 0); PMAP_LOCK(kernel_pmap); moea64_scratchpage_pvo[i] = moea64_pvo_find_va( kernel_pmap, (vm_offset_t)moea64_scratchpage_va[i]); PMAP_UNLOCK(kernel_pmap); } } } /* * Activate a user pmap. This mostly involves setting some non-CPU * state. */ void moea64_activate(mmu_t mmu, struct thread *td) { pmap_t pm; pm = &td->td_proc->p_vmspace->vm_pmap; CPU_SET(PCPU_GET(cpuid), &pm->pm_active); #ifdef __powerpc64__ PCPU_SET(userslb, pm->pm_slb); + __asm __volatile("slbmte %0, %1; isync" :: + "r"(td->td_pcb->pcb_cpu.aim.usr_vsid), "r"(USER_SLB_SLBE)); #else PCPU_SET(curpmap, pm->pmap_phys); + mtsrin(USER_SR << ADDR_SR_SHFT, td->td_pcb->pcb_cpu.aim.usr_vsid); #endif } void moea64_deactivate(mmu_t mmu, struct thread *td) { pmap_t pm; + + __asm __volatile("isync; slbie %0" :: "r"(USER_ADDR)); pm = &td->td_proc->p_vmspace->vm_pmap; CPU_CLR(PCPU_GET(cpuid), &pm->pm_active); #ifdef __powerpc64__ PCPU_SET(userslb, NULL); #else PCPU_SET(curpmap, NULL); #endif } void moea64_unwire(mmu_t mmu, pmap_t pm, vm_offset_t sva, vm_offset_t eva) { struct pvo_entry key, *pvo; vm_page_t m; int64_t refchg; key.pvo_vaddr = sva; PMAP_LOCK(pm); for (pvo = RB_NFIND(pvo_tree, &pm->pmap_pvo, &key); pvo != NULL && PVO_VADDR(pvo) < eva; pvo = RB_NEXT(pvo_tree, &pm->pmap_pvo, pvo)) { if ((pvo->pvo_vaddr & PVO_WIRED) == 0) panic("moea64_unwire: pvo %p is missing PVO_WIRED", pvo); pvo->pvo_vaddr &= ~PVO_WIRED; refchg = MOEA64_PTE_REPLACE(mmu, pvo, 0 /* No invalidation */); if ((pvo->pvo_vaddr & PVO_MANAGED) && (pvo->pvo_pte.prot & VM_PROT_WRITE)) { if (refchg < 0) refchg = LPTE_CHG; m = PHYS_TO_VM_PAGE(pvo->pvo_pte.pa & LPTE_RPGN); refchg |= atomic_readandclear_32(&m->md.mdpg_attrs); if (refchg & LPTE_CHG) vm_page_dirty(m); if (refchg & LPTE_REF) vm_page_aflag_set(m, PGA_REFERENCED); } pm->pm_stats.wired_count--; } PMAP_UNLOCK(pm); } /* * This goes through and sets the physical address of our * special scratch PTE to the PA we want to zero or copy. Because * of locking issues (this can get called in pvo_enter() by * the UMA allocator), we can't use most other utility functions here */ static __inline void moea64_set_scratchpage_pa(mmu_t mmup, int which, vm_offset_t pa) { KASSERT(!hw_direct_map, ("Using OEA64 scratchpage with a direct map!")); mtx_assert(&moea64_scratchpage_mtx, MA_OWNED); moea64_scratchpage_pvo[which]->pvo_pte.pa = moea64_calc_wimg(pa, VM_MEMATTR_DEFAULT) | (uint64_t)pa; MOEA64_PTE_REPLACE(mmup, moea64_scratchpage_pvo[which], MOEA64_PTE_INVALIDATE); isync(); } void moea64_copy_page(mmu_t mmu, vm_page_t msrc, vm_page_t mdst) { vm_offset_t dst; vm_offset_t src; dst = VM_PAGE_TO_PHYS(mdst); src = VM_PAGE_TO_PHYS(msrc); if (hw_direct_map) { bcopy((void *)src, (void *)dst, PAGE_SIZE); } else { mtx_lock(&moea64_scratchpage_mtx); moea64_set_scratchpage_pa(mmu, 0, src); moea64_set_scratchpage_pa(mmu, 1, dst); bcopy((void *)moea64_scratchpage_va[0], (void *)moea64_scratchpage_va[1], PAGE_SIZE); mtx_unlock(&moea64_scratchpage_mtx); } } static inline void moea64_copy_pages_dmap(mmu_t mmu, vm_page_t *ma, vm_offset_t a_offset, vm_page_t *mb, vm_offset_t b_offset, int xfersize) { void *a_cp, *b_cp; vm_offset_t a_pg_offset, b_pg_offset; int cnt; while (xfersize > 0) { a_pg_offset = a_offset & PAGE_MASK; cnt = min(xfersize, PAGE_SIZE - a_pg_offset); a_cp = (char *)VM_PAGE_TO_PHYS(ma[a_offset >> PAGE_SHIFT]) + a_pg_offset; b_pg_offset = b_offset & PAGE_MASK; cnt = min(cnt, PAGE_SIZE - b_pg_offset); b_cp = (char *)VM_PAGE_TO_PHYS(mb[b_offset >> PAGE_SHIFT]) + b_pg_offset; bcopy(a_cp, b_cp, cnt); a_offset += cnt; b_offset += cnt; xfersize -= cnt; } } static inline void moea64_copy_pages_nodmap(mmu_t mmu, vm_page_t *ma, vm_offset_t a_offset, vm_page_t *mb, vm_offset_t b_offset, int xfersize) { void *a_cp, *b_cp; vm_offset_t a_pg_offset, b_pg_offset; int cnt; mtx_lock(&moea64_scratchpage_mtx); while (xfersize > 0) { a_pg_offset = a_offset & PAGE_MASK; cnt = min(xfersize, PAGE_SIZE - a_pg_offset); moea64_set_scratchpage_pa(mmu, 0, VM_PAGE_TO_PHYS(ma[a_offset >> PAGE_SHIFT])); a_cp = (char *)moea64_scratchpage_va[0] + a_pg_offset; b_pg_offset = b_offset & PAGE_MASK; cnt = min(cnt, PAGE_SIZE - b_pg_offset); moea64_set_scratchpage_pa(mmu, 1, VM_PAGE_TO_PHYS(mb[b_offset >> PAGE_SHIFT])); b_cp = (char *)moea64_scratchpage_va[1] + b_pg_offset; bcopy(a_cp, b_cp, cnt); a_offset += cnt; b_offset += cnt; xfersize -= cnt; } mtx_unlock(&moea64_scratchpage_mtx); } void moea64_copy_pages(mmu_t mmu, vm_page_t *ma, vm_offset_t a_offset, vm_page_t *mb, vm_offset_t b_offset, int xfersize) { if (hw_direct_map) { moea64_copy_pages_dmap(mmu, ma, a_offset, mb, b_offset, xfersize); } else { moea64_copy_pages_nodmap(mmu, ma, a_offset, mb, b_offset, xfersize); } } void moea64_zero_page_area(mmu_t mmu, vm_page_t m, int off, int size) { vm_offset_t pa = VM_PAGE_TO_PHYS(m); if (size + off > PAGE_SIZE) panic("moea64_zero_page: size + off > PAGE_SIZE"); if (hw_direct_map) { bzero((caddr_t)pa + off, size); } else { mtx_lock(&moea64_scratchpage_mtx); moea64_set_scratchpage_pa(mmu, 0, pa); bzero((caddr_t)moea64_scratchpage_va[0] + off, size); mtx_unlock(&moea64_scratchpage_mtx); } } /* * Zero a page of physical memory by temporarily mapping it */ void moea64_zero_page(mmu_t mmu, vm_page_t m) { vm_offset_t pa = VM_PAGE_TO_PHYS(m); vm_offset_t va, off; if (!hw_direct_map) { mtx_lock(&moea64_scratchpage_mtx); moea64_set_scratchpage_pa(mmu, 0, pa); va = moea64_scratchpage_va[0]; } else { va = pa; } for (off = 0; off < PAGE_SIZE; off += cacheline_size) __asm __volatile("dcbz 0,%0" :: "r"(va + off)); if (!hw_direct_map) mtx_unlock(&moea64_scratchpage_mtx); } void moea64_zero_page_idle(mmu_t mmu, vm_page_t m) { moea64_zero_page(mmu, m); } /* * Map the given physical page at the specified virtual address in the * target pmap with the protection requested. If specified the page * will be wired down. */ int moea64_enter(mmu_t mmu, pmap_t pmap, vm_offset_t va, vm_page_t m, vm_prot_t prot, u_int flags, int8_t psind) { struct pvo_entry *pvo, *oldpvo; struct pvo_head *pvo_head; uint64_t pte_lo; int error; if ((m->oflags & VPO_UNMANAGED) == 0 && !vm_page_xbusied(m)) VM_OBJECT_ASSERT_LOCKED(m->object); pvo = alloc_pvo_entry(0); pvo->pvo_pmap = NULL; /* to be filled in later */ pvo->pvo_pte.prot = prot; pte_lo = moea64_calc_wimg(VM_PAGE_TO_PHYS(m), pmap_page_get_memattr(m)); pvo->pvo_pte.pa = VM_PAGE_TO_PHYS(m) | pte_lo; if ((flags & PMAP_ENTER_WIRED) != 0) pvo->pvo_vaddr |= PVO_WIRED; if ((m->oflags & VPO_UNMANAGED) != 0 || !moea64_initialized) { pvo_head = NULL; } else { pvo_head = &m->md.mdpg_pvoh; pvo->pvo_vaddr |= PVO_MANAGED; } for (;;) { PV_PAGE_LOCK(m); PMAP_LOCK(pmap); if (pvo->pvo_pmap == NULL) init_pvo_entry(pvo, pmap, va); if (prot & VM_PROT_WRITE) if (pmap_bootstrapped && (m->oflags & VPO_UNMANAGED) == 0) vm_page_aflag_set(m, PGA_WRITEABLE); oldpvo = moea64_pvo_find_va(pmap, va); if (oldpvo != NULL) { if (oldpvo->pvo_vaddr == pvo->pvo_vaddr && oldpvo->pvo_pte.pa == pvo->pvo_pte.pa && oldpvo->pvo_pte.prot == prot) { /* Identical mapping already exists */ error = 0; /* If not in page table, reinsert it */ if (MOEA64_PTE_SYNCH(mmu, oldpvo) < 0) { moea64_pte_overflow--; MOEA64_PTE_INSERT(mmu, oldpvo); } /* Then just clean up and go home */ PV_PAGE_UNLOCK(m); PMAP_UNLOCK(pmap); free_pvo_entry(pvo); break; } /* Otherwise, need to kill it first */ KASSERT(oldpvo->pvo_pmap == pmap, ("pmap of old " "mapping does not match new mapping")); moea64_pvo_remove_from_pmap(mmu, oldpvo); } error = moea64_pvo_enter(mmu, pvo, pvo_head); PV_PAGE_UNLOCK(m); PMAP_UNLOCK(pmap); /* Free any dead pages */ if (oldpvo != NULL) { PV_LOCK(oldpvo->pvo_pte.pa & LPTE_RPGN); moea64_pvo_remove_from_page(mmu, oldpvo); PV_UNLOCK(oldpvo->pvo_pte.pa & LPTE_RPGN); free_pvo_entry(oldpvo); } if (error != ENOMEM) break; if ((flags & PMAP_ENTER_NOSLEEP) != 0) return (KERN_RESOURCE_SHORTAGE); VM_OBJECT_ASSERT_UNLOCKED(m->object); VM_WAIT; } /* * Flush the page from the instruction cache if this page is * mapped executable and cacheable. */ if (pmap != kernel_pmap && !(m->aflags & PGA_EXECUTABLE) && (pte_lo & (LPTE_I | LPTE_G | LPTE_NOEXEC)) == 0) { vm_page_aflag_set(m, PGA_EXECUTABLE); moea64_syncicache(mmu, pmap, va, VM_PAGE_TO_PHYS(m), PAGE_SIZE); } return (KERN_SUCCESS); } static void moea64_syncicache(mmu_t mmu, pmap_t pmap, vm_offset_t va, vm_offset_t pa, vm_size_t sz) { /* * This is much trickier than on older systems because * we can't sync the icache on physical addresses directly * without a direct map. Instead we check a couple of cases * where the memory is already mapped in and, failing that, * use the same trick we use for page zeroing to create * a temporary mapping for this physical address. */ if (!pmap_bootstrapped) { /* * If PMAP is not bootstrapped, we are likely to be * in real mode. */ __syncicache((void *)pa, sz); } else if (pmap == kernel_pmap) { __syncicache((void *)va, sz); } else if (hw_direct_map) { __syncicache((void *)pa, sz); } else { /* Use the scratch page to set up a temp mapping */ mtx_lock(&moea64_scratchpage_mtx); moea64_set_scratchpage_pa(mmu, 1, pa & ~ADDR_POFF); __syncicache((void *)(moea64_scratchpage_va[1] + (va & ADDR_POFF)), sz); mtx_unlock(&moea64_scratchpage_mtx); } } /* * Maps a sequence of resident pages belonging to the same object. * The sequence begins with the given page m_start. This page is * mapped at the given virtual address start. Each subsequent page is * mapped at a virtual address that is offset from start by the same * amount as the page is offset from m_start within the object. The * last page in the sequence is the page with the largest offset from * m_start that can be mapped at a virtual address less than the given * virtual address end. Not every virtual page between start and end * is mapped; only those for which a resident page exists with the * corresponding offset from m_start are mapped. */ void moea64_enter_object(mmu_t mmu, pmap_t pm, vm_offset_t start, vm_offset_t end, vm_page_t m_start, vm_prot_t prot) { vm_page_t m; vm_pindex_t diff, psize; VM_OBJECT_ASSERT_LOCKED(m_start->object); psize = atop(end - start); m = m_start; while (m != NULL && (diff = m->pindex - m_start->pindex) < psize) { moea64_enter(mmu, pm, start + ptoa(diff), m, prot & (VM_PROT_READ | VM_PROT_EXECUTE), PMAP_ENTER_NOSLEEP, 0); m = TAILQ_NEXT(m, listq); } } void moea64_enter_quick(mmu_t mmu, pmap_t pm, vm_offset_t va, vm_page_t m, vm_prot_t prot) { moea64_enter(mmu, pm, va, m, prot & (VM_PROT_READ | VM_PROT_EXECUTE), PMAP_ENTER_NOSLEEP, 0); } vm_paddr_t moea64_extract(mmu_t mmu, pmap_t pm, vm_offset_t va) { struct pvo_entry *pvo; vm_paddr_t pa; PMAP_LOCK(pm); pvo = moea64_pvo_find_va(pm, va); if (pvo == NULL) pa = 0; else pa = (pvo->pvo_pte.pa & LPTE_RPGN) | (va - PVO_VADDR(pvo)); PMAP_UNLOCK(pm); return (pa); } /* * Atomically extract and hold the physical page with the given * pmap and virtual address pair if that mapping permits the given * protection. */ vm_page_t moea64_extract_and_hold(mmu_t mmu, pmap_t pmap, vm_offset_t va, vm_prot_t prot) { struct pvo_entry *pvo; vm_page_t m; vm_paddr_t pa; m = NULL; pa = 0; PMAP_LOCK(pmap); retry: pvo = moea64_pvo_find_va(pmap, va & ~ADDR_POFF); if (pvo != NULL && (pvo->pvo_pte.prot & prot) == prot) { if (vm_page_pa_tryrelock(pmap, pvo->pvo_pte.pa & LPTE_RPGN, &pa)) goto retry; m = PHYS_TO_VM_PAGE(pvo->pvo_pte.pa & LPTE_RPGN); vm_page_hold(m); } PA_UNLOCK_COND(pa); PMAP_UNLOCK(pmap); return (m); } static mmu_t installed_mmu; static void * moea64_uma_page_alloc(uma_zone_t zone, int bytes, u_int8_t *flags, int wait) { struct pvo_entry *pvo; vm_offset_t va; vm_page_t m; int pflags, needed_lock; /* * This entire routine is a horrible hack to avoid bothering kmem * for new KVA addresses. Because this can get called from inside * kmem allocation routines, calling kmem for a new address here * can lead to multiply locking non-recursive mutexes. */ *flags = UMA_SLAB_PRIV; needed_lock = !PMAP_LOCKED(kernel_pmap); pflags = malloc2vm_flags(wait) | VM_ALLOC_WIRED; for (;;) { m = vm_page_alloc(NULL, 0, pflags | VM_ALLOC_NOOBJ); if (m == NULL) { if (wait & M_NOWAIT) return (NULL); VM_WAIT; } else break; } va = VM_PAGE_TO_PHYS(m); pvo = alloc_pvo_entry(1 /* bootstrap */); pvo->pvo_pte.prot = VM_PROT_READ | VM_PROT_WRITE; pvo->pvo_pte.pa = VM_PAGE_TO_PHYS(m) | LPTE_M; if (needed_lock) PMAP_LOCK(kernel_pmap); init_pvo_entry(pvo, kernel_pmap, va); pvo->pvo_vaddr |= PVO_WIRED; moea64_pvo_enter(installed_mmu, pvo, NULL); if (needed_lock) PMAP_UNLOCK(kernel_pmap); if ((wait & M_ZERO) && (m->flags & PG_ZERO) == 0) bzero((void *)va, PAGE_SIZE); return (void *)va; } extern int elf32_nxstack; void moea64_init(mmu_t mmu) { CTR0(KTR_PMAP, "moea64_init"); moea64_pvo_zone = uma_zcreate("UPVO entry", sizeof (struct pvo_entry), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_VM | UMA_ZONE_NOFREE); if (!hw_direct_map) { installed_mmu = mmu; uma_zone_set_allocf(moea64_pvo_zone,moea64_uma_page_alloc); } #ifdef COMPAT_FREEBSD32 elf32_nxstack = 1; #endif moea64_initialized = TRUE; } boolean_t moea64_is_referenced(mmu_t mmu, vm_page_t m) { KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("moea64_is_referenced: page %p is not managed", m)); return (moea64_query_bit(mmu, m, LPTE_REF)); } boolean_t moea64_is_modified(mmu_t mmu, vm_page_t m) { KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("moea64_is_modified: page %p is not managed", m)); /* * If the page is not exclusive busied, then PGA_WRITEABLE cannot be * concurrently set while the object is locked. Thus, if PGA_WRITEABLE * is clear, no PTEs can have LPTE_CHG set. */ VM_OBJECT_ASSERT_LOCKED(m->object); if (!vm_page_xbusied(m) && (m->aflags & PGA_WRITEABLE) == 0) return (FALSE); return (moea64_query_bit(mmu, m, LPTE_CHG)); } boolean_t moea64_is_prefaultable(mmu_t mmu, pmap_t pmap, vm_offset_t va) { struct pvo_entry *pvo; boolean_t rv = TRUE; PMAP_LOCK(pmap); pvo = moea64_pvo_find_va(pmap, va & ~ADDR_POFF); if (pvo != NULL) rv = FALSE; PMAP_UNLOCK(pmap); return (rv); } void moea64_clear_modify(mmu_t mmu, vm_page_t m) { KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("moea64_clear_modify: page %p is not managed", m)); VM_OBJECT_ASSERT_WLOCKED(m->object); KASSERT(!vm_page_xbusied(m), ("moea64_clear_modify: page %p is exclusive busied", m)); /* * If the page is not PGA_WRITEABLE, then no PTEs can have LPTE_CHG * set. If the object containing the page is locked and the page is * not exclusive busied, then PGA_WRITEABLE cannot be concurrently set. */ if ((m->aflags & PGA_WRITEABLE) == 0) return; moea64_clear_bit(mmu, m, LPTE_CHG); } /* * Clear the write and modified bits in each of the given page's mappings. */ void moea64_remove_write(mmu_t mmu, vm_page_t m) { struct pvo_entry *pvo; int64_t refchg, ret; pmap_t pmap; KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("moea64_remove_write: page %p is not managed", m)); /* * If the page is not exclusive busied, then PGA_WRITEABLE cannot be * set by another thread while the object is locked. Thus, * if PGA_WRITEABLE is clear, no page table entries need updating. */ VM_OBJECT_ASSERT_WLOCKED(m->object); if (!vm_page_xbusied(m) && (m->aflags & PGA_WRITEABLE) == 0) return; powerpc_sync(); PV_PAGE_LOCK(m); refchg = 0; LIST_FOREACH(pvo, vm_page_to_pvoh(m), pvo_vlink) { pmap = pvo->pvo_pmap; PMAP_LOCK(pmap); if (!(pvo->pvo_vaddr & PVO_DEAD) && (pvo->pvo_pte.prot & VM_PROT_WRITE)) { pvo->pvo_pte.prot &= ~VM_PROT_WRITE; ret = MOEA64_PTE_REPLACE(mmu, pvo, MOEA64_PTE_PROT_UPDATE); if (ret < 0) ret = LPTE_CHG; refchg |= ret; if (pvo->pvo_pmap == kernel_pmap) isync(); } PMAP_UNLOCK(pmap); } if ((refchg | atomic_readandclear_32(&m->md.mdpg_attrs)) & LPTE_CHG) vm_page_dirty(m); vm_page_aflag_clear(m, PGA_WRITEABLE); PV_PAGE_UNLOCK(m); } /* * moea64_ts_referenced: * * Return a count of reference bits for a page, clearing those bits. * It is not necessary for every reference bit to be cleared, but it * is necessary that 0 only be returned when there are truly no * reference bits set. * * XXX: The exact number of bits to check and clear is a matter that * should be tested and standardized at some point in the future for * optimal aging of shared pages. */ int moea64_ts_referenced(mmu_t mmu, vm_page_t m) { KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("moea64_ts_referenced: page %p is not managed", m)); return (moea64_clear_bit(mmu, m, LPTE_REF)); } /* * Modify the WIMG settings of all mappings for a page. */ void moea64_page_set_memattr(mmu_t mmu, vm_page_t m, vm_memattr_t ma) { struct pvo_entry *pvo; int64_t refchg; pmap_t pmap; uint64_t lo; if ((m->oflags & VPO_UNMANAGED) != 0) { m->md.mdpg_cache_attrs = ma; return; } lo = moea64_calc_wimg(VM_PAGE_TO_PHYS(m), ma); PV_PAGE_LOCK(m); LIST_FOREACH(pvo, vm_page_to_pvoh(m), pvo_vlink) { pmap = pvo->pvo_pmap; PMAP_LOCK(pmap); if (!(pvo->pvo_vaddr & PVO_DEAD)) { pvo->pvo_pte.pa &= ~LPTE_WIMG; pvo->pvo_pte.pa |= lo; refchg = MOEA64_PTE_REPLACE(mmu, pvo, MOEA64_PTE_INVALIDATE); if (refchg < 0) refchg = (pvo->pvo_pte.prot & VM_PROT_WRITE) ? LPTE_CHG : 0; if ((pvo->pvo_vaddr & PVO_MANAGED) && (pvo->pvo_pte.prot & VM_PROT_WRITE)) { refchg |= atomic_readandclear_32(&m->md.mdpg_attrs); if (refchg & LPTE_CHG) vm_page_dirty(m); if (refchg & LPTE_REF) vm_page_aflag_set(m, PGA_REFERENCED); } if (pvo->pvo_pmap == kernel_pmap) isync(); } PMAP_UNLOCK(pmap); } m->md.mdpg_cache_attrs = ma; PV_PAGE_UNLOCK(m); } /* * Map a wired page into kernel virtual address space. */ void moea64_kenter_attr(mmu_t mmu, vm_offset_t va, vm_offset_t pa, vm_memattr_t ma) { int error; struct pvo_entry *pvo, *oldpvo; pvo = alloc_pvo_entry(0); pvo->pvo_pte.prot = VM_PROT_READ | VM_PROT_WRITE | VM_PROT_EXECUTE; pvo->pvo_pte.pa = (pa & ~ADDR_POFF) | moea64_calc_wimg(pa, ma); pvo->pvo_vaddr |= PVO_WIRED; PMAP_LOCK(kernel_pmap); oldpvo = moea64_pvo_find_va(kernel_pmap, va); if (oldpvo != NULL) moea64_pvo_remove_from_pmap(mmu, oldpvo); init_pvo_entry(pvo, kernel_pmap, va); error = moea64_pvo_enter(mmu, pvo, NULL); PMAP_UNLOCK(kernel_pmap); /* Free any dead pages */ if (oldpvo != NULL) { PV_LOCK(oldpvo->pvo_pte.pa & LPTE_RPGN); moea64_pvo_remove_from_page(mmu, oldpvo); PV_UNLOCK(oldpvo->pvo_pte.pa & LPTE_RPGN); free_pvo_entry(oldpvo); } if (error != 0 && error != ENOENT) panic("moea64_kenter: failed to enter va %#zx pa %#zx: %d", va, pa, error); } void moea64_kenter(mmu_t mmu, vm_offset_t va, vm_paddr_t pa) { moea64_kenter_attr(mmu, va, pa, VM_MEMATTR_DEFAULT); } /* * Extract the physical page address associated with the given kernel virtual * address. */ vm_paddr_t moea64_kextract(mmu_t mmu, vm_offset_t va) { struct pvo_entry *pvo; vm_paddr_t pa; /* * Shortcut the direct-mapped case when applicable. We never put * anything but 1:1 mappings below VM_MIN_KERNEL_ADDRESS. */ if (va < VM_MIN_KERNEL_ADDRESS) return (va); PMAP_LOCK(kernel_pmap); pvo = moea64_pvo_find_va(kernel_pmap, va); KASSERT(pvo != NULL, ("moea64_kextract: no addr found for %#" PRIxPTR, va)); pa = (pvo->pvo_pte.pa & LPTE_RPGN) | (va - PVO_VADDR(pvo)); PMAP_UNLOCK(kernel_pmap); return (pa); } /* * Remove a wired page from kernel virtual address space. */ void moea64_kremove(mmu_t mmu, vm_offset_t va) { moea64_remove(mmu, kernel_pmap, va, va + PAGE_SIZE); } /* * Map a range of physical addresses into kernel virtual address space. * * The value passed in *virt is a suggested virtual address for the mapping. * Architectures which can support a direct-mapped physical to virtual region * can return the appropriate address within that region, leaving '*virt' * unchanged. Other architectures should map the pages starting at '*virt' and * update '*virt' with the first usable address after the mapped region. */ vm_offset_t moea64_map(mmu_t mmu, vm_offset_t *virt, vm_paddr_t pa_start, vm_paddr_t pa_end, int prot) { vm_offset_t sva, va; if (hw_direct_map) { /* * Check if every page in the region is covered by the direct * map. The direct map covers all of physical memory. Use * moea64_calc_wimg() as a shortcut to see if the page is in * physical memory as a way to see if the direct map covers it. */ for (va = pa_start; va < pa_end; va += PAGE_SIZE) if (moea64_calc_wimg(va, VM_MEMATTR_DEFAULT) != LPTE_M) break; if (va == pa_end) return (pa_start); } sva = *virt; va = sva; /* XXX respect prot argument */ for (; pa_start < pa_end; pa_start += PAGE_SIZE, va += PAGE_SIZE) moea64_kenter(mmu, va, pa_start); *virt = va; return (sva); } /* * Returns true if the pmap's pv is one of the first * 16 pvs linked to from this page. This count may * be changed upwards or downwards in the future; it * is only necessary that true be returned for a small * subset of pmaps for proper page aging. */ boolean_t moea64_page_exists_quick(mmu_t mmu, pmap_t pmap, vm_page_t m) { int loops; struct pvo_entry *pvo; boolean_t rv; KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("moea64_page_exists_quick: page %p is not managed", m)); loops = 0; rv = FALSE; PV_PAGE_LOCK(m); LIST_FOREACH(pvo, vm_page_to_pvoh(m), pvo_vlink) { if (!(pvo->pvo_vaddr & PVO_DEAD) && pvo->pvo_pmap == pmap) { rv = TRUE; break; } if (++loops >= 16) break; } PV_PAGE_UNLOCK(m); return (rv); } /* * Return the number of managed mappings to the given physical page * that are wired. */ int moea64_page_wired_mappings(mmu_t mmu, vm_page_t m) { struct pvo_entry *pvo; int count; count = 0; if ((m->oflags & VPO_UNMANAGED) != 0) return (count); PV_PAGE_LOCK(m); LIST_FOREACH(pvo, vm_page_to_pvoh(m), pvo_vlink) if ((pvo->pvo_vaddr & (PVO_DEAD | PVO_WIRED)) == PVO_WIRED) count++; PV_PAGE_UNLOCK(m); return (count); } static uintptr_t moea64_vsidcontext; uintptr_t moea64_get_unique_vsid(void) { u_int entropy; register_t hash; uint32_t mask; int i; entropy = 0; __asm __volatile("mftb %0" : "=r"(entropy)); mtx_lock(&moea64_slb_mutex); for (i = 0; i < NVSIDS; i += VSID_NBPW) { u_int n; /* * Create a new value by mutiplying by a prime and adding in * entropy from the timebase register. This is to make the * VSID more random so that the PT hash function collides * less often. (Note that the prime casues gcc to do shifts * instead of a multiply.) */ moea64_vsidcontext = (moea64_vsidcontext * 0x1105) + entropy; hash = moea64_vsidcontext & (NVSIDS - 1); if (hash == 0) /* 0 is special, avoid it */ continue; n = hash >> 5; mask = 1 << (hash & (VSID_NBPW - 1)); hash = (moea64_vsidcontext & VSID_HASHMASK); if (moea64_vsid_bitmap[n] & mask) { /* collision? */ /* anything free in this bucket? */ if (moea64_vsid_bitmap[n] == 0xffffffff) { entropy = (moea64_vsidcontext >> 20); continue; } i = ffs(~moea64_vsid_bitmap[n]) - 1; mask = 1 << i; hash &= VSID_HASHMASK & ~(VSID_NBPW - 1); hash |= i; } KASSERT(!(moea64_vsid_bitmap[n] & mask), ("Allocating in-use VSID %#zx\n", hash)); moea64_vsid_bitmap[n] |= mask; mtx_unlock(&moea64_slb_mutex); return (hash); } mtx_unlock(&moea64_slb_mutex); panic("%s: out of segments",__func__); } #ifdef __powerpc64__ void moea64_pinit(mmu_t mmu, pmap_t pmap) { RB_INIT(&pmap->pmap_pvo); pmap->pm_slb_tree_root = slb_alloc_tree(); pmap->pm_slb = slb_alloc_user_cache(); pmap->pm_slb_len = 0; } #else void moea64_pinit(mmu_t mmu, pmap_t pmap) { int i; uint32_t hash; RB_INIT(&pmap->pmap_pvo); if (pmap_bootstrapped) pmap->pmap_phys = (pmap_t)moea64_kextract(mmu, (vm_offset_t)pmap); else pmap->pmap_phys = pmap; /* * Allocate some segment registers for this pmap. */ hash = moea64_get_unique_vsid(); for (i = 0; i < 16; i++) pmap->pm_sr[i] = VSID_MAKE(i, hash); KASSERT(pmap->pm_sr[0] != 0, ("moea64_pinit: pm_sr[0] = 0")); } #endif /* * Initialize the pmap associated with process 0. */ void moea64_pinit0(mmu_t mmu, pmap_t pm) { PMAP_LOCK_INIT(pm); moea64_pinit(mmu, pm); bzero(&pm->pm_stats, sizeof(pm->pm_stats)); } /* * Set the physical protection on the specified range of this map as requested. */ static void moea64_pvo_protect(mmu_t mmu, pmap_t pm, struct pvo_entry *pvo, vm_prot_t prot) { struct vm_page *pg; vm_prot_t oldprot; int32_t refchg; PMAP_LOCK_ASSERT(pm, MA_OWNED); /* * Change the protection of the page. */ oldprot = pvo->pvo_pte.prot; pvo->pvo_pte.prot = prot; pg = PHYS_TO_VM_PAGE(pvo->pvo_pte.pa & LPTE_RPGN); /* * If the PVO is in the page table, update mapping */ refchg = MOEA64_PTE_REPLACE(mmu, pvo, MOEA64_PTE_PROT_UPDATE); if (refchg < 0) refchg = (oldprot & VM_PROT_WRITE) ? LPTE_CHG : 0; if (pm != kernel_pmap && pg != NULL && !(pg->aflags & PGA_EXECUTABLE) && (pvo->pvo_pte.pa & (LPTE_I | LPTE_G | LPTE_NOEXEC)) == 0) { if ((pg->oflags & VPO_UNMANAGED) == 0) vm_page_aflag_set(pg, PGA_EXECUTABLE); moea64_syncicache(mmu, pm, PVO_VADDR(pvo), pvo->pvo_pte.pa & LPTE_RPGN, PAGE_SIZE); } /* * Update vm about the REF/CHG bits if the page is managed and we have * removed write access. */ if (pg != NULL && (pvo->pvo_vaddr & PVO_MANAGED) && (oldprot & VM_PROT_WRITE)) { refchg |= atomic_readandclear_32(&pg->md.mdpg_attrs); if (refchg & LPTE_CHG) vm_page_dirty(pg); if (refchg & LPTE_REF) vm_page_aflag_set(pg, PGA_REFERENCED); } } void moea64_protect(mmu_t mmu, pmap_t pm, vm_offset_t sva, vm_offset_t eva, vm_prot_t prot) { struct pvo_entry *pvo, *tpvo, key; CTR4(KTR_PMAP, "moea64_protect: pm=%p sva=%#x eva=%#x prot=%#x", pm, sva, eva, prot); KASSERT(pm == &curproc->p_vmspace->vm_pmap || pm == kernel_pmap, ("moea64_protect: non current pmap")); if ((prot & VM_PROT_READ) == VM_PROT_NONE) { moea64_remove(mmu, pm, sva, eva); return; } PMAP_LOCK(pm); key.pvo_vaddr = sva; for (pvo = RB_NFIND(pvo_tree, &pm->pmap_pvo, &key); pvo != NULL && PVO_VADDR(pvo) < eva; pvo = tpvo) { tpvo = RB_NEXT(pvo_tree, &pm->pmap_pvo, pvo); moea64_pvo_protect(mmu, pm, pvo, prot); } PMAP_UNLOCK(pm); } /* * Map a list of wired pages into kernel virtual address space. This is * intended for temporary mappings which do not need page modification or * references recorded. Existing mappings in the region are overwritten. */ void moea64_qenter(mmu_t mmu, vm_offset_t va, vm_page_t *m, int count) { while (count-- > 0) { moea64_kenter(mmu, va, VM_PAGE_TO_PHYS(*m)); va += PAGE_SIZE; m++; } } /* * Remove page mappings from kernel virtual address space. Intended for * temporary mappings entered by moea64_qenter. */ void moea64_qremove(mmu_t mmu, vm_offset_t va, int count) { while (count-- > 0) { moea64_kremove(mmu, va); va += PAGE_SIZE; } } void moea64_release_vsid(uint64_t vsid) { int idx, mask; mtx_lock(&moea64_slb_mutex); idx = vsid & (NVSIDS-1); mask = 1 << (idx % VSID_NBPW); idx /= VSID_NBPW; KASSERT(moea64_vsid_bitmap[idx] & mask, ("Freeing unallocated VSID %#jx", vsid)); moea64_vsid_bitmap[idx] &= ~mask; mtx_unlock(&moea64_slb_mutex); } void moea64_release(mmu_t mmu, pmap_t pmap) { /* * Free segment registers' VSIDs */ #ifdef __powerpc64__ slb_free_tree(pmap); slb_free_user_cache(pmap->pm_slb); #else KASSERT(pmap->pm_sr[0] != 0, ("moea64_release: pm_sr[0] = 0")); moea64_release_vsid(VSID_TO_HASH(pmap->pm_sr[0])); #endif } /* * Remove all pages mapped by the specified pmap */ void moea64_remove_pages(mmu_t mmu, pmap_t pm) { struct pvo_entry *pvo, *tpvo; struct pvo_tree tofree; RB_INIT(&tofree); PMAP_LOCK(pm); RB_FOREACH_SAFE(pvo, pvo_tree, &pm->pmap_pvo, tpvo) { if (pvo->pvo_vaddr & PVO_WIRED) continue; /* * For locking reasons, remove this from the page table and * pmap, but save delinking from the vm_page for a second * pass */ moea64_pvo_remove_from_pmap(mmu, pvo); RB_INSERT(pvo_tree, &tofree, pvo); } PMAP_UNLOCK(pm); RB_FOREACH_SAFE(pvo, pvo_tree, &tofree, tpvo) { PV_LOCK(pvo->pvo_pte.pa & LPTE_RPGN); moea64_pvo_remove_from_page(mmu, pvo); PV_UNLOCK(pvo->pvo_pte.pa & LPTE_RPGN); RB_REMOVE(pvo_tree, &tofree, pvo); free_pvo_entry(pvo); } } /* * Remove the given range of addresses from the specified map. */ void moea64_remove(mmu_t mmu, pmap_t pm, vm_offset_t sva, vm_offset_t eva) { struct pvo_entry *pvo, *tpvo, key; struct pvo_tree tofree; /* * Perform an unsynchronized read. This is, however, safe. */ if (pm->pm_stats.resident_count == 0) return; key.pvo_vaddr = sva; RB_INIT(&tofree); PMAP_LOCK(pm); for (pvo = RB_NFIND(pvo_tree, &pm->pmap_pvo, &key); pvo != NULL && PVO_VADDR(pvo) < eva; pvo = tpvo) { tpvo = RB_NEXT(pvo_tree, &pm->pmap_pvo, pvo); /* * For locking reasons, remove this from the page table and * pmap, but save delinking from the vm_page for a second * pass */ moea64_pvo_remove_from_pmap(mmu, pvo); RB_INSERT(pvo_tree, &tofree, pvo); } PMAP_UNLOCK(pm); RB_FOREACH_SAFE(pvo, pvo_tree, &tofree, tpvo) { PV_LOCK(pvo->pvo_pte.pa & LPTE_RPGN); moea64_pvo_remove_from_page(mmu, pvo); PV_UNLOCK(pvo->pvo_pte.pa & LPTE_RPGN); RB_REMOVE(pvo_tree, &tofree, pvo); free_pvo_entry(pvo); } } /* * Remove physical page from all pmaps in which it resides. moea64_pvo_remove() * will reflect changes in pte's back to the vm_page. */ void moea64_remove_all(mmu_t mmu, vm_page_t m) { struct pvo_entry *pvo, *next_pvo; struct pvo_head freequeue; int wasdead; pmap_t pmap; LIST_INIT(&freequeue); PV_PAGE_LOCK(m); LIST_FOREACH_SAFE(pvo, vm_page_to_pvoh(m), pvo_vlink, next_pvo) { pmap = pvo->pvo_pmap; PMAP_LOCK(pmap); wasdead = (pvo->pvo_vaddr & PVO_DEAD); if (!wasdead) moea64_pvo_remove_from_pmap(mmu, pvo); moea64_pvo_remove_from_page(mmu, pvo); if (!wasdead) LIST_INSERT_HEAD(&freequeue, pvo, pvo_vlink); PMAP_UNLOCK(pmap); } KASSERT(!pmap_page_is_mapped(m), ("Page still has mappings")); KASSERT(!(m->aflags & PGA_WRITEABLE), ("Page still writable")); PV_PAGE_UNLOCK(m); /* Clean up UMA allocations */ LIST_FOREACH_SAFE(pvo, &freequeue, pvo_vlink, next_pvo) free_pvo_entry(pvo); } /* * Allocate a physical page of memory directly from the phys_avail map. * Can only be called from moea64_bootstrap before avail start and end are * calculated. */ vm_offset_t moea64_bootstrap_alloc(vm_size_t size, u_int align) { vm_offset_t s, e; int i, j; size = round_page(size); for (i = 0; phys_avail[i + 1] != 0; i += 2) { if (align != 0) s = (phys_avail[i] + align - 1) & ~(align - 1); else s = phys_avail[i]; e = s + size; if (s < phys_avail[i] || e > phys_avail[i + 1]) continue; if (s + size > platform_real_maxaddr()) continue; if (s == phys_avail[i]) { phys_avail[i] += size; } else if (e == phys_avail[i + 1]) { phys_avail[i + 1] -= size; } else { for (j = phys_avail_count * 2; j > i; j -= 2) { phys_avail[j] = phys_avail[j - 2]; phys_avail[j + 1] = phys_avail[j - 1]; } phys_avail[i + 3] = phys_avail[i + 1]; phys_avail[i + 1] = s; phys_avail[i + 2] = e; phys_avail_count++; } return (s); } panic("moea64_bootstrap_alloc: could not allocate memory"); } static int moea64_pvo_enter(mmu_t mmu, struct pvo_entry *pvo, struct pvo_head *pvo_head) { int first, err; PMAP_LOCK_ASSERT(pvo->pvo_pmap, MA_OWNED); KASSERT(moea64_pvo_find_va(pvo->pvo_pmap, PVO_VADDR(pvo)) == NULL, ("Existing mapping for VA %#jx", (uintmax_t)PVO_VADDR(pvo))); moea64_pvo_enter_calls++; /* * Add to pmap list */ RB_INSERT(pvo_tree, &pvo->pvo_pmap->pmap_pvo, pvo); /* * Remember if the list was empty and therefore will be the first * item. */ if (pvo_head != NULL) { if (LIST_FIRST(pvo_head) == NULL) first = 1; LIST_INSERT_HEAD(pvo_head, pvo, pvo_vlink); } if (pvo->pvo_vaddr & PVO_WIRED) pvo->pvo_pmap->pm_stats.wired_count++; pvo->pvo_pmap->pm_stats.resident_count++; /* * Insert it into the hardware page table */ err = MOEA64_PTE_INSERT(mmu, pvo); if (err != 0) { panic("moea64_pvo_enter: overflow"); } moea64_pvo_entries++; if (pvo->pvo_pmap == kernel_pmap) isync(); #ifdef __powerpc64__ /* * Make sure all our bootstrap mappings are in the SLB as soon * as virtual memory is switched on. */ if (!pmap_bootstrapped) moea64_bootstrap_slb_prefault(PVO_VADDR(pvo), pvo->pvo_vaddr & PVO_LARGE); #endif return (first ? ENOENT : 0); } static void moea64_pvo_remove_from_pmap(mmu_t mmu, struct pvo_entry *pvo) { struct vm_page *pg; int32_t refchg; KASSERT(pvo->pvo_pmap != NULL, ("Trying to remove PVO with no pmap")); PMAP_LOCK_ASSERT(pvo->pvo_pmap, MA_OWNED); KASSERT(!(pvo->pvo_vaddr & PVO_DEAD), ("Trying to remove dead PVO")); /* * If there is an active pte entry, we need to deactivate it */ refchg = MOEA64_PTE_UNSET(mmu, pvo); if (refchg < 0) { /* * If it was evicted from the page table, be pessimistic and * dirty the page. */ if (pvo->pvo_pte.prot & VM_PROT_WRITE) refchg = LPTE_CHG; else refchg = 0; } /* * Update our statistics. */ pvo->pvo_pmap->pm_stats.resident_count--; if (pvo->pvo_vaddr & PVO_WIRED) pvo->pvo_pmap->pm_stats.wired_count--; /* * Remove this PVO from the pmap list. */ RB_REMOVE(pvo_tree, &pvo->pvo_pmap->pmap_pvo, pvo); /* * Mark this for the next sweep */ pvo->pvo_vaddr |= PVO_DEAD; /* Send RC bits to VM */ if ((pvo->pvo_vaddr & PVO_MANAGED) && (pvo->pvo_pte.prot & VM_PROT_WRITE)) { pg = PHYS_TO_VM_PAGE(pvo->pvo_pte.pa & LPTE_RPGN); if (pg != NULL) { refchg |= atomic_readandclear_32(&pg->md.mdpg_attrs); if (refchg & LPTE_CHG) vm_page_dirty(pg); if (refchg & LPTE_REF) vm_page_aflag_set(pg, PGA_REFERENCED); } } } static void moea64_pvo_remove_from_page(mmu_t mmu, struct pvo_entry *pvo) { struct vm_page *pg; KASSERT(pvo->pvo_vaddr & PVO_DEAD, ("Trying to delink live page")); /* Use NULL pmaps as a sentinel for races in page deletion */ if (pvo->pvo_pmap == NULL) return; pvo->pvo_pmap = NULL; /* * Update vm about page writeability/executability if managed */ PV_LOCKASSERT(pvo->pvo_pte.pa & LPTE_RPGN); pg = PHYS_TO_VM_PAGE(pvo->pvo_pte.pa & LPTE_RPGN); if ((pvo->pvo_vaddr & PVO_MANAGED) && pg != NULL) { LIST_REMOVE(pvo, pvo_vlink); if (LIST_EMPTY(vm_page_to_pvoh(pg))) vm_page_aflag_clear(pg, PGA_WRITEABLE | PGA_EXECUTABLE); } moea64_pvo_entries--; moea64_pvo_remove_calls++; } static struct pvo_entry * moea64_pvo_find_va(pmap_t pm, vm_offset_t va) { struct pvo_entry key; PMAP_LOCK_ASSERT(pm, MA_OWNED); key.pvo_vaddr = va & ~ADDR_POFF; return (RB_FIND(pvo_tree, &pm->pmap_pvo, &key)); } static boolean_t moea64_query_bit(mmu_t mmu, vm_page_t m, uint64_t ptebit) { struct pvo_entry *pvo; int64_t ret; boolean_t rv; /* * See if this bit is stored in the page already. */ if (m->md.mdpg_attrs & ptebit) return (TRUE); /* * Examine each PTE. Sync so that any pending REF/CHG bits are * flushed to the PTEs. */ rv = FALSE; powerpc_sync(); PV_PAGE_LOCK(m); LIST_FOREACH(pvo, vm_page_to_pvoh(m), pvo_vlink) { ret = 0; /* * See if this pvo has a valid PTE. if so, fetch the * REF/CHG bits from the valid PTE. If the appropriate * ptebit is set, return success. */ PMAP_LOCK(pvo->pvo_pmap); if (!(pvo->pvo_vaddr & PVO_DEAD)) ret = MOEA64_PTE_SYNCH(mmu, pvo); PMAP_UNLOCK(pvo->pvo_pmap); if (ret > 0) { atomic_set_32(&m->md.mdpg_attrs, ret & (LPTE_CHG | LPTE_REF)); if (ret & ptebit) { rv = TRUE; break; } } } PV_PAGE_UNLOCK(m); return (rv); } static u_int moea64_clear_bit(mmu_t mmu, vm_page_t m, u_int64_t ptebit) { u_int count; struct pvo_entry *pvo; int64_t ret; /* * Sync so that any pending REF/CHG bits are flushed to the PTEs (so * we can reset the right ones). */ powerpc_sync(); /* * For each pvo entry, clear the pte's ptebit. */ count = 0; PV_PAGE_LOCK(m); LIST_FOREACH(pvo, vm_page_to_pvoh(m), pvo_vlink) { ret = 0; PMAP_LOCK(pvo->pvo_pmap); if (!(pvo->pvo_vaddr & PVO_DEAD)) ret = MOEA64_PTE_CLEAR(mmu, pvo, ptebit); PMAP_UNLOCK(pvo->pvo_pmap); if (ret > 0 && (ret & ptebit)) count++; } atomic_clear_32(&m->md.mdpg_attrs, ptebit); PV_PAGE_UNLOCK(m); return (count); } boolean_t moea64_dev_direct_mapped(mmu_t mmu, vm_paddr_t pa, vm_size_t size) { struct pvo_entry *pvo, key; vm_offset_t ppa; int error = 0; PMAP_LOCK(kernel_pmap); key.pvo_vaddr = ppa = pa & ~ADDR_POFF; for (pvo = RB_FIND(pvo_tree, &kernel_pmap->pmap_pvo, &key); ppa < pa + size; ppa += PAGE_SIZE, pvo = RB_NEXT(pvo_tree, &kernel_pmap->pmap_pvo, pvo)) { if (pvo == NULL || (pvo->pvo_pte.pa & LPTE_RPGN) != ppa) { error = EFAULT; break; } } PMAP_UNLOCK(kernel_pmap); return (error); } /* * Map a set of physical memory pages into the kernel virtual * address space. Return a pointer to where it is mapped. This * routine is intended to be used for mapping device memory, * NOT real memory. */ void * moea64_mapdev_attr(mmu_t mmu, vm_offset_t pa, vm_size_t size, vm_memattr_t ma) { vm_offset_t va, tmpva, ppa, offset; ppa = trunc_page(pa); offset = pa & PAGE_MASK; size = roundup2(offset + size, PAGE_SIZE); va = kva_alloc(size); if (!va) panic("moea64_mapdev: Couldn't alloc kernel virtual memory"); for (tmpva = va; size > 0;) { moea64_kenter_attr(mmu, tmpva, ppa, ma); size -= PAGE_SIZE; tmpva += PAGE_SIZE; ppa += PAGE_SIZE; } return ((void *)(va + offset)); } void * moea64_mapdev(mmu_t mmu, vm_paddr_t pa, vm_size_t size) { return moea64_mapdev_attr(mmu, pa, size, VM_MEMATTR_DEFAULT); } void moea64_unmapdev(mmu_t mmu, vm_offset_t va, vm_size_t size) { vm_offset_t base, offset; base = trunc_page(va); offset = va & PAGE_MASK; size = roundup2(offset + size, PAGE_SIZE); kva_free(base, size); } void moea64_sync_icache(mmu_t mmu, pmap_t pm, vm_offset_t va, vm_size_t sz) { struct pvo_entry *pvo; vm_offset_t lim; vm_paddr_t pa; vm_size_t len; PMAP_LOCK(pm); while (sz > 0) { lim = round_page(va); len = MIN(lim - va, sz); pvo = moea64_pvo_find_va(pm, va & ~ADDR_POFF); if (pvo != NULL && !(pvo->pvo_pte.pa & LPTE_I)) { pa = (pvo->pvo_pte.pa & LPTE_RPGN) | (va & ADDR_POFF); moea64_syncicache(mmu, pm, va, pa, len); } va += len; sz -= len; } PMAP_UNLOCK(pm); } void moea64_dumpsys_map(mmu_t mmu, vm_paddr_t pa, size_t sz, void **va) { *va = (void *)pa; } extern struct dump_pa dump_map[PHYS_AVAIL_SZ + 1]; void moea64_scan_init(mmu_t mmu) { struct pvo_entry *pvo; vm_offset_t va; int i; if (!do_minidump) { /* Initialize phys. segments for dumpsys(). */ memset(&dump_map, 0, sizeof(dump_map)); mem_regions(&pregions, &pregions_sz, ®ions, ®ions_sz); for (i = 0; i < pregions_sz; i++) { dump_map[i].pa_start = pregions[i].mr_start; dump_map[i].pa_size = pregions[i].mr_size; } return; } /* Virtual segments for minidumps: */ memset(&dump_map, 0, sizeof(dump_map)); /* 1st: kernel .data and .bss. */ dump_map[0].pa_start = trunc_page((uintptr_t)_etext); dump_map[0].pa_size = round_page((uintptr_t)_end) - dump_map[0].pa_start; /* 2nd: msgbuf and tables (see pmap_bootstrap()). */ dump_map[1].pa_start = (vm_paddr_t)msgbufp->msg_ptr; dump_map[1].pa_size = round_page(msgbufp->msg_size); /* 3rd: kernel VM. */ va = dump_map[1].pa_start + dump_map[1].pa_size; /* Find start of next chunk (from va). */ while (va < virtual_end) { /* Don't dump the buffer cache. */ if (va >= kmi.buffer_sva && va < kmi.buffer_eva) { va = kmi.buffer_eva; continue; } pvo = moea64_pvo_find_va(kernel_pmap, va & ~ADDR_POFF); if (pvo != NULL && !(pvo->pvo_vaddr & PVO_DEAD)) break; va += PAGE_SIZE; } if (va < virtual_end) { dump_map[2].pa_start = va; va += PAGE_SIZE; /* Find last page in chunk. */ while (va < virtual_end) { /* Don't run into the buffer cache. */ if (va == kmi.buffer_sva) break; pvo = moea64_pvo_find_va(kernel_pmap, va & ~ADDR_POFF); if (pvo != NULL && !(pvo->pvo_vaddr & PVO_DEAD)) break; va += PAGE_SIZE; } dump_map[2].pa_size = va - dump_map[2].pa_start; } } Index: projects/cxl_iscsi/sys/powerpc/aim/trap.c =================================================================== --- projects/cxl_iscsi/sys/powerpc/aim/trap.c (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/aim/trap.c (revision 279717) @@ -1,754 +1,752 @@ /*- * Copyright (C) 1995, 1996 Wolfgang Solfrank. * Copyright (C) 1995, 1996 TooLs GmbH. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by TooLs GmbH. * 4. The name of TooLs GmbH may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * $NetBSD: trap.c,v 1.58 2002/03/04 04:07:35 dbj Exp $ */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include static void trap_fatal(struct trapframe *frame); static void printtrap(u_int vector, struct trapframe *frame, int isfatal, int user); static int trap_pfault(struct trapframe *frame, int user); static int fix_unaligned(struct thread *td, struct trapframe *frame); static int handle_onfault(struct trapframe *frame); static void syscall(struct trapframe *frame); #ifdef __powerpc64__ void handle_kernel_slb_spill(int, register_t, register_t); static int handle_user_slb_spill(pmap_t pm, vm_offset_t addr); extern int n_slbs; #endif struct powerpc_exception { u_int vector; char *name; }; #ifdef KDTRACE_HOOKS #include int (*dtrace_invop_jump_addr)(struct trapframe *); #endif static struct powerpc_exception powerpc_exceptions[] = { { 0x0100, "system reset" }, { 0x0200, "machine check" }, { 0x0300, "data storage interrupt" }, { 0x0380, "data segment exception" }, { 0x0400, "instruction storage interrupt" }, { 0x0480, "instruction segment exception" }, { 0x0500, "external interrupt" }, { 0x0600, "alignment" }, { 0x0700, "program" }, { 0x0800, "floating-point unavailable" }, { 0x0900, "decrementer" }, { 0x0c00, "system call" }, { 0x0d00, "trace" }, { 0x0e00, "floating-point assist" }, { 0x0f00, "performance monitoring" }, { 0x0f20, "altivec unavailable" }, { 0x0f40, "vsx unavailable" }, { 0x1000, "instruction tlb miss" }, { 0x1100, "data load tlb miss" }, { 0x1200, "data store tlb miss" }, { 0x1300, "instruction breakpoint" }, { 0x1400, "system management" }, { 0x1600, "altivec assist" }, { 0x1700, "thermal management" }, { 0x2000, "run mode/trace" }, { 0x3000, NULL } }; static const char * trapname(u_int vector) { struct powerpc_exception *pe; for (pe = powerpc_exceptions; pe->vector != 0x3000; pe++) { if (pe->vector == vector) return (pe->name); } return ("unknown"); } void trap(struct trapframe *frame) { struct thread *td; struct proc *p; #ifdef KDTRACE_HOOKS uint32_t inst; #endif int sig, type, user; u_int ucode; ksiginfo_t ksi; PCPU_INC(cnt.v_trap); td = curthread; p = td->td_proc; type = ucode = frame->exc; sig = 0; user = frame->srr1 & PSL_PR; CTR3(KTR_TRAP, "trap: %s type=%s (%s)", td->td_name, trapname(type), user ? "user" : "kernel"); #ifdef KDTRACE_HOOKS /* * A trap can occur while DTrace executes a probe. Before * executing the probe, DTrace blocks re-scheduling and sets * a flag in its per-cpu flags to indicate that it doesn't * want to fault. On returning from the probe, the no-fault * flag is cleared and finally re-scheduling is enabled. * * If the DTrace kernel module has registered a trap handler, * call it and if it returns non-zero, assume that it has * handled the trap and modified the trap frame so that this * function can return normally. */ if (dtrace_trap_func != NULL && (*dtrace_trap_func)(frame, type) != 0) return; #endif if (user) { td->td_pticks = 0; td->td_frame = frame; if (td->td_ucred != p->p_ucred) cred_update_thread(td); /* User Mode Traps */ switch (type) { case EXC_RUNMODETRC: case EXC_TRC: frame->srr1 &= ~PSL_SE; sig = SIGTRAP; ucode = TRAP_TRACE; break; #ifdef __powerpc64__ case EXC_ISE: case EXC_DSE: if (handle_user_slb_spill(&p->p_vmspace->vm_pmap, - (type == EXC_ISE) ? frame->srr0 : - frame->cpu.aim.dar) != 0) { + (type == EXC_ISE) ? frame->srr0 : frame->dar) != 0){ sig = SIGSEGV; ucode = SEGV_MAPERR; } break; #endif case EXC_DSI: case EXC_ISI: sig = trap_pfault(frame, 1); if (sig == SIGSEGV) ucode = SEGV_MAPERR; break; case EXC_SC: syscall(frame); break; case EXC_FPU: KASSERT((td->td_pcb->pcb_flags & PCB_FPU) != PCB_FPU, ("FPU already enabled for thread")); enable_fpu(td); break; case EXC_VEC: KASSERT((td->td_pcb->pcb_flags & PCB_VEC) != PCB_VEC, ("Altivec already enabled for thread")); enable_vec(td); break; case EXC_VSX: KASSERT((td->td_pcb->pcb_flags & PCB_VSX) != PCB_VSX, ("VSX already enabled for thread")); if (!(td->td_pcb->pcb_flags & PCB_VEC)) enable_vec(td); if (!(td->td_pcb->pcb_flags & PCB_FPU)) save_fpu(td); td->td_pcb->pcb_flags |= PCB_VSX; enable_fpu(td); break; case EXC_VECAST_G4: case EXC_VECAST_G5: /* * We get a VPU assist exception for IEEE mode * vector operations on denormalized floats. * Emulating this is a giant pain, so for now, * just switch off IEEE mode and treat them as * zero. */ save_vec(td); td->td_pcb->pcb_vec.vscr |= ALTIVEC_VSCR_NJ; enable_vec(td); break; case EXC_ALI: if (fix_unaligned(td, frame) != 0) { sig = SIGBUS; ucode = BUS_ADRALN; } else frame->srr0 += 4; break; case EXC_PGM: /* Identify the trap reason */ if (frame->srr1 & EXC_PGM_TRAP) { #ifdef KDTRACE_HOOKS inst = fuword32((const void *)frame->srr0); if (inst == 0x0FFFDDDD && dtrace_pid_probe_ptr != NULL) { struct reg regs; fill_regs(td, ®s); (*dtrace_pid_probe_ptr)(®s); break; } #endif sig = SIGTRAP; ucode = TRAP_BRKPT; } else { sig = ppc_instr_emulate(frame, td->td_pcb); if (sig == SIGILL) { if (frame->srr1 & EXC_PGM_PRIV) ucode = ILL_PRVOPC; else if (frame->srr1 & EXC_PGM_ILLEGAL) ucode = ILL_ILLOPC; } else if (sig == SIGFPE) ucode = FPE_FLTINV; /* Punt for now, invalid operation. */ } break; case EXC_MCHK: /* * Note that this may not be recoverable for the user * process, depending on the type of machine check, * but it at least prevents the kernel from dying. */ sig = SIGBUS; ucode = BUS_OBJERR; break; default: trap_fatal(frame); } } else { /* Kernel Mode Traps */ KASSERT(cold || td->td_ucred != NULL, ("kernel trap doesn't have ucred")); switch (type) { #ifdef KDTRACE_HOOKS case EXC_PGM: if (frame->srr1 & EXC_PGM_TRAP) { if (*(uint32_t *)frame->srr0 == EXC_DTRACE) { if (dtrace_invop_jump_addr != NULL) { dtrace_invop_jump_addr(frame); return; } } } break; #endif #ifdef __powerpc64__ case EXC_DSE: - if ((frame->cpu.aim.dar & SEGMENT_MASK) == USER_ADDR) { + if ((frame->dar & SEGMENT_MASK) == USER_ADDR) { __asm __volatile ("slbmte %0, %1" :: "r"(td->td_pcb->pcb_cpu.aim.usr_vsid), "r"(USER_SLB_SLBE)); return; } break; #endif case EXC_DSI: if (trap_pfault(frame, 0) == 0) return; break; case EXC_MCHK: if (handle_onfault(frame)) return; break; default: break; } trap_fatal(frame); } if (sig != 0) { if (p->p_sysent->sv_transtrap != NULL) sig = (p->p_sysent->sv_transtrap)(sig, type); ksiginfo_init_trap(&ksi); ksi.ksi_signo = sig; ksi.ksi_code = (int) ucode; /* XXX, not POSIX */ /* ksi.ksi_addr = ? */ ksi.ksi_trapno = type; trapsignal(td, &ksi); } userret(td, frame); } static void trap_fatal(struct trapframe *frame) { printtrap(frame->exc, frame, 1, (frame->srr1 & PSL_PR)); #ifdef KDB if ((debugger_on_panic || kdb_active) && kdb_trap(frame->exc, 0, frame)) return; #endif panic("%s trap", trapname(frame->exc)); } static void printtrap(u_int vector, struct trapframe *frame, int isfatal, int user) { printf("\n"); printf("%s %s trap:\n", isfatal ? "fatal" : "handled", user ? "user" : "kernel"); printf("\n"); printf(" exception = 0x%x (%s)\n", vector, trapname(vector)); switch (vector) { case EXC_DSE: case EXC_DSI: - printf(" virtual address = 0x%" PRIxPTR "\n", - frame->cpu.aim.dar); + printf(" virtual address = 0x%" PRIxPTR "\n", frame->dar); printf(" dsisr = 0x%" PRIxPTR "\n", frame->cpu.aim.dsisr); break; case EXC_ISE: case EXC_ISI: printf(" virtual address = 0x%" PRIxPTR "\n", frame->srr0); break; } printf(" srr0 = 0x%" PRIxPTR "\n", frame->srr0); printf(" srr1 = 0x%" PRIxPTR "\n", frame->srr1); printf(" lr = 0x%" PRIxPTR "\n", frame->lr); printf(" curthread = %p\n", curthread); if (curthread != NULL) printf(" pid = %d, comm = %s\n", curthread->td_proc->p_pid, curthread->td_name); printf("\n"); } /* * Handles a fatal fault when we have onfault state to recover. Returns * non-zero if there was onfault recovery state available. */ static int handle_onfault(struct trapframe *frame) { struct thread *td; faultbuf *fb; td = curthread; fb = td->td_pcb->pcb_onfault; if (fb != NULL) { frame->srr0 = (*fb)[0]; frame->fixreg[1] = (*fb)[1]; frame->fixreg[2] = (*fb)[2]; frame->fixreg[3] = 1; frame->cr = (*fb)[3]; bcopy(&(*fb)[4], &frame->fixreg[13], 19 * sizeof(register_t)); return (1); } return (0); } int cpu_fetch_syscall_args(struct thread *td, struct syscall_args *sa) { struct proc *p; struct trapframe *frame; caddr_t params; size_t argsz; int error, n, i; p = td->td_proc; frame = td->td_frame; sa->code = frame->fixreg[0]; params = (caddr_t)(frame->fixreg + FIRSTARG); n = NARGREG; if (sa->code == SYS_syscall) { /* * code is first argument, * followed by actual args. */ sa->code = *(register_t *) params; params += sizeof(register_t); n -= 1; } else if (sa->code == SYS___syscall) { /* * Like syscall, but code is a quad, * so as to maintain quad alignment * for the rest of the args. */ if (SV_PROC_FLAG(p, SV_ILP32)) { params += sizeof(register_t); sa->code = *(register_t *) params; params += sizeof(register_t); n -= 2; } else { sa->code = *(register_t *) params; params += sizeof(register_t); n -= 1; } } if (p->p_sysent->sv_mask) sa->code &= p->p_sysent->sv_mask; if (sa->code >= p->p_sysent->sv_size) sa->callp = &p->p_sysent->sv_table[0]; else sa->callp = &p->p_sysent->sv_table[sa->code]; sa->narg = sa->callp->sy_narg; if (SV_PROC_FLAG(p, SV_ILP32)) { argsz = sizeof(uint32_t); for (i = 0; i < n; i++) sa->args[i] = ((u_register_t *)(params))[i] & 0xffffffff; } else { argsz = sizeof(uint64_t); for (i = 0; i < n; i++) sa->args[i] = ((u_register_t *)(params))[i]; } if (sa->narg > n) error = copyin(MOREARGS(frame->fixreg[1]), sa->args + n, (sa->narg - n) * argsz); else error = 0; #ifdef __powerpc64__ if (SV_PROC_FLAG(p, SV_ILP32) && sa->narg > n) { /* Expand the size of arguments copied from the stack */ for (i = sa->narg; i >= n; i--) sa->args[i] = ((uint32_t *)(&sa->args[n]))[i-n]; } #endif if (error == 0) { td->td_retval[0] = 0; td->td_retval[1] = frame->fixreg[FIRSTARG + 1]; } return (error); } #include "../../kern/subr_syscall.c" void syscall(struct trapframe *frame) { struct thread *td; struct syscall_args sa; int error; td = curthread; td->td_frame = frame; #ifdef __powerpc64__ /* * Speculatively restore last user SLB segment, which we know is * invalid already, since we are likely to do copyin()/copyout(). */ __asm __volatile ("slbmte %0, %1; isync" :: "r"(td->td_pcb->pcb_cpu.aim.usr_vsid), "r"(USER_SLB_SLBE)); #endif error = syscallenter(td, &sa); syscallret(td, error, &sa); } #ifdef __powerpc64__ /* Handle kernel SLB faults -- runs in real mode, all seat belts off */ void handle_kernel_slb_spill(int type, register_t dar, register_t srr0) { struct slb *slbcache; uint64_t slbe, slbv; uint64_t esid, addr; int i; addr = (type == EXC_ISE) ? srr0 : dar; slbcache = PCPU_GET(slb); esid = (uintptr_t)addr >> ADDR_SR_SHFT; slbe = (esid << SLBE_ESID_SHIFT) | SLBE_VALID; /* See if the hardware flushed this somehow (can happen in LPARs) */ for (i = 0; i < n_slbs; i++) if (slbcache[i].slbe == (slbe | (uint64_t)i)) return; /* Not in the map, needs to actually be added */ slbv = kernel_va_to_slbv(addr); if (slbcache[USER_SLB_SLOT].slbe == 0) { for (i = 0; i < n_slbs; i++) { if (i == USER_SLB_SLOT) continue; if (!(slbcache[i].slbe & SLBE_VALID)) goto fillkernslb; } if (i == n_slbs) slbcache[USER_SLB_SLOT].slbe = 1; } /* Sacrifice a random SLB entry that is not the user entry */ i = mftb() % n_slbs; if (i == USER_SLB_SLOT) i = (i+1) % n_slbs; fillkernslb: /* Write new entry */ slbcache[i].slbv = slbv; slbcache[i].slbe = slbe | (uint64_t)i; /* Trap handler will restore from cache on exit */ } static int handle_user_slb_spill(pmap_t pm, vm_offset_t addr) { struct slb *user_entry; uint64_t esid; int i; esid = (uintptr_t)addr >> ADDR_SR_SHFT; PMAP_LOCK(pm); user_entry = user_va_to_slb_entry(pm, addr); if (user_entry == NULL) { /* allocate_vsid auto-spills it */ (void)allocate_user_vsid(pm, esid, 0); } else { /* * Check that another CPU has not already mapped this. * XXX: Per-thread SLB caches would be better. */ for (i = 0; i < pm->pm_slb_len; i++) if (pm->pm_slb[i] == user_entry) break; if (i == pm->pm_slb_len) slb_insert_user(pm, user_entry); } PMAP_UNLOCK(pm); return (0); } #endif static int trap_pfault(struct trapframe *frame, int user) { vm_offset_t eva, va; struct thread *td; struct proc *p; vm_map_t map; vm_prot_t ftype; int rv; register_t user_sr; td = curthread; p = td->td_proc; if (frame->exc == EXC_ISI) { eva = frame->srr0; ftype = VM_PROT_EXECUTE; if (frame->srr1 & SRR1_ISI_PFAULT) ftype |= VM_PROT_READ; } else { - eva = frame->cpu.aim.dar; + eva = frame->dar; if (frame->cpu.aim.dsisr & DSISR_STORE) ftype = VM_PROT_WRITE; else ftype = VM_PROT_READ; } if (user) { map = &p->p_vmspace->vm_map; } else { if ((eva >> ADDR_SR_SHFT) == (USER_ADDR >> ADDR_SR_SHFT)) { if (p->p_vmspace == NULL) return (SIGSEGV); map = &p->p_vmspace->vm_map; user_sr = td->td_pcb->pcb_cpu.aim.usr_segm; eva &= ADDR_PIDX | ADDR_POFF; eva |= user_sr << ADDR_SR_SHFT; } else { map = kernel_map; } } va = trunc_page(eva); if (map != kernel_map) { /* * Keep swapout from messing with us during this * critical time. */ PROC_LOCK(p); ++p->p_lock; PROC_UNLOCK(p); /* Fault in the user page: */ rv = vm_fault(map, va, ftype, VM_FAULT_NORMAL); PROC_LOCK(p); --p->p_lock; PROC_UNLOCK(p); /* * XXXDTRACE: add dtrace_doubletrap_func here? */ } else { /* * Don't have to worry about process locking or stacks in the * kernel. */ rv = vm_fault(map, va, ftype, VM_FAULT_NORMAL); } if (rv == KERN_SUCCESS) return (0); if (!user && handle_onfault(frame)) return (0); return (SIGSEGV); } /* * For now, this only deals with the particular unaligned access case * that gcc tends to generate. Eventually it should handle all of the * possibilities that can happen on a 32-bit PowerPC in big-endian mode. */ static int fix_unaligned(struct thread *td, struct trapframe *frame) { struct thread *fputhread; int indicator, reg; double *fpr; indicator = EXC_ALI_OPCODE_INDICATOR(frame->cpu.aim.dsisr); switch (indicator) { case EXC_ALI_LFD: case EXC_ALI_STFD: reg = EXC_ALI_RST(frame->cpu.aim.dsisr); fpr = &td->td_pcb->pcb_fpu.fpr[reg].fpr; fputhread = PCPU_GET(fputhread); /* Juggle the FPU to ensure that we've initialized * the FPRs, and that their current state is in * the PCB. */ if (fputhread != td) { if (fputhread) save_fpu(fputhread); enable_fpu(td); } save_fpu(td); if (indicator == EXC_ALI_LFD) { - if (copyin((void *)frame->cpu.aim.dar, fpr, + if (copyin((void *)frame->dar, fpr, sizeof(double)) != 0) return -1; enable_fpu(td); } else { - if (copyout(fpr, (void *)frame->cpu.aim.dar, + if (copyout(fpr, (void *)frame->dar, sizeof(double)) != 0) return -1; } return 0; break; } return -1; } Index: projects/cxl_iscsi/sys/powerpc/booke/interrupt.c =================================================================== --- projects/cxl_iscsi/sys/powerpc/booke/interrupt.c (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/booke/interrupt.c (revision 279717) @@ -1,144 +1,144 @@ /*- * Copyright (C) 2006 Semihalf, Rafal Jaworowski * Copyright 2002 by Peter Grehan. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. The name of the author may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * */ /* * Interrupts are dispatched to here from locore asm */ #include /* RCS ID & Copyright macro defns */ __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "pic_if.h" extern void decr_intr(struct trapframe *); void powerpc_decr_interrupt(struct trapframe *); void powerpc_extr_interrupt(struct trapframe *); void powerpc_crit_interrupt(struct trapframe *); void powerpc_mchk_interrupt(struct trapframe *); static void dump_frame(struct trapframe *framep); static void dump_frame(struct trapframe *frame) { int i; printf("\n*** *** STACK FRAME DUMP *** ***\n"); printf(" exc = 0x%x\n", frame->exc); printf(" srr0 = 0x%08x\n", frame->srr0); printf(" srr1 = 0x%08x\n", frame->srr1); - printf(" dear = 0x%08x\n", frame->cpu.booke.dear); + printf(" dear = 0x%08x\n", frame->dar); printf(" esr = 0x%08x\n", frame->cpu.booke.esr); printf(" lr = 0x%08x\n", frame->lr); printf(" cr = 0x%08x\n", frame->cr); printf(" sp = 0x%08x\n", frame->fixreg[1]); for (i = 0; i < 32; i++) { printf(" R%02d = 0x%08x", i, frame->fixreg[i]); if ((i & 0x3) == 3) printf("\n"); } printf("\n"); } void powerpc_crit_interrupt(struct trapframe *framep) { printf("powerpc_crit_interrupt: critical interrupt!\n"); dump_frame(framep); trap(framep); } void powerpc_mchk_interrupt(struct trapframe *framep) { printf("powerpc_mchk_interrupt: machine check interrupt!\n"); dump_frame(framep); trap(framep); } /* * Decrementer interrupt routine */ void powerpc_decr_interrupt(struct trapframe *framep) { struct thread *td; struct trapframe *oldframe; td = curthread; critical_enter(); atomic_add_int(&td->td_intr_nesting_level, 1); oldframe = td->td_intr_frame; td->td_intr_frame = framep; decr_intr(framep); td->td_intr_frame = oldframe; atomic_subtract_int(&td->td_intr_nesting_level, 1); critical_exit(); framep->srr1 &= ~PSL_WE; } /* * External input interrupt routine */ void powerpc_extr_interrupt(struct trapframe *framep) { critical_enter(); PIC_DISPATCH(root_pic, framep); critical_exit(); framep->srr1 &= ~PSL_WE; } Index: projects/cxl_iscsi/sys/powerpc/booke/locore.S =================================================================== --- projects/cxl_iscsi/sys/powerpc/booke/locore.S (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/booke/locore.S (revision 279717) @@ -1,769 +1,733 @@ /*- * Copyright (C) 2007-2009 Semihalf, Rafal Jaworowski * Copyright (C) 2006 Semihalf, Marian Balakowicz * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN * NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * $FreeBSD$ */ #include "assym.s" #include #include #include #include #include #include #include #include #define TMPSTACKSZ 16384 .text .globl btext btext: /* * This symbol is here for the benefit of kvm_mkdb, and is supposed to * mark the start of kernel text. */ .globl kernel_text kernel_text: /* * Startup entry. Note, this must be the first thing in the text segment! */ .text .globl __start __start: /* * Assumptions on the boot loader: * - system memory starts from physical address 0 * - it's mapped by a single TBL1 entry * - TLB1 mapping is 1:1 pa to va * - kernel is loaded at 16MB boundary * - all PID registers are set to the same value * - CPU is running in AS=0 * * Registers contents provided by the loader(8): * r1 : stack pointer * r3 : metadata pointer * * We rearrange the TLB1 layout as follows: * - find TLB1 entry we started in * - make sure it's protected, ivalidate other entries * - create temp entry in the second AS (make sure it's not TLB[1]) * - switch to temp mapping * - map 16MB of RAM in TLB1[1] * - use AS=1, set EPN to KERNBASE and RPN to kernel load address * - switch to to TLB1[1] mapping * - invalidate temp mapping * * locore registers use: * r1 : stack pointer * r2 : trace pointer (AP only, for early diagnostics) * r3-r27 : scratch registers * r28 : temp TLB1 entry * r29 : initial TLB1 entry we started in * r30-r31 : arguments (metadata pointer) */ /* * Keep arguments in r30 & r31 for later use. */ mr %r30, %r3 mr %r31, %r4 /* * Initial cleanup */ li %r3, PSL_DE /* Keep debug exceptions for CodeWarrior. */ mtmsr %r3 isync lis %r3, HID0_E500_DEFAULT_SET@h ori %r3, %r3, HID0_E500_DEFAULT_SET@l mtspr SPR_HID0, %r3 isync lis %r3, HID1_E500_DEFAULT_SET@h ori %r3, %r3, HID1_E500_DEFAULT_SET@l mtspr SPR_HID1, %r3 isync /* Invalidate all entries in TLB0 */ li %r3, 0 bl tlb_inval_all cmpwi %r30, 0 beq done_mapping /* * Locate the TLB1 entry that maps this code */ bl 1f 1: mflr %r3 bl tlb1_find_current /* the entry found is returned in r29 */ bl tlb1_inval_all_but_current /* * Create temporary mapping in AS=1 and switch to it */ addi %r3, %r29, 1 bl tlb1_temp_mapping_as1 mfmsr %r3 ori %r3, %r3, (PSL_IS | PSL_DS) bl 2f 2: mflr %r4 addi %r4, %r4, 20 mtspr SPR_SRR0, %r4 mtspr SPR_SRR1, %r3 rfi /* Switch context */ /* * Invalidate initial entry */ mr %r3, %r29 bl tlb1_inval_entry /* * Setup final mapping in TLB1[1] and switch to it */ /* Final kernel mapping, map in 16 MB of RAM */ lis %r3, MAS0_TLBSEL1@h /* Select TLB1 */ li %r4, 0 /* Entry 0 */ rlwimi %r3, %r4, 16, 12, 15 mtspr SPR_MAS0, %r3 isync li %r3, (TLB_SIZE_64M << MAS1_TSIZE_SHIFT)@l oris %r3, %r3, (MAS1_VALID | MAS1_IPROT)@h mtspr SPR_MAS1, %r3 /* note TS was not filled, so it's TS=0 */ isync lis %r3, KERNBASE@h ori %r3, %r3, KERNBASE@l /* EPN = KERNBASE */ #ifdef SMP ori %r3, %r3, MAS2_M@l /* WIMGE = 0b00100 */ #endif mtspr SPR_MAS2, %r3 isync /* Discover phys load address */ bl 3f 3: mflr %r4 /* Use current address */ rlwinm %r4, %r4, 0, 0, 7 /* 16MB alignment mask */ ori %r4, %r4, (MAS3_SX | MAS3_SW | MAS3_SR)@l mtspr SPR_MAS3, %r4 /* Set RPN and protection */ isync tlbwe isync msync /* Switch to the above TLB1[1] mapping */ bl 4f 4: mflr %r4 rlwinm %r4, %r4, 0, 8, 31 /* Current offset from kernel load address */ rlwinm %r3, %r3, 0, 0, 19 add %r4, %r4, %r3 /* Convert to kernel virtual address */ addi %r4, %r4, 36 li %r3, PSL_DE /* Note AS=0 */ mtspr SPR_SRR0, %r4 mtspr SPR_SRR1, %r3 rfi /* * Invalidate temp mapping */ mr %r3, %r28 bl tlb1_inval_entry done_mapping: /* * Setup a temporary stack */ lis %r1, tmpstack@ha addi %r1, %r1, tmpstack@l addi %r1, %r1, (TMPSTACKSZ - 16) /* * Initialise exception vector offsets */ bl ivor_setup /* * Set up arguments and jump to system initialization code */ mr %r3, %r30 mr %r4, %r31 /* Prepare core */ bl booke_init /* Switch to thread0.td_kstack now */ mr %r1, %r3 li %r3, 0 stw %r3, 0(%r1) /* Machine independet part, does not return */ bl mi_startup /* NOT REACHED */ 5: b 5b #ifdef SMP /************************************************************************/ /* AP Boot page */ /************************************************************************/ .text .globl __boot_page .align 12 __boot_page: bl 1f .globl bp_ntlb1s bp_ntlb1s: .long 0 .globl bp_tlb1 bp_tlb1: .space 4 * 3 * 16 .globl bp_tlb1_end bp_tlb1_end: /* * Initial configuration */ 1: mflr %r31 /* r31 hold the address of bp_ntlb1s */ /* Set HIDs */ lis %r3, HID0_E500_DEFAULT_SET@h ori %r3, %r3, HID0_E500_DEFAULT_SET@l mtspr SPR_HID0, %r3 isync lis %r3, HID1_E500_DEFAULT_SET@h ori %r3, %r3, HID1_E500_DEFAULT_SET@l mtspr SPR_HID1, %r3 isync /* Enable branch prediction */ li %r3, BUCSR_BPEN mtspr SPR_BUCSR, %r3 isync /* Invalidate all entries in TLB0 */ li %r3, 0 bl tlb_inval_all /* * Find TLB1 entry which is translating us now */ bl 2f 2: mflr %r3 bl tlb1_find_current /* the entry number found is in r29 */ bl tlb1_inval_all_but_current /* * Create temporary translation in AS=1 and switch to it */ lwz %r3, 0(%r31) bl tlb1_temp_mapping_as1 mfmsr %r3 ori %r3, %r3, (PSL_IS | PSL_DS) bl 3f 3: mflr %r4 addi %r4, %r4, 20 mtspr SPR_SRR0, %r4 mtspr SPR_SRR1, %r3 rfi /* Switch context */ /* * Invalidate initial entry */ mr %r3, %r29 bl tlb1_inval_entry /* * Setup final mapping in TLB1[1] and switch to it */ lwz %r6, 0(%r31) addi %r5, %r31, 4 li %r4, 0 4: lis %r3, MAS0_TLBSEL1@h rlwimi %r3, %r4, 16, 12, 15 mtspr SPR_MAS0, %r3 isync lwz %r3, 0(%r5) mtspr SPR_MAS1, %r3 isync lwz %r3, 4(%r5) mtspr SPR_MAS2, %r3 isync lwz %r3, 8(%r5) mtspr SPR_MAS3, %r3 isync tlbwe isync msync addi %r5, %r5, 12 addi %r4, %r4, 1 cmpw %r4, %r6 blt 4b /* Switch to the final mapping */ lis %r5, __boot_page@ha ori %r5, %r5, __boot_page@l bl 5f 5: mflr %r3 rlwinm %r3, %r3, 0, 0xfff /* Offset from boot page start */ add %r3, %r3, %r5 /* Make this virtual address */ addi %r3, %r3, 32 li %r4, 0 /* Note AS=0 */ mtspr SPR_SRR0, %r3 mtspr SPR_SRR1, %r4 rfi /* * At this point we're running at virtual addresses KERNBASE and beyond so * it's allowed to directly access all locations the kernel was linked * against. */ /* * Invalidate temp mapping */ mr %r3, %r28 bl tlb1_inval_entry /* * Setup a temporary stack */ lis %r1, tmpstack@ha addi %r1, %r1, tmpstack@l addi %r1, %r1, (TMPSTACKSZ - 16) /* * Initialise exception vector offsets */ bl ivor_setup /* * Assign our pcpu instance */ lis %r3, ap_pcpu@h ori %r3, %r3, ap_pcpu@l lwz %r3, 0(%r3) mtsprg0 %r3 bl pmap_bootstrap_ap bl cpudep_ap_bootstrap /* Switch to the idle thread's kstack */ mr %r1, %r3 bl machdep_ap_bootstrap /* NOT REACHED */ 6: b 6b #endif /* SMP */ /* * Invalidate all entries in the given TLB. * * r3 TLBSEL */ tlb_inval_all: rlwinm %r3, %r3, 3, 0x18 /* TLBSEL */ ori %r3, %r3, 0x4 /* INVALL */ tlbivax 0, %r3 isync msync tlbsync msync blr /* * expects address to look up in r3, returns entry number in r29 * * FIXME: the hidden assumption is we are now running in AS=0, but we should * retrieve actual AS from MSR[IS|DS] and put it in MAS6[SAS] */ tlb1_find_current: mfspr %r17, SPR_PID0 slwi %r17, %r17, MAS6_SPID0_SHIFT mtspr SPR_MAS6, %r17 isync tlbsx 0, %r3 mfspr %r17, SPR_MAS0 rlwinm %r29, %r17, 16, 20, 31 /* MAS0[ESEL] -> r29 */ /* Make sure we have IPROT set on the entry */ mfspr %r17, SPR_MAS1 oris %r17, %r17, MAS1_IPROT@h mtspr SPR_MAS1, %r17 isync tlbwe isync msync blr /* * Invalidates a single entry in TLB1. * * r3 ESEL * r4-r5 scratched */ tlb1_inval_entry: lis %r4, MAS0_TLBSEL1@h /* Select TLB1 */ rlwimi %r4, %r3, 16, 12, 15 /* Select our entry */ mtspr SPR_MAS0, %r4 isync tlbre li %r5, 0 /* MAS1[V] = 0 */ mtspr SPR_MAS1, %r5 isync tlbwe isync msync blr /* * r3 entry of temp translation * r29 entry of current translation * r28 returns temp entry passed in r3 * r4-r5 scratched */ tlb1_temp_mapping_as1: mr %r28, %r3 /* Read our current translation */ lis %r3, MAS0_TLBSEL1@h /* Select TLB1 */ rlwimi %r3, %r29, 16, 12, 15 /* Select our current entry */ mtspr SPR_MAS0, %r3 isync tlbre /* Prepare and write temp entry */ lis %r3, MAS0_TLBSEL1@h /* Select TLB1 */ rlwimi %r3, %r28, 16, 12, 15 /* Select temp entry */ mtspr SPR_MAS0, %r3 isync mfspr %r5, SPR_MAS1 li %r4, 1 /* AS=1 */ rlwimi %r5, %r4, 12, 19, 19 li %r4, 0 /* Global mapping, TID=0 */ rlwimi %r5, %r4, 16, 8, 15 oris %r5, %r5, (MAS1_VALID | MAS1_IPROT)@h mtspr SPR_MAS1, %r5 isync tlbwe isync msync blr /* * Loops over TLB1, invalidates all entries skipping the one which currently * maps this code. * * r29 current entry * r3-r5 scratched */ tlb1_inval_all_but_current: mr %r6, %r3 mfspr %r3, SPR_TLB1CFG /* Get number of entries */ andi. %r3, %r3, TLBCFG_NENTRY_MASK@l li %r4, 0 /* Start from Entry 0 */ 1: lis %r5, MAS0_TLBSEL1@h rlwimi %r5, %r4, 16, 12, 15 mtspr SPR_MAS0, %r5 isync tlbre mfspr %r5, SPR_MAS1 cmpw %r4, %r29 /* our current entry? */ beq 2f rlwinm %r5, %r5, 0, 2, 31 /* clear VALID and IPROT bits */ mtspr SPR_MAS1, %r5 isync tlbwe isync msync 2: addi %r4, %r4, 1 cmpw %r4, %r3 /* Check if this is the last entry */ bne 1b blr #ifdef SMP __boot_page_padding: /* * Boot page needs to be exactly 4K, with the last word of this page * acting as the reset vector, so we need to stuff the remainder. * Upon release from holdoff CPU fetches the last word of the boot * page. */ .space 4092 - (__boot_page_padding - __boot_page) b __boot_page #endif /* SMP */ /************************************************************************/ /* locore subroutines */ /************************************************************************/ -ivor_setup: - /* Set base address of interrupt handler routines */ - lis %r3, interrupt_vector_base@h - mtspr SPR_IVPR, %r3 - - /* Assign interrupt handler routines offsets */ - li %r3, int_critical_input@l - mtspr SPR_IVOR0, %r3 - li %r3, int_machine_check@l - mtspr SPR_IVOR1, %r3 - li %r3, int_data_storage@l - mtspr SPR_IVOR2, %r3 - li %r3, int_instr_storage@l - mtspr SPR_IVOR3, %r3 - li %r3, int_external_input@l - mtspr SPR_IVOR4, %r3 - li %r3, int_alignment@l - mtspr SPR_IVOR5, %r3 - li %r3, int_program@l - mtspr SPR_IVOR6, %r3 - li %r3, int_syscall@l - mtspr SPR_IVOR8, %r3 - li %r3, int_decrementer@l - mtspr SPR_IVOR10, %r3 - li %r3, int_fixed_interval_timer@l - mtspr SPR_IVOR11, %r3 - li %r3, int_watchdog@l - mtspr SPR_IVOR12, %r3 - li %r3, int_data_tlb_error@l - mtspr SPR_IVOR13, %r3 - li %r3, int_inst_tlb_error@l - mtspr SPR_IVOR14, %r3 - li %r3, int_debug@l - mtspr SPR_IVOR15, %r3 - blr - /* * void tid_flush(tlbtid_t tid); * * Invalidate all TLB0 entries which match the given TID. Note this is * dedicated for cases when invalidation(s) should NOT be propagated to other * CPUs. * * Global vars tlb0_ways, tlb0_entries_per_way are assumed to have been set up * correctly (by tlb0_get_tlbconf()). * */ ENTRY(tid_flush) cmpwi %r3, TID_KERNEL beq tid_flush_end /* don't evict kernel translations */ /* Number of TLB0 ways */ lis %r4, tlb0_ways@h ori %r4, %r4, tlb0_ways@l lwz %r4, 0(%r4) /* Number of entries / way */ lis %r5, tlb0_entries_per_way@h ori %r5, %r5, tlb0_entries_per_way@l lwz %r5, 0(%r5) /* Disable interrupts */ mfmsr %r10 wrteei 0 li %r6, 0 /* ways counter */ loop_ways: li %r7, 0 /* entries [per way] counter */ loop_entries: /* Select TLB0 and ESEL (way) */ lis %r8, MAS0_TLBSEL0@h rlwimi %r8, %r6, 16, 14, 15 mtspr SPR_MAS0, %r8 isync /* Select EPN (entry within the way) */ rlwinm %r8, %r7, 12, 13, 19 mtspr SPR_MAS2, %r8 isync tlbre /* Check if valid entry */ mfspr %r8, SPR_MAS1 andis. %r9, %r8, MAS1_VALID@h beq next_entry /* invalid entry */ /* Check if this is our TID */ rlwinm %r9, %r8, 16, 24, 31 cmplw %r9, %r3 bne next_entry /* not our TID */ /* Clear VALID bit */ rlwinm %r8, %r8, 0, 1, 31 mtspr SPR_MAS1, %r8 isync tlbwe isync msync next_entry: addi %r7, %r7, 1 cmpw %r7, %r5 bne loop_entries /* Next way */ addi %r6, %r6, 1 cmpw %r6, %r4 bne loop_ways /* Restore MSR (possibly re-enable interrupts) */ mtmsr %r10 isync tid_flush_end: blr /* * Cache disable/enable/inval sequences according * to section 2.16 of E500CORE RM. */ ENTRY(dcache_inval) /* Invalidate d-cache */ mfspr %r3, SPR_L1CSR0 ori %r3, %r3, (L1CSR0_DCFI | L1CSR0_DCLFR)@l msync isync mtspr SPR_L1CSR0, %r3 isync 1: mfspr %r3, SPR_L1CSR0 andi. %r3, %r3, L1CSR0_DCFI bne 1b blr ENTRY(dcache_disable) /* Disable d-cache */ mfspr %r3, SPR_L1CSR0 li %r4, L1CSR0_DCE@l not %r4, %r4 and %r3, %r3, %r4 msync isync mtspr SPR_L1CSR0, %r3 isync blr ENTRY(dcache_enable) /* Enable d-cache */ mfspr %r3, SPR_L1CSR0 oris %r3, %r3, (L1CSR0_DCPE | L1CSR0_DCE)@h ori %r3, %r3, (L1CSR0_DCPE | L1CSR0_DCE)@l msync isync mtspr SPR_L1CSR0, %r3 isync blr ENTRY(icache_inval) /* Invalidate i-cache */ mfspr %r3, SPR_L1CSR1 ori %r3, %r3, (L1CSR1_ICFI | L1CSR1_ICLFR)@l isync mtspr SPR_L1CSR1, %r3 isync 1: mfspr %r3, SPR_L1CSR1 andi. %r3, %r3, L1CSR1_ICFI bne 1b blr ENTRY(icache_disable) /* Disable i-cache */ mfspr %r3, SPR_L1CSR1 li %r4, L1CSR1_ICE@l not %r4, %r4 and %r3, %r3, %r4 isync mtspr SPR_L1CSR1, %r3 isync blr ENTRY(icache_enable) /* Enable i-cache */ mfspr %r3, SPR_L1CSR1 oris %r3, %r3, (L1CSR1_ICPE | L1CSR1_ICE)@h ori %r3, %r3, (L1CSR1_ICPE | L1CSR1_ICE)@l isync mtspr SPR_L1CSR1, %r3 isync blr /* * int setfault() * * Similar to setjmp to setup for handling faults on accesses to user memory. * Any routine using this may only call bcopy, either the form below, * or the (currently used) C code optimized, so it doesn't use any non-volatile * registers. */ .globl setfault setfault: mflr %r0 mfsprg0 %r4 lwz %r4, TD_PCB(%r2) stw %r3, PCB_ONFAULT(%r4) mfcr %r10 mfctr %r11 mfxer %r12 stw %r0, 0(%r3) stw %r1, 4(%r3) stw %r2, 8(%r3) stmw %r10, 12(%r3) /* store CR, CTR, XER, [r13 .. r31] */ li %r3, 0 /* return FALSE */ blr /************************************************************************/ /* Data section */ /************************************************************************/ .data .align 4 tmpstack: .space TMPSTACKSZ tmpstackbound: .space 10240 /* XXX: this really should not be necessary */ /* * Compiled KERNBASE locations */ .globl kernbase .set kernbase, KERNBASE #include Index: projects/cxl_iscsi/sys/powerpc/booke/machdep.c =================================================================== --- projects/cxl_iscsi/sys/powerpc/booke/machdep.c (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/booke/machdep.c (revision 279717) @@ -1,661 +1,706 @@ /*- * Copyright (C) 2006-2012 Semihalf * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN * NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /*- * Copyright (C) 2001 Benno Rice * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY Benno Rice ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * $NetBSD: machdep.c,v 1.74.2.1 2000/11/01 16:13:48 tv Exp $ */ /*- * Copyright (C) 1995, 1996 Wolfgang Solfrank. * Copyright (C) 1995, 1996 TooLs GmbH. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by TooLs GmbH. * 4. The name of TooLs GmbH may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include "opt_compat.h" #include "opt_ddb.h" #include "opt_kstack_pages.h" #include "opt_platform.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef DDB #include #endif #ifdef DEBUG #define debugf(fmt, args...) printf(fmt, ##args) #else #define debugf(fmt, args...) #endif extern unsigned char kernel_text[]; extern unsigned char _etext[]; extern unsigned char _edata[]; extern unsigned char __bss_start[]; extern unsigned char __sbss_start[]; extern unsigned char __sbss_end[]; extern unsigned char _end[]; /* * Bootinfo is passed to us by legacy loaders. Save the address of the * structure to handle backward compatibility. */ uint32_t *bootinfo; struct kva_md_info kmi; struct pcpu __pcpu[MAXCPU]; struct trapframe frame0; int cold = 1; long realmem = 0; long Maxmem = 0; char machine[] = "powerpc"; SYSCTL_STRING(_hw, HW_MACHINE, machine, CTLFLAG_RD, machine, 0, ""); int cacheline_size = 32; SYSCTL_INT(_machdep, CPU_CACHELINE, cacheline_size, CTLFLAG_RD, &cacheline_size, 0, ""); int hw_direct_map = 0; static void cpu_booke_startup(void *); SYSINIT(cpu, SI_SUB_CPU, SI_ORDER_FIRST, cpu_booke_startup, NULL); void print_kernel_section_addr(void); void print_kenv(void); u_int booke_init(uint32_t, uint32_t); +void ivor_setup(void); + +extern void *interrupt_vector_base; +extern void *int_critical_input; +extern void *int_machine_check; +extern void *int_data_storage; +extern void *int_instr_storage; +extern void *int_external_input; +extern void *int_alignment; +extern void *int_program; +extern void *int_syscall; +extern void *int_decrementer; +extern void *int_fixed_interval_timer; +extern void *int_watchdog; +extern void *int_data_tlb_error; +extern void *int_inst_tlb_error; +extern void *int_debug; + +#define SET_TRAP(ivor, handler) \ + KASSERT(((uintptr_t)(&handler) & ~0xffffUL) == \ + ((uintptr_t)(&interrupt_vector_base) & ~0xffffUL), \ + ("Handler " #handler " too far from interrupt vector base")); \ + mtspr(ivor, (uintptr_t)(&handler) & 0xffffUL); + +void +ivor_setup(void) +{ + + mtspr(SPR_IVPR, ((uintptr_t)&interrupt_vector_base) & 0xffff0000); + + SET_TRAP(SPR_IVOR0, int_critical_input); + SET_TRAP(SPR_IVOR1, int_machine_check); + SET_TRAP(SPR_IVOR2, int_data_storage); + SET_TRAP(SPR_IVOR3, int_instr_storage); + SET_TRAP(SPR_IVOR4, int_external_input); + SET_TRAP(SPR_IVOR5, int_alignment); + SET_TRAP(SPR_IVOR6, int_program); + SET_TRAP(SPR_IVOR8, int_syscall); + SET_TRAP(SPR_IVOR10, int_decrementer); + SET_TRAP(SPR_IVOR11, int_fixed_interval_timer); + SET_TRAP(SPR_IVOR12, int_watchdog); + SET_TRAP(SPR_IVOR13, int_data_tlb_error); + SET_TRAP(SPR_IVOR14, int_inst_tlb_error); + SET_TRAP(SPR_IVOR15, int_debug); +} static void cpu_booke_startup(void *dummy) { int indx; unsigned long size; /* Initialise the decrementer-based clock. */ decr_init(); /* Good {morning,afternoon,evening,night}. */ cpu_setup(PCPU_GET(cpuid)); printf("real memory = %lu (%ld MB)\n", ptoa(physmem), ptoa(physmem) / 1048576); realmem = physmem; /* Display any holes after the first chunk of extended memory. */ if (bootverbose) { printf("Physical memory chunk(s):\n"); for (indx = 0; phys_avail[indx + 1] != 0; indx += 2) { size = phys_avail[indx + 1] - phys_avail[indx]; printf("0x%08x - 0x%08x, %lu bytes (%lu pages)\n", phys_avail[indx], phys_avail[indx + 1] - 1, size, size / PAGE_SIZE); } } vm_ksubmap_init(&kmi); printf("avail memory = %lu (%ld MB)\n", ptoa(vm_cnt.v_free_count), ptoa(vm_cnt.v_free_count) / 1048576); /* Set up buffers, so they can be used to read disk labels. */ bufinit(); vm_pager_bufferinit(); } static char * kenv_next(char *cp) { if (cp != NULL) { while (*cp != 0) cp++; cp++; if (*cp == 0) cp = NULL; } return (cp); } void print_kenv(void) { int len; char *cp; debugf("loader passed (static) kenv:\n"); if (kern_envp == NULL) { debugf(" no env, null ptr\n"); return; } debugf(" kern_envp = 0x%08x\n", (u_int32_t)kern_envp); len = 0; for (cp = kern_envp; cp != NULL; cp = kenv_next(cp)) debugf(" %x %s\n", (u_int32_t)cp, cp); } void print_kernel_section_addr(void) { debugf("kernel image addresses:\n"); debugf(" kernel_text = 0x%08x\n", (uint32_t)kernel_text); debugf(" _etext (sdata) = 0x%08x\n", (uint32_t)_etext); debugf(" _edata = 0x%08x\n", (uint32_t)_edata); debugf(" __sbss_start = 0x%08x\n", (uint32_t)__sbss_start); debugf(" __sbss_end = 0x%08x\n", (uint32_t)__sbss_end); debugf(" __sbss_start = 0x%08x\n", (uint32_t)__bss_start); debugf(" _end = 0x%08x\n", (uint32_t)_end); } static int booke_check_for_fdt(uint32_t arg1, vm_offset_t *dtbp) { void *ptr; if (arg1 % 8 != 0) return (-1); ptr = (void *)pmap_early_io_map(arg1, PAGE_SIZE); if (fdt_check_header(ptr) != 0) return (-1); *dtbp = (vm_offset_t)ptr; return (0); } u_int booke_init(uint32_t arg1, uint32_t arg2) { struct pcpu *pc; void *kmdp, *mdp; vm_offset_t dtbp, end; #ifdef DDB vm_offset_t ksym_start; vm_offset_t ksym_end; #endif kmdp = NULL; end = (uintptr_t)_end; dtbp = (vm_offset_t)NULL; /* Set up TLB initially */ bootinfo = NULL; tlb1_init(); /* * Handle the various ways we can get loaded and started: * - FreeBSD's loader passes the pointer to the metadata * in arg1, with arg2 undefined. arg1 has a value that's * relative to the kernel's link address (i.e. larger * than 0xc0000000). * - Juniper's loader passes the metadata pointer in arg2 * and sets arg1 to zero. This is to signal that the * loader maps the kernel and starts it at its link * address (unlike the FreeBSD loader). * - U-Boot passes the standard argc and argv parameters * in arg1 and arg2 (resp). arg1 is between 1 and some * relatively small number, such as 64K. arg2 is the * physical address of the argv vector. * - ePAPR loaders pass an FDT blob in r3 (arg1) and the magic hex * string 0x45504150 ('ePAP') in r6 (which has been lost by now). * r4 (arg2) is supposed to be set to zero, but is not always. */ if (arg1 == 0) /* Juniper loader */ mdp = (void *)arg2; else if (booke_check_for_fdt(arg1, &dtbp) == 0) { /* ePAPR */ end = roundup(end, 8); memmove((void *)end, (void *)dtbp, fdt_totalsize((void *)dtbp)); dtbp = end; end += fdt_totalsize((void *)dtbp); mdp = NULL; } else if (arg1 > (uintptr_t)kernel_text) /* FreeBSD loader */ mdp = (void *)arg1; else /* U-Boot */ mdp = NULL; /* * Parse metadata and fetch parameters. */ if (mdp != NULL) { preload_metadata = mdp; kmdp = preload_search_by_type("elf kernel"); if (kmdp != NULL) { boothowto = MD_FETCH(kmdp, MODINFOMD_HOWTO, int); kern_envp = MD_FETCH(kmdp, MODINFOMD_ENVP, char *); dtbp = MD_FETCH(kmdp, MODINFOMD_DTBP, vm_offset_t); end = MD_FETCH(kmdp, MODINFOMD_KERNEND, vm_offset_t); bootinfo = (uint32_t *)preload_search_info(kmdp, MODINFO_METADATA | MODINFOMD_BOOTINFO); #ifdef DDB ksym_start = MD_FETCH(kmdp, MODINFOMD_SSYM, uintptr_t); ksym_end = MD_FETCH(kmdp, MODINFOMD_ESYM, uintptr_t); db_fetch_ksymtab(ksym_start, ksym_end); #endif } } else { bzero(__sbss_start, __sbss_end - __sbss_start); bzero(__bss_start, _end - __bss_start); } #if defined(FDT_DTB_STATIC) /* * In case the device tree blob was not retrieved (from metadata) try * to use the statically embedded one. */ if (dtbp == (vm_offset_t)NULL) dtbp = (vm_offset_t)&fdt_static_dtb; #endif if (OF_install(OFW_FDT, 0) == FALSE) while (1); if (OF_init((void *)dtbp) != 0) while (1); OF_interpret("perform-fixup", 0); /* Reset TLB1 to get rid of temporary mappings */ tlb1_init(); /* Reset Time Base */ mttb(0); /* Init params/tunables that can be overridden by the loader. */ init_param1(); /* Start initializing proc0 and thread0. */ proc_linkup0(&proc0, &thread0); thread0.td_frame = &frame0; /* Set up per-cpu data and store the pointer in SPR general 0. */ pc = &__pcpu[0]; pcpu_init(pc, 0, sizeof(struct pcpu)); pc->pc_curthread = &thread0; #ifdef __powerpc64__ __asm __volatile("mr 13,%0" :: "r"(pc->pc_curthread)); #else __asm __volatile("mr 2,%0" :: "r"(pc->pc_curthread)); #endif __asm __volatile("mtsprg 0, %0" :: "r"(pc)); /* Initialize system mutexes. */ mutex_init(); /* Initialize the console before printing anything. */ cninit(); /* Print out some debug info... */ debugf("%s: console initialized\n", __func__); debugf(" arg3 mdp = 0x%08x\n", (u_int32_t)mdp); debugf(" end = 0x%08x\n", (u_int32_t)end); debugf(" boothowto = 0x%08x\n", boothowto); debugf(" kernel ccsrbar = 0x%08x\n", CCSRBAR_VA); debugf(" MSR = 0x%08x\n", mfmsr()); #if defined(BOOKE_E500) debugf(" HID0 = 0x%08x\n", mfspr(SPR_HID0)); debugf(" HID1 = 0x%08x\n", mfspr(SPR_HID1)); debugf(" BUCSR = 0x%08x\n", mfspr(SPR_BUCSR)); #endif debugf(" dtbp = 0x%08x\n", (uint32_t)dtbp); print_kernel_section_addr(); print_kenv(); #if defined(BOOKE_E500) //tlb1_print_entries(); //tlb1_print_tlbentries(); #endif kdb_init(); #ifdef KDB if (boothowto & RB_KDB) kdb_enter(KDB_WHY_BOOTFLAGS, "Boot flags requested debugger"); #endif /* Initialise platform module */ platform_probe_and_attach(); /* Initialise virtual memory. */ pmap_mmu_install(MMU_TYPE_BOOKE, 0); pmap_bootstrap((uintptr_t)kernel_text, end); debugf("MSR = 0x%08x\n", mfmsr()); #if defined(BOOKE_E500) //tlb1_print_entries(); //tlb1_print_tlbentries(); #endif /* Initialize params/tunables that are derived from memsize. */ init_param2(physmem); /* Finish setting up thread0. */ thread0.td_pcb = (struct pcb *) ((thread0.td_kstack + thread0.td_kstack_pages * PAGE_SIZE - sizeof(struct pcb)) & ~15); bzero((void *)thread0.td_pcb, sizeof(struct pcb)); pc->pc_curpcb = thread0.td_pcb; /* Initialise the message buffer. */ msgbufinit(msgbufp, msgbufsize); /* Enable Machine Check interrupt. */ mtmsr(mfmsr() | PSL_ME); isync(); /* Enable L1 caches */ booke_enable_l1_cache(); debugf("%s: SP = 0x%08x\n", __func__, ((uintptr_t)thread0.td_pcb - 16) & ~15); return (((uintptr_t)thread0.td_pcb - 16) & ~15); } #define RES_GRANULE 32 extern uint32_t tlb0_miss_locks[]; /* Initialise a struct pcpu. */ void cpu_pcpu_init(struct pcpu *pcpu, int cpuid, size_t sz) { pcpu->pc_tid_next = TID_MIN; #ifdef SMP uint32_t *ptr; int words_per_gran = RES_GRANULE / sizeof(uint32_t); ptr = &tlb0_miss_locks[cpuid * words_per_gran]; pcpu->pc_booke_tlb_lock = ptr; *ptr = TLB_UNLOCKED; *(ptr + 1) = 0; /* recurse counter */ #endif } /* * Flush the D-cache for non-DMA I/O so that the I-cache can * be made coherent later. */ void cpu_flush_dcache(void *ptr, size_t len) { register_t addr, off; /* * Align the address to a cacheline and adjust the length * accordingly. Then round the length to a multiple of the * cacheline for easy looping. */ addr = (uintptr_t)ptr; off = addr & (cacheline_size - 1); addr -= off; len = (len + off + cacheline_size - 1) & ~(cacheline_size - 1); while (len > 0) { __asm __volatile ("dcbf 0,%0" :: "r"(addr)); __asm __volatile ("sync"); addr += cacheline_size; len -= cacheline_size; } } void spinlock_enter(void) { struct thread *td; register_t msr; td = curthread; if (td->td_md.md_spinlock_count == 0) { msr = intr_disable(); td->td_md.md_spinlock_count = 1; td->td_md.md_saved_msr = msr; } else td->td_md.md_spinlock_count++; critical_enter(); } void spinlock_exit(void) { struct thread *td; register_t msr; td = curthread; critical_exit(); msr = td->td_md.md_saved_msr; td->td_md.md_spinlock_count--; if (td->td_md.md_spinlock_count == 0) intr_restore(msr); } /* Shutdown the CPU as much as possible. */ void cpu_halt(void) { mtmsr(mfmsr() & ~(PSL_CE | PSL_EE | PSL_ME | PSL_DE)); while (1) ; } int ptrace_set_pc(struct thread *td, unsigned long addr) { struct trapframe *tf; tf = td->td_frame; tf->srr0 = (register_t)addr; return (0); } int ptrace_single_step(struct thread *td) { struct trapframe *tf; tf = td->td_frame; tf->srr1 |= PSL_DE; tf->cpu.booke.dbcr0 |= (DBCR0_IDM | DBCR0_IC); return (0); } int ptrace_clear_single_step(struct thread *td) { struct trapframe *tf; tf = td->td_frame; tf->srr1 &= ~PSL_DE; tf->cpu.booke.dbcr0 &= ~(DBCR0_IDM | DBCR0_IC); return (0); } void kdb_cpu_clear_singlestep(void) { register_t r; r = mfspr(SPR_DBCR0); mtspr(SPR_DBCR0, r & ~DBCR0_IC); kdb_frame->srr1 &= ~PSL_DE; } void kdb_cpu_set_singlestep(void) { register_t r; r = mfspr(SPR_DBCR0); mtspr(SPR_DBCR0, r | DBCR0_IC | DBCR0_IDM); kdb_frame->srr1 |= PSL_DE; } void bzero(void *buf, size_t len) { caddr_t p; p = buf; while (((vm_offset_t) p & (sizeof(u_long) - 1)) && len) { *p++ = 0; len--; } while (len >= sizeof(u_long) * 8) { *(u_long*) p = 0; *((u_long*) p + 1) = 0; *((u_long*) p + 2) = 0; *((u_long*) p + 3) = 0; len -= sizeof(u_long) * 8; *((u_long*) p + 4) = 0; *((u_long*) p + 5) = 0; *((u_long*) p + 6) = 0; *((u_long*) p + 7) = 0; p += sizeof(u_long) * 8; } while (len >= sizeof(u_long)) { *(u_long*) p = 0; len -= sizeof(u_long); p += sizeof(u_long); } while (len) { *p++ = 0; len--; } } Index: projects/cxl_iscsi/sys/powerpc/booke/pmap.c =================================================================== --- projects/cxl_iscsi/sys/powerpc/booke/pmap.c (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/booke/pmap.c (revision 279717) @@ -1,3319 +1,3323 @@ /*- * Copyright (C) 2007-2009 Semihalf, Rafal Jaworowski * Copyright (C) 2006 Semihalf, Marian Balakowicz * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN * NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * Some hw specific parts of this pmap were derived or influenced * by NetBSD's ibm4xx pmap module. More generic code is shared with * a few other pmap modules from the FreeBSD tree. */ /* * VM layout notes: * * Kernel and user threads run within one common virtual address space * defined by AS=0. * * Virtual address space layout: * ----------------------------- * 0x0000_0000 - 0xafff_ffff : user process * 0xb000_0000 - 0xbfff_ffff : pmap_mapdev()-ed area (PCI/PCIE etc.) * 0xc000_0000 - 0xc0ff_ffff : kernel reserved * 0xc000_0000 - data_end : kernel code+data, env, metadata etc. * 0xc100_0000 - 0xfeef_ffff : KVA * 0xc100_0000 - 0xc100_3fff : reserved for page zero/copy * 0xc100_4000 - 0xc200_3fff : reserved for ptbl bufs * 0xc200_4000 - 0xc200_8fff : guard page + kstack0 * 0xc200_9000 - 0xfeef_ffff : actual free KVA space * 0xfef0_0000 - 0xffff_ffff : I/O devices region */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "mmu_if.h" #ifdef DEBUG #define debugf(fmt, args...) printf(fmt, ##args) #else #define debugf(fmt, args...) #endif #define TODO panic("%s: not implemented", __func__); extern unsigned char _etext[]; extern unsigned char _end[]; extern uint32_t *bootinfo; #ifdef SMP extern uint32_t bp_ntlb1s; #endif vm_paddr_t kernload; vm_offset_t kernstart; vm_size_t kernsize; /* Message buffer and tables. */ static vm_offset_t data_start; static vm_size_t data_end; /* Phys/avail memory regions. */ static struct mem_region *availmem_regions; static int availmem_regions_sz; static struct mem_region *physmem_regions; static int physmem_regions_sz; /* Reserved KVA space and mutex for mmu_booke_zero_page. */ static vm_offset_t zero_page_va; static struct mtx zero_page_mutex; static struct mtx tlbivax_mutex; /* * Reserved KVA space for mmu_booke_zero_page_idle. This is used * by idle thred only, no lock required. */ static vm_offset_t zero_page_idle_va; /* Reserved KVA space and mutex for mmu_booke_copy_page. */ static vm_offset_t copy_page_src_va; static vm_offset_t copy_page_dst_va; static struct mtx copy_page_mutex; /**************************************************************************/ /* PMAP */ /**************************************************************************/ static int mmu_booke_enter_locked(mmu_t, pmap_t, vm_offset_t, vm_page_t, vm_prot_t, u_int flags, int8_t psind); unsigned int kptbl_min; /* Index of the first kernel ptbl. */ unsigned int kernel_ptbls; /* Number of KVA ptbls. */ /* * If user pmap is processed with mmu_booke_remove and the resident count * drops to 0, there are no more pages to remove, so we need not continue. */ #define PMAP_REMOVE_DONE(pmap) \ ((pmap) != kernel_pmap && (pmap)->pm_stats.resident_count == 0) extern void tid_flush(tlbtid_t); extern int elf32_nxstack; /**************************************************************************/ /* TLB and TID handling */ /**************************************************************************/ /* Translation ID busy table */ static volatile pmap_t tidbusy[MAXCPU][TID_MAX + 1]; /* * TLB0 capabilities (entry, way numbers etc.). These can vary between e500 * core revisions and should be read from h/w registers during early config. */ uint32_t tlb0_entries; uint32_t tlb0_ways; uint32_t tlb0_entries_per_way; #define TLB0_ENTRIES (tlb0_entries) #define TLB0_WAYS (tlb0_ways) #define TLB0_ENTRIES_PER_WAY (tlb0_entries_per_way) #define TLB1_ENTRIES 16 /* In-ram copy of the TLB1 */ static tlb_entry_t tlb1[TLB1_ENTRIES]; /* Next free entry in the TLB1 */ static unsigned int tlb1_idx; static vm_offset_t tlb1_map_base = VM_MAX_KERNEL_ADDRESS; static tlbtid_t tid_alloc(struct pmap *); static void tlb_print_entry(int, uint32_t, uint32_t, uint32_t, uint32_t); static int tlb1_set_entry(vm_offset_t, vm_offset_t, vm_size_t, uint32_t); static void tlb1_write_entry(unsigned int); static int tlb1_iomapped(int, vm_paddr_t, vm_size_t, vm_offset_t *); static vm_size_t tlb1_mapin_region(vm_offset_t, vm_paddr_t, vm_size_t); static vm_size_t tsize2size(unsigned int); static unsigned int size2tsize(vm_size_t); static unsigned int ilog2(unsigned int); static void set_mas4_defaults(void); static inline void tlb0_flush_entry(vm_offset_t); static inline unsigned int tlb0_tableidx(vm_offset_t, unsigned int); /**************************************************************************/ /* Page table management */ /**************************************************************************/ static struct rwlock_padalign pvh_global_lock; /* Data for the pv entry allocation mechanism */ static uma_zone_t pvzone; static int pv_entry_count = 0, pv_entry_max = 0, pv_entry_high_water = 0; #define PV_ENTRY_ZONE_MIN 2048 /* min pv entries in uma zone */ #ifndef PMAP_SHPGPERPROC #define PMAP_SHPGPERPROC 200 #endif static void ptbl_init(void); static struct ptbl_buf *ptbl_buf_alloc(void); static void ptbl_buf_free(struct ptbl_buf *); static void ptbl_free_pmap_ptbl(pmap_t, pte_t *); static pte_t *ptbl_alloc(mmu_t, pmap_t, unsigned int, boolean_t); static void ptbl_free(mmu_t, pmap_t, unsigned int); static void ptbl_hold(mmu_t, pmap_t, unsigned int); static int ptbl_unhold(mmu_t, pmap_t, unsigned int); static vm_paddr_t pte_vatopa(mmu_t, pmap_t, vm_offset_t); static pte_t *pte_find(mmu_t, pmap_t, vm_offset_t); static int pte_enter(mmu_t, pmap_t, vm_page_t, vm_offset_t, uint32_t, boolean_t); static int pte_remove(mmu_t, pmap_t, vm_offset_t, uint8_t); static pv_entry_t pv_alloc(void); static void pv_free(pv_entry_t); static void pv_insert(pmap_t, vm_offset_t, vm_page_t); static void pv_remove(pmap_t, vm_offset_t, vm_page_t); /* Number of kva ptbl buffers, each covering one ptbl (PTBL_PAGES). */ #define PTBL_BUFS (128 * 16) struct ptbl_buf { TAILQ_ENTRY(ptbl_buf) link; /* list link */ vm_offset_t kva; /* va of mapping */ }; /* ptbl free list and a lock used for access synchronization. */ static TAILQ_HEAD(, ptbl_buf) ptbl_buf_freelist; static struct mtx ptbl_buf_freelist_lock; /* Base address of kva space allocated fot ptbl bufs. */ static vm_offset_t ptbl_buf_pool_vabase; /* Pointer to ptbl_buf structures. */ static struct ptbl_buf *ptbl_bufs; void pmap_bootstrap_ap(volatile uint32_t *); /* * Kernel MMU interface */ static void mmu_booke_clear_modify(mmu_t, vm_page_t); static void mmu_booke_copy(mmu_t, pmap_t, pmap_t, vm_offset_t, vm_size_t, vm_offset_t); static void mmu_booke_copy_page(mmu_t, vm_page_t, vm_page_t); static void mmu_booke_copy_pages(mmu_t, vm_page_t *, vm_offset_t, vm_page_t *, vm_offset_t, int); static int mmu_booke_enter(mmu_t, pmap_t, vm_offset_t, vm_page_t, vm_prot_t, u_int flags, int8_t psind); static void mmu_booke_enter_object(mmu_t, pmap_t, vm_offset_t, vm_offset_t, vm_page_t, vm_prot_t); static void mmu_booke_enter_quick(mmu_t, pmap_t, vm_offset_t, vm_page_t, vm_prot_t); static vm_paddr_t mmu_booke_extract(mmu_t, pmap_t, vm_offset_t); static vm_page_t mmu_booke_extract_and_hold(mmu_t, pmap_t, vm_offset_t, vm_prot_t); static void mmu_booke_init(mmu_t); static boolean_t mmu_booke_is_modified(mmu_t, vm_page_t); static boolean_t mmu_booke_is_prefaultable(mmu_t, pmap_t, vm_offset_t); static boolean_t mmu_booke_is_referenced(mmu_t, vm_page_t); static int mmu_booke_ts_referenced(mmu_t, vm_page_t); static vm_offset_t mmu_booke_map(mmu_t, vm_offset_t *, vm_paddr_t, vm_paddr_t, int); static int mmu_booke_mincore(mmu_t, pmap_t, vm_offset_t, vm_paddr_t *); static void mmu_booke_object_init_pt(mmu_t, pmap_t, vm_offset_t, vm_object_t, vm_pindex_t, vm_size_t); static boolean_t mmu_booke_page_exists_quick(mmu_t, pmap_t, vm_page_t); static void mmu_booke_page_init(mmu_t, vm_page_t); static int mmu_booke_page_wired_mappings(mmu_t, vm_page_t); static void mmu_booke_pinit(mmu_t, pmap_t); static void mmu_booke_pinit0(mmu_t, pmap_t); static void mmu_booke_protect(mmu_t, pmap_t, vm_offset_t, vm_offset_t, vm_prot_t); static void mmu_booke_qenter(mmu_t, vm_offset_t, vm_page_t *, int); static void mmu_booke_qremove(mmu_t, vm_offset_t, int); static void mmu_booke_release(mmu_t, pmap_t); static void mmu_booke_remove(mmu_t, pmap_t, vm_offset_t, vm_offset_t); static void mmu_booke_remove_all(mmu_t, vm_page_t); static void mmu_booke_remove_write(mmu_t, vm_page_t); static void mmu_booke_unwire(mmu_t, pmap_t, vm_offset_t, vm_offset_t); static void mmu_booke_zero_page(mmu_t, vm_page_t); static void mmu_booke_zero_page_area(mmu_t, vm_page_t, int, int); static void mmu_booke_zero_page_idle(mmu_t, vm_page_t); static void mmu_booke_activate(mmu_t, struct thread *); static void mmu_booke_deactivate(mmu_t, struct thread *); static void mmu_booke_bootstrap(mmu_t, vm_offset_t, vm_offset_t); static void *mmu_booke_mapdev(mmu_t, vm_paddr_t, vm_size_t); static void *mmu_booke_mapdev_attr(mmu_t, vm_paddr_t, vm_size_t, vm_memattr_t); static void mmu_booke_unmapdev(mmu_t, vm_offset_t, vm_size_t); static vm_paddr_t mmu_booke_kextract(mmu_t, vm_offset_t); static void mmu_booke_kenter(mmu_t, vm_offset_t, vm_paddr_t); static void mmu_booke_kenter_attr(mmu_t, vm_offset_t, vm_paddr_t, vm_memattr_t); static void mmu_booke_kremove(mmu_t, vm_offset_t); static boolean_t mmu_booke_dev_direct_mapped(mmu_t, vm_paddr_t, vm_size_t); static void mmu_booke_sync_icache(mmu_t, pmap_t, vm_offset_t, vm_size_t); static void mmu_booke_dumpsys_map(mmu_t, vm_paddr_t pa, size_t, void **); static void mmu_booke_dumpsys_unmap(mmu_t, vm_paddr_t pa, size_t, void *); static void mmu_booke_scan_init(mmu_t); static mmu_method_t mmu_booke_methods[] = { /* pmap dispatcher interface */ MMUMETHOD(mmu_clear_modify, mmu_booke_clear_modify), MMUMETHOD(mmu_copy, mmu_booke_copy), MMUMETHOD(mmu_copy_page, mmu_booke_copy_page), MMUMETHOD(mmu_copy_pages, mmu_booke_copy_pages), MMUMETHOD(mmu_enter, mmu_booke_enter), MMUMETHOD(mmu_enter_object, mmu_booke_enter_object), MMUMETHOD(mmu_enter_quick, mmu_booke_enter_quick), MMUMETHOD(mmu_extract, mmu_booke_extract), MMUMETHOD(mmu_extract_and_hold, mmu_booke_extract_and_hold), MMUMETHOD(mmu_init, mmu_booke_init), MMUMETHOD(mmu_is_modified, mmu_booke_is_modified), MMUMETHOD(mmu_is_prefaultable, mmu_booke_is_prefaultable), MMUMETHOD(mmu_is_referenced, mmu_booke_is_referenced), MMUMETHOD(mmu_ts_referenced, mmu_booke_ts_referenced), MMUMETHOD(mmu_map, mmu_booke_map), MMUMETHOD(mmu_mincore, mmu_booke_mincore), MMUMETHOD(mmu_object_init_pt, mmu_booke_object_init_pt), MMUMETHOD(mmu_page_exists_quick,mmu_booke_page_exists_quick), MMUMETHOD(mmu_page_init, mmu_booke_page_init), MMUMETHOD(mmu_page_wired_mappings, mmu_booke_page_wired_mappings), MMUMETHOD(mmu_pinit, mmu_booke_pinit), MMUMETHOD(mmu_pinit0, mmu_booke_pinit0), MMUMETHOD(mmu_protect, mmu_booke_protect), MMUMETHOD(mmu_qenter, mmu_booke_qenter), MMUMETHOD(mmu_qremove, mmu_booke_qremove), MMUMETHOD(mmu_release, mmu_booke_release), MMUMETHOD(mmu_remove, mmu_booke_remove), MMUMETHOD(mmu_remove_all, mmu_booke_remove_all), MMUMETHOD(mmu_remove_write, mmu_booke_remove_write), MMUMETHOD(mmu_sync_icache, mmu_booke_sync_icache), MMUMETHOD(mmu_unwire, mmu_booke_unwire), MMUMETHOD(mmu_zero_page, mmu_booke_zero_page), MMUMETHOD(mmu_zero_page_area, mmu_booke_zero_page_area), MMUMETHOD(mmu_zero_page_idle, mmu_booke_zero_page_idle), MMUMETHOD(mmu_activate, mmu_booke_activate), MMUMETHOD(mmu_deactivate, mmu_booke_deactivate), /* Internal interfaces */ MMUMETHOD(mmu_bootstrap, mmu_booke_bootstrap), MMUMETHOD(mmu_dev_direct_mapped,mmu_booke_dev_direct_mapped), MMUMETHOD(mmu_mapdev, mmu_booke_mapdev), MMUMETHOD(mmu_mapdev_attr, mmu_booke_mapdev_attr), MMUMETHOD(mmu_kenter, mmu_booke_kenter), MMUMETHOD(mmu_kenter_attr, mmu_booke_kenter_attr), MMUMETHOD(mmu_kextract, mmu_booke_kextract), /* MMUMETHOD(mmu_kremove, mmu_booke_kremove), */ MMUMETHOD(mmu_unmapdev, mmu_booke_unmapdev), /* dumpsys() support */ MMUMETHOD(mmu_dumpsys_map, mmu_booke_dumpsys_map), MMUMETHOD(mmu_dumpsys_unmap, mmu_booke_dumpsys_unmap), MMUMETHOD(mmu_scan_init, mmu_booke_scan_init), { 0, 0 } }; MMU_DEF(booke_mmu, MMU_TYPE_BOOKE, mmu_booke_methods, 0); static __inline uint32_t tlb_calc_wimg(vm_offset_t pa, vm_memattr_t ma) { uint32_t attrib; int i; if (ma != VM_MEMATTR_DEFAULT) { switch (ma) { case VM_MEMATTR_UNCACHEABLE: return (PTE_I | PTE_G); case VM_MEMATTR_WRITE_COMBINING: case VM_MEMATTR_WRITE_BACK: case VM_MEMATTR_PREFETCHABLE: return (PTE_I); case VM_MEMATTR_WRITE_THROUGH: return (PTE_W | PTE_M); } } /* * Assume the page is cache inhibited and access is guarded unless * it's in our available memory array. */ attrib = _TLB_ENTRY_IO; for (i = 0; i < physmem_regions_sz; i++) { if ((pa >= physmem_regions[i].mr_start) && (pa < (physmem_regions[i].mr_start + physmem_regions[i].mr_size))) { attrib = _TLB_ENTRY_MEM; break; } } return (attrib); } static inline void tlb_miss_lock(void) { #ifdef SMP struct pcpu *pc; if (!smp_started) return; STAILQ_FOREACH(pc, &cpuhead, pc_allcpu) { if (pc != pcpup) { CTR3(KTR_PMAP, "%s: tlb miss LOCK of CPU=%d, " "tlb_lock=%p", __func__, pc->pc_cpuid, pc->pc_booke_tlb_lock); KASSERT((pc->pc_cpuid != PCPU_GET(cpuid)), ("tlb_miss_lock: tried to lock self")); tlb_lock(pc->pc_booke_tlb_lock); CTR1(KTR_PMAP, "%s: locked", __func__); } } #endif } static inline void tlb_miss_unlock(void) { #ifdef SMP struct pcpu *pc; if (!smp_started) return; STAILQ_FOREACH(pc, &cpuhead, pc_allcpu) { if (pc != pcpup) { CTR2(KTR_PMAP, "%s: tlb miss UNLOCK of CPU=%d", __func__, pc->pc_cpuid); tlb_unlock(pc->pc_booke_tlb_lock); CTR1(KTR_PMAP, "%s: unlocked", __func__); } } #endif } /* Return number of entries in TLB0. */ static __inline void tlb0_get_tlbconf(void) { uint32_t tlb0_cfg; tlb0_cfg = mfspr(SPR_TLB0CFG); tlb0_entries = tlb0_cfg & TLBCFG_NENTRY_MASK; tlb0_ways = (tlb0_cfg & TLBCFG_ASSOC_MASK) >> TLBCFG_ASSOC_SHIFT; tlb0_entries_per_way = tlb0_entries / tlb0_ways; } /* Initialize pool of kva ptbl buffers. */ static void ptbl_init(void) { int i; CTR3(KTR_PMAP, "%s: s (ptbl_bufs = 0x%08x size 0x%08x)", __func__, (uint32_t)ptbl_bufs, sizeof(struct ptbl_buf) * PTBL_BUFS); CTR3(KTR_PMAP, "%s: s (ptbl_buf_pool_vabase = 0x%08x size = 0x%08x)", __func__, ptbl_buf_pool_vabase, PTBL_BUFS * PTBL_PAGES * PAGE_SIZE); mtx_init(&ptbl_buf_freelist_lock, "ptbl bufs lock", NULL, MTX_DEF); TAILQ_INIT(&ptbl_buf_freelist); for (i = 0; i < PTBL_BUFS; i++) { ptbl_bufs[i].kva = ptbl_buf_pool_vabase + i * PTBL_PAGES * PAGE_SIZE; TAILQ_INSERT_TAIL(&ptbl_buf_freelist, &ptbl_bufs[i], link); } } /* Get a ptbl_buf from the freelist. */ static struct ptbl_buf * ptbl_buf_alloc(void) { struct ptbl_buf *buf; mtx_lock(&ptbl_buf_freelist_lock); buf = TAILQ_FIRST(&ptbl_buf_freelist); if (buf != NULL) TAILQ_REMOVE(&ptbl_buf_freelist, buf, link); mtx_unlock(&ptbl_buf_freelist_lock); CTR2(KTR_PMAP, "%s: buf = %p", __func__, buf); return (buf); } /* Return ptbl buff to free pool. */ static void ptbl_buf_free(struct ptbl_buf *buf) { CTR2(KTR_PMAP, "%s: buf = %p", __func__, buf); mtx_lock(&ptbl_buf_freelist_lock); TAILQ_INSERT_TAIL(&ptbl_buf_freelist, buf, link); mtx_unlock(&ptbl_buf_freelist_lock); } /* * Search the list of allocated ptbl bufs and find on list of allocated ptbls */ static void ptbl_free_pmap_ptbl(pmap_t pmap, pte_t *ptbl) { struct ptbl_buf *pbuf; CTR2(KTR_PMAP, "%s: ptbl = %p", __func__, ptbl); PMAP_LOCK_ASSERT(pmap, MA_OWNED); TAILQ_FOREACH(pbuf, &pmap->pm_ptbl_list, link) if (pbuf->kva == (vm_offset_t)ptbl) { /* Remove from pmap ptbl buf list. */ TAILQ_REMOVE(&pmap->pm_ptbl_list, pbuf, link); /* Free corresponding ptbl buf. */ ptbl_buf_free(pbuf); break; } } /* Allocate page table. */ static pte_t * ptbl_alloc(mmu_t mmu, pmap_t pmap, unsigned int pdir_idx, boolean_t nosleep) { vm_page_t mtbl[PTBL_PAGES]; vm_page_t m; struct ptbl_buf *pbuf; unsigned int pidx; pte_t *ptbl; int i, j; CTR4(KTR_PMAP, "%s: pmap = %p su = %d pdir_idx = %d", __func__, pmap, (pmap == kernel_pmap), pdir_idx); KASSERT((pdir_idx <= (VM_MAXUSER_ADDRESS / PDIR_SIZE)), ("ptbl_alloc: invalid pdir_idx")); KASSERT((pmap->pm_pdir[pdir_idx] == NULL), ("pte_alloc: valid ptbl entry exists!")); pbuf = ptbl_buf_alloc(); if (pbuf == NULL) panic("pte_alloc: couldn't alloc kernel virtual memory"); ptbl = (pte_t *)pbuf->kva; CTR2(KTR_PMAP, "%s: ptbl kva = %p", __func__, ptbl); /* Allocate ptbl pages, this will sleep! */ for (i = 0; i < PTBL_PAGES; i++) { pidx = (PTBL_PAGES * pdir_idx) + i; while ((m = vm_page_alloc(NULL, pidx, VM_ALLOC_NOOBJ | VM_ALLOC_WIRED)) == NULL) { PMAP_UNLOCK(pmap); rw_wunlock(&pvh_global_lock); if (nosleep) { ptbl_free_pmap_ptbl(pmap, ptbl); for (j = 0; j < i; j++) vm_page_free(mtbl[j]); atomic_subtract_int(&vm_cnt.v_wire_count, i); return (NULL); } VM_WAIT; rw_wlock(&pvh_global_lock); PMAP_LOCK(pmap); } mtbl[i] = m; } /* Map allocated pages into kernel_pmap. */ mmu_booke_qenter(mmu, (vm_offset_t)ptbl, mtbl, PTBL_PAGES); /* Zero whole ptbl. */ bzero((caddr_t)ptbl, PTBL_PAGES * PAGE_SIZE); /* Add pbuf to the pmap ptbl bufs list. */ TAILQ_INSERT_TAIL(&pmap->pm_ptbl_list, pbuf, link); return (ptbl); } /* Free ptbl pages and invalidate pdir entry. */ static void ptbl_free(mmu_t mmu, pmap_t pmap, unsigned int pdir_idx) { pte_t *ptbl; vm_paddr_t pa; vm_offset_t va; vm_page_t m; int i; CTR4(KTR_PMAP, "%s: pmap = %p su = %d pdir_idx = %d", __func__, pmap, (pmap == kernel_pmap), pdir_idx); KASSERT((pdir_idx <= (VM_MAXUSER_ADDRESS / PDIR_SIZE)), ("ptbl_free: invalid pdir_idx")); ptbl = pmap->pm_pdir[pdir_idx]; CTR2(KTR_PMAP, "%s: ptbl = %p", __func__, ptbl); KASSERT((ptbl != NULL), ("ptbl_free: null ptbl")); /* * Invalidate the pdir entry as soon as possible, so that other CPUs * don't attempt to look up the page tables we are releasing. */ mtx_lock_spin(&tlbivax_mutex); tlb_miss_lock(); pmap->pm_pdir[pdir_idx] = NULL; tlb_miss_unlock(); mtx_unlock_spin(&tlbivax_mutex); for (i = 0; i < PTBL_PAGES; i++) { va = ((vm_offset_t)ptbl + (i * PAGE_SIZE)); pa = pte_vatopa(mmu, kernel_pmap, va); m = PHYS_TO_VM_PAGE(pa); vm_page_free_zero(m); atomic_subtract_int(&vm_cnt.v_wire_count, 1); mmu_booke_kremove(mmu, va); } ptbl_free_pmap_ptbl(pmap, ptbl); } /* * Decrement ptbl pages hold count and attempt to free ptbl pages. * Called when removing pte entry from ptbl. * * Return 1 if ptbl pages were freed. */ static int ptbl_unhold(mmu_t mmu, pmap_t pmap, unsigned int pdir_idx) { pte_t *ptbl; vm_paddr_t pa; vm_page_t m; int i; CTR4(KTR_PMAP, "%s: pmap = %p su = %d pdir_idx = %d", __func__, pmap, (pmap == kernel_pmap), pdir_idx); KASSERT((pdir_idx <= (VM_MAXUSER_ADDRESS / PDIR_SIZE)), ("ptbl_unhold: invalid pdir_idx")); KASSERT((pmap != kernel_pmap), ("ptbl_unhold: unholding kernel ptbl!")); ptbl = pmap->pm_pdir[pdir_idx]; //debugf("ptbl_unhold: ptbl = 0x%08x\n", (u_int32_t)ptbl); KASSERT(((vm_offset_t)ptbl >= VM_MIN_KERNEL_ADDRESS), ("ptbl_unhold: non kva ptbl")); /* decrement hold count */ for (i = 0; i < PTBL_PAGES; i++) { pa = pte_vatopa(mmu, kernel_pmap, (vm_offset_t)ptbl + (i * PAGE_SIZE)); m = PHYS_TO_VM_PAGE(pa); m->wire_count--; } /* * Free ptbl pages if there are no pte etries in this ptbl. * wire_count has the same value for all ptbl pages, so check the last * page. */ if (m->wire_count == 0) { ptbl_free(mmu, pmap, pdir_idx); //debugf("ptbl_unhold: e (freed ptbl)\n"); return (1); } return (0); } /* * Increment hold count for ptbl pages. This routine is used when a new pte * entry is being inserted into the ptbl. */ static void ptbl_hold(mmu_t mmu, pmap_t pmap, unsigned int pdir_idx) { vm_paddr_t pa; pte_t *ptbl; vm_page_t m; int i; CTR3(KTR_PMAP, "%s: pmap = %p pdir_idx = %d", __func__, pmap, pdir_idx); KASSERT((pdir_idx <= (VM_MAXUSER_ADDRESS / PDIR_SIZE)), ("ptbl_hold: invalid pdir_idx")); KASSERT((pmap != kernel_pmap), ("ptbl_hold: holding kernel ptbl!")); ptbl = pmap->pm_pdir[pdir_idx]; KASSERT((ptbl != NULL), ("ptbl_hold: null ptbl")); for (i = 0; i < PTBL_PAGES; i++) { pa = pte_vatopa(mmu, kernel_pmap, (vm_offset_t)ptbl + (i * PAGE_SIZE)); m = PHYS_TO_VM_PAGE(pa); m->wire_count++; } } /* Allocate pv_entry structure. */ pv_entry_t pv_alloc(void) { pv_entry_t pv; pv_entry_count++; if (pv_entry_count > pv_entry_high_water) pagedaemon_wakeup(); pv = uma_zalloc(pvzone, M_NOWAIT); return (pv); } /* Free pv_entry structure. */ static __inline void pv_free(pv_entry_t pve) { pv_entry_count--; uma_zfree(pvzone, pve); } /* Allocate and initialize pv_entry structure. */ static void pv_insert(pmap_t pmap, vm_offset_t va, vm_page_t m) { pv_entry_t pve; //int su = (pmap == kernel_pmap); //debugf("pv_insert: s (su = %d pmap = 0x%08x va = 0x%08x m = 0x%08x)\n", su, // (u_int32_t)pmap, va, (u_int32_t)m); pve = pv_alloc(); if (pve == NULL) panic("pv_insert: no pv entries!"); pve->pv_pmap = pmap; pve->pv_va = va; /* add to pv_list */ PMAP_LOCK_ASSERT(pmap, MA_OWNED); rw_assert(&pvh_global_lock, RA_WLOCKED); TAILQ_INSERT_TAIL(&m->md.pv_list, pve, pv_link); //debugf("pv_insert: e\n"); } /* Destroy pv entry. */ static void pv_remove(pmap_t pmap, vm_offset_t va, vm_page_t m) { pv_entry_t pve; //int su = (pmap == kernel_pmap); //debugf("pv_remove: s (su = %d pmap = 0x%08x va = 0x%08x)\n", su, (u_int32_t)pmap, va); PMAP_LOCK_ASSERT(pmap, MA_OWNED); rw_assert(&pvh_global_lock, RA_WLOCKED); /* find pv entry */ TAILQ_FOREACH(pve, &m->md.pv_list, pv_link) { if ((pmap == pve->pv_pmap) && (va == pve->pv_va)) { /* remove from pv_list */ TAILQ_REMOVE(&m->md.pv_list, pve, pv_link); if (TAILQ_EMPTY(&m->md.pv_list)) vm_page_aflag_clear(m, PGA_WRITEABLE); /* free pv entry struct */ pv_free(pve); break; } } //debugf("pv_remove: e\n"); } /* * Clean pte entry, try to free page table page if requested. * * Return 1 if ptbl pages were freed, otherwise return 0. */ static int pte_remove(mmu_t mmu, pmap_t pmap, vm_offset_t va, uint8_t flags) { unsigned int pdir_idx = PDIR_IDX(va); unsigned int ptbl_idx = PTBL_IDX(va); vm_page_t m; pte_t *ptbl; pte_t *pte; //int su = (pmap == kernel_pmap); //debugf("pte_remove: s (su = %d pmap = 0x%08x va = 0x%08x flags = %d)\n", // su, (u_int32_t)pmap, va, flags); ptbl = pmap->pm_pdir[pdir_idx]; KASSERT(ptbl, ("pte_remove: null ptbl")); pte = &ptbl[ptbl_idx]; if (pte == NULL || !PTE_ISVALID(pte)) return (0); if (PTE_ISWIRED(pte)) pmap->pm_stats.wired_count--; /* Handle managed entry. */ if (PTE_ISMANAGED(pte)) { /* Get vm_page_t for mapped pte. */ m = PHYS_TO_VM_PAGE(PTE_PA(pte)); if (PTE_ISMODIFIED(pte)) vm_page_dirty(m); if (PTE_ISREFERENCED(pte)) vm_page_aflag_set(m, PGA_REFERENCED); pv_remove(pmap, va, m); } mtx_lock_spin(&tlbivax_mutex); tlb_miss_lock(); tlb0_flush_entry(va); pte->flags = 0; pte->rpn = 0; tlb_miss_unlock(); mtx_unlock_spin(&tlbivax_mutex); pmap->pm_stats.resident_count--; if (flags & PTBL_UNHOLD) { //debugf("pte_remove: e (unhold)\n"); return (ptbl_unhold(mmu, pmap, pdir_idx)); } //debugf("pte_remove: e\n"); return (0); } /* * Insert PTE for a given page and virtual address. */ static int pte_enter(mmu_t mmu, pmap_t pmap, vm_page_t m, vm_offset_t va, uint32_t flags, boolean_t nosleep) { unsigned int pdir_idx = PDIR_IDX(va); unsigned int ptbl_idx = PTBL_IDX(va); pte_t *ptbl, *pte; CTR4(KTR_PMAP, "%s: su = %d pmap = %p va = %p", __func__, pmap == kernel_pmap, pmap, va); /* Get the page table pointer. */ ptbl = pmap->pm_pdir[pdir_idx]; if (ptbl == NULL) { /* Allocate page table pages. */ ptbl = ptbl_alloc(mmu, pmap, pdir_idx, nosleep); if (ptbl == NULL) { KASSERT(nosleep, ("nosleep and NULL ptbl")); return (ENOMEM); } } else { /* * Check if there is valid mapping for requested * va, if there is, remove it. */ pte = &pmap->pm_pdir[pdir_idx][ptbl_idx]; if (PTE_ISVALID(pte)) { pte_remove(mmu, pmap, va, PTBL_HOLD); } else { /* * pte is not used, increment hold count * for ptbl pages. */ if (pmap != kernel_pmap) ptbl_hold(mmu, pmap, pdir_idx); } } /* * Insert pv_entry into pv_list for mapped page if part of managed * memory. */ if ((m->oflags & VPO_UNMANAGED) == 0) { flags |= PTE_MANAGED; /* Create and insert pv entry. */ pv_insert(pmap, va, m); } pmap->pm_stats.resident_count++; mtx_lock_spin(&tlbivax_mutex); tlb_miss_lock(); tlb0_flush_entry(va); if (pmap->pm_pdir[pdir_idx] == NULL) { /* * If we just allocated a new page table, hook it in * the pdir. */ pmap->pm_pdir[pdir_idx] = ptbl; } pte = &(pmap->pm_pdir[pdir_idx][ptbl_idx]); pte->rpn = VM_PAGE_TO_PHYS(m) & ~PTE_PA_MASK; pte->flags |= (PTE_VALID | flags); tlb_miss_unlock(); mtx_unlock_spin(&tlbivax_mutex); return (0); } /* Return the pa for the given pmap/va. */ static vm_paddr_t pte_vatopa(mmu_t mmu, pmap_t pmap, vm_offset_t va) { vm_paddr_t pa = 0; pte_t *pte; pte = pte_find(mmu, pmap, va); if ((pte != NULL) && PTE_ISVALID(pte)) pa = (PTE_PA(pte) | (va & PTE_PA_MASK)); return (pa); } /* Get a pointer to a PTE in a page table. */ static pte_t * pte_find(mmu_t mmu, pmap_t pmap, vm_offset_t va) { unsigned int pdir_idx = PDIR_IDX(va); unsigned int ptbl_idx = PTBL_IDX(va); KASSERT((pmap != NULL), ("pte_find: invalid pmap")); if (pmap->pm_pdir[pdir_idx]) return (&(pmap->pm_pdir[pdir_idx][ptbl_idx])); return (NULL); } /**************************************************************************/ /* PMAP related */ /**************************************************************************/ /* * This is called during booke_init, before the system is really initialized. */ static void mmu_booke_bootstrap(mmu_t mmu, vm_offset_t start, vm_offset_t kernelend) { vm_offset_t phys_kernelend; struct mem_region *mp, *mp1; int cnt, i, j; u_int s, e, sz; u_int phys_avail_count; vm_size_t physsz, hwphyssz, kstack0_sz; vm_offset_t kernel_pdir, kstack0, va; vm_paddr_t kstack0_phys; void *dpcpu; pte_t *pte; debugf("mmu_booke_bootstrap: entered\n"); /* Set interesting system properties */ hw_direct_map = 0; elf32_nxstack = 1; /* Initialize invalidation mutex */ mtx_init(&tlbivax_mutex, "tlbivax", NULL, MTX_SPIN); /* Read TLB0 size and associativity. */ tlb0_get_tlbconf(); /* * Align kernel start and end address (kernel image). * Note that kernel end does not necessarily relate to kernsize. * kernsize is the size of the kernel that is actually mapped. * Also note that "start - 1" is deliberate. With SMP, the * entry point is exactly a page from the actual load address. * As such, trunc_page() has no effect and we're off by a page. * Since we always have the ELF header between the load address * and the entry point, we can safely subtract 1 to compensate. */ kernstart = trunc_page(start - 1); data_start = round_page(kernelend); data_end = data_start; /* * Addresses of preloaded modules (like file systems) use * physical addresses. Make sure we relocate those into * virtual addresses. */ preload_addr_relocate = kernstart - kernload; /* Allocate the dynamic per-cpu area. */ dpcpu = (void *)data_end; data_end += DPCPU_SIZE; /* Allocate space for the message buffer. */ msgbufp = (struct msgbuf *)data_end; data_end += msgbufsize; debugf(" msgbufp at 0x%08x end = 0x%08x\n", (uint32_t)msgbufp, data_end); data_end = round_page(data_end); /* Allocate space for ptbl_bufs. */ ptbl_bufs = (struct ptbl_buf *)data_end; data_end += sizeof(struct ptbl_buf) * PTBL_BUFS; debugf(" ptbl_bufs at 0x%08x end = 0x%08x\n", (uint32_t)ptbl_bufs, data_end); data_end = round_page(data_end); /* Allocate PTE tables for kernel KVA. */ kernel_pdir = data_end; kernel_ptbls = (VM_MAX_KERNEL_ADDRESS - VM_MIN_KERNEL_ADDRESS + PDIR_SIZE - 1) / PDIR_SIZE; data_end += kernel_ptbls * PTBL_PAGES * PAGE_SIZE; debugf(" kernel ptbls: %d\n", kernel_ptbls); debugf(" kernel pdir at 0x%08x end = 0x%08x\n", kernel_pdir, data_end); debugf(" data_end: 0x%08x\n", data_end); if (data_end - kernstart > kernsize) { kernsize += tlb1_mapin_region(kernstart + kernsize, kernload + kernsize, (data_end - kernstart) - kernsize); } data_end = kernstart + kernsize; debugf(" updated data_end: 0x%08x\n", data_end); /* * Clear the structures - note we can only do it safely after the * possible additional TLB1 translations are in place (above) so that * all range up to the currently calculated 'data_end' is covered. */ dpcpu_init(dpcpu, 0); memset((void *)ptbl_bufs, 0, sizeof(struct ptbl_buf) * PTBL_SIZE); memset((void *)kernel_pdir, 0, kernel_ptbls * PTBL_PAGES * PAGE_SIZE); /*******************************************************/ /* Set the start and end of kva. */ /*******************************************************/ virtual_avail = round_page(data_end); virtual_end = VM_MAX_KERNEL_ADDRESS; /* Allocate KVA space for page zero/copy operations. */ zero_page_va = virtual_avail; virtual_avail += PAGE_SIZE; zero_page_idle_va = virtual_avail; virtual_avail += PAGE_SIZE; copy_page_src_va = virtual_avail; virtual_avail += PAGE_SIZE; copy_page_dst_va = virtual_avail; virtual_avail += PAGE_SIZE; debugf("zero_page_va = 0x%08x\n", zero_page_va); debugf("zero_page_idle_va = 0x%08x\n", zero_page_idle_va); debugf("copy_page_src_va = 0x%08x\n", copy_page_src_va); debugf("copy_page_dst_va = 0x%08x\n", copy_page_dst_va); /* Initialize page zero/copy mutexes. */ mtx_init(&zero_page_mutex, "mmu_booke_zero_page", NULL, MTX_DEF); mtx_init(©_page_mutex, "mmu_booke_copy_page", NULL, MTX_DEF); /* Allocate KVA space for ptbl bufs. */ ptbl_buf_pool_vabase = virtual_avail; virtual_avail += PTBL_BUFS * PTBL_PAGES * PAGE_SIZE; debugf("ptbl_buf_pool_vabase = 0x%08x end = 0x%08x\n", ptbl_buf_pool_vabase, virtual_avail); /* Calculate corresponding physical addresses for the kernel region. */ phys_kernelend = kernload + kernsize; debugf("kernel image and allocated data:\n"); debugf(" kernload = 0x%08x\n", kernload); debugf(" kernstart = 0x%08x\n", kernstart); debugf(" kernsize = 0x%08x\n", kernsize); if (sizeof(phys_avail) / sizeof(phys_avail[0]) < availmem_regions_sz) panic("mmu_booke_bootstrap: phys_avail too small"); /* * Remove kernel physical address range from avail regions list. Page * align all regions. Non-page aligned memory isn't very interesting * to us. Also, sort the entries for ascending addresses. */ /* Retrieve phys/avail mem regions */ mem_regions(&physmem_regions, &physmem_regions_sz, &availmem_regions, &availmem_regions_sz); sz = 0; cnt = availmem_regions_sz; debugf("processing avail regions:\n"); for (mp = availmem_regions; mp->mr_size; mp++) { s = mp->mr_start; e = mp->mr_start + mp->mr_size; debugf(" %08x-%08x -> ", s, e); /* Check whether this region holds all of the kernel. */ if (s < kernload && e > phys_kernelend) { availmem_regions[cnt].mr_start = phys_kernelend; availmem_regions[cnt++].mr_size = e - phys_kernelend; e = kernload; } /* Look whether this regions starts within the kernel. */ if (s >= kernload && s < phys_kernelend) { if (e <= phys_kernelend) goto empty; s = phys_kernelend; } /* Now look whether this region ends within the kernel. */ if (e > kernload && e <= phys_kernelend) { if (s >= kernload) goto empty; e = kernload; } /* Now page align the start and size of the region. */ s = round_page(s); e = trunc_page(e); if (e < s) e = s; sz = e - s; debugf("%08x-%08x = %x\n", s, e, sz); /* Check whether some memory is left here. */ if (sz == 0) { empty: memmove(mp, mp + 1, (cnt - (mp - availmem_regions)) * sizeof(*mp)); cnt--; mp--; continue; } /* Do an insertion sort. */ for (mp1 = availmem_regions; mp1 < mp; mp1++) if (s < mp1->mr_start) break; if (mp1 < mp) { memmove(mp1 + 1, mp1, (char *)mp - (char *)mp1); mp1->mr_start = s; mp1->mr_size = sz; } else { mp->mr_start = s; mp->mr_size = sz; } } availmem_regions_sz = cnt; /*******************************************************/ /* Steal physical memory for kernel stack from the end */ /* of the first avail region */ /*******************************************************/ kstack0_sz = KSTACK_PAGES * PAGE_SIZE; kstack0_phys = availmem_regions[0].mr_start + availmem_regions[0].mr_size; kstack0_phys -= kstack0_sz; availmem_regions[0].mr_size -= kstack0_sz; /*******************************************************/ /* Fill in phys_avail table, based on availmem_regions */ /*******************************************************/ phys_avail_count = 0; physsz = 0; hwphyssz = 0; TUNABLE_ULONG_FETCH("hw.physmem", (u_long *) &hwphyssz); debugf("fill in phys_avail:\n"); for (i = 0, j = 0; i < availmem_regions_sz; i++, j += 2) { debugf(" region: 0x%08x - 0x%08x (0x%08x)\n", availmem_regions[i].mr_start, availmem_regions[i].mr_start + availmem_regions[i].mr_size, availmem_regions[i].mr_size); if (hwphyssz != 0 && (physsz + availmem_regions[i].mr_size) >= hwphyssz) { debugf(" hw.physmem adjust\n"); if (physsz < hwphyssz) { phys_avail[j] = availmem_regions[i].mr_start; phys_avail[j + 1] = availmem_regions[i].mr_start + hwphyssz - physsz; physsz = hwphyssz; phys_avail_count++; } break; } phys_avail[j] = availmem_regions[i].mr_start; phys_avail[j + 1] = availmem_regions[i].mr_start + availmem_regions[i].mr_size; phys_avail_count++; physsz += availmem_regions[i].mr_size; } physmem = btoc(physsz); /* Calculate the last available physical address. */ for (i = 0; phys_avail[i + 2] != 0; i += 2) ; Maxmem = powerpc_btop(phys_avail[i + 1]); debugf("Maxmem = 0x%08lx\n", Maxmem); debugf("phys_avail_count = %d\n", phys_avail_count); debugf("physsz = 0x%08x physmem = %ld (0x%08lx)\n", physsz, physmem, physmem); /*******************************************************/ /* Initialize (statically allocated) kernel pmap. */ /*******************************************************/ PMAP_LOCK_INIT(kernel_pmap); kptbl_min = VM_MIN_KERNEL_ADDRESS / PDIR_SIZE; debugf("kernel_pmap = 0x%08x\n", (uint32_t)kernel_pmap); debugf("kptbl_min = %d, kernel_ptbls = %d\n", kptbl_min, kernel_ptbls); debugf("kernel pdir range: 0x%08x - 0x%08x\n", kptbl_min * PDIR_SIZE, (kptbl_min + kernel_ptbls) * PDIR_SIZE - 1); /* Initialize kernel pdir */ for (i = 0; i < kernel_ptbls; i++) kernel_pmap->pm_pdir[kptbl_min + i] = (pte_t *)(kernel_pdir + (i * PAGE_SIZE * PTBL_PAGES)); for (i = 0; i < MAXCPU; i++) { kernel_pmap->pm_tid[i] = TID_KERNEL; /* Initialize each CPU's tidbusy entry 0 with kernel_pmap */ tidbusy[i][0] = kernel_pmap; } /* * Fill in PTEs covering kernel code and data. They are not required * for address translation, as this area is covered by static TLB1 * entries, but for pte_vatopa() to work correctly with kernel area * addresses. */ for (va = kernstart; va < data_end; va += PAGE_SIZE) { pte = &(kernel_pmap->pm_pdir[PDIR_IDX(va)][PTBL_IDX(va)]); pte->rpn = kernload + (va - kernstart); pte->flags = PTE_M | PTE_SR | PTE_SW | PTE_SX | PTE_WIRED | PTE_VALID; } /* Mark kernel_pmap active on all CPUs */ CPU_FILL(&kernel_pmap->pm_active); /* * Initialize the global pv list lock. */ rw_init(&pvh_global_lock, "pmap pv global"); /*******************************************************/ /* Final setup */ /*******************************************************/ /* Enter kstack0 into kernel map, provide guard page */ kstack0 = virtual_avail + KSTACK_GUARD_PAGES * PAGE_SIZE; thread0.td_kstack = kstack0; thread0.td_kstack_pages = KSTACK_PAGES; debugf("kstack_sz = 0x%08x\n", kstack0_sz); debugf("kstack0_phys at 0x%08x - 0x%08x\n", kstack0_phys, kstack0_phys + kstack0_sz); debugf("kstack0 at 0x%08x - 0x%08x\n", kstack0, kstack0 + kstack0_sz); virtual_avail += KSTACK_GUARD_PAGES * PAGE_SIZE + kstack0_sz; for (i = 0; i < KSTACK_PAGES; i++) { mmu_booke_kenter(mmu, kstack0, kstack0_phys); kstack0 += PAGE_SIZE; kstack0_phys += PAGE_SIZE; } pmap_bootstrapped = 1; debugf("virtual_avail = %08x\n", virtual_avail); debugf("virtual_end = %08x\n", virtual_end); debugf("mmu_booke_bootstrap: exit\n"); } void pmap_bootstrap_ap(volatile uint32_t *trcp __unused) { int i; /* * Finish TLB1 configuration: the BSP already set up its TLB1 and we * have the snapshot of its contents in the s/w tlb1[] table, so use * these values directly to (re)program AP's TLB1 hardware. */ for (i = bp_ntlb1s; i < tlb1_idx; i++) { /* Skip invalid entries */ if (!(tlb1[i].mas1 & MAS1_VALID)) continue; tlb1_write_entry(i); } set_mas4_defaults(); } /* * Get the physical page address for the given pmap/virtual address. */ static vm_paddr_t mmu_booke_extract(mmu_t mmu, pmap_t pmap, vm_offset_t va) { vm_paddr_t pa; PMAP_LOCK(pmap); pa = pte_vatopa(mmu, pmap, va); PMAP_UNLOCK(pmap); return (pa); } /* * Extract the physical page address associated with the given * kernel virtual address. */ static vm_paddr_t mmu_booke_kextract(mmu_t mmu, vm_offset_t va) { int i; /* Check TLB1 mappings */ for (i = 0; i < tlb1_idx; i++) { if (!(tlb1[i].mas1 & MAS1_VALID)) continue; if (va >= tlb1[i].virt && va < tlb1[i].virt + tlb1[i].size) return (tlb1[i].phys + (va - tlb1[i].virt)); } return (pte_vatopa(mmu, kernel_pmap, va)); } /* * Initialize the pmap module. * Called by vm_init, to initialize any structures that the pmap * system needs to map virtual memory. */ static void mmu_booke_init(mmu_t mmu) { int shpgperproc = PMAP_SHPGPERPROC; /* * Initialize the address space (zone) for the pv entries. Set a * high water mark so that the system can recover from excessive * numbers of pv entries. */ pvzone = uma_zcreate("PV ENTRY", sizeof(struct pv_entry), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_VM | UMA_ZONE_NOFREE); TUNABLE_INT_FETCH("vm.pmap.shpgperproc", &shpgperproc); pv_entry_max = shpgperproc * maxproc + vm_cnt.v_page_count; TUNABLE_INT_FETCH("vm.pmap.pv_entries", &pv_entry_max); pv_entry_high_water = 9 * (pv_entry_max / 10); uma_zone_reserve_kva(pvzone, pv_entry_max); /* Pre-fill pvzone with initial number of pv entries. */ uma_prealloc(pvzone, PV_ENTRY_ZONE_MIN); /* Initialize ptbl allocation. */ ptbl_init(); } /* * Map a list of wired pages into kernel virtual address space. This is * intended for temporary mappings which do not need page modification or * references recorded. Existing mappings in the region are overwritten. */ static void mmu_booke_qenter(mmu_t mmu, vm_offset_t sva, vm_page_t *m, int count) { vm_offset_t va; va = sva; while (count-- > 0) { mmu_booke_kenter(mmu, va, VM_PAGE_TO_PHYS(*m)); va += PAGE_SIZE; m++; } } /* * Remove page mappings from kernel virtual address space. Intended for * temporary mappings entered by mmu_booke_qenter. */ static void mmu_booke_qremove(mmu_t mmu, vm_offset_t sva, int count) { vm_offset_t va; va = sva; while (count-- > 0) { mmu_booke_kremove(mmu, va); va += PAGE_SIZE; } } /* * Map a wired page into kernel virtual address space. */ static void mmu_booke_kenter(mmu_t mmu, vm_offset_t va, vm_paddr_t pa) { mmu_booke_kenter_attr(mmu, va, pa, VM_MEMATTR_DEFAULT); } static void mmu_booke_kenter_attr(mmu_t mmu, vm_offset_t va, vm_paddr_t pa, vm_memattr_t ma) { unsigned int pdir_idx = PDIR_IDX(va); unsigned int ptbl_idx = PTBL_IDX(va); uint32_t flags; pte_t *pte; KASSERT(((va >= VM_MIN_KERNEL_ADDRESS) && (va <= VM_MAX_KERNEL_ADDRESS)), ("mmu_booke_kenter: invalid va")); flags = PTE_SR | PTE_SW | PTE_SX | PTE_WIRED | PTE_VALID; flags |= tlb_calc_wimg(pa, ma); pte = &(kernel_pmap->pm_pdir[pdir_idx][ptbl_idx]); mtx_lock_spin(&tlbivax_mutex); tlb_miss_lock(); if (PTE_ISVALID(pte)) { CTR1(KTR_PMAP, "%s: replacing entry!", __func__); /* Flush entry from TLB0 */ tlb0_flush_entry(va); } pte->rpn = pa & ~PTE_PA_MASK; pte->flags = flags; //debugf("mmu_booke_kenter: pdir_idx = %d ptbl_idx = %d va=0x%08x " // "pa=0x%08x rpn=0x%08x flags=0x%08x\n", // pdir_idx, ptbl_idx, va, pa, pte->rpn, pte->flags); /* Flush the real memory from the instruction cache. */ if ((flags & (PTE_I | PTE_G)) == 0) { __syncicache((void *)va, PAGE_SIZE); } tlb_miss_unlock(); mtx_unlock_spin(&tlbivax_mutex); } /* * Remove a page from kernel page table. */ static void mmu_booke_kremove(mmu_t mmu, vm_offset_t va) { unsigned int pdir_idx = PDIR_IDX(va); unsigned int ptbl_idx = PTBL_IDX(va); pte_t *pte; // CTR2(KTR_PMAP,("%s: s (va = 0x%08x)\n", __func__, va)); KASSERT(((va >= VM_MIN_KERNEL_ADDRESS) && (va <= VM_MAX_KERNEL_ADDRESS)), ("mmu_booke_kremove: invalid va")); pte = &(kernel_pmap->pm_pdir[pdir_idx][ptbl_idx]); if (!PTE_ISVALID(pte)) { CTR1(KTR_PMAP, "%s: invalid pte", __func__); return; } mtx_lock_spin(&tlbivax_mutex); tlb_miss_lock(); /* Invalidate entry in TLB0, update PTE. */ tlb0_flush_entry(va); pte->flags = 0; pte->rpn = 0; tlb_miss_unlock(); mtx_unlock_spin(&tlbivax_mutex); } /* * Initialize pmap associated with process 0. */ static void mmu_booke_pinit0(mmu_t mmu, pmap_t pmap) { PMAP_LOCK_INIT(pmap); mmu_booke_pinit(mmu, pmap); PCPU_SET(curpmap, pmap); } /* * Initialize a preallocated and zeroed pmap structure, * such as one in a vmspace structure. */ static void mmu_booke_pinit(mmu_t mmu, pmap_t pmap) { int i; CTR4(KTR_PMAP, "%s: pmap = %p, proc %d '%s'", __func__, pmap, curthread->td_proc->p_pid, curthread->td_proc->p_comm); KASSERT((pmap != kernel_pmap), ("pmap_pinit: initializing kernel_pmap")); for (i = 0; i < MAXCPU; i++) pmap->pm_tid[i] = TID_NONE; CPU_ZERO(&kernel_pmap->pm_active); bzero(&pmap->pm_stats, sizeof(pmap->pm_stats)); bzero(&pmap->pm_pdir, sizeof(pte_t *) * PDIR_NENTRIES); TAILQ_INIT(&pmap->pm_ptbl_list); } /* * Release any resources held by the given physical map. * Called when a pmap initialized by mmu_booke_pinit is being released. * Should only be called if the map contains no valid mappings. */ static void mmu_booke_release(mmu_t mmu, pmap_t pmap) { KASSERT(pmap->pm_stats.resident_count == 0, ("pmap_release: pmap resident count %ld != 0", pmap->pm_stats.resident_count)); } /* * Insert the given physical page at the specified virtual address in the * target physical map with the protection requested. If specified the page * will be wired down. */ static int mmu_booke_enter(mmu_t mmu, pmap_t pmap, vm_offset_t va, vm_page_t m, vm_prot_t prot, u_int flags, int8_t psind) { int error; rw_wlock(&pvh_global_lock); PMAP_LOCK(pmap); error = mmu_booke_enter_locked(mmu, pmap, va, m, prot, flags, psind); rw_wunlock(&pvh_global_lock); PMAP_UNLOCK(pmap); return (error); } static int mmu_booke_enter_locked(mmu_t mmu, pmap_t pmap, vm_offset_t va, vm_page_t m, vm_prot_t prot, u_int pmap_flags, int8_t psind __unused) { pte_t *pte; vm_paddr_t pa; uint32_t flags; int error, su, sync; pa = VM_PAGE_TO_PHYS(m); su = (pmap == kernel_pmap); sync = 0; //debugf("mmu_booke_enter_locked: s (pmap=0x%08x su=%d tid=%d m=0x%08x va=0x%08x " // "pa=0x%08x prot=0x%08x flags=%#x)\n", // (u_int32_t)pmap, su, pmap->pm_tid, // (u_int32_t)m, va, pa, prot, flags); if (su) { KASSERT(((va >= virtual_avail) && (va <= VM_MAX_KERNEL_ADDRESS)), ("mmu_booke_enter_locked: kernel pmap, non kernel va")); } else { KASSERT((va <= VM_MAXUSER_ADDRESS), ("mmu_booke_enter_locked: user pmap, non user va")); } if ((m->oflags & VPO_UNMANAGED) == 0 && !vm_page_xbusied(m)) VM_OBJECT_ASSERT_LOCKED(m->object); PMAP_LOCK_ASSERT(pmap, MA_OWNED); /* * If there is an existing mapping, and the physical address has not * changed, must be protection or wiring change. */ if (((pte = pte_find(mmu, pmap, va)) != NULL) && (PTE_ISVALID(pte)) && (PTE_PA(pte) == pa)) { /* * Before actually updating pte->flags we calculate and * prepare its new value in a helper var. */ flags = pte->flags; flags &= ~(PTE_UW | PTE_UX | PTE_SW | PTE_SX | PTE_MODIFIED); /* Wiring change, just update stats. */ if ((pmap_flags & PMAP_ENTER_WIRED) != 0) { if (!PTE_ISWIRED(pte)) { flags |= PTE_WIRED; pmap->pm_stats.wired_count++; } } else { if (PTE_ISWIRED(pte)) { flags &= ~PTE_WIRED; pmap->pm_stats.wired_count--; } } if (prot & VM_PROT_WRITE) { /* Add write permissions. */ flags |= PTE_SW; if (!su) flags |= PTE_UW; if ((flags & PTE_MANAGED) != 0) vm_page_aflag_set(m, PGA_WRITEABLE); } else { /* Handle modified pages, sense modify status. */ /* * The PTE_MODIFIED flag could be set by underlying * TLB misses since we last read it (above), possibly * other CPUs could update it so we check in the PTE * directly rather than rely on that saved local flags * copy. */ if (PTE_ISMODIFIED(pte)) vm_page_dirty(m); } if (prot & VM_PROT_EXECUTE) { flags |= PTE_SX; if (!su) flags |= PTE_UX; /* * Check existing flags for execute permissions: if we * are turning execute permissions on, icache should * be flushed. */ if ((pte->flags & (PTE_UX | PTE_SX)) == 0) sync++; } flags &= ~PTE_REFERENCED; /* * The new flags value is all calculated -- only now actually * update the PTE. */ mtx_lock_spin(&tlbivax_mutex); tlb_miss_lock(); tlb0_flush_entry(va); pte->flags = flags; tlb_miss_unlock(); mtx_unlock_spin(&tlbivax_mutex); } else { /* * If there is an existing mapping, but it's for a different * physical address, pte_enter() will delete the old mapping. */ //if ((pte != NULL) && PTE_ISVALID(pte)) // debugf("mmu_booke_enter_locked: replace\n"); //else // debugf("mmu_booke_enter_locked: new\n"); /* Now set up the flags and install the new mapping. */ flags = (PTE_SR | PTE_VALID); flags |= PTE_M; if (!su) flags |= PTE_UR; if (prot & VM_PROT_WRITE) { flags |= PTE_SW; if (!su) flags |= PTE_UW; if ((m->oflags & VPO_UNMANAGED) == 0) vm_page_aflag_set(m, PGA_WRITEABLE); } if (prot & VM_PROT_EXECUTE) { flags |= PTE_SX; if (!su) flags |= PTE_UX; } /* If its wired update stats. */ if ((pmap_flags & PMAP_ENTER_WIRED) != 0) flags |= PTE_WIRED; error = pte_enter(mmu, pmap, m, va, flags, (pmap_flags & PMAP_ENTER_NOSLEEP) != 0); if (error != 0) return (KERN_RESOURCE_SHORTAGE); if ((flags & PMAP_ENTER_WIRED) != 0) pmap->pm_stats.wired_count++; /* Flush the real memory from the instruction cache. */ if (prot & VM_PROT_EXECUTE) sync++; } if (sync && (su || pmap == PCPU_GET(curpmap))) { __syncicache((void *)va, PAGE_SIZE); sync = 0; } return (KERN_SUCCESS); } /* * Maps a sequence of resident pages belonging to the same object. * The sequence begins with the given page m_start. This page is * mapped at the given virtual address start. Each subsequent page is * mapped at a virtual address that is offset from start by the same * amount as the page is offset from m_start within the object. The * last page in the sequence is the page with the largest offset from * m_start that can be mapped at a virtual address less than the given * virtual address end. Not every virtual page between start and end * is mapped; only those for which a resident page exists with the * corresponding offset from m_start are mapped. */ static void mmu_booke_enter_object(mmu_t mmu, pmap_t pmap, vm_offset_t start, vm_offset_t end, vm_page_t m_start, vm_prot_t prot) { vm_page_t m; vm_pindex_t diff, psize; VM_OBJECT_ASSERT_LOCKED(m_start->object); psize = atop(end - start); m = m_start; rw_wlock(&pvh_global_lock); PMAP_LOCK(pmap); while (m != NULL && (diff = m->pindex - m_start->pindex) < psize) { mmu_booke_enter_locked(mmu, pmap, start + ptoa(diff), m, prot & (VM_PROT_READ | VM_PROT_EXECUTE), PMAP_ENTER_NOSLEEP, 0); m = TAILQ_NEXT(m, listq); } rw_wunlock(&pvh_global_lock); PMAP_UNLOCK(pmap); } static void mmu_booke_enter_quick(mmu_t mmu, pmap_t pmap, vm_offset_t va, vm_page_t m, vm_prot_t prot) { rw_wlock(&pvh_global_lock); PMAP_LOCK(pmap); mmu_booke_enter_locked(mmu, pmap, va, m, prot & (VM_PROT_READ | VM_PROT_EXECUTE), PMAP_ENTER_NOSLEEP, 0); rw_wunlock(&pvh_global_lock); PMAP_UNLOCK(pmap); } /* * Remove the given range of addresses from the specified map. * * It is assumed that the start and end are properly rounded to the page size. */ static void mmu_booke_remove(mmu_t mmu, pmap_t pmap, vm_offset_t va, vm_offset_t endva) { pte_t *pte; uint8_t hold_flag; int su = (pmap == kernel_pmap); //debugf("mmu_booke_remove: s (su = %d pmap=0x%08x tid=%d va=0x%08x endva=0x%08x)\n", // su, (u_int32_t)pmap, pmap->pm_tid, va, endva); if (su) { KASSERT(((va >= virtual_avail) && (va <= VM_MAX_KERNEL_ADDRESS)), ("mmu_booke_remove: kernel pmap, non kernel va")); } else { KASSERT((va <= VM_MAXUSER_ADDRESS), ("mmu_booke_remove: user pmap, non user va")); } if (PMAP_REMOVE_DONE(pmap)) { //debugf("mmu_booke_remove: e (empty)\n"); return; } hold_flag = PTBL_HOLD_FLAG(pmap); //debugf("mmu_booke_remove: hold_flag = %d\n", hold_flag); rw_wlock(&pvh_global_lock); PMAP_LOCK(pmap); for (; va < endva; va += PAGE_SIZE) { pte = pte_find(mmu, pmap, va); if ((pte != NULL) && PTE_ISVALID(pte)) pte_remove(mmu, pmap, va, hold_flag); } PMAP_UNLOCK(pmap); rw_wunlock(&pvh_global_lock); //debugf("mmu_booke_remove: e\n"); } /* * Remove physical page from all pmaps in which it resides. */ static void mmu_booke_remove_all(mmu_t mmu, vm_page_t m) { pv_entry_t pv, pvn; uint8_t hold_flag; rw_wlock(&pvh_global_lock); for (pv = TAILQ_FIRST(&m->md.pv_list); pv != NULL; pv = pvn) { pvn = TAILQ_NEXT(pv, pv_link); PMAP_LOCK(pv->pv_pmap); hold_flag = PTBL_HOLD_FLAG(pv->pv_pmap); pte_remove(mmu, pv->pv_pmap, pv->pv_va, hold_flag); PMAP_UNLOCK(pv->pv_pmap); } vm_page_aflag_clear(m, PGA_WRITEABLE); rw_wunlock(&pvh_global_lock); } /* * Map a range of physical addresses into kernel virtual address space. */ static vm_offset_t mmu_booke_map(mmu_t mmu, vm_offset_t *virt, vm_paddr_t pa_start, vm_paddr_t pa_end, int prot) { vm_offset_t sva = *virt; vm_offset_t va = sva; //debugf("mmu_booke_map: s (sva = 0x%08x pa_start = 0x%08x pa_end = 0x%08x)\n", // sva, pa_start, pa_end); while (pa_start < pa_end) { mmu_booke_kenter(mmu, va, pa_start); va += PAGE_SIZE; pa_start += PAGE_SIZE; } *virt = va; //debugf("mmu_booke_map: e (va = 0x%08x)\n", va); return (sva); } /* * The pmap must be activated before it's address space can be accessed in any * way. */ static void mmu_booke_activate(mmu_t mmu, struct thread *td) { pmap_t pmap; u_int cpuid; pmap = &td->td_proc->p_vmspace->vm_pmap; CTR5(KTR_PMAP, "%s: s (td = %p, proc = '%s', id = %d, pmap = 0x%08x)", __func__, td, td->td_proc->p_comm, td->td_proc->p_pid, pmap); KASSERT((pmap != kernel_pmap), ("mmu_booke_activate: kernel_pmap!")); sched_pin(); cpuid = PCPU_GET(cpuid); CPU_SET_ATOMIC(cpuid, &pmap->pm_active); PCPU_SET(curpmap, pmap); if (pmap->pm_tid[cpuid] == TID_NONE) tid_alloc(pmap); /* Load PID0 register with pmap tid value. */ mtspr(SPR_PID0, pmap->pm_tid[cpuid]); __asm __volatile("isync"); + mtspr(SPR_DBCR0, td->td_pcb->pcb_cpu.booke.dbcr0); + sched_unpin(); CTR3(KTR_PMAP, "%s: e (tid = %d for '%s')", __func__, pmap->pm_tid[PCPU_GET(cpuid)], td->td_proc->p_comm); } /* * Deactivate the specified process's address space. */ static void mmu_booke_deactivate(mmu_t mmu, struct thread *td) { pmap_t pmap; pmap = &td->td_proc->p_vmspace->vm_pmap; CTR5(KTR_PMAP, "%s: td=%p, proc = '%s', id = %d, pmap = 0x%08x", __func__, td, td->td_proc->p_comm, td->td_proc->p_pid, pmap); + + td->td_pcb->pcb_cpu.booke.dbcr0 = mfspr(SPR_DBCR0); CPU_CLR_ATOMIC(PCPU_GET(cpuid), &pmap->pm_active); PCPU_SET(curpmap, NULL); } /* * Copy the range specified by src_addr/len * from the source map to the range dst_addr/len * in the destination map. * * This routine is only advisory and need not do anything. */ static void mmu_booke_copy(mmu_t mmu, pmap_t dst_pmap, pmap_t src_pmap, vm_offset_t dst_addr, vm_size_t len, vm_offset_t src_addr) { } /* * Set the physical protection on the specified range of this map as requested. */ static void mmu_booke_protect(mmu_t mmu, pmap_t pmap, vm_offset_t sva, vm_offset_t eva, vm_prot_t prot) { vm_offset_t va; vm_page_t m; pte_t *pte; if ((prot & VM_PROT_READ) == VM_PROT_NONE) { mmu_booke_remove(mmu, pmap, sva, eva); return; } if (prot & VM_PROT_WRITE) return; PMAP_LOCK(pmap); for (va = sva; va < eva; va += PAGE_SIZE) { if ((pte = pte_find(mmu, pmap, va)) != NULL) { if (PTE_ISVALID(pte)) { m = PHYS_TO_VM_PAGE(PTE_PA(pte)); mtx_lock_spin(&tlbivax_mutex); tlb_miss_lock(); /* Handle modified pages. */ if (PTE_ISMODIFIED(pte) && PTE_ISMANAGED(pte)) vm_page_dirty(m); tlb0_flush_entry(va); pte->flags &= ~(PTE_UW | PTE_SW | PTE_MODIFIED); tlb_miss_unlock(); mtx_unlock_spin(&tlbivax_mutex); } } } PMAP_UNLOCK(pmap); } /* * Clear the write and modified bits in each of the given page's mappings. */ static void mmu_booke_remove_write(mmu_t mmu, vm_page_t m) { pv_entry_t pv; pte_t *pte; KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("mmu_booke_remove_write: page %p is not managed", m)); /* * If the page is not exclusive busied, then PGA_WRITEABLE cannot be * set by another thread while the object is locked. Thus, * if PGA_WRITEABLE is clear, no page table entries need updating. */ VM_OBJECT_ASSERT_WLOCKED(m->object); if (!vm_page_xbusied(m) && (m->aflags & PGA_WRITEABLE) == 0) return; rw_wlock(&pvh_global_lock); TAILQ_FOREACH(pv, &m->md.pv_list, pv_link) { PMAP_LOCK(pv->pv_pmap); if ((pte = pte_find(mmu, pv->pv_pmap, pv->pv_va)) != NULL) { if (PTE_ISVALID(pte)) { m = PHYS_TO_VM_PAGE(PTE_PA(pte)); mtx_lock_spin(&tlbivax_mutex); tlb_miss_lock(); /* Handle modified pages. */ if (PTE_ISMODIFIED(pte)) vm_page_dirty(m); /* Flush mapping from TLB0. */ pte->flags &= ~(PTE_UW | PTE_SW | PTE_MODIFIED); tlb_miss_unlock(); mtx_unlock_spin(&tlbivax_mutex); } } PMAP_UNLOCK(pv->pv_pmap); } vm_page_aflag_clear(m, PGA_WRITEABLE); rw_wunlock(&pvh_global_lock); } static void mmu_booke_sync_icache(mmu_t mmu, pmap_t pm, vm_offset_t va, vm_size_t sz) { pte_t *pte; pmap_t pmap; vm_page_t m; vm_offset_t addr; vm_paddr_t pa = 0; int active, valid; va = trunc_page(va); sz = round_page(sz); rw_wlock(&pvh_global_lock); pmap = PCPU_GET(curpmap); active = (pm == kernel_pmap || pm == pmap) ? 1 : 0; while (sz > 0) { PMAP_LOCK(pm); pte = pte_find(mmu, pm, va); valid = (pte != NULL && PTE_ISVALID(pte)) ? 1 : 0; if (valid) pa = PTE_PA(pte); PMAP_UNLOCK(pm); if (valid) { if (!active) { /* Create a mapping in the active pmap. */ addr = 0; m = PHYS_TO_VM_PAGE(pa); PMAP_LOCK(pmap); pte_enter(mmu, pmap, m, addr, PTE_SR | PTE_VALID | PTE_UR, FALSE); __syncicache((void *)addr, PAGE_SIZE); pte_remove(mmu, pmap, addr, PTBL_UNHOLD); PMAP_UNLOCK(pmap); } else __syncicache((void *)va, PAGE_SIZE); } va += PAGE_SIZE; sz -= PAGE_SIZE; } rw_wunlock(&pvh_global_lock); } /* * Atomically extract and hold the physical page with the given * pmap and virtual address pair if that mapping permits the given * protection. */ static vm_page_t mmu_booke_extract_and_hold(mmu_t mmu, pmap_t pmap, vm_offset_t va, vm_prot_t prot) { pte_t *pte; vm_page_t m; uint32_t pte_wbit; vm_paddr_t pa; m = NULL; pa = 0; PMAP_LOCK(pmap); retry: pte = pte_find(mmu, pmap, va); if ((pte != NULL) && PTE_ISVALID(pte)) { if (pmap == kernel_pmap) pte_wbit = PTE_SW; else pte_wbit = PTE_UW; if ((pte->flags & pte_wbit) || ((prot & VM_PROT_WRITE) == 0)) { if (vm_page_pa_tryrelock(pmap, PTE_PA(pte), &pa)) goto retry; m = PHYS_TO_VM_PAGE(PTE_PA(pte)); vm_page_hold(m); } } PA_UNLOCK_COND(pa); PMAP_UNLOCK(pmap); return (m); } /* * Initialize a vm_page's machine-dependent fields. */ static void mmu_booke_page_init(mmu_t mmu, vm_page_t m) { TAILQ_INIT(&m->md.pv_list); } /* * mmu_booke_zero_page_area zeros the specified hardware page by * mapping it into virtual memory and using bzero to clear * its contents. * * off and size must reside within a single page. */ static void mmu_booke_zero_page_area(mmu_t mmu, vm_page_t m, int off, int size) { vm_offset_t va; /* XXX KASSERT off and size are within a single page? */ mtx_lock(&zero_page_mutex); va = zero_page_va; mmu_booke_kenter(mmu, va, VM_PAGE_TO_PHYS(m)); bzero((caddr_t)va + off, size); mmu_booke_kremove(mmu, va); mtx_unlock(&zero_page_mutex); } /* * mmu_booke_zero_page zeros the specified hardware page. */ static void mmu_booke_zero_page(mmu_t mmu, vm_page_t m) { mmu_booke_zero_page_area(mmu, m, 0, PAGE_SIZE); } /* * mmu_booke_copy_page copies the specified (machine independent) page by * mapping the page into virtual memory and using memcopy to copy the page, * one machine dependent page at a time. */ static void mmu_booke_copy_page(mmu_t mmu, vm_page_t sm, vm_page_t dm) { vm_offset_t sva, dva; sva = copy_page_src_va; dva = copy_page_dst_va; mtx_lock(©_page_mutex); mmu_booke_kenter(mmu, sva, VM_PAGE_TO_PHYS(sm)); mmu_booke_kenter(mmu, dva, VM_PAGE_TO_PHYS(dm)); memcpy((caddr_t)dva, (caddr_t)sva, PAGE_SIZE); mmu_booke_kremove(mmu, dva); mmu_booke_kremove(mmu, sva); mtx_unlock(©_page_mutex); } static inline void mmu_booke_copy_pages(mmu_t mmu, vm_page_t *ma, vm_offset_t a_offset, vm_page_t *mb, vm_offset_t b_offset, int xfersize) { void *a_cp, *b_cp; vm_offset_t a_pg_offset, b_pg_offset; int cnt; mtx_lock(©_page_mutex); while (xfersize > 0) { a_pg_offset = a_offset & PAGE_MASK; cnt = min(xfersize, PAGE_SIZE - a_pg_offset); mmu_booke_kenter(mmu, copy_page_src_va, VM_PAGE_TO_PHYS(ma[a_offset >> PAGE_SHIFT])); a_cp = (char *)copy_page_src_va + a_pg_offset; b_pg_offset = b_offset & PAGE_MASK; cnt = min(cnt, PAGE_SIZE - b_pg_offset); mmu_booke_kenter(mmu, copy_page_dst_va, VM_PAGE_TO_PHYS(mb[b_offset >> PAGE_SHIFT])); b_cp = (char *)copy_page_dst_va + b_pg_offset; bcopy(a_cp, b_cp, cnt); mmu_booke_kremove(mmu, copy_page_dst_va); mmu_booke_kremove(mmu, copy_page_src_va); a_offset += cnt; b_offset += cnt; xfersize -= cnt; } mtx_unlock(©_page_mutex); } /* * mmu_booke_zero_page_idle zeros the specified hardware page by mapping it * into virtual memory and using bzero to clear its contents. This is intended * to be called from the vm_pagezero process only and outside of Giant. No * lock is required. */ static void mmu_booke_zero_page_idle(mmu_t mmu, vm_page_t m) { vm_offset_t va; va = zero_page_idle_va; mmu_booke_kenter(mmu, va, VM_PAGE_TO_PHYS(m)); bzero((caddr_t)va, PAGE_SIZE); mmu_booke_kremove(mmu, va); } /* * Return whether or not the specified physical page was modified * in any of physical maps. */ static boolean_t mmu_booke_is_modified(mmu_t mmu, vm_page_t m) { pte_t *pte; pv_entry_t pv; boolean_t rv; KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("mmu_booke_is_modified: page %p is not managed", m)); rv = FALSE; /* * If the page is not exclusive busied, then PGA_WRITEABLE cannot be * concurrently set while the object is locked. Thus, if PGA_WRITEABLE * is clear, no PTEs can be modified. */ VM_OBJECT_ASSERT_WLOCKED(m->object); if (!vm_page_xbusied(m) && (m->aflags & PGA_WRITEABLE) == 0) return (rv); rw_wlock(&pvh_global_lock); TAILQ_FOREACH(pv, &m->md.pv_list, pv_link) { PMAP_LOCK(pv->pv_pmap); if ((pte = pte_find(mmu, pv->pv_pmap, pv->pv_va)) != NULL && PTE_ISVALID(pte)) { if (PTE_ISMODIFIED(pte)) rv = TRUE; } PMAP_UNLOCK(pv->pv_pmap); if (rv) break; } rw_wunlock(&pvh_global_lock); return (rv); } /* * Return whether or not the specified virtual address is eligible * for prefault. */ static boolean_t mmu_booke_is_prefaultable(mmu_t mmu, pmap_t pmap, vm_offset_t addr) { return (FALSE); } /* * Return whether or not the specified physical page was referenced * in any physical maps. */ static boolean_t mmu_booke_is_referenced(mmu_t mmu, vm_page_t m) { pte_t *pte; pv_entry_t pv; boolean_t rv; KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("mmu_booke_is_referenced: page %p is not managed", m)); rv = FALSE; rw_wlock(&pvh_global_lock); TAILQ_FOREACH(pv, &m->md.pv_list, pv_link) { PMAP_LOCK(pv->pv_pmap); if ((pte = pte_find(mmu, pv->pv_pmap, pv->pv_va)) != NULL && PTE_ISVALID(pte)) { if (PTE_ISREFERENCED(pte)) rv = TRUE; } PMAP_UNLOCK(pv->pv_pmap); if (rv) break; } rw_wunlock(&pvh_global_lock); return (rv); } /* * Clear the modify bits on the specified physical page. */ static void mmu_booke_clear_modify(mmu_t mmu, vm_page_t m) { pte_t *pte; pv_entry_t pv; KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("mmu_booke_clear_modify: page %p is not managed", m)); VM_OBJECT_ASSERT_WLOCKED(m->object); KASSERT(!vm_page_xbusied(m), ("mmu_booke_clear_modify: page %p is exclusive busied", m)); /* * If the page is not PG_AWRITEABLE, then no PTEs can be modified. * If the object containing the page is locked and the page is not * exclusive busied, then PG_AWRITEABLE cannot be concurrently set. */ if ((m->aflags & PGA_WRITEABLE) == 0) return; rw_wlock(&pvh_global_lock); TAILQ_FOREACH(pv, &m->md.pv_list, pv_link) { PMAP_LOCK(pv->pv_pmap); if ((pte = pte_find(mmu, pv->pv_pmap, pv->pv_va)) != NULL && PTE_ISVALID(pte)) { mtx_lock_spin(&tlbivax_mutex); tlb_miss_lock(); if (pte->flags & (PTE_SW | PTE_UW | PTE_MODIFIED)) { tlb0_flush_entry(pv->pv_va); pte->flags &= ~(PTE_SW | PTE_UW | PTE_MODIFIED | PTE_REFERENCED); } tlb_miss_unlock(); mtx_unlock_spin(&tlbivax_mutex); } PMAP_UNLOCK(pv->pv_pmap); } rw_wunlock(&pvh_global_lock); } /* * Return a count of reference bits for a page, clearing those bits. * It is not necessary for every reference bit to be cleared, but it * is necessary that 0 only be returned when there are truly no * reference bits set. * * XXX: The exact number of bits to check and clear is a matter that * should be tested and standardized at some point in the future for * optimal aging of shared pages. */ static int mmu_booke_ts_referenced(mmu_t mmu, vm_page_t m) { pte_t *pte; pv_entry_t pv; int count; KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("mmu_booke_ts_referenced: page %p is not managed", m)); count = 0; rw_wlock(&pvh_global_lock); TAILQ_FOREACH(pv, &m->md.pv_list, pv_link) { PMAP_LOCK(pv->pv_pmap); if ((pte = pte_find(mmu, pv->pv_pmap, pv->pv_va)) != NULL && PTE_ISVALID(pte)) { if (PTE_ISREFERENCED(pte)) { mtx_lock_spin(&tlbivax_mutex); tlb_miss_lock(); tlb0_flush_entry(pv->pv_va); pte->flags &= ~PTE_REFERENCED; tlb_miss_unlock(); mtx_unlock_spin(&tlbivax_mutex); if (++count > 4) { PMAP_UNLOCK(pv->pv_pmap); break; } } } PMAP_UNLOCK(pv->pv_pmap); } rw_wunlock(&pvh_global_lock); return (count); } /* * Clear the wired attribute from the mappings for the specified range of * addresses in the given pmap. Every valid mapping within that range must * have the wired attribute set. In contrast, invalid mappings cannot have * the wired attribute set, so they are ignored. * * The wired attribute of the page table entry is not a hardware feature, so * there is no need to invalidate any TLB entries. */ static void mmu_booke_unwire(mmu_t mmu, pmap_t pmap, vm_offset_t sva, vm_offset_t eva) { vm_offset_t va; pte_t *pte; PMAP_LOCK(pmap); for (va = sva; va < eva; va += PAGE_SIZE) { if ((pte = pte_find(mmu, pmap, va)) != NULL && PTE_ISVALID(pte)) { if (!PTE_ISWIRED(pte)) panic("mmu_booke_unwire: pte %p isn't wired", pte); pte->flags &= ~PTE_WIRED; pmap->pm_stats.wired_count--; } } PMAP_UNLOCK(pmap); } /* * Return true if the pmap's pv is one of the first 16 pvs linked to from this * page. This count may be changed upwards or downwards in the future; it is * only necessary that true be returned for a small subset of pmaps for proper * page aging. */ static boolean_t mmu_booke_page_exists_quick(mmu_t mmu, pmap_t pmap, vm_page_t m) { pv_entry_t pv; int loops; boolean_t rv; KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("mmu_booke_page_exists_quick: page %p is not managed", m)); loops = 0; rv = FALSE; rw_wlock(&pvh_global_lock); TAILQ_FOREACH(pv, &m->md.pv_list, pv_link) { if (pv->pv_pmap == pmap) { rv = TRUE; break; } if (++loops >= 16) break; } rw_wunlock(&pvh_global_lock); return (rv); } /* * Return the number of managed mappings to the given physical page that are * wired. */ static int mmu_booke_page_wired_mappings(mmu_t mmu, vm_page_t m) { pv_entry_t pv; pte_t *pte; int count = 0; if ((m->oflags & VPO_UNMANAGED) != 0) return (count); rw_wlock(&pvh_global_lock); TAILQ_FOREACH(pv, &m->md.pv_list, pv_link) { PMAP_LOCK(pv->pv_pmap); if ((pte = pte_find(mmu, pv->pv_pmap, pv->pv_va)) != NULL) if (PTE_ISVALID(pte) && PTE_ISWIRED(pte)) count++; PMAP_UNLOCK(pv->pv_pmap); } rw_wunlock(&pvh_global_lock); return (count); } static int mmu_booke_dev_direct_mapped(mmu_t mmu, vm_paddr_t pa, vm_size_t size) { int i; vm_offset_t va; /* * This currently does not work for entries that * overlap TLB1 entries. */ for (i = 0; i < tlb1_idx; i ++) { if (tlb1_iomapped(i, pa, size, &va) == 0) return (0); } return (EFAULT); } void mmu_booke_dumpsys_map(mmu_t mmu, vm_paddr_t pa, size_t sz, void **va) { vm_paddr_t ppa; vm_offset_t ofs; vm_size_t gran; /* Minidumps are based on virtual memory addresses. */ if (do_minidump) { *va = (void *)pa; return; } /* Raw physical memory dumps don't have a virtual address. */ /* We always map a 256MB page at 256M. */ gran = 256 * 1024 * 1024; ppa = pa & ~(gran - 1); ofs = pa - ppa; *va = (void *)gran; tlb1_set_entry((vm_offset_t)va, ppa, gran, _TLB_ENTRY_IO); if (sz > (gran - ofs)) tlb1_set_entry((vm_offset_t)(va + gran), ppa + gran, gran, _TLB_ENTRY_IO); } void mmu_booke_dumpsys_unmap(mmu_t mmu, vm_paddr_t pa, size_t sz, void *va) { vm_paddr_t ppa; vm_offset_t ofs; vm_size_t gran; /* Minidumps are based on virtual memory addresses. */ /* Nothing to do... */ if (do_minidump) return; /* Raw physical memory dumps don't have a virtual address. */ tlb1_idx--; tlb1[tlb1_idx].mas1 = 0; tlb1[tlb1_idx].mas2 = 0; tlb1[tlb1_idx].mas3 = 0; tlb1_write_entry(tlb1_idx); gran = 256 * 1024 * 1024; ppa = pa & ~(gran - 1); ofs = pa - ppa; if (sz > (gran - ofs)) { tlb1_idx--; tlb1[tlb1_idx].mas1 = 0; tlb1[tlb1_idx].mas2 = 0; tlb1[tlb1_idx].mas3 = 0; tlb1_write_entry(tlb1_idx); } } extern struct dump_pa dump_map[PHYS_AVAIL_SZ + 1]; void mmu_booke_scan_init(mmu_t mmu) { vm_offset_t va; pte_t *pte; int i; if (!do_minidump) { /* Initialize phys. segments for dumpsys(). */ memset(&dump_map, 0, sizeof(dump_map)); mem_regions(&physmem_regions, &physmem_regions_sz, &availmem_regions, &availmem_regions_sz); for (i = 0; i < physmem_regions_sz; i++) { dump_map[i].pa_start = physmem_regions[i].mr_start; dump_map[i].pa_size = physmem_regions[i].mr_size; } return; } /* Virtual segments for minidumps: */ memset(&dump_map, 0, sizeof(dump_map)); /* 1st: kernel .data and .bss. */ dump_map[0].pa_start = trunc_page((uintptr_t)_etext); dump_map[0].pa_size = round_page((uintptr_t)_end) - dump_map[0].pa_start; /* 2nd: msgbuf and tables (see pmap_bootstrap()). */ dump_map[1].pa_start = data_start; dump_map[1].pa_size = data_end - data_start; /* 3rd: kernel VM. */ va = dump_map[1].pa_start + dump_map[1].pa_size; /* Find start of next chunk (from va). */ while (va < virtual_end) { /* Don't dump the buffer cache. */ if (va >= kmi.buffer_sva && va < kmi.buffer_eva) { va = kmi.buffer_eva; continue; } pte = pte_find(mmu, kernel_pmap, va); if (pte != NULL && PTE_ISVALID(pte)) break; va += PAGE_SIZE; } if (va < virtual_end) { dump_map[2].pa_start = va; va += PAGE_SIZE; /* Find last page in chunk. */ while (va < virtual_end) { /* Don't run into the buffer cache. */ if (va == kmi.buffer_sva) break; pte = pte_find(mmu, kernel_pmap, va); if (pte == NULL || !PTE_ISVALID(pte)) break; va += PAGE_SIZE; } dump_map[2].pa_size = va - dump_map[2].pa_start; } } /* * Map a set of physical memory pages into the kernel virtual address space. * Return a pointer to where it is mapped. This routine is intended to be used * for mapping device memory, NOT real memory. */ static void * mmu_booke_mapdev(mmu_t mmu, vm_paddr_t pa, vm_size_t size) { return (mmu_booke_mapdev_attr(mmu, pa, size, VM_MEMATTR_DEFAULT)); } static void * mmu_booke_mapdev_attr(mmu_t mmu, vm_paddr_t pa, vm_size_t size, vm_memattr_t ma) { void *res; uintptr_t va; vm_size_t sz; int i; /* * Check if this is premapped in TLB1. Note: this should probably also * check whether a sequence of TLB1 entries exist that match the * requirement, but now only checks the easy case. */ if (ma == VM_MEMATTR_DEFAULT) { for (i = 0; i < tlb1_idx; i++) { if (!(tlb1[i].mas1 & MAS1_VALID)) continue; if (pa >= tlb1[i].phys && (pa + size) <= (tlb1[i].phys + tlb1[i].size)) return (void *)(tlb1[i].virt + (pa - tlb1[i].phys)); } } size = roundup(size, PAGE_SIZE); /* * We leave a hole for device direct mapping between the maximum user * address (0x8000000) and the minimum KVA address (0xc0000000). If * devices are in there, just map them 1:1. If not, map them to the * device mapping area about VM_MAX_KERNEL_ADDRESS. These mapped * addresses should be pulled from an allocator, but since we do not * ever free TLB1 entries, it is safe just to increment a counter. * Note that there isn't a lot of address space here (128 MB) and it * is not at all difficult to imagine running out, since that is a 4:1 * compression from the 0xc0000000 - 0xf0000000 address space that gets * mapped there. */ if (pa >= (VM_MAXUSER_ADDRESS + PAGE_SIZE) && (pa + size - 1) < VM_MIN_KERNEL_ADDRESS) va = pa; else va = atomic_fetchadd_int(&tlb1_map_base, size); res = (void *)va; do { sz = 1 << (ilog2(size) & ~1); if (bootverbose) printf("Wiring VA=%x to PA=%x (size=%x), " "using TLB1[%d]\n", va, pa, sz, tlb1_idx); tlb1_set_entry(va, pa, sz, tlb_calc_wimg(pa, ma)); size -= sz; pa += sz; va += sz; } while (size > 0); return (res); } /* * 'Unmap' a range mapped by mmu_booke_mapdev(). */ static void mmu_booke_unmapdev(mmu_t mmu, vm_offset_t va, vm_size_t size) { #ifdef SUPPORTS_SHRINKING_TLB1 vm_offset_t base, offset; /* * Unmap only if this is inside kernel virtual space. */ if ((va >= VM_MIN_KERNEL_ADDRESS) && (va <= VM_MAX_KERNEL_ADDRESS)) { base = trunc_page(va); offset = va & PAGE_MASK; size = roundup(offset + size, PAGE_SIZE); kva_free(base, size); } #endif } /* * mmu_booke_object_init_pt preloads the ptes for a given object into the * specified pmap. This eliminates the blast of soft faults on process startup * and immediately after an mmap. */ static void mmu_booke_object_init_pt(mmu_t mmu, pmap_t pmap, vm_offset_t addr, vm_object_t object, vm_pindex_t pindex, vm_size_t size) { VM_OBJECT_ASSERT_WLOCKED(object); KASSERT(object->type == OBJT_DEVICE || object->type == OBJT_SG, ("mmu_booke_object_init_pt: non-device object")); } /* * Perform the pmap work for mincore. */ static int mmu_booke_mincore(mmu_t mmu, pmap_t pmap, vm_offset_t addr, vm_paddr_t *locked_pa) { /* XXX: this should be implemented at some point */ return (0); } /**************************************************************************/ /* TID handling */ /**************************************************************************/ /* * Allocate a TID. If necessary, steal one from someone else. * The new TID is flushed from the TLB before returning. */ static tlbtid_t tid_alloc(pmap_t pmap) { tlbtid_t tid; int thiscpu; KASSERT((pmap != kernel_pmap), ("tid_alloc: kernel pmap")); CTR2(KTR_PMAP, "%s: s (pmap = %p)", __func__, pmap); thiscpu = PCPU_GET(cpuid); tid = PCPU_GET(tid_next); if (tid > TID_MAX) tid = TID_MIN; PCPU_SET(tid_next, tid + 1); /* If we are stealing TID then clear the relevant pmap's field */ if (tidbusy[thiscpu][tid] != NULL) { CTR2(KTR_PMAP, "%s: warning: stealing tid %d", __func__, tid); tidbusy[thiscpu][tid]->pm_tid[thiscpu] = TID_NONE; /* Flush all entries from TLB0 matching this TID. */ tid_flush(tid); } tidbusy[thiscpu][tid] = pmap; pmap->pm_tid[thiscpu] = tid; __asm __volatile("msync; isync"); CTR3(KTR_PMAP, "%s: e (%02d next = %02d)", __func__, tid, PCPU_GET(tid_next)); return (tid); } /**************************************************************************/ /* TLB0 handling */ /**************************************************************************/ static void tlb_print_entry(int i, uint32_t mas1, uint32_t mas2, uint32_t mas3, uint32_t mas7) { int as; char desc[3]; tlbtid_t tid; vm_size_t size; unsigned int tsize; desc[2] = '\0'; if (mas1 & MAS1_VALID) desc[0] = 'V'; else desc[0] = ' '; if (mas1 & MAS1_IPROT) desc[1] = 'P'; else desc[1] = ' '; as = (mas1 & MAS1_TS_MASK) ? 1 : 0; tid = MAS1_GETTID(mas1); tsize = (mas1 & MAS1_TSIZE_MASK) >> MAS1_TSIZE_SHIFT; size = 0; if (tsize) size = tsize2size(tsize); debugf("%3d: (%s) [AS=%d] " "sz = 0x%08x tsz = %d tid = %d mas1 = 0x%08x " "mas2(va) = 0x%08x mas3(pa) = 0x%08x mas7 = 0x%08x\n", i, desc, as, size, tsize, tid, mas1, mas2, mas3, mas7); } /* Convert TLB0 va and way number to tlb0[] table index. */ static inline unsigned int tlb0_tableidx(vm_offset_t va, unsigned int way) { unsigned int idx; idx = (way * TLB0_ENTRIES_PER_WAY); idx += (va & MAS2_TLB0_ENTRY_IDX_MASK) >> MAS2_TLB0_ENTRY_IDX_SHIFT; return (idx); } /* * Invalidate TLB0 entry. */ static inline void tlb0_flush_entry(vm_offset_t va) { CTR2(KTR_PMAP, "%s: s va=0x%08x", __func__, va); mtx_assert(&tlbivax_mutex, MA_OWNED); __asm __volatile("tlbivax 0, %0" :: "r"(va & MAS2_EPN_MASK)); __asm __volatile("isync; msync"); __asm __volatile("tlbsync; msync"); CTR1(KTR_PMAP, "%s: e", __func__); } /* Print out contents of the MAS registers for each TLB0 entry */ void tlb0_print_tlbentries(void) { uint32_t mas0, mas1, mas2, mas3, mas7; int entryidx, way, idx; debugf("TLB0 entries:\n"); for (way = 0; way < TLB0_WAYS; way ++) for (entryidx = 0; entryidx < TLB0_ENTRIES_PER_WAY; entryidx++) { mas0 = MAS0_TLBSEL(0) | MAS0_ESEL(way); mtspr(SPR_MAS0, mas0); __asm __volatile("isync"); mas2 = entryidx << MAS2_TLB0_ENTRY_IDX_SHIFT; mtspr(SPR_MAS2, mas2); __asm __volatile("isync; tlbre"); mas1 = mfspr(SPR_MAS1); mas2 = mfspr(SPR_MAS2); mas3 = mfspr(SPR_MAS3); mas7 = mfspr(SPR_MAS7); idx = tlb0_tableidx(mas2, way); tlb_print_entry(idx, mas1, mas2, mas3, mas7); } } /**************************************************************************/ /* TLB1 handling */ /**************************************************************************/ /* * TLB1 mapping notes: * * TLB1[0] Kernel text and data. * TLB1[1-15] Additional kernel text and data mappings (if required), PCI * windows, other devices mappings. */ /* * Write given entry to TLB1 hardware. * Use 32 bit pa, clear 4 high-order bits of RPN (mas7). */ static void tlb1_write_entry(unsigned int idx) { uint32_t mas0, mas7; //debugf("tlb1_write_entry: s\n"); /* Clear high order RPN bits */ mas7 = 0; /* Select entry */ mas0 = MAS0_TLBSEL(1) | MAS0_ESEL(idx); //debugf("tlb1_write_entry: mas0 = 0x%08x\n", mas0); mtspr(SPR_MAS0, mas0); __asm __volatile("isync"); mtspr(SPR_MAS1, tlb1[idx].mas1); __asm __volatile("isync"); mtspr(SPR_MAS2, tlb1[idx].mas2); __asm __volatile("isync"); mtspr(SPR_MAS3, tlb1[idx].mas3); __asm __volatile("isync"); mtspr(SPR_MAS7, mas7); __asm __volatile("isync; tlbwe; isync; msync"); //debugf("tlb1_write_entry: e\n"); } /* * Return the largest uint value log such that 2^log <= num. */ static unsigned int ilog2(unsigned int num) { int lz; __asm ("cntlzw %0, %1" : "=r" (lz) : "r" (num)); return (31 - lz); } /* * Convert TLB TSIZE value to mapped region size. */ static vm_size_t tsize2size(unsigned int tsize) { /* * size = 4^tsize KB * size = 4^tsize * 2^10 = 2^(2 * tsize - 10) */ return ((1 << (2 * tsize)) * 1024); } /* * Convert region size (must be power of 4) to TLB TSIZE value. */ static unsigned int size2tsize(vm_size_t size) { return (ilog2(size) / 2 - 5); } /* * Register permanent kernel mapping in TLB1. * * Entries are created starting from index 0 (current free entry is * kept in tlb1_idx) and are not supposed to be invalidated. */ static int tlb1_set_entry(vm_offset_t va, vm_offset_t pa, vm_size_t size, uint32_t flags) { uint32_t ts, tid; int tsize, index; index = atomic_fetchadd_int(&tlb1_idx, 1); if (index >= TLB1_ENTRIES) { printf("tlb1_set_entry: TLB1 full!\n"); return (-1); } /* Convert size to TSIZE */ tsize = size2tsize(size); tid = (TID_KERNEL << MAS1_TID_SHIFT) & MAS1_TID_MASK; /* XXX TS is hard coded to 0 for now as we only use single address space */ ts = (0 << MAS1_TS_SHIFT) & MAS1_TS_MASK; /* * Atomicity is preserved by the atomic increment above since nothing * is ever removed from tlb1. */ tlb1[index].phys = pa; tlb1[index].virt = va; tlb1[index].size = size; tlb1[index].mas1 = MAS1_VALID | MAS1_IPROT | ts | tid; tlb1[index].mas1 |= ((tsize << MAS1_TSIZE_SHIFT) & MAS1_TSIZE_MASK); tlb1[index].mas2 = (va & MAS2_EPN_MASK) | flags; /* Set supervisor RWX permission bits */ tlb1[index].mas3 = (pa & MAS3_RPN) | MAS3_SR | MAS3_SW | MAS3_SX; tlb1_write_entry(index); /* * XXX in general TLB1 updates should be propagated between CPUs, * since current design assumes to have the same TLB1 set-up on all * cores. */ return (0); } /* * Map in contiguous RAM region into the TLB1 using maximum of * KERNEL_REGION_MAX_TLB_ENTRIES entries. * * If necessary round up last entry size and return total size * used by all allocated entries. */ vm_size_t tlb1_mapin_region(vm_offset_t va, vm_paddr_t pa, vm_size_t size) { vm_size_t pgs[KERNEL_REGION_MAX_TLB_ENTRIES]; vm_size_t mapped, pgsz, base, mask; int idx, nents; /* Round up to the next 1M */ size = (size + (1 << 20) - 1) & ~((1 << 20) - 1); mapped = 0; idx = 0; base = va; pgsz = 64*1024*1024; while (mapped < size) { while (mapped < size && idx < KERNEL_REGION_MAX_TLB_ENTRIES) { while (pgsz > (size - mapped)) pgsz >>= 2; pgs[idx++] = pgsz; mapped += pgsz; } /* We under-map. Correct for this. */ if (mapped < size) { while (pgs[idx - 1] == pgsz) { idx--; mapped -= pgsz; } /* XXX We may increase beyond out starting point. */ pgsz <<= 2; pgs[idx++] = pgsz; mapped += pgsz; } } nents = idx; mask = pgs[0] - 1; /* Align address to the boundary */ if (va & mask) { va = (va + mask) & ~mask; pa = (pa + mask) & ~mask; } for (idx = 0; idx < nents; idx++) { pgsz = pgs[idx]; debugf("%u: %x -> %x, size=%x\n", idx, pa, va, pgsz); tlb1_set_entry(va, pa, pgsz, _TLB_ENTRY_MEM); pa += pgsz; va += pgsz; } mapped = (va - base); printf("mapped size 0x%08x (wasted space 0x%08x)\n", mapped, mapped - size); return (mapped); } /* * TLB1 initialization routine, to be called after the very first * assembler level setup done in locore.S. */ void tlb1_init() { uint32_t mas0, mas1, mas2, mas3; uint32_t tsz; u_int i; if (bootinfo != NULL && bootinfo[0] != 1) { tlb1_idx = *((uint16_t *)(bootinfo + 8)); } else tlb1_idx = 1; /* The first entry/entries are used to map the kernel. */ for (i = 0; i < tlb1_idx; i++) { mas0 = MAS0_TLBSEL(1) | MAS0_ESEL(i); mtspr(SPR_MAS0, mas0); __asm __volatile("isync; tlbre"); mas1 = mfspr(SPR_MAS1); if ((mas1 & MAS1_VALID) == 0) continue; mas2 = mfspr(SPR_MAS2); mas3 = mfspr(SPR_MAS3); tlb1[i].mas1 = mas1; tlb1[i].mas2 = mfspr(SPR_MAS2); tlb1[i].mas3 = mas3; tlb1[i].virt = mas2 & MAS2_EPN_MASK; tlb1[i].phys = mas3 & MAS3_RPN; if (i == 0) kernload = mas3 & MAS3_RPN; tsz = (mas1 & MAS1_TSIZE_MASK) >> MAS1_TSIZE_SHIFT; tlb1[i].size = (tsz > 0) ? tsize2size(tsz) : 0; kernsize += tlb1[i].size; } #ifdef SMP bp_ntlb1s = tlb1_idx; #endif /* Purge the remaining entries */ for (i = tlb1_idx; i < TLB1_ENTRIES; i++) tlb1_write_entry(i); /* Setup TLB miss defaults */ set_mas4_defaults(); } vm_offset_t pmap_early_io_map(vm_paddr_t pa, vm_size_t size) { vm_paddr_t pa_base; vm_offset_t va, sz; int i; KASSERT(!pmap_bootstrapped, ("Do not use after PMAP is up!")); for (i = 0; i < tlb1_idx; i++) { if (!(tlb1[i].mas1 & MAS1_VALID)) continue; if (pa >= tlb1[i].phys && (pa + size) <= (tlb1[i].phys + tlb1[i].size)) return (tlb1[i].virt + (pa - tlb1[i].phys)); } pa_base = trunc_page(pa); size = roundup(size + (pa - pa_base), PAGE_SIZE); tlb1_map_base = roundup2(tlb1_map_base, 1 << (ilog2(size) & ~1)); va = tlb1_map_base + (pa - pa_base); do { sz = 1 << (ilog2(size) & ~1); tlb1_set_entry(tlb1_map_base, pa_base, sz, _TLB_ENTRY_IO); size -= sz; pa_base += sz; tlb1_map_base += sz; } while (size > 0); #ifdef SMP bp_ntlb1s = tlb1_idx; #endif return (va); } /* * Setup MAS4 defaults. * These values are loaded to MAS0-2 on a TLB miss. */ static void set_mas4_defaults(void) { uint32_t mas4; /* Defaults: TLB0, PID0, TSIZED=4K */ mas4 = MAS4_TLBSELD0; mas4 |= (TLB_SIZE_4K << MAS4_TSIZED_SHIFT) & MAS4_TSIZED_MASK; #ifdef SMP mas4 |= MAS4_MD; #endif mtspr(SPR_MAS4, mas4); __asm __volatile("isync"); } /* * Print out contents of the MAS registers for each TLB1 entry */ void tlb1_print_tlbentries(void) { uint32_t mas0, mas1, mas2, mas3, mas7; int i; debugf("TLB1 entries:\n"); for (i = 0; i < TLB1_ENTRIES; i++) { mas0 = MAS0_TLBSEL(1) | MAS0_ESEL(i); mtspr(SPR_MAS0, mas0); __asm __volatile("isync; tlbre"); mas1 = mfspr(SPR_MAS1); mas2 = mfspr(SPR_MAS2); mas3 = mfspr(SPR_MAS3); mas7 = mfspr(SPR_MAS7); tlb_print_entry(i, mas1, mas2, mas3, mas7); } } /* * Print out contents of the in-ram tlb1 table. */ void tlb1_print_entries(void) { int i; debugf("tlb1[] table entries:\n"); for (i = 0; i < TLB1_ENTRIES; i++) tlb_print_entry(i, tlb1[i].mas1, tlb1[i].mas2, tlb1[i].mas3, 0); } /* * Return 0 if the physical IO range is encompassed by one of the * the TLB1 entries, otherwise return related error code. */ static int tlb1_iomapped(int i, vm_paddr_t pa, vm_size_t size, vm_offset_t *va) { uint32_t prot; vm_paddr_t pa_start; vm_paddr_t pa_end; unsigned int entry_tsize; vm_size_t entry_size; *va = (vm_offset_t)NULL; /* Skip invalid entries */ if (!(tlb1[i].mas1 & MAS1_VALID)) return (EINVAL); /* * The entry must be cache-inhibited, guarded, and r/w * so it can function as an i/o page */ prot = tlb1[i].mas2 & (MAS2_I | MAS2_G); if (prot != (MAS2_I | MAS2_G)) return (EPERM); prot = tlb1[i].mas3 & (MAS3_SR | MAS3_SW); if (prot != (MAS3_SR | MAS3_SW)) return (EPERM); /* The address should be within the entry range. */ entry_tsize = (tlb1[i].mas1 & MAS1_TSIZE_MASK) >> MAS1_TSIZE_SHIFT; KASSERT((entry_tsize), ("tlb1_iomapped: invalid entry tsize")); entry_size = tsize2size(entry_tsize); pa_start = tlb1[i].mas3 & MAS3_RPN; pa_end = pa_start + entry_size - 1; if ((pa < pa_start) || ((pa + size) > pa_end)) return (ERANGE); /* Return virtual address of this mapping. */ *va = (tlb1[i].mas2 & MAS2_EPN_MASK) + (pa - pa_start); return (0); } Index: projects/cxl_iscsi/sys/powerpc/booke/trap.c =================================================================== --- projects/cxl_iscsi/sys/powerpc/booke/trap.c (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/booke/trap.c (revision 279717) @@ -1,519 +1,519 @@ /*- * Copyright (C) 1995, 1996 Wolfgang Solfrank. * Copyright (C) 1995, 1996 TooLs GmbH. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by TooLs GmbH. * 4. The name of TooLs GmbH may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * $NetBSD: trap.c,v 1.58 2002/03/04 04:07:35 dbj Exp $ */ #include __FBSDID("$FreeBSD$"); #include "opt_fpu_emu.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define FAULTBUF_LR 0 #define FAULTBUF_R1 1 #define FAULTBUF_R2 2 #define FAULTBUF_CR 3 #define FAULTBUF_CTR 4 #define FAULTBUF_XER 5 #define FAULTBUF_R13 6 static void trap_fatal(struct trapframe *frame); static void printtrap(u_int vector, struct trapframe *frame, int isfatal, int user); static int trap_pfault(struct trapframe *frame, int user); static int fix_unaligned(struct thread *td, struct trapframe *frame); static int handle_onfault(struct trapframe *frame); static void syscall(struct trapframe *frame); struct powerpc_exception { u_int vector; char *name; }; static struct powerpc_exception powerpc_exceptions[] = { { EXC_CRIT, "critical input" }, { EXC_MCHK, "machine check" }, { EXC_DSI, "data storage interrupt" }, { EXC_ISI, "instruction storage interrupt" }, { EXC_EXI, "external interrupt" }, { EXC_ALI, "alignment" }, { EXC_PGM, "program" }, { EXC_SC, "system call" }, { EXC_APU, "auxiliary proc unavailable" }, { EXC_DECR, "decrementer" }, { EXC_FIT, "fixed-interval timer" }, { EXC_WDOG, "watchdog timer" }, { EXC_DTMISS, "data tlb miss" }, { EXC_ITMISS, "instruction tlb miss" }, { EXC_DEBUG, "debug" }, { EXC_PERF, "performance monitoring" }, { EXC_LAST, NULL } }; static const char * trapname(u_int vector) { struct powerpc_exception *pe; for (pe = powerpc_exceptions; pe->vector != EXC_LAST; pe++) { if (pe->vector == vector) return (pe->name); } return ("unknown"); } void trap(struct trapframe *frame) { struct thread *td; struct proc *p; int sig, type, user; ksiginfo_t ksi; #ifdef KDB if (kdb_active) { kdb_reenter(); return; } #endif PCPU_INC(cnt.v_trap); td = curthread; p = td->td_proc; type = frame->exc; sig = 0; user = (frame->srr1 & PSL_PR) ? 1 : 0; CTR3(KTR_TRAP, "trap: %s type=%s (%s)", p->p_comm, trapname(type), user ? "user" : "kernel"); if (user) { td->td_frame = frame; if (td->td_ucred != p->p_ucred) cred_update_thread(td); /* User Mode Traps */ switch (type) { case EXC_DSI: case EXC_ISI: sig = trap_pfault(frame, 1); break; case EXC_SC: syscall(frame); break; case EXC_ALI: if (fix_unaligned(td, frame) != 0) sig = SIGBUS; else frame->srr0 += 4; break; case EXC_DEBUG: /* Single stepping */ mtspr(SPR_DBSR, mfspr(SPR_DBSR)); frame->srr1 &= ~PSL_DE; frame->cpu.booke.dbcr0 &= ~(DBCR0_IDM || DBCR0_IC); sig = SIGTRAP; break; case EXC_PGM: /* Program exception */ sig = ppc_instr_emulate(frame, td->td_pcb); break; default: trap_fatal(frame); } } else { /* Kernel Mode Traps */ KASSERT(cold || td->td_ucred != NULL, ("kernel trap doesn't have ucred")); switch (type) { case EXC_DEBUG: mtspr(SPR_DBSR, mfspr(SPR_DBSR)); kdb_trap(frame->exc, 0, frame); return; case EXC_DSI: if (trap_pfault(frame, 0) == 0) return; break; case EXC_MCHK: if (handle_onfault(frame)) return; break; #ifdef KDB case EXC_PGM: if (frame->cpu.booke.esr & ESR_PTR) kdb_trap(EXC_PGM, 0, frame); return; #endif default: break; } trap_fatal(frame); } if (sig != 0) { if (p->p_sysent->sv_transtrap != NULL) sig = (p->p_sysent->sv_transtrap)(sig, type); ksiginfo_init_trap(&ksi); ksi.ksi_signo = sig; ksi.ksi_code = type; /* XXX, not POSIX */ /* ksi.ksi_addr = ? */ ksi.ksi_trapno = type; trapsignal(td, &ksi); } userret(td, frame); } static void trap_fatal(struct trapframe *frame) { printtrap(frame->exc, frame, 1, (frame->srr1 & PSL_PR)); #ifdef KDB if ((debugger_on_panic || kdb_active) && kdb_trap(frame->exc, 0, frame)) return; #endif panic("%s trap", trapname(frame->exc)); } static void printtrap(u_int vector, struct trapframe *frame, int isfatal, int user) { register_t va = 0; printf("\n"); printf("%s %s trap:\n", isfatal ? "fatal" : "handled", user ? "user" : "kernel"); printf("\n"); printf(" exception = 0x%x (%s)\n", vector, trapname(vector)); switch (vector) { case EXC_DTMISS: case EXC_DSI: - va = frame->cpu.booke.dear; + va = frame->dar; break; case EXC_ITMISS: case EXC_ISI: va = frame->srr0; break; } printf(" virtual address = 0x%08x\n", va); printf(" srr0 = 0x%08x\n", frame->srr0); printf(" srr1 = 0x%08x\n", frame->srr1); printf(" curthread = %p\n", curthread); if (curthread != NULL) printf(" pid = %d, comm = %s\n", curthread->td_proc->p_pid, curthread->td_proc->p_comm); printf("\n"); } /* * Handles a fatal fault when we have onfault state to recover. Returns * non-zero if there was onfault recovery state available. */ static int handle_onfault(struct trapframe *frame) { struct thread *td; faultbuf *fb; td = curthread; fb = td->td_pcb->pcb_onfault; if (fb != NULL) { frame->srr0 = (*fb)[FAULTBUF_LR]; frame->fixreg[1] = (*fb)[FAULTBUF_R1]; frame->fixreg[2] = (*fb)[FAULTBUF_R2]; frame->fixreg[3] = 1; frame->cr = (*fb)[FAULTBUF_CR]; frame->ctr = (*fb)[FAULTBUF_CTR]; frame->xer = (*fb)[FAULTBUF_XER]; bcopy(&(*fb)[FAULTBUF_R13], &frame->fixreg[13], 19 * sizeof(register_t)); return (1); } return (0); } int cpu_fetch_syscall_args(struct thread *td, struct syscall_args *sa) { struct proc *p; struct trapframe *frame; caddr_t params; int error, n; p = td->td_proc; frame = td->td_frame; sa->code = frame->fixreg[0]; params = (caddr_t)(frame->fixreg + FIRSTARG); n = NARGREG; if (sa->code == SYS_syscall) { /* * code is first argument, * followed by actual args. */ sa->code = *(u_int *) params; params += sizeof(register_t); n -= 1; } else if (sa->code == SYS___syscall) { /* * Like syscall, but code is a quad, * so as to maintain quad alignment * for the rest of the args. */ params += sizeof(register_t); sa->code = *(u_int *) params; params += sizeof(register_t); n -= 2; } if (p->p_sysent->sv_mask) sa->code &= p->p_sysent->sv_mask; if (sa->code >= p->p_sysent->sv_size) sa->callp = &p->p_sysent->sv_table[0]; else sa->callp = &p->p_sysent->sv_table[sa->code]; sa->narg = sa->callp->sy_narg; bcopy(params, sa->args, n * sizeof(register_t)); if (sa->narg > n) { error = copyin(MOREARGS(frame->fixreg[1]), sa->args + n, (sa->narg - n) * sizeof(register_t)); } else error = 0; if (error == 0) { td->td_retval[0] = 0; td->td_retval[1] = frame->fixreg[FIRSTARG + 1]; } return (error); } #include "../../kern/subr_syscall.c" void syscall(struct trapframe *frame) { struct thread *td; struct syscall_args sa; int error; td = curthread; td->td_frame = frame; error = syscallenter(td, &sa); syscallret(td, error, &sa); } static int trap_pfault(struct trapframe *frame, int user) { vm_offset_t eva, va; struct thread *td; struct proc *p; vm_map_t map; vm_prot_t ftype; int rv; td = curthread; p = td->td_proc; if (frame->exc == EXC_ISI) { eva = frame->srr0; ftype = VM_PROT_READ | VM_PROT_EXECUTE; } else { - eva = frame->cpu.booke.dear; + eva = frame->dar; if (frame->cpu.booke.esr & ESR_ST) ftype = VM_PROT_WRITE; else ftype = VM_PROT_READ; } if (user) { KASSERT(p->p_vmspace != NULL, ("trap_pfault: vmspace NULL")); map = &p->p_vmspace->vm_map; } else { if (eva < VM_MAXUSER_ADDRESS) { if (p->p_vmspace == NULL) return (SIGSEGV); map = &p->p_vmspace->vm_map; } else { map = kernel_map; } } va = trunc_page(eva); if (map != kernel_map) { /* * Keep swapout from messing with us during this * critical time. */ PROC_LOCK(p); ++p->p_lock; PROC_UNLOCK(p); /* Fault in the user page: */ rv = vm_fault(map, va, ftype, VM_FAULT_NORMAL); PROC_LOCK(p); --p->p_lock; PROC_UNLOCK(p); } else { /* * Don't have to worry about process locking or stacks in the * kernel. */ rv = vm_fault(map, va, ftype, VM_FAULT_NORMAL); } if (rv == KERN_SUCCESS) return (0); if (!user && handle_onfault(frame)) return (0); return ((rv == KERN_PROTECTION_FAILURE) ? SIGBUS : SIGSEGV); } /* * For now, this only deals with the particular unaligned access case * that gcc tends to generate. Eventually it should handle all of the * possibilities that can happen on a 32-bit PowerPC in big-endian mode. */ static int fix_unaligned(struct thread *td, struct trapframe *frame) { #if 0 struct thread *fputhread; int indicator, reg; double *fpr; indicator = EXC_ALI_OPCODE_INDICATOR(frame->dsisr); switch (indicator) { case EXC_ALI_LFD: case EXC_ALI_STFD: reg = EXC_ALI_RST(frame->dsisr); fpr = &td->td_pcb->pcb_fpu.fpr[reg]; fputhread = PCPU_GET(fputhread); /* Juggle the FPU to ensure that we've initialized * the FPRs, and that their current state is in * the PCB. */ if (fputhread != td) { if (fputhread) save_fpu(fputhread); enable_fpu(td); } save_fpu(td); if (indicator == EXC_ALI_LFD) { if (copyin((void *)frame->dar, fpr, sizeof(double)) != 0) return -1; enable_fpu(td); } else { if (copyout(fpr, (void *)frame->dar, sizeof(double)) != 0) return -1; } return 0; break; } #endif return (-1); } #ifdef KDB int db_trap_glue(struct trapframe *); int db_trap_glue(struct trapframe *tf) { if (!(tf->srr1 & PSL_PR)) return (kdb_trap(tf->exc, 0, tf)); return (0); } #endif Index: projects/cxl_iscsi/sys/powerpc/include/frame.h =================================================================== --- projects/cxl_iscsi/sys/powerpc/include/frame.h (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/include/frame.h (revision 279717) @@ -1,116 +1,115 @@ /*- * Copyright (C) 1995, 1996 Wolfgang Solfrank. * Copyright (C) 1995, 1996 TooLs GmbH. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by TooLs GmbH. * 4. The name of TooLs GmbH may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * $NetBSD: frame.h,v 1.2 1999/01/10 10:13:15 tsubai Exp $ * $FreeBSD$ */ #ifndef _MACHINE_FRAME_H_ #define _MACHINE_FRAME_H_ #include /* * We have to save all registers on every trap, because * 1. user could attach this process every time * 2. we must be able to restore all user registers in case of fork * Actually, we do not save the fp registers on trap, since * these are not used by the kernel. They are saved only when switching * between processes using the FPU. * * Change ordering to cluster together these register_t's. XXX */ struct trapframe { register_t fixreg[32]; register_t lr; register_t cr; register_t xer; register_t ctr; register_t srr0; register_t srr1; register_t exc; + register_t dar; /* DAR/DEAR filled in on DSI traps */ union { struct { - /* dar & dsisr are only filled on a DSI trap */ - register_t dar; + /* dsisr only filled on a DSI trap */ register_t dsisr; } aim; struct { - register_t dear; register_t esr; register_t dbcr0; } booke; } cpu; }; /* * FRAMELEN is the size of the stack region used by the low-level trap * handler. It is the size of its data (trapframe) plus the callframe * header (sizeof(struct callframe) - 3 register widths). It must also * be 16-byte aligned. */ #define FRAMELEN roundup(sizeof(struct trapframe) + \ sizeof(struct callframe) - 3*sizeof(register_t), 16) #define trapframe(td) ((td)->td_frame) /* * Call frame for PowerPC used during fork. */ #ifdef __powerpc64__ struct callframe { register_t cf_dummy_fp; /* dummy frame pointer */ register_t cf_cr; register_t cf_lr; register_t cf_compiler; register_t cf_linkeditor; register_t cf_toc; register_t cf_func; register_t cf_arg0; register_t cf_arg1; register_t _padding; /* Maintain 16-byte alignment */ }; #else struct callframe { register_t cf_dummy_fp; /* dummy frame pointer */ register_t cf_lr; /* space for link register save */ register_t cf_func; register_t cf_arg0; register_t cf_arg1; register_t _padding; /* Maintain 16-byte alignment */ }; #endif /* Definitions for syscalls */ #define FIRSTARG 3 /* first arg in reg 3 */ #define NARGREG 8 /* 8 args in regs */ #define MOREARGS(sp) ((caddr_t)((uintptr_t)(sp) + \ sizeof(struct callframe) - 3*sizeof(register_t))) /* more args go here */ #endif /* _MACHINE_FRAME_H_ */ Index: projects/cxl_iscsi/sys/powerpc/include/pte.h =================================================================== --- projects/cxl_iscsi/sys/powerpc/include/pte.h (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/include/pte.h (revision 279717) @@ -1,283 +1,277 @@ /*- * Copyright (C) 1995, 1996 Wolfgang Solfrank. * Copyright (C) 1995, 1996 TooLs GmbH. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by TooLs GmbH. * 4. The name of TooLs GmbH may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * $NetBSD: pte.h,v 1.2 1998/08/31 14:43:40 tsubai Exp $ * $FreeBSD$ */ #ifndef _MACHINE_PTE_H_ #define _MACHINE_PTE_H_ #if defined(AIM) /* * Page Table Entries */ #ifndef LOCORE /* 32-bit PTE */ struct pte { u_int32_t pte_hi; u_int32_t pte_lo; }; struct pteg { struct pte pt[8]; }; /* 64-bit (long) PTE */ struct lpte { u_int64_t pte_hi; u_int64_t pte_lo; }; struct lpteg { struct lpte pt[8]; }; #endif /* LOCORE */ /* 32-bit PTE definitions */ /* High word: */ #define PTE_VALID 0x80000000 #define PTE_VSID_SHFT 7 #define PTE_HID 0x00000040 #define PTE_API 0x0000003f /* Low word: */ #define PTE_RPGN 0xfffff000 #define PTE_REF 0x00000100 #define PTE_CHG 0x00000080 #define PTE_WIMG 0x00000078 #define PTE_W 0x00000040 #define PTE_I 0x00000020 #define PTE_M 0x00000010 #define PTE_G 0x00000008 #define PTE_PP 0x00000003 #define PTE_SO 0x00000000 /* Super. Only (U: XX, S: RW) */ #define PTE_SW 0x00000001 /* Super. Write-Only (U: RO, S: RW) */ #define PTE_BW 0x00000002 /* Supervisor (U: RW, S: RW) */ #define PTE_BR 0x00000003 /* Both Read Only (U: RO, S: RO) */ #define PTE_RW PTE_BW #define PTE_RO PTE_BR #define PTE_EXEC 0x00000200 /* pseudo bit in attrs; page is exec */ /* 64-bit PTE definitions */ /* High quadword: */ #define LPTE_VSID_SHIFT 12 #define LPTE_AVPN_MASK 0xFFFFFFFFFFFFFF80ULL #define LPTE_API 0x0000000000000F80ULL #define LPTE_SWBITS 0x0000000000000078ULL #define LPTE_WIRED 0x0000000000000010ULL #define LPTE_LOCKED 0x0000000000000008ULL #define LPTE_BIG 0x0000000000000004ULL /* 4kb/16Mb page */ #define LPTE_HID 0x0000000000000002ULL #define LPTE_VALID 0x0000000000000001ULL /* Low quadword: */ #define EXTEND_PTE(x) UINT64_C(x) /* make constants 64-bit */ #define LPTE_RPGN 0xfffffffffffff000ULL #define LPTE_REF EXTEND_PTE( PTE_REF ) #define LPTE_CHG EXTEND_PTE( PTE_CHG ) #define LPTE_WIMG EXTEND_PTE( PTE_WIMG ) #define LPTE_W EXTEND_PTE( PTE_W ) #define LPTE_I EXTEND_PTE( PTE_I ) #define LPTE_M EXTEND_PTE( PTE_M ) #define LPTE_G EXTEND_PTE( PTE_G ) #define LPTE_NOEXEC 0x0000000000000004ULL #define LPTE_PP EXTEND_PTE( PTE_PP ) #define LPTE_SO EXTEND_PTE( PTE_SO ) /* Super. Only */ #define LPTE_SW EXTEND_PTE( PTE_SW ) /* Super. Write-Only */ #define LPTE_BW EXTEND_PTE( PTE_BW ) /* Supervisor */ #define LPTE_BR EXTEND_PTE( PTE_BR ) /* Both Read Only */ #define LPTE_RW LPTE_BW #define LPTE_RO LPTE_BR #ifndef LOCORE typedef struct pte pte_t; typedef struct lpte lpte_t; #endif /* LOCORE */ /* * Extract bits from address */ #define ADDR_SR_SHFT 28 #define ADDR_PIDX 0x0ffff000UL #define ADDR_PIDX_SHFT 12 #define ADDR_API_SHFT 22 #define ADDR_API_SHFT64 16 #define ADDR_POFF 0x00000fffUL /* * Bits in DSISR: */ #define DSISR_DIRECT 0x80000000 #define DSISR_NOTFOUND 0x40000000 #define DSISR_PROTECT 0x08000000 #define DSISR_INVRX 0x04000000 #define DSISR_STORE 0x02000000 #define DSISR_DABR 0x00400000 #define DSISR_SEGMENT 0x00200000 #define DSISR_EAR 0x00100000 /* * Bits in SRR1 on ISI: */ #define ISSRR1_NOTFOUND 0x40000000 #define ISSRR1_DIRECT 0x10000000 #define ISSRR1_PROTECT 0x08000000 #define ISSRR1_SEGMENT 0x00200000 -#ifdef _KERNEL -#ifndef LOCORE -extern u_int dsisr(void); -#endif /* _KERNEL */ -#endif /* LOCORE */ - #else /* BOOKE */ #include /* * 1st level - page table directory (pdir) * * pdir consists of 1024 entries, each being a pointer to * second level entity, i.e. the actual page table (ptbl). */ #define PDIR_SHIFT 22 #define PDIR_SIZE (1 << PDIR_SHIFT) /* va range mapped by pdir */ #define PDIR_MASK (~(PDIR_SIZE - 1)) #define PDIR_NENTRIES 1024 /* number of page tables in pdir */ /* Returns pdir entry number for given va */ #define PDIR_IDX(va) ((va) >> PDIR_SHIFT) #define PDIR_ENTRY_SHIFT 2 /* entry size is 2^2 = 4 bytes */ /* * 2nd level - page table (ptbl) * * Page table covers 1024 page table entries. Page * table entry (pte) is 32 bit wide and defines mapping * for a single page. */ #define PTBL_SHIFT PAGE_SHIFT #define PTBL_SIZE PAGE_SIZE /* va range mapped by ptbl entry */ #define PTBL_MASK ((PDIR_SIZE - 1) & ~((1 << PAGE_SHIFT) - 1)) #define PTBL_NENTRIES 1024 /* number of pages mapped by ptbl */ /* Returns ptbl entry number for given va */ #define PTBL_IDX(va) (((va) & PTBL_MASK) >> PTBL_SHIFT) /* Size of ptbl in pages, 1024 entries, each sizeof(struct pte_entry). */ #define PTBL_PAGES 2 #define PTBL_ENTRY_SHIFT 3 /* entry size is 2^3 = 8 bytes */ /* * Flags for pte_remove() routine. */ #define PTBL_HOLD 0x00000001 /* do not unhold ptbl pages */ #define PTBL_UNHOLD 0x00000002 /* unhold and attempt to free ptbl pages */ #define PTBL_HOLD_FLAG(pmap) (((pmap) == kernel_pmap) ? PTBL_HOLD : PTBL_UNHOLD) /* * Page Table Entry definitions and macros. */ #ifndef LOCORE struct pte { vm_offset_t rpn; uint32_t flags; }; typedef struct pte pte_t; #endif /* RPN mask, TLB0 4K pages */ #define PTE_PA_MASK PAGE_MASK #if defined(BOOKE_E500) /* PTE bits assigned to MAS2, MAS3 flags */ #define PTE_W MAS2_W #define PTE_I MAS2_I #define PTE_M MAS2_M #define PTE_G MAS2_G #define PTE_MAS2_MASK (MAS2_G | MAS2_M | MAS2_I | MAS2_W) #define PTE_MAS3_SHIFT 8 #define PTE_UX (MAS3_UX << PTE_MAS3_SHIFT) #define PTE_SX (MAS3_SX << PTE_MAS3_SHIFT) #define PTE_UW (MAS3_UW << PTE_MAS3_SHIFT) #define PTE_SW (MAS3_SW << PTE_MAS3_SHIFT) #define PTE_UR (MAS3_UR << PTE_MAS3_SHIFT) #define PTE_SR (MAS3_SR << PTE_MAS3_SHIFT) #define PTE_MAS3_MASK ((MAS3_UX | MAS3_SX | MAS3_UW \ | MAS3_SW | MAS3_UR | MAS3_SR) << PTE_MAS3_SHIFT) #elif defined(BOOKE_PPC4XX) #define PTE_WL1 TLB_WL1 #define PTE_IL2I TLB_IL2I #define PTE_IL2D TLB_IL2D #define PTE_W TLB_W #define PTE_I TLB_I #define PTE_M TLB_M #define PTE_G TLB_G #define PTE_UX TLB_UX #define PTE_SX TLB_SX #define PTE_UW TLB_UW #define PTE_SW TLB_SW #define PTE_UR TLB_UR #define PTE_SR TLB_SR #endif /* Other PTE flags */ #define PTE_VALID 0x80000000 /* Valid */ #define PTE_MODIFIED 0x40000000 /* Modified */ #define PTE_WIRED 0x20000000 /* Wired */ #define PTE_MANAGED 0x10000000 /* Managed */ #define PTE_REFERENCED 0x04000000 /* Referenced */ /* Macro argument must of pte_t type. */ #define PTE_PA(pte) ((pte)->rpn & ~PTE_PA_MASK) #define PTE_ISVALID(pte) ((pte)->flags & PTE_VALID) #define PTE_ISWIRED(pte) ((pte)->flags & PTE_WIRED) #define PTE_ISMANAGED(pte) ((pte)->flags & PTE_MANAGED) #define PTE_ISMODIFIED(pte) ((pte)->flags & PTE_MODIFIED) #define PTE_ISREFERENCED(pte) ((pte)->flags & PTE_REFERENCED) #endif /* BOOKE_PPC4XX */ #endif /* _MACHINE_PTE_H_ */ Index: projects/cxl_iscsi/sys/powerpc/powerpc/db_trace.c =================================================================== --- projects/cxl_iscsi/sys/powerpc/powerpc/db_trace.c (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/powerpc/db_trace.c (revision 279717) @@ -1,313 +1,310 @@ /* $FreeBSD$ */ /* $NetBSD: db_trace.c,v 1.20 2002/05/13 20:30:09 matt Exp $ */ /* $OpenBSD: db_trace.c,v 1.3 1997/03/21 02:10:48 niklas Exp $ */ /*- * Mach Operating System * Copyright (c) 1992 Carnegie Mellon University * All Rights Reserved. * * Permission to use, copy, modify and distribute this software and its * documentation is hereby granted, provided that both the copyright * notice and this permission notice appear in all copies of the * software, derivative works or modified versions, and any portions * thereof, and that both notices appear in supporting documentation. * * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND FOR * ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. * * Carnegie Mellon requests users of this software to return to * * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU * School of Computer Science * Carnegie Mellon University * Pittsburgh PA 15213-3890 * * any improvements or extensions that they make and grant Carnegie Mellon * the rights to redistribute these changes. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include static db_varfcn_t db_frame; #define DB_OFFSET(x) (db_expr_t *)offsetof(struct trapframe, x) #ifdef __powerpc64__ #define CALLOFFSET 8 /* Include TOC reload slot */ #else #define CALLOFFSET 4 #endif struct db_variable db_regs[] = { { "r0", DB_OFFSET(fixreg[0]), db_frame }, { "r1", DB_OFFSET(fixreg[1]), db_frame }, { "r2", DB_OFFSET(fixreg[2]), db_frame }, { "r3", DB_OFFSET(fixreg[3]), db_frame }, { "r4", DB_OFFSET(fixreg[4]), db_frame }, { "r5", DB_OFFSET(fixreg[5]), db_frame }, { "r6", DB_OFFSET(fixreg[6]), db_frame }, { "r7", DB_OFFSET(fixreg[7]), db_frame }, { "r8", DB_OFFSET(fixreg[8]), db_frame }, { "r9", DB_OFFSET(fixreg[9]), db_frame }, { "r10", DB_OFFSET(fixreg[10]), db_frame }, { "r11", DB_OFFSET(fixreg[11]), db_frame }, { "r12", DB_OFFSET(fixreg[12]), db_frame }, { "r13", DB_OFFSET(fixreg[13]), db_frame }, { "r14", DB_OFFSET(fixreg[14]), db_frame }, { "r15", DB_OFFSET(fixreg[15]), db_frame }, { "r16", DB_OFFSET(fixreg[16]), db_frame }, { "r17", DB_OFFSET(fixreg[17]), db_frame }, { "r18", DB_OFFSET(fixreg[18]), db_frame }, { "r19", DB_OFFSET(fixreg[19]), db_frame }, { "r20", DB_OFFSET(fixreg[20]), db_frame }, { "r21", DB_OFFSET(fixreg[21]), db_frame }, { "r22", DB_OFFSET(fixreg[22]), db_frame }, { "r23", DB_OFFSET(fixreg[23]), db_frame }, { "r24", DB_OFFSET(fixreg[24]), db_frame }, { "r25", DB_OFFSET(fixreg[25]), db_frame }, { "r26", DB_OFFSET(fixreg[26]), db_frame }, { "r27", DB_OFFSET(fixreg[27]), db_frame }, { "r28", DB_OFFSET(fixreg[28]), db_frame }, { "r29", DB_OFFSET(fixreg[29]), db_frame }, { "r30", DB_OFFSET(fixreg[30]), db_frame }, { "r31", DB_OFFSET(fixreg[31]), db_frame }, { "srr0", DB_OFFSET(srr0), db_frame }, { "srr1", DB_OFFSET(srr1), db_frame }, { "lr", DB_OFFSET(lr), db_frame }, { "ctr", DB_OFFSET(ctr), db_frame }, { "cr", DB_OFFSET(cr), db_frame }, { "xer", DB_OFFSET(xer), db_frame }, + { "dar", DB_OFFSET(dar), db_frame }, #ifdef AIM - { "dar", DB_OFFSET(cpu.aim.dar), db_frame }, { "dsisr", DB_OFFSET(cpu.aim.dsisr), db_frame }, #endif #if defined(BOOKE) - { "dear", DB_OFFSET(cpu.booke.dear), db_frame }, { "esr", DB_OFFSET(cpu.booke.esr), db_frame }, #endif }; struct db_variable *db_eregs = db_regs + sizeof (db_regs)/sizeof (db_regs[0]); /* * register variable handling */ static int db_frame(struct db_variable *vp, db_expr_t *valuep, int op) { register_t *reg; if (kdb_frame == NULL) return (0); reg = (register_t*)((uintptr_t)kdb_frame + (uintptr_t)vp->valuep); if (op == DB_VAR_GET) *valuep = *reg; else *reg = *valuep; return (1); } /* * Frame tracing. */ static int db_backtrace(struct thread *td, db_addr_t fp, int count) { db_addr_t stackframe, lr, *args; boolean_t kernel_only = TRUE; boolean_t full = FALSE; #if 0 { register char *cp = modif; register char c; while ((c = *cp++) != 0) { if (c == 't') trace_thread = TRUE; if (c == 'u') kernel_only = FALSE; if (c == 'f') full = TRUE; } } #endif stackframe = fp; while (!db_pager_quit) { if (stackframe < PAGE_SIZE) break; /* * Locate the next frame by grabbing the backchain ptr * from frame[0] */ stackframe = *(db_addr_t *)stackframe; next_frame: #ifdef __powerpc64__ /* The saved arg values start at frame[6] */ args = (db_addr_t *)(stackframe + 48); #else /* The saved arg values start at frame[2] */ args = (db_addr_t *)(stackframe + 8); #endif if (stackframe < PAGE_SIZE) break; if (count-- == 0) break; /* * Extract link register from frame and subtract * 4 to convert into calling address (as opposed to * return address) */ #ifdef __powerpc64__ lr = *(db_addr_t *)(stackframe + 16) - 4; #else lr = *(db_addr_t *)(stackframe + 4) - 4; #endif if ((lr & 3) || (lr < 0x100)) { db_printf("saved LR(0x%zx) is invalid.", lr); break; } #ifdef __powerpc64__ db_printf("0x%016lx: ", stackframe); #else db_printf("0x%08x: ", stackframe); #endif /* * The trap code labels the return addresses from the * call to C code as 'trapexit' and 'asttrapexit. Use this * to determine if the callframe has to traverse a saved * trap context */ if ((lr + CALLOFFSET == (db_addr_t) &trapexit) || (lr + CALLOFFSET == (db_addr_t) &asttrapexit)) { const char *trapstr; struct trapframe *tf = (struct trapframe *)(args); db_printf("%s ", tf->srr1 & PSL_PR ? "user" : "kernel"); switch (tf->exc) { case EXC_DSI: /* XXX take advantage of the union. */ db_printf("DSI %s trap @ %#zx by ", (tf->cpu.aim.dsisr & DSISR_STORE) ? "write" - : "read", tf->cpu.aim.dar); + : "read", tf->dar); goto print_trap; case EXC_ALI: /* XXX take advantage of the union. */ db_printf("ALI trap @ %#zx (xSR %#x) ", - tf->cpu.aim.dar, - (uint32_t)tf->cpu.aim.dsisr); + tf->dar, (uint32_t)tf->cpu.aim.dsisr); goto print_trap; #ifdef __powerpc64__ case EXC_DSE: - db_printf("DSE trap @ %#zx by ", - tf->cpu.aim.dar); + db_printf("DSE trap @ %#zx by ", tf->dar); goto print_trap; case EXC_ISE: db_printf("ISE trap @ %#zx by ", tf->srr0); goto print_trap; #endif case EXC_ISI: trapstr = "ISI"; break; case EXC_PGM: trapstr = "PGM"; break; case EXC_SC: trapstr = "SC"; break; case EXC_EXI: trapstr = "EXI"; break; case EXC_MCHK: trapstr = "MCHK"; break; #if !defined(BOOKE) case EXC_VEC: trapstr = "VEC"; break; case EXC_FPA: trapstr = "FPA"; break; case EXC_BPT: trapstr = "BPT"; break; case EXC_TRC: trapstr = "TRC"; break; case EXC_RUNMODETRC: trapstr = "RUNMODETRC"; break; case EXC_SMI: trapstr = "SMI"; break; case EXC_RST: trapstr = "RST"; break; #endif case EXC_FPU: trapstr = "FPU"; break; case EXC_DECR: trapstr = "DECR"; break; case EXC_PERF: trapstr = "PERF"; break; case EXC_VSX: trapstr = "VSX"; break; default: trapstr = NULL; break; } if (trapstr != NULL) { db_printf("%s trap by ", trapstr); } else { db_printf("trap %#zx by ", tf->exc); } print_trap: lr = (db_addr_t) tf->srr0; db_printsym(lr, DB_STGY_ANY); db_printf(": srr1=%#zx\n", tf->srr1); db_printf("%-10s r1=%#zx cr=%#x xer=%#x ctr=%#zx", "", tf->fixreg[1], (uint32_t)tf->cr, (uint32_t)tf->xer, tf->ctr); #ifdef __powerpc64__ db_printf(" r2=%#zx", tf->fixreg[2]); #endif if (tf->exc == EXC_DSI) db_printf(" sr=%#x", (uint32_t)tf->cpu.aim.dsisr); db_printf("\n"); stackframe = (db_addr_t) tf->fixreg[1]; if (kernel_only && (tf->srr1 & PSL_PR)) break; goto next_frame; } db_printf("at "); db_printsym(lr, DB_STGY_PROC); if (full) /* Print all the args stored in that stackframe. */ db_printf("(%zx, %zx, %zx, %zx, %zx, %zx, %zx, %zx)", args[0], args[1], args[2], args[3], args[4], args[5], args[6], args[7]); db_printf("\n"); } return (0); } void db_trace_self(void) { db_addr_t addr; addr = (db_addr_t)__builtin_frame_address(1); db_backtrace(curthread, addr, -1); } int db_trace_thread(struct thread *td, int count) { struct pcb *ctx; ctx = kdb_thr_ctx(td); return (db_backtrace(td, (db_addr_t)ctx->pcb_sp, count)); } Index: projects/cxl_iscsi/sys/powerpc/powerpc/exec_machdep.c =================================================================== --- projects/cxl_iscsi/sys/powerpc/powerpc/exec_machdep.c (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/powerpc/exec_machdep.c (revision 279717) @@ -1,1091 +1,1081 @@ /*- * Copyright (C) 1995, 1996 Wolfgang Solfrank. * Copyright (C) 1995, 1996 TooLs GmbH. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by TooLs GmbH. * 4. The name of TooLs GmbH may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /*- * Copyright (C) 2001 Benno Rice * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY Benno Rice ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * $NetBSD: machdep.c,v 1.74.2.1 2000/11/01 16:13:48 tv Exp $ */ #include __FBSDID("$FreeBSD$"); #include "opt_compat.h" #include "opt_fpu_emu.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef FPU_EMU #include #endif #ifdef COMPAT_FREEBSD32 #include #include #include typedef struct __ucontext32 { sigset_t uc_sigmask; mcontext32_t uc_mcontext; uint32_t uc_link; struct sigaltstack32 uc_stack; uint32_t uc_flags; uint32_t __spare__[4]; } ucontext32_t; struct sigframe32 { ucontext32_t sf_uc; struct siginfo32 sf_si; }; static int grab_mcontext32(struct thread *td, mcontext32_t *, int flags); #endif static int grab_mcontext(struct thread *, mcontext_t *, int); void sendsig(sig_t catcher, ksiginfo_t *ksi, sigset_t *mask) { struct trapframe *tf; struct sigacts *psp; struct sigframe sf; struct thread *td; struct proc *p; #ifdef COMPAT_FREEBSD32 struct siginfo32 siginfo32; struct sigframe32 sf32; #endif size_t sfpsize; caddr_t sfp, usfp; int oonstack, rndfsize; int sig; int code; td = curthread; p = td->td_proc; PROC_LOCK_ASSERT(p, MA_OWNED); psp = p->p_sigacts; mtx_assert(&psp->ps_mtx, MA_OWNED); tf = td->td_frame; oonstack = sigonstack(tf->fixreg[1]); /* * Fill siginfo structure. */ ksi->ksi_info.si_signo = ksi->ksi_signo; - #ifdef AIM ksi->ksi_info.si_addr = (void *)((tf->exc == EXC_DSI) ? - tf->cpu.aim.dar : tf->srr0); - #else - ksi->ksi_info.si_addr = (void *)((tf->exc == EXC_DSI) ? - tf->cpu.booke.dear : tf->srr0); - #endif + tf->dar : tf->srr0); #ifdef COMPAT_FREEBSD32 if (SV_PROC_FLAG(p, SV_ILP32)) { siginfo_to_siginfo32(&ksi->ksi_info, &siginfo32); sig = siginfo32.si_signo; code = siginfo32.si_code; sfp = (caddr_t)&sf32; sfpsize = sizeof(sf32); rndfsize = ((sizeof(sf32) + 15) / 16) * 16; /* * Save user context */ memset(&sf32, 0, sizeof(sf32)); grab_mcontext32(td, &sf32.sf_uc.uc_mcontext, 0); sf32.sf_uc.uc_sigmask = *mask; sf32.sf_uc.uc_stack.ss_sp = (uintptr_t)td->td_sigstk.ss_sp; sf32.sf_uc.uc_stack.ss_size = (uint32_t)td->td_sigstk.ss_size; sf32.sf_uc.uc_stack.ss_flags = (td->td_pflags & TDP_ALTSTACK) ? ((oonstack) ? SS_ONSTACK : 0) : SS_DISABLE; sf32.sf_uc.uc_mcontext.mc_onstack = (oonstack) ? 1 : 0; } else { #endif sig = ksi->ksi_signo; code = ksi->ksi_code; sfp = (caddr_t)&sf; sfpsize = sizeof(sf); #ifdef __powerpc64__ /* * 64-bit PPC defines a 288 byte scratch region * below the stack. */ rndfsize = 288 + ((sizeof(sf) + 47) / 48) * 48; #else rndfsize = ((sizeof(sf) + 15) / 16) * 16; #endif /* * Save user context */ memset(&sf, 0, sizeof(sf)); grab_mcontext(td, &sf.sf_uc.uc_mcontext, 0); sf.sf_uc.uc_sigmask = *mask; sf.sf_uc.uc_stack = td->td_sigstk; sf.sf_uc.uc_stack.ss_flags = (td->td_pflags & TDP_ALTSTACK) ? ((oonstack) ? SS_ONSTACK : 0) : SS_DISABLE; sf.sf_uc.uc_mcontext.mc_onstack = (oonstack) ? 1 : 0; #ifdef COMPAT_FREEBSD32 } #endif CTR4(KTR_SIG, "sendsig: td=%p (%s) catcher=%p sig=%d", td, p->p_comm, catcher, sig); /* * Allocate and validate space for the signal handler context. */ if ((td->td_pflags & TDP_ALTSTACK) != 0 && !oonstack && SIGISMEMBER(psp->ps_sigonstack, sig)) { usfp = (void *)(td->td_sigstk.ss_sp + td->td_sigstk.ss_size - rndfsize); } else { usfp = (void *)(tf->fixreg[1] - rndfsize); } /* * Translate the signal if appropriate (Linux emu ?) */ if (p->p_sysent->sv_sigtbl && sig <= p->p_sysent->sv_sigsize) sig = p->p_sysent->sv_sigtbl[_SIG_IDX(sig)]; /* * Save the floating-point state, if necessary, then copy it. */ /* XXX */ /* * Set up the registers to return to sigcode. * * r1/sp - sigframe ptr * lr - sig function, dispatched to by blrl in trampoline * r3 - sig number * r4 - SIGINFO ? &siginfo : exception code * r5 - user context * srr0 - trampoline function addr */ tf->lr = (register_t)catcher; tf->fixreg[1] = (register_t)usfp; tf->fixreg[FIRSTARG] = sig; #ifdef COMPAT_FREEBSD32 tf->fixreg[FIRSTARG+2] = (register_t)usfp + ((SV_PROC_FLAG(p, SV_ILP32)) ? offsetof(struct sigframe32, sf_uc) : offsetof(struct sigframe, sf_uc)); #else tf->fixreg[FIRSTARG+2] = (register_t)usfp + offsetof(struct sigframe, sf_uc); #endif if (SIGISMEMBER(psp->ps_siginfo, sig)) { /* * Signal handler installed with SA_SIGINFO. */ #ifdef COMPAT_FREEBSD32 if (SV_PROC_FLAG(p, SV_ILP32)) { sf32.sf_si = siginfo32; tf->fixreg[FIRSTARG+1] = (register_t)usfp + offsetof(struct sigframe32, sf_si); sf32.sf_si = siginfo32; } else { #endif tf->fixreg[FIRSTARG+1] = (register_t)usfp + offsetof(struct sigframe, sf_si); sf.sf_si = ksi->ksi_info; #ifdef COMPAT_FREEBSD32 } #endif } else { /* Old FreeBSD-style arguments. */ tf->fixreg[FIRSTARG+1] = code; - #ifdef AIM tf->fixreg[FIRSTARG+3] = (tf->exc == EXC_DSI) ? - tf->cpu.aim.dar : tf->srr0; - #else - tf->fixreg[FIRSTARG+3] = (tf->exc == EXC_DSI) ? - tf->cpu.booke.dear : tf->srr0; - #endif + tf->dar : tf->srr0; } mtx_unlock(&psp->ps_mtx); PROC_UNLOCK(p); tf->srr0 = (register_t)p->p_sysent->sv_sigcode_base; /* * copy the frame out to userland. */ if (copyout(sfp, usfp, sfpsize) != 0) { /* * Process has trashed its stack. Kill it. */ CTR2(KTR_SIG, "sendsig: sigexit td=%p sfp=%p", td, sfp); PROC_LOCK(p); sigexit(td, SIGILL); } CTR3(KTR_SIG, "sendsig: return td=%p pc=%#x sp=%#x", td, tf->srr0, tf->fixreg[1]); PROC_LOCK(p); mtx_lock(&psp->ps_mtx); } int sys_sigreturn(struct thread *td, struct sigreturn_args *uap) { ucontext_t uc; int error; CTR2(KTR_SIG, "sigreturn: td=%p ucp=%p", td, uap->sigcntxp); if (copyin(uap->sigcntxp, &uc, sizeof(uc)) != 0) { CTR1(KTR_SIG, "sigreturn: efault td=%p", td); return (EFAULT); } error = set_mcontext(td, &uc.uc_mcontext); if (error != 0) return (error); kern_sigprocmask(td, SIG_SETMASK, &uc.uc_sigmask, NULL, 0); CTR3(KTR_SIG, "sigreturn: return td=%p pc=%#x sp=%#x", td, uc.uc_mcontext.mc_srr0, uc.uc_mcontext.mc_gpr[1]); return (EJUSTRETURN); } #ifdef COMPAT_FREEBSD4 int freebsd4_sigreturn(struct thread *td, struct freebsd4_sigreturn_args *uap) { return sys_sigreturn(td, (struct sigreturn_args *)uap); } #endif /* * Construct a PCB from a trapframe. This is called from kdb_trap() where * we want to start a backtrace from the function that caused us to enter * the debugger. We have the context in the trapframe, but base the trace * on the PCB. The PCB doesn't have to be perfect, as long as it contains * enough for a backtrace. */ void makectx(struct trapframe *tf, struct pcb *pcb) { pcb->pcb_lr = tf->srr0; pcb->pcb_sp = tf->fixreg[1]; } /* * get_mcontext/sendsig helper routine that doesn't touch the * proc lock */ static int grab_mcontext(struct thread *td, mcontext_t *mcp, int flags) { struct pcb *pcb; int i; pcb = td->td_pcb; memset(mcp, 0, sizeof(mcontext_t)); mcp->mc_vers = _MC_VERSION; mcp->mc_flags = 0; memcpy(&mcp->mc_frame, td->td_frame, sizeof(struct trapframe)); if (flags & GET_MC_CLEAR_RET) { mcp->mc_gpr[3] = 0; mcp->mc_gpr[4] = 0; } /* * This assumes that floating-point context is *not* lazy, * so if the thread has used FP there would have been a * FP-unavailable exception that would have set things up * correctly. */ if (pcb->pcb_flags & PCB_FPREGS) { if (pcb->pcb_flags & PCB_FPU) { KASSERT(td == curthread, ("get_mcontext: fp save not curthread")); critical_enter(); save_fpu(td); critical_exit(); } mcp->mc_flags |= _MC_FP_VALID; memcpy(&mcp->mc_fpscr, &pcb->pcb_fpu.fpscr, sizeof(double)); memcpy(mcp->mc_fpreg, pcb->pcb_fpu.fpr, 32*sizeof(double)); for (i = 0; i < 32; i++) memcpy(&mcp->mc_fpreg[i], &pcb->pcb_fpu.fpr[i].fpr, sizeof(double)); } /* * Repeat for Altivec context */ if (pcb->pcb_flags & PCB_VEC) { KASSERT(td == curthread, ("get_mcontext: fp save not curthread")); critical_enter(); save_vec(td); critical_exit(); mcp->mc_flags |= _MC_AV_VALID; mcp->mc_vscr = pcb->pcb_vec.vscr; mcp->mc_vrsave = pcb->pcb_vec.vrsave; memcpy(mcp->mc_avec, pcb->pcb_vec.vr, sizeof(mcp->mc_avec)); } /* XXX VSX context */ mcp->mc_len = sizeof(*mcp); return (0); } int get_mcontext(struct thread *td, mcontext_t *mcp, int flags) { int error; error = grab_mcontext(td, mcp, flags); if (error == 0) { PROC_LOCK(curthread->td_proc); mcp->mc_onstack = sigonstack(td->td_frame->fixreg[1]); PROC_UNLOCK(curthread->td_proc); } return (error); } int set_mcontext(struct thread *td, mcontext_t *mcp) { struct pcb *pcb; struct trapframe *tf; register_t tls; int i; pcb = td->td_pcb; tf = td->td_frame; if (mcp->mc_vers != _MC_VERSION || mcp->mc_len != sizeof(*mcp)) return (EINVAL); /* * Don't let the user set privileged MSR bits */ if ((mcp->mc_srr1 & PSL_USERSTATIC) != (tf->srr1 & PSL_USERSTATIC)) { return (EINVAL); } /* Copy trapframe, preserving TLS pointer across context change */ if (SV_PROC_FLAG(td->td_proc, SV_LP64)) tls = tf->fixreg[13]; else tls = tf->fixreg[2]; memcpy(tf, mcp->mc_frame, sizeof(mcp->mc_frame)); if (SV_PROC_FLAG(td->td_proc, SV_LP64)) tf->fixreg[13] = tls; else tf->fixreg[2] = tls; if (mcp->mc_flags & _MC_FP_VALID) { /* enable_fpu() will happen lazily on a fault */ pcb->pcb_flags |= PCB_FPREGS; memcpy(&pcb->pcb_fpu.fpscr, &mcp->mc_fpscr, sizeof(double)); bzero(pcb->pcb_fpu.fpr, sizeof(pcb->pcb_fpu.fpr)); for (i = 0; i < 32; i++) memcpy(&pcb->pcb_fpu.fpr[i].fpr, &mcp->mc_fpreg[i], sizeof(double)); } if (mcp->mc_flags & _MC_AV_VALID) { if ((pcb->pcb_flags & PCB_VEC) != PCB_VEC) { critical_enter(); enable_vec(td); critical_exit(); } pcb->pcb_vec.vscr = mcp->mc_vscr; pcb->pcb_vec.vrsave = mcp->mc_vrsave; memcpy(pcb->pcb_vec.vr, mcp->mc_avec, sizeof(mcp->mc_avec)); } /* XXX VSX context */ return (0); } /* * Set set up registers on exec. */ void exec_setregs(struct thread *td, struct image_params *imgp, u_long stack) { struct trapframe *tf; register_t argc; #ifdef __powerpc64__ register_t entry_desc[3]; #endif tf = trapframe(td); bzero(tf, sizeof *tf); #ifdef __powerpc64__ tf->fixreg[1] = -roundup(-stack + 48, 16); #else tf->fixreg[1] = -roundup(-stack + 8, 16); #endif /* * Set up arguments for _start(): * _start(argc, argv, envp, obj, cleanup, ps_strings); * * Notes: * - obj and cleanup are the auxilliary and termination * vectors. They are fixed up by ld.elf_so. * - ps_strings is a NetBSD extention, and will be * ignored by executables which are strictly * compliant with the SVR4 ABI. * * XXX We have to set both regs and retval here due to different * XXX calling convention in trap.c and init_main.c. */ /* Collect argc from the user stack */ argc = fuword((void *)stack); /* * XXX PG: these get overwritten in the syscall return code. * execve() should return EJUSTRETURN, like it does on NetBSD. * Emulate by setting the syscall return value cells. The * registers still have to be set for init's fork trampoline. */ td->td_retval[0] = argc; td->td_retval[1] = stack + sizeof(register_t); tf->fixreg[3] = argc; tf->fixreg[4] = stack + sizeof(register_t); tf->fixreg[5] = stack + (2 + argc)*sizeof(register_t); tf->fixreg[6] = 0; /* auxillary vector */ tf->fixreg[7] = 0; /* termination vector */ tf->fixreg[8] = (register_t)imgp->ps_strings; /* NetBSD extension */ #ifdef __powerpc64__ /* * For 64-bit, we need to disentangle the function descriptor * * 0. entry point * 1. TOC value (r2) * 2. Environment pointer (r11) */ (void)copyin((void *)imgp->entry_addr, entry_desc, sizeof(entry_desc)); tf->srr0 = entry_desc[0] + imgp->reloc_base; tf->fixreg[2] = entry_desc[1] + imgp->reloc_base; tf->fixreg[11] = entry_desc[2] + imgp->reloc_base; tf->srr1 = PSL_SF | PSL_USERSET | PSL_FE_DFLT; if (mfmsr() & PSL_HV) tf->srr1 |= PSL_HV; #else tf->srr0 = imgp->entry_addr; tf->srr1 = PSL_USERSET | PSL_FE_DFLT; #endif td->td_pcb->pcb_flags = 0; } #ifdef COMPAT_FREEBSD32 void ppc32_setregs(struct thread *td, struct image_params *imgp, u_long stack) { struct trapframe *tf; uint32_t argc; tf = trapframe(td); bzero(tf, sizeof *tf); tf->fixreg[1] = -roundup(-stack + 8, 16); argc = fuword32((void *)stack); td->td_retval[0] = argc; td->td_retval[1] = stack + sizeof(uint32_t); tf->fixreg[3] = argc; tf->fixreg[4] = stack + sizeof(uint32_t); tf->fixreg[5] = stack + (2 + argc)*sizeof(uint32_t); tf->fixreg[6] = 0; /* auxillary vector */ tf->fixreg[7] = 0; /* termination vector */ tf->fixreg[8] = (register_t)imgp->ps_strings; /* NetBSD extension */ tf->srr0 = imgp->entry_addr; tf->srr1 = PSL_USERSET | PSL_FE_DFLT; tf->srr1 &= ~PSL_SF; if (mfmsr() & PSL_HV) tf->srr1 |= PSL_HV; td->td_pcb->pcb_flags = 0; } #endif int fill_regs(struct thread *td, struct reg *regs) { struct trapframe *tf; tf = td->td_frame; memcpy(regs, tf, sizeof(struct reg)); return (0); } int fill_dbregs(struct thread *td, struct dbreg *dbregs) { /* No debug registers on PowerPC */ return (ENOSYS); } int fill_fpregs(struct thread *td, struct fpreg *fpregs) { struct pcb *pcb; pcb = td->td_pcb; if ((pcb->pcb_flags & PCB_FPREGS) == 0) memset(fpregs, 0, sizeof(struct fpreg)); else memcpy(fpregs, &pcb->pcb_fpu, sizeof(struct fpreg)); return (0); } int set_regs(struct thread *td, struct reg *regs) { struct trapframe *tf; tf = td->td_frame; memcpy(tf, regs, sizeof(struct reg)); return (0); } int set_dbregs(struct thread *td, struct dbreg *dbregs) { /* No debug registers on PowerPC */ return (ENOSYS); } int set_fpregs(struct thread *td, struct fpreg *fpregs) { struct pcb *pcb; pcb = td->td_pcb; pcb->pcb_flags |= PCB_FPREGS; memcpy(&pcb->pcb_fpu, fpregs, sizeof(struct fpreg)); return (0); } #ifdef COMPAT_FREEBSD32 int set_regs32(struct thread *td, struct reg32 *regs) { struct trapframe *tf; int i; tf = td->td_frame; for (i = 0; i < 32; i++) tf->fixreg[i] = regs->fixreg[i]; tf->lr = regs->lr; tf->cr = regs->cr; tf->xer = regs->xer; tf->ctr = regs->ctr; tf->srr0 = regs->pc; return (0); } int fill_regs32(struct thread *td, struct reg32 *regs) { struct trapframe *tf; int i; tf = td->td_frame; for (i = 0; i < 32; i++) regs->fixreg[i] = tf->fixreg[i]; regs->lr = tf->lr; regs->cr = tf->cr; regs->xer = tf->xer; regs->ctr = tf->ctr; regs->pc = tf->srr0; return (0); } static int grab_mcontext32(struct thread *td, mcontext32_t *mcp, int flags) { mcontext_t mcp64; int i, error; error = grab_mcontext(td, &mcp64, flags); if (error != 0) return (error); mcp->mc_vers = mcp64.mc_vers; mcp->mc_flags = mcp64.mc_flags; mcp->mc_onstack = mcp64.mc_onstack; mcp->mc_len = mcp64.mc_len; memcpy(mcp->mc_avec,mcp64.mc_avec,sizeof(mcp64.mc_avec)); memcpy(mcp->mc_av,mcp64.mc_av,sizeof(mcp64.mc_av)); for (i = 0; i < 42; i++) mcp->mc_frame[i] = mcp64.mc_frame[i]; memcpy(mcp->mc_fpreg,mcp64.mc_fpreg,sizeof(mcp64.mc_fpreg)); return (0); } static int get_mcontext32(struct thread *td, mcontext32_t *mcp, int flags) { int error; error = grab_mcontext32(td, mcp, flags); if (error == 0) { PROC_LOCK(curthread->td_proc); mcp->mc_onstack = sigonstack(td->td_frame->fixreg[1]); PROC_UNLOCK(curthread->td_proc); } return (error); } static int set_mcontext32(struct thread *td, mcontext32_t *mcp) { mcontext_t mcp64; int i, error; mcp64.mc_vers = mcp->mc_vers; mcp64.mc_flags = mcp->mc_flags; mcp64.mc_onstack = mcp->mc_onstack; mcp64.mc_len = mcp->mc_len; memcpy(mcp64.mc_avec,mcp->mc_avec,sizeof(mcp64.mc_avec)); memcpy(mcp64.mc_av,mcp->mc_av,sizeof(mcp64.mc_av)); for (i = 0; i < 42; i++) mcp64.mc_frame[i] = mcp->mc_frame[i]; mcp64.mc_srr1 |= (td->td_frame->srr1 & 0xFFFFFFFF00000000ULL); memcpy(mcp64.mc_fpreg,mcp->mc_fpreg,sizeof(mcp64.mc_fpreg)); error = set_mcontext(td, &mcp64); return (error); } #endif #ifdef COMPAT_FREEBSD32 int freebsd32_sigreturn(struct thread *td, struct freebsd32_sigreturn_args *uap) { ucontext32_t uc; int error; CTR2(KTR_SIG, "sigreturn: td=%p ucp=%p", td, uap->sigcntxp); if (copyin(uap->sigcntxp, &uc, sizeof(uc)) != 0) { CTR1(KTR_SIG, "sigreturn: efault td=%p", td); return (EFAULT); } error = set_mcontext32(td, &uc.uc_mcontext); if (error != 0) return (error); kern_sigprocmask(td, SIG_SETMASK, &uc.uc_sigmask, NULL, 0); CTR3(KTR_SIG, "sigreturn: return td=%p pc=%#x sp=%#x", td, uc.uc_mcontext.mc_srr0, uc.uc_mcontext.mc_gpr[1]); return (EJUSTRETURN); } /* * The first two fields of a ucontext_t are the signal mask and the machine * context. The next field is uc_link; we want to avoid destroying the link * when copying out contexts. */ #define UC32_COPY_SIZE offsetof(ucontext32_t, uc_link) int freebsd32_getcontext(struct thread *td, struct freebsd32_getcontext_args *uap) { ucontext32_t uc; int ret; if (uap->ucp == NULL) ret = EINVAL; else { get_mcontext32(td, &uc.uc_mcontext, GET_MC_CLEAR_RET); PROC_LOCK(td->td_proc); uc.uc_sigmask = td->td_sigmask; PROC_UNLOCK(td->td_proc); ret = copyout(&uc, uap->ucp, UC32_COPY_SIZE); } return (ret); } int freebsd32_setcontext(struct thread *td, struct freebsd32_setcontext_args *uap) { ucontext32_t uc; int ret; if (uap->ucp == NULL) ret = EINVAL; else { ret = copyin(uap->ucp, &uc, UC32_COPY_SIZE); if (ret == 0) { ret = set_mcontext32(td, &uc.uc_mcontext); if (ret == 0) { kern_sigprocmask(td, SIG_SETMASK, &uc.uc_sigmask, NULL, 0); } } } return (ret == 0 ? EJUSTRETURN : ret); } int freebsd32_swapcontext(struct thread *td, struct freebsd32_swapcontext_args *uap) { ucontext32_t uc; int ret; if (uap->oucp == NULL || uap->ucp == NULL) ret = EINVAL; else { get_mcontext32(td, &uc.uc_mcontext, GET_MC_CLEAR_RET); PROC_LOCK(td->td_proc); uc.uc_sigmask = td->td_sigmask; PROC_UNLOCK(td->td_proc); ret = copyout(&uc, uap->oucp, UC32_COPY_SIZE); if (ret == 0) { ret = copyin(uap->ucp, &uc, UC32_COPY_SIZE); if (ret == 0) { ret = set_mcontext32(td, &uc.uc_mcontext); if (ret == 0) { kern_sigprocmask(td, SIG_SETMASK, &uc.uc_sigmask, NULL, 0); } } } } return (ret == 0 ? EJUSTRETURN : ret); } #endif void cpu_set_syscall_retval(struct thread *td, int error) { struct proc *p; struct trapframe *tf; int fixup; if (error == EJUSTRETURN) return; p = td->td_proc; tf = td->td_frame; if (tf->fixreg[0] == SYS___syscall && (SV_PROC_FLAG(p, SV_ILP32))) { int code = tf->fixreg[FIRSTARG + 1]; if (p->p_sysent->sv_mask) code &= p->p_sysent->sv_mask; fixup = (code != SYS_freebsd6_lseek && code != SYS_lseek) ? 1 : 0; } else fixup = 0; switch (error) { case 0: if (fixup) { /* * 64-bit return, 32-bit syscall. Fixup byte order */ tf->fixreg[FIRSTARG] = 0; tf->fixreg[FIRSTARG + 1] = td->td_retval[0]; } else { tf->fixreg[FIRSTARG] = td->td_retval[0]; tf->fixreg[FIRSTARG + 1] = td->td_retval[1]; } tf->cr &= ~0x10000000; /* Unset summary overflow */ break; case ERESTART: /* * Set user's pc back to redo the system call. */ tf->srr0 -= 4; break; default: if (p->p_sysent->sv_errsize) { error = (error < p->p_sysent->sv_errsize) ? p->p_sysent->sv_errtbl[error] : -1; } tf->fixreg[FIRSTARG] = error; tf->cr |= 0x10000000; /* Set summary overflow */ break; } } /* * Threading functions */ void cpu_thread_exit(struct thread *td) { } void cpu_thread_clean(struct thread *td) { } void cpu_thread_alloc(struct thread *td) { struct pcb *pcb; pcb = (struct pcb *)((td->td_kstack + td->td_kstack_pages * PAGE_SIZE - sizeof(struct pcb)) & ~0x2fUL); td->td_pcb = pcb; td->td_frame = (struct trapframe *)pcb - 1; } void cpu_thread_free(struct thread *td) { } int cpu_set_user_tls(struct thread *td, void *tls_base) { if (SV_PROC_FLAG(td->td_proc, SV_LP64)) td->td_frame->fixreg[13] = (register_t)tls_base + 0x7010; else td->td_frame->fixreg[2] = (register_t)tls_base + 0x7008; return (0); } void cpu_set_upcall(struct thread *td, struct thread *td0) { struct pcb *pcb2; struct trapframe *tf; struct callframe *cf; pcb2 = td->td_pcb; /* Copy the upcall pcb */ bcopy(td0->td_pcb, pcb2, sizeof(*pcb2)); /* Create a stack for the new thread */ tf = td->td_frame; bcopy(td0->td_frame, tf, sizeof(struct trapframe)); tf->fixreg[FIRSTARG] = 0; tf->fixreg[FIRSTARG + 1] = 0; tf->cr &= ~0x10000000; /* Set registers for trampoline to user mode. */ cf = (struct callframe *)tf - 1; memset(cf, 0, sizeof(struct callframe)); cf->cf_func = (register_t)fork_return; cf->cf_arg0 = (register_t)td; cf->cf_arg1 = (register_t)tf; pcb2->pcb_sp = (register_t)cf; #ifdef __powerpc64__ pcb2->pcb_lr = ((register_t *)fork_trampoline)[0]; pcb2->pcb_toc = ((register_t *)fork_trampoline)[1]; #else pcb2->pcb_lr = (register_t)fork_trampoline; #endif pcb2->pcb_cpu.aim.usr_vsid = 0; /* Setup to release spin count in fork_exit(). */ td->td_md.md_spinlock_count = 1; td->td_md.md_saved_msr = PSL_KERNSET; } void cpu_set_upcall_kse(struct thread *td, void (*entry)(void *), void *arg, stack_t *stack) { struct trapframe *tf; uintptr_t sp; tf = td->td_frame; /* align stack and alloc space for frame ptr and saved LR */ #ifdef __powerpc64__ sp = ((uintptr_t)stack->ss_sp + stack->ss_size - 48) & ~0x1f; #else sp = ((uintptr_t)stack->ss_sp + stack->ss_size - 8) & ~0x1f; #endif bzero(tf, sizeof(struct trapframe)); tf->fixreg[1] = (register_t)sp; tf->fixreg[3] = (register_t)arg; if (SV_PROC_FLAG(td->td_proc, SV_ILP32)) { tf->srr0 = (register_t)entry; tf->srr1 = PSL_USERSET | PSL_FE_DFLT; #ifdef __powerpc64__ tf->srr1 &= ~PSL_SF; #endif } else { #ifdef __powerpc64__ register_t entry_desc[3]; (void)copyin((void *)entry, entry_desc, sizeof(entry_desc)); tf->srr0 = entry_desc[0]; tf->fixreg[2] = entry_desc[1]; tf->fixreg[11] = entry_desc[2]; tf->srr1 = PSL_SF | PSL_USERSET | PSL_FE_DFLT; #endif } #ifdef __powerpc64__ if (mfmsr() & PSL_HV) tf->srr1 |= PSL_HV; #endif td->td_pcb->pcb_flags = 0; td->td_retval[0] = (register_t)entry; td->td_retval[1] = 0; } int ppc_instr_emulate(struct trapframe *frame, struct pcb *pcb) { uint32_t instr; int reg, sig; instr = fuword32((void *)frame->srr0); sig = SIGILL; if ((instr & 0xfc1fffff) == 0x7c1f42a6) { /* mfpvr */ reg = (instr & ~0xfc1fffff) >> 21; frame->fixreg[reg] = mfpvr(); frame->srr0 += 4; return (0); } if ((instr & 0xfc000ffe) == 0x7c0004ac) { /* various sync */ powerpc_sync(); /* Do a heavy-weight sync */ frame->srr0 += 4; return (0); } #ifdef FPU_EMU if (!(pcb->pcb_flags & PCB_FPREGS)) { bzero(&pcb->pcb_fpu, sizeof(pcb->pcb_fpu)); pcb->pcb_flags |= PCB_FPREGS; } sig = fpu_emulate(frame, (struct fpreg *)&pcb->pcb_fpu); #endif return (sig); } Index: projects/cxl_iscsi/sys/powerpc/powerpc/genassym.c =================================================================== --- projects/cxl_iscsi/sys/powerpc/powerpc/genassym.c (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/powerpc/genassym.c (revision 279717) @@ -1,259 +1,259 @@ /*- * Copyright (c) 1982, 1990 The Regents of the University of California. * All rights reserved. * * This code is derived from software contributed to Berkeley by * William Jolitz. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)genassym.c 5.11 (Berkeley) 5/10/91 * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include ASSYM(PC_CURTHREAD, offsetof(struct pcpu, pc_curthread)); ASSYM(PC_CURPCB, offsetof(struct pcpu, pc_curpcb)); ASSYM(PC_CURPMAP, offsetof(struct pcpu, pc_curpmap)); ASSYM(PC_TEMPSAVE, offsetof(struct pcpu, pc_tempsave)); ASSYM(PC_DISISAVE, offsetof(struct pcpu, pc_disisave)); ASSYM(PC_DBSAVE, offsetof(struct pcpu, pc_dbsave)); ASSYM(PC_RESTORE, offsetof(struct pcpu, pc_restore)); #if defined(BOOKE) ASSYM(PC_BOOKE_CRITSAVE, offsetof(struct pcpu, pc_booke_critsave)); ASSYM(PC_BOOKE_MCHKSAVE, offsetof(struct pcpu, pc_booke_mchksave)); ASSYM(PC_BOOKE_TLBSAVE, offsetof(struct pcpu, pc_booke_tlbsave)); ASSYM(PC_BOOKE_TLB_LEVEL, offsetof(struct pcpu, pc_booke_tlb_level)); ASSYM(PC_BOOKE_TLB_LOCK, offsetof(struct pcpu, pc_booke_tlb_lock)); #endif ASSYM(CPUSAVE_R27, CPUSAVE_R27*sizeof(register_t)); ASSYM(CPUSAVE_R28, CPUSAVE_R28*sizeof(register_t)); ASSYM(CPUSAVE_R29, CPUSAVE_R29*sizeof(register_t)); ASSYM(CPUSAVE_R30, CPUSAVE_R30*sizeof(register_t)); ASSYM(CPUSAVE_R31, CPUSAVE_R31*sizeof(register_t)); ASSYM(CPUSAVE_SRR0, CPUSAVE_SRR0*sizeof(register_t)); ASSYM(CPUSAVE_SRR1, CPUSAVE_SRR1*sizeof(register_t)); ASSYM(CPUSAVE_AIM_DAR, CPUSAVE_AIM_DAR*sizeof(register_t)); ASSYM(CPUSAVE_AIM_DSISR, CPUSAVE_AIM_DSISR*sizeof(register_t)); ASSYM(CPUSAVE_BOOKE_DEAR, CPUSAVE_BOOKE_DEAR*sizeof(register_t)); ASSYM(CPUSAVE_BOOKE_ESR, CPUSAVE_BOOKE_ESR*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_LR, TLBSAVE_BOOKE_LR*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_CR, TLBSAVE_BOOKE_CR*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_SRR0, TLBSAVE_BOOKE_SRR0*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_SRR1, TLBSAVE_BOOKE_SRR1*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_R20, TLBSAVE_BOOKE_R20*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_R21, TLBSAVE_BOOKE_R21*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_R22, TLBSAVE_BOOKE_R22*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_R23, TLBSAVE_BOOKE_R23*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_R24, TLBSAVE_BOOKE_R24*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_R25, TLBSAVE_BOOKE_R25*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_R26, TLBSAVE_BOOKE_R26*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_R27, TLBSAVE_BOOKE_R27*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_R28, TLBSAVE_BOOKE_R28*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_R29, TLBSAVE_BOOKE_R29*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_R30, TLBSAVE_BOOKE_R30*sizeof(register_t)); ASSYM(TLBSAVE_BOOKE_R31, TLBSAVE_BOOKE_R31*sizeof(register_t)); ASSYM(MTX_LOCK, offsetof(struct mtx, mtx_lock)); #if defined(AIM) ASSYM(USER_ADDR, USER_ADDR); #ifdef __powerpc64__ ASSYM(PC_KERNSLB, offsetof(struct pcpu, pc_slb)); ASSYM(PC_USERSLB, offsetof(struct pcpu, pc_userslb)); ASSYM(PC_SLBSAVE, offsetof(struct pcpu, pc_slbsave)); ASSYM(PC_SLBSTACK, offsetof(struct pcpu, pc_slbstack)); ASSYM(USER_SLB_SLOT, USER_SLB_SLOT); ASSYM(USER_SLB_SLBE, USER_SLB_SLBE); ASSYM(SEGMENT_MASK, SEGMENT_MASK); #else ASSYM(PM_SR, offsetof(struct pmap, pm_sr)); ASSYM(USER_SR, USER_SR); #endif #elif defined(BOOKE) ASSYM(PM_PDIR, offsetof(struct pmap, pm_pdir)); ASSYM(PTE_RPN, offsetof(struct pte, rpn)); ASSYM(PTE_FLAGS, offsetof(struct pte, flags)); #if defined(BOOKE_E500) ASSYM(TLB0_ENTRY_SIZE, sizeof(struct tlb_entry)); #endif #endif #ifdef __powerpc64__ ASSYM(FSP, 48); #else ASSYM(FSP, 8); #endif ASSYM(FRAMELEN, FRAMELEN); ASSYM(FRAME_0, offsetof(struct trapframe, fixreg[0])); ASSYM(FRAME_1, offsetof(struct trapframe, fixreg[1])); ASSYM(FRAME_2, offsetof(struct trapframe, fixreg[2])); ASSYM(FRAME_3, offsetof(struct trapframe, fixreg[3])); ASSYM(FRAME_4, offsetof(struct trapframe, fixreg[4])); ASSYM(FRAME_5, offsetof(struct trapframe, fixreg[5])); ASSYM(FRAME_6, offsetof(struct trapframe, fixreg[6])); ASSYM(FRAME_7, offsetof(struct trapframe, fixreg[7])); ASSYM(FRAME_8, offsetof(struct trapframe, fixreg[8])); ASSYM(FRAME_9, offsetof(struct trapframe, fixreg[9])); ASSYM(FRAME_10, offsetof(struct trapframe, fixreg[10])); ASSYM(FRAME_11, offsetof(struct trapframe, fixreg[11])); ASSYM(FRAME_12, offsetof(struct trapframe, fixreg[12])); ASSYM(FRAME_13, offsetof(struct trapframe, fixreg[13])); ASSYM(FRAME_14, offsetof(struct trapframe, fixreg[14])); ASSYM(FRAME_15, offsetof(struct trapframe, fixreg[15])); ASSYM(FRAME_16, offsetof(struct trapframe, fixreg[16])); ASSYM(FRAME_17, offsetof(struct trapframe, fixreg[17])); ASSYM(FRAME_18, offsetof(struct trapframe, fixreg[18])); ASSYM(FRAME_19, offsetof(struct trapframe, fixreg[19])); ASSYM(FRAME_20, offsetof(struct trapframe, fixreg[20])); ASSYM(FRAME_21, offsetof(struct trapframe, fixreg[21])); ASSYM(FRAME_22, offsetof(struct trapframe, fixreg[22])); ASSYM(FRAME_23, offsetof(struct trapframe, fixreg[23])); ASSYM(FRAME_24, offsetof(struct trapframe, fixreg[24])); ASSYM(FRAME_25, offsetof(struct trapframe, fixreg[25])); ASSYM(FRAME_26, offsetof(struct trapframe, fixreg[26])); ASSYM(FRAME_27, offsetof(struct trapframe, fixreg[27])); ASSYM(FRAME_28, offsetof(struct trapframe, fixreg[28])); ASSYM(FRAME_29, offsetof(struct trapframe, fixreg[29])); ASSYM(FRAME_30, offsetof(struct trapframe, fixreg[30])); ASSYM(FRAME_31, offsetof(struct trapframe, fixreg[31])); ASSYM(FRAME_LR, offsetof(struct trapframe, lr)); ASSYM(FRAME_CR, offsetof(struct trapframe, cr)); ASSYM(FRAME_CTR, offsetof(struct trapframe, ctr)); ASSYM(FRAME_XER, offsetof(struct trapframe, xer)); ASSYM(FRAME_SRR0, offsetof(struct trapframe, srr0)); ASSYM(FRAME_SRR1, offsetof(struct trapframe, srr1)); ASSYM(FRAME_EXC, offsetof(struct trapframe, exc)); -ASSYM(FRAME_AIM_DAR, offsetof(struct trapframe, cpu.aim.dar)); +ASSYM(FRAME_AIM_DAR, offsetof(struct trapframe, dar)); ASSYM(FRAME_AIM_DSISR, offsetof(struct trapframe, cpu.aim.dsisr)); -ASSYM(FRAME_BOOKE_DEAR, offsetof(struct trapframe, cpu.booke.dear)); +ASSYM(FRAME_BOOKE_DEAR, offsetof(struct trapframe, dar)); ASSYM(FRAME_BOOKE_ESR, offsetof(struct trapframe, cpu.booke.esr)); ASSYM(FRAME_BOOKE_DBCR0, offsetof(struct trapframe, cpu.booke.dbcr0)); ASSYM(CF_FUNC, offsetof(struct callframe, cf_func)); ASSYM(CF_ARG0, offsetof(struct callframe, cf_arg0)); ASSYM(CF_ARG1, offsetof(struct callframe, cf_arg1)); ASSYM(CF_SIZE, sizeof(struct callframe)); ASSYM(PCB_CONTEXT, offsetof(struct pcb, pcb_context)); ASSYM(PCB_CR, offsetof(struct pcb, pcb_cr)); ASSYM(PCB_SP, offsetof(struct pcb, pcb_sp)); ASSYM(PCB_TOC, offsetof(struct pcb, pcb_toc)); ASSYM(PCB_LR, offsetof(struct pcb, pcb_lr)); ASSYM(PCB_ONFAULT, offsetof(struct pcb, pcb_onfault)); ASSYM(PCB_FLAGS, offsetof(struct pcb, pcb_flags)); ASSYM(PCB_FPU, PCB_FPU); ASSYM(PCB_VEC, PCB_VEC); ASSYM(PCB_AIM_USR_VSID, offsetof(struct pcb, pcb_cpu.aim.usr_vsid)); ASSYM(PCB_BOOKE_DBCR0, offsetof(struct pcb, pcb_cpu.booke.dbcr0)); ASSYM(TD_LOCK, offsetof(struct thread, td_lock)); ASSYM(TD_PROC, offsetof(struct thread, td_proc)); ASSYM(TD_PCB, offsetof(struct thread, td_pcb)); ASSYM(P_VMSPACE, offsetof(struct proc, p_vmspace)); ASSYM(VM_PMAP, offsetof(struct vmspace, vm_pmap)); ASSYM(TD_FLAGS, offsetof(struct thread, td_flags)); ASSYM(TDF_ASTPENDING, TDF_ASTPENDING); ASSYM(TDF_NEEDRESCHED, TDF_NEEDRESCHED); ASSYM(SF_UC, offsetof(struct sigframe, sf_uc)); ASSYM(KERNBASE, KERNBASE); ASSYM(MAXCOMLEN, MAXCOMLEN); ASSYM(PSL_DE, PSL_DE); ASSYM(PSL_DS, PSL_DS); ASSYM(PSL_IS, PSL_IS); ASSYM(PSL_CE, PSL_CE); ASSYM(PSL_UCLE, PSL_UCLE); ASSYM(PSL_WE, PSL_WE); ASSYM(PSL_UBLE, PSL_UBLE); #if defined(BOOKE_E500) ASSYM(PSL_KERNSET_INIT, PSL_KERNSET_INIT); #endif #if defined(AIM) && defined(__powerpc64__) ASSYM(PSL_SF, PSL_SF); ASSYM(PSL_HV, PSL_HV); #endif ASSYM(PSL_POW, PSL_POW); ASSYM(PSL_ILE, PSL_ILE); ASSYM(PSL_LE, PSL_LE); ASSYM(PSL_SE, PSL_SE); ASSYM(PSL_RI, PSL_RI); ASSYM(PSL_DR, PSL_DR); ASSYM(PSL_IP, PSL_IP); ASSYM(PSL_IR, PSL_IR); ASSYM(PSL_FE_DIS, PSL_FE_DIS); ASSYM(PSL_FE_NONREC, PSL_FE_NONREC); ASSYM(PSL_FE_PREC, PSL_FE_PREC); ASSYM(PSL_FE_REC, PSL_FE_REC); ASSYM(PSL_VEC, PSL_VEC); ASSYM(PSL_BE, PSL_BE); ASSYM(PSL_EE, PSL_EE); ASSYM(PSL_FE0, PSL_FE0); ASSYM(PSL_FE1, PSL_FE1); ASSYM(PSL_FP, PSL_FP); ASSYM(PSL_ME, PSL_ME); ASSYM(PSL_PR, PSL_PR); ASSYM(PSL_PMM, PSL_PMM); ASSYM(PSL_KERNSET, PSL_KERNSET); ASSYM(PSL_USERSET, PSL_USERSET); ASSYM(PSL_USERSTATIC, PSL_USERSTATIC); Index: projects/cxl_iscsi/sys/powerpc/powerpc/swtch32.S =================================================================== --- projects/cxl_iscsi/sys/powerpc/powerpc/swtch32.S (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/powerpc/swtch32.S (revision 279717) @@ -1,218 +1,204 @@ /* $FreeBSD$ */ /* $NetBSD: locore.S,v 1.24 2000/05/31 05:09:17 thorpej Exp $ */ /*- * Copyright (C) 2001 Benno Rice * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY Benno Rice ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /*- * Copyright (C) 1995, 1996 Wolfgang Solfrank. * Copyright (C) 1995, 1996 TooLs GmbH. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by TooLs GmbH. * 4. The name of TooLs GmbH may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "assym.s" #include "opt_sched.h" #include #include #include #include #include /* * void cpu_throw(struct thread *old, struct thread *new) */ ENTRY(cpu_throw) mr %r2, %r4 li %r14,0 /* Tell cpu_switchin not to release a thread */ b cpu_switchin /* * void cpu_switch(struct thread *old, * struct thread *new, * struct mutex *mtx); * * Switch to a new thread saving the current state in the old thread. */ ENTRY(cpu_switch) lwz %r6,TD_PCB(%r3) /* Get the old thread's PCB ptr */ stmw %r12,PCB_CONTEXT(%r6) /* Save the non-volatile GP regs. These can now be used for scratch */ mfcr %r16 /* Save the condition register */ stw %r16,PCB_CR(%r6) mflr %r16 /* Save the link register */ stw %r16,PCB_LR(%r6) -#ifdef BOOKE - mfspr %r16,SPR_DBCR0 - stw %r16,PCB_BOOKE_DBCR0(%r6) -#endif stw %r1,PCB_SP(%r6) /* Save the stack pointer */ mr %r14,%r3 /* Copy the old thread ptr... */ mr %r2,%r4 /* and the new thread ptr in curthread */ mr %r16,%r5 /* and the new lock */ mr %r17,%r6 /* and the PCB */ lwz %r7,PCB_FLAGS(%r17) /* Save FPU context if needed */ andi. %r7, %r7, PCB_FPU beq .L1 bl save_fpu .L1: mr %r3,%r14 /* restore old thread ptr */ lwz %r7,PCB_FLAGS(%r17) /* Save Altivec context if needed */ andi. %r7, %r7, PCB_VEC beq .L2 bl save_vec .L2: mr %r3,%r14 /* restore old thread ptr */ bl pmap_deactivate /* Deactivate the current pmap */ sync /* Make sure all of that finished */ cpu_switchin: #if defined(SMP) && defined(SCHED_ULE) /* Wait for the new thread to become unblocked */ lis %r6,blocked_lock@ha addi %r6,%r6,blocked_lock@l blocked_loop: lwz %r7,TD_LOCK(%r2) cmpw %r6,%r7 beq- blocked_loop isync #endif lwz %r17,TD_PCB(%r2) /* Get new current PCB */ lwz %r1,PCB_SP(%r17) /* Load new stack pointer */ /* Release old thread now that we have a stack pointer set up */ cmpwi %r14,0 beq- 1f stw %r16,TD_LOCK(%r14) /* ULE: update old thread's lock */ 1: mfsprg %r7,0 /* Get the pcpu pointer */ stw %r2,PC_CURTHREAD(%r7) /* Store new current thread */ lwz %r17,TD_PCB(%r2) /* Store new current PCB */ stw %r17,PC_CURPCB(%r7) mr %r3,%r2 /* Get new thread ptr */ bl pmap_activate /* Activate the new address space */ lwz %r6, PCB_FLAGS(%r17) /* Restore FPU context if needed */ andi. %r6, %r6, PCB_FPU beq .L3 mr %r3,%r2 /* Pass curthread to enable_fpu */ bl enable_fpu .L3: lwz %r6, PCB_FLAGS(%r17) /* Restore Altivec context if needed */ andi. %r6, %r6, PCB_VEC beq .L4 mr %r3,%r2 /* Pass curthread to enable_vec */ bl enable_vec .L4: /* thread to restore is in r3 */ mr %r3,%r17 /* Recover PCB ptr */ lmw %r12,PCB_CONTEXT(%r3) /* Load the non-volatile GP regs */ lwz %r5,PCB_CR(%r3) /* Load the condition register */ mtcr %r5 lwz %r5,PCB_LR(%r3) /* Load the link register */ mtlr %r5 -#ifdef AIM - lwz %r5,PCB_AIM_USR_VSID(%r3) /* Load the USER_SR segment reg */ - isync - mtsr USER_SR,%r5 - isync -#endif -#ifdef BOOKE - lwz %r5,PCB_BOOKE_DBCR0(%r3) - mtspr SPR_DBCR0,%r5 -#endif lwz %r1,PCB_SP(%r3) /* Load the stack pointer */ /* * Perform a dummy stwcx. to clear any reservations we may have * inherited from the previous thread. It doesn't matter if the * stwcx succeeds or not. pcb_context[0] can be clobbered. */ stwcx. %r1, 0, %r3 blr /* * savectx(pcb) * Update pcb, saving current processor state */ ENTRY(savectx) stmw %r12,PCB_CONTEXT(%r3) /* Save the non-volatile GP regs */ mfcr %r4 /* Save the condition register */ stw %r4,PCB_CR(%r3) blr /* * fork_trampoline() * Set up the return from cpu_fork() */ ENTRY(fork_trampoline) lwz %r3,CF_FUNC(%r1) lwz %r4,CF_ARG0(%r1) lwz %r5,CF_ARG1(%r1) bl fork_exit addi %r1,%r1,CF_SIZE-FSP /* Allow 8 bytes in front of trapframe to simulate FRAME_SETUP does when allocating space for a frame pointer/saved LR */ b trapexit Index: projects/cxl_iscsi/sys/powerpc/powerpc/swtch64.S =================================================================== --- projects/cxl_iscsi/sys/powerpc/powerpc/swtch64.S (revision 279716) +++ projects/cxl_iscsi/sys/powerpc/powerpc/swtch64.S (revision 279717) @@ -1,293 +1,278 @@ /* $FreeBSD$ */ /* $NetBSD: locore.S,v 1.24 2000/05/31 05:09:17 thorpej Exp $ */ /*- * Copyright (C) 2001 Benno Rice * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY Benno Rice ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /*- * Copyright (C) 1995, 1996 Wolfgang Solfrank. * Copyright (C) 1995, 1996 TooLs GmbH. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by TooLs GmbH. * 4. The name of TooLs GmbH may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "assym.s" #include "opt_sched.h" #include #include #include #include TOC_ENTRY(blocked_lock) /* * void cpu_throw(struct thread *old, struct thread *new) */ ENTRY(cpu_throw) mr %r13, %r4 li %r14,0 /* Tell cpu_switchin not to release a thread */ b cpu_switchin /* * void cpu_switch(struct thread *old, * struct thread *new, * struct mutex *mtx); * * Switch to a new thread saving the current state in the old thread. */ ENTRY(cpu_switch) ld %r6,TD_PCB(%r3) /* Get the old thread's PCB ptr */ std %r12,PCB_CONTEXT(%r6) /* Save the non-volatile GP regs. These can now be used for scratch */ std %r14,PCB_CONTEXT+2*8(%r6) std %r15,PCB_CONTEXT+3*8(%r6) std %r16,PCB_CONTEXT+4*8(%r6) std %r17,PCB_CONTEXT+5*8(%r6) std %r18,PCB_CONTEXT+6*8(%r6) std %r19,PCB_CONTEXT+7*8(%r6) std %r20,PCB_CONTEXT+8*8(%r6) std %r21,PCB_CONTEXT+9*8(%r6) std %r22,PCB_CONTEXT+10*8(%r6) std %r23,PCB_CONTEXT+11*8(%r6) std %r24,PCB_CONTEXT+12*8(%r6) std %r25,PCB_CONTEXT+13*8(%r6) std %r26,PCB_CONTEXT+14*8(%r6) std %r27,PCB_CONTEXT+15*8(%r6) std %r28,PCB_CONTEXT+16*8(%r6) std %r29,PCB_CONTEXT+17*8(%r6) std %r30,PCB_CONTEXT+18*8(%r6) std %r31,PCB_CONTEXT+19*8(%r6) mfcr %r16 /* Save the condition register */ std %r16,PCB_CR(%r6) mflr %r16 /* Save the link register */ std %r16,PCB_LR(%r6) std %r1,PCB_SP(%r6) /* Save the stack pointer */ std %r2,PCB_TOC(%r6) /* Save the TOC pointer */ mr %r14,%r3 /* Copy the old thread ptr... */ mr %r13,%r4 /* and the new thread ptr in curthread*/ mr %r16,%r5 /* and the new lock */ mr %r17,%r6 /* and the PCB */ stdu %r1,-48(%r1) lwz %r7,PCB_FLAGS(%r17) /* Save FPU context if needed */ andi. %r7, %r7, PCB_FPU beq .L1 bl save_fpu nop .L1: mr %r3,%r14 /* restore old thread ptr */ lwz %r7,PCB_FLAGS(%r17) /* Save Altivec context if needed */ andi. %r7, %r7, PCB_VEC beq .L2 bl save_vec nop .L2: mr %r3,%r14 /* restore old thread ptr */ bl pmap_deactivate /* Deactivate the current pmap */ nop sync /* Make sure all of that finished */ cpu_switchin: #if defined(SMP) && defined(SCHED_ULE) /* Wait for the new thread to become unblocked */ ld %r6,TOC_REF(blocked_lock)(%r2) blocked_loop: ld %r7,TD_LOCK(%r13) cmpd %r6,%r7 beq- blocked_loop isync #endif ld %r17,TD_PCB(%r13) /* Get new PCB */ ld %r1,PCB_SP(%r17) /* Load the stack pointer */ /* Release old thread now that we have a stack pointer set up */ cmpdi %r14,0 beq- 1f std %r16,TD_LOCK(%r14) /* ULE: update old thread's lock */ 1: mfsprg %r7,0 /* Get the pcpu pointer */ std %r13,PC_CURTHREAD(%r7) /* Store new current thread */ ld %r17,TD_PCB(%r13) /* Store new current PCB */ std %r17,PC_CURPCB(%r7) mr %r3,%r13 /* Get new thread ptr */ bl pmap_activate /* Activate the new address space */ nop lwz %r6, PCB_FLAGS(%r17) /* Restore FPU context if needed */ andi. %r6, %r6, PCB_FPU beq .L3 mr %r3,%r13 /* Pass curthread to enable_fpu */ bl enable_fpu nop .L3: lwz %r6, PCB_FLAGS(%r17) /* Restore Altivec context if needed */ andi. %r6, %r6, PCB_VEC beq .L4 mr %r3,%r13 /* Pass curthread to enable_vec */ bl enable_vec nop /* thread to restore is in r3 */ .L4: addi %r1,%r1,48 mr %r3,%r17 /* Recover PCB ptr */ ld %r12,PCB_CONTEXT(%r3) /* Load the non-volatile GP regs. */ ld %r14,PCB_CONTEXT+2*8(%r3) ld %r15,PCB_CONTEXT+3*8(%r3) ld %r16,PCB_CONTEXT+4*8(%r3) ld %r17,PCB_CONTEXT+5*8(%r3) ld %r18,PCB_CONTEXT+6*8(%r3) ld %r19,PCB_CONTEXT+7*8(%r3) ld %r20,PCB_CONTEXT+8*8(%r3) ld %r21,PCB_CONTEXT+9*8(%r3) ld %r22,PCB_CONTEXT+10*8(%r3) ld %r23,PCB_CONTEXT+11*8(%r3) ld %r24,PCB_CONTEXT+12*8(%r3) ld %r25,PCB_CONTEXT+13*8(%r3) ld %r26,PCB_CONTEXT+14*8(%r3) ld %r27,PCB_CONTEXT+15*8(%r3) ld %r28,PCB_CONTEXT+16*8(%r3) ld %r29,PCB_CONTEXT+17*8(%r3) ld %r30,PCB_CONTEXT+18*8(%r3) ld %r31,PCB_CONTEXT+19*8(%r3) ld %r5,PCB_CR(%r3) /* Load the condition register */ mtcr %r5 ld %r5,PCB_LR(%r3) /* Load the link register */ mtlr %r5 ld %r1,PCB_SP(%r3) /* Load the stack pointer */ ld %r2,PCB_TOC(%r3) /* Load the TOC pointer */ - lis %r5,USER_ADDR@highesta /* Load the copyin/out segment reg */ - ori %r5,%r5,USER_ADDR@highera - sldi %r5,%r5,32 - oris %r5,%r5,USER_ADDR@ha - isync - slbie %r5 - lis %r6,USER_SLB_SLBE@highesta - ori %r6,%r6,USER_SLB_SLBE@highera - sldi %r6,%r6,32 - oris %r6,%r6,USER_SLB_SLBE@ha - ori %r6,%r6,USER_SLB_SLBE@l - ld %r5,PCB_AIM_USR_VSID(%r3) - slbmte %r5,%r6 - isync - /* * Perform a dummy stdcx. to clear any reservations we may have * inherited from the previous thread. It doesn't matter if the * stdcx succeeds or not. pcb_context[0] can be clobbered. */ stdcx. %r1, 0, %r3 blr /* * savectx(pcb) * Update pcb, saving current processor state */ ENTRY(savectx) std %r12,PCB_CONTEXT(%r3) /* Save the non-volatile GP regs. */ std %r13,PCB_CONTEXT+1*8(%r3) std %r14,PCB_CONTEXT+2*8(%r3) std %r15,PCB_CONTEXT+3*8(%r3) std %r16,PCB_CONTEXT+4*8(%r3) std %r17,PCB_CONTEXT+5*8(%r3) std %r18,PCB_CONTEXT+6*8(%r3) std %r19,PCB_CONTEXT+7*8(%r3) std %r20,PCB_CONTEXT+8*8(%r3) std %r21,PCB_CONTEXT+9*8(%r3) std %r22,PCB_CONTEXT+10*8(%r3) std %r23,PCB_CONTEXT+11*8(%r3) std %r24,PCB_CONTEXT+12*8(%r3) std %r25,PCB_CONTEXT+13*8(%r3) std %r26,PCB_CONTEXT+14*8(%r3) std %r27,PCB_CONTEXT+15*8(%r3) std %r28,PCB_CONTEXT+16*8(%r3) std %r29,PCB_CONTEXT+17*8(%r3) std %r30,PCB_CONTEXT+18*8(%r3) std %r31,PCB_CONTEXT+19*8(%r3) mfcr %r4 /* Save the condition register */ std %r4,PCB_CR(%r3) std %r2,PCB_TOC(%r3) /* Save the TOC pointer */ blr /* * fork_trampoline() * Set up the return from cpu_fork() */ ENTRY_NOPROF(fork_trampoline) ld %r3,CF_FUNC(%r1) ld %r4,CF_ARG0(%r1) ld %r5,CF_ARG1(%r1) stdu %r1,-48(%r1) bl fork_exit nop addi %r1,%r1,48+CF_SIZE-FSP /* Allow 8 bytes in front of trapframe to simulate FRAME_SETUP does when allocating space for a frame pointer/saved LR */ b trapexit nop Index: projects/cxl_iscsi/sys/sys/buf.h =================================================================== --- projects/cxl_iscsi/sys/sys/buf.h (revision 279716) +++ projects/cxl_iscsi/sys/sys/buf.h (revision 279717) @@ -1,550 +1,552 @@ /*- * Copyright (c) 1982, 1986, 1989, 1993 * The Regents of the University of California. All rights reserved. * (c) UNIX System Laboratories, Inc. * All or some portions of this file are derived from material licensed * to the University of California by American Telephone and Telegraph * Co. or Unix System Laboratories, Inc. and are reproduced herein with * the permission of UNIX System Laboratories, Inc. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)buf.h 8.9 (Berkeley) 3/30/95 * $FreeBSD$ */ #ifndef _SYS_BUF_H_ #define _SYS_BUF_H_ #include #include #include #include struct bio; struct buf; struct bufobj; struct mount; struct vnode; struct uio; /* * To avoid including */ LIST_HEAD(workhead, worklist); /* * These are currently used only by the soft dependency code, hence * are stored once in a global variable. If other subsystems wanted * to use these hooks, a pointer to a set of bio_ops could be added * to each buffer. */ extern struct bio_ops { void (*io_start)(struct buf *); void (*io_complete)(struct buf *); void (*io_deallocate)(struct buf *); int (*io_countdeps)(struct buf *, int); } bioops; struct vm_object; typedef unsigned char b_xflags_t; /* * The buffer header describes an I/O operation in the kernel. * * NOTES: * b_bufsize, b_bcount. b_bufsize is the allocation size of the * buffer, either DEV_BSIZE or PAGE_SIZE aligned. b_bcount is the * originally requested buffer size and can serve as a bounds check * against EOF. For most, but not all uses, b_bcount == b_bufsize. * * b_dirtyoff, b_dirtyend. Buffers support piecemeal, unaligned * ranges of dirty data that need to be written to backing store. * The range is typically clipped at b_bcount ( not b_bufsize ). * * b_resid. Number of bytes remaining in I/O. After an I/O operation * completes, b_resid is usually 0 indicating 100% success. * * All fields are protected by the buffer lock except those marked: * V - Protected by owning bufobj lock * Q - Protected by the buf queue lock * D - Protected by an dependency implementation specific lock */ struct buf { struct bufobj *b_bufobj; long b_bcount; void *b_caller1; caddr_t b_data; int b_error; uint8_t b_iocmd; uint8_t b_ioflags; off_t b_iooffset; long b_resid; void (*b_iodone)(struct buf *); daddr_t b_blkno; /* Underlying physical block number. */ off_t b_offset; /* Offset into file. */ TAILQ_ENTRY(buf) b_bobufs; /* (V) Buffer's associated vnode. */ uint32_t b_vflags; /* (V) BV_* flags */ unsigned short b_qindex; /* (Q) buffer queue index */ uint32_t b_flags; /* B_* flags. */ b_xflags_t b_xflags; /* extra flags */ struct lock b_lock; /* Buffer lock */ long b_bufsize; /* Allocated buffer size. */ long b_runningbufspace; /* when I/O is running, pipelining */ caddr_t b_kvabase; /* base kva for buffer */ caddr_t b_kvaalloc; /* allocated kva for B_KVAALLOC */ int b_kvasize; /* size of kva for buffer */ daddr_t b_lblkno; /* Logical block number. */ struct vnode *b_vp; /* Device vnode. */ int b_dirtyoff; /* Offset in buffer of dirty region. */ int b_dirtyend; /* Offset of end of dirty region. */ struct ucred *b_rcred; /* Read credentials reference. */ struct ucred *b_wcred; /* Write credentials reference. */ void *b_saveaddr; /* Original b_addr for physio. */ union { TAILQ_ENTRY(buf) bu_freelist; /* (Q) */ struct { void (*pg_iodone)(void *, vm_page_t *, int, int); int pg_reqpage; } bu_pager; } b_union; #define b_freelist b_union.bu_freelist #define b_pager b_union.bu_pager union cluster_info { TAILQ_HEAD(cluster_list_head, buf) cluster_head; TAILQ_ENTRY(buf) cluster_entry; } b_cluster; struct vm_page *b_pages[btoc(MAXPHYS)]; int b_npages; struct workhead b_dep; /* (D) List of filesystem dependencies. */ void *b_fsprivate1; void *b_fsprivate2; void *b_fsprivate3; int b_pin_count; }; #define b_object b_bufobj->bo_object /* * These flags are kept in b_flags. * * Notes: * * B_ASYNC VOP calls on bp's are usually async whether or not * B_ASYNC is set, but some subsystems, such as NFS, like * to know what is best for the caller so they can * optimize the I/O. * * B_PAGING Indicates that bp is being used by the paging system or * some paging system and that the bp is not linked into * the b_vp's clean/dirty linked lists or ref counts. * Buffer vp reassignments are illegal in this case. * * B_CACHE This may only be set if the buffer is entirely valid. * The situation where B_DELWRI is set and B_CACHE is * clear MUST be committed to disk by getblk() so * B_DELWRI can also be cleared. See the comments for * getblk() in kern/vfs_bio.c. If B_CACHE is clear, * the caller is expected to clear BIO_ERROR and B_INVAL, * set BIO_READ, and initiate an I/O. * * The 'entire buffer' is defined to be the range from * 0 through b_bcount. * * B_MALLOC Request that the buffer be allocated from the malloc * pool, DEV_BSIZE aligned instead of PAGE_SIZE aligned. * * B_CLUSTEROK This flag is typically set for B_DELWRI buffers * by filesystems that allow clustering when the buffer * is fully dirty and indicates that it may be clustered * with other adjacent dirty buffers. Note the clustering * may not be used with the stage 1 data write under NFS * but may be used for the commit rpc portion. * * B_VMIO Indicates that the buffer is tied into an VM object. * The buffer's data is always PAGE_SIZE aligned even * if b_bufsize and b_bcount are not. ( b_bufsize is * always at least DEV_BSIZE aligned, though ). * * B_DIRECT Hint that we should attempt to completely free * the pages underlying the buffer. B_DIRECT is * sticky until the buffer is released and typically * only has an effect when B_RELBUF is also set. * */ #define B_AGE 0x00000001 /* Move to age queue when I/O done. */ #define B_NEEDCOMMIT 0x00000002 /* Append-write in progress. */ #define B_ASYNC 0x00000004 /* Start I/O, do not wait. */ #define B_DIRECT 0x00000008 /* direct I/O flag (pls free vmio) */ #define B_DEFERRED 0x00000010 /* Skipped over for cleaning */ #define B_CACHE 0x00000020 /* Bread found us in the cache. */ #define B_VALIDSUSPWRT 0x00000040 /* Valid write during suspension. */ #define B_DELWRI 0x00000080 /* Delay I/O until buffer reused. */ #define B_PERSISTENT 0x00000100 /* Perm. ref'ed while EXT2FS mounted. */ #define B_DONE 0x00000200 /* I/O completed. */ #define B_EINTR 0x00000400 /* I/O was interrupted */ #define B_UNMAPPED 0x00000800 /* KVA is not mapped. */ #define B_KVAALLOC 0x00001000 /* But allocated. */ #define B_INVAL 0x00002000 /* Does not contain valid info. */ #define B_BARRIER 0x00004000 /* Write this and all preceeding first. */ #define B_NOCACHE 0x00008000 /* Do not cache block after use. */ #define B_MALLOC 0x00010000 /* malloced b_data */ #define B_CLUSTEROK 0x00020000 /* Pagein op, so swap() can count it. */ #define B_000400000 0x00040000 /* Available flag. */ #define B_000800000 0x00080000 /* Available flag. */ #define B_001000000 0x00100000 /* Available flag. */ #define B_DIRTY 0x00200000 /* Needs writing later (in EXT2FS). */ #define B_RELBUF 0x00400000 /* Release VMIO buffer. */ #define B_FS_FLAG1 0x00800000 /* Available flag for FS use. */ #define B_NOCOPY 0x01000000 /* Don't copy-on-write this buf. */ #define B_INFREECNT 0x02000000 /* buf is counted in numfreebufs */ #define B_PAGING 0x04000000 /* volatile paging I/O -- bypass VMIO */ #define B_MANAGED 0x08000000 /* Managed by FS. */ #define B_RAM 0x10000000 /* Read ahead mark (flag) */ #define B_VMIO 0x20000000 /* VMIO flag */ #define B_CLUSTER 0x40000000 /* pagein op, so swap() can count it */ #define B_REMFREE 0x80000000 /* Delayed bremfree */ #define PRINT_BUF_FLAGS "\20\40remfree\37cluster\36vmio\35ram\34managed" \ "\33paging\32infreecnt\31nocopy\30b23\27relbuf\26dirty\25b20" \ "\24b19\23b18\22clusterok\21malloc\20nocache\17b14\16inval" \ "\15b12\14b11\13eintr\12done\11persist\10delwri\7validsuspwrt" \ "\6cache\5deferred\4direct\3async\2needcommit\1age" /* * These flags are kept in b_xflags. */ #define BX_VNDIRTY 0x00000001 /* On vnode dirty list */ #define BX_VNCLEAN 0x00000002 /* On vnode clean list */ #define BX_BKGRDWRITE 0x00000010 /* Do writes in background */ #define BX_BKGRDMARKER 0x00000020 /* Mark buffer for splay tree */ #define BX_ALTDATA 0x00000040 /* Holds extended data */ #define PRINT_BUF_XFLAGS "\20\7altdata\6bkgrdmarker\5bkgrdwrite\2clean\1dirty" #define NOOFFSET (-1LL) /* No buffer offset calculated yet */ /* * These flags are kept in b_vflags. */ #define BV_SCANNED 0x00000001 /* VOP_FSYNC funcs mark written bufs */ #define BV_BKGRDINPROG 0x00000002 /* Background write in progress */ #define BV_BKGRDWAIT 0x00000004 /* Background write waiting */ #define PRINT_BUF_VFLAGS "\20\3bkgrdwait\2bkgrdinprog\1scanned" #ifdef _KERNEL /* * Buffer locking */ extern const char *buf_wmesg; /* Default buffer lock message */ #define BUF_WMESG "bufwait" #include /* XXX for curthread */ #include /* * Initialize a lock. */ #define BUF_LOCKINIT(bp) \ lockinit(&(bp)->b_lock, PRIBIO + 4, buf_wmesg, 0, 0) /* * * Get a lock sleeping non-interruptably until it becomes available. */ #define BUF_LOCK(bp, locktype, interlock) \ _lockmgr_args_rw(&(bp)->b_lock, (locktype), (interlock), \ LK_WMESG_DEFAULT, LK_PRIO_DEFAULT, LK_TIMO_DEFAULT, \ LOCK_FILE, LOCK_LINE) /* * Get a lock sleeping with specified interruptably and timeout. */ #define BUF_TIMELOCK(bp, locktype, interlock, wmesg, catch, timo) \ _lockmgr_args_rw(&(bp)->b_lock, (locktype) | LK_TIMELOCK, \ (interlock), (wmesg), (PRIBIO + 4) | (catch), (timo), \ LOCK_FILE, LOCK_LINE) /* * Release a lock. Only the acquiring process may free the lock unless * it has been handed off to biodone. */ #define BUF_UNLOCK(bp) do { \ KASSERT(((bp)->b_flags & B_REMFREE) == 0, \ ("BUF_UNLOCK %p while B_REMFREE is still set.", (bp))); \ \ (void)_lockmgr_args(&(bp)->b_lock, LK_RELEASE, NULL, \ LK_WMESG_DEFAULT, LK_PRIO_DEFAULT, LK_TIMO_DEFAULT, \ LOCK_FILE, LOCK_LINE); \ } while (0) /* * Check if a buffer lock is recursed. */ #define BUF_LOCKRECURSED(bp) \ lockmgr_recursed(&(bp)->b_lock) /* * Check if a buffer lock is currently held. */ #define BUF_ISLOCKED(bp) \ lockstatus(&(bp)->b_lock) /* * Free a buffer lock. */ #define BUF_LOCKFREE(bp) \ lockdestroy(&(bp)->b_lock) /* * Print informations on a buffer lock. */ #define BUF_LOCKPRINTINFO(bp) \ lockmgr_printinfo(&(bp)->b_lock) /* * Buffer lock assertions. */ #if defined(INVARIANTS) && defined(INVARIANT_SUPPORT) #define BUF_ASSERT_LOCKED(bp) \ _lockmgr_assert(&(bp)->b_lock, KA_LOCKED, LOCK_FILE, LOCK_LINE) #define BUF_ASSERT_SLOCKED(bp) \ _lockmgr_assert(&(bp)->b_lock, KA_SLOCKED, LOCK_FILE, LOCK_LINE) #define BUF_ASSERT_XLOCKED(bp) \ _lockmgr_assert(&(bp)->b_lock, KA_XLOCKED, LOCK_FILE, LOCK_LINE) #define BUF_ASSERT_UNLOCKED(bp) \ _lockmgr_assert(&(bp)->b_lock, KA_UNLOCKED, LOCK_FILE, LOCK_LINE) #define BUF_ASSERT_HELD(bp) #define BUF_ASSERT_UNHELD(bp) #else #define BUF_ASSERT_LOCKED(bp) #define BUF_ASSERT_SLOCKED(bp) #define BUF_ASSERT_XLOCKED(bp) #define BUF_ASSERT_UNLOCKED(bp) #define BUF_ASSERT_HELD(bp) #define BUF_ASSERT_UNHELD(bp) #endif #ifdef _SYS_PROC_H_ /* Avoid #include pollution */ /* * When initiating asynchronous I/O, change ownership of the lock to the * kernel. Once done, the lock may legally released by biodone. The * original owning process can no longer acquire it recursively, but must * wait until the I/O is completed and the lock has been freed by biodone. */ #define BUF_KERNPROC(bp) \ _lockmgr_disown(&(bp)->b_lock, LOCK_FILE, LOCK_LINE) #endif /* * Find out if the lock has waiters or not. */ #define BUF_LOCKWAITERS(bp) \ lockmgr_waiters(&(bp)->b_lock) #endif /* _KERNEL */ struct buf_queue_head { TAILQ_HEAD(buf_queue, buf) queue; daddr_t last_pblkno; struct buf *insert_point; struct buf *switch_point; }; /* * This structure describes a clustered I/O. It is stored in the b_saveaddr * field of the buffer on which I/O is done. At I/O completion, cluster * callback uses the structure to parcel I/O's to individual buffers, and * then free's this structure. */ struct cluster_save { long bs_bcount; /* Saved b_bcount. */ long bs_bufsize; /* Saved b_bufsize. */ void *bs_saveaddr; /* Saved b_addr. */ int bs_nchildren; /* Number of associated buffers. */ struct buf **bs_children; /* List of associated buffers. */ }; #ifdef _KERNEL static __inline int bwrite(struct buf *bp) { KASSERT(bp->b_bufobj != NULL, ("bwrite: no bufobj bp=%p", bp)); KASSERT(bp->b_bufobj->bo_ops != NULL, ("bwrite: no bo_ops bp=%p", bp)); KASSERT(bp->b_bufobj->bo_ops->bop_write != NULL, ("bwrite: no bop_write bp=%p", bp)); return (BO_WRITE(bp->b_bufobj, bp)); } static __inline void bstrategy(struct buf *bp) { KASSERT(bp->b_bufobj != NULL, ("bstrategy: no bufobj bp=%p", bp)); KASSERT(bp->b_bufobj->bo_ops != NULL, ("bstrategy: no bo_ops bp=%p", bp)); KASSERT(bp->b_bufobj->bo_ops->bop_strategy != NULL, ("bstrategy: no bop_strategy bp=%p", bp)); BO_STRATEGY(bp->b_bufobj, bp); } static __inline void buf_start(struct buf *bp) { if (bioops.io_start) (*bioops.io_start)(bp); } static __inline void buf_complete(struct buf *bp) { if (bioops.io_complete) (*bioops.io_complete)(bp); } static __inline void buf_deallocate(struct buf *bp) { if (bioops.io_deallocate) (*bioops.io_deallocate)(bp); } static __inline int buf_countdeps(struct buf *bp, int i) { if (bioops.io_countdeps) return ((*bioops.io_countdeps)(bp, i)); else return (0); } #endif /* _KERNEL */ /* * Zero out the buffer's data area. */ #define clrbuf(bp) { \ bzero((bp)->b_data, (u_int)(bp)->b_bcount); \ (bp)->b_resid = 0; \ } /* * Flags for getblk's last parameter. */ #define GB_LOCK_NOWAIT 0x0001 /* Fail if we block on a buf lock. */ #define GB_NOCREAT 0x0002 /* Don't create a buf if not found. */ #define GB_NOWAIT_BD 0x0004 /* Do not wait for bufdaemon. */ #define GB_UNMAPPED 0x0008 /* Do not mmap buffer pages. */ #define GB_KVAALLOC 0x0010 /* But allocate KVA. */ #ifdef _KERNEL extern int nbuf; /* The number of buffer headers */ extern long maxswzone; /* Max KVA for swap structures */ extern long maxbcache; /* Max KVA for buffer cache */ extern long runningbufspace; extern long hibufspace; extern int dirtybufthresh; extern int bdwriteskip; extern int dirtybufferflushes; extern int altbufferflushes; extern struct buf *buf; /* The buffer headers. */ extern struct buf *swbuf; /* Swap I/O buffer headers. */ extern int nswbuf; /* Number of swap I/O buffer headers. */ extern int cluster_pbuf_freecnt; /* Number of pbufs for clusters */ extern int vnode_pbuf_freecnt; /* Number of pbufs for vnode pager */ +extern int vnode_async_pbuf_freecnt; /* Number of pbufs for vnode pager, + asynchronous reads */ extern caddr_t unmapped_buf; void runningbufwakeup(struct buf *); void waitrunningbufspace(void); caddr_t kern_vfs_bio_buffer_alloc(caddr_t v, long physmem_est); void bufinit(void); void bdata2bio(struct buf *bp, struct bio *bip); void bwillwrite(void); int buf_dirty_count_severe(void); void bremfree(struct buf *); void bremfreef(struct buf *); /* XXX Force bremfree, only for nfs. */ #define bread(vp, blkno, size, cred, bpp) \ breadn_flags(vp, blkno, size, NULL, NULL, 0, cred, 0, bpp) #define bread_gb(vp, blkno, size, cred, gbflags, bpp) \ breadn_flags(vp, blkno, size, NULL, NULL, 0, cred, \ gbflags, bpp) #define breadn(vp, blkno, size, rablkno, rabsize, cnt, cred, bpp) \ breadn_flags(vp, blkno, size, rablkno, rabsize, cnt, cred, 0, bpp) int breadn_flags(struct vnode *, daddr_t, int, daddr_t *, int *, int, struct ucred *, int, struct buf **); void breada(struct vnode *, daddr_t *, int *, int, struct ucred *); void bdwrite(struct buf *); void bawrite(struct buf *); void babarrierwrite(struct buf *); int bbarrierwrite(struct buf *); void bdirty(struct buf *); void bundirty(struct buf *); void bufstrategy(struct bufobj *, struct buf *); void brelse(struct buf *); void bqrelse(struct buf *); int vfs_bio_awrite(struct buf *); void vfs_drain_busy_pages(struct buf *bp); struct buf * getpbuf(int *); struct buf *incore(struct bufobj *, daddr_t); struct buf *gbincore(struct bufobj *, daddr_t); struct buf *getblk(struct vnode *, daddr_t, int, int, int, int); struct buf *geteblk(int, int); int bufwait(struct buf *); int bufwrite(struct buf *); void bufdone(struct buf *); void bufdone_finish(struct buf *); void bd_speedup(void); int cluster_read(struct vnode *, u_quad_t, daddr_t, long, struct ucred *, long, int, int, struct buf **); int cluster_wbuild(struct vnode *, long, daddr_t, int, int); void cluster_write(struct vnode *, struct buf *, u_quad_t, int, int); void vfs_bio_bzero_buf(struct buf *bp, int base, int size); void vfs_bio_set_valid(struct buf *, int base, int size); void vfs_bio_clrbuf(struct buf *); void vfs_busy_pages(struct buf *, int clear_modify); void vfs_unbusy_pages(struct buf *); int vmapbuf(struct buf *, int); void vunmapbuf(struct buf *); void relpbuf(struct buf *, int *); void brelvp(struct buf *); void bgetvp(struct vnode *, struct buf *); void pbgetbo(struct bufobj *bo, struct buf *bp); void pbgetvp(struct vnode *, struct buf *); void pbrelbo(struct buf *); void pbrelvp(struct buf *); int allocbuf(struct buf *bp, int size); void reassignbuf(struct buf *); struct buf *trypbuf(int *); void bwait(struct buf *, u_char, const char *); void bdone(struct buf *); void bpin(struct buf *); void bunpin(struct buf *); void bunpin_wait(struct buf *); #endif /* _KERNEL */ #endif /* !_SYS_BUF_H_ */ Index: projects/cxl_iscsi/sys/sys/elf_common.h =================================================================== --- projects/cxl_iscsi/sys/sys/elf_common.h (revision 279716) +++ projects/cxl_iscsi/sys/sys/elf_common.h (revision 279717) @@ -1,1242 +1,1244 @@ /*- * Copyright (c) 2000, 2001, 2008, 2011, David E. O'Brien * Copyright (c) 1998 John D. Polstra. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #ifndef _SYS_ELF_COMMON_H_ #define _SYS_ELF_COMMON_H_ 1 /* * ELF definitions that are independent of architecture or word size. */ /* * Note header. The ".note" section contains an array of notes. Each * begins with this header, aligned to a word boundary. Immediately * following the note header is n_namesz bytes of name, padded to the * next word boundary. Then comes n_descsz bytes of descriptor, again * padded to a word boundary. The values of n_namesz and n_descsz do * not include the padding. */ typedef struct { u_int32_t n_namesz; /* Length of name. */ u_int32_t n_descsz; /* Length of descriptor. */ u_int32_t n_type; /* Type of this note. */ } Elf_Note; /* * Option kinds. */ #define ODK_NULL 0 /* undefined */ #define ODK_REGINFO 1 /* register usage info */ #define ODK_EXCEPTIONS 2 /* exception processing info */ #define ODK_PAD 3 /* section padding */ #define ODK_HWPATCH 4 /* hardware patch applied */ #define ODK_FILL 5 /* fill value used by the linker */ #define ODK_TAGS 6 /* reserved space for tools */ #define ODK_HWAND 7 /* hardware AND patch applied */ #define ODK_HWOR 8 /* hardware OR patch applied */ #define ODK_GP_GROUP 9 /* GP group for text/data sections */ #define ODK_IDENT 10 /* ID information */ #define ODK_PAGESIZE 11 /* page size information */ /* * ODK_EXCEPTIONS info field masks. */ #define OEX_FPU_MIN 0x0000001f /* min FPU exception required */ #define OEX_FPU_MAX 0x00001f00 /* max FPU exception allowed */ #define OEX_PAGE0 0x00010000 /* page zero must be mapped */ #define OEX_SMM 0x00020000 /* run in sequential memory mode */ #define OEX_PRECISEFP 0x00040000 /* run in precise FP exception mode */ #define OEX_DISMISS 0x00080000 /* dismiss invalid address traps */ /* * ODK_PAD info field masks. */ #define OPAD_PREFIX 0x0001 #define OPAD_POSTFIX 0x0002 #define OPAD_SYMBOL 0x0004 /* * ODK_HWPATCH info field masks. */ #define OHW_R4KEOP 0x00000001 /* patch for R4000 branch at end-of-page bug */ #define OHW_R8KPFETCH 0x00000002 /* R8000 prefetch bug may occur */ #define OHW_R5KEOP 0x00000004 /* patch for R5000 branch at end-of-page bug */ #define OHW_R5KCVTL 0x00000008 /* R5000 cvt.[ds].l bug: clean == 1 */ #define OHW_R10KLDL 0x00000010UL /* need patch for R10000 misaligned load */ /* * ODK_HWAND/ODK_HWOR info field and hwp_flags[12] masks. */ #define OHWA0_R4KEOP_CHECKED 0x00000001 /* object checked for R4000 end-of-page bug */ #define OHWA0_R4KEOP_CLEAN 0x00000002 /* object verified clean for R4000 end-of-page bug */ #define OHWO0_FIXADE 0x00000001 /* object requires call to fixade */ /* * ODK_IDENT/ODK_GP_GROUP info field masks. */ #define OGP_GROUP 0x0000ffff /* GP group number */ #define OGP_SELF 0x00010000 /* GP group is self-contained */ /* * The header for GNU-style hash sections. */ typedef struct { u_int32_t gh_nbuckets; /* Number of hash buckets. */ u_int32_t gh_symndx; /* First visible symbol in .dynsym. */ u_int32_t gh_maskwords; /* #maskwords used in bloom filter. */ u_int32_t gh_shift2; /* Bloom filter shift count. */ } Elf_GNU_Hash_Header; /* Indexes into the e_ident array. Keep synced with http://www.sco.com/developers/gabi/latest/ch4.eheader.html */ #define EI_MAG0 0 /* Magic number, byte 0. */ #define EI_MAG1 1 /* Magic number, byte 1. */ #define EI_MAG2 2 /* Magic number, byte 2. */ #define EI_MAG3 3 /* Magic number, byte 3. */ #define EI_CLASS 4 /* Class of machine. */ #define EI_DATA 5 /* Data format. */ #define EI_VERSION 6 /* ELF format version. */ #define EI_OSABI 7 /* Operating system / ABI identification */ #define EI_ABIVERSION 8 /* ABI version */ #define OLD_EI_BRAND 8 /* Start of architecture identification. */ #define EI_PAD 9 /* Start of padding (per SVR4 ABI). */ #define EI_NIDENT 16 /* Size of e_ident array. */ /* Values for the magic number bytes. */ #define ELFMAG0 0x7f #define ELFMAG1 'E' #define ELFMAG2 'L' #define ELFMAG3 'F' #define ELFMAG "\177ELF" /* magic string */ #define SELFMAG 4 /* magic string size */ /* Values for e_ident[EI_VERSION] and e_version. */ #define EV_NONE 0 #define EV_CURRENT 1 /* Values for e_ident[EI_CLASS]. */ #define ELFCLASSNONE 0 /* Unknown class. */ #define ELFCLASS32 1 /* 32-bit architecture. */ #define ELFCLASS64 2 /* 64-bit architecture. */ /* Values for e_ident[EI_DATA]. */ #define ELFDATANONE 0 /* Unknown data format. */ #define ELFDATA2LSB 1 /* 2's complement little-endian. */ #define ELFDATA2MSB 2 /* 2's complement big-endian. */ /* Values for e_ident[EI_OSABI]. */ #define ELFOSABI_NONE 0 /* UNIX System V ABI */ #define ELFOSABI_HPUX 1 /* HP-UX operating system */ #define ELFOSABI_NETBSD 2 /* NetBSD */ #define ELFOSABI_LINUX 3 /* GNU/Linux */ #define ELFOSABI_HURD 4 /* GNU/Hurd */ #define ELFOSABI_86OPEN 5 /* 86Open common IA32 ABI */ #define ELFOSABI_SOLARIS 6 /* Solaris */ #define ELFOSABI_AIX 7 /* AIX */ #define ELFOSABI_IRIX 8 /* IRIX */ #define ELFOSABI_FREEBSD 9 /* FreeBSD */ #define ELFOSABI_TRU64 10 /* TRU64 UNIX */ #define ELFOSABI_MODESTO 11 /* Novell Modesto */ #define ELFOSABI_OPENBSD 12 /* OpenBSD */ #define ELFOSABI_OPENVMS 13 /* Open VMS */ #define ELFOSABI_NSK 14 /* HP Non-Stop Kernel */ #define ELFOSABI_AROS 15 /* Amiga Research OS */ +#define ELFOSABI_FENIXOS 16 /* FenixOS */ +#define ELFOSABI_CLOUDABI 17 /* Nuxi CloudABI */ #define ELFOSABI_ARM 97 /* ARM */ #define ELFOSABI_STANDALONE 255 /* Standalone (embedded) application */ #define ELFOSABI_SYSV ELFOSABI_NONE /* symbol used in old spec */ #define ELFOSABI_MONTEREY ELFOSABI_AIX /* Monterey */ #define ELFOSABI_GNU ELFOSABI_LINUX /* e_ident */ #define IS_ELF(ehdr) ((ehdr).e_ident[EI_MAG0] == ELFMAG0 && \ (ehdr).e_ident[EI_MAG1] == ELFMAG1 && \ (ehdr).e_ident[EI_MAG2] == ELFMAG2 && \ (ehdr).e_ident[EI_MAG3] == ELFMAG3) /* Values for e_type. */ #define ET_NONE 0 /* Unknown type. */ #define ET_REL 1 /* Relocatable. */ #define ET_EXEC 2 /* Executable. */ #define ET_DYN 3 /* Shared object. */ #define ET_CORE 4 /* Core file. */ #define ET_LOOS 0xfe00 /* First operating system specific. */ #define ET_HIOS 0xfeff /* Last operating system-specific. */ #define ET_LOPROC 0xff00 /* First processor-specific. */ #define ET_HIPROC 0xffff /* Last processor-specific. */ /* Values for e_machine. */ #define EM_NONE 0 /* Unknown machine. */ #define EM_M32 1 /* AT&T WE32100. */ #define EM_SPARC 2 /* Sun SPARC. */ #define EM_386 3 /* Intel i386. */ #define EM_68K 4 /* Motorola 68000. */ #define EM_88K 5 /* Motorola 88000. */ #define EM_860 7 /* Intel i860. */ #define EM_MIPS 8 /* MIPS R3000 Big-Endian only. */ #define EM_S370 9 /* IBM System/370. */ #define EM_MIPS_RS3_LE 10 /* MIPS R3000 Little-Endian. */ #define EM_PARISC 15 /* HP PA-RISC. */ #define EM_VPP500 17 /* Fujitsu VPP500. */ #define EM_SPARC32PLUS 18 /* SPARC v8plus. */ #define EM_960 19 /* Intel 80960. */ #define EM_PPC 20 /* PowerPC 32-bit. */ #define EM_PPC64 21 /* PowerPC 64-bit. */ #define EM_S390 22 /* IBM System/390. */ #define EM_V800 36 /* NEC V800. */ #define EM_FR20 37 /* Fujitsu FR20. */ #define EM_RH32 38 /* TRW RH-32. */ #define EM_RCE 39 /* Motorola RCE. */ #define EM_ARM 40 /* ARM. */ #define EM_SH 42 /* Hitachi SH. */ #define EM_SPARCV9 43 /* SPARC v9 64-bit. */ #define EM_TRICORE 44 /* Siemens TriCore embedded processor. */ #define EM_ARC 45 /* Argonaut RISC Core. */ #define EM_H8_300 46 /* Hitachi H8/300. */ #define EM_H8_300H 47 /* Hitachi H8/300H. */ #define EM_H8S 48 /* Hitachi H8S. */ #define EM_H8_500 49 /* Hitachi H8/500. */ #define EM_IA_64 50 /* Intel IA-64 Processor. */ #define EM_MIPS_X 51 /* Stanford MIPS-X. */ #define EM_COLDFIRE 52 /* Motorola ColdFire. */ #define EM_68HC12 53 /* Motorola M68HC12. */ #define EM_MMA 54 /* Fujitsu MMA. */ #define EM_PCP 55 /* Siemens PCP. */ #define EM_NCPU 56 /* Sony nCPU. */ #define EM_NDR1 57 /* Denso NDR1 microprocessor. */ #define EM_STARCORE 58 /* Motorola Star*Core processor. */ #define EM_ME16 59 /* Toyota ME16 processor. */ #define EM_ST100 60 /* STMicroelectronics ST100 processor. */ #define EM_TINYJ 61 /* Advanced Logic Corp. TinyJ processor. */ #define EM_X86_64 62 /* Advanced Micro Devices x86-64 */ #define EM_AMD64 EM_X86_64 /* Advanced Micro Devices x86-64 (compat) */ #define EM_PDSP 63 /* Sony DSP Processor. */ #define EM_FX66 66 /* Siemens FX66 microcontroller. */ #define EM_ST9PLUS 67 /* STMicroelectronics ST9+ 8/16 microcontroller. */ #define EM_ST7 68 /* STmicroelectronics ST7 8-bit microcontroller. */ #define EM_68HC16 69 /* Motorola MC68HC16 microcontroller. */ #define EM_68HC11 70 /* Motorola MC68HC11 microcontroller. */ #define EM_68HC08 71 /* Motorola MC68HC08 microcontroller. */ #define EM_68HC05 72 /* Motorola MC68HC05 microcontroller. */ #define EM_SVX 73 /* Silicon Graphics SVx. */ #define EM_ST19 74 /* STMicroelectronics ST19 8-bit mc. */ #define EM_VAX 75 /* Digital VAX. */ #define EM_CRIS 76 /* Axis Communications 32-bit embedded processor. */ #define EM_JAVELIN 77 /* Infineon Technologies 32-bit embedded processor. */ #define EM_FIREPATH 78 /* Element 14 64-bit DSP Processor. */ #define EM_ZSP 79 /* LSI Logic 16-bit DSP Processor. */ #define EM_MMIX 80 /* Donald Knuth's educational 64-bit proc. */ #define EM_HUANY 81 /* Harvard University machine-independent object files. */ #define EM_PRISM 82 /* SiTera Prism. */ #define EM_AVR 83 /* Atmel AVR 8-bit microcontroller. */ #define EM_FR30 84 /* Fujitsu FR30. */ #define EM_D10V 85 /* Mitsubishi D10V. */ #define EM_D30V 86 /* Mitsubishi D30V. */ #define EM_V850 87 /* NEC v850. */ #define EM_M32R 88 /* Mitsubishi M32R. */ #define EM_MN10300 89 /* Matsushita MN10300. */ #define EM_MN10200 90 /* Matsushita MN10200. */ #define EM_PJ 91 /* picoJava. */ #define EM_OPENRISC 92 /* OpenRISC 32-bit embedded processor. */ #define EM_ARC_A5 93 /* ARC Cores Tangent-A5. */ #define EM_XTENSA 94 /* Tensilica Xtensa Architecture. */ #define EM_VIDEOCORE 95 /* Alphamosaic VideoCore processor. */ #define EM_TMM_GPP 96 /* Thompson Multimedia General Purpose Processor. */ #define EM_NS32K 97 /* National Semiconductor 32000 series. */ #define EM_TPC 98 /* Tenor Network TPC processor. */ #define EM_SNP1K 99 /* Trebia SNP 1000 processor. */ #define EM_ST200 100 /* STMicroelectronics ST200 microcontroller. */ #define EM_IP2K 101 /* Ubicom IP2xxx microcontroller family. */ #define EM_MAX 102 /* MAX Processor. */ #define EM_CR 103 /* National Semiconductor CompactRISC microprocessor. */ #define EM_F2MC16 104 /* Fujitsu F2MC16. */ #define EM_MSP430 105 /* Texas Instruments embedded microcontroller msp430. */ #define EM_BLACKFIN 106 /* Analog Devices Blackfin (DSP) processor. */ #define EM_SE_C33 107 /* S1C33 Family of Seiko Epson processors. */ #define EM_SEP 108 /* Sharp embedded microprocessor. */ #define EM_ARCA 109 /* Arca RISC Microprocessor. */ #define EM_UNICORE 110 /* Microprocessor series from PKU-Unity Ltd. and MPRC of Peking University */ #define EM_AARCH64 183 /* AArch64 (64-bit ARM) */ /* Non-standard or deprecated. */ #define EM_486 6 /* Intel i486. */ #define EM_MIPS_RS4_BE 10 /* MIPS R4000 Big-Endian */ #define EM_ALPHA_STD 41 /* Digital Alpha (standard value). */ #define EM_ALPHA 0x9026 /* Alpha (written in the absence of an ABI) */ /** * e_flags */ #define EF_ARM_RELEXEC 0x1 #define EF_ARM_HASENTRY 0x2 #define EF_ARM_SYMSARESORTED 0x4 #define EF_ARM_DYNSYMSUSESEGIDX 0x8 #define EF_ARM_MAPSYMSFIRST 0x10 #define EF_ARM_LE8 0x00400000 #define EF_ARM_BE8 0x00800000 #define EF_ARM_EABIMASK 0xFF000000 #define EF_ARM_EABI_UNKNOWN 0x00000000 #define EF_ARM_EABI_VER1 0x01000000 #define EF_ARM_EABI_VER2 0x02000000 #define EF_ARM_EABI_VER3 0x03000000 #define EF_ARM_EABI_VER4 0x04000000 #define EF_ARM_EABI_VER5 0x05000000 #define EF_ARM_INTERWORK 0x00000004 #define EF_ARM_APCS_26 0x00000008 #define EF_ARM_APCS_FLOAT 0x00000010 #define EF_ARM_PIC 0x00000020 #define EF_ARM_ALIGN8 0x00000040 #define EF_ARM_NEW_ABI 0x00000080 #define EF_ARM_OLD_ABI 0x00000100 #define EF_ARM_SOFT_FLOAT 0x00000200 #define EF_ARM_VFP_FLOAT 0x00000400 #define EF_ARM_MAVERICK_FLOAT 0x00000800 #define EF_MIPS_NOREORDER 0x00000001 #define EF_MIPS_PIC 0x00000002 /* Contains PIC code */ #define EF_MIPS_CPIC 0x00000004 /* STD PIC calling sequence */ #define EF_MIPS_UCODE 0x00000010 #define EF_MIPS_ABI2 0x00000020 /* N32 */ #define EF_MIPS_OPTIONS_FIRST 0x00000080 #define EF_MIPS_ARCH_ASE 0x0F000000 /* Architectural extensions */ #define EF_MIPS_ARCH_ASE_MDMX 0x08000000 /* MDMX multimedia extension */ #define EF_MIPS_ARCH_ASE_M16 0x04000000 /* MIPS-16 ISA extensions */ #define EF_MIPS_ARCH 0xF0000000 /* Architecture field */ #define EF_PPC_EMB 0x80000000 #define EF_PPC_RELOCATABLE 0x00010000 #define EF_PPC_RELOCATABLE_LIB 0x00008000 #define EF_SPARC_EXT_MASK 0x00ffff00 #define EF_SPARC_32PLUS 0x00000100 #define EF_SPARC_SUN_US1 0x00000200 #define EF_SPARC_HAL_R1 0x00000200 #define EF_SPARC_SUN_US3 0x00000800 #define EF_SPARCV9_MM 0x00000003 #define EF_SPARCV9_TSO 0x00000000 #define EF_SPARCV9_PSO 0x00000001 #define EF_SPARCV9_RMO 0x00000002 /* Special section indexes. */ #define SHN_UNDEF 0 /* Undefined, missing, irrelevant. */ #define SHN_LORESERVE 0xff00 /* First of reserved range. */ #define SHN_LOPROC 0xff00 /* First processor-specific. */ #define SHN_HIPROC 0xff1f /* Last processor-specific. */ #define SHN_LOOS 0xff20 /* First operating system-specific. */ #define SHN_HIOS 0xff3f /* Last operating system-specific. */ #define SHN_ABS 0xfff1 /* Absolute values. */ #define SHN_COMMON 0xfff2 /* Common data. */ #define SHN_XINDEX 0xffff /* Escape -- index stored elsewhere. */ #define SHN_HIRESERVE 0xffff /* Last of reserved range. */ /* sh_type */ #define SHT_NULL 0 /* inactive */ #define SHT_PROGBITS 1 /* program defined information */ #define SHT_SYMTAB 2 /* symbol table section */ #define SHT_STRTAB 3 /* string table section */ #define SHT_RELA 4 /* relocation section with addends */ #define SHT_HASH 5 /* symbol hash table section */ #define SHT_DYNAMIC 6 /* dynamic section */ #define SHT_NOTE 7 /* note section */ #define SHT_NOBITS 8 /* no space section */ #define SHT_REL 9 /* relocation section - no addends */ #define SHT_SHLIB 10 /* reserved - purpose unknown */ #define SHT_DYNSYM 11 /* dynamic symbol table section */ #define SHT_INIT_ARRAY 14 /* Initialization function pointers. */ #define SHT_FINI_ARRAY 15 /* Termination function pointers. */ #define SHT_PREINIT_ARRAY 16 /* Pre-initialization function ptrs. */ #define SHT_GROUP 17 /* Section group. */ #define SHT_SYMTAB_SHNDX 18 /* Section indexes (see SHN_XINDEX). */ #define SHT_LOOS 0x60000000 /* First of OS specific semantics */ #define SHT_LOSUNW 0x6ffffff4 #define SHT_SUNW_dof 0x6ffffff4 #define SHT_SUNW_cap 0x6ffffff5 #define SHT_GNU_ATTRIBUTES 0x6ffffff5 #define SHT_SUNW_SIGNATURE 0x6ffffff6 #define SHT_GNU_HASH 0x6ffffff6 #define SHT_GNU_LIBLIST 0x6ffffff7 #define SHT_SUNW_ANNOTATE 0x6ffffff7 #define SHT_SUNW_DEBUGSTR 0x6ffffff8 #define SHT_SUNW_DEBUG 0x6ffffff9 #define SHT_SUNW_move 0x6ffffffa #define SHT_SUNW_COMDAT 0x6ffffffb #define SHT_SUNW_syminfo 0x6ffffffc #define SHT_SUNW_verdef 0x6ffffffd #define SHT_GNU_verdef 0x6ffffffd /* Symbol versions provided */ #define SHT_SUNW_verneed 0x6ffffffe #define SHT_GNU_verneed 0x6ffffffe /* Symbol versions required */ #define SHT_SUNW_versym 0x6fffffff #define SHT_GNU_versym 0x6fffffff /* Symbol version table */ #define SHT_HISUNW 0x6fffffff #define SHT_HIOS 0x6fffffff /* Last of OS specific semantics */ #define SHT_LOPROC 0x70000000 /* reserved range for processor */ #define SHT_AMD64_UNWIND 0x70000001 /* unwind information */ #define SHT_ARM_EXIDX 0x70000001 /* Exception index table. */ #define SHT_ARM_PREEMPTMAP 0x70000002 /* BPABI DLL dynamic linking pre-emption map. */ #define SHT_ARM_ATTRIBUTES 0x70000003 /* Object file compatibility attributes. */ #define SHT_ARM_DEBUGOVERLAY 0x70000004 /* See DBGOVL for details. */ #define SHT_ARM_OVERLAYSECTION 0x70000005 /* See DBGOVL for details. */ #define SHT_MIPS_LIBLIST 0x70000000 #define SHT_MIPS_MSYM 0x70000001 #define SHT_MIPS_CONFLICT 0x70000002 #define SHT_MIPS_GPTAB 0x70000003 #define SHT_MIPS_UCODE 0x70000004 #define SHT_MIPS_DEBUG 0x70000005 #define SHT_MIPS_REGINFO 0x70000006 #define SHT_MIPS_PACKAGE 0x70000007 #define SHT_MIPS_PACKSYM 0x70000008 #define SHT_MIPS_RELD 0x70000009 #define SHT_MIPS_IFACE 0x7000000b #define SHT_MIPS_CONTENT 0x7000000c #define SHT_MIPS_OPTIONS 0x7000000d #define SHT_MIPS_DELTASYM 0x7000001b #define SHT_MIPS_DELTAINST 0x7000001c #define SHT_MIPS_DELTACLASS 0x7000001d #define SHT_MIPS_DWARF 0x7000001e /* MIPS gcc uses MIPS_DWARF */ #define SHT_MIPS_DELTADECL 0x7000001f #define SHT_MIPS_SYMBOL_LIB 0x70000020 #define SHT_MIPS_EVENTS 0x70000021 #define SHT_MIPS_TRANSLATE 0x70000022 #define SHT_MIPS_PIXIE 0x70000023 #define SHT_MIPS_XLATE 0x70000024 #define SHT_MIPS_XLATE_DEBUG 0x70000025 #define SHT_MIPS_WHIRL 0x70000026 #define SHT_MIPS_EH_REGION 0x70000027 #define SHT_MIPS_XLATE_OLD 0x70000028 #define SHT_MIPS_PDR_EXCEPTION 0x70000029 #define SHT_SPARC_GOTDATA 0x70000000 #define SHTORDERED #define SHT_HIPROC 0x7fffffff /* specific section header types */ #define SHT_LOUSER 0x80000000 /* reserved range for application */ #define SHT_HIUSER 0xffffffff /* specific indexes */ /* Flags for sh_flags. */ #define SHF_WRITE 0x1 /* Section contains writable data. */ #define SHF_ALLOC 0x2 /* Section occupies memory. */ #define SHF_EXECINSTR 0x4 /* Section contains instructions. */ #define SHF_MERGE 0x10 /* Section may be merged. */ #define SHF_STRINGS 0x20 /* Section contains strings. */ #define SHF_INFO_LINK 0x40 /* sh_info holds section index. */ #define SHF_LINK_ORDER 0x80 /* Special ordering requirements. */ #define SHF_OS_NONCONFORMING 0x100 /* OS-specific processing required. */ #define SHF_GROUP 0x200 /* Member of section group. */ #define SHF_TLS 0x400 /* Section contains TLS data. */ #define SHF_MASKOS 0x0ff00000 /* OS-specific semantics. */ #define SHF_MASKPROC 0xf0000000 /* Processor-specific semantics. */ /* Values for p_type. */ #define PT_NULL 0 /* Unused entry. */ #define PT_LOAD 1 /* Loadable segment. */ #define PT_DYNAMIC 2 /* Dynamic linking information segment. */ #define PT_INTERP 3 /* Pathname of interpreter. */ #define PT_NOTE 4 /* Auxiliary information. */ #define PT_SHLIB 5 /* Reserved (not used). */ #define PT_PHDR 6 /* Location of program header itself. */ #define PT_TLS 7 /* Thread local storage segment */ #define PT_LOOS 0x60000000 /* First OS-specific. */ #define PT_SUNW_UNWIND 0x6464e550 /* amd64 UNWIND program header */ #define PT_GNU_EH_FRAME 0x6474e550 #define PT_GNU_STACK 0x6474e551 #define PT_GNU_RELRO 0x6474e552 #define PT_DUMP_DELTA 0x6fb5d000 /* va->pa map for kernel dumps (currently arm). */ #define PT_LOSUNW 0x6ffffffa #define PT_SUNWBSS 0x6ffffffa /* Sun Specific segment */ #define PT_SUNWSTACK 0x6ffffffb /* describes the stack segment */ #define PT_SUNWDTRACE 0x6ffffffc /* private */ #define PT_SUNWCAP 0x6ffffffd /* hard/soft capabilities segment */ #define PT_HISUNW 0x6fffffff #define PT_HIOS 0x6fffffff /* Last OS-specific. */ #define PT_LOPROC 0x70000000 /* First processor-specific type. */ #define PT_HIPROC 0x7fffffff /* Last processor-specific type. */ /* Values for p_flags. */ #define PF_X 0x1 /* Executable. */ #define PF_W 0x2 /* Writable. */ #define PF_R 0x4 /* Readable. */ #define PF_MASKOS 0x0ff00000 /* Operating system-specific. */ #define PF_MASKPROC 0xf0000000 /* Processor-specific. */ /* Extended program header index. */ #define PN_XNUM 0xffff /* Values for d_tag. */ #define DT_NULL 0 /* Terminating entry. */ #define DT_NEEDED 1 /* String table offset of a needed shared library. */ #define DT_PLTRELSZ 2 /* Total size in bytes of PLT relocations. */ #define DT_PLTGOT 3 /* Processor-dependent address. */ #define DT_HASH 4 /* Address of symbol hash table. */ #define DT_STRTAB 5 /* Address of string table. */ #define DT_SYMTAB 6 /* Address of symbol table. */ #define DT_RELA 7 /* Address of ElfNN_Rela relocations. */ #define DT_RELASZ 8 /* Total size of ElfNN_Rela relocations. */ #define DT_RELAENT 9 /* Size of each ElfNN_Rela relocation entry. */ #define DT_STRSZ 10 /* Size of string table. */ #define DT_SYMENT 11 /* Size of each symbol table entry. */ #define DT_INIT 12 /* Address of initialization function. */ #define DT_FINI 13 /* Address of finalization function. */ #define DT_SONAME 14 /* String table offset of shared object name. */ #define DT_RPATH 15 /* String table offset of library path. [sup] */ #define DT_SYMBOLIC 16 /* Indicates "symbolic" linking. [sup] */ #define DT_REL 17 /* Address of ElfNN_Rel relocations. */ #define DT_RELSZ 18 /* Total size of ElfNN_Rel relocations. */ #define DT_RELENT 19 /* Size of each ElfNN_Rel relocation. */ #define DT_PLTREL 20 /* Type of relocation used for PLT. */ #define DT_DEBUG 21 /* Reserved (not used). */ #define DT_TEXTREL 22 /* Indicates there may be relocations in non-writable segments. [sup] */ #define DT_JMPREL 23 /* Address of PLT relocations. */ #define DT_BIND_NOW 24 /* [sup] */ #define DT_INIT_ARRAY 25 /* Address of the array of pointers to initialization functions */ #define DT_FINI_ARRAY 26 /* Address of the array of pointers to termination functions */ #define DT_INIT_ARRAYSZ 27 /* Size in bytes of the array of initialization functions. */ #define DT_FINI_ARRAYSZ 28 /* Size in bytes of the array of termination functions. */ #define DT_RUNPATH 29 /* String table offset of a null-terminated library search path string. */ #define DT_FLAGS 30 /* Object specific flag values. */ #define DT_ENCODING 32 /* Values greater than or equal to DT_ENCODING and less than DT_LOOS follow the rules for the interpretation of the d_un union as follows: even == 'd_ptr', odd == 'd_val' or none */ #define DT_PREINIT_ARRAY 32 /* Address of the array of pointers to pre-initialization functions. */ #define DT_PREINIT_ARRAYSZ 33 /* Size in bytes of the array of pre-initialization functions. */ #define DT_MAXPOSTAGS 34 /* number of positive tags */ #define DT_LOOS 0x6000000d /* First OS-specific */ #define DT_SUNW_AUXILIARY 0x6000000d /* symbol auxiliary name */ #define DT_SUNW_RTLDINF 0x6000000e /* ld.so.1 info (private) */ #define DT_SUNW_FILTER 0x6000000f /* symbol filter name */ #define DT_SUNW_CAP 0x60000010 /* hardware/software */ #define DT_HIOS 0x6ffff000 /* Last OS-specific */ /* * DT_* entries which fall between DT_VALRNGHI & DT_VALRNGLO use the * Dyn.d_un.d_val field of the Elf*_Dyn structure. */ #define DT_VALRNGLO 0x6ffffd00 #define DT_GNU_PRELINKED 0x6ffffdf5 /* prelinking timestamp */ #define DT_GNU_CONFLICTSZ 0x6ffffdf6 /* size of conflict section */ #define DT_GNU_LIBLISTSZ 0x6ffffdf7 /* size of library list */ #define DT_CHECKSUM 0x6ffffdf8 /* elf checksum */ #define DT_PLTPADSZ 0x6ffffdf9 /* pltpadding size */ #define DT_MOVEENT 0x6ffffdfa /* move table entry size */ #define DT_MOVESZ 0x6ffffdfb /* move table size */ #define DT_FEATURE 0x6ffffdfc /* feature holder */ #define DT_FEATURE_1 DT_FEATURE #define DT_POSFLAG_1 0x6ffffdfd /* flags for DT_* entries, effecting */ /* the following DT_* entry. */ /* See DF_P1_* definitions */ #define DT_SYMINSZ 0x6ffffdfe /* syminfo table size (in bytes) */ #define DT_SYMINENT 0x6ffffdff /* syminfo entry size (in bytes) */ #define DT_VALRNGHI 0x6ffffdff /* * DT_* entries which fall between DT_ADDRRNGHI & DT_ADDRRNGLO use the * Dyn.d_un.d_ptr field of the Elf*_Dyn structure. * * If any adjustment is made to the ELF object after it has been * built, these entries will need to be adjusted. */ #define DT_ADDRRNGLO 0x6ffffe00 #define DT_GNU_HASH 0x6ffffef5 /* GNU-style hash table */ #define DT_GNU_CONFLICT 0x6ffffef8 /* address of conflict section */ #define DT_GNU_LIBLIST 0x6ffffef9 /* address of library list */ #define DT_CONFIG 0x6ffffefa /* configuration information */ #define DT_DEPAUDIT 0x6ffffefb /* dependency auditing */ #define DT_AUDIT 0x6ffffefc /* object auditing */ #define DT_PLTPAD 0x6ffffefd /* pltpadding (sparcv9) */ #define DT_MOVETAB 0x6ffffefe /* move table */ #define DT_SYMINFO 0x6ffffeff /* syminfo table */ #define DT_ADDRRNGHI 0x6ffffeff #define DT_VERSYM 0x6ffffff0 /* Address of versym section. */ #define DT_RELACOUNT 0x6ffffff9 /* number of RELATIVE relocations */ #define DT_RELCOUNT 0x6ffffffa /* number of RELATIVE relocations */ #define DT_FLAGS_1 0x6ffffffb /* state flags - see DF_1_* defs */ #define DT_VERDEF 0x6ffffffc /* Address of verdef section. */ #define DT_VERDEFNUM 0x6ffffffd /* Number of elems in verdef section */ #define DT_VERNEED 0x6ffffffe /* Address of verneed section. */ #define DT_VERNEEDNUM 0x6fffffff /* Number of elems in verneed section */ #define DT_LOPROC 0x70000000 /* First processor-specific type. */ #define DT_ARM_SYMTABSZ 0x70000001 #define DT_ARM_PREEMPTMAP 0x70000002 #define DT_SPARC_REGISTER 0x70000001 #define DT_DEPRECATED_SPARC_REGISTER 0x7000001 #define DT_MIPS_RLD_VERSION 0x70000001 #define DT_MIPS_TIME_STAMP 0x70000002 #define DT_MIPS_ICHECKSUM 0x70000003 #define DT_MIPS_IVERSION 0x70000004 #define DT_MIPS_FLAGS 0x70000005 #define DT_MIPS_BASE_ADDRESS 0x70000006 #define DT_MIPS_CONFLICT 0x70000008 #define DT_MIPS_LIBLIST 0x70000009 #define DT_MIPS_LOCAL_GOTNO 0x7000000a #define DT_MIPS_CONFLICTNO 0x7000000b #define DT_MIPS_LIBLISTNO 0x70000010 #define DT_MIPS_SYMTABNO 0x70000011 #define DT_MIPS_UNREFEXTNO 0x70000012 #define DT_MIPS_GOTSYM 0x70000013 #define DT_MIPS_HIPAGENO 0x70000014 #define DT_MIPS_RLD_MAP 0x70000016 #define DT_MIPS_DELTA_CLASS 0x70000017 #define DT_MIPS_DELTA_CLASS_NO 0x70000018 #define DT_MIPS_DELTA_INSTANCE 0x70000019 #define DT_MIPS_DELTA_INSTANCE_NO 0x7000001A #define DT_MIPS_DELTA_RELOC 0x7000001B #define DT_MIPS_DELTA_RELOC_NO 0x7000001C #define DT_MIPS_DELTA_SYM 0x7000001D #define DT_MIPS_DELTA_SYM_NO 0x7000001E #define DT_MIPS_DELTA_CLASSSYM 0x70000020 #define DT_MIPS_DELTA_CLASSSYM_NO 0x70000021 #define DT_MIPS_CXX_FLAGS 0x70000022 #define DT_MIPS_PIXIE_INIT 0x70000023 #define DT_MIPS_SYMBOL_LIB 0x70000024 #define DT_MIPS_LOCALPAGE_GOTIDX 0x70000025 #define DT_MIPS_LOCAL_GOTIDX 0x70000026 #define DT_MIPS_HIDDEN_GOTIDX 0x70000027 #define DT_MIPS_PROTECTED_GOTIDX 0x70000028 #define DT_MIPS_OPTIONS 0x70000029 #define DT_MIPS_INTERFACE 0x7000002A #define DT_MIPS_DYNSTR_ALIGN 0x7000002B #define DT_MIPS_INTERFACE_SIZE 0x7000002C #define DT_MIPS_RLD_TEXT_RESOLVE_ADDR 0x7000002D #define DT_MIPS_PERF_SUFFIX 0x7000002E #define DT_MIPS_COMPACT_SIZE 0x7000002F #define DT_MIPS_GP_VALUE 0x70000030 #define DT_MIPS_AUX_DYNAMIC 0x70000031 #define DT_MIPS_PLTGOT 0x70000032 #define DT_MIPS_RLD_OBJ_UPDATE 0x70000033 #define DT_MIPS_RWPLT 0x70000034 #define DT_PPC_GOT 0x70000000 #define DT_PPC_TLSOPT 0x70000001 #define DT_PPC64_GLINK 0x70000000 #define DT_PPC64_OPD 0x70000001 #define DT_PPC64_OPDSZ 0x70000002 #define DT_PPC64_TLSOPT 0x70000003 #define DT_AUXILIARY 0x7ffffffd /* shared library auxiliary name */ #define DT_USED 0x7ffffffe /* ignored - same as needed */ #define DT_FILTER 0x7fffffff /* shared library filter name */ #define DT_HIPROC 0x7fffffff /* Last processor-specific type. */ /* Values for DT_FLAGS */ #define DF_ORIGIN 0x0001 /* Indicates that the object being loaded may make reference to the $ORIGIN substitution string */ #define DF_SYMBOLIC 0x0002 /* Indicates "symbolic" linking. */ #define DF_TEXTREL 0x0004 /* Indicates there may be relocations in non-writable segments. */ #define DF_BIND_NOW 0x0008 /* Indicates that the dynamic linker should process all relocations for the object containing this entry before transferring control to the program. */ #define DF_STATIC_TLS 0x0010 /* Indicates that the shared object or executable contains code using a static thread-local storage scheme. */ /* Values for DT_FLAGS_1 */ #define DF_1_BIND_NOW 0x00000001 /* Same as DF_BIND_NOW */ #define DF_1_GLOBAL 0x00000002 /* Set the RTLD_GLOBAL for object */ #define DF_1_NODELETE 0x00000008 /* Set the RTLD_NODELETE for object */ #define DF_1_LOADFLTR 0x00000010 /* Immediate loading of filtees */ #define DF_1_NOOPEN 0x00000040 /* Do not allow loading on dlopen() */ #define DF_1_ORIGIN 0x00000080 /* Process $ORIGIN */ #define DF_1_INTERPOSE 0x00000400 /* Interpose all objects but main */ #define DF_1_NODEFLIB 0x00000800 /* Do not search default paths */ /* Values for l_flags. */ #define LL_NONE 0x0 /* no flags */ #define LL_EXACT_MATCH 0x1 /* require an exact match */ #define LL_IGNORE_INT_VER 0x2 /* ignore version incompatibilities */ #define LL_REQUIRE_MINOR 0x4 #define LL_EXPORTS 0x8 #define LL_DELAY_LOAD 0x10 #define LL_DELTA 0x20 /* Values for n_type. Used in core files. */ #define NT_PRSTATUS 1 /* Process status. */ #define NT_FPREGSET 2 /* Floating point registers. */ #define NT_PRPSINFO 3 /* Process state info. */ #define NT_THRMISC 7 /* Thread miscellaneous info. */ #define NT_PROCSTAT_PROC 8 /* Procstat proc data. */ #define NT_PROCSTAT_FILES 9 /* Procstat files data. */ #define NT_PROCSTAT_VMMAP 10 /* Procstat vmmap data. */ #define NT_PROCSTAT_GROUPS 11 /* Procstat groups data. */ #define NT_PROCSTAT_UMASK 12 /* Procstat umask data. */ #define NT_PROCSTAT_RLIMIT 13 /* Procstat rlimit data. */ #define NT_PROCSTAT_OSREL 14 /* Procstat osreldate data. */ #define NT_PROCSTAT_PSSTRINGS 15 /* Procstat ps_strings data. */ #define NT_PROCSTAT_AUXV 16 /* Procstat auxv data. */ #define NT_PPC_VMX 0x100 /* PowerPC Altivec/VMX registers */ #define NT_X86_XSTATE 0x202 /* x86 XSAVE extended state. */ /* Symbol Binding - ELFNN_ST_BIND - st_info */ #define STB_LOCAL 0 /* Local symbol */ #define STB_GLOBAL 1 /* Global symbol */ #define STB_WEAK 2 /* like global - lower precedence */ #define STB_LOOS 10 /* Reserved range for operating system */ #define STB_HIOS 12 /* specific semantics. */ #define STB_LOPROC 13 /* reserved range for processor */ #define STB_HIPROC 15 /* specific semantics. */ /* Symbol type - ELFNN_ST_TYPE - st_info */ #define STT_NOTYPE 0 /* Unspecified type. */ #define STT_OBJECT 1 /* Data object. */ #define STT_FUNC 2 /* Function. */ #define STT_SECTION 3 /* Section. */ #define STT_FILE 4 /* Source file. */ #define STT_COMMON 5 /* Uninitialized common block. */ #define STT_TLS 6 /* TLS object. */ #define STT_NUM 7 #define STT_LOOS 10 /* Reserved range for operating system */ #define STT_GNU_IFUNC 10 #define STT_HIOS 12 /* specific semantics. */ #define STT_LOPROC 13 /* reserved range for processor */ #define STT_HIPROC 15 /* specific semantics. */ /* Symbol visibility - ELFNN_ST_VISIBILITY - st_other */ #define STV_DEFAULT 0x0 /* Default visibility (see binding). */ #define STV_INTERNAL 0x1 /* Special meaning in relocatable objects. */ #define STV_HIDDEN 0x2 /* Not visible. */ #define STV_PROTECTED 0x3 /* Visible but not preemptible. */ #define STV_EXPORTED 0x4 #define STV_SINGLETON 0x5 #define STV_ELIMINATE 0x6 /* Special symbol table indexes. */ #define STN_UNDEF 0 /* Undefined symbol index. */ /* Symbol versioning flags. */ #define VER_DEF_CURRENT 1 #define VER_DEF_IDX(x) VER_NDX(x) #define VER_FLG_BASE 0x01 #define VER_FLG_WEAK 0x02 #define VER_NEED_CURRENT 1 #define VER_NEED_WEAK (1u << 15) #define VER_NEED_HIDDEN VER_NDX_HIDDEN #define VER_NEED_IDX(x) VER_NDX(x) #define VER_NDX_LOCAL 0 #define VER_NDX_GLOBAL 1 #define VER_NDX_GIVEN 2 #define VER_NDX_HIDDEN (1u << 15) #define VER_NDX(x) ((x) & ~(1u << 15)) #define CA_SUNW_NULL 0 #define CA_SUNW_HW_1 1 /* first hardware capabilities entry */ #define CA_SUNW_SF_1 2 /* first software capabilities entry */ /* * Syminfo flag values */ #define SYMINFO_FLG_DIRECT 0x0001 /* symbol ref has direct association */ /* to object containing defn. */ #define SYMINFO_FLG_PASSTHRU 0x0002 /* ignored - see SYMINFO_FLG_FILTER */ #define SYMINFO_FLG_COPY 0x0004 /* symbol is a copy-reloc */ #define SYMINFO_FLG_LAZYLOAD 0x0008 /* object containing defn should be */ /* lazily-loaded */ #define SYMINFO_FLG_DIRECTBIND 0x0010 /* ref should be bound directly to */ /* object containing defn. */ #define SYMINFO_FLG_NOEXTDIRECT 0x0020 /* don't let an external reference */ /* directly bind to this symbol */ #define SYMINFO_FLG_FILTER 0x0002 /* symbol ref is associated to a */ #define SYMINFO_FLG_AUXILIARY 0x0040 /* standard or auxiliary filter */ /* * Syminfo.si_boundto values. */ #define SYMINFO_BT_SELF 0xffff /* symbol bound to self */ #define SYMINFO_BT_PARENT 0xfffe /* symbol bound to parent */ #define SYMINFO_BT_NONE 0xfffd /* no special symbol binding */ #define SYMINFO_BT_EXTERN 0xfffc /* symbol defined as external */ #define SYMINFO_BT_LOWRESERVE 0xff00 /* beginning of reserved entries */ /* * Syminfo version values. */ #define SYMINFO_NONE 0 /* Syminfo version */ #define SYMINFO_CURRENT 1 #define SYMINFO_NUM 2 /* * Relocation types. * * All machine architectures are defined here to allow tools on one to * handle others. */ #define R_386_NONE 0 /* No relocation. */ #define R_386_32 1 /* Add symbol value. */ #define R_386_PC32 2 /* Add PC-relative symbol value. */ #define R_386_GOT32 3 /* Add PC-relative GOT offset. */ #define R_386_PLT32 4 /* Add PC-relative PLT offset. */ #define R_386_COPY 5 /* Copy data from shared object. */ #define R_386_GLOB_DAT 6 /* Set GOT entry to data address. */ #define R_386_JMP_SLOT 7 /* Set GOT entry to code address. */ #define R_386_RELATIVE 8 /* Add load address of shared object. */ #define R_386_GOTOFF 9 /* Add GOT-relative symbol address. */ #define R_386_GOTPC 10 /* Add PC-relative GOT table address. */ #define R_386_TLS_TPOFF 14 /* Negative offset in static TLS block */ #define R_386_TLS_IE 15 /* Absolute address of GOT for -ve static TLS */ #define R_386_TLS_GOTIE 16 /* GOT entry for negative static TLS block */ #define R_386_TLS_LE 17 /* Negative offset relative to static TLS */ #define R_386_TLS_GD 18 /* 32 bit offset to GOT (index,off) pair */ #define R_386_TLS_LDM 19 /* 32 bit offset to GOT (index,zero) pair */ #define R_386_TLS_GD_32 24 /* 32 bit offset to GOT (index,off) pair */ #define R_386_TLS_GD_PUSH 25 /* pushl instruction for Sun ABI GD sequence */ #define R_386_TLS_GD_CALL 26 /* call instruction for Sun ABI GD sequence */ #define R_386_TLS_GD_POP 27 /* popl instruction for Sun ABI GD sequence */ #define R_386_TLS_LDM_32 28 /* 32 bit offset to GOT (index,zero) pair */ #define R_386_TLS_LDM_PUSH 29 /* pushl instruction for Sun ABI LD sequence */ #define R_386_TLS_LDM_CALL 30 /* call instruction for Sun ABI LD sequence */ #define R_386_TLS_LDM_POP 31 /* popl instruction for Sun ABI LD sequence */ #define R_386_TLS_LDO_32 32 /* 32 bit offset from start of TLS block */ #define R_386_TLS_IE_32 33 /* 32 bit offset to GOT static TLS offset entry */ #define R_386_TLS_LE_32 34 /* 32 bit offset within static TLS block */ #define R_386_TLS_DTPMOD32 35 /* GOT entry containing TLS index */ #define R_386_TLS_DTPOFF32 36 /* GOT entry containing TLS offset */ #define R_386_TLS_TPOFF32 37 /* GOT entry of -ve static TLS offset */ #define R_386_IRELATIVE 42 /* PLT entry resolved indirectly at runtime */ #define R_AARCH64_ABS64 257 /* Absolute offset */ #define R_AARCH64_ABS32 258 /* Absolute, 32-bit overflow check */ #define R_AARCH64_ABS16 259 /* Absolute, 16-bit overflow check */ #define R_AARCH64_PREL64 260 /* PC relative */ #define R_AARCH64_PREL32 261 /* PC relative, 32-bit overflow check */ #define R_AARCH64_PREL16 262 /* PC relative, 16-bit overflow check */ #define R_AARCH64_COPY 1024 /* Copy data from shared object */ #define R_AARCH64_GLOB_DAT 1025 /* Set GOT entry to data address */ #define R_AARCH64_JUMP_SLOT 1026 /* Set GOT entry to code address */ #define R_AARCH64_RELATIVE 1027 /* Add load address of shared object */ #define R_AARCH64_TLSDESC 1031 /* Identify the TLS descriptor */ #define R_ARM_NONE 0 /* No relocation. */ #define R_ARM_PC24 1 #define R_ARM_ABS32 2 #define R_ARM_REL32 3 #define R_ARM_PC13 4 #define R_ARM_ABS16 5 #define R_ARM_ABS12 6 #define R_ARM_THM_ABS5 7 #define R_ARM_ABS8 8 #define R_ARM_SBREL32 9 #define R_ARM_THM_PC22 10 #define R_ARM_THM_PC8 11 #define R_ARM_AMP_VCALL9 12 #define R_ARM_SWI24 13 #define R_ARM_THM_SWI8 14 #define R_ARM_XPC25 15 #define R_ARM_THM_XPC22 16 /* TLS relocations */ #define R_ARM_TLS_DTPMOD32 17 /* ID of module containing symbol */ #define R_ARM_TLS_DTPOFF32 18 /* Offset in TLS block */ #define R_ARM_TLS_TPOFF32 19 /* Offset in static TLS block */ #define R_ARM_COPY 20 /* Copy data from shared object. */ #define R_ARM_GLOB_DAT 21 /* Set GOT entry to data address. */ #define R_ARM_JUMP_SLOT 22 /* Set GOT entry to code address. */ #define R_ARM_RELATIVE 23 /* Add load address of shared object. */ #define R_ARM_GOTOFF 24 /* Add GOT-relative symbol address. */ #define R_ARM_GOTPC 25 /* Add PC-relative GOT table address. */ #define R_ARM_GOT32 26 /* Add PC-relative GOT offset. */ #define R_ARM_PLT32 27 /* Add PC-relative PLT offset. */ #define R_ARM_GNU_VTENTRY 100 #define R_ARM_GNU_VTINHERIT 101 #define R_ARM_RSBREL32 250 #define R_ARM_THM_RPC22 251 #define R_ARM_RREL32 252 #define R_ARM_RABS32 253 #define R_ARM_RPC24 254 #define R_ARM_RBASE 255 /* Name Value Field Calculation */ #define R_IA_64_NONE 0 /* None */ #define R_IA_64_IMM14 0x21 /* immediate14 S + A */ #define R_IA_64_IMM22 0x22 /* immediate22 S + A */ #define R_IA_64_IMM64 0x23 /* immediate64 S + A */ #define R_IA_64_DIR32MSB 0x24 /* word32 MSB S + A */ #define R_IA_64_DIR32LSB 0x25 /* word32 LSB S + A */ #define R_IA_64_DIR64MSB 0x26 /* word64 MSB S + A */ #define R_IA_64_DIR64LSB 0x27 /* word64 LSB S + A */ #define R_IA_64_GPREL22 0x2a /* immediate22 @gprel(S + A) */ #define R_IA_64_GPREL64I 0x2b /* immediate64 @gprel(S + A) */ #define R_IA_64_GPREL32MSB 0x2c /* word32 MSB @gprel(S + A) */ #define R_IA_64_GPREL32LSB 0x2d /* word32 LSB @gprel(S + A) */ #define R_IA_64_GPREL64MSB 0x2e /* word64 MSB @gprel(S + A) */ #define R_IA_64_GPREL64LSB 0x2f /* word64 LSB @gprel(S + A) */ #define R_IA_64_LTOFF22 0x32 /* immediate22 @ltoff(S + A) */ #define R_IA_64_LTOFF64I 0x33 /* immediate64 @ltoff(S + A) */ #define R_IA_64_PLTOFF22 0x3a /* immediate22 @pltoff(S + A) */ #define R_IA_64_PLTOFF64I 0x3b /* immediate64 @pltoff(S + A) */ #define R_IA_64_PLTOFF64MSB 0x3e /* word64 MSB @pltoff(S + A) */ #define R_IA_64_PLTOFF64LSB 0x3f /* word64 LSB @pltoff(S + A) */ #define R_IA_64_FPTR64I 0x43 /* immediate64 @fptr(S + A) */ #define R_IA_64_FPTR32MSB 0x44 /* word32 MSB @fptr(S + A) */ #define R_IA_64_FPTR32LSB 0x45 /* word32 LSB @fptr(S + A) */ #define R_IA_64_FPTR64MSB 0x46 /* word64 MSB @fptr(S + A) */ #define R_IA_64_FPTR64LSB 0x47 /* word64 LSB @fptr(S + A) */ #define R_IA_64_PCREL60B 0x48 /* immediate60 form1 S + A - P */ #define R_IA_64_PCREL21B 0x49 /* immediate21 form1 S + A - P */ #define R_IA_64_PCREL21M 0x4a /* immediate21 form2 S + A - P */ #define R_IA_64_PCREL21F 0x4b /* immediate21 form3 S + A - P */ #define R_IA_64_PCREL32MSB 0x4c /* word32 MSB S + A - P */ #define R_IA_64_PCREL32LSB 0x4d /* word32 LSB S + A - P */ #define R_IA_64_PCREL64MSB 0x4e /* word64 MSB S + A - P */ #define R_IA_64_PCREL64LSB 0x4f /* word64 LSB S + A - P */ #define R_IA_64_LTOFF_FPTR22 0x52 /* immediate22 @ltoff(@fptr(S + A)) */ #define R_IA_64_LTOFF_FPTR64I 0x53 /* immediate64 @ltoff(@fptr(S + A)) */ #define R_IA_64_LTOFF_FPTR32MSB 0x54 /* word32 MSB @ltoff(@fptr(S + A)) */ #define R_IA_64_LTOFF_FPTR32LSB 0x55 /* word32 LSB @ltoff(@fptr(S + A)) */ #define R_IA_64_LTOFF_FPTR64MSB 0x56 /* word64 MSB @ltoff(@fptr(S + A)) */ #define R_IA_64_LTOFF_FPTR64LSB 0x57 /* word64 LSB @ltoff(@fptr(S + A)) */ #define R_IA_64_SEGREL32MSB 0x5c /* word32 MSB @segrel(S + A) */ #define R_IA_64_SEGREL32LSB 0x5d /* word32 LSB @segrel(S + A) */ #define R_IA_64_SEGREL64MSB 0x5e /* word64 MSB @segrel(S + A) */ #define R_IA_64_SEGREL64LSB 0x5f /* word64 LSB @segrel(S + A) */ #define R_IA_64_SECREL32MSB 0x64 /* word32 MSB @secrel(S + A) */ #define R_IA_64_SECREL32LSB 0x65 /* word32 LSB @secrel(S + A) */ #define R_IA_64_SECREL64MSB 0x66 /* word64 MSB @secrel(S + A) */ #define R_IA_64_SECREL64LSB 0x67 /* word64 LSB @secrel(S + A) */ #define R_IA_64_REL32MSB 0x6c /* word32 MSB BD + A */ #define R_IA_64_REL32LSB 0x6d /* word32 LSB BD + A */ #define R_IA_64_REL64MSB 0x6e /* word64 MSB BD + A */ #define R_IA_64_REL64LSB 0x6f /* word64 LSB BD + A */ #define R_IA_64_LTV32MSB 0x74 /* word32 MSB S + A */ #define R_IA_64_LTV32LSB 0x75 /* word32 LSB S + A */ #define R_IA_64_LTV64MSB 0x76 /* word64 MSB S + A */ #define R_IA_64_LTV64LSB 0x77 /* word64 LSB S + A */ #define R_IA_64_PCREL21BI 0x79 /* immediate21 form1 S + A - P */ #define R_IA_64_PCREL22 0x7a /* immediate22 S + A - P */ #define R_IA_64_PCREL64I 0x7b /* immediate64 S + A - P */ #define R_IA_64_IPLTMSB 0x80 /* function descriptor MSB special */ #define R_IA_64_IPLTLSB 0x81 /* function descriptor LSB speciaal */ #define R_IA_64_SUB 0x85 /* immediate64 A - S */ #define R_IA_64_LTOFF22X 0x86 /* immediate22 special */ #define R_IA_64_LDXMOV 0x87 /* immediate22 special */ #define R_IA_64_TPREL14 0x91 /* imm14 @tprel(S + A) */ #define R_IA_64_TPREL22 0x92 /* imm22 @tprel(S + A) */ #define R_IA_64_TPREL64I 0x93 /* imm64 @tprel(S + A) */ #define R_IA_64_TPREL64MSB 0x96 /* word64 MSB @tprel(S + A) */ #define R_IA_64_TPREL64LSB 0x97 /* word64 LSB @tprel(S + A) */ #define R_IA_64_LTOFF_TPREL22 0x9a /* imm22 @ltoff(@tprel(S+A)) */ #define R_IA_64_DTPMOD64MSB 0xa6 /* word64 MSB @dtpmod(S + A) */ #define R_IA_64_DTPMOD64LSB 0xa7 /* word64 LSB @dtpmod(S + A) */ #define R_IA_64_LTOFF_DTPMOD22 0xaa /* imm22 @ltoff(@dtpmod(S+A)) */ #define R_IA_64_DTPREL14 0xb1 /* imm14 @dtprel(S + A) */ #define R_IA_64_DTPREL22 0xb2 /* imm22 @dtprel(S + A) */ #define R_IA_64_DTPREL64I 0xb3 /* imm64 @dtprel(S + A) */ #define R_IA_64_DTPREL32MSB 0xb4 /* word32 MSB @dtprel(S + A) */ #define R_IA_64_DTPREL32LSB 0xb5 /* word32 LSB @dtprel(S + A) */ #define R_IA_64_DTPREL64MSB 0xb6 /* word64 MSB @dtprel(S + A) */ #define R_IA_64_DTPREL64LSB 0xb7 /* word64 LSB @dtprel(S + A) */ #define R_IA_64_LTOFF_DTPREL22 0xba /* imm22 @ltoff(@dtprel(S+A)) */ #define R_MIPS_NONE 0 /* No reloc */ #define R_MIPS_16 1 /* Direct 16 bit */ #define R_MIPS_32 2 /* Direct 32 bit */ #define R_MIPS_REL32 3 /* PC relative 32 bit */ #define R_MIPS_26 4 /* Direct 26 bit shifted */ #define R_MIPS_HI16 5 /* High 16 bit */ #define R_MIPS_LO16 6 /* Low 16 bit */ #define R_MIPS_GPREL16 7 /* GP relative 16 bit */ #define R_MIPS_LITERAL 8 /* 16 bit literal entry */ #define R_MIPS_GOT16 9 /* 16 bit GOT entry */ #define R_MIPS_PC16 10 /* PC relative 16 bit */ #define R_MIPS_CALL16 11 /* 16 bit GOT entry for function */ #define R_MIPS_GPREL32 12 /* GP relative 32 bit */ #define R_MIPS_64 18 /* Direct 64 bit */ #define R_MIPS_GOTHI16 21 /* GOT HI 16 bit */ #define R_MIPS_GOTLO16 22 /* GOT LO 16 bit */ #define R_MIPS_CALLHI16 30 /* upper 16 bit GOT entry for function */ #define R_MIPS_CALLLO16 31 /* lower 16 bit GOT entry for function */ #define R_PPC_NONE 0 /* No relocation. */ #define R_PPC_ADDR32 1 #define R_PPC_ADDR24 2 #define R_PPC_ADDR16 3 #define R_PPC_ADDR16_LO 4 #define R_PPC_ADDR16_HI 5 #define R_PPC_ADDR16_HA 6 #define R_PPC_ADDR14 7 #define R_PPC_ADDR14_BRTAKEN 8 #define R_PPC_ADDR14_BRNTAKEN 9 #define R_PPC_REL24 10 #define R_PPC_REL14 11 #define R_PPC_REL14_BRTAKEN 12 #define R_PPC_REL14_BRNTAKEN 13 #define R_PPC_GOT16 14 #define R_PPC_GOT16_LO 15 #define R_PPC_GOT16_HI 16 #define R_PPC_GOT16_HA 17 #define R_PPC_PLTREL24 18 #define R_PPC_COPY 19 #define R_PPC_GLOB_DAT 20 #define R_PPC_JMP_SLOT 21 #define R_PPC_RELATIVE 22 #define R_PPC_LOCAL24PC 23 #define R_PPC_UADDR32 24 #define R_PPC_UADDR16 25 #define R_PPC_REL32 26 #define R_PPC_PLT32 27 #define R_PPC_PLTREL32 28 #define R_PPC_PLT16_LO 29 #define R_PPC_PLT16_HI 30 #define R_PPC_PLT16_HA 31 #define R_PPC_SDAREL16 32 #define R_PPC_SECTOFF 33 #define R_PPC_SECTOFF_LO 34 #define R_PPC_SECTOFF_HI 35 #define R_PPC_SECTOFF_HA 36 /* * 64-bit relocations */ #define R_PPC64_ADDR64 38 #define R_PPC64_ADDR16_HIGHER 39 #define R_PPC64_ADDR16_HIGHERA 40 #define R_PPC64_ADDR16_HIGHEST 41 #define R_PPC64_ADDR16_HIGHESTA 42 #define R_PPC64_UADDR64 43 #define R_PPC64_REL64 44 #define R_PPC64_PLT64 45 #define R_PPC64_PLTREL64 46 #define R_PPC64_TOC16 47 #define R_PPC64_TOC16_LO 48 #define R_PPC64_TOC16_HI 49 #define R_PPC64_TOC16_HA 50 #define R_PPC64_TOC 51 #define R_PPC64_DTPMOD64 68 #define R_PPC64_TPREL64 73 #define R_PPC64_DTPREL64 78 /* * TLS relocations */ #define R_PPC_TLS 67 #define R_PPC_DTPMOD32 68 #define R_PPC_TPREL16 69 #define R_PPC_TPREL16_LO 70 #define R_PPC_TPREL16_HI 71 #define R_PPC_TPREL16_HA 72 #define R_PPC_TPREL32 73 #define R_PPC_DTPREL16 74 #define R_PPC_DTPREL16_LO 75 #define R_PPC_DTPREL16_HI 76 #define R_PPC_DTPREL16_HA 77 #define R_PPC_DTPREL32 78 #define R_PPC_GOT_TLSGD16 79 #define R_PPC_GOT_TLSGD16_LO 80 #define R_PPC_GOT_TLSGD16_HI 81 #define R_PPC_GOT_TLSGD16_HA 82 #define R_PPC_GOT_TLSLD16 83 #define R_PPC_GOT_TLSLD16_LO 84 #define R_PPC_GOT_TLSLD16_HI 85 #define R_PPC_GOT_TLSLD16_HA 86 #define R_PPC_GOT_TPREL16 87 #define R_PPC_GOT_TPREL16_LO 88 #define R_PPC_GOT_TPREL16_HI 89 #define R_PPC_GOT_TPREL16_HA 90 /* * The remaining relocs are from the Embedded ELF ABI, and are not in the * SVR4 ELF ABI. */ #define R_PPC_EMB_NADDR32 101 #define R_PPC_EMB_NADDR16 102 #define R_PPC_EMB_NADDR16_LO 103 #define R_PPC_EMB_NADDR16_HI 104 #define R_PPC_EMB_NADDR16_HA 105 #define R_PPC_EMB_SDAI16 106 #define R_PPC_EMB_SDA2I16 107 #define R_PPC_EMB_SDA2REL 108 #define R_PPC_EMB_SDA21 109 #define R_PPC_EMB_MRKREF 110 #define R_PPC_EMB_RELSEC16 111 #define R_PPC_EMB_RELST_LO 112 #define R_PPC_EMB_RELST_HI 113 #define R_PPC_EMB_RELST_HA 114 #define R_PPC_EMB_BIT_FLD 115 #define R_PPC_EMB_RELSDA 116 #define R_SPARC_NONE 0 #define R_SPARC_8 1 #define R_SPARC_16 2 #define R_SPARC_32 3 #define R_SPARC_DISP8 4 #define R_SPARC_DISP16 5 #define R_SPARC_DISP32 6 #define R_SPARC_WDISP30 7 #define R_SPARC_WDISP22 8 #define R_SPARC_HI22 9 #define R_SPARC_22 10 #define R_SPARC_13 11 #define R_SPARC_LO10 12 #define R_SPARC_GOT10 13 #define R_SPARC_GOT13 14 #define R_SPARC_GOT22 15 #define R_SPARC_PC10 16 #define R_SPARC_PC22 17 #define R_SPARC_WPLT30 18 #define R_SPARC_COPY 19 #define R_SPARC_GLOB_DAT 20 #define R_SPARC_JMP_SLOT 21 #define R_SPARC_RELATIVE 22 #define R_SPARC_UA32 23 #define R_SPARC_PLT32 24 #define R_SPARC_HIPLT22 25 #define R_SPARC_LOPLT10 26 #define R_SPARC_PCPLT32 27 #define R_SPARC_PCPLT22 28 #define R_SPARC_PCPLT10 29 #define R_SPARC_10 30 #define R_SPARC_11 31 #define R_SPARC_64 32 #define R_SPARC_OLO10 33 #define R_SPARC_HH22 34 #define R_SPARC_HM10 35 #define R_SPARC_LM22 36 #define R_SPARC_PC_HH22 37 #define R_SPARC_PC_HM10 38 #define R_SPARC_PC_LM22 39 #define R_SPARC_WDISP16 40 #define R_SPARC_WDISP19 41 #define R_SPARC_GLOB_JMP 42 #define R_SPARC_7 43 #define R_SPARC_5 44 #define R_SPARC_6 45 #define R_SPARC_DISP64 46 #define R_SPARC_PLT64 47 #define R_SPARC_HIX22 48 #define R_SPARC_LOX10 49 #define R_SPARC_H44 50 #define R_SPARC_M44 51 #define R_SPARC_L44 52 #define R_SPARC_REGISTER 53 #define R_SPARC_UA64 54 #define R_SPARC_UA16 55 #define R_SPARC_TLS_GD_HI22 56 #define R_SPARC_TLS_GD_LO10 57 #define R_SPARC_TLS_GD_ADD 58 #define R_SPARC_TLS_GD_CALL 59 #define R_SPARC_TLS_LDM_HI22 60 #define R_SPARC_TLS_LDM_LO10 61 #define R_SPARC_TLS_LDM_ADD 62 #define R_SPARC_TLS_LDM_CALL 63 #define R_SPARC_TLS_LDO_HIX22 64 #define R_SPARC_TLS_LDO_LOX10 65 #define R_SPARC_TLS_LDO_ADD 66 #define R_SPARC_TLS_IE_HI22 67 #define R_SPARC_TLS_IE_LO10 68 #define R_SPARC_TLS_IE_LD 69 #define R_SPARC_TLS_IE_LDX 70 #define R_SPARC_TLS_IE_ADD 71 #define R_SPARC_TLS_LE_HIX22 72 #define R_SPARC_TLS_LE_LOX10 73 #define R_SPARC_TLS_DTPMOD32 74 #define R_SPARC_TLS_DTPMOD64 75 #define R_SPARC_TLS_DTPOFF32 76 #define R_SPARC_TLS_DTPOFF64 77 #define R_SPARC_TLS_TPOFF32 78 #define R_SPARC_TLS_TPOFF64 79 #define R_X86_64_NONE 0 /* No relocation. */ #define R_X86_64_64 1 /* Add 64 bit symbol value. */ #define R_X86_64_PC32 2 /* PC-relative 32 bit signed sym value. */ #define R_X86_64_GOT32 3 /* PC-relative 32 bit GOT offset. */ #define R_X86_64_PLT32 4 /* PC-relative 32 bit PLT offset. */ #define R_X86_64_COPY 5 /* Copy data from shared object. */ #define R_X86_64_GLOB_DAT 6 /* Set GOT entry to data address. */ #define R_X86_64_JMP_SLOT 7 /* Set GOT entry to code address. */ #define R_X86_64_RELATIVE 8 /* Add load address of shared object. */ #define R_X86_64_GOTPCREL 9 /* Add 32 bit signed pcrel offset to GOT. */ #define R_X86_64_32 10 /* Add 32 bit zero extended symbol value */ #define R_X86_64_32S 11 /* Add 32 bit sign extended symbol value */ #define R_X86_64_16 12 /* Add 16 bit zero extended symbol value */ #define R_X86_64_PC16 13 /* Add 16 bit signed extended pc relative symbol value */ #define R_X86_64_8 14 /* Add 8 bit zero extended symbol value */ #define R_X86_64_PC8 15 /* Add 8 bit signed extended pc relative symbol value */ #define R_X86_64_DTPMOD64 16 /* ID of module containing symbol */ #define R_X86_64_DTPOFF64 17 /* Offset in TLS block */ #define R_X86_64_TPOFF64 18 /* Offset in static TLS block */ #define R_X86_64_TLSGD 19 /* PC relative offset to GD GOT entry */ #define R_X86_64_TLSLD 20 /* PC relative offset to LD GOT entry */ #define R_X86_64_DTPOFF32 21 /* Offset in TLS block */ #define R_X86_64_GOTTPOFF 22 /* PC relative offset to IE GOT entry */ #define R_X86_64_TPOFF32 23 /* Offset in static TLS block */ #define R_X86_64_PC64 24 /* PC-relative 64 bit signed sym value. */ #define R_X86_64_GOTOFF64 25 #define R_X86_64_GOTPC32 26 #define R_X86_64_GOT64 27 #define R_X86_64_GOTPCREL64 28 #define R_X86_64_GOTPC64 29 #define R_X86_64_GOTPLT64 30 #define R_X86_64_PLTOFF64 31 #define R_X86_64_SIZE32 32 #define R_X86_64_SIZE64 33 #define R_X86_64_GOTPC32_TLSDESC 34 #define R_X86_64_TLSDESC_CALL 35 #define R_X86_64_TLSDESC 36 #define R_X86_64_IRELATIVE 37 #endif /* !_SYS_ELF_COMMON_H_ */ Index: projects/cxl_iscsi/sys/vm/vm_pager.c =================================================================== --- projects/cxl_iscsi/sys/vm/vm_pager.c (revision 279716) +++ projects/cxl_iscsi/sys/vm/vm_pager.c (revision 279717) @@ -1,517 +1,518 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * The Mach Operating System project at Carnegie-Mellon University. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)vm_pager.c 8.6 (Berkeley) 1/12/94 * * * Copyright (c) 1987, 1990 Carnegie-Mellon University. * All rights reserved. * * Authors: Avadis Tevanian, Jr., Michael Wayne Young * * Permission to use, copy, modify and distribute this software and * its documentation is hereby granted, provided that both the copyright * notice and this permission notice appear in all copies of the * software, derivative works or modified versions, and any portions * thereof, and that both notices appear in supporting documentation. * * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. * * Carnegie Mellon requests users of this software to return to * * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU * School of Computer Science * Carnegie Mellon University * Pittsburgh PA 15213-3890 * * any improvements or extensions that they make and grant Carnegie the * rights to redistribute these changes. */ /* * Paging space routine stubs. Emulates a matchmaker-like interface * for builtin pagers. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include int cluster_pbuf_freecnt = -1; /* unlimited to begin with */ static int dead_pager_getpages(vm_object_t, vm_page_t *, int, int); static vm_object_t dead_pager_alloc(void *, vm_ooffset_t, vm_prot_t, vm_ooffset_t, struct ucred *); static void dead_pager_putpages(vm_object_t, vm_page_t *, int, int, int *); static boolean_t dead_pager_haspage(vm_object_t, vm_pindex_t, int *, int *); static void dead_pager_dealloc(vm_object_t); static int dead_pager_getpages(obj, ma, count, req) vm_object_t obj; vm_page_t *ma; int count; int req; { return VM_PAGER_FAIL; } static vm_object_t dead_pager_alloc(void *handle, vm_ooffset_t size, vm_prot_t prot, vm_ooffset_t off, struct ucred *cred) { return NULL; } static void dead_pager_putpages(object, m, count, flags, rtvals) vm_object_t object; vm_page_t *m; int count; int flags; int *rtvals; { int i; for (i = 0; i < count; i++) { rtvals[i] = VM_PAGER_AGAIN; } } static int dead_pager_haspage(object, pindex, prev, next) vm_object_t object; vm_pindex_t pindex; int *prev; int *next; { if (prev) *prev = 0; if (next) *next = 0; return FALSE; } static void dead_pager_dealloc(object) vm_object_t object; { return; } static struct pagerops deadpagerops = { .pgo_alloc = dead_pager_alloc, .pgo_dealloc = dead_pager_dealloc, .pgo_getpages = dead_pager_getpages, .pgo_putpages = dead_pager_putpages, .pgo_haspage = dead_pager_haspage, }; struct pagerops *pagertab[] = { &defaultpagerops, /* OBJT_DEFAULT */ &swappagerops, /* OBJT_SWAP */ &vnodepagerops, /* OBJT_VNODE */ &devicepagerops, /* OBJT_DEVICE */ &physpagerops, /* OBJT_PHYS */ &deadpagerops, /* OBJT_DEAD */ &sgpagerops, /* OBJT_SG */ &mgtdevicepagerops, /* OBJT_MGTDEVICE */ }; static const int npagers = sizeof(pagertab) / sizeof(pagertab[0]); /* * Kernel address space for mapping pages. * Used by pagers where KVAs are needed for IO. * * XXX needs to be large enough to support the number of pending async * cleaning requests (NPENDINGIO == 64) * the maximum swap cluster size * (MAXPHYS == 64k) if you want to get the most efficiency. */ struct mtx_padalign pbuf_mtx; static TAILQ_HEAD(swqueue, buf) bswlist; static int bswneeded; vm_offset_t swapbkva; /* swap buffers kva */ void vm_pager_init() { struct pagerops **pgops; TAILQ_INIT(&bswlist); /* * Initialize known pagers */ for (pgops = pagertab; pgops < &pagertab[npagers]; pgops++) if ((*pgops)->pgo_init != NULL) (*(*pgops)->pgo_init) (); } void vm_pager_bufferinit() { struct buf *bp; int i; mtx_init(&pbuf_mtx, "pbuf mutex", NULL, MTX_DEF); bp = swbuf; /* * Now set up swap and physical I/O buffer headers. */ for (i = 0; i < nswbuf; i++, bp++) { TAILQ_INSERT_HEAD(&bswlist, bp, b_freelist); BUF_LOCKINIT(bp); LIST_INIT(&bp->b_dep); bp->b_rcred = bp->b_wcred = NOCRED; bp->b_xflags = 0; } cluster_pbuf_freecnt = nswbuf / 2; vnode_pbuf_freecnt = nswbuf / 2 + 1; + vnode_async_pbuf_freecnt = nswbuf / 2; } /* * Allocate an instance of a pager of the given type. * Size, protection and offset parameters are passed in for pagers that * need to perform page-level validation (e.g. the device pager). */ vm_object_t vm_pager_allocate(objtype_t type, void *handle, vm_ooffset_t size, vm_prot_t prot, vm_ooffset_t off, struct ucred *cred) { vm_object_t ret; struct pagerops *ops; ops = pagertab[type]; if (ops) ret = (*ops->pgo_alloc) (handle, size, prot, off, cred); else ret = NULL; return (ret); } /* * The object must be locked. */ void vm_pager_deallocate(object) vm_object_t object; { VM_OBJECT_ASSERT_WLOCKED(object); (*pagertab[object->type]->pgo_dealloc) (object); } /* * vm_pager_get_pages() - inline, see vm/vm_pager.h * vm_pager_put_pages() - inline, see vm/vm_pager.h * vm_pager_has_page() - inline, see vm/vm_pager.h */ /* * Search the specified pager object list for an object with the * specified handle. If an object with the specified handle is found, * increase its reference count and return it. Otherwise, return NULL. * * The pager object list must be locked. */ vm_object_t vm_pager_object_lookup(struct pagerlst *pg_list, void *handle) { vm_object_t object; TAILQ_FOREACH(object, pg_list, pager_object_list) { if (object->handle == handle) { VM_OBJECT_WLOCK(object); if ((object->flags & OBJ_DEAD) == 0) { vm_object_reference_locked(object); VM_OBJECT_WUNLOCK(object); break; } VM_OBJECT_WUNLOCK(object); } } return (object); } /* * Free the non-requested pages from the given array. */ void vm_pager_free_nonreq(vm_object_t object, vm_page_t ma[], int reqpage, int npages) { int i; boolean_t object_locked; VM_OBJECT_ASSERT_UNLOCKED(object); object_locked = FALSE; for (i = 0; i < npages; ++i) { if (i != reqpage) { if (!object_locked) { VM_OBJECT_WLOCK(object); object_locked = TRUE; } vm_page_lock(ma[i]); vm_page_free(ma[i]); vm_page_unlock(ma[i]); } } if (object_locked) VM_OBJECT_WUNLOCK(object); } /* * initialize a physical buffer */ /* * XXX This probably belongs in vfs_bio.c */ static void initpbuf(struct buf *bp) { KASSERT(bp->b_bufobj == NULL, ("initpbuf with bufobj")); KASSERT(bp->b_vp == NULL, ("initpbuf with vp")); bp->b_rcred = NOCRED; bp->b_wcred = NOCRED; bp->b_qindex = 0; /* On no queue (QUEUE_NONE) */ bp->b_saveaddr = (caddr_t) (MAXPHYS * (bp - swbuf)) + swapbkva; bp->b_data = bp->b_saveaddr; bp->b_kvabase = bp->b_saveaddr; bp->b_kvasize = MAXPHYS; bp->b_xflags = 0; bp->b_flags = 0; bp->b_ioflags = 0; bp->b_iodone = NULL; bp->b_error = 0; BUF_LOCK(bp, LK_EXCLUSIVE, NULL); } /* * allocate a physical buffer * * There are a limited number (nswbuf) of physical buffers. We need * to make sure that no single subsystem is able to hog all of them, * so each subsystem implements a counter which is typically initialized * to 1/2 nswbuf. getpbuf() decrements this counter in allocation and * increments it on release, and blocks if the counter hits zero. A * subsystem may initialize the counter to -1 to disable the feature, * but it must still be sure to match up all uses of getpbuf() with * relpbuf() using the same variable. * * NOTE: pfreecnt can be NULL, but this 'feature' will be removed * relatively soon when the rest of the subsystems get smart about it. XXX */ struct buf * getpbuf(int *pfreecnt) { struct buf *bp; mtx_lock(&pbuf_mtx); for (;;) { if (pfreecnt) { while (*pfreecnt == 0) { msleep(pfreecnt, &pbuf_mtx, PVM, "wswbuf0", 0); } } /* get a bp from the swap buffer header pool */ if ((bp = TAILQ_FIRST(&bswlist)) != NULL) break; bswneeded = 1; msleep(&bswneeded, &pbuf_mtx, PVM, "wswbuf1", 0); /* loop in case someone else grabbed one */ } TAILQ_REMOVE(&bswlist, bp, b_freelist); if (pfreecnt) --*pfreecnt; mtx_unlock(&pbuf_mtx); initpbuf(bp); return bp; } /* * allocate a physical buffer, if one is available. * * Note that there is no NULL hack here - all subsystems using this * call understand how to use pfreecnt. */ struct buf * trypbuf(int *pfreecnt) { struct buf *bp; mtx_lock(&pbuf_mtx); if (*pfreecnt == 0 || (bp = TAILQ_FIRST(&bswlist)) == NULL) { mtx_unlock(&pbuf_mtx); return NULL; } TAILQ_REMOVE(&bswlist, bp, b_freelist); --*pfreecnt; mtx_unlock(&pbuf_mtx); initpbuf(bp); return bp; } /* * release a physical buffer * * NOTE: pfreecnt can be NULL, but this 'feature' will be removed * relatively soon when the rest of the subsystems get smart about it. XXX */ void relpbuf(struct buf *bp, int *pfreecnt) { if (bp->b_rcred != NOCRED) { crfree(bp->b_rcred); bp->b_rcred = NOCRED; } if (bp->b_wcred != NOCRED) { crfree(bp->b_wcred); bp->b_wcred = NOCRED; } KASSERT(bp->b_vp == NULL, ("relpbuf with vp")); KASSERT(bp->b_bufobj == NULL, ("relpbuf with bufobj")); BUF_UNLOCK(bp); mtx_lock(&pbuf_mtx); TAILQ_INSERT_HEAD(&bswlist, bp, b_freelist); if (bswneeded) { bswneeded = 0; wakeup(&bswneeded); } if (pfreecnt) { if (++*pfreecnt == 1) wakeup(pfreecnt); } mtx_unlock(&pbuf_mtx); } /* * Associate a p-buffer with a vnode. * * Also sets B_PAGING flag to indicate that vnode is not fully associated * with the buffer. i.e. the bp has not been linked into the vnode or * ref-counted. */ void pbgetvp(struct vnode *vp, struct buf *bp) { KASSERT(bp->b_vp == NULL, ("pbgetvp: not free")); KASSERT(bp->b_bufobj == NULL, ("pbgetvp: not free (bufobj)")); bp->b_vp = vp; bp->b_flags |= B_PAGING; bp->b_bufobj = &vp->v_bufobj; } /* * Associate a p-buffer with a vnode. * * Also sets B_PAGING flag to indicate that vnode is not fully associated * with the buffer. i.e. the bp has not been linked into the vnode or * ref-counted. */ void pbgetbo(struct bufobj *bo, struct buf *bp) { KASSERT(bp->b_vp == NULL, ("pbgetbo: not free (vnode)")); KASSERT(bp->b_bufobj == NULL, ("pbgetbo: not free (bufobj)")); bp->b_flags |= B_PAGING; bp->b_bufobj = bo; } /* * Disassociate a p-buffer from a vnode. */ void pbrelvp(struct buf *bp) { KASSERT(bp->b_vp != NULL, ("pbrelvp: NULL")); KASSERT(bp->b_bufobj != NULL, ("pbrelvp: NULL bufobj")); KASSERT((bp->b_xflags & (BX_VNDIRTY | BX_VNCLEAN)) == 0, ("pbrelvp: pager buf on vnode list.")); bp->b_vp = NULL; bp->b_bufobj = NULL; bp->b_flags &= ~B_PAGING; } /* * Disassociate a p-buffer from a bufobj. */ void pbrelbo(struct buf *bp) { KASSERT(bp->b_vp == NULL, ("pbrelbo: vnode")); KASSERT(bp->b_bufobj != NULL, ("pbrelbo: NULL bufobj")); KASSERT((bp->b_xflags & (BX_VNDIRTY | BX_VNCLEAN)) == 0, ("pbrelbo: pager buf on vnode list.")); bp->b_bufobj = NULL; bp->b_flags &= ~B_PAGING; } Index: projects/cxl_iscsi/sys/vm/vnode_pager.c =================================================================== --- projects/cxl_iscsi/sys/vm/vnode_pager.c (revision 279716) +++ projects/cxl_iscsi/sys/vm/vnode_pager.c (revision 279717) @@ -1,1369 +1,1378 @@ /*- * Copyright (c) 1990 University of Utah. * Copyright (c) 1991 The Regents of the University of California. * All rights reserved. * Copyright (c) 1993, 1994 John S. Dyson * Copyright (c) 1995, David Greenman * * This code is derived from software contributed to Berkeley by * the Systems Programming Group of the University of Utah Computer * Science Department. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)vnode_pager.c 7.5 (Berkeley) 4/20/91 */ /* * Page to/from files (vnodes). */ /* * TODO: * Implement VOP_GETPAGES/PUTPAGES interface for filesystems. Will * greatly re-simplify the vnode_pager. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include static int vnode_pager_addr(struct vnode *vp, vm_ooffset_t address, daddr_t *rtaddress, int *run); static int vnode_pager_input_smlfs(vm_object_t object, vm_page_t m); static int vnode_pager_input_old(vm_object_t object, vm_page_t m); static void vnode_pager_dealloc(vm_object_t); static int vnode_pager_local_getpages0(struct vnode *, vm_page_t *, int, int, vop_getpages_iodone_t, void *); static int vnode_pager_getpages(vm_object_t, vm_page_t *, int, int); static int vnode_pager_getpages_async(vm_object_t, vm_page_t *, int, int, vop_getpages_iodone_t, void *); static void vnode_pager_putpages(vm_object_t, vm_page_t *, int, int, int *); static boolean_t vnode_pager_haspage(vm_object_t, vm_pindex_t, int *, int *); static vm_object_t vnode_pager_alloc(void *, vm_ooffset_t, vm_prot_t, vm_ooffset_t, struct ucred *cred); static int vnode_pager_generic_getpages_done(struct buf *); static void vnode_pager_generic_getpages_done_async(struct buf *); struct pagerops vnodepagerops = { .pgo_alloc = vnode_pager_alloc, .pgo_dealloc = vnode_pager_dealloc, .pgo_getpages = vnode_pager_getpages, .pgo_getpages_async = vnode_pager_getpages_async, .pgo_putpages = vnode_pager_putpages, .pgo_haspage = vnode_pager_haspage, }; int vnode_pbuf_freecnt; +int vnode_async_pbuf_freecnt; /* Create the VM system backing object for this vnode */ int vnode_create_vobject(struct vnode *vp, off_t isize, struct thread *td) { vm_object_t object; vm_ooffset_t size = isize; struct vattr va; if (!vn_isdisk(vp, NULL) && vn_canvmio(vp) == FALSE) return (0); while ((object = vp->v_object) != NULL) { VM_OBJECT_WLOCK(object); if (!(object->flags & OBJ_DEAD)) { VM_OBJECT_WUNLOCK(object); return (0); } VOP_UNLOCK(vp, 0); vm_object_set_flag(object, OBJ_DISCONNECTWNT); VM_OBJECT_SLEEP(object, object, PDROP | PVM, "vodead", 0); vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); } if (size == 0) { if (vn_isdisk(vp, NULL)) { size = IDX_TO_OFF(INT_MAX); } else { if (VOP_GETATTR(vp, &va, td->td_ucred)) return (0); size = va.va_size; } } object = vnode_pager_alloc(vp, size, 0, 0, td->td_ucred); /* * Dereference the reference we just created. This assumes * that the object is associated with the vp. */ VM_OBJECT_WLOCK(object); object->ref_count--; VM_OBJECT_WUNLOCK(object); vrele(vp); KASSERT(vp->v_object != NULL, ("vnode_create_vobject: NULL object")); return (0); } void vnode_destroy_vobject(struct vnode *vp) { struct vm_object *obj; obj = vp->v_object; if (obj == NULL) return; ASSERT_VOP_ELOCKED(vp, "vnode_destroy_vobject"); VM_OBJECT_WLOCK(obj); if (obj->ref_count == 0) { /* * don't double-terminate the object */ if ((obj->flags & OBJ_DEAD) == 0) vm_object_terminate(obj); else VM_OBJECT_WUNLOCK(obj); } else { /* * Woe to the process that tries to page now :-). */ vm_pager_deallocate(obj); VM_OBJECT_WUNLOCK(obj); } vp->v_object = NULL; } /* * Allocate (or lookup) pager for a vnode. * Handle is a vnode pointer. * * MPSAFE */ vm_object_t vnode_pager_alloc(void *handle, vm_ooffset_t size, vm_prot_t prot, vm_ooffset_t offset, struct ucred *cred) { vm_object_t object; struct vnode *vp; /* * Pageout to vnode, no can do yet. */ if (handle == NULL) return (NULL); vp = (struct vnode *) handle; /* * If the object is being terminated, wait for it to * go away. */ retry: while ((object = vp->v_object) != NULL) { VM_OBJECT_WLOCK(object); if ((object->flags & OBJ_DEAD) == 0) break; vm_object_set_flag(object, OBJ_DISCONNECTWNT); VM_OBJECT_SLEEP(object, object, PDROP | PVM, "vadead", 0); } KASSERT(vp->v_usecount != 0, ("vnode_pager_alloc: no vnode reference")); if (object == NULL) { /* * Add an object of the appropriate size */ object = vm_object_allocate(OBJT_VNODE, OFF_TO_IDX(round_page(size))); object->un_pager.vnp.vnp_size = size; object->un_pager.vnp.writemappings = 0; object->handle = handle; VI_LOCK(vp); if (vp->v_object != NULL) { /* * Object has been created while we were sleeping */ VI_UNLOCK(vp); vm_object_destroy(object); goto retry; } vp->v_object = object; VI_UNLOCK(vp); } else { object->ref_count++; VM_OBJECT_WUNLOCK(object); } vref(vp); return (object); } /* * The object must be locked. */ static void vnode_pager_dealloc(vm_object_t object) { struct vnode *vp; int refs; vp = object->handle; if (vp == NULL) panic("vnode_pager_dealloc: pager already dealloced"); VM_OBJECT_ASSERT_WLOCKED(object); vm_object_pip_wait(object, "vnpdea"); refs = object->ref_count; object->handle = NULL; object->type = OBJT_DEAD; if (object->flags & OBJ_DISCONNECTWNT) { vm_object_clear_flag(object, OBJ_DISCONNECTWNT); wakeup(object); } ASSERT_VOP_ELOCKED(vp, "vnode_pager_dealloc"); if (object->un_pager.vnp.writemappings > 0) { object->un_pager.vnp.writemappings = 0; VOP_ADD_WRITECOUNT(vp, -1); CTR3(KTR_VFS, "%s: vp %p v_writecount decreased to %d", __func__, vp, vp->v_writecount); } vp->v_object = NULL; VOP_UNSET_TEXT(vp); VM_OBJECT_WUNLOCK(object); while (refs-- > 0) vunref(vp); VM_OBJECT_WLOCK(object); } static boolean_t vnode_pager_haspage(vm_object_t object, vm_pindex_t pindex, int *before, int *after) { struct vnode *vp = object->handle; daddr_t bn; int err; daddr_t reqblock; int poff; int bsize; int pagesperblock, blocksperpage; VM_OBJECT_ASSERT_WLOCKED(object); /* * If no vp or vp is doomed or marked transparent to VM, we do not * have the page. */ if (vp == NULL || vp->v_iflag & VI_DOOMED) return FALSE; /* * If the offset is beyond end of file we do * not have the page. */ if (IDX_TO_OFF(pindex) >= object->un_pager.vnp.vnp_size) return FALSE; bsize = vp->v_mount->mnt_stat.f_iosize; pagesperblock = bsize / PAGE_SIZE; blocksperpage = 0; if (pagesperblock > 0) { reqblock = pindex / pagesperblock; } else { blocksperpage = (PAGE_SIZE / bsize); reqblock = pindex * blocksperpage; } VM_OBJECT_WUNLOCK(object); err = VOP_BMAP(vp, reqblock, NULL, &bn, after, before); VM_OBJECT_WLOCK(object); if (err) return TRUE; if (bn == -1) return FALSE; if (pagesperblock > 0) { poff = pindex - (reqblock * pagesperblock); if (before) { *before *= pagesperblock; *before += poff; } if (after) { int numafter; *after *= pagesperblock; numafter = pagesperblock - (poff + 1); if (IDX_TO_OFF(pindex + numafter) > object->un_pager.vnp.vnp_size) { numafter = OFF_TO_IDX(object->un_pager.vnp.vnp_size) - pindex; } *after += numafter; } } else { if (before) { *before /= blocksperpage; } if (after) { *after /= blocksperpage; } } return TRUE; } /* * Lets the VM system know about a change in size for a file. * We adjust our own internal size and flush any cached pages in * the associated object that are affected by the size change. * * Note: this routine may be invoked as a result of a pager put * operation (possibly at object termination time), so we must be careful. */ void vnode_pager_setsize(struct vnode *vp, vm_ooffset_t nsize) { vm_object_t object; vm_page_t m; vm_pindex_t nobjsize; if ((object = vp->v_object) == NULL) return; /* ASSERT_VOP_ELOCKED(vp, "vnode_pager_setsize and not locked vnode"); */ VM_OBJECT_WLOCK(object); if (object->type == OBJT_DEAD) { VM_OBJECT_WUNLOCK(object); return; } KASSERT(object->type == OBJT_VNODE, ("not vnode-backed object %p", object)); if (nsize == object->un_pager.vnp.vnp_size) { /* * Hasn't changed size */ VM_OBJECT_WUNLOCK(object); return; } nobjsize = OFF_TO_IDX(nsize + PAGE_MASK); if (nsize < object->un_pager.vnp.vnp_size) { /* * File has shrunk. Toss any cached pages beyond the new EOF. */ if (nobjsize < object->size) vm_object_page_remove(object, nobjsize, object->size, 0); /* * this gets rid of garbage at the end of a page that is now * only partially backed by the vnode. * * XXX for some reason (I don't know yet), if we take a * completely invalid page and mark it partially valid * it can screw up NFS reads, so we don't allow the case. */ if ((nsize & PAGE_MASK) && (m = vm_page_lookup(object, OFF_TO_IDX(nsize))) != NULL && m->valid != 0) { int base = (int)nsize & PAGE_MASK; int size = PAGE_SIZE - base; /* * Clear out partial-page garbage in case * the page has been mapped. */ pmap_zero_page_area(m, base, size); /* * Update the valid bits to reflect the blocks that * have been zeroed. Some of these valid bits may * have already been set. */ vm_page_set_valid_range(m, base, size); /* * Round "base" to the next block boundary so that the * dirty bit for a partially zeroed block is not * cleared. */ base = roundup2(base, DEV_BSIZE); /* * Clear out partial-page dirty bits. * * note that we do not clear out the valid * bits. This would prevent bogus_page * replacement from working properly. */ vm_page_clear_dirty(m, base, PAGE_SIZE - base); } else if ((nsize & PAGE_MASK) && vm_page_is_cached(object, OFF_TO_IDX(nsize))) { vm_page_cache_free(object, OFF_TO_IDX(nsize), nobjsize); } } object->un_pager.vnp.vnp_size = nsize; object->size = nobjsize; VM_OBJECT_WUNLOCK(object); } /* * calculate the linear (byte) disk address of specified virtual * file address */ static int vnode_pager_addr(struct vnode *vp, vm_ooffset_t address, daddr_t *rtaddress, int *run) { int bsize; int err; daddr_t vblock; daddr_t voffset; if (address < 0) return -1; if (vp->v_iflag & VI_DOOMED) return -1; bsize = vp->v_mount->mnt_stat.f_iosize; vblock = address / bsize; voffset = address % bsize; err = VOP_BMAP(vp, vblock, NULL, rtaddress, run, NULL); if (err == 0) { if (*rtaddress != -1) *rtaddress += voffset / DEV_BSIZE; if (run) { *run += 1; *run *= bsize/PAGE_SIZE; *run -= voffset/PAGE_SIZE; } } return (err); } /* * small block filesystem vnode pager input */ static int vnode_pager_input_smlfs(vm_object_t object, vm_page_t m) { struct vnode *vp; struct bufobj *bo; struct buf *bp; struct sf_buf *sf; daddr_t fileaddr; vm_offset_t bsize; vm_page_bits_t bits; int error, i; error = 0; vp = object->handle; if (vp->v_iflag & VI_DOOMED) return VM_PAGER_BAD; bsize = vp->v_mount->mnt_stat.f_iosize; VOP_BMAP(vp, 0, &bo, 0, NULL, NULL); sf = sf_buf_alloc(m, 0); for (i = 0; i < PAGE_SIZE / bsize; i++) { vm_ooffset_t address; bits = vm_page_bits(i * bsize, bsize); if (m->valid & bits) continue; address = IDX_TO_OFF(m->pindex) + i * bsize; if (address >= object->un_pager.vnp.vnp_size) { fileaddr = -1; } else { error = vnode_pager_addr(vp, address, &fileaddr, NULL); if (error) break; } if (fileaddr != -1) { bp = getpbuf(&vnode_pbuf_freecnt); /* build a minimal buffer header */ bp->b_iocmd = BIO_READ; bp->b_iodone = bdone; KASSERT(bp->b_rcred == NOCRED, ("leaking read ucred")); KASSERT(bp->b_wcred == NOCRED, ("leaking write ucred")); bp->b_rcred = crhold(curthread->td_ucred); bp->b_wcred = crhold(curthread->td_ucred); bp->b_data = (caddr_t)sf_buf_kva(sf) + i * bsize; bp->b_blkno = fileaddr; pbgetbo(bo, bp); bp->b_vp = vp; bp->b_bcount = bsize; bp->b_bufsize = bsize; bp->b_runningbufspace = bp->b_bufsize; atomic_add_long(&runningbufspace, bp->b_runningbufspace); /* do the input */ bp->b_iooffset = dbtob(bp->b_blkno); bstrategy(bp); bwait(bp, PVM, "vnsrd"); if ((bp->b_ioflags & BIO_ERROR) != 0) error = EIO; /* * free the buffer header back to the swap buffer pool */ bp->b_vp = NULL; pbrelbo(bp); relpbuf(bp, &vnode_pbuf_freecnt); if (error) break; } else bzero((caddr_t)sf_buf_kva(sf) + i * bsize, bsize); KASSERT((m->dirty & bits) == 0, ("vnode_pager_input_smlfs: page %p is dirty", m)); VM_OBJECT_WLOCK(object); m->valid |= bits; VM_OBJECT_WUNLOCK(object); } sf_buf_free(sf); if (error) { return VM_PAGER_ERROR; } return VM_PAGER_OK; } /* * old style vnode pager input routine */ static int vnode_pager_input_old(vm_object_t object, vm_page_t m) { struct uio auio; struct iovec aiov; int error; int size; struct sf_buf *sf; struct vnode *vp; VM_OBJECT_ASSERT_WLOCKED(object); error = 0; /* * Return failure if beyond current EOF */ if (IDX_TO_OFF(m->pindex) >= object->un_pager.vnp.vnp_size) { return VM_PAGER_BAD; } else { size = PAGE_SIZE; if (IDX_TO_OFF(m->pindex) + size > object->un_pager.vnp.vnp_size) size = object->un_pager.vnp.vnp_size - IDX_TO_OFF(m->pindex); vp = object->handle; VM_OBJECT_WUNLOCK(object); /* * Allocate a kernel virtual address and initialize so that * we can use VOP_READ/WRITE routines. */ sf = sf_buf_alloc(m, 0); aiov.iov_base = (caddr_t)sf_buf_kva(sf); aiov.iov_len = size; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_offset = IDX_TO_OFF(m->pindex); auio.uio_segflg = UIO_SYSSPACE; auio.uio_rw = UIO_READ; auio.uio_resid = size; auio.uio_td = curthread; error = VOP_READ(vp, &auio, 0, curthread->td_ucred); if (!error) { int count = size - auio.uio_resid; if (count == 0) error = EINVAL; else if (count != PAGE_SIZE) bzero((caddr_t)sf_buf_kva(sf) + count, PAGE_SIZE - count); } sf_buf_free(sf); VM_OBJECT_WLOCK(object); } KASSERT(m->dirty == 0, ("vnode_pager_input_old: page %p is dirty", m)); if (!error) m->valid = VM_PAGE_BITS_ALL; return error ? VM_PAGER_ERROR : VM_PAGER_OK; } /* * generic vnode pager input routine */ /* * Local media VFS's that do not implement their own VOP_GETPAGES * should have their VOP_GETPAGES call to vnode_pager_generic_getpages() * to implement the previous behaviour. * * All other FS's should use the bypass to get to the local media * backing vp's VOP_GETPAGES. */ static int vnode_pager_getpages(vm_object_t object, vm_page_t *m, int count, int reqpage) { int rtval; struct vnode *vp; int bytes = count * PAGE_SIZE; vp = object->handle; VM_OBJECT_WUNLOCK(object); rtval = VOP_GETPAGES(vp, m, bytes, reqpage); KASSERT(rtval != EOPNOTSUPP, ("vnode_pager: FS getpages not implemented\n")); VM_OBJECT_WLOCK(object); return rtval; } static int vnode_pager_getpages_async(vm_object_t object, vm_page_t *m, int count, int reqpage, vop_getpages_iodone_t iodone, void *arg) { struct vnode *vp; int rtval; vp = object->handle; VM_OBJECT_WUNLOCK(object); rtval = VOP_GETPAGES_ASYNC(vp, m, count * PAGE_SIZE, reqpage, 0, iodone, arg); KASSERT(rtval != EOPNOTSUPP, ("vnode_pager: FS getpages_async not implemented\n")); VM_OBJECT_WLOCK(object); return (rtval); } /* * The implementation of VOP_GETPAGES() and VOP_GETPAGES_ASYNC() for * local filesystems, where partially valid pages can only occur at * the end of file. */ int vnode_pager_local_getpages(struct vop_getpages_args *ap) { return (vnode_pager_local_getpages0(ap->a_vp, ap->a_m, ap->a_count, ap->a_reqpage, NULL, NULL)); } int vnode_pager_local_getpages_async(struct vop_getpages_async_args *ap) { return (vnode_pager_local_getpages0(ap->a_vp, ap->a_m, ap->a_count, ap->a_reqpage, ap->a_iodone, ap->a_arg)); } static int vnode_pager_local_getpages0(struct vnode *vp, vm_page_t *m, int bytecount, int reqpage, vop_getpages_iodone_t iodone, void *arg) { vm_page_t mreq; mreq = m[reqpage]; /* * Since the caller has busied the requested page, that page's valid * field will not be changed by other threads. */ vm_page_assert_xbusied(mreq); /* * The requested page has valid blocks. Invalid part can only * exist at the end of file, and the page is made fully valid * by zeroing in vm_pager_getpages(). Free non-requested * pages, since no i/o is done to read its content. */ if (mreq->valid != 0) { vm_pager_free_nonreq(mreq->object, m, reqpage, round_page(bytecount) / PAGE_SIZE); if (iodone != NULL) iodone(arg, m, reqpage, 0); return (VM_PAGER_OK); } return (vnode_pager_generic_getpages(vp, m, bytecount, reqpage, iodone, arg)); } /* * This is now called from local media FS's to operate against their * own vnodes if they fail to implement VOP_GETPAGES. */ int vnode_pager_generic_getpages(struct vnode *vp, vm_page_t *m, int bytecount, int reqpage, vop_getpages_iodone_t iodone, void *arg) { vm_object_t object; off_t foff; - int i, j, size, bsize, first; + int i, j, size, bsize, first, *freecnt; daddr_t firstaddr, reqblock; struct bufobj *bo; int runpg; int runend; struct buf *bp; int count; int error; object = vp->v_object; count = bytecount / PAGE_SIZE; KASSERT(vp->v_type != VCHR && vp->v_type != VBLK, ("vnode_pager_generic_getpages does not support devices")); if (vp->v_iflag & VI_DOOMED) return VM_PAGER_BAD; bsize = vp->v_mount->mnt_stat.f_iosize; foff = IDX_TO_OFF(m[reqpage]->pindex); + freecnt = iodone != NULL ? + &vnode_async_pbuf_freecnt : &vnode_pbuf_freecnt; + bp = getpbuf(freecnt); + /* * Get the underlying device blocks for the file with VOP_BMAP(). * If the file system doesn't support VOP_BMAP, use old way of * getting pages via VOP_READ. */ error = VOP_BMAP(vp, foff / bsize, &bo, &reqblock, NULL, NULL); if (error == EOPNOTSUPP) { + relpbuf(bp, freecnt); VM_OBJECT_WLOCK(object); - for (i = 0; i < count; i++) if (i != reqpage) { vm_page_lock(m[i]); vm_page_free(m[i]); vm_page_unlock(m[i]); } PCPU_INC(cnt.v_vnodein); PCPU_INC(cnt.v_vnodepgsin); error = vnode_pager_input_old(object, m[reqpage]); VM_OBJECT_WUNLOCK(object); return (error); } else if (error != 0) { + relpbuf(bp, freecnt); vm_pager_free_nonreq(object, m, reqpage, count); return (VM_PAGER_ERROR); /* * if the blocksize is smaller than a page size, then use * special small filesystem code. NFS sometimes has a small * blocksize, but it can handle large reads itself. */ } else if ((PAGE_SIZE / bsize) > 1 && (vp->v_mount->mnt_stat.f_type != nfs_mount_type)) { + relpbuf(bp, freecnt); vm_pager_free_nonreq(object, m, reqpage, count); PCPU_INC(cnt.v_vnodein); PCPU_INC(cnt.v_vnodepgsin); return vnode_pager_input_smlfs(object, m[reqpage]); } /* * Since the caller has busied the requested page, that page's valid * field will not be changed by other threads. */ vm_page_assert_xbusied(m[reqpage]); /* * If we have a completely valid page available to us, we can * clean up and return. Otherwise we have to re-read the * media. */ if (m[reqpage]->valid == VM_PAGE_BITS_ALL) { + relpbuf(bp, freecnt); vm_pager_free_nonreq(object, m, reqpage, count); return (VM_PAGER_OK); } else if (reqblock == -1) { + relpbuf(bp, freecnt); pmap_zero_page(m[reqpage]); KASSERT(m[reqpage]->dirty == 0, ("vnode_pager_generic_getpages: page %p is dirty", m)); VM_OBJECT_WLOCK(object); m[reqpage]->valid = VM_PAGE_BITS_ALL; for (i = 0; i < count; i++) if (i != reqpage) { vm_page_lock(m[i]); vm_page_free(m[i]); vm_page_unlock(m[i]); } VM_OBJECT_WUNLOCK(object); return (VM_PAGER_OK); } else if (m[reqpage]->valid != 0) { VM_OBJECT_WLOCK(object); m[reqpage]->valid = 0; VM_OBJECT_WUNLOCK(object); } /* * here on direct device I/O */ firstaddr = -1; /* * calculate the run that includes the required page */ for (first = 0, i = 0; i < count; i = runend) { if (vnode_pager_addr(vp, IDX_TO_OFF(m[i]->pindex), &firstaddr, &runpg) != 0) { + relpbuf(bp, freecnt); VM_OBJECT_WLOCK(object); for (; i < count; i++) if (i != reqpage) { vm_page_lock(m[i]); vm_page_free(m[i]); vm_page_unlock(m[i]); } VM_OBJECT_WUNLOCK(object); return (VM_PAGER_ERROR); } if (firstaddr == -1) { VM_OBJECT_WLOCK(object); if (i == reqpage && foff < object->un_pager.vnp.vnp_size) { panic("vnode_pager_getpages: unexpected missing page: firstaddr: %jd, foff: 0x%jx%08jx, vnp_size: 0x%jx%08jx", (intmax_t)firstaddr, (uintmax_t)(foff >> 32), (uintmax_t)foff, (uintmax_t) (object->un_pager.vnp.vnp_size >> 32), (uintmax_t)object->un_pager.vnp.vnp_size); } vm_page_lock(m[i]); vm_page_free(m[i]); vm_page_unlock(m[i]); VM_OBJECT_WUNLOCK(object); runend = i + 1; first = runend; continue; } runend = i + runpg; if (runend <= reqpage) { VM_OBJECT_WLOCK(object); for (j = i; j < runend; j++) { vm_page_lock(m[j]); vm_page_free(m[j]); vm_page_unlock(m[j]); } VM_OBJECT_WUNLOCK(object); } else { if (runpg < (count - first)) { VM_OBJECT_WLOCK(object); for (i = first + runpg; i < count; i++) { vm_page_lock(m[i]); vm_page_free(m[i]); vm_page_unlock(m[i]); } VM_OBJECT_WUNLOCK(object); count = first + runpg; } break; } first = runend; } /* * the first and last page have been calculated now, move input pages * to be zero based... */ if (first != 0) { m += first; count -= first; reqpage -= first; } /* * calculate the file virtual address for the transfer */ foff = IDX_TO_OFF(m[0]->pindex); /* * calculate the size of the transfer */ size = count * PAGE_SIZE; KASSERT(count > 0, ("zero count")); if ((foff + size) > object->un_pager.vnp.vnp_size) size = object->un_pager.vnp.vnp_size - foff; KASSERT(size > 0, ("zero size")); /* * round up physical size for real devices. */ if (1) { int secmask = bo->bo_bsize - 1; KASSERT(secmask < PAGE_SIZE && secmask > 0, ("vnode_pager_generic_getpages: sector size %d too large", secmask + 1)); size = (size + secmask) & ~secmask; } - bp = getpbuf(&vnode_pbuf_freecnt); bp->b_kvaalloc = bp->b_data; /* * and map the pages to be read into the kva, if the filesystem * requires mapped buffers. */ if ((vp->v_mount->mnt_kern_flag & MNTK_UNMAPPED_BUFS) != 0 && unmapped_buf_allowed) { bp->b_data = unmapped_buf; bp->b_kvabase = unmapped_buf; bp->b_offset = 0; bp->b_flags |= B_UNMAPPED; } else pmap_qenter((vm_offset_t)bp->b_kvaalloc, m, count); /* build a minimal buffer header */ bp->b_iocmd = BIO_READ; KASSERT(bp->b_rcred == NOCRED, ("leaking read ucred")); KASSERT(bp->b_wcred == NOCRED, ("leaking write ucred")); bp->b_rcred = crhold(curthread->td_ucred); bp->b_wcred = crhold(curthread->td_ucred); bp->b_blkno = firstaddr; pbgetbo(bo, bp); bp->b_vp = vp; bp->b_bcount = size; bp->b_bufsize = size; bp->b_runningbufspace = bp->b_bufsize; for (i = 0; i < count; i++) bp->b_pages[i] = m[i]; bp->b_npages = count; bp->b_pager.pg_reqpage = reqpage; atomic_add_long(&runningbufspace, bp->b_runningbufspace); PCPU_INC(cnt.v_vnodein); PCPU_ADD(cnt.v_vnodepgsin, count); /* do the input */ bp->b_iooffset = dbtob(bp->b_blkno); if (iodone != NULL) { /* async */ bp->b_pager.pg_iodone = iodone; bp->b_caller1 = arg; bp->b_iodone = vnode_pager_generic_getpages_done_async; bp->b_flags |= B_ASYNC; BUF_KERNPROC(bp); bstrategy(bp); /* Good bye! */ } else { bp->b_iodone = bdone; bstrategy(bp); bwait(bp, PVM, "vnread"); error = vnode_pager_generic_getpages_done(bp); for (i = 0; i < bp->b_npages; i++) bp->b_pages[i] = NULL; bp->b_vp = NULL; pbrelbo(bp); relpbuf(bp, &vnode_pbuf_freecnt); } return (error != 0 ? VM_PAGER_ERROR : VM_PAGER_OK); } static void vnode_pager_generic_getpages_done_async(struct buf *bp) { int error; error = vnode_pager_generic_getpages_done(bp); bp->b_pager.pg_iodone(bp->b_caller1, bp->b_pages, bp->b_pager.pg_reqpage, error); for (int i = 0; i < bp->b_npages; i++) bp->b_pages[i] = NULL; bp->b_vp = NULL; pbrelbo(bp); - relpbuf(bp, &vnode_pbuf_freecnt); + relpbuf(bp, &vnode_async_pbuf_freecnt); } static int vnode_pager_generic_getpages_done(struct buf *bp) { vm_object_t object; off_t tfoff, nextoff; int i, error; error = (bp->b_ioflags & BIO_ERROR) != 0 ? EIO : 0; object = bp->b_vp->v_object; if (error == 0 && bp->b_bcount != bp->b_npages * PAGE_SIZE) { if ((bp->b_flags & B_UNMAPPED) != 0) { bp->b_flags &= ~B_UNMAPPED; pmap_qenter((vm_offset_t)bp->b_kvaalloc, bp->b_pages, bp->b_npages); } bzero(bp->b_kvaalloc + bp->b_bcount, PAGE_SIZE * bp->b_npages - bp->b_bcount); } if ((bp->b_flags & B_UNMAPPED) == 0) pmap_qremove((vm_offset_t)bp->b_kvaalloc, bp->b_npages); if ((bp->b_vp->v_mount->mnt_kern_flag & MNTK_UNMAPPED_BUFS) != 0) { bp->b_data = bp->b_kvaalloc; bp->b_kvabase = bp->b_kvaalloc; bp->b_flags &= ~B_UNMAPPED; } VM_OBJECT_WLOCK(object); for (i = 0, tfoff = IDX_TO_OFF(bp->b_pages[0]->pindex); i < bp->b_npages; i++, tfoff = nextoff) { vm_page_t mt; nextoff = tfoff + PAGE_SIZE; mt = bp->b_pages[i]; if (nextoff <= object->un_pager.vnp.vnp_size) { /* * Read filled up entire page. */ mt->valid = VM_PAGE_BITS_ALL; KASSERT(mt->dirty == 0, ("%s: page %p is dirty", __func__, mt)); KASSERT(!pmap_page_is_mapped(mt), ("%s: page %p is mapped", __func__, mt)); } else { /* * Read did not fill up entire page. * * Currently we do not set the entire page valid, * we just try to clear the piece that we couldn't * read. */ vm_page_set_valid_range(mt, 0, object->un_pager.vnp.vnp_size - tfoff); KASSERT((mt->dirty & vm_page_bits(0, object->un_pager.vnp.vnp_size - tfoff)) == 0, ("%s: page %p is dirty", __func__, mt)); } if (i != bp->b_pager.pg_reqpage) vm_page_readahead_finish(mt); } VM_OBJECT_WUNLOCK(object); if (error != 0) printf("%s: I/O read error %d\n", __func__, error); return (error); } /* * EOPNOTSUPP is no longer legal. For local media VFS's that do not * implement their own VOP_PUTPAGES, their VOP_PUTPAGES should call to * vnode_pager_generic_putpages() to implement the previous behaviour. * * All other FS's should use the bypass to get to the local media * backing vp's VOP_PUTPAGES. */ static void vnode_pager_putpages(vm_object_t object, vm_page_t *m, int count, int flags, int *rtvals) { int rtval; struct vnode *vp; int bytes = count * PAGE_SIZE; /* * Force synchronous operation if we are extremely low on memory * to prevent a low-memory deadlock. VOP operations often need to * allocate more memory to initiate the I/O ( i.e. do a BMAP * operation ). The swapper handles the case by limiting the amount * of asynchronous I/O, but that sort of solution doesn't scale well * for the vnode pager without a lot of work. * * Also, the backing vnode's iodone routine may not wake the pageout * daemon up. This should be probably be addressed XXX. */ if (vm_cnt.v_free_count + vm_cnt.v_cache_count < vm_cnt.v_pageout_free_min) flags |= VM_PAGER_PUT_SYNC; /* * Call device-specific putpages function */ vp = object->handle; VM_OBJECT_WUNLOCK(object); rtval = VOP_PUTPAGES(vp, m, bytes, flags, rtvals); KASSERT(rtval != EOPNOTSUPP, ("vnode_pager: stale FS putpages\n")); VM_OBJECT_WLOCK(object); } /* * This is now called from local media FS's to operate against their * own vnodes if they fail to implement VOP_PUTPAGES. * * This is typically called indirectly via the pageout daemon and * clustering has already typically occured, so in general we ask the * underlying filesystem to write the data out asynchronously rather * then delayed. */ int vnode_pager_generic_putpages(struct vnode *vp, vm_page_t *ma, int bytecount, int flags, int *rtvals) { int i; vm_object_t object; vm_page_t m; int count; int maxsize, ncount; vm_ooffset_t poffset; struct uio auio; struct iovec aiov; int error; int ioflags; int ppscheck = 0; static struct timeval lastfail; static int curfail; object = vp->v_object; count = bytecount / PAGE_SIZE; for (i = 0; i < count; i++) rtvals[i] = VM_PAGER_ERROR; if ((int64_t)ma[0]->pindex < 0) { printf("vnode_pager_putpages: attempt to write meta-data!!! -- 0x%lx(%lx)\n", (long)ma[0]->pindex, (u_long)ma[0]->dirty); rtvals[0] = VM_PAGER_BAD; return VM_PAGER_BAD; } maxsize = count * PAGE_SIZE; ncount = count; poffset = IDX_TO_OFF(ma[0]->pindex); /* * If the page-aligned write is larger then the actual file we * have to invalidate pages occuring beyond the file EOF. However, * there is an edge case where a file may not be page-aligned where * the last page is partially invalid. In this case the filesystem * may not properly clear the dirty bits for the entire page (which * could be VM_PAGE_BITS_ALL due to the page having been mmap()d). * With the page locked we are free to fix-up the dirty bits here. * * We do not under any circumstances truncate the valid bits, as * this will screw up bogus page replacement. */ VM_OBJECT_WLOCK(object); if (maxsize + poffset > object->un_pager.vnp.vnp_size) { if (object->un_pager.vnp.vnp_size > poffset) { int pgoff; maxsize = object->un_pager.vnp.vnp_size - poffset; ncount = btoc(maxsize); if ((pgoff = (int)maxsize & PAGE_MASK) != 0) { /* * If the object is locked and the following * conditions hold, then the page's dirty * field cannot be concurrently changed by a * pmap operation. */ m = ma[ncount - 1]; vm_page_assert_sbusied(m); KASSERT(!pmap_page_is_write_mapped(m), ("vnode_pager_generic_putpages: page %p is not read-only", m)); vm_page_clear_dirty(m, pgoff, PAGE_SIZE - pgoff); } } else { maxsize = 0; ncount = 0; } if (ncount < count) { for (i = ncount; i < count; i++) { rtvals[i] = VM_PAGER_BAD; } } } VM_OBJECT_WUNLOCK(object); /* * pageouts are already clustered, use IO_ASYNC to force a bawrite() * rather then a bdwrite() to prevent paging I/O from saturating * the buffer cache. Dummy-up the sequential heuristic to cause * large ranges to cluster. If neither IO_SYNC or IO_ASYNC is set, * the system decides how to cluster. */ ioflags = IO_VMIO; if (flags & (VM_PAGER_PUT_SYNC | VM_PAGER_PUT_INVAL)) ioflags |= IO_SYNC; else if ((flags & VM_PAGER_CLUSTER_OK) == 0) ioflags |= IO_ASYNC; ioflags |= (flags & VM_PAGER_PUT_INVAL) ? IO_INVAL: 0; ioflags |= IO_SEQMAX << IO_SEQSHIFT; aiov.iov_base = (caddr_t) 0; aiov.iov_len = maxsize; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_offset = poffset; auio.uio_segflg = UIO_NOCOPY; auio.uio_rw = UIO_WRITE; auio.uio_resid = maxsize; auio.uio_td = (struct thread *) 0; error = VOP_WRITE(vp, &auio, ioflags, curthread->td_ucred); PCPU_INC(cnt.v_vnodeout); PCPU_ADD(cnt.v_vnodepgsout, ncount); if (error) { if ((ppscheck = ppsratecheck(&lastfail, &curfail, 1))) printf("vnode_pager_putpages: I/O error %d\n", error); } if (auio.uio_resid) { if (ppscheck || ppsratecheck(&lastfail, &curfail, 1)) printf("vnode_pager_putpages: residual I/O %zd at %lu\n", auio.uio_resid, (u_long)ma[0]->pindex); } for (i = 0; i < ncount; i++) { rtvals[i] = VM_PAGER_OK; } return rtvals[0]; } void vnode_pager_undirty_pages(vm_page_t *ma, int *rtvals, int written) { vm_object_t obj; int i, pos; if (written == 0) return; obj = ma[0]->object; VM_OBJECT_WLOCK(obj); for (i = 0, pos = 0; pos < written; i++, pos += PAGE_SIZE) { if (pos < trunc_page(written)) { rtvals[i] = VM_PAGER_OK; vm_page_undirty(ma[i]); } else { /* Partially written page. */ rtvals[i] = VM_PAGER_AGAIN; vm_page_clear_dirty(ma[i], 0, written & PAGE_MASK); } } VM_OBJECT_WUNLOCK(obj); } void vnode_pager_update_writecount(vm_object_t object, vm_offset_t start, vm_offset_t end) { struct vnode *vp; vm_ooffset_t old_wm; VM_OBJECT_WLOCK(object); if (object->type != OBJT_VNODE) { VM_OBJECT_WUNLOCK(object); return; } old_wm = object->un_pager.vnp.writemappings; object->un_pager.vnp.writemappings += (vm_ooffset_t)end - start; vp = object->handle; if (old_wm == 0 && object->un_pager.vnp.writemappings != 0) { ASSERT_VOP_ELOCKED(vp, "v_writecount inc"); VOP_ADD_WRITECOUNT(vp, 1); CTR3(KTR_VFS, "%s: vp %p v_writecount increased to %d", __func__, vp, vp->v_writecount); } else if (old_wm != 0 && object->un_pager.vnp.writemappings == 0) { ASSERT_VOP_ELOCKED(vp, "v_writecount dec"); VOP_ADD_WRITECOUNT(vp, -1); CTR3(KTR_VFS, "%s: vp %p v_writecount decreased to %d", __func__, vp, vp->v_writecount); } VM_OBJECT_WUNLOCK(object); } void vnode_pager_release_writecount(vm_object_t object, vm_offset_t start, vm_offset_t end) { struct vnode *vp; struct mount *mp; vm_offset_t inc; VM_OBJECT_WLOCK(object); /* * First, recheck the object type to account for the race when * the vnode is reclaimed. */ if (object->type != OBJT_VNODE) { VM_OBJECT_WUNLOCK(object); return; } /* * Optimize for the case when writemappings is not going to * zero. */ inc = end - start; if (object->un_pager.vnp.writemappings != inc) { object->un_pager.vnp.writemappings -= inc; VM_OBJECT_WUNLOCK(object); return; } vp = object->handle; vhold(vp); VM_OBJECT_WUNLOCK(object); mp = NULL; vn_start_write(vp, &mp, V_WAIT); vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); /* * Decrement the object's writemappings, by swapping the start * and end arguments for vnode_pager_update_writecount(). If * there was not a race with vnode reclaimation, then the * vnode's v_writecount is decremented. */ vnode_pager_update_writecount(object, end, start); VOP_UNLOCK(vp, 0); vdrop(vp); if (mp != NULL) vn_finished_write(mp); } Index: projects/cxl_iscsi/sys =================================================================== --- projects/cxl_iscsi/sys (revision 279716) +++ projects/cxl_iscsi/sys (revision 279717) Property changes on: projects/cxl_iscsi/sys ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys:r279555-279716 Index: projects/cxl_iscsi/usr.bin/calendar/calcpp.c =================================================================== --- projects/cxl_iscsi/usr.bin/calendar/calcpp.c (revision 279716) +++ projects/cxl_iscsi/usr.bin/calendar/calcpp.c (nonexistent) @@ -1,232 +0,0 @@ -/*- - * Copyright (c) 2013 Diane Bruce - * All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * $FreeBSD$ - */ - -/* calendar fake cpp does a very limited cpp version */ - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include "pathnames.h" -#include "calendar.h" - -#define MAXFPSTACK 50 -static FILE *fpstack[MAXFPSTACK]; -static int curfpi; -static void pushfp(FILE *fp); -static FILE *popfp(void); -static int tokenscpp(char *buf, char *string); - -#define T_INVALID -1 -#define T_INCLUDE 0 -#define T_DEFINE 1 -#define T_IFNDEF 2 -#define T_ENDIF 3 - -#define MAXSYMS 100 -static char *symtable[MAXSYMS]; -static void addsym(const char *name); -static int findsym(const char *name); - -FILE * -fincludegets(char *buf, int size, FILE *fp) -{ - char name[MAXPATHLEN]; - FILE *nfp=NULL; - char *p; - int ch; - - if (fp == NULL) - return(NULL); - - if (fgets(buf, size, fp) == NULL) { - *buf = '\0'; - fclose(fp); - fp = popfp(); - return (fp); - } - if ((p = strchr(buf, '\n')) != NULL) - *p = '\0'; - else { - /* Flush this line */ - while ((ch = fgetc(fp)) != '\n' && ch != EOF); - if (ch == EOF) { - *buf = '\0'; - fclose(fp); - fp = popfp(); - return(fp); - } - } - switch (tokenscpp(buf, name)) { - case T_INCLUDE: - *buf = '\0'; - if ((nfp = fopen(name, "r")) != NULL) { - pushfp(fp); - fp = nfp; - } - break; - case T_DEFINE: - addsym(name); - break; - case T_IFNDEF: - if (findsym(name)) { - fclose(fp); - fp = popfp(); - *buf = '\0'; - } - break; - case T_ENDIF: - *buf = '\0'; - break; - default: - break; - } - return (fp); -} - -static int -tokenscpp(char *buf, char *string) -{ - char *p; - char *s; - - if ((p = strstr(buf, "#define")) != NULL) { - p += 8; - while (isspace((unsigned char)*p)) - p++; - s = p; - while(!isspace((unsigned char)*p)) - p++; - strlcpy(string, s, MAXPATHLEN); - return(T_DEFINE); - } else if ((p = strstr(buf, "#ifndef")) != NULL) { - p += 8; - while (isspace((unsigned char)*p)) - p++; - s = p; - while(!isspace((unsigned char)*p)) - p++; - *p = '\0'; - strncpy(string, s, MAXPATHLEN); - return(T_IFNDEF); - } else if ((p = strstr(buf, "#endif")) != NULL) { - return(T_ENDIF); - } if ((p = strstr(buf, "#include")) != NULL) { - p += 8; - while (isspace((unsigned char)*p)) - p++; - if (*p == '<') { - s = p+1; - if ((p = strchr(s, '>')) != NULL) - *p = '\0'; - if (*s != '/') - snprintf (string, MAXPATHLEN, "%s/%s", - _PATH_INCLUDE, s); - else - strncpy(string, s, MAXPATHLEN); - } else if (*p == '(') { - s = p+1; - if ((p = strchr(p, '>')) != NULL) - *p = '\0'; - snprintf (string, MAXPATHLEN, "%s", s); - } - return(T_INCLUDE); - } - return(T_INVALID); -} - -static void -pushfp(FILE *fp) -{ - curfpi++; - if (curfpi == MAXFPSTACK) - errx(1, "Max #include reached"); - fpstack[curfpi] = fp; -} - -static -FILE *popfp(void) -{ - FILE *tmp; - - assert(curfpi >= 0); - tmp = fpstack[curfpi]; - curfpi--; - return(tmp); -} - -void -initcpp(void) -{ - int i; - - for (i=0; i < MAXSYMS; i++) - symtable[i] = NULL; - fpstack[0] = NULL; - curfpi = 0; -} - - -static void -addsym(const char *name) -{ - int i; - - if (!findsym(name)) - for (i=0; i < MAXSYMS; i++) { - if (symtable[i] == NULL) { - symtable[i] = strdup(name); - if (symtable[i] == NULL) - errx(1, "malloc error in addsym"); - return; - } - } - errx(1, "symbol table full\n"); -} - -static int -findsym(const char *name) -{ - int i; - - for (i=0; i < MAXSYMS; i++) - if (symtable[i] != NULL && strcmp(symtable[i],name) == 0) - return (1); - return (0); -} Property changes on: projects/cxl_iscsi/usr.bin/calendar/calcpp.c ___________________________________________________________________ Deleted: svn:eol-style ## -1 +0,0 ## -native \ No newline at end of property Deleted: svn:keywords ## -1 +0,0 ## -FreeBSD=%H \ No newline at end of property Deleted: svn:mime-type ## -1 +0,0 ## -text/plain \ No newline at end of property Index: projects/cxl_iscsi/usr.bin/calendar/Makefile =================================================================== --- projects/cxl_iscsi/usr.bin/calendar/Makefile (revision 279716) +++ projects/cxl_iscsi/usr.bin/calendar/Makefile (revision 279717) @@ -1,38 +1,38 @@ # @(#)Makefile 8.1 (Berkeley) 6/6/93 # $FreeBSD$ .include PROG= calendar SRCS= calendar.c locale.c events.c dates.c parsedata.c io.c day.c \ - ostern.c paskha.c pom.c sunpos.c calcpp.c + ostern.c paskha.c pom.c sunpos.c LIBADD= m INTER= de_AT.ISO_8859-15 de_DE.ISO8859-1 fr_FR.ISO8859-1 \ hr_HR.ISO8859-2 hu_HU.ISO8859-2 pt_BR.ISO8859-1 \ pt_BR.UTF-8 ru_RU.KOI8-R ru_RU.UTF-8 uk_UA.KOI8-U DE_LINKS= de_DE.ISO8859-15 FR_LINKS= fr_FR.ISO8859-15 TEXTMODE?= 444 beforeinstall: ${INSTALL} -o ${BINOWN} -g ${BINGRP} -m ${TEXTMODE} \ ${.CURDIR}/calendars/calendar.* ${DESTDIR}${SHAREDIR}/calendar .for lang in ${INTER} ${INSTALL} -o ${BINOWN} -g ${BINGRP} -m ${TEXTMODE} \ ${.CURDIR}/calendars/${lang}/calendar.* \ ${DESTDIR}${SHAREDIR}/calendar/${lang} .endfor .for link in ${DE_LINKS} rm -rf ${DESTDIR}${SHAREDIR}/calendar/${link} ln -s de_DE.ISO8859-1 ${DESTDIR}${SHAREDIR}/calendar/${link} .endfor .for link in ${FR_LINKS} rm -rf ${DESTDIR}${SHAREDIR}/calendar/${link} ln -s fr_FR.ISO8859-1 ${DESTDIR}${SHAREDIR}/calendar/${link} .endfor .if ${MK_TESTS} != "no" SUBDIR+= tests .endif .include Index: projects/cxl_iscsi/usr.bin/calendar/calendar.h =================================================================== --- projects/cxl_iscsi/usr.bin/calendar/calendar.h (revision 279716) +++ projects/cxl_iscsi/usr.bin/calendar/calendar.h (revision 279717) @@ -1,201 +1,197 @@ /*- * Copyright (c) 1989, 1993, 1994 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include #include #define SECSPERDAY (24 * 60 * 60) #define SECSPERHOUR (60 * 60) #define SECSPERMINUTE (60) #define MINSPERHOUR (60) #define HOURSPERDAY (24) #define FSECSPERDAY (24.0 * 60.0 * 60.0) #define FSECSPERHOUR (60.0 * 60.0) #define FSECSPERMINUTE (60.0) #define FMINSPERHOUR (60.0) #define FHOURSPERDAY (24.0) #define DAYSPERYEAR 365 #define DAYSPERLEAPYEAR 366 /* Not yet categorized */ extern struct passwd *pw; extern int doall; extern time_t t1, t2; extern const char *calendarFile; extern int yrdays; extern struct fixs neaster, npaskha, ncny, nfullmoon, nnewmoon; extern struct fixs nmarequinox, nsepequinox, njunsolstice, ndecsolstice; extern double UTCOffset; extern int EastLongitude; #define isleap(y) ((((y) % 4) == 0 && ((y) % 100) != 0) || ((y) % 400) == 0) /* Flags to determine the returned values by determinestyle() in parsedata.c */ #define F_NONE 0x00000 #define F_MONTH 0x00001 #define F_DAYOFWEEK 0x00002 #define F_DAYOFMONTH 0x00004 #define F_MODIFIERINDEX 0x00008 #define F_MODIFIEROFFSET 0x00010 #define F_SPECIALDAY 0x00020 #define F_ALLMONTH 0x00040 #define F_ALLDAY 0x00080 #define F_VARIABLE 0x00100 #define F_EASTER 0x00200 #define F_CNY 0x00400 #define F_PASKHA 0x00800 #define F_NEWMOON 0x01000 #define F_FULLMOON 0x02000 #define F_MAREQUINOX 0x04000 #define F_SEPEQUINOX 0x08000 #define F_JUNSOLSTICE 0x10000 #define F_DECSOLSTICE 0x20000 #define F_YEAR 0x40000 #define STRING_EASTER "Easter" #define STRING_PASKHA "Paskha" #define STRING_CNY "ChineseNewYear" #define STRING_NEWMOON "NewMoon" #define STRING_FULLMOON "FullMoon" #define STRING_MAREQUINOX "MarEquinox" #define STRING_SEPEQUINOX "SepEquinox" #define STRING_JUNSOLSTICE "JunSolstice" #define STRING_DECSOLSTICE "DecSolstice" #define MAXCOUNT 125 /* Random number of maximum number of * repeats of an event. Should be 52 * (number of weeks per year), if you * want to show two years then it * should be 104. If you are seeing * more than this you are using this * program wrong. */ /* * All the astronomical calculations are carried out for the meridian 120 * degrees east of Greenwich. */ #define UTCOFFSET_CNY 8.0 extern int debug; /* show parsing of the input */ extern int year1, year2; /* events.c */ /* * Event sorting related functions: * - Use event_add() to create a new event * - Use event_continue() to add more text to the last added event * - Use event_print_all() to display them in time chronological order */ struct event *event_add(int, int, int, char *, int, char *, char *); void event_continue(struct event *events, char *txt); void event_print_all(FILE *fp); struct event { int year; int month; int day; int var; char *date; char *text; char *extra; struct event *next; }; /* locale.c */ struct fixs { char *name; size_t len; }; extern const char *days[]; extern const char *fdays[]; extern const char *fmonths[]; extern const char *months[]; extern const char *sequences[]; extern struct fixs fndays[8]; /* full national days names */ extern struct fixs fnmonths[13]; /* full national months names */ extern struct fixs ndays[8]; /* short national days names */ extern struct fixs nmonths[13]; /* short national month names */ extern struct fixs nsequences[10]; void setnnames(void); void setnsequences(char *); /* day.c */ extern const struct tm tm0; extern char dayname[]; void settimes(time_t,int before, int after, int friday, struct tm *tp1, struct tm *tp2); time_t Mktime(char *); /* parsedata.c */ int parsedaymonth(char *, int *, int *, int *, int *, char **); void dodebug(char *type); /* io.c */ void cal(void); void closecal(FILE *); FILE *opencalin(void); FILE *opencalout(void); -/* calcpp.c */ -void initcpp(void); -FILE *fincludegets(char *buf, int size, FILE *fp); - /* ostern.c / paskha.c */ int paskha(int); int easter(int); int j2g(int); /* dates.c */ extern int cumdaytab[][14]; extern int monthdaytab[][14]; extern int debug_remember; void generatedates(struct tm *tp1, struct tm *tp2); void dumpdates(void); int remember_ymd(int y, int m, int d); int remember_yd(int y, int d, int *rm, int *rd); int first_dayofweek_of_year(int y); int first_dayofweek_of_month(int y, int m); int walkthrough_dates(struct event **e); void addtodate(struct event *e, int year, int month, int day); /* pom.c */ #define MAXMOONS 18 void pom(int year, double UTCoffset, int *fms, int *nms); void fpom(int year, double utcoffset, double *ffms, double *fnms); /* sunpos.c */ void equinoxsolstice(int year, double UTCoffset, int *equinoxdays, int *solsticedays); void fequinoxsolstice(int year, double UTCoffset, double *equinoxdays, double *solsticedays); int calculatesunlongitude30(int year, int degreeGMToffset, int *ichinesemonths); Index: projects/cxl_iscsi/usr.bin/calendar/io.c =================================================================== --- projects/cxl_iscsi/usr.bin/calendar/io.c (revision 279716) +++ projects/cxl_iscsi/usr.bin/calendar/io.c (revision 279717) @@ -1,334 +1,501 @@ /*- * Copyright (c) 1989, 1993, 1994 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef lint static const char copyright[] = "@(#) Copyright (c) 1989, 1993\n\ The Regents of the University of California. All rights reserved.\n"; #endif #if 0 #ifndef lint static char sccsid[] = "@(#)calendar.c 8.3 (Berkeley) 3/25/94"; #endif #endif #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include +#include +#define _WITH_GETLINE #include #include #include +#include #include #include "pathnames.h" #include "calendar.h" +enum { + T_OK = 0, + T_ERR, + T_PROCESS, +}; + const char *calendarFile = "calendar"; /* default calendar file */ static const char *calendarHomes[] = {".calendar", _PATH_INCLUDE}; /* HOME */ static const char *calendarNoMail = "nomail";/* don't sent mail if file exist */ static char path[MAXPATHLEN]; struct fixs neaster, npaskha, ncny, nfullmoon, nnewmoon; struct fixs nmarequinox, nsepequinox, njunsolstice, ndecsolstice; +static int cal_parse(FILE *in, FILE *out); + +static StringList *definitions = NULL; +static struct event *events[MAXCOUNT]; +static char *extradata[MAXCOUNT]; + +static void +trimlr(char **buf) +{ + char *walk = *buf; + + while (isspace(*walk)) + walk++; + while (isspace(walk[strlen(walk) -1])) + walk[strlen(walk) -1] = '\0'; + + *buf = walk; +} + +static FILE * +cal_fopen(const char *file) +{ + FILE *fp; + char *home = getenv("HOME"); + unsigned int i; + + if (home == NULL || *home == '\0') { + warnx("Cannot get home directory"); + return (NULL); + } + + if (chdir(home) != 0) { + warnx("Cannot enter home directory"); + return (NULL); + } + + for (i = 0; i < sizeof(calendarHomes)/sizeof(calendarHomes[0]) ; i++) { + if (chdir(calendarHomes[i]) != 0) + continue; + + if ((fp = fopen(file, "r")) != NULL) + return (fp); + } + + warnx("can't open calendar file \"%s\"", file); + + return (NULL); +} + +static int +token(char *line, FILE *out, bool *skip) +{ + char *walk, c, a; + + if (strncmp(line, "endif", 5) == 0) { + *skip = false; + return (T_OK); + } + + if (*skip) + return (T_OK); + + if (strncmp(line, "include", 7) == 0) { + walk = line + 7; + + trimlr(&walk); + + if (*walk == '\0') { + warnx("Expecting arguments after #include"); + return (T_ERR); + } + + if (*walk != '<' && *walk != '\"') { + warnx("Excecting '<' or '\"' after #include"); + return (T_ERR); + } + + a = *walk; + walk++; + c = walk[strlen(walk) - 1]; + + switch(c) { + case '>': + if (a != '<') { + warnx("Unterminated include expecting '\"'"); + return (T_ERR); + } + break; + case '\"': + if (a != '\"') { + warnx("Unterminated include expecting '>'"); + return (T_ERR); + } + break; + default: + warnx("Unterminated include expecting '%c'", + a == '<' ? '>' : '\"' ); + return (T_ERR); + } + walk[strlen(walk) - 1] = '\0'; + + if (cal_parse(cal_fopen(walk), out)) + return (T_ERR); + + return (T_OK); + } + + if (strncmp(line, "define", 6) == 0) { + if (definitions == NULL) + definitions = sl_init(); + walk = line + 6; + trimlr(&walk); + + if (*walk == '\0') { + warnx("Expecting arguments after #define"); + return (T_ERR); + } + + sl_add(definitions, strdup(walk)); + return (T_OK); + } + + if (strncmp(line, "ifndef", 6) == 0) { + walk = line + 6; + trimlr(&walk); + + if (*walk == '\0') { + warnx("Expecting arguments after #ifndef"); + return (T_ERR); + } + + if (definitions != NULL && sl_find(definitions, walk) != NULL) + *skip = true; + + return (T_OK); + } + + return (T_PROCESS); + +} + #define REPLACE(string, slen, struct_) \ if (strncasecmp(buf, (string), (slen)) == 0 && buf[(slen)]) { \ if (struct_.name != NULL) \ free(struct_.name); \ if ((struct_.name = strdup(buf + (slen))) == NULL) \ errx(1, "cannot allocate memory"); \ struct_.len = strlen(buf + (slen)); \ continue; \ } -void -cal(void) +static int +cal_parse(FILE *in, FILE *out) { - char *pp, p; - FILE *fpin; - FILE *fpout; - int l; - int count, i; + char *line = NULL; + char *buf; + size_t linecap = 0; + ssize_t linelen; + ssize_t l; + static int d_first = -1; + static int count = 0; + int i; int month[MAXCOUNT]; int day[MAXCOUNT]; int year[MAXCOUNT]; - char **extradata; /* strings of 20 length */ - int flags; - static int d_first = -1; - char buf[2048 + 1]; - struct event *events[MAXCOUNT]; - struct tm tm; + bool skip = false; char dbuf[80]; + char *pp, p; + struct tm tm; + int flags; - initcpp(); - extradata = (char **)calloc(MAXCOUNT, sizeof(char *)); - for (i = 0; i < MAXCOUNT; i++) { - extradata[i] = (char *)calloc(1, 20); - } - /* Unused */ tm.tm_sec = 0; tm.tm_min = 0; tm.tm_hour = 0; tm.tm_wday = 0; - count = 0; - if ((fpin = opencalin()) == NULL) { - free(extradata); - return; - } - if ((fpout = opencalout()) == NULL) { - fclose(fpin); - free(extradata); - return; - } - while ((fpin = fincludegets(buf, sizeof(buf), fpin)) != NULL) { - if (*buf == '\0') + if (in == NULL) + return (1); + + while ((linelen = getline(&line, &linecap, in)) > 0) { + if (linelen == 0) continue; - for (l = strlen(buf); + + if (*line == '#') { + switch (token(line+1, out, &skip)) { + case T_ERR: + free(line); + return (1); + case T_OK: + continue; + case T_PROCESS: + break; + default: + break; + } + } + + if (skip) + continue; + + buf = line; + for (l = linelen; l > 0 && isspace((unsigned char)buf[l - 1]); l--) ; buf[l] = '\0'; if (buf[0] == '\0') continue; /* Parse special definitions: LANG, Easter, Paskha etc */ if (strncmp(buf, "LANG=", 5) == 0) { (void)setlocale(LC_ALL, buf + 5); d_first = (*nl_langinfo(D_MD_ORDER) == 'd'); setnnames(); continue; } REPLACE("Easter=", 7, neaster); REPLACE("Paskha=", 7, npaskha); REPLACE("ChineseNewYear=", 15, ncny); REPLACE("NewMoon=", 8, nnewmoon); REPLACE("FullMoon=", 9, nfullmoon); REPLACE("MarEquinox=", 11, nmarequinox); REPLACE("SepEquinox=", 11, nsepequinox); REPLACE("JunSolstice=", 12, njunsolstice); REPLACE("DecSolstice=", 12, ndecsolstice); if (strncmp(buf, "SEQUENCE=", 9) == 0) { setnsequences(buf + 9); continue; } /* * If the line starts with a tab, the data has to be * added to the previous line */ if (buf[0] == '\t') { for (i = 0; i < count; i++) event_continue(events[i], buf); continue; } /* Get rid of leading spaces (non-standard) */ while (isspace((unsigned char)buf[0])) memcpy(buf, buf + 1, strlen(buf)); /* No tab in the line, then not a valid line */ if ((pp = strchr(buf, '\t')) == NULL) continue; /* Trim spaces in front of the tab */ while (isspace((unsigned char)pp[-1])) pp--; p = *pp; *pp = '\0'; if ((count = parsedaymonth(buf, year, month, day, &flags, extradata)) == 0) continue; *pp = p; if (count < 0) { /* Show error status based on return value */ if (debug) fprintf(stderr, "Ignored: %s\n", buf); if (count == -1) continue; count = -count + 1; } /* Find the last tab */ while (pp[1] == '\t') pp++; if (d_first < 0) d_first = (*nl_langinfo(D_MD_ORDER) == 'd'); for (i = 0; i < count; i++) { tm.tm_mon = month[i] - 1; tm.tm_mday = day[i]; tm.tm_year = year[i] - 1900; (void)strftime(dbuf, sizeof(dbuf), d_first ? "%e %b" : "%b %e", &tm); if (debug) fprintf(stderr, "got %s\n", pp); events[i] = event_add(year[i], month[i], day[i], dbuf, ((flags &= F_VARIABLE) != 0) ? 1 : 0, pp, extradata[i]); } } + free(line); + fclose(in); + + return (0); +} + +void +cal(void) +{ + FILE *fpin; + FILE *fpout; + int i; + + for (i = 0; i < MAXCOUNT; i++) + extradata[i] = (char *)calloc(1, 20); + + + if ((fpin = opencalin()) == NULL) + return; + + if ((fpout = opencalout()) == NULL) { + fclose(fpin); + return; + } + + if (cal_parse(fpin, fpout)) + return; + event_print_all(fpout); closecal(fpout); - free(extradata); } FILE * opencalin(void) { - size_t i; - int found; struct stat sbuf; FILE *fpin; /* open up calendar file */ if ((fpin = fopen(calendarFile, "r")) == NULL) { if (doall) { if (chdir(calendarHomes[0]) != 0) return (NULL); if (stat(calendarNoMail, &sbuf) == 0) return (NULL); if ((fpin = fopen(calendarFile, "r")) == NULL) return (NULL); } else { - char *home = getenv("HOME"); - if (home == NULL || *home == '\0') - errx(1, "cannot get home directory"); - if (chdir(home) != 0) - errx(1, "cannot enter home directory"); - for (found = i = 0; i < sizeof(calendarHomes) / - sizeof(calendarHomes[0]); i++) - if (chdir(calendarHomes[i]) == 0 && - (fpin = fopen(calendarFile, "r")) != NULL) { - found = 1; - break; - } - if (!found) - errx(1, - "can't open calendar file \"%s\": %s (%d)", - calendarFile, strerror(errno), errno); + fpin = cal_fopen(calendarFile); } } return (fpin); } FILE * opencalout(void) { int fd; /* not reading all calendar files, just set output to stdout */ if (!doall) return (stdout); /* set output to a temporary file, so if no output don't send mail */ snprintf(path, sizeof(path), "%s/_calXXXXXX", _PATH_TMP); if ((fd = mkstemp(path)) < 0) return (NULL); return (fdopen(fd, "w+")); } void closecal(FILE *fp) { uid_t uid; struct stat sbuf; int nread, pdes[2], status; char buf[1024]; if (!doall) return; rewind(fp); if (fstat(fileno(fp), &sbuf) || !sbuf.st_size) goto done; if (pipe(pdes) < 0) goto done; switch (fork()) { case -1: /* error */ (void)close(pdes[0]); (void)close(pdes[1]); goto done; case 0: /* child -- set stdin to pipe output */ if (pdes[0] != STDIN_FILENO) { (void)dup2(pdes[0], STDIN_FILENO); (void)close(pdes[0]); } (void)close(pdes[1]); uid = geteuid(); if (setuid(getuid()) < 0) { warnx("setuid failed"); _exit(1); }; if (setgid(getegid()) < 0) { warnx("setgid failed"); _exit(1); } if (setuid(uid) < 0) { warnx("setuid failed"); _exit(1); } execl(_PATH_SENDMAIL, "sendmail", "-i", "-t", "-F", "\"Reminder Service\"", (char *)NULL); warn(_PATH_SENDMAIL); _exit(1); } /* parent -- write to pipe input */ (void)close(pdes[0]); write(pdes[1], "From: \"Reminder Service\" <", 26); write(pdes[1], pw->pw_name, strlen(pw->pw_name)); write(pdes[1], ">\nTo: <", 7); write(pdes[1], pw->pw_name, strlen(pw->pw_name)); write(pdes[1], ">\nSubject: ", 11); write(pdes[1], dayname, strlen(dayname)); write(pdes[1], "'s Calendar\nPrecedence: bulk\n\n", 30); while ((nread = read(fileno(fp), buf, sizeof(buf))) > 0) (void)write(pdes[1], buf, nread); (void)close(pdes[1]); done: (void)fclose(fp); (void)unlink(path); while (wait(&status) >= 0); } Index: projects/cxl_iscsi/usr.bin/calendar =================================================================== --- projects/cxl_iscsi/usr.bin/calendar (revision 279716) +++ projects/cxl_iscsi/usr.bin/calendar (revision 279717) Property changes on: projects/cxl_iscsi/usr.bin/calendar ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/usr.bin/calendar:r278538-279716 Index: projects/cxl_iscsi/usr.bin/mt/mt.1 =================================================================== --- projects/cxl_iscsi/usr.bin/mt/mt.1 (revision 279716) +++ projects/cxl_iscsi/usr.bin/mt/mt.1 (revision 279717) @@ -1,623 +1,625 @@ .\" Copyright (c) 1981, 1990, 1993 .\" The Regents of the University of California. All rights reserved. .\" .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" 4. Neither the name of the University nor the names of its contributors .\" may be used to endorse or promote products derived from this software .\" without specific prior written permission. .\" .\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" @(#)mt.1 8.1 (Berkeley) 6/6/93 .\" $FreeBSD$ .\" -.Dd February 12, 2014 +.Dd March 3, 2014 .Dt MT 1 .Os .Sh NAME .Nm mt .Nd magnetic tape manipulating program .Sh SYNOPSIS .Nm .Op Fl f Ar tapename .Ar command .Op Ar count .Nm .Op Fl f Ar tapename .Ar command .Ar argument .Sh DESCRIPTION The .Nm utility is used to command a magnetic tape drive for operations other than reading or writing data. .Pp The .Fl f option's .Ar tapename overrides the .Ev TAPE environment variable described below. .Pp The available commands are listed below. Only as many characters as are required to uniquely identify a command need be specified. .Pp The following commands optionally take a .Ar count , which defaults to 1. .Bl -tag -width ".Cm erase" .It Cm weof Write .Ar count end-of-file (EOF) marks at the current position. This returns when the file mark has been written to the media. .It Cm weofi Write .Ar count end-of-file (EOF) marks at the current position. This returns as soon as the command has been validated by the tape drive. .It Cm smk Write .Ar count setmarks at the current position (DDS drives only). .It Cm fsf Forward space .Ar count files. .It Cm fsr Forward space .Ar count records. .It Cm fss Forward space .Ar count setmarks (DDS drives only). .It Cm bsf Backward space .Ar count files. .It Cm bsr Backward space .Ar count records. .It Cm bss Backward space .Ar count setmarks (DDS drives only). .It Cm erase Erase the tape using a long (often very long) method. With a .Ar count of 0, it will erase the tape using a quick method. Operation is not guaranteed if the tape is not at its beginning. The tape will be at its beginning upon completion. .El .Pp The following commands ignore .Ar count . .Bl -tag -width ".Cm geteotmodel" .It Cm rdhpos Read the hardware block position. The block number reported is specific for that hardware only. With drive data compression especially, this position may have more to do with the amount of data sent to the drive than the amount of data written to tape. Some drives do not support this. .It Cm rdspos Read the SCSI logical block position. This typically is greater than the hardware position by the number of end-of-file marks. Some drives do not support this. .It Cm rewind Rewind the tape. .It Cm offline , rewoffl Rewind the tape and place the drive off line. Some drives are never off line. .It Cm load Load the tape into the drive. .It Cm retension Re-tension the tape. This winds the tape from the current position to the end and then to the beginning. This sometimes improves subsequent reading and writing, particularly for streaming drives. Some drives do not support this. .It Cm ostatus Output status information about the drive. For SCSI magnetic tape devices, the current operating modes of density, blocksize, and whether compression is enabled is reported. The current state of the driver (what it thinks that it is doing with the device) is reported. If the driver knows the relative position from BOT (in terms of filemarks and records), it outputs that. Note that this information is not definitive (only BOT, End of Recorded Media, and hardware or SCSI logical block position (if the drive supports such) are considered definitive tape positions). .Pp Also note that this is the old status command, and will be eliminated in favor of the new status command (see below) in a future release. .It Cm errstat Output (and clear) error status information about this device. For every normal operation (e.g., a read or a write) and every control operation (e.g,, a rewind), the driver stores up the last command executed and it is associated status and any residual counts (if any). This command retrieves and outputs this information. If possible, this also clears any latched error information. .It Cm geteotmodel Output the current EOT filemark model. The model states how many filemarks will be written at close if a tape was being written. .It Cm eod , eom Wind the tape to the end of the recorded data, typically after an EOF mark where another file may be written. .It Cm rblim Report the block limits of the tape drive, including the minimum and maximum block size, and the block granularity if any. .El .Pp The following commands may require an .Ar argument . .Bl -tag -width ".Cm seteotmodel" .It Cm sethpos Set the hardware block position. The .Ar argument is a hardware block number to which to position the tape. Some drives do not support this. .It Cm setspos Set the SCSI logical block position. The .Ar argument is a SCSI logical block number to which to position the tape. Some drives do not support this. .It Cm blocksize Set the block size for the drive. The .Ar argument is the number of bytes per block, except 0 commands the drive to use variable-length blocks. .It Cm seteotmodel Set the EOT filemark model to .Ar argument and output the old and new models. Typically this will be 2 filemarks, but some devices (typically QIC cartridge drives) can only write 1 filemark. You may only choose a value of .Ar 1 or .Ar 2 . .It Cm status Output status information about the drive. For SCSI magnetic tape devices, the current operating modes of density, blocksize, and whether compression is enabled is reported. The current state of the driver (what it thinks that it is doing with the device) is reported. .Pp If the driver knows the relative position from BOT (in terms of filemarks and records), it outputs that. If the tape drive supports the long form report of the .Tn SCSI READ POSITION command, the Reported File Number and Reported Record Number will be numbers other than -1, and there may be Flags reported as well. .Pp The BOP flag means that the logical position of the drive is at the beginning of the partition. .Pp The EOP flag means that the logical position of the drive is between Early Warning and End of Partition. .Pp The BPEW flag means that the logical position of the drive is in a Programmable Early Warning Zone or on the EOP side of Early Warning. .Pp Note that the Reported Record Number is the tape block or object number relative to the beginning of the partition. The Calculated Record Number is the tape block or object number relative to the previous file mark. .Pp Note that the Calculated File and Record Numbers are not definitive. The Reported File and Record Numbers are definitive, if they are numbers other than -1. .Bl -tag -width 6n .It Fl v Print additional status information, such as the maximum supported I/O size. .It Fl x Print all available status data to stdout in XML format. .El .It Cm getdensity Report density support information for the tape drive and any media that is loaded. Most drives will report at least basic density information similar to that reported by .Nm status command. Newer tape drives that conform to the T-10 SSC and newer tape specifications may report more detailed information about the types of tapes they support and the tape currently in the drive. .Bl -tag -width 6n .It Fl x Print all available density data to stdout in XML format. Because density information is currently included in the general status XML report used for .Nm status command, this will be the same XML output via .Do .Nm status .Fl x .Dc .El .It Cm param Display or set parameters. One of .Fl l , .Fl s , or .Fl x must be specified to indicate which operation to perform. .Bl -tag -width 8n .It Fl l List parameters, values and descriptions. By default all parameters will be displayed. To display a specific parameter, specify the parameter with .Fl p . .It Fl p Ar name Specify the parameter name to list (with .Fl l ) or set (with .Fl s ) . .It Fl q Enable quiet mode for parameter listing. This will suppress printing of parameter descriptions. .It Fl s Ar value Specify the parameter value to set. The general type of this argument (integer, unsigned integer, string) is determined by the type of the variable indicated by the .Xr sa 4 driver. More detailed argument checking is done by the .Xr sa 4 driver. .It Fl x Print out all parameter information in XML format. .El .It Cm protect Display or set drive protection parameters. This is used to control checking and reporting a per-block checksum for tape drives that support it. Some drives may only support some parameters. .Bl -tag -width 8n .It Fl b Ar 0|1 Set the Recover Buffered Data Protected bit. If set, this indicates that checksums are transferred with the logical blocks transferred by the RECOVERED BUFFERED DATA .Tn SCSI command. .It Fl d Disable all protection information settings. .It Fl e Enable all protection information settings. The default protection method used is Reed-Solomon CRC (protection method 1), as specified in ECMA-319. The default protection information length used with Reed-Solomon CRC is 4 bytes. To enable all settings except one more more settings, specify the .Fl e argument and then explicitly disable settings that you do not wish to enable. For example, specifying .Fl e .Fl w Ar 0 will enable all settings except for LBP_W. .It Fl l List available protection parmeters and their current settings. .It Fl L Ar len Set the length of the protection information in bytes. For Reed-Solomon CRC, the protection information length should be 4 bytes. .It Fl m Ar num Specify the numeric value for the protection method. The numeric value for Reed-Solomon CRC is 1. .It Fl r Ar 0|1 Set the LBP_R parameter. When set, this indicates that each block read from the tape drive will have a checksum at the end. .It Fl v Enable verbose mode for parameter listing. This will include descriptions of each parameter. .It Fl w Ar 0|1 Set the LBP_W parameter. When set, this indicates that each block written to the tape drive will have a checksum at the end. The drive will verify the checksum before writing the block to tape. .El .It Cm locate Set the tape drive's logical position. One of .Fl b , .Fl e , .Fl f , or .Fl s must be specified to indicate the type of position. If the partition number is specified, the drive will first relocate to the given partition (if it exists) and then to the position indicated within that partition. If the partition number is not specified, the drive will relocate to the given position within the current partition. .Bl -tag -width 14n .It Fl b Ar block_addr Relocate to the given tape block or logical object identifier. Note that the block number is the Reported Record Number that is relative to the beginning of the partition (or beginning of tape). .It Fl e Relocate to the end of data. .It Fl f Ar fileno Relocate to the given file number. .It Fl p Ar partition Specify the partition to change to. .It Fl s Ar setmark Relocate to the given set mark. .El .It Cm comp Set the drive's compression mode. The non-numeric values of .Ar argument are: .Pp .Bl -tag -width 9n -compact .It off Turn compression off. .It on Turn compression on. .It none Same as .Ar off . .It enable Same as .Ar on . .It IDRC IBM Improved Data Recording Capability compression (0x10). .It DCLZ DCLZ compression algorithm (0x20). .El .Pp In addition to the above recognized compression keywords, the user can supply a numeric compression algorithm for the drive to use. In most cases, simply turning the compression .Sq on will have the desired effect of enabling the default compression algorithm supported by the drive. If this is not the case (see the .Cm status display to see which compression algorithm is currently in use), the user can manually specify one of the supported compression keywords (above), or supply a numeric compression value from the drive's specifications. .Pp Note that for some older tape drives (for example the Exabyte 8200 and 8500 series drives) it is necessary to switch to a different density to tell the drive to record data in its compressed format. If the user attempts to turn compression on while the uncompressed density is selected, the drive will return an error. This is generally not an issue for modern tape drives. .It Cm density Set the density for the drive. For the density codes, see below. The density value could be given either numerically, or as a string, corresponding to the .Dq Reference field. If the string is abbreviated, it will be resolved in the order shown in the table, and the first matching entry will be used. If the given string and the resulting canonical density name do not match exactly, an informational message is output about what the given string has been taken for. .El .Pp The initial version of the density table below was taken from the .Sq Historical sequential access density codes table (A-1) in Revision 11 of the SCSI-3 Stream Device Commands (SSC) working draft, dated November 11, 1997. Subsequent additions have come from a number of sources. .Pp The density codes are: .Bd -literal -offset 2n 0x0 default for device 0xE reserved for ECMA Value Width Tracks Density Code Type Reference Note mm in bpmm bpi 0x01 12.7 (0.5) 9 32 (800) NRZI R X3.22-1983 2 0x02 12.7 (0.5) 9 63 (1,600) PE R X3.39-1986 2 0x03 12.7 (0.5) 9 246 (6,250) GCR R X3.54-1986 2 0x05 6.3 (0.25) 4/9 315 (8,000) GCR C X3.136-1986 1 0x06 12.7 (0.5) 9 126 (3,200) PE R X3.157-1987 2 0x07 6.3 (0.25) 4 252 (6,400) IMFM C X3.116-1986 1 0x08 3.81 (0.15) 4 315 (8,000) GCR CS X3.158-1987 1 0x09 12.7 (0.5) 18 1,491 (37,871) GCR C X3.180 2 0x0A 12.7 (0.5) 22 262 (6,667) MFM C X3B5/86-199 1 0x0B 6.3 (0.25) 4 63 (1,600) PE C X3.56-1986 1 0x0C 12.7 (0.5) 24 500 (12,690) GCR C HI-TC1 1,6 0x0D 12.7 (0.5) 24 999 (25,380) GCR C HI-TC2 1,6 0x0F 6.3 (0.25) 15 394 (10,000) GCR C QIC-120 1,6 0x10 6.3 (0.25) 18 394 (10,000) GCR C QIC-150 1,6 0x11 6.3 (0.25) 26 630 (16,000) GCR C QIC-320 1,6 0x12 6.3 (0.25) 30 2,034 (51,667) RLL C QIC-1350 1,6 0x13 3.81 (0.15) 1 2,400 (61,000) DDS CS X3B5/88-185A 5 0x14 8.0 (0.315) 1 1,703 (43,245) RLL CS X3.202-1991 5,11 0x15 8.0 (0.315) 1 1,789 (45,434) RLL CS ECMA TC17 5,12 0x16 12.7 (0.5) 48 394 (10,000) MFM C X3.193-1990 1 0x17 12.7 (0.5) 48 1,673 (42,500) MFM C X3B5/91-174 1 0x18 12.7 (0.5) 112 1,673 (42,500) MFM C X3B5/92-50 1 0x19 12.7 (0.5) 128 2,460 (62,500) RLL C DLTapeIII 6,7 0x1A 12.7 (0.5) 128 3,214 (81,633) RLL C DLTapeIV(20) 6,7 0x1B 12.7 (0.5) 208 3,383 (85,937) RLL C DLTapeIV(35) 6,7 0x1C 6.3 (0.25) 34 1,654 (42,000) MFM C QIC-385M 1,6 0x1D 6.3 (0.25) 32 1,512 (38,400) GCR C QIC-410M 1,6 0x1E 6.3 (0.25) 30 1,385 (36,000) GCR C QIC-1000C 1,6 0x1F 6.3 (0.25) 30 2,666 (67,733) RLL C QIC-2100C 1,6 0x20 6.3 (0.25) 144 2,666 (67,733) RLL C QIC-6GB(M) 1,6 0x21 6.3 (0.25) 144 2,666 (67,733) RLL C QIC-20GB(C) 1,6 0x22 6.3 (0.25) 42 1,600 (40,640) GCR C QIC-2GB(C) ? 0x23 6.3 (0.25) 38 2,666 (67,733) RLL C QIC-875M ? 0x24 3.81 (0.15) 1 2,400 (61,000) CS DDS-2 5 0x25 3.81 (0.15) 1 3,816 (97,000) CS DDS-3 5 0x26 3.81 (0.15) 1 3,816 (97,000) CS DDS-4 5 0x27 8.0 (0.315) 1 3,056 (77,611) RLL CS Mammoth 5 0x28 12.7 (0.5) 36 1,491 (37,871) GCR C X3.224 1 0x29 12.7 (0.5) 0x2A 0x2B 12.7 (0.5) 3 ? ? ? C X3.267 5 0x40 12.7 (0.5) 384 4,800 (123,952) C LTO-1 0x41 12.7 (0.5) 208 3,868 (98,250) RLL C DLTapeIV(40) 6,7 0x42 12.7 (0.5) 512 7,398 (187,909) C LTO-2 0x44 12.7 (0.5) 704 9,638 (244,805) C LTO-3 0x46 12.7 (0.5) 896 12,725 (323,215) C LTO-4 -0x48 12.7 (0.5) 448 5,236 (133,000) PRML C SDLTapeI(110) 6,8 +0x47 3.81 (0.25) ? 6,417 (163,000) CS DAT-72 +0x48 12.7 (0.5) 448 5,236 (133,000) PRML C SDLTapeI(110) 6,8,13 0x49 12.7 (0.5) 448 7,598 (193,000) PRML C SDLTapeI(160) 6,8 0x4A 12.7 (0.5) 768 ? C T10000A 10 0x4B 12.7 (0.5) 1152 ? C T10000B 10 0x4C 12.7 (0.5) 3584 ? C T10000C 10 0x4D 12.7 (0.5) 4608 ? C T10000D 10 0x51 12.7 (0.5) 512 11,800 (299,720) C 3592A1 (unencrypted) 0x52 12.7 (0.5) 896 11,800 (299,720) C 3592A2 (unencrypted) 0x53 12.7 (0.5) 1152 13,452 (341,681) C 3592A3 (unencrypted) 0x54 12.7 (0.5) 2560 19,686 (500,024) C 3592A4 (unencrypted) 0x55 12.7 (0.5) 5120 20,670 (525,018) C 3592A5 (unencrypted) 0x58 12.7 (0.5) 1280 15,142 (384,607) C LTO-5 0x5A 12.7 (0.5) 2176 15,142 (384,607) C LTO-6 0x71 12.7 (0.5) 512 11,800 (299,720) C 3592A1 (encrypted) 0x72 12.7 (0.5) 896 11,800 (299,720) C 3592A2 (encrypted) 0x73 12.7 (0.5) 1152 13,452 (341,681) C 3592A3 (encrypted) 0x74 12.7 (0.5) 2560 19,686 (500,024) C 3592A4 (encrypted) 0x75 12.7 (0.5) 5120 20,670 (525,018) C 3592A5 (encrypted) 0x8c 8.0 (0.315) 1 1,789 (45,434) RLL CS EXB-8500c 5,9 0x90 8.0 (0.315) 1 1,703 (43,245) RLL CS EXB-8200c 5,9 .Ed .Bd -literal -offset 2n Code Description Type Description ---- -------------------------------------- ---- ----------- NRZI Non return to zero, change on ones R Reel-to-reel GCR Group code recording C Cartridge PE Phase encoded CS Cassette IMFM Inverted modified frequency modulation MFM Modified frequency modulation DDS DAT data storage RLL Run length limited PRML Partial Response Maximum Likelihood .Ed .Bd -literal -offset 2n NOTES 1. Serial recorded. 2. Parallel recorded. 3. Old format known as QIC-11. 5. Helical scan. 6. This is not an American National Standard. The reference is based on an industry standard definition of the media format. 7. DLT recording: serially recorded track pairs (DLTapeIII and DLTapeIV(20)), or track quads (DLTapeIV(35) and DLTapeIV(40)). 8. Super DLT (SDLT) recording: 56 serially recorded logical tracks with 8 physical tracks each. 9. Vendor-specific Exabyte density code for compressed format. 10. bpi/bpmm values for the Oracle/StorageTek T10000 tape drives are not listed in the manual. Someone with access to a drive can supply the necessary values by running 'mt getdensity'. 11. This is Exabyte 8200 uncompressed format. The compressed format density code is 0x90. 12. This is Exabyte 8500 uncompressed format. The compressed format density code is 0x8c. +13. This density code (0x48) was also used for DAT-160. .Ed .Sh ENVIRONMENT .Bl -tag -width ".Ev TAPE" .It Ev TAPE This is the pathname of the tape drive. The default (if the variable is unset, but not if it is null) is .Pa /dev/nsa0 . It may be overridden with the .Fl f option. .El .Sh FILES .Bl -tag -width ".Pa /dev/*sa[0-9]*" -compact .It Pa /dev/*sa[0-9]* SCSI magnetic tape interface .El .Sh DIAGNOSTICS The exit status will be 0 when the drive operations were successful, 2 when the drive operations were unsuccessful, and 1 for other problems like an unrecognized command or a missing drive device. .Sh COMPATIBILITY Some undocumented commands support old software. .Sh SEE ALSO .Xr dd 1 , .Xr ioctl 2 , .Xr mtio 4 , .Xr sa 4 , .Xr environ 7 .Sh HISTORY The .Nm command appeared in .Bx 4.3 . .Pp Extensions regarding the .Xr st 4 driver appeared in .Bx 386 0.1 as a separate .Nm st command, and have been merged into the .Nm command in .Fx 2.1 . .Pp The former .Cm eof command that used to be a synonym for .Cm weof has been abandoned in .Fx 2.1 since it was often confused with .Cm eom , which is fairly dangerous. .Sh BUGS The utility cannot be interrupted or killed during a long erase (which can be longer than an hour), and it is easy to forget that the default erase is long. .Pp Hardware block numbers do not always correspond to blocks on the tape when the drive uses internal compression. .Pp Erasure is not guaranteed if the tape is not at its beginning. .Pp Tape-related documentation is poor, here and elsewhere. Index: projects/cxl_iscsi/usr.bin/netstat/if.c =================================================================== --- projects/cxl_iscsi/usr.bin/netstat/if.c (revision 279716) +++ projects/cxl_iscsi/usr.bin/netstat/if.c (revision 279717) @@ -1,673 +1,673 @@ /*- * Copyright (c) 2013 Gleb Smirnoff * Copyright (c) 1983, 1988, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #if 0 #ifndef lint static char sccsid[] = "@(#)if.c 8.3 (Berkeley) 4/28/95"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef PF #include #include #endif #include #include #include #include #ifdef INET6 #include #endif #include #include #include #include #include #include #include #include #include #include "netstat.h" static void sidewaysintpr(int); #ifdef INET6 static char addr_buf[NI_MAXHOST]; /* for getnameinfo() */ #endif #ifdef PF static const char* pfsyncacts[] = { /* PFSYNC_ACT_CLR */ "clear all request", /* PFSYNC_ACT_INS */ "state insert", /* PFSYNC_ACT_INS_ACK */ "state inserted ack", /* PFSYNC_ACT_UPD */ "state update", /* PFSYNC_ACT_UPD_C */ "compressed state update", /* PFSYNC_ACT_UPD_REQ */ "uncompressed state request", /* PFSYNC_ACT_DEL */ "state delete", /* PFSYNC_ACT_DEL_C */ "compressed state delete", /* PFSYNC_ACT_INS_F */ "fragment insert", /* PFSYNC_ACT_DEL_F */ "fragment delete", /* PFSYNC_ACT_BUS */ "bulk update mark", /* PFSYNC_ACT_TDB */ "TDB replay counter update", /* PFSYNC_ACT_EOF */ "end of frame mark", }; static const char* pfsyncacts_name[] = { /* PFSYNC_ACT_CLR */ "clear-all-request", /* PFSYNC_ACT_INS */ "state-insert", /* PFSYNC_ACT_INS_ACK */ "state-inserted-ack", /* PFSYNC_ACT_UPD */ "state-update", /* PFSYNC_ACT_UPD_C */ "compressed-state-update", /* PFSYNC_ACT_UPD_REQ */ "uncompressed-state-request", /* PFSYNC_ACT_DEL */ "state-delete", /* PFSYNC_ACT_DEL_C */ "compressed-state-delete", /* PFSYNC_ACT_INS_F */ "fragment-insert", /* PFSYNC_ACT_DEL_F */ "fragment-delete", /* PFSYNC_ACT_BUS */ "bulk-update-mark", /* PFSYNC_ACT_TDB */ "TDB-replay-counter-update", /* PFSYNC_ACT_EOF */ "end-of-frame-mark", }; static void pfsync_acts_stats(const char *list, const char *desc, uint64_t *a) { int i; xo_open_list(list); for (i = 0; i < PFSYNC_ACT_MAX; i++, a++) { if (*a || sflag <= 1) { xo_open_instance(list); xo_emit("\t\t{e:name}{:count/%ju} {N:/%s%s %s}\n", pfsyncacts_name[i], (uintmax_t)(*a), pfsyncacts[i], plural(*a), desc); xo_close_instance(list); } } xo_close_list(list); } /* * Dump pfsync statistics structure. */ void pfsync_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct pfsyncstats pfsyncstat, zerostat; size_t len = sizeof(struct pfsyncstats); if (live) { if (zflag) memset(&zerostat, 0, len); if (sysctlbyname("net.pfsync.stats", &pfsyncstat, &len, zflag ? &zerostat : NULL, zflag ? len : 0) < 0) { if (errno != ENOENT) warn("sysctl: net.pfsync.stats"); return; } } else kread(off, &pfsyncstat, len); xo_emit("{T:/%s}:\n", name); xo_open_container(name); #define p(f, m) if (pfsyncstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)pfsyncstat.f, plural(pfsyncstat.f)) p(pfsyncs_ipackets, "\t{:received-inet-packets/%ju} " "{N:/packet%s received (IPv4)}\n"); p(pfsyncs_ipackets6, "\t{:received-inet6-packets/%ju} " "{N:/packet%s received (IPv6)}\n"); pfsync_acts_stats("input-histogram", "received", &pfsyncstat.pfsyncs_iacts[0]); p(pfsyncs_badif, "\t\t/{:dropped-bad-interface/%ju} " "{N:/packet%s discarded for bad interface}\n"); p(pfsyncs_badttl, "\t\t{:dropped-bad-ttl/%ju} " "{N:/packet%s discarded for bad ttl}\n"); p(pfsyncs_hdrops, "\t\t{:dropped-short-header/%ju} " "{N:/packet%s shorter than header}\n"); p(pfsyncs_badver, "\t\t{:dropped-bad-version/%ju} " "{N:/packet%s discarded for bad version}\n"); p(pfsyncs_badauth, "\t\t{:dropped-bad-auth/%ju} " "{N:/packet%s discarded for bad HMAC}\n"); p(pfsyncs_badact,"\t\t{:dropped-bad-action/%ju} " "{N:/packet%s discarded for bad action}\n"); p(pfsyncs_badlen, "\t\t{:dropped-short/%ju} " "{N:/packet%s discarded for short packet}\n"); p(pfsyncs_badval, "\t\t{:dropped-bad-values/%ju} " "{N:/state%s discarded for bad values}\n"); p(pfsyncs_stale, "\t\t{:dropped-stale-state/%ju} " "{N:/stale state%s}\n"); p(pfsyncs_badstate, "\t\t{:dropped-failed-lookup/%ju} " "{N:/failed state lookup\\/insert%s}\n"); p(pfsyncs_opackets, "\t{:sent-inet-packets/%ju} " "{N:/packet%s sent (IPv4})\n"); p(pfsyncs_opackets6, "\t{:send-inet6-packets/%ju} " "{N:/packet%s sent (IPv6})\n"); pfsync_acts_stats("output-histogram", "sent", &pfsyncstat.pfsyncs_oacts[0]); p(pfsyncs_onomem, "\t\t{:discarded-no-memory/%ju} " "{N:/failure%s due to mbuf memory error}\n"); p(pfsyncs_oerrors, "\t\t{:send-errors/%ju} " "{N:/send error%s}\n"); #undef p xo_close_container(name); } #endif /* PF */ /* * Display a formatted value, or a '-' in the same space. */ static void show_stat(const char *fmt, int width, const char *name, u_long value, short showvalue) { const char *lsep, *rsep; char newfmt[64]; lsep = ""; if (strncmp(fmt, "LS", 2) == 0) { lsep = " "; fmt += 2; } rsep = " "; if (strncmp(fmt, "NRS", 3) == 0) { rsep = ""; fmt += 3; } if (showvalue == 0) { /* Print just dash. */ xo_emit("{P:/%s}{D:/%*s}{P:/%s}", lsep, width, "-", rsep); return; } /* * XXX: workaround {P:} modifier can't be empty and doesn't seem to * take args... so we need to conditionally include it in the format. */ #define maybe_pad(pad) do { \ if (strlen(pad)) { \ snprintf(newfmt, sizeof(newfmt), "{P:%s}", pad); \ xo_emit(newfmt); \ } \ } while (0) if (hflag) { char buf[5]; /* Format in human readable form. */ humanize_number(buf, sizeof(buf), (int64_t)value, "", HN_AUTOSCALE, HN_NOSPACE | HN_DECIMAL); maybe_pad(lsep); snprintf(newfmt, sizeof(newfmt), "{:%s/%%%ds}", name, width); xo_emit(newfmt, buf); maybe_pad(rsep); } else { /* Construct the format string. */ maybe_pad(lsep); snprintf(newfmt, sizeof(newfmt), "{:%s/%%%d%s}", name, width, fmt); xo_emit(newfmt, value); maybe_pad(rsep); } } /* * Find next multiaddr for a given interface name. */ static struct ifmaddrs * next_ifma(struct ifmaddrs *ifma, const char *name, const sa_family_t family) { for(; ifma != NULL; ifma = ifma->ifma_next) { struct sockaddr_dl *sdl; sdl = (struct sockaddr_dl *)ifma->ifma_name; if (ifma->ifma_addr->sa_family == family && strcmp(sdl->sdl_data, name) == 0) break; } return (ifma); } /* * Print a description of the network interfaces. */ void intpr(int interval, void (*pfunc)(char *), int af) { struct ifaddrs *ifap, *ifa; struct ifmaddrs *ifmap, *ifma; if (interval) return sidewaysintpr(interval); if (getifaddrs(&ifap) != 0) err(EX_OSERR, "getifaddrs"); if (aflag && getifmaddrs(&ifmap) != 0) err(EX_OSERR, "getifmaddrs"); xo_open_list("interface"); if (!pfunc) { if (Wflag) xo_emit("{T:/%-7.7s}", "Name"); else xo_emit("{T:/%-5.5s}", "Name"); xo_emit(" {T:/%5.5s} {T:/%-13.13s} {T:/%-17.17s} {T:/%8.8s} " "{T:/%5.5s} {T:/%5.5s}", "Mtu", "Network", "Address", "Ipkts", "Ierrs", "Idrop"); if (bflag) xo_emit(" {T:/%10.10s}","Ibytes"); xo_emit(" {T:/%8.8s} {T:/%5.5s}", "Opkts", "Oerrs"); if (bflag) xo_emit(" {T:/%10.10s}","Obytes"); xo_emit(" {T:/%5s}", "Coll"); if (dflag) xo_emit(" {T:/%s}", "Drop"); xo_emit("\n"); } for (ifa = ifap; ifa; ifa = ifa->ifa_next) { bool network = false, link = false; char *name; if (interface != NULL && strcmp(ifa->ifa_name, interface) != 0) continue; name = ifa->ifa_name; if (pfunc) { (*pfunc)(name); /* * Skip all ifaddrs belonging to same interface. */ while(ifa->ifa_next != NULL && (strcmp(ifa->ifa_next->ifa_name, name) == 0)) { ifa = ifa->ifa_next; } continue; } if (af != AF_UNSPEC && ifa->ifa_addr->sa_family != af) continue; xo_open_instance("interface"); if (Wflag) xo_emit("{tk:name/%-7.7s}", name); else xo_emit("{tk:name/%-5.5s}", name); #define IFA_MTU(ifa) (((struct if_data *)(ifa)->ifa_data)->ifi_mtu) show_stat("lu", 6, "mtu", IFA_MTU(ifa), IFA_MTU(ifa)); #undef IFA_MTU switch (ifa->ifa_addr->sa_family) { case AF_UNSPEC: xo_emit("{:network/%-13.13s} ", "none"); xo_emit("{:address/%-15.15s} ", "none"); break; case AF_INET: { struct sockaddr_in *sin, *mask; sin = (struct sockaddr_in *)ifa->ifa_addr; mask = (struct sockaddr_in *)ifa->ifa_netmask; xo_emit("{t:network/%-13.13s} ", netname(sin->sin_addr.s_addr, mask->sin_addr.s_addr)); xo_emit("{t:address/%-17.17s} ", routename(sin->sin_addr.s_addr)); network = true; break; } #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *sin6, *mask; sin6 = (struct sockaddr_in6 *)ifa->ifa_addr; mask = (struct sockaddr_in6 *)ifa->ifa_netmask; xo_emit("{t:network/%-13.13s} ", netname6(sin6, &mask->sin6_addr)); getnameinfo(ifa->ifa_addr, ifa->ifa_addr->sa_len, addr_buf, sizeof(addr_buf), 0, 0, NI_NUMERICHOST); xo_emit("{t:address/%-17.17s} ", addr_buf); network = 1; break; } #endif /* INET6 */ case AF_LINK: { struct sockaddr_dl *sdl; char *cp, linknum[10]; int len = 32; char buf[len]; int n, z; sdl = (struct sockaddr_dl *)ifa->ifa_addr; cp = (char *)LLADDR(sdl); n = sdl->sdl_alen; sprintf(linknum, "", sdl->sdl_index); xo_emit("{t:network/%-13.13s} ", linknum); buf[0] = '\0'; z = 0; while ((--n >= 0) && (z < len)) { snprintf(buf + z, len - z, "%02x%c", *cp++ & 0xff, n > 0 ? ':' : ' '); z += 3; } if (z > 0) xo_emit("{:address/%*s}", 32 - z, buf); else xo_emit("{P: }"); link = 1; break; } } #define IFA_STAT(s) (((struct if_data *)ifa->ifa_data)->ifi_ ## s) show_stat("lu", 8, "received-packets", IFA_STAT(ipackets), link|network); show_stat("lu", 5, "received-errors", IFA_STAT(ierrors), link); - show_stat("lu", 5, "dropped-packet", IFA_STAT(iqdrops), link); + show_stat("lu", 5, "dropped-packets", IFA_STAT(iqdrops), link); if (bflag) show_stat("lu", 10, "received-bytes", IFA_STAT(ibytes), link|network); show_stat("lu", 8, "sent-packets", IFA_STAT(opackets), link|network); show_stat("lu", 5, "send-errors", IFA_STAT(oerrors), link); if (bflag) show_stat("lu", 10, "sent-bytes", IFA_STAT(obytes), link|network); show_stat("NRSlu", 5, "collisions", IFA_STAT(collisions), link); if (dflag) show_stat("LSlu", 5, "dropped-packets", IFA_STAT(oqdrops), link); xo_emit("\n"); if (!aflag) { xo_close_instance("interface"); continue; } /* * Print family's multicast addresses. */ xo_open_list("multicast-address"); for (ifma = next_ifma(ifmap, ifa->ifa_name, ifa->ifa_addr->sa_family); ifma != NULL; ifma = next_ifma(ifma, ifa->ifa_name, ifa->ifa_addr->sa_family)) { const char *fmt = NULL; xo_open_instance("multicast-address"); switch (ifma->ifma_addr->sa_family) { case AF_INET: { struct sockaddr_in *sin; sin = (struct sockaddr_in *)ifma->ifma_addr; fmt = routename(sin->sin_addr.s_addr); break; } #ifdef INET6 case AF_INET6: /* in6_fillscopeid(&msa.in6); */ getnameinfo(ifma->ifma_addr, ifma->ifma_addr->sa_len, addr_buf, sizeof(addr_buf), 0, 0, NI_NUMERICHOST); xo_emit("{P:/%*s }{t:address/%-19.19s}", Wflag ? 27 : 25, "", addr_buf); break; #endif /* INET6 */ case AF_LINK: { struct sockaddr_dl *sdl; sdl = (struct sockaddr_dl *)ifma->ifma_addr; switch (sdl->sdl_type) { case IFT_ETHER: case IFT_FDDI: fmt = ether_ntoa( (struct ether_addr *)LLADDR(sdl)); break; } break; } } if (fmt) { xo_emit("{P:/%*s }{t:address/%-17.17s/}", Wflag ? 27 : 25, "", fmt); if (ifma->ifma_addr->sa_family == AF_LINK) { xo_emit(" {:received-packets/%8lu}", IFA_STAT(imcasts)); xo_emit("{P:/%*s}", bflag? 17 : 6, ""); xo_emit(" {:sent-packets/%8lu}", IFA_STAT(omcasts)); } xo_emit("\n"); } xo_close_instance("multicast-address"); ifma = ifma->ifma_next; } xo_close_list("multicast-address"); xo_close_instance("interface"); } xo_close_list("interface"); freeifaddrs(ifap); if (aflag) freeifmaddrs(ifmap); } struct iftot { u_long ift_ip; /* input packets */ u_long ift_ie; /* input errors */ u_long ift_id; /* input drops */ u_long ift_op; /* output packets */ u_long ift_oe; /* output errors */ u_long ift_od; /* output drops */ u_long ift_co; /* collisions */ u_long ift_ib; /* input bytes */ u_long ift_ob; /* output bytes */ }; /* * Obtain stats for interface(s). */ static void fill_iftot(struct iftot *st) { struct ifaddrs *ifap, *ifa; bool found = false; if (getifaddrs(&ifap) != 0) xo_err(EX_OSERR, "getifaddrs"); bzero(st, sizeof(*st)); for (ifa = ifap; ifa; ifa = ifa->ifa_next) { if (ifa->ifa_addr->sa_family != AF_LINK) continue; if (interface) { if (strcmp(ifa->ifa_name, interface) == 0) found = true; else continue; } st->ift_ip += IFA_STAT(ipackets); st->ift_ie += IFA_STAT(ierrors); st->ift_id += IFA_STAT(iqdrops); st->ift_ib += IFA_STAT(ibytes); st->ift_op += IFA_STAT(opackets); st->ift_oe += IFA_STAT(oerrors); st->ift_od += IFA_STAT(oqdrops); st->ift_ob += IFA_STAT(obytes); st->ift_co += IFA_STAT(collisions); } if (interface && found == false) xo_err(EX_DATAERR, "interface %s not found", interface); freeifaddrs(ifap); } /* * Set a flag to indicate that a signal from the periodic itimer has been * caught. */ static sig_atomic_t signalled; static void catchalarm(int signo __unused) { signalled = true; } /* * Print a running summary of interface statistics. * Repeat display every interval seconds, showing statistics * collected over that interval. Assumes that interval is non-zero. * First line printed at top of screen is always cumulative. */ static void sidewaysintpr(int interval) { struct iftot ift[2], *new, *old; struct itimerval interval_it; int oldmask, line; new = &ift[0]; old = &ift[1]; fill_iftot(old); (void)signal(SIGALRM, catchalarm); signalled = false; interval_it.it_interval.tv_sec = interval; interval_it.it_interval.tv_usec = 0; interval_it.it_value = interval_it.it_interval; setitimer(ITIMER_REAL, &interval_it, NULL); xo_open_list("interface-statistics"); banner: xo_emit("{T:/%17s} {T:/%14s} {T:/%16s}\n", "input", interface != NULL ? interface : "(Total)", "output"); xo_emit("{T:/%10s} {T:/%5s} {T:/%5s} {T:/%10s} {T:/%10s} {T:/%5s} " "{T:/%10s} {T:/%5s}", "packets", "errs", "idrops", "bytes", "packets", "errs", "bytes", "colls"); if (dflag) xo_emit(" {T:/%5.5s}", "drops"); xo_emit("\n"); xo_flush(); line = 0; loop: if ((noutputs != 0) && (--noutputs == 0)) { xo_close_list("interface-statistics"); return; } oldmask = sigblock(sigmask(SIGALRM)); while (!signalled) sigpause(0); signalled = false; sigsetmask(oldmask); line++; fill_iftot(new); xo_open_instance("stats"); show_stat("lu", 10, "received-packets", new->ift_ip - old->ift_ip, 1); show_stat("lu", 5, "received-errors", new->ift_ie - old->ift_ie, 1); show_stat("lu", 5, "dropped-packets", new->ift_id - old->ift_id, 1); show_stat("lu", 10, "received-bytes", new->ift_ib - old->ift_ib, 1); show_stat("lu", 10, "sent-packets", new->ift_op - old->ift_op, 1); show_stat("lu", 5, "send-errors", new->ift_oe - old->ift_oe, 1); show_stat("lu", 10, "sent-bytes", new->ift_ob - old->ift_ob, 1); show_stat("NRSlu", 5, "collisions", new->ift_co - old->ift_co, 1); if (dflag) show_stat("LSlu", 5, "dropped-packets", new->ift_od - old->ift_od, 1); xo_close_instance("stats"); xo_emit("\n"); xo_flush(); if (new == &ift[0]) { new = &ift[1]; old = &ift[0]; } else { new = &ift[0]; old = &ift[1]; } if (line == 21) goto banner; else goto loop; /* NOTREACHED */ } Index: projects/cxl_iscsi/usr.bin/rlogin/Makefile =================================================================== --- projects/cxl_iscsi/usr.bin/rlogin/Makefile (revision 279716) +++ projects/cxl_iscsi/usr.bin/rlogin/Makefile (revision 279717) @@ -1,10 +1,9 @@ # @(#)Makefile 8.1 (Berkeley) 7/19/93 # $FreeBSD$ PROG= rlogin BINOWN= root BINMODE=4555 -PRECIOUSPROG= .include Index: projects/cxl_iscsi/usr.bin/rlogin/rlogin.c =================================================================== --- projects/cxl_iscsi/usr.bin/rlogin/rlogin.c (revision 279716) +++ projects/cxl_iscsi/usr.bin/rlogin/rlogin.c (revision 279717) @@ -1,722 +1,722 @@ /* * Copyright (c) 1983, 1990, 1993 * The Regents of the University of California. All rights reserved. * Copyright (c) 2002 Networks Associates Technology, Inc. * All rights reserved. * * Portions of this software were developed for the FreeBSD Project by * ThinkSec AS and NAI Labs, the Security Research Division of Network * Associates, Inc. under DARPA/SPAWAR contract N66001-01-C-8035 * ("CBOSS"), as part of the DARPA CHATS research program. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef lint static const char copyright[] = "@(#) Copyright (c) 1983, 1990, 1993\n\ The Regents of the University of California. All rights reserved.\n"; #endif /* not lint */ #if 0 #ifndef lint static const char sccsid[] = "@(#)rlogin.c 8.1 (Berkeley) 6/6/93"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); /* * rlogin - remote login */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifndef TIOCPKT_WINDOW #define TIOCPKT_WINDOW 0x80 #endif /* concession to Sun */ #ifndef SIGUSR1 #define SIGUSR1 30 #endif static int eight, rem; static struct termios deftty; static int family = PF_UNSPEC; static int noescape; static u_char escapechar = '~'; #define get_window_size(fd, wp) ioctl(fd, TIOCGWINSZ, wp) static struct winsize winsize; static void catch_child(int); static void copytochild(int); static _Noreturn void doit(long); static _Noreturn void done(int); static void echo(char); static u_int getescape(const char *); static void lostpeer(int); static void mode(int); static void msg(const char *); static void oob(int); static int reader(int); static void sendwindow(void); static void setsignal(int); static void sigwinch(int); static void stop(char); static _Noreturn void usage(void); static void writer(void); static void writeroob(int); int main(int argc, char *argv[]) { struct passwd *pw; struct servent *sp; struct termios tty; long omask; int argoff, ch, dflag, Dflag, one; uid_t uid; - char *host, *localname, *p, *user, term[1024]; + char *host, *localname, *p, *user, term[1024] = "network"; speed_t ospeed; struct sockaddr_storage ss; socklen_t sslen; size_t len, len2; int i; argoff = dflag = Dflag = 0; one = 1; host = localname = user = NULL; if ((p = strrchr(argv[0], '/'))) ++p; else p = argv[0]; if (strcmp(p, "rlogin")) host = p; /* handle "rlogin host flags" */ if (!host && argc > 2 && argv[1][0] != '-') { host = argv[1]; argoff = 1; } #define OPTIONS "468DEde:i:l:" while ((ch = getopt(argc - argoff, argv + argoff, OPTIONS)) != -1) switch(ch) { case '4': family = PF_INET; break; case '6': family = PF_INET6; break; case '8': eight = 1; break; case 'D': Dflag = 1; break; case 'E': noescape = 1; break; case 'd': dflag = 1; break; case 'e': noescape = 0; escapechar = getescape(optarg); break; case 'i': if (getuid() != 0) errx(1, "-i user: permission denied"); localname = optarg; break; case 'l': user = optarg; break; case '?': default: usage(); } optind += argoff; /* if haven't gotten a host yet, do so */ if (!host && !(host = argv[optind++])) usage(); if (argv[optind]) usage(); if (!(pw = getpwuid(uid = getuid()))) errx(1, "unknown user id"); if (!user) user = pw->pw_name; if (!localname) localname = pw->pw_name; sp = NULL; sp = getservbyname("login", "tcp"); if (sp == NULL) errx(1, "login/tcp: unknown service"); if ((p = getenv("TERM")) != NULL) (void)strlcpy(term, p, sizeof(term)); len = strlen(term); if (len < (sizeof(term) - 1) && tcgetattr(0, &tty) == 0) { /* start at 2 to include the / */ for (ospeed = i = cfgetospeed(&tty), len2 = 2; i > 9; len2++) i /= 10; if (len + len2 < sizeof(term)) (void)snprintf(term + len, len2 + 1, "/%d", ospeed); } (void)get_window_size(0, &winsize); (void)signal(SIGPIPE, lostpeer); /* will use SIGUSR1 for window size hack, so hold it off */ omask = sigblock(sigmask(SIGURG) | sigmask(SIGUSR1)); /* * We set SIGURG and SIGUSR1 below so that an * incoming signal will be held pending rather than being * discarded. Note that these routines will be ready to get * a signal by the time that they are unblocked below. */ (void)signal(SIGURG, copytochild); (void)signal(SIGUSR1, writeroob); rem = rcmd_af(&host, sp->s_port, localname, user, term, 0, family); if (rem < 0) exit(1); if (dflag && setsockopt(rem, SOL_SOCKET, SO_DEBUG, &one, sizeof(one)) < 0) warn("setsockopt"); if (Dflag && setsockopt(rem, IPPROTO_TCP, TCP_NODELAY, &one, sizeof(one)) < 0) warn("setsockopt NODELAY (ignored)"); sslen = sizeof(ss); one = IPTOS_LOWDELAY; if (getsockname(rem, (struct sockaddr *)&ss, &sslen) == 0 && ss.ss_family == AF_INET) { if (setsockopt(rem, IPPROTO_IP, IP_TOS, (char *)&one, sizeof(int)) < 0) warn("setsockopt TOS (ignored)"); } else if (ss.ss_family == AF_INET) warn("setsockopt getsockname failed"); (void)setuid(uid); doit(omask); /*NOTREACHED*/ } static int child; static void doit(long omask) { (void)signal(SIGINT, SIG_IGN); setsignal(SIGHUP); setsignal(SIGQUIT); mode(1); child = fork(); if (child == -1) { warn("fork"); done(1); } if (child == 0) { if (reader(omask) == 0) { msg("connection closed"); exit(0); } sleep(1); msg("\007connection closed"); exit(1); } /* * We may still own the socket, and may have a pending SIGURG (or might * receive one soon) that we really want to send to the reader. When * one of these comes in, the trap copytochild simply copies such * signals to the child. We can now unblock SIGURG and SIGUSR1 * that were set above. */ (void)sigsetmask(omask); (void)signal(SIGCHLD, catch_child); writer(); msg("closed connection"); done(0); } /* trap a signal, unless it is being ignored. */ static void setsignal(int sig) { int omask = sigblock(sigmask(sig)); if (signal(sig, exit) == SIG_IGN) (void)signal(sig, SIG_IGN); (void)sigsetmask(omask); } static void done(int status) { int w, wstatus; mode(0); if (child > 0) { /* make sure catch_child does not snap it up */ (void)signal(SIGCHLD, SIG_DFL); if (kill(child, SIGKILL) >= 0) while ((w = wait(&wstatus)) > 0 && w != child); } exit(status); } static int dosigwinch; /* * This is called when the reader process gets the out-of-band (urgent) * request to turn on the window-changing protocol. */ /* ARGSUSED */ static void writeroob(int signo __unused) { if (dosigwinch == 0) { sendwindow(); (void)signal(SIGWINCH, sigwinch); } dosigwinch = 1; } /* ARGSUSED */ static void catch_child(int signo __unused) { pid_t pid; int status; for (;;) { pid = wait3(&status, WNOHANG|WUNTRACED, NULL); if (pid == 0) return; /* if the child (reader) dies, just quit */ if (pid < 0 || (pid == child && !WIFSTOPPED(status))) done(WTERMSIG(status) | WEXITSTATUS(status)); } /* NOTREACHED */ } /* * writer: write to remote: 0 -> line. * ~. terminate * ~^Z suspend rlogin process. * ~ suspend rlogin process, but leave reader alone. */ static void writer(void) { int bol, local, n; char c; bol = 1; /* beginning of line */ local = 0; for (;;) { n = read(STDIN_FILENO, &c, 1); if (n <= 0) { if (n < 0 && errno == EINTR) continue; break; } /* * If we're at the beginning of the line and recognize a * command character, then we echo locally. Otherwise, * characters are echo'd remotely. If the command character * is doubled, this acts as a force and local echo is * suppressed. */ if (bol) { bol = 0; if (!noescape && c == escapechar) { local = 1; continue; } } else if (local) { local = 0; if (c == '.' || CCEQ(deftty.c_cc[VEOF], c)) { echo(c); break; } if (CCEQ(deftty.c_cc[VSUSP], c) || CCEQ(deftty.c_cc[VDSUSP], c)) { bol = 1; echo(c); stop(c); continue; } if (c != escapechar) (void)write(rem, &escapechar, 1); } if (write(rem, &c, 1) == 0) { msg("line gone"); break; } bol = CCEQ(deftty.c_cc[VKILL], c) || CCEQ(deftty.c_cc[VEOF], c) || CCEQ(deftty.c_cc[VINTR], c) || CCEQ(deftty.c_cc[VSUSP], c) || c == '\r' || c == '\n'; } } static void echo(char c) { char *p; char buf[8]; p = buf; c &= 0177; *p++ = escapechar; if (c < ' ') { *p++ = '^'; *p++ = c + '@'; } else if (c == 0177) { *p++ = '^'; *p++ = '?'; } else *p++ = c; *p++ = '\r'; *p++ = '\n'; (void)write(STDOUT_FILENO, buf, p - buf); } static void stop(char cmdc) { mode(0); (void)signal(SIGCHLD, SIG_IGN); (void)kill(CCEQ(deftty.c_cc[VSUSP], cmdc) ? 0 : getpid(), SIGTSTP); (void)signal(SIGCHLD, catch_child); mode(1); sigwinch(0); /* check for size changes */ } /* ARGSUSED */ static void sigwinch(int signo __unused) { struct winsize ws; if (dosigwinch && get_window_size(0, &ws) == 0 && bcmp(&ws, &winsize, sizeof(ws))) { winsize = ws; sendwindow(); } } /* * Send the window size to the server via the magic escape */ static void sendwindow(void) { struct winsize ws; char obuf[4 + sizeof (struct winsize)]; obuf[0] = 0377; obuf[1] = 0377; obuf[2] = 's'; obuf[3] = 's'; ws.ws_row = htons(winsize.ws_row); ws.ws_col = htons(winsize.ws_col); ws.ws_xpixel = htons(winsize.ws_xpixel); ws.ws_ypixel = htons(winsize.ws_ypixel); bcopy(&ws, obuf + 4, sizeof(ws)); (void)write(rem, obuf, sizeof(obuf)); } /* * reader: read from remote: line -> 1 */ #define READING 1 #define WRITING 2 static jmp_buf rcvtop; static int rcvcnt, rcvstate; static pid_t ppid; static char rcvbuf[8 * 1024]; /* ARGSUSED */ static void oob(int signo __unused) { struct termios tty; int atmark, n, rcvd; char waste[BUFSIZ], mark; rcvd = 0; while (recv(rem, &mark, 1, MSG_OOB) < 0) { switch (errno) { case EWOULDBLOCK: /* * Urgent data not here yet. It may not be possible * to send it yet if we are blocked for output and * our input buffer is full. */ if (rcvcnt < (int)sizeof(rcvbuf)) { n = read(rem, rcvbuf + rcvcnt, sizeof(rcvbuf) - rcvcnt); if (n <= 0) return; rcvd += n; } else { n = read(rem, waste, sizeof(waste)); if (n <= 0) return; } continue; default: return; } } if (mark & TIOCPKT_WINDOW) { /* Let server know about window size changes */ (void)kill(ppid, SIGUSR1); } if (!eight && (mark & TIOCPKT_NOSTOP)) { (void)tcgetattr(0, &tty); tty.c_iflag &= ~IXON; (void)tcsetattr(0, TCSANOW, &tty); } if (!eight && (mark & TIOCPKT_DOSTOP)) { (void)tcgetattr(0, &tty); tty.c_iflag |= (deftty.c_iflag & IXON); (void)tcsetattr(0, TCSANOW, &tty); } if (mark & TIOCPKT_FLUSHWRITE) { (void)tcflush(1, TCIOFLUSH); for (;;) { if (ioctl(rem, SIOCATMARK, &atmark) < 0) { warn("ioctl"); break; } if (atmark) break; n = read(rem, waste, sizeof (waste)); if (n <= 0) break; } /* * Don't want any pending data to be output, so clear the recv * buffer. If we were hanging on a write when interrupted, * don't want it to restart. If we were reading, restart * anyway. */ rcvcnt = 0; longjmp(rcvtop, 1); } /* oob does not do FLUSHREAD (alas!) */ /* * If we filled the receive buffer while a read was pending, longjmp * to the top to restart appropriately. Don't abort a pending write, * however, or we won't know how much was written. */ if (rcvd && rcvstate == READING) longjmp(rcvtop, 1); } /* reader: read from remote: line -> 1 */ static int reader(int omask) { int n, remaining; char *bufp; pid_t pid; pid = getpid(); (void)signal(SIGTTOU, SIG_IGN); (void)signal(SIGURG, oob); (void)signal(SIGUSR1, oob); /* When propogating SIGURG from parent */ ppid = getppid(); (void)fcntl(rem, F_SETOWN, pid); (void)setjmp(rcvtop); (void)sigsetmask(omask); bufp = rcvbuf; for (;;) { while ((remaining = rcvcnt - (bufp - rcvbuf)) > 0) { rcvstate = WRITING; n = write(STDOUT_FILENO, bufp, remaining); if (n < 0) { if (errno != EINTR) return (-1); continue; } bufp += n; } bufp = rcvbuf; rcvcnt = 0; rcvstate = READING; rcvcnt = read(rem, rcvbuf, sizeof (rcvbuf)); if (rcvcnt == 0) return (0); if (rcvcnt < 0) { if (errno == EINTR) continue; warn("read"); return (-1); } } } static void mode(int f) { struct termios tty; switch (f) { case 0: (void)tcsetattr(0, TCSANOW, &deftty); break; case 1: (void)tcgetattr(0, &deftty); tty = deftty; /* This is loosely derived from sys/kern/tty_compat.c. */ tty.c_lflag &= ~(ECHO|ICANON|ISIG|IEXTEN); tty.c_iflag &= ~ICRNL; tty.c_oflag &= ~OPOST; tty.c_cc[VMIN] = 1; tty.c_cc[VTIME] = 0; if (eight) { tty.c_iflag &= IXOFF; tty.c_cflag &= ~(CSIZE|PARENB); tty.c_cflag |= CS8; } (void)tcsetattr(0, TCSANOW, &tty); break; default: return; } } /* ARGSUSED */ static void lostpeer(int signo __unused) { (void)signal(SIGPIPE, SIG_IGN); msg("\007connection closed"); done(1); } /* copy SIGURGs to the child process via SIGUSR1. */ /* ARGSUSED */ static void copytochild(int signo __unused) { (void)kill(child, SIGUSR1); } static void msg(const char *str) { (void)fprintf(stderr, "rlogin: %s\r\n", str); } static void usage(void) { (void)fprintf(stderr, "usage: rlogin [-46%s]%s[-e char] [-i localname] [-l username] host\n", "8DEd", " "); exit(1); } static u_int getescape(const char *p) { long val; size_t len; if ((len = strlen(p)) == 1) /* use any single char, including '\' */ return ((u_int)*p); /* otherwise, \nnn */ if (*p == '\\' && len >= 2 && len <= 4) { val = strtol(++p, NULL, 8); for (;;) { if (!*++p) return ((u_int)val); if (*p < '0' || *p > '8') break; } } msg("illegal option value -- e"); usage(); /* NOTREACHED */ } Index: projects/cxl_iscsi/usr.bin/rsh/Makefile =================================================================== --- projects/cxl_iscsi/usr.bin/rsh/Makefile (revision 279716) +++ projects/cxl_iscsi/usr.bin/rsh/Makefile (revision 279717) @@ -1,13 +1,12 @@ # @(#)Makefile 8.1 (Berkeley) 7/19/93 # $FreeBSD$ PROG= rsh CFLAGS+=-I${.CURDIR}/../../libexec/rlogind WARNS?= 2 BINOWN= root BINMODE=4555 -PRECIOUSPROG= .include Index: projects/cxl_iscsi/usr.sbin/bhyve/block_if.c =================================================================== --- projects/cxl_iscsi/usr.sbin/bhyve/block_if.c (revision 279716) +++ projects/cxl_iscsi/usr.sbin/bhyve/block_if.c (revision 279717) @@ -1,612 +1,636 @@ /*- * Copyright (c) 2013 Peter Grehan * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "bhyverun.h" #include "mevent.h" #include "block_if.h" #define BLOCKIF_SIG 0xb109b109 #define BLOCKIF_MAXREQ 33 enum blockop { BOP_READ, BOP_WRITE, BOP_FLUSH }; enum blockstat { BST_FREE, BST_PEND, BST_BUSY, BST_DONE }; struct blockif_elem { TAILQ_ENTRY(blockif_elem) be_link; struct blockif_req *be_req; enum blockop be_op; enum blockstat be_status; pthread_t be_tid; }; struct blockif_ctxt { int bc_magic; int bc_fd; + int bc_ischr; int bc_rdonly; off_t bc_size; int bc_sectsz; + int bc_psectsz; + int bc_psectoff; pthread_t bc_btid; pthread_mutex_t bc_mtx; pthread_cond_t bc_cond; int bc_closing; /* Request elements and free/pending/busy queues */ TAILQ_HEAD(, blockif_elem) bc_freeq; TAILQ_HEAD(, blockif_elem) bc_pendq; TAILQ_HEAD(, blockif_elem) bc_busyq; u_int bc_req_count; struct blockif_elem bc_reqs[BLOCKIF_MAXREQ]; }; static pthread_once_t blockif_once = PTHREAD_ONCE_INIT; struct blockif_sig_elem { pthread_mutex_t bse_mtx; pthread_cond_t bse_cond; int bse_pending; struct blockif_sig_elem *bse_next; }; static struct blockif_sig_elem *blockif_bse_head; static int blockif_enqueue(struct blockif_ctxt *bc, struct blockif_req *breq, enum blockop op) { struct blockif_elem *be; assert(bc->bc_req_count < BLOCKIF_MAXREQ); be = TAILQ_FIRST(&bc->bc_freeq); assert(be != NULL); assert(be->be_status == BST_FREE); TAILQ_REMOVE(&bc->bc_freeq, be, be_link); be->be_status = BST_PEND; be->be_req = breq; be->be_op = op; TAILQ_INSERT_TAIL(&bc->bc_pendq, be, be_link); bc->bc_req_count++; return (0); } static int blockif_dequeue(struct blockif_ctxt *bc, struct blockif_elem **bep) { struct blockif_elem *be; if (bc->bc_req_count == 0) return (ENOENT); be = TAILQ_FIRST(&bc->bc_pendq); assert(be != NULL); assert(be->be_status == BST_PEND); TAILQ_REMOVE(&bc->bc_pendq, be, be_link); be->be_status = BST_BUSY; be->be_tid = bc->bc_btid; TAILQ_INSERT_TAIL(&bc->bc_busyq, be, be_link); *bep = be; return (0); } static void blockif_complete(struct blockif_ctxt *bc, struct blockif_elem *be) { assert(be->be_status == BST_DONE); TAILQ_REMOVE(&bc->bc_busyq, be, be_link); be->be_tid = 0; be->be_status = BST_FREE; be->be_req = NULL; TAILQ_INSERT_TAIL(&bc->bc_freeq, be, be_link); bc->bc_req_count--; } static void blockif_proc(struct blockif_ctxt *bc, struct blockif_elem *be) { struct blockif_req *br; int err; br = be->be_req; err = 0; switch (be->be_op) { case BOP_READ: if (preadv(bc->bc_fd, br->br_iov, br->br_iovcnt, br->br_offset) < 0) err = errno; break; case BOP_WRITE: if (bc->bc_rdonly) err = EROFS; else if (pwritev(bc->bc_fd, br->br_iov, br->br_iovcnt, br->br_offset) < 0) err = errno; break; case BOP_FLUSH: + if (bc->bc_ischr) { + if (ioctl(bc->bc_fd, DIOCGFLUSH)) + err = errno; + } else if (fsync(bc->bc_fd)) + err = errno; break; default: err = EINVAL; break; } be->be_status = BST_DONE; (*br->br_callback)(br, err); } static void * blockif_thr(void *arg) { struct blockif_ctxt *bc; struct blockif_elem *be; bc = arg; for (;;) { pthread_mutex_lock(&bc->bc_mtx); while (!blockif_dequeue(bc, &be)) { pthread_mutex_unlock(&bc->bc_mtx); blockif_proc(bc, be); pthread_mutex_lock(&bc->bc_mtx); blockif_complete(bc, be); } pthread_cond_wait(&bc->bc_cond, &bc->bc_mtx); pthread_mutex_unlock(&bc->bc_mtx); /* * Check ctxt status here to see if exit requested */ if (bc->bc_closing) pthread_exit(NULL); } /* Not reached */ return (NULL); } static void blockif_sigcont_handler(int signal, enum ev_type type, void *arg) { struct blockif_sig_elem *bse; for (;;) { /* * Process the entire list even if not intended for * this thread. */ do { bse = blockif_bse_head; if (bse == NULL) return; } while (!atomic_cmpset_ptr((uintptr_t *)&blockif_bse_head, (uintptr_t)bse, (uintptr_t)bse->bse_next)); pthread_mutex_lock(&bse->bse_mtx); bse->bse_pending = 0; pthread_cond_signal(&bse->bse_cond); pthread_mutex_unlock(&bse->bse_mtx); } } static void blockif_init(void) { mevent_add(SIGCONT, EVF_SIGNAL, blockif_sigcont_handler, NULL); (void) signal(SIGCONT, SIG_IGN); } struct blockif_ctxt * blockif_open(const char *optstr, const char *ident) { char tname[MAXCOMLEN + 1]; char *nopt, *xopts; struct blockif_ctxt *bc; struct stat sbuf; - off_t size; + off_t size, psectsz, psectoff; int extra, fd, i, sectsz; int nocache, sync, ro; pthread_once(&blockif_once, blockif_init); nocache = 0; sync = 0; ro = 0; /* * The first element in the optstring is always a pathname. * Optional elements follow */ nopt = strdup(optstr); for (xopts = strtok(nopt, ","); xopts != NULL; xopts = strtok(NULL, ",")) { if (!strcmp(xopts, "nocache")) nocache = 1; else if (!strcmp(xopts, "sync")) sync = 1; else if (!strcmp(xopts, "ro")) ro = 1; } extra = 0; if (nocache) extra |= O_DIRECT; if (sync) extra |= O_SYNC; fd = open(nopt, (ro ? O_RDONLY : O_RDWR) | extra); if (fd < 0 && !ro) { /* Attempt a r/w fail with a r/o open */ fd = open(nopt, O_RDONLY | extra); ro = 1; } if (fd < 0) { perror("Could not open backing file"); return (NULL); } if (fstat(fd, &sbuf) < 0) { perror("Could not stat backing file"); close(fd); return (NULL); } /* * Deal with raw devices */ size = sbuf.st_size; sectsz = DEV_BSIZE; + psectsz = psectoff = 0; if (S_ISCHR(sbuf.st_mode)) { if (ioctl(fd, DIOCGMEDIASIZE, &size) < 0 || ioctl(fd, DIOCGSECTORSIZE, §sz)) { perror("Could not fetch dev blk/sector size"); close(fd); return (NULL); } assert(size != 0); assert(sectsz != 0); - } + if (ioctl(fd, DIOCGSTRIPESIZE, &psectsz) == 0 && psectsz > 0) + ioctl(fd, DIOCGSTRIPEOFFSET, &psectoff); + } else + psectsz = sbuf.st_blksize; bc = calloc(1, sizeof(struct blockif_ctxt)); if (bc == NULL) { close(fd); return (NULL); } bc->bc_magic = BLOCKIF_SIG; bc->bc_fd = fd; + bc->bc_ischr = S_ISCHR(sbuf.st_mode); bc->bc_rdonly = ro; bc->bc_size = size; bc->bc_sectsz = sectsz; + bc->bc_psectsz = psectsz; + bc->bc_psectoff = psectoff; pthread_mutex_init(&bc->bc_mtx, NULL); pthread_cond_init(&bc->bc_cond, NULL); TAILQ_INIT(&bc->bc_freeq); TAILQ_INIT(&bc->bc_pendq); TAILQ_INIT(&bc->bc_busyq); bc->bc_req_count = 0; for (i = 0; i < BLOCKIF_MAXREQ; i++) { bc->bc_reqs[i].be_status = BST_FREE; TAILQ_INSERT_HEAD(&bc->bc_freeq, &bc->bc_reqs[i], be_link); } pthread_create(&bc->bc_btid, NULL, blockif_thr, bc); snprintf(tname, sizeof(tname), "blk-%s", ident); pthread_set_name_np(bc->bc_btid, tname); return (bc); } static int blockif_request(struct blockif_ctxt *bc, struct blockif_req *breq, enum blockop op) { int err; err = 0; pthread_mutex_lock(&bc->bc_mtx); if (bc->bc_req_count < BLOCKIF_MAXREQ) { /* * Enqueue and inform the block i/o thread * that there is work available */ blockif_enqueue(bc, breq, op); pthread_cond_signal(&bc->bc_cond); } else { /* * Callers are not allowed to enqueue more than * the specified blockif queue limit. Return an * error to indicate that the queue length has been * exceeded. */ err = E2BIG; } pthread_mutex_unlock(&bc->bc_mtx); return (err); } int blockif_read(struct blockif_ctxt *bc, struct blockif_req *breq) { assert(bc->bc_magic == BLOCKIF_SIG); return (blockif_request(bc, breq, BOP_READ)); } int blockif_write(struct blockif_ctxt *bc, struct blockif_req *breq) { assert(bc->bc_magic == BLOCKIF_SIG); return (blockif_request(bc, breq, BOP_WRITE)); } int blockif_flush(struct blockif_ctxt *bc, struct blockif_req *breq) { assert(bc->bc_magic == BLOCKIF_SIG); return (blockif_request(bc, breq, BOP_FLUSH)); } int blockif_cancel(struct blockif_ctxt *bc, struct blockif_req *breq) { struct blockif_elem *be; assert(bc->bc_magic == BLOCKIF_SIG); pthread_mutex_lock(&bc->bc_mtx); /* * Check pending requests. */ TAILQ_FOREACH(be, &bc->bc_pendq, be_link) { if (be->be_req == breq) break; } if (be != NULL) { /* * Found it. */ TAILQ_REMOVE(&bc->bc_pendq, be, be_link); be->be_status = BST_FREE; be->be_req = NULL; TAILQ_INSERT_TAIL(&bc->bc_freeq, be, be_link); bc->bc_req_count--; pthread_mutex_unlock(&bc->bc_mtx); return (0); } /* * Check in-flight requests. */ TAILQ_FOREACH(be, &bc->bc_busyq, be_link) { if (be->be_req == breq) break; } if (be == NULL) { /* * Didn't find it. */ pthread_mutex_unlock(&bc->bc_mtx); return (EINVAL); } /* * Interrupt the processing thread to force it return * prematurely via it's normal callback path. */ while (be->be_status == BST_BUSY) { struct blockif_sig_elem bse, *old_head; pthread_mutex_init(&bse.bse_mtx, NULL); pthread_cond_init(&bse.bse_cond, NULL); bse.bse_pending = 1; do { old_head = blockif_bse_head; bse.bse_next = old_head; } while (!atomic_cmpset_ptr((uintptr_t *)&blockif_bse_head, (uintptr_t)old_head, (uintptr_t)&bse)); pthread_kill(be->be_tid, SIGCONT); pthread_mutex_lock(&bse.bse_mtx); while (bse.bse_pending) pthread_cond_wait(&bse.bse_cond, &bse.bse_mtx); pthread_mutex_unlock(&bse.bse_mtx); } pthread_mutex_unlock(&bc->bc_mtx); /* * The processing thread has been interrupted. Since it's not * clear if the callback has been invoked yet, return EBUSY. */ return (EBUSY); } int blockif_close(struct blockif_ctxt *bc) { void *jval; int err; err = 0; assert(bc->bc_magic == BLOCKIF_SIG); /* * Stop the block i/o thread */ bc->bc_closing = 1; pthread_cond_signal(&bc->bc_cond); pthread_join(bc->bc_btid, &jval); /* XXX Cancel queued i/o's ??? */ /* * Release resources */ bc->bc_magic = 0; close(bc->bc_fd); free(bc); return (0); } /* * Return virtual C/H/S values for a given block. Use the algorithm * outlined in the VHD specification to calculate values. */ void blockif_chs(struct blockif_ctxt *bc, uint16_t *c, uint8_t *h, uint8_t *s) { off_t sectors; /* total sectors of the block dev */ off_t hcyl; /* cylinders times heads */ uint16_t secpt; /* sectors per track */ uint8_t heads; assert(bc->bc_magic == BLOCKIF_SIG); sectors = bc->bc_size / bc->bc_sectsz; /* Clamp the size to the largest possible with CHS */ if (sectors > 65535UL*16*255) sectors = 65535UL*16*255; if (sectors >= 65536UL*16*63) { secpt = 255; heads = 16; hcyl = sectors / secpt; } else { secpt = 17; hcyl = sectors / secpt; heads = (hcyl + 1023) / 1024; if (heads < 4) heads = 4; if (hcyl >= (heads * 1024) || heads > 16) { secpt = 31; heads = 16; hcyl = sectors / secpt; } if (hcyl >= (heads * 1024)) { secpt = 63; heads = 16; hcyl = sectors / secpt; } } *c = hcyl / heads; *h = heads; *s = secpt; } /* * Accessors */ off_t blockif_size(struct blockif_ctxt *bc) { assert(bc->bc_magic == BLOCKIF_SIG); return (bc->bc_size); } int blockif_sectsz(struct blockif_ctxt *bc) { assert(bc->bc_magic == BLOCKIF_SIG); return (bc->bc_sectsz); +} + +void +blockif_psectsz(struct blockif_ctxt *bc, int *size, int *off) +{ + + assert(bc->bc_magic == BLOCKIF_SIG); + *size = bc->bc_psectsz; + *off = bc->bc_psectoff; } int blockif_queuesz(struct blockif_ctxt *bc) { assert(bc->bc_magic == BLOCKIF_SIG); return (BLOCKIF_MAXREQ - 1); } int blockif_is_ro(struct blockif_ctxt *bc) { assert(bc->bc_magic == BLOCKIF_SIG); return (bc->bc_rdonly); } Index: projects/cxl_iscsi/usr.sbin/bhyve/block_if.h =================================================================== --- projects/cxl_iscsi/usr.sbin/bhyve/block_if.h (revision 279716) +++ projects/cxl_iscsi/usr.sbin/bhyve/block_if.h (revision 279717) @@ -1,66 +1,67 @@ /*- * Copyright (c) 2013 Peter Grehan * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ /* * The block API to be used by bhyve block-device emulations. The routines * are thread safe, with no assumptions about the context of the completion * callback - it may occur in the caller's context, or asynchronously in * another thread. */ #ifndef _BLOCK_IF_H_ #define _BLOCK_IF_H_ #include #include #define BLOCKIF_IOV_MAX 32 /* not practical to be IOV_MAX */ struct blockif_req { struct iovec br_iov[BLOCKIF_IOV_MAX]; int br_iovcnt; off_t br_offset; void (*br_callback)(struct blockif_req *req, int err); void *br_param; }; struct blockif_ctxt; struct blockif_ctxt *blockif_open(const char *optstr, const char *ident); off_t blockif_size(struct blockif_ctxt *bc); void blockif_chs(struct blockif_ctxt *bc, uint16_t *c, uint8_t *h, uint8_t *s); int blockif_sectsz(struct blockif_ctxt *bc); +void blockif_psectsz(struct blockif_ctxt *bc, int *size, int *off); int blockif_queuesz(struct blockif_ctxt *bc); int blockif_is_ro(struct blockif_ctxt *bc); int blockif_read(struct blockif_ctxt *bc, struct blockif_req *breq); int blockif_write(struct blockif_ctxt *bc, struct blockif_req *breq); int blockif_flush(struct blockif_ctxt *bc, struct blockif_req *breq); int blockif_cancel(struct blockif_ctxt *bc, struct blockif_req *breq); int blockif_close(struct blockif_ctxt *bc); #endif /* _BLOCK_IF_H_ */ Index: projects/cxl_iscsi/usr.sbin/bhyve/pci_ahci.c =================================================================== --- projects/cxl_iscsi/usr.sbin/bhyve/pci_ahci.c (revision 279716) +++ projects/cxl_iscsi/usr.sbin/bhyve/pci_ahci.c (revision 279717) @@ -1,2010 +1,2025 @@ /*- * Copyright (c) 2013 Zhixiang Yu * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "bhyverun.h" #include "pci_emul.h" #include "ahci.h" #include "block_if.h" #define MAX_PORTS 6 /* Intel ICH8 AHCI supports 6 ports */ #define PxSIG_ATA 0x00000101 /* ATA drive */ #define PxSIG_ATAPI 0xeb140101 /* ATAPI drive */ enum sata_fis_type { FIS_TYPE_REGH2D = 0x27, /* Register FIS - host to device */ FIS_TYPE_REGD2H = 0x34, /* Register FIS - device to host */ FIS_TYPE_DMAACT = 0x39, /* DMA activate FIS - device to host */ FIS_TYPE_DMASETUP = 0x41, /* DMA setup FIS - bidirectional */ FIS_TYPE_DATA = 0x46, /* Data FIS - bidirectional */ FIS_TYPE_BIST = 0x58, /* BIST activate FIS - bidirectional */ FIS_TYPE_PIOSETUP = 0x5F, /* PIO setup FIS - device to host */ FIS_TYPE_SETDEVBITS = 0xA1, /* Set dev bits FIS - device to host */ }; /* * SCSI opcodes */ #define TEST_UNIT_READY 0x00 #define REQUEST_SENSE 0x03 #define INQUIRY 0x12 #define START_STOP_UNIT 0x1B #define PREVENT_ALLOW 0x1E #define READ_CAPACITY 0x25 #define READ_10 0x28 #define POSITION_TO_ELEMENT 0x2B #define READ_TOC 0x43 #define GET_EVENT_STATUS_NOTIFICATION 0x4A #define MODE_SENSE_10 0x5A #define READ_12 0xA8 #define READ_CD 0xBE /* * SCSI mode page codes */ #define MODEPAGE_RW_ERROR_RECOVERY 0x01 #define MODEPAGE_CD_CAPABILITIES 0x2A /* * ATA commands */ #define ATA_SF_ENAB_SATA_SF 0x10 #define ATA_SATA_SF_AN 0x05 #define ATA_SF_DIS_SATA_SF 0x90 /* * Debug printf */ #ifdef AHCI_DEBUG static FILE *dbg; #define DPRINTF(format, arg...) do{fprintf(dbg, format, ##arg);fflush(dbg);}while(0) #else #define DPRINTF(format, arg...) #endif #define WPRINTF(format, arg...) printf(format, ##arg) struct ahci_ioreq { struct blockif_req io_req; struct ahci_port *io_pr; STAILQ_ENTRY(ahci_ioreq) io_flist; TAILQ_ENTRY(ahci_ioreq) io_blist; uint8_t *cfis; uint32_t len; uint32_t done; int slot; int prdtl; }; struct ahci_port { struct blockif_ctxt *bctx; struct pci_ahci_softc *pr_sc; uint8_t *cmd_lst; uint8_t *rfis; int atapi; int reset; int mult_sectors; uint8_t xfermode; uint8_t sense_key; uint8_t asc; uint32_t pending; uint32_t clb; uint32_t clbu; uint32_t fb; uint32_t fbu; uint32_t is; uint32_t ie; uint32_t cmd; uint32_t unused0; uint32_t tfd; uint32_t sig; uint32_t ssts; uint32_t sctl; uint32_t serr; uint32_t sact; uint32_t ci; uint32_t sntf; uint32_t fbs; /* * i/o request info */ struct ahci_ioreq *ioreq; int ioqsz; STAILQ_HEAD(ahci_fhead, ahci_ioreq) iofhd; TAILQ_HEAD(ahci_bhead, ahci_ioreq) iobhd; }; struct ahci_cmd_hdr { uint16_t flags; uint16_t prdtl; uint32_t prdbc; uint64_t ctba; uint32_t reserved[4]; }; struct ahci_prdt_entry { uint64_t dba; uint32_t reserved; #define DBCMASK 0x3fffff uint32_t dbc; }; struct pci_ahci_softc { struct pci_devinst *asc_pi; pthread_mutex_t mtx; int ports; uint32_t cap; uint32_t ghc; uint32_t is; uint32_t pi; uint32_t vs; uint32_t ccc_ctl; uint32_t ccc_pts; uint32_t em_loc; uint32_t em_ctl; uint32_t cap2; uint32_t bohc; uint32_t lintr; struct ahci_port port[MAX_PORTS]; }; #define ahci_ctx(sc) ((sc)->asc_pi->pi_vmctx) static inline void lba_to_msf(uint8_t *buf, int lba) { lba += 150; buf[0] = (lba / 75) / 60; buf[1] = (lba / 75) % 60; buf[2] = lba % 75; } /* * generate HBA intr depending on whether or not ports within * the controller have an interrupt pending. */ static void ahci_generate_intr(struct pci_ahci_softc *sc) { struct pci_devinst *pi; int i; pi = sc->asc_pi; for (i = 0; i < sc->ports; i++) { struct ahci_port *pr; pr = &sc->port[i]; if (pr->is & pr->ie) sc->is |= (1 << i); } DPRINTF("%s %x\n", __func__, sc->is); if (sc->is && (sc->ghc & AHCI_GHC_IE)) { if (pci_msi_enabled(pi)) { /* * Generate an MSI interrupt on every edge */ pci_generate_msi(pi, 0); } else if (!sc->lintr) { /* * Only generate a pin-based interrupt if one wasn't * in progress */ sc->lintr = 1; pci_lintr_assert(pi); } } else if (sc->lintr) { /* * No interrupts: deassert pin-based signal if it had * been asserted */ pci_lintr_deassert(pi); sc->lintr = 0; } } static void ahci_write_fis(struct ahci_port *p, enum sata_fis_type ft, uint8_t *fis) { int offset, len, irq; if (p->rfis == NULL || !(p->cmd & AHCI_P_CMD_FRE)) return; switch (ft) { case FIS_TYPE_REGD2H: offset = 0x40; len = 20; irq = AHCI_P_IX_DHR; break; case FIS_TYPE_SETDEVBITS: offset = 0x58; len = 8; irq = AHCI_P_IX_SDB; break; case FIS_TYPE_PIOSETUP: offset = 0x20; len = 20; irq = 0; break; default: WPRINTF("unsupported fis type %d\n", ft); return; } memcpy(p->rfis + offset, fis, len); if (irq) { p->is |= irq; ahci_generate_intr(p->pr_sc); } } static void ahci_write_fis_piosetup(struct ahci_port *p) { uint8_t fis[20]; memset(fis, 0, sizeof(fis)); fis[0] = FIS_TYPE_PIOSETUP; ahci_write_fis(p, FIS_TYPE_PIOSETUP, fis); } static void ahci_write_fis_sdb(struct ahci_port *p, int slot, uint32_t tfd) { uint8_t fis[8]; uint8_t error; error = (tfd >> 8) & 0xff; memset(fis, 0, sizeof(fis)); fis[0] = error; fis[2] = tfd & 0x77; *(uint32_t *)(fis + 4) = (1 << slot); if (fis[2] & ATA_S_ERROR) p->is |= AHCI_P_IX_TFE; p->tfd = tfd; ahci_write_fis(p, FIS_TYPE_SETDEVBITS, fis); } static void ahci_write_fis_d2h(struct ahci_port *p, int slot, uint8_t *cfis, uint32_t tfd) { uint8_t fis[20]; uint8_t error; error = (tfd >> 8) & 0xff; memset(fis, 0, sizeof(fis)); fis[0] = FIS_TYPE_REGD2H; fis[1] = (1 << 6); fis[2] = tfd & 0xff; fis[3] = error; fis[4] = cfis[4]; fis[5] = cfis[5]; fis[6] = cfis[6]; fis[7] = cfis[7]; fis[8] = cfis[8]; fis[9] = cfis[9]; fis[10] = cfis[10]; fis[11] = cfis[11]; fis[12] = cfis[12]; fis[13] = cfis[13]; if (fis[2] & ATA_S_ERROR) p->is |= AHCI_P_IX_TFE; else p->ci &= ~(1 << slot); p->tfd = tfd; ahci_write_fis(p, FIS_TYPE_REGD2H, fis); } static void ahci_write_reset_fis_d2h(struct ahci_port *p) { uint8_t fis[20]; memset(fis, 0, sizeof(fis)); fis[0] = FIS_TYPE_REGD2H; fis[3] = 1; fis[4] = 1; if (p->atapi) { fis[5] = 0x14; fis[6] = 0xeb; } fis[12] = 1; ahci_write_fis(p, FIS_TYPE_REGD2H, fis); } static void ahci_check_stopped(struct ahci_port *p) { /* * If we are no longer processing the command list and nothing * is in-flight, clear the running bit, the current command * slot, the command issue and active bits. */ if (!(p->cmd & AHCI_P_CMD_ST)) { if (p->pending == 0) { p->cmd &= ~(AHCI_P_CMD_CR | AHCI_P_CMD_CCS_MASK); p->ci = 0; p->sact = 0; } } } static void ahci_port_stop(struct ahci_port *p) { struct ahci_ioreq *aior; uint8_t *cfis; int slot; int ncq; int error; assert(pthread_mutex_isowned_np(&p->pr_sc->mtx)); TAILQ_FOREACH(aior, &p->iobhd, io_blist) { /* * Try to cancel the outstanding blockif request. */ error = blockif_cancel(p->bctx, &aior->io_req); if (error != 0) continue; slot = aior->slot; cfis = aior->cfis; if (cfis[2] == ATA_WRITE_FPDMA_QUEUED || cfis[2] == ATA_READ_FPDMA_QUEUED) ncq = 1; if (ncq) p->sact &= ~(1 << slot); else p->ci &= ~(1 << slot); /* * This command is now done. */ p->pending &= ~(1 << slot); /* * Delete the blockif request from the busy list */ TAILQ_REMOVE(&p->iobhd, aior, io_blist); /* * Move the blockif request back to the free list */ STAILQ_INSERT_TAIL(&p->iofhd, aior, io_flist); } ahci_check_stopped(p); } static void ahci_port_reset(struct ahci_port *pr) { pr->sctl = 0; pr->serr = 0; pr->sact = 0; pr->xfermode = ATA_UDMA6; pr->mult_sectors = 128; if (!pr->bctx) { pr->ssts = ATA_SS_DET_NO_DEVICE; pr->sig = 0xFFFFFFFF; pr->tfd = 0x7F; return; } pr->ssts = ATA_SS_DET_PHY_ONLINE | ATA_SS_SPD_GEN2 | ATA_SS_IPM_ACTIVE; pr->tfd = (1 << 8) | ATA_S_DSC | ATA_S_DMA; if (!pr->atapi) { pr->sig = PxSIG_ATA; pr->tfd |= ATA_S_READY; } else pr->sig = PxSIG_ATAPI; ahci_write_reset_fis_d2h(pr); } static void ahci_reset(struct pci_ahci_softc *sc) { int i; sc->ghc = AHCI_GHC_AE; sc->is = 0; if (sc->lintr) { pci_lintr_deassert(sc->asc_pi); sc->lintr = 0; } for (i = 0; i < sc->ports; i++) { sc->port[i].ie = 0; sc->port[i].is = 0; ahci_port_reset(&sc->port[i]); } } static void ata_string(uint8_t *dest, const char *src, int len) { int i; for (i = 0; i < len; i++) { if (*src) dest[i ^ 1] = *src++; else dest[i ^ 1] = ' '; } } static void atapi_string(uint8_t *dest, const char *src, int len) { int i; for (i = 0; i < len; i++) { if (*src) dest[i] = *src++; else dest[i] = ' '; } } static void ahci_handle_dma(struct ahci_port *p, int slot, uint8_t *cfis, uint32_t done, int seek) { struct ahci_ioreq *aior; struct blockif_req *breq; struct pci_ahci_softc *sc; struct ahci_prdt_entry *prdt; struct ahci_cmd_hdr *hdr; uint64_t lba; uint32_t len; int i, err, iovcnt, ncq, readop; sc = p->pr_sc; prdt = (struct ahci_prdt_entry *)(cfis + 0x80); hdr = (struct ahci_cmd_hdr *)(p->cmd_lst + slot * AHCI_CL_SIZE); ncq = 0; readop = 1; prdt += seek; if (cfis[2] == ATA_WRITE_DMA || cfis[2] == ATA_WRITE_DMA48 || cfis[2] == ATA_WRITE_FPDMA_QUEUED) readop = 0; if (cfis[2] == ATA_WRITE_FPDMA_QUEUED || cfis[2] == ATA_READ_FPDMA_QUEUED) { lba = ((uint64_t)cfis[10] << 40) | ((uint64_t)cfis[9] << 32) | ((uint64_t)cfis[8] << 24) | ((uint64_t)cfis[6] << 16) | ((uint64_t)cfis[5] << 8) | cfis[4]; len = cfis[11] << 8 | cfis[3]; if (!len) len = 65536; ncq = 1; } else if (cfis[2] == ATA_READ_DMA48 || cfis[2] == ATA_WRITE_DMA48) { lba = ((uint64_t)cfis[10] << 40) | ((uint64_t)cfis[9] << 32) | ((uint64_t)cfis[8] << 24) | ((uint64_t)cfis[6] << 16) | ((uint64_t)cfis[5] << 8) | cfis[4]; len = cfis[13] << 8 | cfis[12]; if (!len) len = 65536; } else { lba = ((cfis[7] & 0xf) << 24) | (cfis[6] << 16) | (cfis[5] << 8) | cfis[4]; len = cfis[12]; if (!len) len = 256; } lba *= blockif_sectsz(p->bctx); len *= blockif_sectsz(p->bctx); /* * Pull request off free list */ aior = STAILQ_FIRST(&p->iofhd); assert(aior != NULL); STAILQ_REMOVE_HEAD(&p->iofhd, io_flist); aior->cfis = cfis; aior->slot = slot; aior->len = len; aior->done = done; breq = &aior->io_req; breq->br_offset = lba + done; iovcnt = hdr->prdtl - seek; if (iovcnt > BLOCKIF_IOV_MAX) { aior->prdtl = iovcnt - BLOCKIF_IOV_MAX; iovcnt = BLOCKIF_IOV_MAX; } else aior->prdtl = 0; breq->br_iovcnt = iovcnt; /* * Mark this command in-flight. */ p->pending |= 1 << slot; /* * Stuff request onto busy list */ TAILQ_INSERT_HEAD(&p->iobhd, aior, io_blist); /* * Build up the iovec based on the prdt */ for (i = 0; i < iovcnt; i++) { uint32_t dbcsz; dbcsz = (prdt->dbc & DBCMASK) + 1; breq->br_iov[i].iov_base = paddr_guest2host(ahci_ctx(sc), prdt->dba, dbcsz); breq->br_iov[i].iov_len = dbcsz; aior->done += dbcsz; prdt++; } if (readop) err = blockif_read(p->bctx, breq); else err = blockif_write(p->bctx, breq); assert(err == 0); if (ncq) p->ci &= ~(1 << slot); } static void ahci_handle_flush(struct ahci_port *p, int slot, uint8_t *cfis) { struct ahci_ioreq *aior; struct blockif_req *breq; int err; /* * Pull request off free list */ aior = STAILQ_FIRST(&p->iofhd); assert(aior != NULL); STAILQ_REMOVE_HEAD(&p->iofhd, io_flist); aior->cfis = cfis; aior->slot = slot; aior->len = 0; aior->done = 0; aior->prdtl = 0; breq = &aior->io_req; /* * Mark this command in-flight. */ p->pending |= 1 << slot; /* * Stuff request onto busy list */ TAILQ_INSERT_HEAD(&p->iobhd, aior, io_blist); err = blockif_flush(p->bctx, breq); assert(err == 0); } static inline void write_prdt(struct ahci_port *p, int slot, uint8_t *cfis, void *buf, int size) { struct ahci_cmd_hdr *hdr; struct ahci_prdt_entry *prdt; void *from; int i, len; hdr = (struct ahci_cmd_hdr *)(p->cmd_lst + slot * AHCI_CL_SIZE); len = size; from = buf; prdt = (struct ahci_prdt_entry *)(cfis + 0x80); for (i = 0; i < hdr->prdtl && len; i++) { uint8_t *ptr; uint32_t dbcsz; int sublen; dbcsz = (prdt->dbc & DBCMASK) + 1; ptr = paddr_guest2host(ahci_ctx(p->pr_sc), prdt->dba, dbcsz); sublen = len < dbcsz ? len : dbcsz; memcpy(ptr, from, sublen); len -= sublen; from += sublen; prdt++; } hdr->prdbc = size - len; } static void handle_identify(struct ahci_port *p, int slot, uint8_t *cfis) { struct ahci_cmd_hdr *hdr; hdr = (struct ahci_cmd_hdr *)(p->cmd_lst + slot * AHCI_CL_SIZE); if (p->atapi || hdr->prdtl == 0) { p->tfd = (ATA_E_ABORT << 8) | ATA_S_READY | ATA_S_ERROR; p->is |= AHCI_P_IX_TFE; } else { uint16_t buf[256]; uint64_t sectors; + int sectsz, psectsz, psectoff; uint16_t cyl; uint8_t sech, heads; - sectors = blockif_size(p->bctx) / blockif_sectsz(p->bctx); + sectsz = blockif_sectsz(p->bctx); + sectors = blockif_size(p->bctx) / sectsz; blockif_chs(p->bctx, &cyl, &heads, &sech); + blockif_psectsz(p->bctx, &psectsz, &psectoff); memset(buf, 0, sizeof(buf)); buf[0] = 0x0040; buf[1] = cyl; buf[3] = heads; buf[6] = sech; /* TODO emulate different serial? */ ata_string((uint8_t *)(buf+10), "123456", 20); ata_string((uint8_t *)(buf+23), "001", 8); ata_string((uint8_t *)(buf+27), "BHYVE SATA DISK", 40); buf[47] = (0x8000 | 128); buf[48] = 0x1; buf[49] = (1 << 8 | 1 << 9 | 1 << 11); buf[50] = (1 << 14); buf[53] = (1 << 1 | 1 << 2); if (p->mult_sectors) buf[59] = (0x100 | p->mult_sectors); buf[60] = sectors; buf[61] = (sectors >> 16); buf[63] = 0x7; if (p->xfermode & ATA_WDMA0) buf[63] |= (1 << ((p->xfermode & 7) + 8)); buf[64] = 0x3; buf[65] = 100; buf[66] = 100; buf[67] = 100; buf[68] = 100; buf[75] = 31; buf[76] = (1 << 8 | 1 << 2); buf[80] = 0x1f0; buf[81] = 0x28; buf[82] = (1 << 5 | 1 << 14); buf[83] = (1 << 10 | 1 << 12 | 1 << 13 | 1 << 14); buf[84] = (1 << 14); buf[85] = (1 << 5 | 1 << 14); buf[86] = (1 << 10 | 1 << 12 | 1 << 13); buf[87] = (1 << 14); buf[88] = 0x7f; if (p->xfermode & ATA_UDMA0) buf[88] |= (1 << ((p->xfermode & 7) + 8)); buf[93] = (1 | 1 <<14); buf[100] = sectors; buf[101] = (sectors >> 16); buf[102] = (sectors >> 32); buf[103] = (sectors >> 48); + buf[106] = 0x4000; + buf[209] = 0x4000; + if (psectsz > sectsz) { + buf[106] |= 0x2000; + buf[106] |= ffsl(psectsz / sectsz) - 1; + buf[209] |= (psectoff / sectsz); + } + if (sectsz > 512) { + buf[106] |= 0x1000; + buf[117] = sectsz / 2; + buf[118] = ((sectsz / 2) >> 16); + } ahci_write_fis_piosetup(p); write_prdt(p, slot, cfis, (void *)buf, sizeof(buf)); p->tfd = ATA_S_DSC | ATA_S_READY; p->is |= AHCI_P_IX_DP; p->ci &= ~(1 << slot); } ahci_generate_intr(p->pr_sc); } static void handle_atapi_identify(struct ahci_port *p, int slot, uint8_t *cfis) { if (!p->atapi) { p->tfd = (ATA_E_ABORT << 8) | ATA_S_READY | ATA_S_ERROR; p->is |= AHCI_P_IX_TFE; } else { uint16_t buf[256]; memset(buf, 0, sizeof(buf)); buf[0] = (2 << 14 | 5 << 8 | 1 << 7 | 2 << 5); /* TODO emulate different serial? */ ata_string((uint8_t *)(buf+10), "123456", 20); ata_string((uint8_t *)(buf+23), "001", 8); ata_string((uint8_t *)(buf+27), "BHYVE SATA DVD ROM", 40); buf[49] = (1 << 9 | 1 << 8); buf[50] = (1 << 14 | 1); buf[53] = (1 << 2 | 1 << 1); buf[62] = 0x3f; buf[63] = 7; buf[64] = 3; buf[65] = 100; buf[66] = 100; buf[67] = 100; buf[68] = 100; buf[76] = (1 << 2 | 1 << 1); buf[78] = (1 << 5); buf[80] = (0x1f << 4); buf[82] = (1 << 4); buf[83] = (1 << 14); buf[84] = (1 << 14); buf[85] = (1 << 4); buf[87] = (1 << 14); buf[88] = (1 << 14 | 0x7f); ahci_write_fis_piosetup(p); write_prdt(p, slot, cfis, (void *)buf, sizeof(buf)); p->tfd = ATA_S_DSC | ATA_S_READY; p->is |= AHCI_P_IX_DHR; p->ci &= ~(1 << slot); } ahci_generate_intr(p->pr_sc); } static void atapi_inquiry(struct ahci_port *p, int slot, uint8_t *cfis) { uint8_t buf[36]; uint8_t *acmd; int len; acmd = cfis + 0x40; buf[0] = 0x05; buf[1] = 0x80; buf[2] = 0x00; buf[3] = 0x21; buf[4] = 31; buf[5] = 0; buf[6] = 0; buf[7] = 0; atapi_string(buf + 8, "BHYVE", 8); atapi_string(buf + 16, "BHYVE DVD-ROM", 16); atapi_string(buf + 32, "001", 4); len = sizeof(buf); if (len > acmd[4]) len = acmd[4]; cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; write_prdt(p, slot, cfis, buf, len); ahci_write_fis_d2h(p, slot, cfis, ATA_S_READY | ATA_S_DSC); } static void atapi_read_capacity(struct ahci_port *p, int slot, uint8_t *cfis) { uint8_t buf[8]; uint64_t sectors; sectors = blockif_size(p->bctx) / 2048; be32enc(buf, sectors - 1); be32enc(buf + 4, 2048); cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; write_prdt(p, slot, cfis, buf, sizeof(buf)); ahci_write_fis_d2h(p, slot, cfis, ATA_S_READY | ATA_S_DSC); } static void atapi_read_toc(struct ahci_port *p, int slot, uint8_t *cfis) { uint8_t *acmd; uint8_t format; int len; acmd = cfis + 0x40; len = be16dec(acmd + 7); format = acmd[9] >> 6; switch (format) { case 0: { int msf, size; uint64_t sectors; uint8_t start_track, buf[20], *bp; msf = (acmd[1] >> 1) & 1; start_track = acmd[6]; if (start_track > 1 && start_track != 0xaa) { uint32_t tfd; p->sense_key = ATA_SENSE_ILLEGAL_REQUEST; p->asc = 0x24; tfd = (p->sense_key << 12) | ATA_S_READY | ATA_S_ERROR; cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; ahci_write_fis_d2h(p, slot, cfis, tfd); return; } bp = buf + 2; *bp++ = 1; *bp++ = 1; if (start_track <= 1) { *bp++ = 0; *bp++ = 0x14; *bp++ = 1; *bp++ = 0; if (msf) { *bp++ = 0; lba_to_msf(bp, 0); bp += 3; } else { *bp++ = 0; *bp++ = 0; *bp++ = 0; *bp++ = 0; } } *bp++ = 0; *bp++ = 0x14; *bp++ = 0xaa; *bp++ = 0; sectors = blockif_size(p->bctx) / blockif_sectsz(p->bctx); sectors >>= 2; if (msf) { *bp++ = 0; lba_to_msf(bp, sectors); bp += 3; } else { be32enc(bp, sectors); bp += 4; } size = bp - buf; be16enc(buf, size - 2); if (len > size) len = size; write_prdt(p, slot, cfis, buf, len); cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; ahci_write_fis_d2h(p, slot, cfis, ATA_S_READY | ATA_S_DSC); break; } case 1: { uint8_t buf[12]; memset(buf, 0, sizeof(buf)); buf[1] = 0xa; buf[2] = 0x1; buf[3] = 0x1; if (len > sizeof(buf)) len = sizeof(buf); write_prdt(p, slot, cfis, buf, len); cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; ahci_write_fis_d2h(p, slot, cfis, ATA_S_READY | ATA_S_DSC); break; } case 2: { int msf, size; uint64_t sectors; uint8_t start_track, *bp, buf[50]; msf = (acmd[1] >> 1) & 1; start_track = acmd[6]; bp = buf + 2; *bp++ = 1; *bp++ = 1; *bp++ = 1; *bp++ = 0x14; *bp++ = 0; *bp++ = 0xa0; *bp++ = 0; *bp++ = 0; *bp++ = 0; *bp++ = 0; *bp++ = 1; *bp++ = 0; *bp++ = 0; *bp++ = 1; *bp++ = 0x14; *bp++ = 0; *bp++ = 0xa1; *bp++ = 0; *bp++ = 0; *bp++ = 0; *bp++ = 0; *bp++ = 1; *bp++ = 0; *bp++ = 0; *bp++ = 1; *bp++ = 0x14; *bp++ = 0; *bp++ = 0xa2; *bp++ = 0; *bp++ = 0; *bp++ = 0; sectors = blockif_size(p->bctx) / blockif_sectsz(p->bctx); sectors >>= 2; if (msf) { *bp++ = 0; lba_to_msf(bp, sectors); bp += 3; } else { be32enc(bp, sectors); bp += 4; } *bp++ = 1; *bp++ = 0x14; *bp++ = 0; *bp++ = 1; *bp++ = 0; *bp++ = 0; *bp++ = 0; if (msf) { *bp++ = 0; lba_to_msf(bp, 0); bp += 3; } else { *bp++ = 0; *bp++ = 0; *bp++ = 0; *bp++ = 0; } size = bp - buf; be16enc(buf, size - 2); if (len > size) len = size; write_prdt(p, slot, cfis, buf, len); cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; ahci_write_fis_d2h(p, slot, cfis, ATA_S_READY | ATA_S_DSC); break; } default: { uint32_t tfd; p->sense_key = ATA_SENSE_ILLEGAL_REQUEST; p->asc = 0x24; tfd = (p->sense_key << 12) | ATA_S_READY | ATA_S_ERROR; cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; ahci_write_fis_d2h(p, slot, cfis, tfd); break; } } } static void atapi_read(struct ahci_port *p, int slot, uint8_t *cfis, uint32_t done, int seek) { struct ahci_ioreq *aior; struct ahci_cmd_hdr *hdr; struct ahci_prdt_entry *prdt; struct blockif_req *breq; struct pci_ahci_softc *sc; uint8_t *acmd; uint64_t lba; uint32_t len; int i, err, iovcnt; sc = p->pr_sc; acmd = cfis + 0x40; hdr = (struct ahci_cmd_hdr *)(p->cmd_lst + slot * AHCI_CL_SIZE); prdt = (struct ahci_prdt_entry *)(cfis + 0x80); prdt += seek; lba = be32dec(acmd + 2); if (acmd[0] == READ_10) len = be16dec(acmd + 7); else len = be32dec(acmd + 6); if (len == 0) { cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; ahci_write_fis_d2h(p, slot, cfis, ATA_S_READY | ATA_S_DSC); } lba *= 2048; len *= 2048; /* * Pull request off free list */ aior = STAILQ_FIRST(&p->iofhd); assert(aior != NULL); STAILQ_REMOVE_HEAD(&p->iofhd, io_flist); aior->cfis = cfis; aior->slot = slot; aior->len = len; aior->done = done; breq = &aior->io_req; breq->br_offset = lba + done; iovcnt = hdr->prdtl - seek; if (iovcnt > BLOCKIF_IOV_MAX) { aior->prdtl = iovcnt - BLOCKIF_IOV_MAX; iovcnt = BLOCKIF_IOV_MAX; } else aior->prdtl = 0; breq->br_iovcnt = iovcnt; /* * Mark this command in-flight. */ p->pending |= 1 << slot; /* * Stuff request onto busy list */ TAILQ_INSERT_HEAD(&p->iobhd, aior, io_blist); /* * Build up the iovec based on the prdt */ for (i = 0; i < iovcnt; i++) { uint32_t dbcsz; dbcsz = (prdt->dbc & DBCMASK) + 1; breq->br_iov[i].iov_base = paddr_guest2host(ahci_ctx(sc), prdt->dba, dbcsz); breq->br_iov[i].iov_len = dbcsz; aior->done += dbcsz; prdt++; } err = blockif_read(p->bctx, breq); assert(err == 0); } static void atapi_request_sense(struct ahci_port *p, int slot, uint8_t *cfis) { uint8_t buf[64]; uint8_t *acmd; int len; acmd = cfis + 0x40; len = acmd[4]; if (len > sizeof(buf)) len = sizeof(buf); memset(buf, 0, len); buf[0] = 0x70 | (1 << 7); buf[2] = p->sense_key; buf[7] = 10; buf[12] = p->asc; write_prdt(p, slot, cfis, buf, len); cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; ahci_write_fis_d2h(p, slot, cfis, ATA_S_READY | ATA_S_DSC); } static void atapi_start_stop_unit(struct ahci_port *p, int slot, uint8_t *cfis) { uint8_t *acmd = cfis + 0x40; uint32_t tfd; switch (acmd[4] & 3) { case 0: case 1: case 3: cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; tfd = ATA_S_READY | ATA_S_DSC; break; case 2: /* TODO eject media */ cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; p->sense_key = ATA_SENSE_ILLEGAL_REQUEST; p->asc = 0x53; tfd = (p->sense_key << 12) | ATA_S_READY | ATA_S_ERROR; break; } ahci_write_fis_d2h(p, slot, cfis, tfd); } static void atapi_mode_sense(struct ahci_port *p, int slot, uint8_t *cfis) { uint8_t *acmd; uint32_t tfd; uint8_t pc, code; int len; acmd = cfis + 0x40; len = be16dec(acmd + 7); pc = acmd[2] >> 6; code = acmd[2] & 0x3f; switch (pc) { case 0: switch (code) { case MODEPAGE_RW_ERROR_RECOVERY: { uint8_t buf[16]; if (len > sizeof(buf)) len = sizeof(buf); memset(buf, 0, sizeof(buf)); be16enc(buf, 16 - 2); buf[2] = 0x70; buf[8] = 0x01; buf[9] = 16 - 10; buf[11] = 0x05; write_prdt(p, slot, cfis, buf, len); tfd = ATA_S_READY | ATA_S_DSC; break; } case MODEPAGE_CD_CAPABILITIES: { uint8_t buf[30]; if (len > sizeof(buf)) len = sizeof(buf); memset(buf, 0, sizeof(buf)); be16enc(buf, 30 - 2); buf[2] = 0x70; buf[8] = 0x2A; buf[9] = 30 - 10; buf[10] = 0x08; buf[12] = 0x71; be16enc(&buf[18], 2); be16enc(&buf[20], 512); write_prdt(p, slot, cfis, buf, len); tfd = ATA_S_READY | ATA_S_DSC; break; } default: goto error; break; } break; case 3: p->sense_key = ATA_SENSE_ILLEGAL_REQUEST; p->asc = 0x39; tfd = (p->sense_key << 12) | ATA_S_READY | ATA_S_ERROR; break; error: case 1: case 2: p->sense_key = ATA_SENSE_ILLEGAL_REQUEST; p->asc = 0x24; tfd = (p->sense_key << 12) | ATA_S_READY | ATA_S_ERROR; break; } cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; ahci_write_fis_d2h(p, slot, cfis, tfd); } static void atapi_get_event_status_notification(struct ahci_port *p, int slot, uint8_t *cfis) { uint8_t *acmd; uint32_t tfd; acmd = cfis + 0x40; /* we don't support asynchronous operation */ if (!(acmd[1] & 1)) { p->sense_key = ATA_SENSE_ILLEGAL_REQUEST; p->asc = 0x24; tfd = (p->sense_key << 12) | ATA_S_READY | ATA_S_ERROR; } else { uint8_t buf[8]; int len; len = be16dec(acmd + 7); if (len > sizeof(buf)) len = sizeof(buf); memset(buf, 0, sizeof(buf)); be16enc(buf, 8 - 2); buf[2] = 0x04; buf[3] = 0x10; buf[5] = 0x02; write_prdt(p, slot, cfis, buf, len); tfd = ATA_S_READY | ATA_S_DSC; } cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; ahci_write_fis_d2h(p, slot, cfis, tfd); } static void handle_packet_cmd(struct ahci_port *p, int slot, uint8_t *cfis) { uint8_t *acmd; acmd = cfis + 0x40; #ifdef AHCI_DEBUG { int i; DPRINTF("ACMD:"); for (i = 0; i < 16; i++) DPRINTF("%02x ", acmd[i]); DPRINTF("\n"); } #endif switch (acmd[0]) { case TEST_UNIT_READY: cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; ahci_write_fis_d2h(p, slot, cfis, ATA_S_READY | ATA_S_DSC); break; case INQUIRY: atapi_inquiry(p, slot, cfis); break; case READ_CAPACITY: atapi_read_capacity(p, slot, cfis); break; case PREVENT_ALLOW: /* TODO */ cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; ahci_write_fis_d2h(p, slot, cfis, ATA_S_READY | ATA_S_DSC); break; case READ_TOC: atapi_read_toc(p, slot, cfis); break; case READ_10: case READ_12: atapi_read(p, slot, cfis, 0, 0); break; case REQUEST_SENSE: atapi_request_sense(p, slot, cfis); break; case START_STOP_UNIT: atapi_start_stop_unit(p, slot, cfis); break; case MODE_SENSE_10: atapi_mode_sense(p, slot, cfis); break; case GET_EVENT_STATUS_NOTIFICATION: atapi_get_event_status_notification(p, slot, cfis); break; default: cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; p->sense_key = ATA_SENSE_ILLEGAL_REQUEST; p->asc = 0x20; ahci_write_fis_d2h(p, slot, cfis, (p->sense_key << 12) | ATA_S_READY | ATA_S_ERROR); break; } } static void ahci_handle_cmd(struct ahci_port *p, int slot, uint8_t *cfis) { switch (cfis[2]) { case ATA_ATA_IDENTIFY: handle_identify(p, slot, cfis); break; case ATA_SETFEATURES: { switch (cfis[3]) { case ATA_SF_ENAB_SATA_SF: switch (cfis[12]) { case ATA_SATA_SF_AN: p->tfd = ATA_S_DSC | ATA_S_READY; break; default: p->tfd = ATA_S_ERROR | ATA_S_READY; p->tfd |= (ATA_ERROR_ABORT << 8); break; } break; case ATA_SF_ENAB_WCACHE: case ATA_SF_DIS_WCACHE: case ATA_SF_ENAB_RCACHE: case ATA_SF_DIS_RCACHE: p->tfd = ATA_S_DSC | ATA_S_READY; break; case ATA_SF_SETXFER: { switch (cfis[12] & 0xf8) { case ATA_PIO: case ATA_PIO0: break; case ATA_WDMA0: case ATA_UDMA0: p->xfermode = (cfis[12] & 0x7); break; } p->tfd = ATA_S_DSC | ATA_S_READY; break; } default: p->tfd = ATA_S_ERROR | ATA_S_READY; p->tfd |= (ATA_ERROR_ABORT << 8); break; } ahci_write_fis_d2h(p, slot, cfis, p->tfd); break; } case ATA_SET_MULTI: if (cfis[12] != 0 && (cfis[12] > 128 || (cfis[12] & (cfis[12] - 1)))) { p->tfd = ATA_S_ERROR | ATA_S_READY; p->tfd |= (ATA_ERROR_ABORT << 8); } else { p->mult_sectors = cfis[12]; p->tfd = ATA_S_DSC | ATA_S_READY; } p->is |= AHCI_P_IX_DP; p->ci &= ~(1 << slot); ahci_generate_intr(p->pr_sc); break; case ATA_READ_DMA: case ATA_WRITE_DMA: case ATA_READ_DMA48: case ATA_WRITE_DMA48: case ATA_READ_FPDMA_QUEUED: case ATA_WRITE_FPDMA_QUEUED: ahci_handle_dma(p, slot, cfis, 0, 0); break; case ATA_FLUSHCACHE: case ATA_FLUSHCACHE48: ahci_handle_flush(p, slot, cfis); break; case ATA_STANDBY_CMD: break; case ATA_NOP: case ATA_STANDBY_IMMEDIATE: case ATA_IDLE_IMMEDIATE: case ATA_SLEEP: ahci_write_fis_d2h(p, slot, cfis, ATA_S_READY | ATA_S_DSC); break; case ATA_ATAPI_IDENTIFY: handle_atapi_identify(p, slot, cfis); break; case ATA_PACKET_CMD: if (!p->atapi) { p->tfd = (ATA_E_ABORT << 8) | ATA_S_READY | ATA_S_ERROR; p->is |= AHCI_P_IX_TFE; ahci_generate_intr(p->pr_sc); } else handle_packet_cmd(p, slot, cfis); break; default: WPRINTF("Unsupported cmd:%02x\n", cfis[2]); p->tfd = (ATA_E_ABORT << 8) | ATA_S_READY | ATA_S_ERROR; p->is |= AHCI_P_IX_TFE; ahci_generate_intr(p->pr_sc); break; } } static void ahci_handle_slot(struct ahci_port *p, int slot) { struct ahci_cmd_hdr *hdr; struct ahci_prdt_entry *prdt; struct pci_ahci_softc *sc; uint8_t *cfis; int cfl; sc = p->pr_sc; hdr = (struct ahci_cmd_hdr *)(p->cmd_lst + slot * AHCI_CL_SIZE); cfl = (hdr->flags & 0x1f) * 4; cfis = paddr_guest2host(ahci_ctx(sc), hdr->ctba, 0x80 + hdr->prdtl * sizeof(struct ahci_prdt_entry)); prdt = (struct ahci_prdt_entry *)(cfis + 0x80); #ifdef AHCI_DEBUG DPRINTF("\ncfis:"); for (i = 0; i < cfl; i++) { if (i % 10 == 0) DPRINTF("\n"); DPRINTF("%02x ", cfis[i]); } DPRINTF("\n"); for (i = 0; i < hdr->prdtl; i++) { DPRINTF("%d@%08"PRIx64"\n", prdt->dbc & 0x3fffff, prdt->dba); prdt++; } #endif if (cfis[0] != FIS_TYPE_REGH2D) { WPRINTF("Not a H2D FIS:%02x\n", cfis[0]); return; } if (cfis[1] & 0x80) { ahci_handle_cmd(p, slot, cfis); } else { if (cfis[15] & (1 << 2)) p->reset = 1; else if (p->reset) { p->reset = 0; ahci_port_reset(p); } p->ci &= ~(1 << slot); } } static void ahci_handle_port(struct ahci_port *p) { int i; if (!(p->cmd & AHCI_P_CMD_ST)) return; /* * Search for any new commands to issue ignoring those that * are already in-flight. */ for (i = 0; (i < 32) && p->ci; i++) { if ((p->ci & (1 << i)) && !(p->pending & (1 << i))) { p->cmd &= ~AHCI_P_CMD_CCS_MASK; p->cmd |= i << AHCI_P_CMD_CCS_SHIFT; ahci_handle_slot(p, i); } } } /* * blockif callback routine - this runs in the context of the blockif * i/o thread, so the mutex needs to be acquired. */ static void ata_ioreq_cb(struct blockif_req *br, int err) { struct ahci_cmd_hdr *hdr; struct ahci_ioreq *aior; struct ahci_port *p; struct pci_ahci_softc *sc; uint32_t tfd; uint8_t *cfis; int pending, slot, ncq; DPRINTF("%s %d\n", __func__, err); ncq = 0; aior = br->br_param; p = aior->io_pr; cfis = aior->cfis; slot = aior->slot; pending = aior->prdtl; sc = p->pr_sc; hdr = (struct ahci_cmd_hdr *)(p->cmd_lst + slot * AHCI_CL_SIZE); if (cfis[2] == ATA_WRITE_FPDMA_QUEUED || cfis[2] == ATA_READ_FPDMA_QUEUED) ncq = 1; pthread_mutex_lock(&sc->mtx); /* * Delete the blockif request from the busy list */ TAILQ_REMOVE(&p->iobhd, aior, io_blist); /* * Move the blockif request back to the free list */ STAILQ_INSERT_TAIL(&p->iofhd, aior, io_flist); if (pending && !err) { ahci_handle_dma(p, slot, cfis, aior->done, hdr->prdtl - pending); goto out; } if (!err && aior->done == aior->len) { tfd = ATA_S_READY | ATA_S_DSC; if (ncq) hdr->prdbc = 0; else hdr->prdbc = aior->len; } else { tfd = (ATA_E_ABORT << 8) | ATA_S_READY | ATA_S_ERROR; hdr->prdbc = 0; if (ncq) p->serr |= (1 << slot); } if (ncq) { p->sact &= ~(1 << slot); ahci_write_fis_sdb(p, slot, tfd); } else ahci_write_fis_d2h(p, slot, cfis, tfd); /* * This command is now complete. */ p->pending &= ~(1 << slot); ahci_check_stopped(p); out: pthread_mutex_unlock(&sc->mtx); DPRINTF("%s exit\n", __func__); } static void atapi_ioreq_cb(struct blockif_req *br, int err) { struct ahci_cmd_hdr *hdr; struct ahci_ioreq *aior; struct ahci_port *p; struct pci_ahci_softc *sc; uint8_t *cfis; uint32_t tfd; int pending, slot; DPRINTF("%s %d\n", __func__, err); aior = br->br_param; p = aior->io_pr; cfis = aior->cfis; slot = aior->slot; pending = aior->prdtl; sc = p->pr_sc; hdr = (struct ahci_cmd_hdr *)(p->cmd_lst + aior->slot * AHCI_CL_SIZE); pthread_mutex_lock(&sc->mtx); /* * Delete the blockif request from the busy list */ TAILQ_REMOVE(&p->iobhd, aior, io_blist); /* * Move the blockif request back to the free list */ STAILQ_INSERT_TAIL(&p->iofhd, aior, io_flist); if (pending && !err) { atapi_read(p, slot, cfis, aior->done, hdr->prdtl - pending); goto out; } if (!err && aior->done == aior->len) { tfd = ATA_S_READY | ATA_S_DSC; hdr->prdbc = aior->len; } else { p->sense_key = ATA_SENSE_ILLEGAL_REQUEST; p->asc = 0x21; tfd = (p->sense_key << 12) | ATA_S_READY | ATA_S_ERROR; hdr->prdbc = 0; } cfis[4] = (cfis[4] & ~7) | ATA_I_CMD | ATA_I_IN; ahci_write_fis_d2h(p, slot, cfis, tfd); /* * This command is now complete. */ p->pending &= ~(1 << slot); ahci_check_stopped(p); out: pthread_mutex_unlock(&sc->mtx); DPRINTF("%s exit\n", __func__); } static void pci_ahci_ioreq_init(struct ahci_port *pr) { struct ahci_ioreq *vr; int i; pr->ioqsz = blockif_queuesz(pr->bctx); pr->ioreq = calloc(pr->ioqsz, sizeof(struct ahci_ioreq)); STAILQ_INIT(&pr->iofhd); /* * Add all i/o request entries to the free queue */ for (i = 0; i < pr->ioqsz; i++) { vr = &pr->ioreq[i]; vr->io_pr = pr; if (!pr->atapi) vr->io_req.br_callback = ata_ioreq_cb; else vr->io_req.br_callback = atapi_ioreq_cb; vr->io_req.br_param = vr; STAILQ_INSERT_TAIL(&pr->iofhd, vr, io_flist); } TAILQ_INIT(&pr->iobhd); } static void pci_ahci_port_write(struct pci_ahci_softc *sc, uint64_t offset, uint64_t value) { int port = (offset - AHCI_OFFSET) / AHCI_STEP; offset = (offset - AHCI_OFFSET) % AHCI_STEP; struct ahci_port *p = &sc->port[port]; DPRINTF("pci_ahci_port %d: write offset 0x%"PRIx64" value 0x%"PRIx64"\n", port, offset, value); switch (offset) { case AHCI_P_CLB: p->clb = value; break; case AHCI_P_CLBU: p->clbu = value; break; case AHCI_P_FB: p->fb = value; break; case AHCI_P_FBU: p->fbu = value; break; case AHCI_P_IS: p->is &= ~value; break; case AHCI_P_IE: p->ie = value & 0xFDC000FF; ahci_generate_intr(sc); break; case AHCI_P_CMD: { p->cmd = value; if (!(value & AHCI_P_CMD_ST)) { ahci_port_stop(p); } else { uint64_t clb; p->cmd |= AHCI_P_CMD_CR; clb = (uint64_t)p->clbu << 32 | p->clb; p->cmd_lst = paddr_guest2host(ahci_ctx(sc), clb, AHCI_CL_SIZE * AHCI_MAX_SLOTS); } if (value & AHCI_P_CMD_FRE) { uint64_t fb; p->cmd |= AHCI_P_CMD_FR; fb = (uint64_t)p->fbu << 32 | p->fb; /* we don't support FBSCP, so rfis size is 256Bytes */ p->rfis = paddr_guest2host(ahci_ctx(sc), fb, 256); } else { p->cmd &= ~AHCI_P_CMD_FR; } if (value & AHCI_P_CMD_CLO) { p->tfd = 0; p->cmd &= ~AHCI_P_CMD_CLO; } ahci_handle_port(p); break; } case AHCI_P_TFD: case AHCI_P_SIG: case AHCI_P_SSTS: WPRINTF("pci_ahci_port: read only registers 0x%"PRIx64"\n", offset); break; case AHCI_P_SCTL: if (!(p->cmd & AHCI_P_CMD_ST)) { if (value & ATA_SC_DET_RESET) ahci_port_reset(p); p->sctl = value; } break; case AHCI_P_SERR: p->serr &= ~value; break; case AHCI_P_SACT: p->sact |= value; break; case AHCI_P_CI: p->ci |= value; ahci_handle_port(p); break; case AHCI_P_SNTF: case AHCI_P_FBS: default: break; } } static void pci_ahci_host_write(struct pci_ahci_softc *sc, uint64_t offset, uint64_t value) { DPRINTF("pci_ahci_host: write offset 0x%"PRIx64" value 0x%"PRIx64"\n", offset, value); switch (offset) { case AHCI_CAP: case AHCI_PI: case AHCI_VS: case AHCI_CAP2: DPRINTF("pci_ahci_host: read only registers 0x%"PRIx64"\n", offset); break; case AHCI_GHC: if (value & AHCI_GHC_HR) ahci_reset(sc); else if (value & AHCI_GHC_IE) { sc->ghc |= AHCI_GHC_IE; ahci_generate_intr(sc); } break; case AHCI_IS: sc->is &= ~value; ahci_generate_intr(sc); break; default: break; } } static void pci_ahci_write(struct vmctx *ctx, int vcpu, struct pci_devinst *pi, int baridx, uint64_t offset, int size, uint64_t value) { struct pci_ahci_softc *sc = pi->pi_arg; assert(baridx == 5); assert(size == 4); pthread_mutex_lock(&sc->mtx); if (offset < AHCI_OFFSET) pci_ahci_host_write(sc, offset, value); else if (offset < AHCI_OFFSET + sc->ports * AHCI_STEP) pci_ahci_port_write(sc, offset, value); else WPRINTF("pci_ahci: unknown i/o write offset 0x%"PRIx64"\n", offset); pthread_mutex_unlock(&sc->mtx); } static uint64_t pci_ahci_host_read(struct pci_ahci_softc *sc, uint64_t offset) { uint32_t value; switch (offset) { case AHCI_CAP: case AHCI_GHC: case AHCI_IS: case AHCI_PI: case AHCI_VS: case AHCI_CCCC: case AHCI_CCCP: case AHCI_EM_LOC: case AHCI_EM_CTL: case AHCI_CAP2: { uint32_t *p = &sc->cap; p += (offset - AHCI_CAP) / sizeof(uint32_t); value = *p; break; } default: value = 0; break; } DPRINTF("pci_ahci_host: read offset 0x%"PRIx64" value 0x%x\n", offset, value); return (value); } static uint64_t pci_ahci_port_read(struct pci_ahci_softc *sc, uint64_t offset) { uint32_t value; int port = (offset - AHCI_OFFSET) / AHCI_STEP; offset = (offset - AHCI_OFFSET) % AHCI_STEP; switch (offset) { case AHCI_P_CLB: case AHCI_P_CLBU: case AHCI_P_FB: case AHCI_P_FBU: case AHCI_P_IS: case AHCI_P_IE: case AHCI_P_CMD: case AHCI_P_TFD: case AHCI_P_SIG: case AHCI_P_SSTS: case AHCI_P_SCTL: case AHCI_P_SERR: case AHCI_P_SACT: case AHCI_P_CI: case AHCI_P_SNTF: case AHCI_P_FBS: { uint32_t *p= &sc->port[port].clb; p += (offset - AHCI_P_CLB) / sizeof(uint32_t); value = *p; break; } default: value = 0; break; } DPRINTF("pci_ahci_port %d: read offset 0x%"PRIx64" value 0x%x\n", port, offset, value); return value; } static uint64_t pci_ahci_read(struct vmctx *ctx, int vcpu, struct pci_devinst *pi, int baridx, uint64_t offset, int size) { struct pci_ahci_softc *sc = pi->pi_arg; uint32_t value; assert(baridx == 5); assert(size == 4); pthread_mutex_lock(&sc->mtx); if (offset < AHCI_OFFSET) value = pci_ahci_host_read(sc, offset); else if (offset < AHCI_OFFSET + sc->ports * AHCI_STEP) value = pci_ahci_port_read(sc, offset); else { value = 0; WPRINTF("pci_ahci: unknown i/o read offset 0x%"PRIx64"\n", offset); } pthread_mutex_unlock(&sc->mtx); return (value); } static int pci_ahci_init(struct vmctx *ctx, struct pci_devinst *pi, char *opts, int atapi) { char bident[sizeof("XX:X:X")]; struct blockif_ctxt *bctxt; struct pci_ahci_softc *sc; int ret, slots; ret = 0; if (opts == NULL) { fprintf(stderr, "pci_ahci: backing device required\n"); return (1); } #ifdef AHCI_DEBUG dbg = fopen("/tmp/log", "w+"); #endif sc = calloc(1, sizeof(struct pci_ahci_softc)); pi->pi_arg = sc; sc->asc_pi = pi; sc->ports = MAX_PORTS; /* * Only use port 0 for a backing device. All other ports will be * marked as unused */ sc->port[0].atapi = atapi; /* * Attempt to open the backing image. Use the PCI * slot/func for the identifier string. */ snprintf(bident, sizeof(bident), "%d:%d", pi->pi_slot, pi->pi_func); bctxt = blockif_open(opts, bident); if (bctxt == NULL) { ret = 1; goto open_fail; } sc->port[0].bctx = bctxt; sc->port[0].pr_sc = sc; /* * Allocate blockif request structures and add them * to the free list */ pci_ahci_ioreq_init(&sc->port[0]); pthread_mutex_init(&sc->mtx, NULL); /* Intel ICH8 AHCI */ slots = sc->port[0].ioqsz; if (slots > 32) slots = 32; --slots; sc->cap = AHCI_CAP_64BIT | AHCI_CAP_SNCQ | AHCI_CAP_SSNTF | AHCI_CAP_SMPS | AHCI_CAP_SSS | AHCI_CAP_SALP | AHCI_CAP_SAL | AHCI_CAP_SCLO | (0x3 << AHCI_CAP_ISS_SHIFT)| AHCI_CAP_PMD | AHCI_CAP_SSC | AHCI_CAP_PSC | (slots << AHCI_CAP_NCS_SHIFT) | AHCI_CAP_SXS | (sc->ports - 1); /* Only port 0 implemented */ sc->pi = 1; sc->vs = 0x10300; sc->cap2 = AHCI_CAP2_APST; ahci_reset(sc); pci_set_cfgdata16(pi, PCIR_DEVICE, 0x2821); pci_set_cfgdata16(pi, PCIR_VENDOR, 0x8086); pci_set_cfgdata8(pi, PCIR_CLASS, PCIC_STORAGE); pci_set_cfgdata8(pi, PCIR_SUBCLASS, PCIS_STORAGE_SATA); pci_set_cfgdata8(pi, PCIR_PROGIF, PCIP_STORAGE_SATA_AHCI_1_0); pci_emul_add_msicap(pi, 1); pci_emul_alloc_bar(pi, 5, PCIBAR_MEM32, AHCI_OFFSET + sc->ports * AHCI_STEP); pci_lintr_request(pi); open_fail: if (ret) { if (sc->port[0].bctx != NULL) blockif_close(sc->port[0].bctx); free(sc); } return (ret); } static int pci_ahci_hd_init(struct vmctx *ctx, struct pci_devinst *pi, char *opts) { return (pci_ahci_init(ctx, pi, opts, 0)); } static int pci_ahci_atapi_init(struct vmctx *ctx, struct pci_devinst *pi, char *opts) { return (pci_ahci_init(ctx, pi, opts, 1)); } /* * Use separate emulation names to distinguish drive and atapi devices */ struct pci_devemu pci_de_ahci_hd = { .pe_emu = "ahci-hd", .pe_init = pci_ahci_hd_init, .pe_barwrite = pci_ahci_write, .pe_barread = pci_ahci_read }; PCI_EMUL_SET(pci_de_ahci_hd); struct pci_devemu pci_de_ahci_cd = { .pe_emu = "ahci-cd", .pe_init = pci_ahci_atapi_init, .pe_barwrite = pci_ahci_write, .pe_barread = pci_ahci_read }; PCI_EMUL_SET(pci_de_ahci_cd); Index: projects/cxl_iscsi/usr.sbin/bhyve/pci_virtio_block.c =================================================================== --- projects/cxl_iscsi/usr.sbin/bhyve/pci_virtio_block.c (revision 279716) +++ projects/cxl_iscsi/usr.sbin/bhyve/pci_virtio_block.c (revision 279717) @@ -1,383 +1,411 @@ /*- * Copyright (c) 2011 NetApp, Inc. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY NETAPP, INC ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL NETAPP, INC OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "bhyverun.h" #include "pci_emul.h" #include "virtio.h" #define VTBLK_RINGSZ 64 #define VTBLK_MAXSEGS 32 #define VTBLK_S_OK 0 #define VTBLK_S_IOERR 1 #define VTBLK_S_UNSUPP 2 #define VTBLK_BLK_ID_BYTES 20 /* Capability bits */ #define VTBLK_F_SEG_MAX (1 << 2) /* Maximum request segments */ -#define VTBLK_F_BLK_SIZE (1 << 6) /* cfg block size valid */ +#define VTBLK_F_BLK_SIZE (1 << 6) /* cfg block size valid */ +#define VTBLK_F_TOPOLOGY (1 << 10) /* Optimal I/O alignment */ /* * Host capabilities */ #define VTBLK_S_HOSTCAPS \ ( VTBLK_F_SEG_MAX | \ VTBLK_F_BLK_SIZE | \ + VTBLK_F_TOPOLOGY | \ VIRTIO_RING_F_INDIRECT_DESC ) /* indirect descriptors */ /* * Config space "registers" */ struct vtblk_config { uint64_t vbc_capacity; uint32_t vbc_size_max; uint32_t vbc_seg_max; - uint16_t vbc_geom_c; - uint8_t vbc_geom_h; - uint8_t vbc_geom_s; + struct { + uint16_t cylinders; + uint8_t heads; + uint8_t sectors; + } vbc_geometry; uint32_t vbc_blk_size; - uint32_t vbc_sectors_max; + struct { + uint8_t physical_block_exp; + uint8_t alignment_offset; + uint16_t min_io_size; + uint32_t opt_io_size; + } vbc_topology; + uint8_t vbc_writeback; } __packed; /* * Fixed-size block header */ struct virtio_blk_hdr { #define VBH_OP_READ 0 #define VBH_OP_WRITE 1 #define VBH_OP_FLUSH 4 #define VBH_OP_FLUSH_OUT 5 #define VBH_OP_IDENT 8 #define VBH_FLAG_BARRIER 0x80000000 /* OR'ed into vbh_type */ uint32_t vbh_type; uint32_t vbh_ioprio; uint64_t vbh_sector; } __packed; /* * Debug printf */ static int pci_vtblk_debug; #define DPRINTF(params) if (pci_vtblk_debug) printf params #define WPRINTF(params) printf params /* * Per-device softc */ struct pci_vtblk_softc { struct virtio_softc vbsc_vs; pthread_mutex_t vsc_mtx; struct vqueue_info vbsc_vq; int vbsc_fd; + int vbsc_ischr; struct vtblk_config vbsc_cfg; char vbsc_ident[VTBLK_BLK_ID_BYTES]; }; static void pci_vtblk_reset(void *); static void pci_vtblk_notify(void *, struct vqueue_info *); static int pci_vtblk_cfgread(void *, int, int, uint32_t *); static int pci_vtblk_cfgwrite(void *, int, int, uint32_t); static struct virtio_consts vtblk_vi_consts = { "vtblk", /* our name */ 1, /* we support 1 virtqueue */ sizeof(struct vtblk_config), /* config reg size */ pci_vtblk_reset, /* reset */ pci_vtblk_notify, /* device-wide qnotify */ pci_vtblk_cfgread, /* read PCI config */ pci_vtblk_cfgwrite, /* write PCI config */ NULL, /* apply negotiated features */ VTBLK_S_HOSTCAPS, /* our capabilities */ }; static void pci_vtblk_reset(void *vsc) { struct pci_vtblk_softc *sc = vsc; DPRINTF(("vtblk: device reset requested !\n")); vi_reset_dev(&sc->vbsc_vs); } static void pci_vtblk_proc(struct pci_vtblk_softc *sc, struct vqueue_info *vq) { struct virtio_blk_hdr *vbh; uint8_t *status; int i, n; int err; int iolen; int writeop, type; off_t offset; struct iovec iov[VTBLK_MAXSEGS + 2]; uint16_t flags[VTBLK_MAXSEGS + 2]; n = vq_getchain(vq, iov, VTBLK_MAXSEGS + 2, flags); /* * The first descriptor will be the read-only fixed header, * and the last is for status (hence +2 above and below). * The remaining iov's are the actual data I/O vectors. * * XXX - note - this fails on crash dump, which does a * VIRTIO_BLK_T_FLUSH with a zero transfer length */ assert(n >= 2 && n <= VTBLK_MAXSEGS + 2); assert((flags[0] & VRING_DESC_F_WRITE) == 0); assert(iov[0].iov_len == sizeof(struct virtio_blk_hdr)); vbh = iov[0].iov_base; status = iov[--n].iov_base; assert(iov[n].iov_len == 1); assert(flags[n] & VRING_DESC_F_WRITE); /* * XXX * The guest should not be setting the BARRIER flag because * we don't advertise the capability. */ type = vbh->vbh_type & ~VBH_FLAG_BARRIER; writeop = (type == VBH_OP_WRITE); offset = vbh->vbh_sector * DEV_BSIZE; iolen = 0; for (i = 1; i < n; i++) { /* * - write op implies read-only descriptor, * - read/ident op implies write-only descriptor, * therefore test the inverse of the descriptor bit * to the op. */ assert(((flags[i] & VRING_DESC_F_WRITE) == 0) == writeop); iolen += iov[i].iov_len; } DPRINTF(("virtio-block: %s op, %d bytes, %d segs, offset %ld\n\r", writeop ? "write" : "read/ident", iolen, i - 1, offset)); + err = 0; switch (type) { case VBH_OP_WRITE: - err = pwritev(sc->vbsc_fd, iov + 1, i - 1, offset); + if (pwritev(sc->vbsc_fd, iov + 1, i - 1, offset) < 0) + err = errno; break; case VBH_OP_READ: - err = preadv(sc->vbsc_fd, iov + 1, i - 1, offset); + if (preadv(sc->vbsc_fd, iov + 1, i - 1, offset) < 0) + err = errno; break; case VBH_OP_IDENT: /* Assume a single buffer */ strlcpy(iov[1].iov_base, sc->vbsc_ident, MIN(iov[1].iov_len, sizeof(sc->vbsc_ident))); err = 0; break; case VBH_OP_FLUSH: case VBH_OP_FLUSH_OUT: - err = fsync(sc->vbsc_fd); + if (sc->vbsc_ischr) { + if (ioctl(sc->vbsc_fd, DIOCGFLUSH)) + err = errno; + } else if (fsync(sc->vbsc_fd)) + err = errno; break; default: err = -ENOSYS; break; } /* convert errno into a virtio block error return */ - if (err < 0) { - if (err == -ENOSYS) - *status = VTBLK_S_UNSUPP; - else - *status = VTBLK_S_IOERR; - } else + if (err == -ENOSYS) + *status = VTBLK_S_UNSUPP; + else if (err != 0) + *status = VTBLK_S_IOERR; + else *status = VTBLK_S_OK; /* * Return the descriptor back to the host. * We wrote 1 byte (our status) to host. */ vq_relchain(vq, 1); } static void pci_vtblk_notify(void *vsc, struct vqueue_info *vq) { struct pci_vtblk_softc *sc = vsc; vq_startchains(vq); while (vq_has_descs(vq)) pci_vtblk_proc(sc, vq); vq_endchains(vq, 1); /* Generate interrupt if appropriate. */ } static int pci_vtblk_init(struct vmctx *ctx, struct pci_devinst *pi, char *opts) { struct stat sbuf; MD5_CTX mdctx; u_char digest[16]; struct pci_vtblk_softc *sc; - off_t size; + off_t size, sts, sto; int fd; int sectsz; if (opts == NULL) { printf("virtio-block: backing device required\n"); return (1); } /* * The supplied backing file has to exist */ fd = open(opts, O_RDWR); if (fd < 0) { perror("Could not open backing file"); return (1); } if (fstat(fd, &sbuf) < 0) { perror("Could not stat backing file"); close(fd); return (1); } /* * Deal with raw devices */ size = sbuf.st_size; sectsz = DEV_BSIZE; + sts = sto = 0; if (S_ISCHR(sbuf.st_mode)) { if (ioctl(fd, DIOCGMEDIASIZE, &size) < 0 || ioctl(fd, DIOCGSECTORSIZE, §sz)) { perror("Could not fetch dev blk/sector size"); close(fd); return (1); } assert(size != 0); assert(sectsz != 0); - } + if (ioctl(fd, DIOCGSTRIPESIZE, &sts) == 0 && sts > 0) + ioctl(fd, DIOCGSTRIPEOFFSET, &sto); + } else + sts = sbuf.st_blksize; sc = calloc(1, sizeof(struct pci_vtblk_softc)); /* record fd of storage device/file */ sc->vbsc_fd = fd; + sc->vbsc_ischr = S_ISCHR(sbuf.st_mode); pthread_mutex_init(&sc->vsc_mtx, NULL); /* init virtio softc and virtqueues */ vi_softc_linkup(&sc->vbsc_vs, &vtblk_vi_consts, sc, pi, &sc->vbsc_vq); sc->vbsc_vs.vs_mtx = &sc->vsc_mtx; sc->vbsc_vq.vq_qsize = VTBLK_RINGSZ; /* sc->vbsc_vq.vq_notify = we have no per-queue notify */ /* * Create an identifier for the backing file. Use parts of the * md5 sum of the filename */ MD5Init(&mdctx); MD5Update(&mdctx, opts, strlen(opts)); MD5Final(digest, &mdctx); sprintf(sc->vbsc_ident, "BHYVE-%02X%02X-%02X%02X-%02X%02X", digest[0], digest[1], digest[2], digest[3], digest[4], digest[5]); /* setup virtio block config space */ sc->vbsc_cfg.vbc_capacity = size / DEV_BSIZE; /* 512-byte units */ + sc->vbsc_cfg.vbc_size_max = 0; /* not negotiated */ sc->vbsc_cfg.vbc_seg_max = VTBLK_MAXSEGS; + sc->vbsc_cfg.vbc_geometry.cylinders = 0; /* no geometry */ + sc->vbsc_cfg.vbc_geometry.heads = 0; + sc->vbsc_cfg.vbc_geometry.sectors = 0; sc->vbsc_cfg.vbc_blk_size = sectsz; - sc->vbsc_cfg.vbc_size_max = 0; /* not negotiated */ - sc->vbsc_cfg.vbc_geom_c = 0; /* no geometry */ - sc->vbsc_cfg.vbc_geom_h = 0; - sc->vbsc_cfg.vbc_geom_s = 0; - sc->vbsc_cfg.vbc_sectors_max = 0; + sc->vbsc_cfg.vbc_topology.physical_block_exp = + (sts > sectsz) ? (ffsll(sts / sectsz) - 1) : 0; + sc->vbsc_cfg.vbc_topology.alignment_offset = + (sto != 0) ? ((sts - sto) / sectsz) : 0; + sc->vbsc_cfg.vbc_topology.min_io_size = 0; + sc->vbsc_cfg.vbc_topology.opt_io_size = 0; + sc->vbsc_cfg.vbc_writeback = 0; /* * Should we move some of this into virtio.c? Could * have the device, class, and subdev_0 as fields in * the virtio constants structure. */ pci_set_cfgdata16(pi, PCIR_DEVICE, VIRTIO_DEV_BLOCK); pci_set_cfgdata16(pi, PCIR_VENDOR, VIRTIO_VENDOR); pci_set_cfgdata8(pi, PCIR_CLASS, PCIC_STORAGE); pci_set_cfgdata16(pi, PCIR_SUBDEV_0, VIRTIO_TYPE_BLOCK); pci_lintr_request(pi); if (vi_intr_init(&sc->vbsc_vs, 1, fbsdrun_virtio_msix())) return (1); vi_set_io_bar(&sc->vbsc_vs, 0); return (0); } static int pci_vtblk_cfgwrite(void *vsc, int offset, int size, uint32_t value) { DPRINTF(("vtblk: write to readonly reg %d\n\r", offset)); return (1); } static int pci_vtblk_cfgread(void *vsc, int offset, int size, uint32_t *retval) { struct pci_vtblk_softc *sc = vsc; void *ptr; /* our caller has already verified offset and size */ ptr = (uint8_t *)&sc->vbsc_cfg + offset; memcpy(retval, ptr, size); return (0); } struct pci_devemu pci_de_vblk = { .pe_emu = "virtio-blk", .pe_init = pci_vtblk_init, .pe_barwrite = vi_pci_write, .pe_barread = vi_pci_read }; PCI_EMUL_SET(pci_de_vblk); Index: projects/cxl_iscsi/usr.sbin/bhyve =================================================================== --- projects/cxl_iscsi/usr.sbin/bhyve (revision 279716) +++ projects/cxl_iscsi/usr.sbin/bhyve (revision 279717) Property changes on: projects/cxl_iscsi/usr.sbin/bhyve ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/usr.sbin/bhyve:r279515-279716 Index: projects/cxl_iscsi/usr.sbin/ctld/ctl.conf.5 =================================================================== --- projects/cxl_iscsi/usr.sbin/ctld/ctl.conf.5 (revision 279716) +++ projects/cxl_iscsi/usr.sbin/ctld/ctl.conf.5 (revision 279717) @@ -1,444 +1,446 @@ .\" Copyright (c) 2012 The FreeBSD Foundation .\" All rights reserved. .\" .\" This software was developed by Edward Tomasz Napierala under sponsorship .\" from the FreeBSD Foundation. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" $FreeBSD$ .\" -.Dd February 27, 2015 +.Dd March 4, 2015 .Dt CTL.CONF 5 .Os .Sh NAME .Nm ctl.conf .Nd CAM Target Layer / iSCSI target daemon configuration file .Sh DESCRIPTION The .Nm configuration file is used by the .Xr ctld 8 daemon. Lines starting with .Ql # are interpreted as comments. The general syntax of the .Nm file is: .Bd -literal -offset indent .No pidfile Ar path .No auth-group Ar name No { .Dl chap Ar user Ar secret .Dl ... } .No portal-group Ar name No { .Dl listen Ar address .\".Dl listen-iser Ar address .Dl discovery-auth-group Ar name .Dl ... } .No lun Ar name No { .Dl path Ar path } .No target Ar name { .Dl auth-group Ar name .Dl portal-group Ar name Op Ar agname .Dl port Ar name .Dl lun Ar number Ar name .Dl lun Ar number No { .Dl path Ar path .Dl } .Dl ... } .Ed .Ss Global Context .Bl -tag -width indent .It Ic auth-group Ar name Create an .Sy auth-group configuration context, defining a new auth-group, which can then be assigned to any number of targets. .It Ic debug Ar level The debug verbosity level. The default is 0. .It Ic maxproc Ar number The limit for concurrently running child processes handling incoming connections. The default is 30. A setting of 0 disables the limit. .It Ic pidfile Ar path The path to the pidfile. The default is .Pa /var/run/ctld.pid . .It Ic portal-group Ar name Create a .Sy portal-group configuration context, defining a new portal-group, which can then be assigned to any number of targets. .It Ic lun Ar name Create a .Sy lun configuration context, defining a LUN to be exported by some target(s). .It Ic target Ar name Create a .Sy target configuration context, which can contain one or more .Sy lun contexts. .It Ic timeout Ar seconds The timeout for login sessions, after which the connection will be forcibly terminated. The default is 60. A setting of 0 disables the timeout. .It Ic isns-server Ar address An IPv4 or IPv6 address and optionally port of iSNS server to register on. .It Ic isns-period Ar seconds iSNS registration period. Registered Network Entity not updated during this period will be unregistered. The default is 900. .It Ic isns-timeout Ar seconds Timeout for iSNS requests. The default is 5. .El .Ss auth-group Context .Bl -tag -width indent .It Ic auth-type Ar type Sets the authentication type. Type can be either .Qq Ar none , .Qq Ar deny , .Qq Ar chap , or .Qq Ar chap-mutual . In most cases it is not necessary to set the type using this clause; it is usually used to disable authentication for a given .Sy auth-group . .It Ic chap Ar user Ar secret A set of CHAP authentication credentials. Note that for any .Sy auth-group , the configuration may only contain either .Sy chap or .Sy chap-mutual entries; it is an error to mix them. .It Ic chap-mutual Ar user Ar secret Ar mutualuser Ar mutualsecret A set of mutual CHAP authentication credentials. Note that for any .Sy auth-group , the configuration may only contain either .Sy chap or .Sy chap-mutual entries; it is an error to mix them. .It Ic initiator-name Ar initiator-name An iSCSI initiator name. Only initiators with a name matching one of the defined names will be allowed to connect. If not defined, there will be no restrictions based on initiator name. .It Ic initiator-portal Ar address Ns Op / Ns Ar prefixlen An iSCSI initiator portal: an IPv4 or IPv6 address, optionally followed by a literal slash and a prefix length. Only initiators with an address matching one of the defined addresses will be allowed to connect. If not defined, there will be no restrictions based on initiator address. .El .Ss portal-group Context .Bl -tag -width indent .It Ic discovery-auth-group Ar name Assign a previously defined authentication group to the portal group, to be used for target discovery. By default, portal groups are assigned predefined .Sy auth-group .Qq Ar default , which denies discovery. Another predefined .Sy auth-group , .Qq Ar no-authentication , may be used to permit discovery without authentication. .It Ic discovery-filter Ar filter Determines which targets are returned during discovery. Filter can be either .Qq Ar none , .Qq Ar portal , .Qq Ar portal-name , or .Qq Ar portal-name-auth . When set to .Qq Ar none , discovery will return all targets assigned to that portal group. When set to .Qq Ar portal , discovery will not return targets that cannot be accessed by the initiator because of their .Sy initiator-portal . When set to .Qq Ar portal-name , the check will include both .Sy initiator-portal and .Sy initiator-name . When set to .Qq Ar portal-name-auth , the check will include .Sy initiator-portal , .Sy initiator-name , and authentication credentials. The target is returned if it does not require CHAP authentication, or if the CHAP user and secret used during discovery match those used by the target. Note that when using .Qq Ar portal-name-auth , targets that require CHAP authentication will only be returned if .Sy discovery-auth-group requires CHAP. The default is .Qq Ar none . .It Ic listen Ar address An IPv4 or IPv6 address and port to listen on for incoming connections. .\".It Ic listen-iser Ar address .\"An IPv4 or IPv6 address and port to listen on for incoming connections .\"using iSER (iSCSI over RDMA) protocol. .It Ic offload Ar driver Define iSCSI hardware offload driver to use for this .Sy portal-group . .It Ic redirect Ar address IPv4 or IPv6 address to redirect initiators to. When configured, all initiators attempting to connect to portal belonging to this .Sy portal-group will get redirected using "Target moved temporarily" login response. Redirection happens before authentication and any .Sy initiator-name or .Sy initiator-portal checks are skipped. .El .Ss target Context .Bl -tag -width indent .It Ic alias Ar text Assign a human-readable description to the target. There is no default. .It Ic auth-group Ar name Assign a previously defined authentication group to the target. By default, targets that do not specify their own auth settings, using clauses such as .Sy chap or .Sy initiator-name , are assigned predefined .Sy auth-group .Qq Ar default , which denies all access. Another predefined .Sy auth-group , .Qq Ar no-authentication , may be used to permit access without authentication. Note that targets must only use one of .Sy auth-group , chap , No or Sy chap-mutual ; it is a configuration error to mix multiple types in one target. .It Ic auth-type Ar type Sets the authentication type. Type can be either .Qq Ar none , .Qq Ar deny , .Qq Ar chap , or .Qq Ar chap-mutual . In most cases it is not necessary to set the type using this clause; it is usually used to disable authentication for a given .Sy target . This clause is mutually exclusive with .Sy auth-group ; one cannot use both in a single target. .It Ic chap Ar user Ar secret A set of CHAP authentication credentials. Note that targets must only use one of .Sy auth-group , chap , No or Sy chap-mutual ; it is a configuration error to mix multiple types in one target. .It Ic chap-mutual Ar user Ar secret Ar mutualuser Ar mutualsecret A set of mutual CHAP authentication credentials. Note that targets must only use one of .Sy auth-group , chap , No or Sy chap-mutual ; it is a configuration error to mix multiple types in one target. .It Ic initiator-name Ar initiator-name An iSCSI initiator name. Only initiators with a name matching one of the defined names will be allowed to connect. If not defined, there will be no restrictions based on initiator name. This clause is mutually exclusive with .Sy auth-group ; one cannot use both in a single target. .It Ic initiator-portal Ar address Ns Op / Ns Ar prefixlen An iSCSI initiator portal: an IPv4 or IPv6 address, optionally followed by a literal slash and a prefix length. Only initiators with an address matching one of the defined addresses will be allowed to connect. If not defined, there will be no restrictions based on initiator address. This clause is mutually exclusive with .Sy auth-group ; one cannot use both in a single target. .It Ic portal-group Ar name Op Ar agname Assign a previously defined portal group to the target. The default portal group is .Qq Ar default , which makes the target available on TCP port 3260 on all configured IPv4 and IPv6 addresses. Optional second argument specifies auth group name for connections to this specific portal group. If second argument is not specified, target auth group is used. .It Ic port Ar name Assign specified CTL port (such as "isp0") to the target. On startup ctld configures LUN mapping and enables all assigned ports. Each port can be assigned to only one target. .It Ic redirect Ar address IPv4 or IPv6 address to redirect initiators to. When configured, all initiators attempting to connect to this target will get redirected using "Target moved temporarily" login response. Redirection happens after successful authentication. .It Ic lun Ar number Ar name Export previously defined .Sy lun by the parent target. .It Ic lun Ar number Create a .Sy lun configuration context, defining a LUN exported by the parent target. .El .Ss lun Context .Bl -tag -width indent .It Ic backend Ar block No | Ar ramdisk The CTL backend to use for a given LUN. Valid choices are .Qq Ar block and .Qq Ar ramdisk ; block is used for LUNs backed by files or disk device nodes; ramdisk is a bitsink device, used mostly for testing. The default backend is block. .It Ic blocksize Ar size The blocksize visible to the initiator. The default blocksize is 512. .It Ic device-id Ar string The SCSI Device Identification string presented to the initiator. .It Ic option Ar name Ar value The CTL-specific options passed to the kernel. All CTL-specific options are documented in the .Sx OPTIONS section of .Xr ctladm 8 . .It Ic path Ar path The path to the file or device node used to back the LUN. .It Ic serial Ar string The SCSI serial number presented to the initiator. .It Ic size Ar size The LUN size, in bytes. .El .Sh FILES .Bl -tag -width ".Pa /etc/ctl.conf" -compact .It Pa /etc/ctl.conf The default location of the .Xr ctld 8 configuration file. .El .Sh EXAMPLES .Bd -literal auth-group ag0 { chap-mutual "user" "secret" "mutualuser" "mutualsecret" chap-mutual "user2" "secret2" "mutualuser" "mutualsecret" initiator-portal 192.168.1.1/16 } auth-group ag1 { auth-type none initiator-name "iqn.2012-06.com.example:initiatorhost1" initiator-name "iqn.2012-06.com.example:initiatorhost2" initiator-portal 192.168.1.1/24 initiator-portal [2001:db8::de:ef] } portal-group pg0 { discovery-auth-group no-authentication listen 0.0.0.0:3260 listen [::]:3260 listen [fe80::be:ef]:3261 } target iqn.2012-06.com.example:target0 { alias "Example target" auth-group no-authentication lun 0 { path /dev/zvol/tank/example_0 blocksize 4096 size 4G } } lun example_1 { path /dev/zvol/tank/example_1 + option naa 0x50015178f369f093 } target iqn.2012-06.com.example:target1 { - chap chapuser chapsecret - lun 0 example_1 -} - -target iqn.2012-06.com.example:target2 { auth-group ag0 portal-group pg0 lun 0 example_1 lun 1 { path /dev/zvol/tank/example_2 option foo bar } +} + +target naa.50015178f369f092 { + port isp0 + port isp1 + lun 0 example_1 } .Ed .Sh SEE ALSO .Xr ctl 4 , .Xr ctladm 8 , .Xr ctld 8 .Sh AUTHORS The .Nm configuration file functionality for .Xr ctld 8 was developed by .An Edward Tomasz Napierala Aq Mt trasz@FreeBSD.org under sponsorship from the FreeBSD Foundation. Index: projects/cxl_iscsi/usr.sbin/ctld/discovery.c =================================================================== --- projects/cxl_iscsi/usr.sbin/ctld/discovery.c (revision 279716) +++ projects/cxl_iscsi/usr.sbin/ctld/discovery.c (revision 279717) @@ -1,333 +1,337 @@ /*- * Copyright (c) 2012 The FreeBSD Foundation * All rights reserved. * * This software was developed by Edward Tomasz Napierala under sponsorship * from the FreeBSD Foundation. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include "ctld.h" #include "iscsi_proto.h" static struct pdu * text_receive(struct connection *conn) { struct pdu *request; struct iscsi_bhs_text_request *bhstr; request = pdu_new(conn); pdu_receive(request); if ((request->pdu_bhs->bhs_opcode & ~ISCSI_BHS_OPCODE_IMMEDIATE) != ISCSI_BHS_OPCODE_TEXT_REQUEST) log_errx(1, "protocol error: received invalid opcode 0x%x", request->pdu_bhs->bhs_opcode); bhstr = (struct iscsi_bhs_text_request *)request->pdu_bhs; #if 0 if ((bhstr->bhstr_flags & ISCSI_BHSTR_FLAGS_FINAL) == 0) log_errx(1, "received Text PDU without the \"F\" flag"); #endif /* * XXX: Implement the C flag some day. */ if ((bhstr->bhstr_flags & BHSTR_FLAGS_CONTINUE) != 0) log_errx(1, "received Text PDU with unsupported \"C\" flag"); if (ISCSI_SNLT(ntohl(bhstr->bhstr_cmdsn), conn->conn_cmdsn)) { log_errx(1, "received Text PDU with decreasing CmdSN: " "was %u, is %u", conn->conn_cmdsn, ntohl(bhstr->bhstr_cmdsn)); } if (ntohl(bhstr->bhstr_expstatsn) != conn->conn_statsn) { log_errx(1, "received Text PDU with wrong StatSN: " "is %u, should be %u", ntohl(bhstr->bhstr_expstatsn), conn->conn_statsn); } conn->conn_cmdsn = ntohl(bhstr->bhstr_cmdsn); + if ((bhstr->bhstr_opcode & ISCSI_BHS_OPCODE_IMMEDIATE) == 0) + conn->conn_cmdsn++; return (request); } static struct pdu * text_new_response(struct pdu *request) { struct pdu *response; struct connection *conn; struct iscsi_bhs_text_request *bhstr; struct iscsi_bhs_text_response *bhstr2; bhstr = (struct iscsi_bhs_text_request *)request->pdu_bhs; conn = request->pdu_connection; response = pdu_new_response(request); bhstr2 = (struct iscsi_bhs_text_response *)response->pdu_bhs; bhstr2->bhstr_opcode = ISCSI_BHS_OPCODE_TEXT_RESPONSE; bhstr2->bhstr_flags = BHSTR_FLAGS_FINAL; bhstr2->bhstr_lun = bhstr->bhstr_lun; bhstr2->bhstr_initiator_task_tag = bhstr->bhstr_initiator_task_tag; bhstr2->bhstr_target_transfer_tag = bhstr->bhstr_target_transfer_tag; bhstr2->bhstr_statsn = htonl(conn->conn_statsn++); bhstr2->bhstr_expcmdsn = htonl(conn->conn_cmdsn); bhstr2->bhstr_maxcmdsn = htonl(conn->conn_cmdsn); return (response); } static struct pdu * logout_receive(struct connection *conn) { struct pdu *request; struct iscsi_bhs_logout_request *bhslr; request = pdu_new(conn); pdu_receive(request); if ((request->pdu_bhs->bhs_opcode & ~ISCSI_BHS_OPCODE_IMMEDIATE) != ISCSI_BHS_OPCODE_LOGOUT_REQUEST) log_errx(1, "protocol error: received invalid opcode 0x%x", request->pdu_bhs->bhs_opcode); bhslr = (struct iscsi_bhs_logout_request *)request->pdu_bhs; if ((bhslr->bhslr_reason & 0x7f) != BHSLR_REASON_CLOSE_SESSION) log_debugx("received Logout PDU with invalid reason 0x%x; " "continuing anyway", bhslr->bhslr_reason & 0x7f); if (ISCSI_SNLT(ntohl(bhslr->bhslr_cmdsn), conn->conn_cmdsn)) { log_errx(1, "received Logout PDU with decreasing CmdSN: " "was %u, is %u", conn->conn_cmdsn, ntohl(bhslr->bhslr_cmdsn)); } if (ntohl(bhslr->bhslr_expstatsn) != conn->conn_statsn) { log_errx(1, "received Logout PDU with wrong StatSN: " "is %u, should be %u", ntohl(bhslr->bhslr_expstatsn), conn->conn_statsn); } conn->conn_cmdsn = ntohl(bhslr->bhslr_cmdsn); + if ((bhslr->bhslr_opcode & ISCSI_BHS_OPCODE_IMMEDIATE) == 0) + conn->conn_cmdsn++; return (request); } static struct pdu * logout_new_response(struct pdu *request) { struct pdu *response; struct connection *conn; struct iscsi_bhs_logout_request *bhslr; struct iscsi_bhs_logout_response *bhslr2; bhslr = (struct iscsi_bhs_logout_request *)request->pdu_bhs; conn = request->pdu_connection; response = pdu_new_response(request); bhslr2 = (struct iscsi_bhs_logout_response *)response->pdu_bhs; bhslr2->bhslr_opcode = ISCSI_BHS_OPCODE_LOGOUT_RESPONSE; bhslr2->bhslr_flags = 0x80; bhslr2->bhslr_response = BHSLR_RESPONSE_CLOSED_SUCCESSFULLY; bhslr2->bhslr_initiator_task_tag = bhslr->bhslr_initiator_task_tag; bhslr2->bhslr_statsn = htonl(conn->conn_statsn++); bhslr2->bhslr_expcmdsn = htonl(conn->conn_cmdsn); bhslr2->bhslr_maxcmdsn = htonl(conn->conn_cmdsn); return (response); } static void discovery_add_target(struct keys *response_keys, const struct target *targ) { struct port *port; struct portal *portal; char *buf; char hbuf[NI_MAXHOST], sbuf[NI_MAXSERV]; struct addrinfo *ai; int ret; keys_add(response_keys, "TargetName", targ->t_name); TAILQ_FOREACH(port, &targ->t_ports, p_ts) { if (port->p_portal_group == NULL) continue; TAILQ_FOREACH(portal, &port->p_portal_group->pg_portals, p_next) { ai = portal->p_ai; ret = getnameinfo(ai->ai_addr, ai->ai_addrlen, hbuf, sizeof(hbuf), sbuf, sizeof(sbuf), NI_NUMERICHOST | NI_NUMERICSERV); if (ret != 0) { log_warnx("getnameinfo: %s", gai_strerror(ret)); continue; } switch (ai->ai_addr->sa_family) { case AF_INET: if (strcmp(hbuf, "0.0.0.0") == 0) continue; ret = asprintf(&buf, "%s:%s,%d", hbuf, sbuf, port->p_portal_group->pg_tag); break; case AF_INET6: if (strcmp(hbuf, "::") == 0) continue; ret = asprintf(&buf, "[%s]:%s,%d", hbuf, sbuf, port->p_portal_group->pg_tag); break; default: continue; } if (ret <= 0) log_err(1, "asprintf"); keys_add(response_keys, "TargetAddress", buf); free(buf); } } } static bool discovery_target_filtered_out(const struct connection *conn, const struct port *port) { const struct auth_group *ag; const struct portal_group *pg; const struct target *targ; const struct auth *auth; int error; targ = port->p_target; ag = port->p_auth_group; if (ag == NULL) ag = targ->t_auth_group; pg = conn->conn_portal->p_portal_group; assert(pg->pg_discovery_auth_group != PG_FILTER_UNKNOWN); if (pg->pg_discovery_filter >= PG_FILTER_PORTAL && auth_portal_check(ag, &conn->conn_initiator_sa) != 0) { log_debugx("initiator does not match initiator portals " "allowed for target \"%s\"; skipping", targ->t_name); return (true); } if (pg->pg_discovery_filter >= PG_FILTER_PORTAL_NAME && auth_name_check(ag, conn->conn_initiator_name) != 0) { log_debugx("initiator does not match initiator names " "allowed for target \"%s\"; skipping", targ->t_name); return (true); } if (pg->pg_discovery_filter >= PG_FILTER_PORTAL_NAME_AUTH && ag->ag_type != AG_TYPE_NO_AUTHENTICATION) { if (conn->conn_chap == NULL) { assert(pg->pg_discovery_auth_group->ag_type == AG_TYPE_NO_AUTHENTICATION); log_debugx("initiator didn't authenticate, but target " "\"%s\" requires CHAP; skipping", targ->t_name); return (true); } assert(conn->conn_user != NULL); auth = auth_find(ag, conn->conn_user); if (auth == NULL) { log_debugx("CHAP user \"%s\" doesn't match target " "\"%s\"; skipping", conn->conn_user, targ->t_name); return (true); } error = chap_authenticate(conn->conn_chap, auth->a_secret); if (error != 0) { log_debugx("password for CHAP user \"%s\" doesn't " "match target \"%s\"; skipping", conn->conn_user, targ->t_name); return (true); } } return (false); } void discovery(struct connection *conn) { struct pdu *request, *response; struct keys *request_keys, *response_keys; const struct port *port; const struct portal_group *pg; const char *send_targets; pg = conn->conn_portal->p_portal_group; log_debugx("beginning discovery session; waiting for Text PDU"); request = text_receive(conn); request_keys = keys_new(); keys_load(request_keys, request); send_targets = keys_find(request_keys, "SendTargets"); if (send_targets == NULL) log_errx(1, "received Text PDU without SendTargets"); response = text_new_response(request); response_keys = keys_new(); if (strcmp(send_targets, "All") == 0) { TAILQ_FOREACH(port, &pg->pg_ports, p_pgs) { if (discovery_target_filtered_out(conn, port)) { /* Ignore this target. */ continue; } discovery_add_target(response_keys, port->p_target); } } else { port = port_find_in_pg(pg, send_targets); if (port == NULL) { log_debugx("initiator requested information on unknown " "target \"%s\"; returning nothing", send_targets); } else { if (discovery_target_filtered_out(conn, port)) { /* Ignore this target. */ } else { discovery_add_target(response_keys, port->p_target); } } } keys_save(response_keys, response); pdu_send(response); pdu_delete(response); keys_delete(response_keys); pdu_delete(request); keys_delete(request_keys); log_debugx("done sending targets; waiting for Logout PDU"); request = logout_receive(conn); response = logout_new_response(request); pdu_send(response); pdu_delete(response); pdu_delete(request); log_debugx("discovery session done"); } Index: projects/cxl_iscsi/usr.sbin/ctld/kernel.c =================================================================== --- projects/cxl_iscsi/usr.sbin/ctld/kernel.c (revision 279716) +++ projects/cxl_iscsi/usr.sbin/ctld/kernel.c (revision 279717) @@ -1,1191 +1,1203 @@ /*- * Copyright (c) 2003, 2004 Silicon Graphics International Corp. * Copyright (c) 1997-2007 Kenneth D. Merry * Copyright (c) 2012 The FreeBSD Foundation * All rights reserved. * * Portions of this software were developed by Edward Tomasz Napierala * under sponsorship from the FreeBSD Foundation. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions, and the following disclaimer, * without modification. * 2. Redistributions in binary form must reproduce at minimum a disclaimer * substantially similar to the "NO WARRANTY" disclaimer below * ("Disclaimer") and any redistribution must be conditioned upon * including a substantially similar Disclaimer requirement for further * binary redistribution. * * NO WARRANTY * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT * HOLDERS OR CONTRIBUTORS BE LIABLE FOR SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGES. * */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "ctld.h" #ifdef ICL_KERNEL_PROXY #include #endif extern bool proxy_mode; static int ctl_fd = 0; void kernel_init(void) { int retval, saved_errno; ctl_fd = open(CTL_DEFAULT_DEV, O_RDWR); if (ctl_fd < 0 && errno == ENOENT) { saved_errno = errno; retval = kldload("ctl"); if (retval != -1) ctl_fd = open(CTL_DEFAULT_DEV, O_RDWR); else errno = saved_errno; } if (ctl_fd < 0) log_err(1, "failed to open %s", CTL_DEFAULT_DEV); } /* * Name/value pair used for per-LUN attributes. */ struct cctl_lun_nv { char *name; char *value; STAILQ_ENTRY(cctl_lun_nv) links; }; /* * Backend LUN information. */ struct cctl_lun { uint64_t lun_id; char *backend_type; uint64_t size_blocks; uint32_t blocksize; char *serial_number; char *device_id; char *ctld_name; STAILQ_HEAD(,cctl_lun_nv) attr_list; STAILQ_ENTRY(cctl_lun) links; }; struct cctl_port { uint32_t port_id; char *port_name; int cfiscsi_state; char *cfiscsi_target; uint16_t cfiscsi_portal_group_tag; char *ctld_portal_group_name; STAILQ_HEAD(,cctl_lun_nv) attr_list; STAILQ_ENTRY(cctl_port) links; }; struct cctl_devlist_data { int num_luns; STAILQ_HEAD(,cctl_lun) lun_list; struct cctl_lun *cur_lun; int num_ports; STAILQ_HEAD(,cctl_port) port_list; struct cctl_port *cur_port; int level; struct sbuf *cur_sb[32]; }; static void cctl_start_element(void *user_data, const char *name, const char **attr) { int i; struct cctl_devlist_data *devlist; struct cctl_lun *cur_lun; devlist = (struct cctl_devlist_data *)user_data; cur_lun = devlist->cur_lun; devlist->level++; if ((u_int)devlist->level >= (sizeof(devlist->cur_sb) / sizeof(devlist->cur_sb[0]))) log_errx(1, "%s: too many nesting levels, %zd max", __func__, sizeof(devlist->cur_sb) / sizeof(devlist->cur_sb[0])); devlist->cur_sb[devlist->level] = sbuf_new_auto(); if (devlist->cur_sb[devlist->level] == NULL) log_err(1, "%s: unable to allocate sbuf", __func__); if (strcmp(name, "lun") == 0) { if (cur_lun != NULL) log_errx(1, "%s: improper lun element nesting", __func__); cur_lun = calloc(1, sizeof(*cur_lun)); if (cur_lun == NULL) log_err(1, "%s: cannot allocate %zd bytes", __func__, sizeof(*cur_lun)); devlist->num_luns++; devlist->cur_lun = cur_lun; STAILQ_INIT(&cur_lun->attr_list); STAILQ_INSERT_TAIL(&devlist->lun_list, cur_lun, links); for (i = 0; attr[i] != NULL; i += 2) { if (strcmp(attr[i], "id") == 0) { cur_lun->lun_id = strtoull(attr[i+1], NULL, 0); } else { log_errx(1, "%s: invalid LUN attribute %s = %s", __func__, attr[i], attr[i+1]); } } } } static void cctl_end_element(void *user_data, const char *name) { struct cctl_devlist_data *devlist; struct cctl_lun *cur_lun; char *str; devlist = (struct cctl_devlist_data *)user_data; cur_lun = devlist->cur_lun; if ((cur_lun == NULL) && (strcmp(name, "ctllunlist") != 0)) log_errx(1, "%s: cur_lun == NULL! (name = %s)", __func__, name); if (devlist->cur_sb[devlist->level] == NULL) log_errx(1, "%s: no valid sbuf at level %d (name %s)", __func__, devlist->level, name); sbuf_finish(devlist->cur_sb[devlist->level]); str = checked_strdup(sbuf_data(devlist->cur_sb[devlist->level])); if (strlen(str) == 0) { free(str); str = NULL; } sbuf_delete(devlist->cur_sb[devlist->level]); devlist->cur_sb[devlist->level] = NULL; devlist->level--; if (strcmp(name, "backend_type") == 0) { cur_lun->backend_type = str; str = NULL; } else if (strcmp(name, "size") == 0) { cur_lun->size_blocks = strtoull(str, NULL, 0); } else if (strcmp(name, "blocksize") == 0) { cur_lun->blocksize = strtoul(str, NULL, 0); } else if (strcmp(name, "serial_number") == 0) { cur_lun->serial_number = str; str = NULL; } else if (strcmp(name, "device_id") == 0) { cur_lun->device_id = str; str = NULL; } else if (strcmp(name, "ctld_name") == 0) { cur_lun->ctld_name = str; str = NULL; } else if (strcmp(name, "lun") == 0) { devlist->cur_lun = NULL; } else if (strcmp(name, "ctllunlist") == 0) { /* Nothing. */ } else { struct cctl_lun_nv *nv; nv = calloc(1, sizeof(*nv)); if (nv == NULL) log_err(1, "%s: can't allocate %zd bytes for nv pair", __func__, sizeof(*nv)); nv->name = checked_strdup(name); nv->value = str; str = NULL; STAILQ_INSERT_TAIL(&cur_lun->attr_list, nv, links); } free(str); } static void cctl_start_pelement(void *user_data, const char *name, const char **attr) { int i; struct cctl_devlist_data *devlist; struct cctl_port *cur_port; devlist = (struct cctl_devlist_data *)user_data; cur_port = devlist->cur_port; devlist->level++; if ((u_int)devlist->level >= (sizeof(devlist->cur_sb) / sizeof(devlist->cur_sb[0]))) log_errx(1, "%s: too many nesting levels, %zd max", __func__, sizeof(devlist->cur_sb) / sizeof(devlist->cur_sb[0])); devlist->cur_sb[devlist->level] = sbuf_new_auto(); if (devlist->cur_sb[devlist->level] == NULL) log_err(1, "%s: unable to allocate sbuf", __func__); if (strcmp(name, "targ_port") == 0) { if (cur_port != NULL) log_errx(1, "%s: improper port element nesting (%s)", __func__, name); cur_port = calloc(1, sizeof(*cur_port)); if (cur_port == NULL) log_err(1, "%s: cannot allocate %zd bytes", __func__, sizeof(*cur_port)); devlist->num_ports++; devlist->cur_port = cur_port; STAILQ_INIT(&cur_port->attr_list); STAILQ_INSERT_TAIL(&devlist->port_list, cur_port, links); for (i = 0; attr[i] != NULL; i += 2) { if (strcmp(attr[i], "id") == 0) { cur_port->port_id = strtoul(attr[i+1], NULL, 0); } else { log_errx(1, "%s: invalid LUN attribute %s = %s", __func__, attr[i], attr[i+1]); } } } } static void cctl_end_pelement(void *user_data, const char *name) { struct cctl_devlist_data *devlist; struct cctl_port *cur_port; char *str; devlist = (struct cctl_devlist_data *)user_data; cur_port = devlist->cur_port; if ((cur_port == NULL) && (strcmp(name, "ctlportlist") != 0)) log_errx(1, "%s: cur_port == NULL! (name = %s)", __func__, name); if (devlist->cur_sb[devlist->level] == NULL) log_errx(1, "%s: no valid sbuf at level %d (name %s)", __func__, devlist->level, name); sbuf_finish(devlist->cur_sb[devlist->level]); str = checked_strdup(sbuf_data(devlist->cur_sb[devlist->level])); if (strlen(str) == 0) { free(str); str = NULL; } sbuf_delete(devlist->cur_sb[devlist->level]); devlist->cur_sb[devlist->level] = NULL; devlist->level--; if (strcmp(name, "port_name") == 0) { cur_port->port_name = str; str = NULL; } else if (strcmp(name, "cfiscsi_target") == 0) { cur_port->cfiscsi_target = str; str = NULL; } else if (strcmp(name, "cfiscsi_state") == 0) { cur_port->cfiscsi_state = strtoul(str, NULL, 0); } else if (strcmp(name, "cfiscsi_portal_group_tag") == 0) { cur_port->cfiscsi_portal_group_tag = strtoul(str, NULL, 0); } else if (strcmp(name, "ctld_portal_group_name") == 0) { cur_port->ctld_portal_group_name = str; str = NULL; } else if (strcmp(name, "targ_port") == 0) { devlist->cur_port = NULL; } else if (strcmp(name, "ctlportlist") == 0) { /* Nothing. */ } else { struct cctl_lun_nv *nv; nv = calloc(1, sizeof(*nv)); if (nv == NULL) log_err(1, "%s: can't allocate %zd bytes for nv pair", __func__, sizeof(*nv)); nv->name = checked_strdup(name); nv->value = str; str = NULL; STAILQ_INSERT_TAIL(&cur_port->attr_list, nv, links); } free(str); } static void cctl_char_handler(void *user_data, const XML_Char *str, int len) { struct cctl_devlist_data *devlist; devlist = (struct cctl_devlist_data *)user_data; sbuf_bcat(devlist->cur_sb[devlist->level], str, len); } struct conf * conf_new_from_kernel(void) { struct conf *conf = NULL; struct target *targ; struct portal_group *pg; struct pport *pp; struct port *cp; struct lun *cl; struct lun_option *lo; struct ctl_lun_list list; struct cctl_devlist_data devlist; struct cctl_lun *lun; struct cctl_port *port; XML_Parser parser; char *str; int len, retval; bzero(&devlist, sizeof(devlist)); STAILQ_INIT(&devlist.lun_list); STAILQ_INIT(&devlist.port_list); log_debugx("obtaining previously configured CTL luns from the kernel"); str = NULL; len = 4096; retry: str = realloc(str, len); if (str == NULL) log_err(1, "realloc"); bzero(&list, sizeof(list)); list.alloc_len = len; list.status = CTL_LUN_LIST_NONE; list.lun_xml = str; if (ioctl(ctl_fd, CTL_LUN_LIST, &list) == -1) { log_warn("error issuing CTL_LUN_LIST ioctl"); free(str); return (NULL); } if (list.status == CTL_LUN_LIST_ERROR) { log_warnx("error returned from CTL_LUN_LIST ioctl: %s", list.error_str); free(str); return (NULL); } if (list.status == CTL_LUN_LIST_NEED_MORE_SPACE) { len = len << 1; goto retry; } parser = XML_ParserCreate(NULL); if (parser == NULL) { log_warnx("unable to create XML parser"); free(str); return (NULL); } XML_SetUserData(parser, &devlist); XML_SetElementHandler(parser, cctl_start_element, cctl_end_element); XML_SetCharacterDataHandler(parser, cctl_char_handler); retval = XML_Parse(parser, str, strlen(str), 1); XML_ParserFree(parser); free(str); if (retval != 1) { log_warnx("XML_Parse failed"); return (NULL); } str = NULL; len = 4096; retry_port: str = realloc(str, len); if (str == NULL) log_err(1, "realloc"); bzero(&list, sizeof(list)); list.alloc_len = len; list.status = CTL_LUN_LIST_NONE; list.lun_xml = str; if (ioctl(ctl_fd, CTL_PORT_LIST, &list) == -1) { log_warn("error issuing CTL_PORT_LIST ioctl"); free(str); return (NULL); } if (list.status == CTL_PORT_LIST_ERROR) { log_warnx("error returned from CTL_PORT_LIST ioctl: %s", list.error_str); free(str); return (NULL); } if (list.status == CTL_LUN_LIST_NEED_MORE_SPACE) { len = len << 1; goto retry_port; } parser = XML_ParserCreate(NULL); if (parser == NULL) { log_warnx("unable to create XML parser"); free(str); return (NULL); } XML_SetUserData(parser, &devlist); XML_SetElementHandler(parser, cctl_start_pelement, cctl_end_pelement); XML_SetCharacterDataHandler(parser, cctl_char_handler); retval = XML_Parse(parser, str, strlen(str), 1); XML_ParserFree(parser); free(str); if (retval != 1) { log_warnx("XML_Parse failed"); return (NULL); } conf = conf_new(); STAILQ_FOREACH(port, &devlist.port_list, links) { if (port->cfiscsi_target == NULL) { log_debugx("CTL port %u \"%s\" wasn't managed by ctld; ", port->port_id, port->port_name); pp = pport_find(conf, port->port_name); if (pp == NULL) { #if 0 log_debugx("found new kernel port %u \"%s\"", port->port_id, port->port_name); #endif pp = pport_new(conf, port->port_name, port->port_id); if (pp == NULL) { log_warnx("pport_new failed"); continue; } } continue; } if (port->cfiscsi_state != 1) { log_debugx("CTL port %ju is not active (%d); ignoring", (uintmax_t)port->port_id, port->cfiscsi_state); continue; } targ = target_find(conf, port->cfiscsi_target); if (targ == NULL) { #if 0 log_debugx("found new kernel target %s for CTL port %ld", port->cfiscsi_target, port->port_id); #endif targ = target_new(conf, port->cfiscsi_target); if (targ == NULL) { log_warnx("target_new failed"); continue; } } if (port->ctld_portal_group_name == NULL) continue; pg = portal_group_find(conf, port->ctld_portal_group_name); if (pg == NULL) { #if 0 log_debugx("found new kernel portal group %s for CTL port %ld", port->ctld_portal_group_name, port->port_id); #endif pg = portal_group_new(conf, port->ctld_portal_group_name); if (pg == NULL) { log_warnx("portal_group_new failed"); continue; } } pg->pg_tag = port->cfiscsi_portal_group_tag; cp = port_new(conf, targ, pg); if (cp == NULL) { log_warnx("port_new failed"); continue; } cp->p_ctl_port = port->port_id; } STAILQ_FOREACH(lun, &devlist.lun_list, links) { struct cctl_lun_nv *nv; if (lun->ctld_name == NULL) { log_debugx("CTL lun %ju wasn't managed by ctld; " "ignoring", (uintmax_t)lun->lun_id); continue; } cl = lun_find(conf, lun->ctld_name); if (cl != NULL) { log_warnx("found CTL lun %ju \"%s\", " "also backed by CTL lun %d; ignoring", (uintmax_t)lun->lun_id, lun->ctld_name, cl->l_ctl_lun); continue; } log_debugx("found CTL lun %ju \"%s\"", (uintmax_t)lun->lun_id, lun->ctld_name); cl = lun_new(conf, lun->ctld_name); if (cl == NULL) { log_warnx("lun_new failed"); continue; } lun_set_backend(cl, lun->backend_type); lun_set_blocksize(cl, lun->blocksize); lun_set_device_id(cl, lun->device_id); lun_set_serial(cl, lun->serial_number); lun_set_size(cl, lun->size_blocks * cl->l_blocksize); lun_set_ctl_lun(cl, lun->lun_id); STAILQ_FOREACH(nv, &lun->attr_list, links) { if (strcmp(nv->name, "file") == 0 || strcmp(nv->name, "dev") == 0) { lun_set_path(cl, nv->value); continue; } lo = lun_option_new(cl, nv->name, nv->value); if (lo == NULL) log_warnx("unable to add CTL lun option %s " "for CTL lun %ju \"%s\"", nv->name, (uintmax_t) lun->lun_id, cl->l_name); } } return (conf); } static void str_arg(struct ctl_be_arg *arg, const char *name, const char *value) { arg->namelen = strlen(name) + 1; arg->name = __DECONST(char *, name); arg->vallen = strlen(value) + 1; arg->value = __DECONST(char *, value); arg->flags = CTL_BEARG_ASCII | CTL_BEARG_RD; } int kernel_lun_add(struct lun *lun) { struct lun_option *lo; struct ctl_lun_req req; int error, i, num_options; bzero(&req, sizeof(req)); strlcpy(req.backend, lun->l_backend, sizeof(req.backend)); req.reqtype = CTL_LUNREQ_CREATE; req.reqdata.create.blocksize_bytes = lun->l_blocksize; if (lun->l_size != 0) req.reqdata.create.lun_size_bytes = lun->l_size; req.reqdata.create.flags |= CTL_LUN_FLAG_DEV_TYPE; req.reqdata.create.device_type = T_DIRECT; if (lun->l_serial != NULL) { strncpy(req.reqdata.create.serial_num, lun->l_serial, sizeof(req.reqdata.create.serial_num)); req.reqdata.create.flags |= CTL_LUN_FLAG_SERIAL_NUM; } if (lun->l_device_id != NULL) { strncpy(req.reqdata.create.device_id, lun->l_device_id, sizeof(req.reqdata.create.device_id)); req.reqdata.create.flags |= CTL_LUN_FLAG_DEVID; } if (lun->l_path != NULL) { lo = lun_option_find(lun, "file"); if (lo != NULL) { lun_option_set(lo, lun->l_path); } else { lo = lun_option_new(lun, "file", lun->l_path); assert(lo != NULL); } } lo = lun_option_find(lun, "ctld_name"); if (lo != NULL) { lun_option_set(lo, lun->l_name); } else { lo = lun_option_new(lun, "ctld_name", lun->l_name); assert(lo != NULL); } lo = lun_option_find(lun, "scsiname"); if (lo == NULL && lun->l_scsiname != NULL) { lo = lun_option_new(lun, "scsiname", lun->l_scsiname); assert(lo != NULL); } num_options = 0; TAILQ_FOREACH(lo, &lun->l_options, lo_next) num_options++; req.num_be_args = num_options; if (num_options > 0) { req.be_args = malloc(num_options * sizeof(*req.be_args)); if (req.be_args == NULL) { log_warn("error allocating %zd bytes", num_options * sizeof(*req.be_args)); return (1); } i = 0; TAILQ_FOREACH(lo, &lun->l_options, lo_next) { str_arg(&req.be_args[i], lo->lo_name, lo->lo_value); i++; } assert(i == num_options); } error = ioctl(ctl_fd, CTL_LUN_REQ, &req); free(req.be_args); if (error != 0) { log_warn("error issuing CTL_LUN_REQ ioctl"); return (1); } switch (req.status) { case CTL_LUN_ERROR: log_warnx("LUN creation error: %s", req.error_str); return (1); case CTL_LUN_WARNING: log_warnx("LUN creation warning: %s", req.error_str); break; case CTL_LUN_OK: break; default: log_warnx("unknown LUN creation status: %d", req.status); return (1); } lun_set_ctl_lun(lun, req.reqdata.create.req_lun_id); return (0); } int kernel_lun_resize(struct lun *lun) { struct ctl_lun_req req; bzero(&req, sizeof(req)); strlcpy(req.backend, lun->l_backend, sizeof(req.backend)); req.reqtype = CTL_LUNREQ_MODIFY; req.reqdata.modify.lun_id = lun->l_ctl_lun; req.reqdata.modify.lun_size_bytes = lun->l_size; if (ioctl(ctl_fd, CTL_LUN_REQ, &req) == -1) { log_warn("error issuing CTL_LUN_REQ ioctl"); return (1); } switch (req.status) { case CTL_LUN_ERROR: log_warnx("LUN modification error: %s", req.error_str); return (1); case CTL_LUN_WARNING: log_warnx("LUN modification warning: %s", req.error_str); break; case CTL_LUN_OK: break; default: log_warnx("unknown LUN modification status: %d", req.status); return (1); } return (0); } int kernel_lun_remove(struct lun *lun) { struct ctl_lun_req req; bzero(&req, sizeof(req)); strlcpy(req.backend, lun->l_backend, sizeof(req.backend)); req.reqtype = CTL_LUNREQ_RM; req.reqdata.rm.lun_id = lun->l_ctl_lun; if (ioctl(ctl_fd, CTL_LUN_REQ, &req) == -1) { log_warn("error issuing CTL_LUN_REQ ioctl"); return (1); } switch (req.status) { case CTL_LUN_ERROR: log_warnx("LUN removal error: %s", req.error_str); return (1); case CTL_LUN_WARNING: log_warnx("LUN removal warning: %s", req.error_str); break; case CTL_LUN_OK: break; default: log_warnx("unknown LUN removal status: %d", req.status); return (1); } return (0); } void kernel_handoff(struct connection *conn) { struct ctl_iscsi req; bzero(&req, sizeof(req)); req.type = CTL_ISCSI_HANDOFF; strlcpy(req.data.handoff.initiator_name, conn->conn_initiator_name, sizeof(req.data.handoff.initiator_name)); strlcpy(req.data.handoff.initiator_addr, conn->conn_initiator_addr, sizeof(req.data.handoff.initiator_addr)); if (conn->conn_initiator_alias != NULL) { strlcpy(req.data.handoff.initiator_alias, conn->conn_initiator_alias, sizeof(req.data.handoff.initiator_alias)); } memcpy(req.data.handoff.initiator_isid, conn->conn_initiator_isid, sizeof(req.data.handoff.initiator_isid)); strlcpy(req.data.handoff.target_name, conn->conn_target->t_name, sizeof(req.data.handoff.target_name)); if (conn->conn_portal->p_portal_group->pg_offload != NULL) { strlcpy(req.data.handoff.offload, conn->conn_portal->p_portal_group->pg_offload, sizeof(req.data.handoff.offload)); } #ifdef ICL_KERNEL_PROXY if (proxy_mode) req.data.handoff.connection_id = conn->conn_socket; else req.data.handoff.socket = conn->conn_socket; #else req.data.handoff.socket = conn->conn_socket; #endif req.data.handoff.portal_group_tag = conn->conn_portal->p_portal_group->pg_tag; if (conn->conn_header_digest == CONN_DIGEST_CRC32C) req.data.handoff.header_digest = CTL_ISCSI_DIGEST_CRC32C; if (conn->conn_data_digest == CONN_DIGEST_CRC32C) req.data.handoff.data_digest = CTL_ISCSI_DIGEST_CRC32C; req.data.handoff.cmdsn = conn->conn_cmdsn; req.data.handoff.statsn = conn->conn_statsn; req.data.handoff.max_recv_data_segment_length = conn->conn_max_data_segment_length; req.data.handoff.max_burst_length = conn->conn_max_burst_length; req.data.handoff.immediate_data = conn->conn_immediate_data; if (ioctl(ctl_fd, CTL_ISCSI, &req) == -1) { log_err(1, "error issuing CTL_ISCSI ioctl; " "dropping connection"); } if (req.status != CTL_ISCSI_OK) { log_errx(1, "error returned from CTL iSCSI handoff request: " "%s; dropping connection", req.error_str); } } void kernel_limits(const char *offload, size_t *max_data_segment_length) { struct ctl_iscsi req; bzero(&req, sizeof(req)); req.type = CTL_ISCSI_LIMITS; if (offload != NULL) { strlcpy(req.data.limits.offload, offload, sizeof(req.data.limits.offload)); } if (ioctl(ctl_fd, CTL_ISCSI, &req) == -1) { log_err(1, "error issuing CTL_ISCSI ioctl; " "dropping connection"); } if (req.status != CTL_ISCSI_OK) { log_errx(1, "error returned from CTL iSCSI limits request: " "%s; dropping connection", req.error_str); } *max_data_segment_length = req.data.limits.data_segment_limit; if (offload != NULL) { log_debugx("MaxRecvDataSegment kernel limit for offload " "\"%s\" is %zd", offload, *max_data_segment_length); } else { log_debugx("MaxRecvDataSegment kernel limit is %zd", *max_data_segment_length); } } int kernel_port_add(struct port *port) { struct ctl_port_entry entry; struct ctl_req req; struct ctl_lun_map lm; struct target *targ = port->p_target; struct portal_group *pg = port->p_portal_group; char tagstr[16]; int error, i, n; /* Create iSCSI port. */ if (port->p_portal_group) { bzero(&req, sizeof(req)); strlcpy(req.driver, "iscsi", sizeof(req.driver)); req.reqtype = CTL_REQ_CREATE; req.num_args = 5; req.args = malloc(req.num_args * sizeof(*req.args)); if (req.args == NULL) log_err(1, "malloc"); n = 0; req.args[n].namelen = sizeof("port_id"); req.args[n].name = __DECONST(char *, "port_id"); req.args[n].vallen = sizeof(port->p_ctl_port); req.args[n].value = &port->p_ctl_port; req.args[n++].flags = CTL_BEARG_WR; str_arg(&req.args[n++], "cfiscsi_target", targ->t_name); snprintf(tagstr, sizeof(tagstr), "%d", pg->pg_tag); str_arg(&req.args[n++], "cfiscsi_portal_group_tag", tagstr); if (targ->t_alias) str_arg(&req.args[n++], "cfiscsi_target_alias", targ->t_alias); str_arg(&req.args[n++], "ctld_portal_group_name", pg->pg_name); req.num_args = n; error = ioctl(ctl_fd, CTL_PORT_REQ, &req); free(req.args); if (error != 0) { log_warn("error issuing CTL_PORT_REQ ioctl"); return (1); } if (req.status == CTL_LUN_ERROR) { log_warnx("error returned from port creation request: %s", req.error_str); return (1); } if (req.status != CTL_LUN_OK) { log_warnx("unknown port creation request status %d", req.status); return (1); } - } else if (port->p_pport) + } else if (port->p_pport) { port->p_ctl_port = port->p_pport->pp_ctl_port; + + if (strncmp(targ->t_name, "naa.", 4) == 0 && + strlen(targ->t_name) == 20) { + bzero(&entry, sizeof(entry)); + entry.port_type = CTL_PORT_NONE; + entry.targ_port = port->p_ctl_port; + entry.flags |= CTL_PORT_WWNN_VALID; + entry.wwnn = strtoull(targ->t_name + 4, NULL, 16); + if (ioctl(ctl_fd, CTL_SET_PORT_WWNS, &entry) == -1) + log_warn("CTL_SET_PORT_WWNS ioctl failed"); + } + } /* Explicitly enable mapping to block any access except allowed. */ lm.port = port->p_ctl_port; lm.plun = UINT32_MAX; lm.lun = 0; error = ioctl(ctl_fd, CTL_LUN_MAP, &lm); if (error != 0) log_warn("CTL_LUN_MAP ioctl failed"); /* Map configured LUNs */ for (i = 0; i < MAX_LUNS; i++) { if (targ->t_luns[i] == NULL) continue; lm.port = port->p_ctl_port; lm.plun = i; lm.lun = targ->t_luns[i]->l_ctl_lun; error = ioctl(ctl_fd, CTL_LUN_MAP, &lm); if (error != 0) log_warn("CTL_LUN_MAP ioctl failed"); } /* Enable port */ bzero(&entry, sizeof(entry)); entry.targ_port = port->p_ctl_port; error = ioctl(ctl_fd, CTL_ENABLE_PORT, &entry); if (error != 0) { log_warn("CTL_ENABLE_PORT ioctl failed"); return (-1); } return (0); } int kernel_port_update(struct port *port) { struct ctl_lun_map lm; struct target *targ = port->p_target; int error, i; /* Map configured LUNs and unmap others */ for (i = 0; i < MAX_LUNS; i++) { lm.port = port->p_ctl_port; lm.plun = i; if (targ->t_luns[i] == NULL) lm.lun = UINT32_MAX; else lm.lun = targ->t_luns[i]->l_ctl_lun; error = ioctl(ctl_fd, CTL_LUN_MAP, &lm); if (error != 0) log_warn("CTL_LUN_MAP ioctl failed"); } return (0); } int kernel_port_remove(struct port *port) { struct ctl_port_entry entry; struct ctl_lun_map lm; struct ctl_req req; char tagstr[16]; struct target *targ = port->p_target; struct portal_group *pg = port->p_portal_group; int error; /* Disable port */ bzero(&entry, sizeof(entry)); entry.targ_port = port->p_ctl_port; error = ioctl(ctl_fd, CTL_DISABLE_PORT, &entry); if (error != 0) { log_warn("CTL_DISABLE_PORT ioctl failed"); return (-1); } /* Remove iSCSI port. */ if (port->p_portal_group) { bzero(&req, sizeof(req)); strlcpy(req.driver, "iscsi", sizeof(req.driver)); req.reqtype = CTL_REQ_REMOVE; req.num_args = 2; req.args = malloc(req.num_args * sizeof(*req.args)); if (req.args == NULL) log_err(1, "malloc"); str_arg(&req.args[0], "cfiscsi_target", targ->t_name); snprintf(tagstr, sizeof(tagstr), "%d", pg->pg_tag); str_arg(&req.args[1], "cfiscsi_portal_group_tag", tagstr); error = ioctl(ctl_fd, CTL_PORT_REQ, &req); free(req.args); if (error != 0) { log_warn("error issuing CTL_PORT_REQ ioctl"); return (1); } if (req.status == CTL_LUN_ERROR) { log_warnx("error returned from port removal request: %s", req.error_str); return (1); } if (req.status != CTL_LUN_OK) { log_warnx("unknown port removal request status %d", req.status); return (1); } } else { /* Disable LUN mapping. */ lm.port = port->p_ctl_port; lm.plun = UINT32_MAX; lm.lun = UINT32_MAX; error = ioctl(ctl_fd, CTL_LUN_MAP, &lm); if (error != 0) log_warn("CTL_LUN_MAP ioctl failed"); } return (0); } #ifdef ICL_KERNEL_PROXY void kernel_listen(struct addrinfo *ai, bool iser, int portal_id) { struct ctl_iscsi req; bzero(&req, sizeof(req)); req.type = CTL_ISCSI_LISTEN; req.data.listen.iser = iser; req.data.listen.domain = ai->ai_family; req.data.listen.socktype = ai->ai_socktype; req.data.listen.protocol = ai->ai_protocol; req.data.listen.addr = ai->ai_addr; req.data.listen.addrlen = ai->ai_addrlen; req.data.listen.portal_id = portal_id; if (ioctl(ctl_fd, CTL_ISCSI, &req) == -1) log_err(1, "error issuing CTL_ISCSI ioctl"); if (req.status != CTL_ISCSI_OK) { log_errx(1, "error returned from CTL iSCSI listen: %s", req.error_str); } } void kernel_accept(int *connection_id, int *portal_id, struct sockaddr *client_sa, socklen_t *client_salen) { struct ctl_iscsi req; struct sockaddr_storage ss; bzero(&req, sizeof(req)); req.type = CTL_ISCSI_ACCEPT; req.data.accept.initiator_addr = (struct sockaddr *)&ss; if (ioctl(ctl_fd, CTL_ISCSI, &req) == -1) log_err(1, "error issuing CTL_ISCSI ioctl"); if (req.status != CTL_ISCSI_OK) { log_errx(1, "error returned from CTL iSCSI accept: %s", req.error_str); } *connection_id = req.data.accept.connection_id; *portal_id = req.data.accept.portal_id; *client_salen = req.data.accept.initiator_addrlen; memcpy(client_sa, &ss, *client_salen); } void kernel_send(struct pdu *pdu) { struct ctl_iscsi req; bzero(&req, sizeof(req)); req.type = CTL_ISCSI_SEND; req.data.send.connection_id = pdu->pdu_connection->conn_socket; req.data.send.bhs = pdu->pdu_bhs; req.data.send.data_segment_len = pdu->pdu_data_len; req.data.send.data_segment = pdu->pdu_data; if (ioctl(ctl_fd, CTL_ISCSI, &req) == -1) { log_err(1, "error issuing CTL_ISCSI ioctl; " "dropping connection"); } if (req.status != CTL_ISCSI_OK) { log_errx(1, "error returned from CTL iSCSI send: " "%s; dropping connection", req.error_str); } } void kernel_receive(struct pdu *pdu) { struct ctl_iscsi req; pdu->pdu_data = malloc(MAX_DATA_SEGMENT_LENGTH); if (pdu->pdu_data == NULL) log_err(1, "malloc"); bzero(&req, sizeof(req)); req.type = CTL_ISCSI_RECEIVE; req.data.receive.connection_id = pdu->pdu_connection->conn_socket; req.data.receive.bhs = pdu->pdu_bhs; req.data.receive.data_segment_len = MAX_DATA_SEGMENT_LENGTH; req.data.receive.data_segment = pdu->pdu_data; if (ioctl(ctl_fd, CTL_ISCSI, &req) == -1) { log_err(1, "error issuing CTL_ISCSI ioctl; " "dropping connection"); } if (req.status != CTL_ISCSI_OK) { log_errx(1, "error returned from CTL iSCSI receive: " "%s; dropping connection", req.error_str); } } #endif /* ICL_KERNEL_PROXY */ /* * XXX: I CANT INTO LATIN */ void kernel_capsicate(void) { int error; cap_rights_t rights; const unsigned long cmds[] = { CTL_ISCSI }; cap_rights_init(&rights, CAP_IOCTL); error = cap_rights_limit(ctl_fd, &rights); if (error != 0 && errno != ENOSYS) log_err(1, "cap_rights_limit"); error = cap_ioctls_limit(ctl_fd, cmds, sizeof(cmds) / sizeof(cmds[0])); if (error != 0 && errno != ENOSYS) log_err(1, "cap_ioctls_limit"); error = cap_enter(); if (error != 0 && errno != ENOSYS) log_err(1, "cap_enter"); if (cap_sandboxed()) log_debugx("Capsicum capability mode enabled"); else log_warnx("Capsicum capability mode not supported"); } Index: projects/cxl_iscsi/usr.sbin/freebsd-update/freebsd-update.8 =================================================================== --- projects/cxl_iscsi/usr.sbin/freebsd-update/freebsd-update.8 (revision 279716) +++ projects/cxl_iscsi/usr.sbin/freebsd-update/freebsd-update.8 (revision 279717) @@ -1,177 +1,192 @@ .\"- .\" Copyright 2006, 2007 Colin Percival .\" All rights reserved .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted providing that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR .\" IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED .\" WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY .\" DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, .\" STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING .\" IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE .\" POSSIBILITY OF SUCH DAMAGE. .\" .\" $FreeBSD$ .\" -.Dd July 14, 2010 +.Dd March 2, 2015 .Dt FREEBSD-UPDATE 8 .Os FreeBSD .Sh NAME .Nm freebsd-update .Nd fetch and install binary updates to FreeBSD .Sh SYNOPSIS .Nm .Op Fl b Ar basedir .Op Fl d Ar workdir .Op Fl f Ar conffile +.Op Fl F .Op Fl k Ar KEY .Op Fl r Ar newrelease .Op Fl s Ar server .Op Fl t Ar address +.Op Fl -not-running-from-cron .Cm command ... .Sh DESCRIPTION The .Nm tool is used to fetch, install, and rollback binary updates to the FreeBSD base system. Note that updates are only available if they are being built for the FreeBSD release and architecture being used; in particular, the .Fx Security Team only builds updates for releases shipped in binary form by the .Fx Release Engineering Team, e.g., .Fx -7.3-RELEASE and +9.3-RELEASE and .Fx -8.0-RELEASE, but not +10.1-RELEASE, but not .Fx -6.3-STABLE or +9.3-STABLE or .Fx -9.0-CURRENT. +11-CURRENT. .Sh OPTIONS The following options are supported: -.Bl -tag -width "-f conffile" +.Bl -tag -width "-r newrelease" .It Fl b Ar basedir Operate on a system mounted at .Ar basedir . (default: .Pa / , or as given in the configuration file.) .It Fl d Ar workdir Store working files in .Ar workdir . (default: .Pa /var/db/freebsd-update/ , or as given in the configuration file.) .It Fl f Ar conffile Read configuration options from .Ar conffile . (default: .Pa /etc/freebsd-update.conf ) +.It Fl F +Force +.Nm Cm fetch +to proceed where it normally would not, such as an unfinished upgrade .It Fl k Ar KEY Trust an RSA key with SHA256 of .Ar KEY . (default: read value from configuration file.) .It Fl r Ar newrelease Specify the new release to which .Nm should upgrade (upgrade command only). .It Fl s Ar server Fetch files from the specified server or server pool. (default: read value from configuration file.) .It Fl t Ar address Mail output of .Cm cron command, if any, to .Ar address . (default: root, or as given in the configuration file.) +.It Fl -not-running-from-cron +Force +.Nm Cm fetch +to proceed when there is no controlling tty. +This is for use by automated scripts and orchestration tools. +Please do not run +.Nm Cm fetch +from crontab or similar using this flag, see: +.Nm Cm cron .El .Sh COMMANDS The .Cm command can be any one of the following: -.Bl -tag -width "-f conffile" +.Bl -tag -width "rollback" .It Cm fetch Based on the currently installed world and the configuration options set, fetch all available binary updates. .It Cm cron Sleep a random amount of time between 1 and 3600 seconds, then download updates as if the .Cm fetch command was used. If updates are downloaded, an email will be sent (to root or a different address if specified via the .Fl t option or in the configuration file). As the name suggests, this command is designed for running from .Xr cron 8 ; the random delay serves to minimize the probability that a large number of machines will simultaneously attempt to fetch updates. .It Cm upgrade Fetch files necessary for upgrading to a new release. Before using this command, make sure that you read the announcement and release notes for the new release in case there are any special steps needed for upgrading. Note that this command may require up to 500 MB of space in .Ar workdir depending on which components of the .Fx base system are installed. .It Cm install Install the most recently fetched updates or upgrade. .It Cm rollback Uninstall the most recently installed updates. .It Cm IDS Compare the system against a "known good" index of the installed release. .El .Sh TIPS .Bl -bullet .It If your clock is set to local time, adding the line .Pp .Dl 0 3 * * * root /usr/sbin/freebsd-update cron .Pp to /etc/crontab will check for updates every night. If your clock is set to UTC, please pick a random time other than 3AM, to avoid overly imposing an uneven load on the server(s) hosting the updates. .It In spite of its name, .Nm IDS should not be relied upon as an "Intrusion Detection System", since if the system has been tampered with it cannot be trusted to operate correctly. If you intend to use this command for intrusion-detection purposes, make sure you boot from a secure disk (e.g., a CD). .El .Sh FILES .Bl -tag -width "/etc/freebsd-update.conf" .It Pa /etc/freebsd-update.conf Default location of the .Nm configuration file. .It Pa /var/db/freebsd-update/ Default location where .Nm stores temporary files and downloaded updates. .El .Sh SEE ALSO .Xr freebsd-update.conf 5 .Sh AUTHORS .An Colin Percival Aq Mt cperciva@FreeBSD.org Index: projects/cxl_iscsi/usr.sbin/freebsd-update/freebsd-update.sh =================================================================== --- projects/cxl_iscsi/usr.sbin/freebsd-update/freebsd-update.sh (revision 279716) +++ projects/cxl_iscsi/usr.sbin/freebsd-update/freebsd-update.sh (revision 279717) @@ -1,3268 +1,3291 @@ #!/bin/sh #- # Copyright 2004-2007 Colin Percival # All rights reserved # # Redistribution and use in source and binary forms, with or without # modification, are permitted providing that the following conditions # are met: # 1. Redistributions of source code must retain the above copyright # notice, this list of conditions and the following disclaimer. # 2. Redistributions in binary form must reproduce the above copyright # notice, this list of conditions and the following disclaimer in the # documentation and/or other materials provided with the distribution. # # THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR # IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED # WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE # ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY # DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL # DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS # OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) # HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, # STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING # IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE # POSSIBILITY OF SUCH DAMAGE. # $FreeBSD$ #### Usage function -- called from command-line handling code. # Usage instructions. Options not listed: # --debug -- don't filter output from utilities # --no-stats -- don't show progress statistics while fetching files usage () { cat < ${LINE}" exit 1 fi done < ${CONFFILE} # Merge the settings read from the configuration file with those # provided at the command line. mergeconfig } # Provide some default parameters default_params () { # Save any parameters already configured, and clear the slate saveconfig nullconfig # Default configurations config_WorkDir /var/db/freebsd-update config_MailTo root config_AllowAdd yes config_AllowDelete yes config_KeepModifiedMetadata yes config_BaseDir / config_VerboseLevel stats config_StrictComponents no config_BackupKernel yes config_BackupKernelDir /boot/kernel.old config_BackupKernelSymbolFiles no # Merge these defaults into the earlier-configured settings mergeconfig } # Set utility output filtering options, based on ${VERBOSELEVEL} fetch_setup_verboselevel () { case ${VERBOSELEVEL} in debug) QUIETREDIR="/dev/stderr" QUIETFLAG=" " STATSREDIR="/dev/stderr" DDSTATS=".." XARGST="-t" NDEBUG=" " ;; nostats) QUIETREDIR="" QUIETFLAG="" STATSREDIR="/dev/null" DDSTATS=".." XARGST="" NDEBUG="" ;; stats) QUIETREDIR="/dev/null" QUIETFLAG="-q" STATSREDIR="/dev/stdout" DDSTATS="" XARGST="" NDEBUG="-n" ;; esac } # Perform sanity checks and set some final parameters # in preparation for fetching files. Figure out which # set of updates should be downloaded: If the user is # running *-p[0-9]+, strip off the last part; if the # user is running -SECURITY, call it -RELEASE. Chdir # into the working directory. fetchupgrade_check_params () { export HTTP_USER_AGENT="freebsd-update (${COMMAND}, `uname -r`)" _SERVERNAME_z=\ "SERVERNAME must be given via command line or configuration file." _KEYPRINT_z="Key must be given via -k option or configuration file." _KEYPRINT_bad="Invalid key fingerprint: " _WORKDIR_bad="Directory does not exist or is not writable: " _WORKDIR_bad2="Directory is not on a persistent filesystem: " if [ -z "${SERVERNAME}" ]; then echo -n "`basename $0`: " echo "${_SERVERNAME_z}" exit 1 fi if [ -z "${KEYPRINT}" ]; then echo -n "`basename $0`: " echo "${_KEYPRINT_z}" exit 1 fi if ! echo "${KEYPRINT}" | grep -qE "^[0-9a-f]{64}$"; then echo -n "`basename $0`: " echo -n "${_KEYPRINT_bad}" echo ${KEYPRINT} exit 1 fi if ! [ -d "${WORKDIR}" -a -w "${WORKDIR}" ]; then echo -n "`basename $0`: " echo -n "${_WORKDIR_bad}" echo ${WORKDIR} exit 1 fi case `df -T ${WORKDIR}` in */dev/md[0-9]* | *tmpfs*) echo -n "`basename $0`: " echo -n "${_WORKDIR_bad2}" echo ${WORKDIR} exit 1 ;; esac chmod 700 ${WORKDIR} cd ${WORKDIR} || exit 1 # Generate release number. The s/SECURITY/RELEASE/ bit exists # to provide an upgrade path for FreeBSD Update 1.x users, since # the kernels provided by FreeBSD Update 1.x are always labelled # as X.Y-SECURITY. RELNUM=`uname -r | sed -E 's,-p[0-9]+,,' | sed -E 's,-SECURITY,-RELEASE,'` ARCH=`uname -m` FETCHDIR=${RELNUM}/${ARCH} PATCHDIR=${RELNUM}/${ARCH}/bp # Figure out what directory contains the running kernel BOOTFILE=`sysctl -n kern.bootfile` KERNELDIR=${BOOTFILE%/kernel} if ! [ -d ${KERNELDIR} ]; then echo "Cannot identify running kernel" exit 1 fi # Figure out what kernel configuration is running. We start with # the output of `uname -i`, and then make the following adjustments: # 1. Replace "SMP-GENERIC" with "SMP". Why the SMP kernel config # file says "ident SMP-GENERIC", I don't know... # 2. If the kernel claims to be GENERIC _and_ ${ARCH} is "amd64" # _and_ `sysctl kern.version` contains a line which ends "/SMP", then # we're running an SMP kernel. This mis-identification is a bug # which was fixed in 6.2-STABLE. KERNCONF=`uname -i` if [ ${KERNCONF} = "SMP-GENERIC" ]; then KERNCONF=SMP fi if [ ${KERNCONF} = "GENERIC" ] && [ ${ARCH} = "amd64" ]; then if sysctl kern.version | grep -qE '/SMP$'; then KERNCONF=SMP fi fi # Define some paths BSPATCH=/usr/bin/bspatch SHA256=/sbin/sha256 PHTTPGET=/usr/libexec/phttpget # Set up variables relating to VERBOSELEVEL fetch_setup_verboselevel # Construct a unique name from ${BASEDIR} BDHASH=`echo ${BASEDIR} | sha256 -q` } # Perform sanity checks etc. before fetching updates. fetch_check_params () { fetchupgrade_check_params if ! [ -z "${TARGETRELEASE}" ]; then echo -n "`basename $0`: " echo -n "-r option is meaningless with 'fetch' command. " echo "(Did you mean 'upgrade' instead?)" exit 1 fi + + # Check that we have updates ready to install + if [ -f ${BDHASH}-install/kerneldone && $FORCEFETCH -eq 0 ]; then + echo "You have a partially completed upgrade pending" + echo "Run '$0 install' first." + echo "Run '$0 fetch -F' to proceed anyway." + exit 1 + fi } # Perform sanity checks etc. before fetching upgrades. upgrade_check_params () { fetchupgrade_check_params # Unless set otherwise, we're upgrading to the same kernel config. NKERNCONF=${KERNCONF} # We need TARGETRELEASE set _TARGETRELEASE_z="Release target must be specified via -r option." if [ -z "${TARGETRELEASE}" ]; then echo -n "`basename $0`: " echo "${_TARGETRELEASE_z}" exit 1 fi # The target release should be != the current release. if [ "${TARGETRELEASE}" = "${RELNUM}" ]; then echo -n "`basename $0`: " echo "Cannot upgrade from ${RELNUM} to itself" exit 1 fi # Turning off AllowAdd or AllowDelete is a bad idea for upgrades. if [ "${ALLOWADD}" = "no" ]; then echo -n "`basename $0`: " echo -n "WARNING: \"AllowAdd no\" is a bad idea " echo "when upgrading between releases." echo fi if [ "${ALLOWDELETE}" = "no" ]; then echo -n "`basename $0`: " echo -n "WARNING: \"AllowDelete no\" is a bad idea " echo "when upgrading between releases." echo fi # Set EDITOR to /usr/bin/vi if it isn't already set : ${EDITOR:='/usr/bin/vi'} } # Perform sanity checks and set some final parameters in # preparation for installing updates. install_check_params () { # Check that we are root. All sorts of things won't work otherwise. if [ `id -u` != 0 ]; then echo "You must be root to run this." exit 1 fi # Check that securelevel <= 0. Otherwise we can't update schg files. if [ `sysctl -n kern.securelevel` -gt 0 ]; then echo "Updates cannot be installed when the system securelevel" echo "is greater than zero." exit 1 fi # Check that we have a working directory _WORKDIR_bad="Directory does not exist or is not writable: " if ! [ -d "${WORKDIR}" -a -w "${WORKDIR}" ]; then echo -n "`basename $0`: " echo -n "${_WORKDIR_bad}" echo ${WORKDIR} exit 1 fi cd ${WORKDIR} || exit 1 # Construct a unique name from ${BASEDIR} BDHASH=`echo ${BASEDIR} | sha256 -q` # Check that we have updates ready to install if ! [ -L ${BDHASH}-install ]; then echo "No updates are available to install." echo "Run '$0 fetch' first." exit 1 fi if ! [ -f ${BDHASH}-install/INDEX-OLD ] || ! [ -f ${BDHASH}-install/INDEX-NEW ]; then echo "Update manifest is corrupt -- this should never happen." echo "Re-run '$0 fetch'." exit 1 fi # Figure out what directory contains the running kernel BOOTFILE=`sysctl -n kern.bootfile` KERNELDIR=${BOOTFILE%/kernel} if ! [ -d ${KERNELDIR} ]; then echo "Cannot identify running kernel" exit 1 fi } # Perform sanity checks and set some final parameters in # preparation for UNinstalling updates. rollback_check_params () { # Check that we are root. All sorts of things won't work otherwise. if [ `id -u` != 0 ]; then echo "You must be root to run this." exit 1 fi # Check that we have a working directory _WORKDIR_bad="Directory does not exist or is not writable: " if ! [ -d "${WORKDIR}" -a -w "${WORKDIR}" ]; then echo -n "`basename $0`: " echo -n "${_WORKDIR_bad}" echo ${WORKDIR} exit 1 fi cd ${WORKDIR} || exit 1 # Construct a unique name from ${BASEDIR} BDHASH=`echo ${BASEDIR} | sha256 -q` # Check that we have updates ready to rollback if ! [ -L ${BDHASH}-rollback ]; then echo "No rollback directory found." exit 1 fi if ! [ -f ${BDHASH}-rollback/INDEX-OLD ] || ! [ -f ${BDHASH}-rollback/INDEX-NEW ]; then echo "Update manifest is corrupt -- this should never happen." exit 1 fi } # Perform sanity checks and set some final parameters # in preparation for comparing the system against the # published index. Figure out which index we should # compare against: If the user is running *-p[0-9]+, # strip off the last part; if the user is running # -SECURITY, call it -RELEASE. Chdir into the working # directory. IDS_check_params () { export HTTP_USER_AGENT="freebsd-update (${COMMAND}, `uname -r`)" _SERVERNAME_z=\ "SERVERNAME must be given via command line or configuration file." _KEYPRINT_z="Key must be given via -k option or configuration file." _KEYPRINT_bad="Invalid key fingerprint: " _WORKDIR_bad="Directory does not exist or is not writable: " if [ -z "${SERVERNAME}" ]; then echo -n "`basename $0`: " echo "${_SERVERNAME_z}" exit 1 fi if [ -z "${KEYPRINT}" ]; then echo -n "`basename $0`: " echo "${_KEYPRINT_z}" exit 1 fi if ! echo "${KEYPRINT}" | grep -qE "^[0-9a-f]{64}$"; then echo -n "`basename $0`: " echo -n "${_KEYPRINT_bad}" echo ${KEYPRINT} exit 1 fi if ! [ -d "${WORKDIR}" -a -w "${WORKDIR}" ]; then echo -n "`basename $0`: " echo -n "${_WORKDIR_bad}" echo ${WORKDIR} exit 1 fi cd ${WORKDIR} || exit 1 # Generate release number. The s/SECURITY/RELEASE/ bit exists # to provide an upgrade path for FreeBSD Update 1.x users, since # the kernels provided by FreeBSD Update 1.x are always labelled # as X.Y-SECURITY. RELNUM=`uname -r | sed -E 's,-p[0-9]+,,' | sed -E 's,-SECURITY,-RELEASE,'` ARCH=`uname -m` FETCHDIR=${RELNUM}/${ARCH} PATCHDIR=${RELNUM}/${ARCH}/bp # Figure out what directory contains the running kernel BOOTFILE=`sysctl -n kern.bootfile` KERNELDIR=${BOOTFILE%/kernel} if ! [ -d ${KERNELDIR} ]; then echo "Cannot identify running kernel" exit 1 fi # Figure out what kernel configuration is running. We start with # the output of `uname -i`, and then make the following adjustments: # 1. Replace "SMP-GENERIC" with "SMP". Why the SMP kernel config # file says "ident SMP-GENERIC", I don't know... # 2. If the kernel claims to be GENERIC _and_ ${ARCH} is "amd64" # _and_ `sysctl kern.version` contains a line which ends "/SMP", then # we're running an SMP kernel. This mis-identification is a bug # which was fixed in 6.2-STABLE. KERNCONF=`uname -i` if [ ${KERNCONF} = "SMP-GENERIC" ]; then KERNCONF=SMP fi if [ ${KERNCONF} = "GENERIC" ] && [ ${ARCH} = "amd64" ]; then if sysctl kern.version | grep -qE '/SMP$'; then KERNCONF=SMP fi fi # Define some paths SHA256=/sbin/sha256 PHTTPGET=/usr/libexec/phttpget # Set up variables relating to VERBOSELEVEL fetch_setup_verboselevel } #### Core functionality -- the actual work gets done here # Use an SRV query to pick a server. If the SRV query doesn't provide # a useful answer, use the server name specified by the user. # Put another way... look up _http._tcp.${SERVERNAME} and pick a server # from that; or if no servers are returned, use ${SERVERNAME}. # This allows a user to specify "portsnap.freebsd.org" (in which case # portsnap will select one of the mirrors) or "portsnap5.tld.freebsd.org" # (in which case portsnap will use that particular server, since there # won't be an SRV entry for that name). # # We ignore the Port field, since we are always going to use port 80. # Fetch the mirror list, but do not pick a mirror yet. Returns 1 if # no mirrors are available for any reason. fetch_pick_server_init () { : > serverlist_tried # Check that host(1) exists (i.e., that the system wasn't built with the # WITHOUT_BIND set) and don't try to find a mirror if it doesn't exist. if ! which -s host; then : > serverlist_full return 1 fi echo -n "Looking up ${SERVERNAME} mirrors... " # Issue the SRV query and pull out the Priority, Weight, and Target fields. # BIND 9 prints "$name has SRV record ..." while BIND 8 prints # "$name server selection ..."; we allow either format. MLIST="_http._tcp.${SERVERNAME}" host -t srv "${MLIST}" | sed -nE "s/${MLIST} (has SRV record|server selection) //p" | cut -f 1,2,4 -d ' ' | sed -e 's/\.$//' | sort > serverlist_full # If no records, give up -- we'll just use the server name we were given. if [ `wc -l < serverlist_full` -eq 0 ]; then echo "none found." return 1 fi # Report how many mirrors we found. echo `wc -l < serverlist_full` "mirrors found." # Generate a random seed for use in picking mirrors. If HTTP_PROXY # is set, this will be used to generate the seed; otherwise, the seed # will be random. if [ -n "${HTTP_PROXY}${http_proxy}" ]; then RANDVALUE=`sha256 -qs "${HTTP_PROXY}${http_proxy}" | tr -d 'a-f' | cut -c 1-9` else RANDVALUE=`jot -r 1 0 999999999` fi } # Pick a mirror. Returns 1 if we have run out of mirrors to try. fetch_pick_server () { # Generate a list of not-yet-tried mirrors sort serverlist_tried | comm -23 serverlist_full - > serverlist # Have we run out of mirrors? if [ `wc -l < serverlist` -eq 0 ]; then echo "No mirrors remaining, giving up." return 1 fi # Find the highest priority level (lowest numeric value). SRV_PRIORITY=`cut -f 1 -d ' ' serverlist | sort -n | head -1` # Add up the weights of the response lines at that priority level. SRV_WSUM=0; while read X; do case "$X" in ${SRV_PRIORITY}\ *) SRV_W=`echo $X | cut -f 2 -d ' '` SRV_WSUM=$(($SRV_WSUM + $SRV_W)) ;; esac done < serverlist # If all the weights are 0, pretend that they are all 1 instead. if [ ${SRV_WSUM} -eq 0 ]; then SRV_WSUM=`grep -E "^${SRV_PRIORITY} " serverlist | wc -l` SRV_W_ADD=1 else SRV_W_ADD=0 fi # Pick a value between 0 and the sum of the weights - 1 SRV_RND=`expr ${RANDVALUE} % ${SRV_WSUM}` # Read through the list of mirrors and set SERVERNAME. Write the line # corresponding to the mirror we selected into serverlist_tried so that # we won't try it again. while read X; do case "$X" in ${SRV_PRIORITY}\ *) SRV_W=`echo $X | cut -f 2 -d ' '` SRV_W=$(($SRV_W + $SRV_W_ADD)) if [ $SRV_RND -lt $SRV_W ]; then SERVERNAME=`echo $X | cut -f 3 -d ' '` echo "$X" >> serverlist_tried break else SRV_RND=$(($SRV_RND - $SRV_W)) fi ;; esac done < serverlist } # Take a list of ${oldhash}|${newhash} and output a list of needed patches, # i.e., those for which we have ${oldhash} and don't have ${newhash}. fetch_make_patchlist () { grep -vE "^([0-9a-f]{64})\|\1$" | tr '|' ' ' | while read X Y; do if [ -f "files/${Y}.gz" ] || [ ! -f "files/${X}.gz" ]; then continue fi echo "${X}|${Y}" done | uniq } # Print user-friendly progress statistics fetch_progress () { LNC=0 while read x; do LNC=$(($LNC + 1)) if [ $(($LNC % 10)) = 0 ]; then echo -n $LNC elif [ $(($LNC % 2)) = 0 ]; then echo -n . fi done echo -n " " } # Function for asking the user if everything is ok continuep () { while read -p "Does this look reasonable (y/n)? " CONTINUE; do case "${CONTINUE}" in y*) return 0 ;; n*) return 1 ;; esac done } # Initialize the working directory workdir_init () { mkdir -p files touch tINDEX.present } # Check that we have a public key with an appropriate hash, or # fetch the key if it doesn't exist. Returns 1 if the key has # not yet been fetched. fetch_key () { if [ -r pub.ssl ] && [ `${SHA256} -q pub.ssl` = ${KEYPRINT} ]; then return 0 fi echo -n "Fetching public key from ${SERVERNAME}... " rm -f pub.ssl fetch ${QUIETFLAG} http://${SERVERNAME}/${FETCHDIR}/pub.ssl \ 2>${QUIETREDIR} || true if ! [ -r pub.ssl ]; then echo "failed." return 1 fi if ! [ `${SHA256} -q pub.ssl` = ${KEYPRINT} ]; then echo "key has incorrect hash." rm -f pub.ssl return 1 fi echo "done." } # Fetch metadata signature, aka "tag". fetch_tag () { echo -n "Fetching metadata signature " echo ${NDEBUG} "for ${RELNUM} from ${SERVERNAME}... " rm -f latest.ssl fetch ${QUIETFLAG} http://${SERVERNAME}/${FETCHDIR}/latest.ssl \ 2>${QUIETREDIR} || true if ! [ -r latest.ssl ]; then echo "failed." return 1 fi openssl rsautl -pubin -inkey pub.ssl -verify \ < latest.ssl > tag.new 2>${QUIETREDIR} || true rm latest.ssl if ! [ `wc -l < tag.new` = 1 ] || ! grep -qE \ "^freebsd-update\|${ARCH}\|${RELNUM}\|[0-9]+\|[0-9a-f]{64}\|[0-9]{10}" \ tag.new; then echo "invalid signature." return 1 fi echo "done." RELPATCHNUM=`cut -f 4 -d '|' < tag.new` TINDEXHASH=`cut -f 5 -d '|' < tag.new` EOLTIME=`cut -f 6 -d '|' < tag.new` } # Sanity-check the patch number in a tag, to make sure that we're not # going to "update" backwards and to prevent replay attacks. fetch_tagsanity () { # Check that we're not going to move from -pX to -pY with Y < X. RELPX=`uname -r | sed -E 's,.*-,,'` if echo ${RELPX} | grep -qE '^p[0-9]+$'; then RELPX=`echo ${RELPX} | cut -c 2-` else RELPX=0 fi if [ "${RELPATCHNUM}" -lt "${RELPX}" ]; then echo echo -n "Files on mirror (${RELNUM}-p${RELPATCHNUM})" echo " appear older than what" echo "we are currently running (`uname -r`)!" echo "Cowardly refusing to proceed any further." return 1 fi # If "tag" exists and corresponds to ${RELNUM}, make sure that # it contains a patch number <= RELPATCHNUM, in order to protect # against rollback (replay) attacks. if [ -f tag ] && grep -qE \ "^freebsd-update\|${ARCH}\|${RELNUM}\|[0-9]+\|[0-9a-f]{64}\|[0-9]{10}" \ tag; then LASTRELPATCHNUM=`cut -f 4 -d '|' < tag` if [ "${RELPATCHNUM}" -lt "${LASTRELPATCHNUM}" ]; then echo echo -n "Files on mirror (${RELNUM}-p${RELPATCHNUM})" echo " are older than the" echo -n "most recently seen updates" echo " (${RELNUM}-p${LASTRELPATCHNUM})." echo "Cowardly refusing to proceed any further." return 1 fi fi } # Fetch metadata index file fetch_metadata_index () { echo ${NDEBUG} "Fetching metadata index... " rm -f ${TINDEXHASH} fetch ${QUIETFLAG} http://${SERVERNAME}/${FETCHDIR}/t/${TINDEXHASH} 2>${QUIETREDIR} if ! [ -f ${TINDEXHASH} ]; then echo "failed." return 1 fi if [ `${SHA256} -q ${TINDEXHASH}` != ${TINDEXHASH} ]; then echo "update metadata index corrupt." return 1 fi echo "done." } # Print an error message about signed metadata being bogus. fetch_metadata_bogus () { echo echo "The update metadata$1 is correctly signed, but" echo "failed an integrity check." echo "Cowardly refusing to proceed any further." return 1 } # Construct tINDEX.new by merging the lines named in $1 from ${TINDEXHASH} # with the lines not named in $@ from tINDEX.present (if that file exists). fetch_metadata_index_merge () { for METAFILE in $@; do if [ `grep -E "^${METAFILE}\|" ${TINDEXHASH} | wc -l` \ -ne 1 ]; then fetch_metadata_bogus " index" return 1 fi grep -E "${METAFILE}\|" ${TINDEXHASH} done | sort > tINDEX.wanted if [ -f tINDEX.present ]; then join -t '|' -v 2 tINDEX.wanted tINDEX.present | sort -m - tINDEX.wanted > tINDEX.new rm tINDEX.wanted else mv tINDEX.wanted tINDEX.new fi } # Sanity check all the lines of tINDEX.new. Even if more metadata lines # are added by future versions of the server, this won't cause problems, # since the only lines which appear in tINDEX.new are the ones which we # specifically grepped out of ${TINDEXHASH}. fetch_metadata_index_sanity () { if grep -qvE '^[0-9A-Z.-]+\|[0-9a-f]{64}$' tINDEX.new; then fetch_metadata_bogus " index" return 1 fi } # Sanity check the metadata file $1. fetch_metadata_sanity () { # Some aliases to save space later: ${P} is a character which can # appear in a path; ${M} is the four numeric metadata fields; and # ${H} is a sha256 hash. P="[-+./:=%@_[~[:alnum:]]" M="[0-9]+\|[0-9]+\|[0-9]+\|[0-9]+" H="[0-9a-f]{64}" # Check that the first four fields make sense. if gunzip -c < files/$1.gz | grep -qvE "^[a-z]+\|[0-9a-z]+\|${P}+\|[fdL-]\|"; then fetch_metadata_bogus "" return 1 fi # Remove the first three fields. gunzip -c < files/$1.gz | cut -f 4- -d '|' > sanitycheck.tmp # Sanity check entries with type 'f' if grep -E '^f' sanitycheck.tmp | grep -qvE "^f\|${M}\|${H}\|${P}*\$"; then fetch_metadata_bogus "" return 1 fi # Sanity check entries with type 'd' if grep -E '^d' sanitycheck.tmp | grep -qvE "^d\|${M}\|\|\$"; then fetch_metadata_bogus "" return 1 fi # Sanity check entries with type 'L' if grep -E '^L' sanitycheck.tmp | grep -qvE "^L\|${M}\|${P}*\|\$"; then fetch_metadata_bogus "" return 1 fi # Sanity check entries with type '-' if grep -E '^-' sanitycheck.tmp | grep -qvE "^-\|\|\|\|\|\|"; then fetch_metadata_bogus "" return 1 fi # Clean up rm sanitycheck.tmp } # Fetch the metadata index and metadata files listed in $@, # taking advantage of metadata patches where possible. fetch_metadata () { fetch_metadata_index || return 1 fetch_metadata_index_merge $@ || return 1 fetch_metadata_index_sanity || return 1 # Generate a list of wanted metadata patches join -t '|' -o 1.2,2.2 tINDEX.present tINDEX.new | fetch_make_patchlist > patchlist if [ -s patchlist ]; then # Attempt to fetch metadata patches echo -n "Fetching `wc -l < patchlist | tr -d ' '` " echo ${NDEBUG} "metadata patches.${DDSTATS}" tr '|' '-' < patchlist | lam -s "${FETCHDIR}/tp/" - -s ".gz" | xargs ${XARGST} ${PHTTPGET} ${SERVERNAME} \ 2>${STATSREDIR} | fetch_progress echo "done." # Attempt to apply metadata patches echo -n "Applying metadata patches... " tr '|' ' ' < patchlist | while read X Y; do if [ ! -f "${X}-${Y}.gz" ]; then continue; fi gunzip -c < ${X}-${Y}.gz > diff gunzip -c < files/${X}.gz > diff-OLD # Figure out which lines are being added and removed grep -E '^-' diff | cut -c 2- | while read PREFIX; do look "${PREFIX}" diff-OLD done | sort > diff-rm grep -E '^\+' diff | cut -c 2- > diff-add # Generate the new file comm -23 diff-OLD diff-rm | sort - diff-add > diff-NEW if [ `${SHA256} -q diff-NEW` = ${Y} ]; then mv diff-NEW files/${Y} gzip -n files/${Y} else mv diff-NEW ${Y}.bad fi rm -f ${X}-${Y}.gz diff rm -f diff-OLD diff-NEW diff-add diff-rm done 2>${QUIETREDIR} echo "done." fi # Update metadata without patches cut -f 2 -d '|' < tINDEX.new | while read Y; do if [ ! -f "files/${Y}.gz" ]; then echo ${Y}; fi done | sort -u > filelist if [ -s filelist ]; then echo -n "Fetching `wc -l < filelist | tr -d ' '` " echo ${NDEBUG} "metadata files... " lam -s "${FETCHDIR}/m/" - -s ".gz" < filelist | xargs ${XARGST} ${PHTTPGET} ${SERVERNAME} \ 2>${QUIETREDIR} while read Y; do if ! [ -f ${Y}.gz ]; then echo "failed." return 1 fi if [ `gunzip -c < ${Y}.gz | ${SHA256} -q` = ${Y} ]; then mv ${Y}.gz files/${Y}.gz else echo "metadata is corrupt." return 1 fi done < filelist echo "done." fi # Sanity-check the metadata files. cut -f 2 -d '|' tINDEX.new > filelist while read X; do fetch_metadata_sanity ${X} || return 1 done < filelist # Remove files which are no longer needed cut -f 2 -d '|' tINDEX.present | sort > oldfiles cut -f 2 -d '|' tINDEX.new | sort | comm -13 - oldfiles | lam -s "files/" - -s ".gz" | xargs rm -f rm patchlist filelist oldfiles rm ${TINDEXHASH} # We're done! mv tINDEX.new tINDEX.present mv tag.new tag return 0 } # Extract a subset of a downloaded metadata file containing only the parts # which are listed in COMPONENTS. fetch_filter_metadata_components () { METAHASH=`look "$1|" tINDEX.present | cut -f 2 -d '|'` gunzip -c < files/${METAHASH}.gz > $1.all # Fish out the lines belonging to components we care about. for C in ${COMPONENTS}; do look "`echo ${C} | tr '/' '|'`|" $1.all done > $1 # Remove temporary file. rm $1.all } # Generate a filtered version of the metadata file $1 from the downloaded # file, by fishing out the lines corresponding to components we're trying # to keep updated, and then removing lines corresponding to paths we want # to ignore. fetch_filter_metadata () { # Fish out the lines belonging to components we care about. fetch_filter_metadata_components $1 # Canonicalize directory names by removing any trailing / in # order to avoid listing directories multiple times if they # belong to multiple components. Turning "/" into "" doesn't # matter, since we add a leading "/" when we use paths later. cut -f 3- -d '|' $1 | sed -e 's,/|d|,|d|,' | sed -e 's,/|-|,|-|,' | sort -u > $1.tmp # Figure out which lines to ignore and remove them. for X in ${IGNOREPATHS}; do grep -E "^${X}" $1.tmp done | sort -u | comm -13 - $1.tmp > $1 # Remove temporary files. rm $1.tmp } # Filter the metadata file $1 by adding lines with "/boot/$2" # replaced by ${KERNELDIR} (which is `sysctl -n kern.bootfile` minus the # trailing "/kernel"); and if "/boot/$2" does not exist, remove # the original lines which start with that. # Put another way: Deal with the fact that the FOO kernel is sometimes # installed in /boot/FOO/ and is sometimes installed elsewhere. fetch_filter_kernel_names () { grep ^/boot/$2 $1 | sed -e "s,/boot/$2,${KERNELDIR},g" | sort - $1 > $1.tmp mv $1.tmp $1 if ! [ -d /boot/$2 ]; then grep -v ^/boot/$2 $1 > $1.tmp mv $1.tmp $1 fi } # For all paths appearing in $1 or $3, inspect the system # and generate $2 describing what is currently installed. fetch_inspect_system () { # No errors yet... rm -f .err # Tell the user why his disk is suddenly making lots of noise echo -n "Inspecting system... " # Generate list of files to inspect cat $1 $3 | cut -f 1 -d '|' | sort -u > filelist # Examine each file and output lines of the form # /path/to/file|type|device-inum|user|group|perm|flags|value # sorted by device and inode number. while read F; do # If the symlink/file/directory does not exist, record this. if ! [ -e ${BASEDIR}/${F} ]; then echo "${F}|-||||||" continue fi if ! [ -r ${BASEDIR}/${F} ]; then echo "Cannot read file: ${BASEDIR}/${F}" \ >/dev/stderr touch .err return 1 fi # Otherwise, output an index line. if [ -L ${BASEDIR}/${F} ]; then echo -n "${F}|L|" stat -n -f '%d-%i|%u|%g|%Mp%Lp|%Of|' ${BASEDIR}/${F}; readlink ${BASEDIR}/${F}; elif [ -f ${BASEDIR}/${F} ]; then echo -n "${F}|f|" stat -n -f '%d-%i|%u|%g|%Mp%Lp|%Of|' ${BASEDIR}/${F}; sha256 -q ${BASEDIR}/${F}; elif [ -d ${BASEDIR}/${F} ]; then echo -n "${F}|d|" stat -f '%d-%i|%u|%g|%Mp%Lp|%Of|' ${BASEDIR}/${F}; else echo "Unknown file type: ${BASEDIR}/${F}" \ >/dev/stderr touch .err return 1 fi done < filelist | sort -k 3,3 -t '|' > $2.tmp rm filelist # Check if an error occurred during system inspection if [ -f .err ]; then return 1 fi # Convert to the form # /path/to/file|type|user|group|perm|flags|value|hlink # by resolving identical device and inode numbers into hard links. cut -f 1,3 -d '|' $2.tmp | sort -k 1,1 -t '|' | sort -s -u -k 2,2 -t '|' | join -1 2 -2 3 -t '|' - $2.tmp | awk -F \| -v OFS=\| \ '{ if (($2 == $3) || ($4 == "-")) print $3,$4,$5,$6,$7,$8,$9,"" else print $3,$4,$5,$6,$7,$8,$9,$2 }' | sort > $2 rm $2.tmp # We're finished looking around echo "done." } # For any paths matching ${MERGECHANGES}, compare $1 and $2 and find any # files which differ; generate $3 containing these paths and the old hashes. fetch_filter_mergechanges () { # Pull out the paths and hashes of the files matching ${MERGECHANGES}. for F in $1 $2; do for X in ${MERGECHANGES}; do grep -E "^${X}" ${F} done | cut -f 1,2,7 -d '|' | sort > ${F}-values done # Any line in $2-values which doesn't appear in $1-values and is a # file means that we should list the path in $3. comm -13 $1-values $2-values | fgrep '|f|' | cut -f 1 -d '|' > $2-paths # For each path, pull out one (and only one!) entry from $1-values. # Note that we cannot distinguish which "old" version the user made # changes to; but hopefully any changes which occur due to security # updates will exist in both the "new" version and the version which # the user has installed, so the merging will still work. while read X; do look "${X}|" $1-values | head -1 done < $2-paths > $3 # Clean up rm $1-values $2-values $2-paths } # For any paths matching ${UPDATEIFUNMODIFIED}, remove lines from $[123] # which correspond to lines in $2 with hashes not matching $1 or $3, unless # the paths are listed in $4. For entries in $2 marked "not present" # (aka. type -), remove lines from $[123] unless there is a corresponding # entry in $1. fetch_filter_unmodified_notpresent () { # Figure out which lines of $1 and $3 correspond to bits which # should only be updated if they haven't changed, and fish out # the (path, type, value) tuples. # NOTE: We don't consider a file to be "modified" if it matches # the hash from $3. for X in ${UPDATEIFUNMODIFIED}; do grep -E "^${X}" $1 grep -E "^${X}" $3 done | cut -f 1,2,7 -d '|' | sort > $1-values # Do the same for $2. for X in ${UPDATEIFUNMODIFIED}; do grep -E "^${X}" $2 done | cut -f 1,2,7 -d '|' | sort > $2-values # Any entry in $2-values which is not in $1-values corresponds to # a path which we need to remove from $1, $2, and $3, unless it # that path appears in $4. comm -13 $1-values $2-values | sort -t '|' -k 1,1 > mlines.tmp cut -f 1 -d '|' $4 | sort | join -v 2 -t '|' - mlines.tmp | sort > mlines rm $1-values $2-values mlines.tmp # Any lines in $2 which are not in $1 AND are "not present" lines # also belong in mlines. comm -13 $1 $2 | cut -f 1,2,7 -d '|' | fgrep '|-|' >> mlines # Remove lines from $1, $2, and $3 for X in $1 $2 $3; do sort -t '|' -k 1,1 ${X} > ${X}.tmp cut -f 1 -d '|' < mlines | sort | join -v 2 -t '|' - ${X}.tmp | sort > ${X} rm ${X}.tmp done # Store a list of the modified files, for future reference fgrep -v '|-|' mlines | cut -f 1 -d '|' > modifiedfiles rm mlines } # For each entry in $1 of type -, remove any corresponding # entry from $2 if ${ALLOWADD} != "yes". Remove all entries # of type - from $1. fetch_filter_allowadd () { cut -f 1,2 -d '|' < $1 | fgrep '|-' | cut -f 1 -d '|' > filesnotpresent if [ ${ALLOWADD} != "yes" ]; then sort < $2 | join -v 1 -t '|' - filesnotpresent | sort > $2.tmp mv $2.tmp $2 fi sort < $1 | join -v 1 -t '|' - filesnotpresent | sort > $1.tmp mv $1.tmp $1 rm filesnotpresent } # If ${ALLOWDELETE} != "yes", then remove any entries from $1 # which don't correspond to entries in $2. fetch_filter_allowdelete () { # Produce a lists ${PATH}|${TYPE} for X in $1 $2; do cut -f 1-2 -d '|' < ${X} | sort -u > ${X}.nodes done # Figure out which lines need to be removed from $1. if [ ${ALLOWDELETE} != "yes" ]; then comm -23 $1.nodes $2.nodes > $1.badnodes else : > $1.badnodes fi # Remove the relevant lines from $1 while read X; do look "${X}|" $1 done < $1.badnodes | comm -13 - $1 > $1.tmp mv $1.tmp $1 rm $1.badnodes $1.nodes $2.nodes } # If ${KEEPMODIFIEDMETADATA} == "yes", then for each entry in $2 # with metadata not matching any entry in $1, replace the corresponding # line of $3 with one having the same metadata as the entry in $2. fetch_filter_modified_metadata () { # Fish out the metadata from $1 and $2 for X in $1 $2; do cut -f 1-6 -d '|' < ${X} > ${X}.metadata done # Find the metadata we need to keep if [ ${KEEPMODIFIEDMETADATA} = "yes" ]; then comm -13 $1.metadata $2.metadata > keepmeta else : > keepmeta fi # Extract the lines which we need to remove from $3, and # construct the lines which we need to add to $3. : > $3.remove : > $3.add while read LINE; do NODE=`echo "${LINE}" | cut -f 1-2 -d '|'` look "${NODE}|" $3 >> $3.remove look "${NODE}|" $3 | cut -f 7- -d '|' | lam -s "${LINE}|" - >> $3.add done < keepmeta # Remove the specified lines and add the new lines. sort $3.remove | comm -13 - $3 | sort -u - $3.add > $3.tmp mv $3.tmp $3 rm keepmeta $1.metadata $2.metadata $3.add $3.remove } # Remove lines from $1 and $2 which are identical; # no need to update a file if it isn't changing. fetch_filter_uptodate () { comm -23 $1 $2 > $1.tmp comm -13 $1 $2 > $2.tmp mv $1.tmp $1 mv $2.tmp $2 } # Fetch any "clean" old versions of files we need for merging changes. fetch_files_premerge () { # We only need to do anything if $1 is non-empty. if [ -s $1 ]; then # Tell the user what we're doing echo -n "Fetching files from ${OLDRELNUM} for merging... " # List of files wanted fgrep '|f|' < $1 | cut -f 3 -d '|' | sort -u > files.wanted # Only fetch the files we don't already have while read Y; do if [ ! -f "files/${Y}.gz" ]; then echo ${Y}; fi done < files.wanted > filelist # Actually fetch them lam -s "${OLDFETCHDIR}/f/" - -s ".gz" < filelist | xargs ${XARGST} ${PHTTPGET} ${SERVERNAME} \ 2>${QUIETREDIR} # Make sure we got them all, and move them into /files/ while read Y; do if ! [ -f ${Y}.gz ]; then echo "failed." return 1 fi if [ `gunzip -c < ${Y}.gz | ${SHA256} -q` = ${Y} ]; then mv ${Y}.gz files/${Y}.gz else echo "${Y} has incorrect hash." return 1 fi done < filelist echo "done." # Clean up rm filelist files.wanted fi } # Prepare to fetch files: Generate a list of the files we need, # copy the unmodified files we have into /files/, and generate # a list of patches to download. fetch_files_prepare () { # Tell the user why his disk is suddenly making lots of noise echo -n "Preparing to download files... " # Reduce indices to ${PATH}|${HASH} pairs for X in $1 $2 $3; do cut -f 1,2,7 -d '|' < ${X} | fgrep '|f|' | cut -f 1,3 -d '|' | sort > ${X}.hashes done # List of files wanted cut -f 2 -d '|' < $3.hashes | sort -u | while read HASH; do if ! [ -f files/${HASH}.gz ]; then echo ${HASH} fi done > files.wanted # Generate a list of unmodified files comm -12 $1.hashes $2.hashes | sort -k 1,1 -t '|' > unmodified.files # Copy all files into /files/. We only need the unmodified files # for use in patching; but we'll want all of them if the user asks # to rollback the updates later. while read LINE; do F=`echo "${LINE}" | cut -f 1 -d '|'` HASH=`echo "${LINE}" | cut -f 2 -d '|'` # Skip files we already have. if [ -f files/${HASH}.gz ]; then continue fi # Make sure the file hasn't changed. cp "${BASEDIR}/${F}" tmpfile if [ `sha256 -q tmpfile` != ${HASH} ]; then echo echo "File changed while FreeBSD Update running: ${F}" return 1 fi # Place the file into storage. gzip -c < tmpfile > files/${HASH}.gz rm tmpfile done < $2.hashes # Produce a list of patches to download sort -k 1,1 -t '|' $3.hashes | join -t '|' -o 2.2,1.2 - unmodified.files | fetch_make_patchlist > patchlist # Garbage collect rm unmodified.files $1.hashes $2.hashes $3.hashes # We don't need the list of possible old files any more. rm $1 # We're finished making noise echo "done." } # Fetch files. fetch_files () { # Attempt to fetch patches if [ -s patchlist ]; then echo -n "Fetching `wc -l < patchlist | tr -d ' '` " echo ${NDEBUG} "patches.${DDSTATS}" tr '|' '-' < patchlist | lam -s "${PATCHDIR}/" - | xargs ${XARGST} ${PHTTPGET} ${SERVERNAME} \ 2>${STATSREDIR} | fetch_progress echo "done." # Attempt to apply patches echo -n "Applying patches... " tr '|' ' ' < patchlist | while read X Y; do if [ ! -f "${X}-${Y}" ]; then continue; fi gunzip -c < files/${X}.gz > OLD bspatch OLD NEW ${X}-${Y} if [ `${SHA256} -q NEW` = ${Y} ]; then mv NEW files/${Y} gzip -n files/${Y} fi rm -f diff OLD NEW ${X}-${Y} done 2>${QUIETREDIR} echo "done." fi # Download files which couldn't be generate via patching while read Y; do if [ ! -f "files/${Y}.gz" ]; then echo ${Y}; fi done < files.wanted > filelist if [ -s filelist ]; then echo -n "Fetching `wc -l < filelist | tr -d ' '` " echo ${NDEBUG} "files... " lam -s "${FETCHDIR}/f/" - -s ".gz" < filelist | xargs ${XARGST} ${PHTTPGET} ${SERVERNAME} \ 2>${QUIETREDIR} while read Y; do if ! [ -f ${Y}.gz ]; then echo "failed." return 1 fi if [ `gunzip -c < ${Y}.gz | ${SHA256} -q` = ${Y} ]; then mv ${Y}.gz files/${Y}.gz else echo "${Y} has incorrect hash." return 1 fi done < filelist echo "done." fi # Clean up rm files.wanted filelist patchlist } # Create and populate install manifest directory; and report what updates # are available. fetch_create_manifest () { # If we have an existing install manifest, nuke it. if [ -L "${BDHASH}-install" ]; then rm -r ${BDHASH}-install/ rm ${BDHASH}-install fi # Report to the user if any updates were avoided due to local changes if [ -s modifiedfiles ]; then echo echo -n "The following files are affected by updates, " echo "but no changes have" echo -n "been downloaded because the files have been " echo "modified locally:" cat modifiedfiles fi | $PAGER rm modifiedfiles # If no files will be updated, tell the user and exit if ! [ -s INDEX-PRESENT ] && ! [ -s INDEX-NEW ]; then rm INDEX-PRESENT INDEX-NEW echo echo -n "No updates needed to update system to " echo "${RELNUM}-p${RELPATCHNUM}." return fi # Divide files into (a) removed files, (b) added files, and # (c) updated files. cut -f 1 -d '|' < INDEX-PRESENT | sort > INDEX-PRESENT.flist cut -f 1 -d '|' < INDEX-NEW | sort > INDEX-NEW.flist comm -23 INDEX-PRESENT.flist INDEX-NEW.flist > files.removed comm -13 INDEX-PRESENT.flist INDEX-NEW.flist > files.added comm -12 INDEX-PRESENT.flist INDEX-NEW.flist > files.updated rm INDEX-PRESENT.flist INDEX-NEW.flist # Report removed files, if any if [ -s files.removed ]; then echo echo -n "The following files will be removed " echo "as part of updating to ${RELNUM}-p${RELPATCHNUM}:" cat files.removed fi | $PAGER rm files.removed # Report added files, if any if [ -s files.added ]; then echo echo -n "The following files will be added " echo "as part of updating to ${RELNUM}-p${RELPATCHNUM}:" cat files.added fi | $PAGER rm files.added # Report updated files, if any if [ -s files.updated ]; then echo echo -n "The following files will be updated " echo "as part of updating to ${RELNUM}-p${RELPATCHNUM}:" cat files.updated fi | $PAGER rm files.updated # Create a directory for the install manifest. MDIR=`mktemp -d install.XXXXXX` || return 1 # Populate it mv INDEX-PRESENT ${MDIR}/INDEX-OLD mv INDEX-NEW ${MDIR}/INDEX-NEW # Link it into place ln -s ${MDIR} ${BDHASH}-install } # Warn about any upcoming EoL fetch_warn_eol () { # What's the current time? NOWTIME=`date "+%s"` # When did we last warn about the EoL date? if [ -f lasteolwarn ]; then LASTWARN=`cat lasteolwarn` else LASTWARN=`expr ${NOWTIME} - 63072000` fi # If the EoL time is past, warn. if [ ${EOLTIME} -lt ${NOWTIME} ]; then echo cat <<-EOF WARNING: `uname -sr` HAS PASSED ITS END-OF-LIFE DATE. Any security issues discovered after `date -r ${EOLTIME}` will not have been corrected. EOF return 1 fi # Figure out how long it has been since we last warned about the # upcoming EoL, and how much longer we have left. SINCEWARN=`expr ${NOWTIME} - ${LASTWARN}` TIMELEFT=`expr ${EOLTIME} - ${NOWTIME}` # Don't warn if the EoL is more than 3 months away if [ ${TIMELEFT} -gt 7884000 ]; then return 0 fi # Don't warn if the time remaining is more than 3 times the time # since the last warning. if [ ${TIMELEFT} -gt `expr ${SINCEWARN} \* 3` ]; then return 0 fi # Figure out what time units to use. if [ ${TIMELEFT} -lt 604800 ]; then UNIT="day" SIZE=86400 elif [ ${TIMELEFT} -lt 2678400 ]; then UNIT="week" SIZE=604800 else UNIT="month" SIZE=2678400 fi # Compute the right number of units NUM=`expr ${TIMELEFT} / ${SIZE}` if [ ${NUM} != 1 ]; then UNIT="${UNIT}s" fi # Print the warning echo cat <<-EOF WARNING: `uname -sr` is approaching its End-of-Life date. It is strongly recommended that you upgrade to a newer release within the next ${NUM} ${UNIT}. EOF # Update the stored time of last warning echo ${NOWTIME} > lasteolwarn } # Do the actual work involved in "fetch" / "cron". fetch_run () { workdir_init || return 1 # Prepare the mirror list. fetch_pick_server_init && fetch_pick_server # Try to fetch the public key until we run out of servers. while ! fetch_key; do fetch_pick_server || return 1 done # Try to fetch the metadata index signature ("tag") until we run # out of available servers; and sanity check the downloaded tag. while ! fetch_tag; do fetch_pick_server || return 1 done fetch_tagsanity || return 1 # Fetch the latest INDEX-NEW and INDEX-OLD files. fetch_metadata INDEX-NEW INDEX-OLD || return 1 # Generate filtered INDEX-NEW and INDEX-OLD files containing only # the lines which (a) belong to components we care about, and (b) # don't correspond to paths we're explicitly ignoring. fetch_filter_metadata INDEX-NEW || return 1 fetch_filter_metadata INDEX-OLD || return 1 # Translate /boot/${KERNCONF} into ${KERNELDIR} fetch_filter_kernel_names INDEX-NEW ${KERNCONF} fetch_filter_kernel_names INDEX-OLD ${KERNCONF} # For all paths appearing in INDEX-OLD or INDEX-NEW, inspect the # system and generate an INDEX-PRESENT file. fetch_inspect_system INDEX-OLD INDEX-PRESENT INDEX-NEW || return 1 # Based on ${UPDATEIFUNMODIFIED}, remove lines from INDEX-* which # correspond to lines in INDEX-PRESENT with hashes not appearing # in INDEX-OLD or INDEX-NEW. Also remove lines where the entry in # INDEX-PRESENT has type - and there isn't a corresponding entry in # INDEX-OLD with type -. fetch_filter_unmodified_notpresent \ INDEX-OLD INDEX-PRESENT INDEX-NEW /dev/null # For each entry in INDEX-PRESENT of type -, remove any corresponding # entry from INDEX-NEW if ${ALLOWADD} != "yes". Remove all entries # of type - from INDEX-PRESENT. fetch_filter_allowadd INDEX-PRESENT INDEX-NEW # If ${ALLOWDELETE} != "yes", then remove any entries from # INDEX-PRESENT which don't correspond to entries in INDEX-NEW. fetch_filter_allowdelete INDEX-PRESENT INDEX-NEW # If ${KEEPMODIFIEDMETADATA} == "yes", then for each entry in # INDEX-PRESENT with metadata not matching any entry in INDEX-OLD, # replace the corresponding line of INDEX-NEW with one having the # same metadata as the entry in INDEX-PRESENT. fetch_filter_modified_metadata INDEX-OLD INDEX-PRESENT INDEX-NEW # Remove lines from INDEX-PRESENT and INDEX-NEW which are identical; # no need to update a file if it isn't changing. fetch_filter_uptodate INDEX-PRESENT INDEX-NEW # Prepare to fetch files: Generate a list of the files we need, # copy the unmodified files we have into /files/, and generate # a list of patches to download. fetch_files_prepare INDEX-OLD INDEX-PRESENT INDEX-NEW || return 1 # Fetch files. fetch_files || return 1 # Create and populate install manifest directory; and report what # updates are available. fetch_create_manifest || return 1 # Warn about any upcoming EoL fetch_warn_eol || return 1 } # If StrictComponents is not "yes", generate a new components list # with only the components which appear to be installed. upgrade_guess_components () { if [ "${STRICTCOMPONENTS}" = "no" ]; then # Generate filtered INDEX-ALL with only the components listed # in COMPONENTS. fetch_filter_metadata_components $1 || return 1 # Tell the user why his disk is suddenly making lots of noise echo -n "Inspecting system... " # Look at the files on disk, and assume that a component is # supposed to be present if it is more than half-present. cut -f 1-3 -d '|' < INDEX-ALL | tr '|' ' ' | while read C S F; do if [ -e ${BASEDIR}/${F} ]; then echo "+ ${C}|${S}" fi echo "= ${C}|${S}" done | sort | uniq -c | sed -E 's,^ +,,' > compfreq grep ' = ' compfreq | cut -f 1,3 -d ' ' | sort -k 2,2 -t ' ' > compfreq.total grep ' + ' compfreq | cut -f 1,3 -d ' ' | sort -k 2,2 -t ' ' > compfreq.present join -t ' ' -1 2 -2 2 compfreq.present compfreq.total | while read S P T; do if [ ${P} -gt `expr ${T} / 2` ]; then echo ${S} fi done > comp.present cut -f 2 -d ' ' < compfreq.total > comp.total rm INDEX-ALL compfreq compfreq.total compfreq.present # We're done making noise. echo "done." # Sometimes the kernel isn't installed where INDEX-ALL # thinks that it should be: In particular, it is often in # /boot/kernel instead of /boot/GENERIC or /boot/SMP. To # deal with this, if "kernel|X" is listed in comp.total # (i.e., is a component which would be upgraded if it is # found to be present) we will add it to comp.present. # If "kernel|" is in comp.total but "kernel|X" is # not, we print a warning -- the user is running a kernel # which isn't part of the release. KCOMP=`echo ${KERNCONF} | tr 'A-Z' 'a-z'` grep -E "^kernel\|${KCOMP}\$" comp.total >> comp.present if grep -qE "^kernel\|" comp.total && ! grep -qE "^kernel\|${KCOMP}\$" comp.total; then cat <<-EOF WARNING: This system is running a "${KCOMP}" kernel, which is not a kernel configuration distributed as part of FreeBSD ${RELNUM}. This kernel will not be updated: you MUST update the kernel manually before running "$0 install". EOF fi # Re-sort the list of installed components and generate # the list of non-installed components. sort -u < comp.present > comp.present.tmp mv comp.present.tmp comp.present comm -13 comp.present comp.total > comp.absent # Ask the user to confirm that what we have is correct. To # reduce user confusion, translate "X|Y" back to "X/Y" (as # subcomponents must be listed in the configuration file). echo echo -n "The following components of FreeBSD " echo "seem to be installed:" tr '|' '/' < comp.present | fmt -72 echo echo -n "The following components of FreeBSD " echo "do not seem to be installed:" tr '|' '/' < comp.absent | fmt -72 echo continuep || return 1 echo # Suck the generated list of components into ${COMPONENTS}. # Note that comp.present.tmp is used due to issues with # pipelines and setting variables. COMPONENTS="" tr '|' '/' < comp.present > comp.present.tmp while read C; do COMPONENTS="${COMPONENTS} ${C}" done < comp.present.tmp # Delete temporary files rm comp.present comp.present.tmp comp.absent comp.total fi } # If StrictComponents is not "yes", COMPONENTS contains an entry # corresponding to the currently running kernel, and said kernel # does not exist in the new release, add "kernel/generic" to the # list of components. upgrade_guess_new_kernel () { if [ "${STRICTCOMPONENTS}" = "no" ]; then # Grab the unfiltered metadata file. METAHASH=`look "$1|" tINDEX.present | cut -f 2 -d '|'` gunzip -c < files/${METAHASH}.gz > $1.all # If "kernel/${KCOMP}" is in ${COMPONENTS} and that component # isn't in $1.all, we need to add kernel/generic. for C in ${COMPONENTS}; do if [ ${C} = "kernel/${KCOMP}" ] && ! grep -qE "^kernel\|${KCOMP}\|" $1.all; then COMPONENTS="${COMPONENTS} kernel/generic" NKERNCONF="GENERIC" cat <<-EOF WARNING: This system is running a "${KCOMP}" kernel, which is not a kernel configuration distributed as part of FreeBSD ${RELNUM}. As part of upgrading to FreeBSD ${RELNUM}, this kernel will be replaced with a "generic" kernel. EOF continuep || return 1 fi done # Don't need this any more... rm $1.all fi } # Convert INDEX-OLD (last release) and INDEX-ALL (new release) into # INDEX-OLD and INDEX-NEW files (in the sense of normal upgrades). upgrade_oldall_to_oldnew () { # For each ${F}|... which appears in INDEX-ALL but does not appear # in INDEX-OLD, add ${F}|-|||||| to INDEX-OLD. cut -f 1 -d '|' < $1 | sort -u > $1.paths cut -f 1 -d '|' < $2 | sort -u | comm -13 $1.paths - | lam - -s "|-||||||" | sort - $1 > $1.tmp mv $1.tmp $1 # Remove lines from INDEX-OLD which also appear in INDEX-ALL comm -23 $1 $2 > $1.tmp mv $1.tmp $1 # Remove lines from INDEX-ALL which have a file name not appearing # anywhere in INDEX-OLD (since these must be files which haven't # changed -- if they were new, there would be an entry of type "-"). cut -f 1 -d '|' < $1 | sort -u > $1.paths sort -k 1,1 -t '|' < $2 | join -t '|' - $1.paths | sort > $2.tmp rm $1.paths mv $2.tmp $2 # Rename INDEX-ALL to INDEX-NEW. mv $2 $3 } # Helper for upgrade_merge: Return zero true iff the two files differ only # in the contents of their RCS tags. samef () { X=`sed -E 's/\\$FreeBSD.*\\$/\$FreeBSD\$/' < $1 | ${SHA256}` Y=`sed -E 's/\\$FreeBSD.*\\$/\$FreeBSD\$/' < $2 | ${SHA256}` if [ $X = $Y ]; then return 0; else return 1; fi } # From the list of "old" files in $1, merge changes in $2 with those in $3, # and update $3 to reflect the hashes of merged files. upgrade_merge () { # We only need to do anything if $1 is non-empty. if [ -s $1 ]; then cut -f 1 -d '|' $1 | sort > $1-paths # Create staging area for merging files rm -rf merge/ while read F; do D=`dirname ${F}` mkdir -p merge/old/${D} mkdir -p merge/${OLDRELNUM}/${D} mkdir -p merge/${RELNUM}/${D} mkdir -p merge/new/${D} done < $1-paths # Copy in files while read F; do # Currently installed file V=`look "${F}|" $2 | cut -f 7 -d '|'` gunzip < files/${V}.gz > merge/old/${F} # Old release if look "${F}|" $1 | fgrep -q "|f|"; then V=`look "${F}|" $1 | cut -f 3 -d '|'` gunzip < files/${V}.gz \ > merge/${OLDRELNUM}/${F} fi # New release if look "${F}|" $3 | cut -f 1,2,7 -d '|' | fgrep -q "|f|"; then V=`look "${F}|" $3 | cut -f 7 -d '|'` gunzip < files/${V}.gz \ > merge/${RELNUM}/${F} fi done < $1-paths # Attempt to automatically merge changes echo -n "Attempting to automatically merge " echo -n "changes in files..." : > failed.merges while read F; do # If the file doesn't exist in the new release, # the result of "merging changes" is having the file # not exist. if ! [ -f merge/${RELNUM}/${F} ]; then continue fi # If the file didn't exist in the old release, we're # going to throw away the existing file and hope that # the version from the new release is what we want. if ! [ -f merge/${OLDRELNUM}/${F} ]; then cp merge/${RELNUM}/${F} merge/new/${F} continue fi # Some files need special treatment. case ${F} in /etc/spwd.db | /etc/pwd.db | /etc/login.conf.db) # Don't merge these -- we're rebuild them # after updates are installed. cp merge/old/${F} merge/new/${F} ;; *) if ! merge -p -L "current version" \ -L "${OLDRELNUM}" -L "${RELNUM}" \ merge/old/${F} \ merge/${OLDRELNUM}/${F} \ merge/${RELNUM}/${F} \ > merge/new/${F} 2>/dev/null; then echo ${F} >> failed.merges fi ;; esac done < $1-paths echo " done." # Ask the user to handle any files which didn't merge. while read F; do # If the installed file differs from the version in # the old release only due to RCS tag expansion # then just use the version in the new release. if samef merge/old/${F} merge/${OLDRELNUM}/${F}; then cp merge/${RELNUM}/${F} merge/new/${F} continue fi cat <<-EOF The following file could not be merged automatically: ${F} Press Enter to edit this file in ${EDITOR} and resolve the conflicts manually... EOF read dummy files/${V}.gz echo "${F}|${V}" fi done < $1-paths > newhashes # Pull lines out from $3 which need to be updated to # reflect merged files. while read F; do look "${F}|" $3 done < $1-paths > $3-oldlines # Update lines to reflect merged files join -t '|' -o 1.1,1.2,1.3,1.4,1.5,1.6,2.2,1.8 \ $3-oldlines newhashes > $3-newlines # Remove old lines from $3 and add new lines. sort $3-oldlines | comm -13 - $3 | sort - $3-newlines > $3.tmp mv $3.tmp $3 # Clean up rm $1-paths newhashes $3-oldlines $3-newlines rm -rf merge/ fi # We're done with merging files. rm $1 } # Do the work involved in fetching upgrades to a new release upgrade_run () { workdir_init || return 1 # Prepare the mirror list. fetch_pick_server_init && fetch_pick_server # Try to fetch the public key until we run out of servers. while ! fetch_key; do fetch_pick_server || return 1 done # Try to fetch the metadata index signature ("tag") until we run # out of available servers; and sanity check the downloaded tag. while ! fetch_tag; do fetch_pick_server || return 1 done fetch_tagsanity || return 1 # Fetch the INDEX-OLD and INDEX-ALL. fetch_metadata INDEX-OLD INDEX-ALL || return 1 # If StrictComponents is not "yes", generate a new components list # with only the components which appear to be installed. upgrade_guess_components INDEX-ALL || return 1 # Generate filtered INDEX-OLD and INDEX-ALL files containing only # the components we want and without anything marked as "Ignore". fetch_filter_metadata INDEX-OLD || return 1 fetch_filter_metadata INDEX-ALL || return 1 # Merge the INDEX-OLD and INDEX-ALL files into INDEX-OLD. sort INDEX-OLD INDEX-ALL > INDEX-OLD.tmp mv INDEX-OLD.tmp INDEX-OLD rm INDEX-ALL # Adjust variables for fetching files from the new release. OLDRELNUM=${RELNUM} RELNUM=${TARGETRELEASE} OLDFETCHDIR=${FETCHDIR} FETCHDIR=${RELNUM}/${ARCH} # Try to fetch the NEW metadata index signature ("tag") until we run # out of available servers; and sanity check the downloaded tag. while ! fetch_tag; do fetch_pick_server || return 1 done # Fetch the new INDEX-ALL. fetch_metadata INDEX-ALL || return 1 # If StrictComponents is not "yes", COMPONENTS contains an entry # corresponding to the currently running kernel, and said kernel # does not exist in the new release, add "kernel/generic" to the # list of components. upgrade_guess_new_kernel INDEX-ALL || return 1 # Filter INDEX-ALL to contain only the components we want and without # anything marked as "Ignore". fetch_filter_metadata INDEX-ALL || return 1 # Convert INDEX-OLD (last release) and INDEX-ALL (new release) into # INDEX-OLD and INDEX-NEW files (in the sense of normal upgrades). upgrade_oldall_to_oldnew INDEX-OLD INDEX-ALL INDEX-NEW # Translate /boot/${KERNCONF} or /boot/${NKERNCONF} into ${KERNELDIR} fetch_filter_kernel_names INDEX-NEW ${NKERNCONF} fetch_filter_kernel_names INDEX-OLD ${KERNCONF} # For all paths appearing in INDEX-OLD or INDEX-NEW, inspect the # system and generate an INDEX-PRESENT file. fetch_inspect_system INDEX-OLD INDEX-PRESENT INDEX-NEW || return 1 # Based on ${MERGECHANGES}, generate a file tomerge-old with the # paths and hashes of old versions of files to merge. fetch_filter_mergechanges INDEX-OLD INDEX-PRESENT tomerge-old # Based on ${UPDATEIFUNMODIFIED}, remove lines from INDEX-* which # correspond to lines in INDEX-PRESENT with hashes not appearing # in INDEX-OLD or INDEX-NEW. Also remove lines where the entry in # INDEX-PRESENT has type - and there isn't a corresponding entry in # INDEX-OLD with type -. fetch_filter_unmodified_notpresent \ INDEX-OLD INDEX-PRESENT INDEX-NEW tomerge-old # For each entry in INDEX-PRESENT of type -, remove any corresponding # entry from INDEX-NEW if ${ALLOWADD} != "yes". Remove all entries # of type - from INDEX-PRESENT. fetch_filter_allowadd INDEX-PRESENT INDEX-NEW # If ${ALLOWDELETE} != "yes", then remove any entries from # INDEX-PRESENT which don't correspond to entries in INDEX-NEW. fetch_filter_allowdelete INDEX-PRESENT INDEX-NEW # If ${KEEPMODIFIEDMETADATA} == "yes", then for each entry in # INDEX-PRESENT with metadata not matching any entry in INDEX-OLD, # replace the corresponding line of INDEX-NEW with one having the # same metadata as the entry in INDEX-PRESENT. fetch_filter_modified_metadata INDEX-OLD INDEX-PRESENT INDEX-NEW # Remove lines from INDEX-PRESENT and INDEX-NEW which are identical; # no need to update a file if it isn't changing. fetch_filter_uptodate INDEX-PRESENT INDEX-NEW # Fetch "clean" files from the old release for merging changes. fetch_files_premerge tomerge-old # Prepare to fetch files: Generate a list of the files we need, # copy the unmodified files we have into /files/, and generate # a list of patches to download. fetch_files_prepare INDEX-OLD INDEX-PRESENT INDEX-NEW || return 1 # Fetch patches from to-${RELNUM}/${ARCH}/bp/ PATCHDIR=to-${RELNUM}/${ARCH}/bp fetch_files || return 1 # Merge configuration file changes. upgrade_merge tomerge-old INDEX-PRESENT INDEX-NEW || return 1 # Create and populate install manifest directory; and report what # updates are available. fetch_create_manifest || return 1 # Leave a note behind to tell the "install" command that the kernel # needs to be installed before the world. touch ${BDHASH}-install/kernelfirst # Remind the user that they need to run "freebsd-update install" # to install the downloaded bits, in case they didn't RTFM. echo "To install the downloaded upgrades, run \"$0 install\"." } # Make sure that all the file hashes mentioned in $@ have corresponding # gzipped files stored in /files/. install_verify () { # Generate a list of hashes cat $@ | cut -f 2,7 -d '|' | grep -E '^f' | cut -f 2 -d '|' | sort -u > filelist # Make sure all the hashes exist while read HASH; do if ! [ -f files/${HASH}.gz ]; then echo -n "Update files missing -- " echo "this should never happen." echo "Re-run '$0 fetch'." return 1 fi done < filelist # Clean up rm filelist } # Remove the system immutable flag from files install_unschg () { # Generate file list cat $@ | cut -f 1 -d '|' > filelist # Remove flags while read F; do if ! [ -e ${BASEDIR}/${F} ]; then continue fi chflags noschg ${BASEDIR}/${F} || return 1 done < filelist # Clean up rm filelist } # Decide which directory name to use for kernel backups. backup_kernel_finddir () { CNT=0 while true ; do # Pathname does not exist, so it is OK use that name # for backup directory. if [ ! -e $BASEDIR/$BACKUPKERNELDIR ]; then return 0 fi # If directory do exist, we only use if it has our # marker file. if [ -d $BASEDIR/$BACKUPKERNELDIR -a \ -e $BASEDIR/$BACKUPKERNELDIR/.freebsd-update ]; then return 0 fi # We could not use current directory name, so add counter to # the end and try again. CNT=$((CNT + 1)) if [ $CNT -gt 9 ]; then echo "Could not find valid backup dir ($BASEDIR/$BACKUPKERNELDIR)" exit 1 fi BACKUPKERNELDIR="`echo $BACKUPKERNELDIR | sed -Ee 's/[0-9]\$//'`" BACKUPKERNELDIR="${BACKUPKERNELDIR}${CNT}" done } # Backup the current kernel using hardlinks, if not disabled by user. # Since we delete all files in the directory used for previous backups # we create a marker file called ".freebsd-update" in the directory so # we can determine on the next run that the directory was created by # freebsd-update and we then do not accidentally remove user files in # the unlikely case that the user has created a directory with a # conflicting name. backup_kernel () { # Only make kernel backup is so configured. if [ $BACKUPKERNEL != yes ]; then return 0 fi # Decide which directory name to use for kernel backups. backup_kernel_finddir # Remove old kernel backup files. If $BACKUPKERNELDIR was # "not ours", backup_kernel_finddir would have exited, so # deleting the directory content is as safe as we can make it. if [ -d $BASEDIR/$BACKUPKERNELDIR ]; then rm -fr $BASEDIR/$BACKUPKERNELDIR fi # Create directories for backup. mkdir -p $BASEDIR/$BACKUPKERNELDIR mtree -cdn -p "${BASEDIR}/${KERNELDIR}" | \ mtree -Ue -p "${BASEDIR}/${BACKUPKERNELDIR}" > /dev/null # Mark the directory as having been created by freebsd-update. touch $BASEDIR/$BACKUPKERNELDIR/.freebsd-update if [ $? -ne 0 ]; then echo "Could not create kernel backup directory" exit 1 fi # Disable pathname expansion to be sure *.symbols is not # expanded. set -f # Use find to ignore symbol files, unless disabled by user. if [ $BACKUPKERNELSYMBOLFILES = yes ]; then FINDFILTER="" else FINDFILTER=-"a ! -name *.symbols" fi # Backup all the kernel files using hardlinks. (cd ${BASEDIR}/${KERNELDIR} && find . -type f $FINDFILTER -exec \ cp -pl '{}' ${BASEDIR}/${BACKUPKERNELDIR}/'{}' \;) # Re-enable patchname expansion. set +f } # Install new files install_from_index () { # First pass: Do everything apart from setting file flags. We # can't set flags yet, because schg inhibits hard linking. sort -k 1,1 -t '|' $1 | tr '|' ' ' | while read FPATH TYPE OWNER GROUP PERM FLAGS HASH LINK; do case ${TYPE} in d) # Create a directory install -d -o ${OWNER} -g ${GROUP} \ -m ${PERM} ${BASEDIR}/${FPATH} ;; f) if [ -z "${LINK}" ]; then # Create a file, without setting flags. gunzip < files/${HASH}.gz > ${HASH} install -S -o ${OWNER} -g ${GROUP} \ -m ${PERM} ${HASH} ${BASEDIR}/${FPATH} rm ${HASH} else # Create a hard link. ln -f ${BASEDIR}/${LINK} ${BASEDIR}/${FPATH} fi ;; L) # Create a symlink ln -sfh ${HASH} ${BASEDIR}/${FPATH} ;; esac done # Perform a second pass, adding file flags. tr '|' ' ' < $1 | while read FPATH TYPE OWNER GROUP PERM FLAGS HASH LINK; do if [ ${TYPE} = "f" ] && ! [ ${FLAGS} = "0" ]; then chflags ${FLAGS} ${BASEDIR}/${FPATH} fi done } # Remove files which we want to delete install_delete () { # Generate list of new files cut -f 1 -d '|' < $2 | sort > newfiles # Generate subindex of old files we want to nuke sort -k 1,1 -t '|' $1 | join -t '|' -v 1 - newfiles | sort -r -k 1,1 -t '|' | cut -f 1,2 -d '|' | tr '|' ' ' > killfiles # Remove the offending bits while read FPATH TYPE; do case ${TYPE} in d) rmdir ${BASEDIR}/${FPATH} ;; f) rm ${BASEDIR}/${FPATH} ;; L) rm ${BASEDIR}/${FPATH} ;; esac done < killfiles # Clean up rm newfiles killfiles } # Install new files, delete old files, and update linker.hints install_files () { # If we haven't already dealt with the kernel, deal with it. if ! [ -f $1/kerneldone ]; then grep -E '^/boot/' $1/INDEX-OLD > INDEX-OLD grep -E '^/boot/' $1/INDEX-NEW > INDEX-NEW # Backup current kernel before installing a new one backup_kernel || return 1 # Install new files install_from_index INDEX-NEW || return 1 # Remove files which need to be deleted install_delete INDEX-OLD INDEX-NEW || return 1 # Update linker.hints if necessary if [ -s INDEX-OLD -o -s INDEX-NEW ]; then kldxref -R ${BASEDIR}/boot/ 2>/dev/null fi # We've finished updating the kernel. touch $1/kerneldone # Do we need to ask for a reboot now? if [ -f $1/kernelfirst ] && [ -s INDEX-OLD -o -s INDEX-NEW ]; then cat <<-EOF Kernel updates have been installed. Please reboot and run "$0 install" again to finish installing updates. EOF exit 0 fi fi # If we haven't already dealt with the world, deal with it. if ! [ -f $1/worlddone ]; then # Create any necessary directories first grep -vE '^/boot/' $1/INDEX-NEW | grep -E '^[^|]+\|d\|' > INDEX-NEW install_from_index INDEX-NEW || return 1 # Install new runtime linker grep -vE '^/boot/' $1/INDEX-NEW | grep -vE '^[^|]+\|d\|' | grep -E '^/libexec/ld-elf[^|]*\.so\.[0-9]+\|' > INDEX-NEW install_from_index INDEX-NEW || return 1 # Install new shared libraries next grep -vE '^/boot/' $1/INDEX-NEW | grep -vE '^[^|]+\|d\|' | grep -vE '^/libexec/ld-elf[^|]*\.so\.[0-9]+\|' | grep -E '^[^|]*/lib/[^|]*\.so\.[0-9]+\|' > INDEX-NEW install_from_index INDEX-NEW || return 1 # Deal with everything else grep -vE '^/boot/' $1/INDEX-OLD | grep -vE '^[^|]+\|d\|' | grep -vE '^/libexec/ld-elf[^|]*\.so\.[0-9]+\|' | grep -vE '^[^|]*/lib/[^|]*\.so\.[0-9]+\|' > INDEX-OLD grep -vE '^/boot/' $1/INDEX-NEW | grep -vE '^[^|]+\|d\|' | grep -vE '^/libexec/ld-elf[^|]*\.so\.[0-9]+\|' | grep -vE '^[^|]*/lib/[^|]*\.so\.[0-9]+\|' > INDEX-NEW install_from_index INDEX-NEW || return 1 install_delete INDEX-OLD INDEX-NEW || return 1 # Rebuild /etc/spwd.db and /etc/pwd.db if necessary. if [ ${BASEDIR}/etc/master.passwd -nt ${BASEDIR}/etc/spwd.db ] || [ ${BASEDIR}/etc/master.passwd -nt ${BASEDIR}/etc/pwd.db ]; then pwd_mkdb -d ${BASEDIR}/etc ${BASEDIR}/etc/master.passwd fi # Rebuild /etc/login.conf.db if necessary. if [ ${BASEDIR}/etc/login.conf -nt ${BASEDIR}/etc/login.conf.db ]; then cap_mkdb ${BASEDIR}/etc/login.conf fi # We've finished installing the world and deleting old files # which are not shared libraries. touch $1/worlddone # Do we need to ask the user to portupgrade now? grep -vE '^/boot/' $1/INDEX-NEW | grep -E '^[^|]*/lib/[^|]*\.so\.[0-9]+\|' | cut -f 1 -d '|' | sort > newfiles if grep -vE '^/boot/' $1/INDEX-OLD | grep -E '^[^|]*/lib/[^|]*\.so\.[0-9]+\|' | cut -f 1 -d '|' | sort | join -v 1 - newfiles | grep -q .; then cat <<-EOF Completing this upgrade requires removing old shared object files. Please rebuild all installed 3rd party software (e.g., programs installed from the ports tree) and then run "$0 install" again to finish installing updates. EOF rm newfiles exit 0 fi rm newfiles fi # Remove old shared libraries grep -vE '^/boot/' $1/INDEX-NEW | grep -vE '^[^|]+\|d\|' | grep -E '^[^|]*/lib/[^|]*\.so\.[0-9]+\|' > INDEX-NEW grep -vE '^/boot/' $1/INDEX-OLD | grep -vE '^[^|]+\|d\|' | grep -E '^[^|]*/lib/[^|]*\.so\.[0-9]+\|' > INDEX-OLD install_delete INDEX-OLD INDEX-NEW || return 1 # Remove old directories grep -vE '^/boot/' $1/INDEX-NEW | grep -E '^[^|]+\|d\|' > INDEX-NEW grep -vE '^/boot/' $1/INDEX-OLD | grep -E '^[^|]+\|d\|' > INDEX-OLD install_delete INDEX-OLD INDEX-NEW || return 1 # Remove temporary files rm INDEX-OLD INDEX-NEW } # Rearrange bits to allow the installed updates to be rolled back install_setup_rollback () { # Remove the "reboot after installing kernel", "kernel updated", and # "finished installing the world" flags if present -- they are # irrelevant when rolling back updates. if [ -f ${BDHASH}-install/kernelfirst ]; then rm ${BDHASH}-install/kernelfirst rm ${BDHASH}-install/kerneldone fi if [ -f ${BDHASH}-install/worlddone ]; then rm ${BDHASH}-install/worlddone fi if [ -L ${BDHASH}-rollback ]; then mv ${BDHASH}-rollback ${BDHASH}-install/rollback fi mv ${BDHASH}-install ${BDHASH}-rollback } # Actually install updates install_run () { echo -n "Installing updates..." # Make sure we have all the files we should have install_verify ${BDHASH}-install/INDEX-OLD \ ${BDHASH}-install/INDEX-NEW || return 1 # Remove system immutable flag from files install_unschg ${BDHASH}-install/INDEX-OLD \ ${BDHASH}-install/INDEX-NEW || return 1 # Install new files, delete old files, and update linker.hints install_files ${BDHASH}-install || return 1 # Rearrange bits to allow the installed updates to be rolled back install_setup_rollback echo " done." } # Rearrange bits to allow the previous set of updates to be rolled back next. rollback_setup_rollback () { if [ -L ${BDHASH}-rollback/rollback ]; then mv ${BDHASH}-rollback/rollback rollback-tmp rm -r ${BDHASH}-rollback/ rm ${BDHASH}-rollback mv rollback-tmp ${BDHASH}-rollback else rm -r ${BDHASH}-rollback/ rm ${BDHASH}-rollback fi } # Install old files, delete new files, and update linker.hints rollback_files () { # Install old shared library files which don't have the same path as # a new shared library file. grep -vE '^/boot/' $1/INDEX-NEW | grep -E '/lib/.*\.so\.[0-9]+\|' | cut -f 1 -d '|' | sort > INDEX-NEW.libs.flist grep -vE '^/boot/' $1/INDEX-OLD | grep -E '/lib/.*\.so\.[0-9]+\|' | sort -k 1,1 -t '|' - | join -t '|' -v 1 - INDEX-NEW.libs.flist > INDEX-OLD install_from_index INDEX-OLD || return 1 # Deal with files which are neither kernel nor shared library grep -vE '^/boot/' $1/INDEX-OLD | grep -vE '/lib/.*\.so\.[0-9]+\|' > INDEX-OLD grep -vE '^/boot/' $1/INDEX-NEW | grep -vE '/lib/.*\.so\.[0-9]+\|' > INDEX-NEW install_from_index INDEX-OLD || return 1 install_delete INDEX-NEW INDEX-OLD || return 1 # Install any old shared library files which we didn't install above. grep -vE '^/boot/' $1/INDEX-OLD | grep -E '/lib/.*\.so\.[0-9]+\|' | sort -k 1,1 -t '|' - | join -t '|' - INDEX-NEW.libs.flist > INDEX-OLD install_from_index INDEX-OLD || return 1 # Delete unneeded shared library files grep -vE '^/boot/' $1/INDEX-OLD | grep -E '/lib/.*\.so\.[0-9]+\|' > INDEX-OLD grep -vE '^/boot/' $1/INDEX-NEW | grep -E '/lib/.*\.so\.[0-9]+\|' > INDEX-NEW install_delete INDEX-NEW INDEX-OLD || return 1 # Deal with kernel files grep -E '^/boot/' $1/INDEX-OLD > INDEX-OLD grep -E '^/boot/' $1/INDEX-NEW > INDEX-NEW install_from_index INDEX-OLD || return 1 install_delete INDEX-NEW INDEX-OLD || return 1 if [ -s INDEX-OLD -o -s INDEX-NEW ]; then kldxref -R /boot/ 2>/dev/null fi # Remove temporary files rm INDEX-OLD INDEX-NEW INDEX-NEW.libs.flist } # Actually rollback updates rollback_run () { echo -n "Uninstalling updates..." # If there are updates waiting to be installed, remove them; we # want the user to re-run 'fetch' after rolling back updates. if [ -L ${BDHASH}-install ]; then rm -r ${BDHASH}-install/ rm ${BDHASH}-install fi # Make sure we have all the files we should have install_verify ${BDHASH}-rollback/INDEX-NEW \ ${BDHASH}-rollback/INDEX-OLD || return 1 # Remove system immutable flag from files install_unschg ${BDHASH}-rollback/INDEX-NEW \ ${BDHASH}-rollback/INDEX-OLD || return 1 # Install old files, delete new files, and update linker.hints rollback_files ${BDHASH}-rollback || return 1 # Remove the rollback directory and the symlink pointing to it; and # rearrange bits to allow the previous set of updates to be rolled # back next. rollback_setup_rollback echo " done." } # Compare INDEX-ALL and INDEX-PRESENT and print warnings about differences. IDS_compare () { # Get all the lines which mismatch in something other than file # flags. We ignore file flags because sysinstall doesn't seem to # set them when it installs FreeBSD; warning about these adds a # very large amount of noise. cut -f 1-5,7-8 -d '|' $1 > $1.noflags sort -k 1,1 -t '|' $1.noflags > $1.sorted cut -f 1-5,7-8 -d '|' $2 | comm -13 $1.noflags - | fgrep -v '|-|||||' | sort -k 1,1 -t '|' | join -t '|' $1.sorted - > INDEX-NOTMATCHING # Ignore files which match IDSIGNOREPATHS. for X in ${IDSIGNOREPATHS}; do grep -E "^${X}" INDEX-NOTMATCHING done | sort -u | comm -13 - INDEX-NOTMATCHING > INDEX-NOTMATCHING.tmp mv INDEX-NOTMATCHING.tmp INDEX-NOTMATCHING # Go through the lines and print warnings. local IFS='|' while read FPATH TYPE OWNER GROUP PERM HASH LINK P_TYPE P_OWNER P_GROUP P_PERM P_HASH P_LINK; do # Warn about different object types. if ! [ "${TYPE}" = "${P_TYPE}" ]; then echo -n "${FPATH} is a " case "${P_TYPE}" in f) echo -n "regular file, " ;; d) echo -n "directory, " ;; L) echo -n "symlink, " ;; esac echo -n "but should be a " case "${TYPE}" in f) echo -n "regular file." ;; d) echo -n "directory." ;; L) echo -n "symlink." ;; esac echo # Skip other tests, since they don't make sense if # we're comparing different object types. continue fi # Warn about different owners. if ! [ "${OWNER}" = "${P_OWNER}" ]; then echo -n "${FPATH} is owned by user id ${P_OWNER}, " echo "but should be owned by user id ${OWNER}." fi # Warn about different groups. if ! [ "${GROUP}" = "${P_GROUP}" ]; then echo -n "${FPATH} is owned by group id ${P_GROUP}, " echo "but should be owned by group id ${GROUP}." fi # Warn about different permissions. We do not warn about # different permissions on symlinks, since some archivers # don't extract symlink permissions correctly and they are # ignored anyway. if ! [ "${PERM}" = "${P_PERM}" ] && ! [ "${TYPE}" = "L" ]; then echo -n "${FPATH} has ${P_PERM} permissions, " echo "but should have ${PERM} permissions." fi # Warn about different file hashes / symlink destinations. if ! [ "${HASH}" = "${P_HASH}" ]; then if [ "${TYPE}" = "L" ]; then echo -n "${FPATH} is a symlink to ${P_HASH}, " echo "but should be a symlink to ${HASH}." fi if [ "${TYPE}" = "f" ]; then echo -n "${FPATH} has SHA256 hash ${P_HASH}, " echo "but should have SHA256 hash ${HASH}." fi fi # We don't warn about different hard links, since some # some archivers break hard links, and as long as the # underlying data is correct they really don't matter. done < INDEX-NOTMATCHING # Clean up rm $1 $1.noflags $1.sorted $2 INDEX-NOTMATCHING } # Do the work involved in comparing the system to a "known good" index IDS_run () { workdir_init || return 1 # Prepare the mirror list. fetch_pick_server_init && fetch_pick_server # Try to fetch the public key until we run out of servers. while ! fetch_key; do fetch_pick_server || return 1 done # Try to fetch the metadata index signature ("tag") until we run # out of available servers; and sanity check the downloaded tag. while ! fetch_tag; do fetch_pick_server || return 1 done fetch_tagsanity || return 1 # Fetch INDEX-OLD and INDEX-ALL. fetch_metadata INDEX-OLD INDEX-ALL || return 1 # Generate filtered INDEX-OLD and INDEX-ALL files containing only # the components we want and without anything marked as "Ignore". fetch_filter_metadata INDEX-OLD || return 1 fetch_filter_metadata INDEX-ALL || return 1 # Merge the INDEX-OLD and INDEX-ALL files into INDEX-ALL. sort INDEX-OLD INDEX-ALL > INDEX-ALL.tmp mv INDEX-ALL.tmp INDEX-ALL rm INDEX-OLD # Translate /boot/${KERNCONF} to ${KERNELDIR} fetch_filter_kernel_names INDEX-ALL ${KERNCONF} # Inspect the system and generate an INDEX-PRESENT file. fetch_inspect_system INDEX-ALL INDEX-PRESENT /dev/null || return 1 # Compare INDEX-ALL and INDEX-PRESENT and print warnings about any # differences. IDS_compare INDEX-ALL INDEX-PRESENT } #### Main functions -- call parameter-handling and core functions # Using the command line, configuration file, and defaults, # set all the parameters which are needed later. get_params () { init_params parse_cmdline $@ parse_conffile default_params } # Fetch command. Make sure that we're being called # interactively, then run fetch_check_params and fetch_run cmd_fetch () { - if [ ! -t 0 ]; then + if [ ! -t 0 && $NOTTYOK -eq 0 ]; then echo -n "`basename $0` fetch should not " echo "be run non-interactively." echo "Run `basename $0` cron instead." exit 1 fi fetch_check_params fetch_run || exit 1 } # Cron command. Make sure the parameters are sensible; wait # rand(3600) seconds; then fetch updates. While fetching updates, # send output to a temporary file; only print that file if the # fetching failed. cmd_cron () { fetch_check_params sleep `jot -r 1 0 3600` TMPFILE=`mktemp /tmp/freebsd-update.XXXXXX` || exit 1 if ! fetch_run >> ${TMPFILE} || ! grep -q "No updates needed" ${TMPFILE} || [ ${VERBOSELEVEL} = "debug" ]; then mail -s "`hostname` security updates" ${MAILTO} < ${TMPFILE} fi rm ${TMPFILE} } # Fetch files for upgrading to a new release. cmd_upgrade () { upgrade_check_params upgrade_run || exit 1 } # Install downloaded updates. cmd_install () { install_check_params install_run || exit 1 } # Rollback most recently installed updates. cmd_rollback () { rollback_check_params rollback_run || exit 1 } # Compare system against a "known good" index. cmd_IDS () { IDS_check_params IDS_run || exit 1 } #### Entry point # Make sure we find utilities from the base system export PATH=/sbin:/bin:/usr/sbin:/usr/bin:${PATH} # Set a pager if the user doesn't if [ -z "$PAGER" ]; then PAGER=/usr/bin/more fi # Set LC_ALL in order to avoid problems with character ranges like [A-Z]. export LC_ALL=C get_params $@ for COMMAND in ${COMMANDS}; do cmd_${COMMAND} done Index: projects/cxl_iscsi/usr.sbin/syslogd/syslogd.8 =================================================================== --- projects/cxl_iscsi/usr.sbin/syslogd/syslogd.8 (revision 279716) +++ projects/cxl_iscsi/usr.sbin/syslogd/syslogd.8 (revision 279717) @@ -1,414 +1,425 @@ .\" Copyright (c) 1983, 1986, 1991, 1993 .\" The Regents of the University of California. All rights reserved. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" 4. Neither the name of the University nor the names of its contributors .\" may be used to endorse or promote products derived from this software .\" without specific prior written permission. .\" .\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" @(#)syslogd.8 8.1 (Berkeley) 6/6/93 .\" $FreeBSD$ .\" -.Dd May 13, 2008 +.Dd March 3, 2015 .Dt SYSLOGD 8 .Os .Sh NAME .Nm syslogd .Nd log systems messages .Sh SYNOPSIS .Nm -.Op Fl 468ACcdkNnosTuv +.Op Fl 468ACcdFkNnosTuv .Op Fl a Ar allowed_peer .Op Fl b Ar bind_address .Op Fl f Ar config_file .Op Fl l Oo Ar mode : Oc Ns Ar path .Op Fl m Ar mark_interval .Op Fl P Ar pid_file .Op Fl p Ar log_socket .Sh DESCRIPTION The .Nm utility reads and logs messages to the system console, log files, other machines and/or users as specified by its configuration file. .Pp The options are as follows: .Bl -tag -width indent .It Fl 4 Force .Nm to use IPv4 addresses only. .It Fl 6 Force .Nm to use IPv6 addresses only. .It Fl 8 Tells .Nm not to interfere with 8-bit data. Normally .Nm will replace C1 control characters .Pq ISO 8859 and Unicode characters with their .Dq M- Ns Em x equivalent. Note, this option does not change the way .Nm alters control characters .Pq see Xr iscntrl 3 . They will always be replaced with their .Dq ^ Ns Em x equivalent. .It Fl A Ordinarily, .Nm tries to send the message to only one address even if the host has more than one A or AAAA record. If this option is specified, .Nm tries to send the message to all addresses. .It Fl a Ar allowed_peer Allow .Ar allowed_peer to log to this .Nm using UDP datagrams. Multiple .Fl a options may be specified. .Pp The .Ar allowed_peer option may be any of the following: .Bl -tag -width "ipaddr/masklen[:service]XX" .It Xo .Sm off .Ar ipaddr .No / Ar masklen .Op : Ar service .Sm on .Xc Accept datagrams from .Ar ipaddr (in the usual dotted quad notation) with .Ar masklen bits being taken into account when doing the address comparison. .Ar ipaddr can be also IPv6 address by enclosing the address with .Ql \&[ and .Ql \&] . If specified, .Ar service is the name or number of an UDP service (see .Xr services 5 ) the source packet must belong to. A .Ar service of .Ql \&* allows packets being sent from any UDP port. The default .Ar service is .Ql syslog . If .Ar ipaddr is IPv4 address, a missing .Ar masklen will be substituted by the historic class A or class B netmasks if .Ar ipaddr belongs into the address range of class A or B, respectively, or by 24 otherwise. If .Ar ipaddr is IPv6 address, a missing .Ar masklen will be substituted by 128. .It Xo .Sm off .Ar domainname Op : Ar service .Sm on .Xc Accept datagrams where the reverse address lookup yields .Ar domainname for the sender address. The meaning of .Ar service is as explained above. .It Xo .Sm off .No * Ar domainname Op : Ar service .Sm on .Xc Same as before, except that any source host whose name .Em ends in .Ar domainname will get permission. .El .Pp The .Fl a options are ignored if the .Fl s option is also specified. .It Xo .Fl b .Sm off .Ar bind_address Op : Ar service .Sm on .Xc .It Xo .Fl b .Sm off .Li : Ar service .Sm on .Xc Bind to a specific address and/or port. The address can be specified as a hostname, and the port as a service name. If an IPv6 address is specified, it should be enclosed with .Ql \&[ and .Ql \&] . The default .Ar service is .Ql syslog . .It Fl C Create log files that do not exist (permission is set to .Li 0600 ) . .It Fl c Disable the compression of repeated instances of the same line into a single line of the form .Dq Li "last message repeated N times" when the output is a pipe to another program. If specified twice, disable this compression in all cases. .It Fl d Put .Nm into debugging mode. This is probably only of use to developers working on .Nm . .It Fl f Specify the pathname of an alternate configuration file; the default is .Pa /etc/syslog.conf . +.It Fl F +Run +.Nm +in the foreground, rather than going into daemon mode. This is useful if +some other process uses +.Xr fork 2 +and +.Xr exec 3 +to run +.Nm , +and wants to monitor when and how it exits. .It Fl k Disable the translation of messages received with facility .Dq kern to facility .Dq user . Usually the .Dq kern facility is reserved for messages read directly from .Pa /dev/klog . .It Fl m Select the number of minutes between .Dq mark messages; the default is 20 minutes. .It Fl N Disable binding on UDP sockets. RFC 3164 recommends that outgoing syslogd messages should originate from the privileged port, this option .Em disables the recommended behavior. This option inherits .Fl s . .It Fl n Disable dns query for every request. .It Fl o Prefix kernel messages with the full kernel boot file as determined by .Xr getbootfile 3 . Without this, the kernel message prefix is always .Dq Li kernel: . .It Fl p Specify the pathname of an alternate log socket to be used instead; the default is .Pa /var/run/log . .It Fl P Specify an alternative file in which to store the process ID. The default is .Pa /var/run/syslog.pid . .It Fl S Specify the pathname of an alternate log socket for privileged applications to be used instead; the default is .Pa /var/run/logpriv . .It Fl l Specify a location where .Nm should place an additional log socket. The primary use for this is to place additional log sockets in .Pa /var/run/log of various chroot filespaces. File permissions for socket can be specified in octal representation before socket name, delimited with a colon. Path to socket location must be absolute. .It Fl s Operate in secure mode. Do not log messages from remote machines. If specified twice, no network socket will be opened at all, which also disables logging to remote machines. .It Fl T Always use the local time and date for messages received from the network, instead of the timestamp field supplied in the message by the remote host. This is useful if some of the originating hosts can't keep time properly or are unable to generate a correct timestamp. .It Fl u Unique priority logging. Only log messages at the specified priority. Without this option, messages at the stated priority or higher are logged. This option changes the default comparison from .Dq => to .Dq = . .It Fl v Verbose logging. If specified once, the numeric facility and priority are logged with each locally-written message. If specified more than once, the names of the facility and priority are logged with each locally-written message. .El .Pp The .Nm utility reads its configuration file when it starts up and whenever it receives a hangup signal. For information on the format of the configuration file, see .Xr syslog.conf 5 . .Pp The .Nm utility reads messages from the .Ux domain sockets .Pa /var/run/log and .Pa /var/run/logpriv , from an Internet domain socket specified in .Pa /etc/services , and from the special device .Pa /dev/klog (to read kernel messages). .Pp The .Nm utility creates its process ID file, by default .Pa /var/run/syslog.pid , and stores its process ID there. This can be used to kill or reconfigure .Nm . .Pp The message sent to .Nm should consist of a single line. The message can contain a priority code, which should be a preceding decimal number in angle braces, for example, .Sq Aq 5 . This priority code should map into the priorities defined in the include file .In sys/syslog.h . .Pp For security reasons, .Nm will not append to log files that do not exist (unless .Fl C option is specified); therefore, they must be created manually before running .Nm . .Pp The date and time are taken from the received message. If the format of the timestamp field is incorrect, time obtained from the local host is used instead. This can be overridden by the .Fl T flag. .Sh FILES .Bl -tag -width /var/run/syslog.pid -compact .It Pa /etc/syslog.conf configuration file .It Pa /var/run/syslog.pid default process ID file .It Pa /var/run/log name of the .Ux domain datagram log socket .It Pa /var/run/logpriv .Ux socket for privileged applications .It Pa /dev/klog kernel log device .El .Sh SEE ALSO .Xr logger 1 , .Xr syslog 3 , .Xr services 5 , .Xr syslog.conf 5 , .Xr newsyslog 8 .Sh HISTORY The .Nm utility appeared in .Bx 4.3 . .Pp The .Fl a , .Fl s , .Fl u , and .Fl v options are .Fx 2.2 extensions. .Sh BUGS The ability to log messages received in UDP packets is equivalent to an unauthenticated remote disk-filling service, and should probably be disabled by default. Some sort of .No inter- Ns Nm syslogd authentication mechanism ought to be worked out. To prevent the worst abuse, use of the .Fl a option is therefore highly recommended. .Pp The .Fl a matching algorithm does not pretend to be very efficient; use of numeric IP addresses is faster than domain name comparison. Since the allowed peer list is being walked linearly, peer groups where frequent messages are being anticipated from should be put early into the .Fl a list. .Pp The log socket was moved from .Pa /dev to ease the use of a read-only root file system. This may confuse some old binaries so that a symbolic link might be used for a transitional period. Index: projects/cxl_iscsi/usr.sbin/syslogd/syslogd.c =================================================================== --- projects/cxl_iscsi/usr.sbin/syslogd/syslogd.c (revision 279716) +++ projects/cxl_iscsi/usr.sbin/syslogd/syslogd.c (revision 279717) @@ -1,2758 +1,2762 @@ /* * Copyright (c) 1983, 1988, 1993, 1994 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef lint static const char copyright[] = "@(#) Copyright (c) 1983, 1988, 1993, 1994\n\ The Regents of the University of California. All rights reserved.\n"; #endif /* not lint */ #ifndef lint #if 0 static char sccsid[] = "@(#)syslogd.c 8.3 (Berkeley) 4/4/94"; #endif #endif /* not lint */ #include __FBSDID("$FreeBSD$"); /* * syslogd -- log system messages * * This program implements a system log. It takes a series of lines. * Each line may have a priority, signified as "" as * the first characters of the line. If this is * not present, a default priority is used. * * To kill syslogd, send a signal 15 (terminate). A signal 1 (hup) will * cause it to reread its configuration file. * * Defined Constants: * * MAXLINE -- the maximum line length that can be handled. * DEFUPRI -- the default priority for user messages * DEFSPRI -- the default priority for kernel messages * * Author: Eric Allman * extensive changes by Ralph Campbell * more extensive changes by Eric Allman (again) * Extension to log by program name as well as facility and priority * by Peter da Silva. * -u and -v by Harlan Stenn. * Priority comparison code by Harlan Stenn. */ #define MAXLINE 1024 /* maximum line length */ #define MAXSVLINE 120 /* maximum saved line length */ #define DEFUPRI (LOG_USER|LOG_NOTICE) #define DEFSPRI (LOG_KERN|LOG_CRIT) #define TIMERINTVL 30 /* interval for checking flush, mark */ #define TTYMSGTIME 1 /* timeout passed to ttymsg */ #define RCVBUF_MINSIZE (80 * 1024) /* minimum size of dgram rcv buffer */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "pathnames.h" #include "ttymsg.h" #define SYSLOG_NAMES #include const char *ConfFile = _PATH_LOGCONF; const char *PidFile = _PATH_LOGPID; const char ctty[] = _PATH_CONSOLE; #define dprintf if (Debug) printf #define MAXUNAMES 20 /* maximum number of user names */ /* * Unix sockets. * We have two default sockets, one with 666 permissions, * and one for privileged programs. */ struct funix { int s; const char *name; mode_t mode; STAILQ_ENTRY(funix) next; }; struct funix funix_secure = { -1, _PATH_LOG_PRIV, S_IRUSR | S_IWUSR, { NULL } }; struct funix funix_default = { -1, _PATH_LOG, DEFFILEMODE, { &funix_secure } }; STAILQ_HEAD(, funix) funixes = { &funix_default, &(funix_secure.next.stqe_next) }; /* * Flags to logmsg(). */ #define IGN_CONS 0x001 /* don't print on console */ #define SYNC_FILE 0x002 /* do fsync on file after printing */ #define ADDDATE 0x004 /* add a date to the message */ #define MARK 0x008 /* this message is a mark */ #define ISKERNEL 0x010 /* kernel generated message */ /* * This structure represents the files that will have log * copies printed. * We require f_file to be valid if f_type is F_FILE, F_CONSOLE, F_TTY * or if f_type if F_PIPE and f_pid > 0. */ struct filed { struct filed *f_next; /* next in linked list */ short f_type; /* entry type, see below */ short f_file; /* file descriptor */ time_t f_time; /* time this was last written */ char *f_host; /* host from which to recd. */ u_char f_pmask[LOG_NFACILITIES+1]; /* priority mask */ u_char f_pcmp[LOG_NFACILITIES+1]; /* compare priority */ #define PRI_LT 0x1 #define PRI_EQ 0x2 #define PRI_GT 0x4 char *f_program; /* program this applies to */ union { char f_uname[MAXUNAMES][MAXLOGNAME]; struct { char f_hname[MAXHOSTNAMELEN]; struct addrinfo *f_addr; } f_forw; /* forwarding address */ char f_fname[MAXPATHLEN]; struct { char f_pname[MAXPATHLEN]; pid_t f_pid; } f_pipe; } f_un; char f_prevline[MAXSVLINE]; /* last message logged */ char f_lasttime[16]; /* time of last occurrence */ char f_prevhost[MAXHOSTNAMELEN]; /* host from which recd. */ int f_prevpri; /* pri of f_prevline */ int f_prevlen; /* length of f_prevline */ int f_prevcount; /* repetition cnt of prevline */ u_int f_repeatcount; /* number of "repeated" msgs */ int f_flags; /* file-specific flags */ #define FFLAG_SYNC 0x01 #define FFLAG_NEEDSYNC 0x02 }; /* * Queue of about-to-be dead processes we should watch out for. */ TAILQ_HEAD(stailhead, deadq_entry) deadq_head; struct stailhead *deadq_headp; struct deadq_entry { pid_t dq_pid; int dq_timeout; TAILQ_ENTRY(deadq_entry) dq_entries; }; /* * The timeout to apply to processes waiting on the dead queue. Unit * of measure is `mark intervals', i.e. 20 minutes by default. * Processes on the dead queue will be terminated after that time. */ #define DQ_TIMO_INIT 2 typedef struct deadq_entry *dq_t; /* * Struct to hold records of network addresses that are allowed to log * to us. */ struct allowedpeer { int isnumeric; u_short port; union { struct { struct sockaddr_storage addr; struct sockaddr_storage mask; } numeric; char *name; } u; #define a_addr u.numeric.addr #define a_mask u.numeric.mask #define a_name u.name }; /* * Intervals at which we flush out "message repeated" messages, * in seconds after previous message is logged. After each flush, * we move to the next interval until we reach the largest. */ int repeatinterval[] = { 30, 120, 600 }; /* # of secs before flush */ #define MAXREPEAT ((sizeof(repeatinterval) / sizeof(repeatinterval[0])) - 1) #define REPEATTIME(f) ((f)->f_time + repeatinterval[(f)->f_repeatcount]) #define BACKOFF(f) { if (++(f)->f_repeatcount > MAXREPEAT) \ (f)->f_repeatcount = MAXREPEAT; \ } /* values for f_type */ #define F_UNUSED 0 /* unused entry */ #define F_FILE 1 /* regular file */ #define F_TTY 2 /* terminal */ #define F_CONSOLE 3 /* console terminal */ #define F_FORW 4 /* remote machine */ #define F_USERS 5 /* list of users */ #define F_WALL 6 /* everyone logged on */ #define F_PIPE 7 /* pipe to program */ const char *TypeNames[8] = { "UNUSED", "FILE", "TTY", "CONSOLE", "FORW", "USERS", "WALL", "PIPE" }; static struct filed *Files; /* Log files that we write to */ static struct filed consfile; /* Console */ static int Debug; /* debug flag */ +static int Foreground = 0; /* Run in foreground, instead of daemonizing */ static int resolve = 1; /* resolve hostname */ static char LocalHostName[MAXHOSTNAMELEN]; /* our hostname */ static const char *LocalDomain; /* our local domain name */ static int *finet; /* Internet datagram socket */ static int fklog = -1; /* /dev/klog */ static int Initialized; /* set when we have initialized ourselves */ static int MarkInterval = 20 * 60; /* interval between marks in seconds */ static int MarkSeq; /* mark sequence number */ static int NoBind; /* don't bind() as suggested by RFC 3164 */ static int SecureMode; /* when true, receive only unix domain socks */ #ifdef INET6 static int family = PF_UNSPEC; /* protocol family (IPv4, IPv6 or both) */ #else static int family = PF_INET; /* protocol family (IPv4 only) */ #endif static int mask_C1 = 1; /* mask characters from 0x80 - 0x9F */ static int send_to_all; /* send message to all IPv4/IPv6 addresses */ static int use_bootfile; /* log entire bootfile for every kern msg */ static int no_compress; /* don't compress messages (1=pipes, 2=all) */ static int logflags = O_WRONLY|O_APPEND; /* flags used to open log files */ static char bootfile[MAXLINE+1]; /* booted kernel file */ struct allowedpeer *AllowedPeers; /* List of allowed peers */ static int NumAllowed; /* Number of entries in AllowedPeers */ static int RemoteAddDate; /* Always set the date on remote messages */ static int UniquePriority; /* Only log specified priority? */ static int LogFacPri; /* Put facility and priority in log message: */ /* 0=no, 1=numeric, 2=names */ static int KeepKernFac; /* Keep remotely logged kernel facility */ static int needdofsync = 0; /* Are any file(s) waiting to be fsynced? */ static struct pidfh *pfh; volatile sig_atomic_t MarkSet, WantDie; static int allowaddr(char *); static void cfline(const char *, struct filed *, const char *, const char *); static const char *cvthname(struct sockaddr *); static void deadq_enter(pid_t, const char *); static int deadq_remove(pid_t); static int decode(const char *, const CODE *); static void die(int); static void dodie(int); static void dofsync(void); static void domark(int); static void fprintlog(struct filed *, int, const char *); static int *socksetup(int, char *); static void init(int); static void logerror(const char *); static void logmsg(int, const char *, const char *, int); static void log_deadchild(pid_t, int, const char *); static void markit(void); static int skip_message(const char *, const char *, int); static void printline(const char *, char *, int); static void printsys(char *); static int p_open(const char *, pid_t *); static void readklog(void); static void reapchild(int); static void usage(void); static int validate(struct sockaddr *, const char *); static void unmapped(struct sockaddr *); static void wallmsg(struct filed *, struct iovec *, const int iovlen); static int waitdaemon(int, int, int); static void timedout(int); static void increase_rcvbuf(int); int main(int argc, char *argv[]) { int ch, i, fdsrmax = 0, l; struct sockaddr_un sunx, fromunix; struct sockaddr_storage frominet; fd_set *fdsr = NULL; char line[MAXLINE + 1]; char *bindhostname; const char *hname; struct timeval tv, *tvp; struct sigaction sact; struct funix *fx, *fx1; sigset_t mask; pid_t ppid = 1, spid; socklen_t len; if (madvise(NULL, 0, MADV_PROTECT) != 0) dprintf("madvise() failed: %s\n", strerror(errno)); bindhostname = NULL; - while ((ch = getopt(argc, argv, "468Aa:b:cCdf:kl:m:nNop:P:sS:Tuv")) + while ((ch = getopt(argc, argv, "468Aa:b:cCdf:Fkl:m:nNop:P:sS:Tuv")) != -1) switch (ch) { case '4': family = PF_INET; break; #ifdef INET6 case '6': family = PF_INET6; break; #endif case '8': mask_C1 = 0; break; case 'A': send_to_all++; break; case 'a': /* allow specific network addresses only */ if (allowaddr(optarg) == -1) usage(); break; case 'b': bindhostname = optarg; break; case 'c': no_compress++; break; case 'C': logflags |= O_CREAT; break; case 'd': /* debug */ Debug++; break; case 'f': /* configuration file */ ConfFile = optarg; break; + case 'F': /* run in foreground instead of daemon */ + Foreground++; + break; case 'k': /* keep remote kern fac */ KeepKernFac = 1; break; case 'l': { long perml; mode_t mode; char *name, *ep; if (optarg[0] == '/') { mode = DEFFILEMODE; name = optarg; } else if ((name = strchr(optarg, ':')) != NULL) { *name++ = '\0'; if (name[0] != '/') errx(1, "socket name must be absolute " "path"); if (isdigit(*optarg)) { perml = strtol(optarg, &ep, 8); if (*ep || perml < 0 || perml & ~(S_IRWXU|S_IRWXG|S_IRWXO)) errx(1, "invalid mode %s, exiting", optarg); mode = (mode_t )perml; } else errx(1, "invalid mode %s, exiting", optarg); } else /* doesn't begin with '/', and no ':' */ errx(1, "can't parse path %s", optarg); if (strlen(name) >= sizeof(sunx.sun_path)) errx(1, "%s path too long, exiting", name); if ((fx = malloc(sizeof(struct funix))) == NULL) errx(1, "malloc failed"); fx->s = -1; fx->name = name; fx->mode = mode; STAILQ_INSERT_TAIL(&funixes, fx, next); break; } case 'm': /* mark interval */ MarkInterval = atoi(optarg) * 60; break; case 'N': NoBind = 1; SecureMode = 1; break; case 'n': resolve = 0; break; case 'o': use_bootfile = 1; break; case 'p': /* path */ if (strlen(optarg) >= sizeof(sunx.sun_path)) errx(1, "%s path too long, exiting", optarg); funix_default.name = optarg; break; case 'P': /* path for alt. PID */ PidFile = optarg; break; case 's': /* no network mode */ SecureMode++; break; case 'S': /* path for privileged originator */ if (strlen(optarg) >= sizeof(sunx.sun_path)) errx(1, "%s path too long, exiting", optarg); funix_secure.name = optarg; break; case 'T': RemoteAddDate = 1; break; case 'u': /* only log specified priority */ UniquePriority++; break; case 'v': /* log facility and priority */ LogFacPri++; break; default: usage(); } if ((argc -= optind) != 0) usage(); pfh = pidfile_open(PidFile, 0600, &spid); if (pfh == NULL) { if (errno == EEXIST) errx(1, "syslogd already running, pid: %d", spid); warn("cannot open pid file"); } - if (!Debug) { + if ((!Foreground) && (!Debug)) { ppid = waitdaemon(0, 0, 30); if (ppid < 0) { warn("could not become daemon"); pidfile_remove(pfh); exit(1); } - } else { + } else if (Debug) { setlinebuf(stdout); } if (NumAllowed) endservent(); consfile.f_type = F_CONSOLE; (void)strlcpy(consfile.f_un.f_fname, ctty + sizeof _PATH_DEV - 1, sizeof(consfile.f_un.f_fname)); (void)strlcpy(bootfile, getbootfile(), sizeof(bootfile)); (void)signal(SIGTERM, dodie); (void)signal(SIGINT, Debug ? dodie : SIG_IGN); (void)signal(SIGQUIT, Debug ? dodie : SIG_IGN); /* * We don't want the SIGCHLD and SIGHUP handlers to interfere * with each other; they are likely candidates for being called * simultaneously (SIGHUP closes pipe descriptor, process dies, * SIGCHLD happens). */ sigemptyset(&mask); sigaddset(&mask, SIGHUP); sact.sa_handler = reapchild; sact.sa_mask = mask; sact.sa_flags = SA_RESTART; (void)sigaction(SIGCHLD, &sact, NULL); (void)signal(SIGALRM, domark); (void)signal(SIGPIPE, SIG_IGN); /* We'll catch EPIPE instead. */ (void)alarm(TIMERINTVL); TAILQ_INIT(&deadq_head); #ifndef SUN_LEN #define SUN_LEN(unp) (strlen((unp)->sun_path) + 2) #endif STAILQ_FOREACH_SAFE(fx, &funixes, next, fx1) { (void)unlink(fx->name); memset(&sunx, 0, sizeof(sunx)); sunx.sun_family = AF_LOCAL; (void)strlcpy(sunx.sun_path, fx->name, sizeof(sunx.sun_path)); fx->s = socket(PF_LOCAL, SOCK_DGRAM, 0); if (fx->s < 0 || bind(fx->s, (struct sockaddr *)&sunx, SUN_LEN(&sunx)) < 0 || chmod(fx->name, fx->mode) < 0) { (void)snprintf(line, sizeof line, "cannot create %s", fx->name); logerror(line); dprintf("cannot create %s (%d)\n", fx->name, errno); if (fx == &funix_default || fx == &funix_secure) die(0); else { STAILQ_REMOVE(&funixes, fx, funix, next); continue; } } increase_rcvbuf(fx->s); } if (SecureMode <= 1) finet = socksetup(family, bindhostname); if (finet) { if (SecureMode) { for (i = 0; i < *finet; i++) { if (shutdown(finet[i+1], SHUT_RD) < 0 && errno != ENOTCONN) { logerror("shutdown"); if (!Debug) die(0); } } } else { dprintf("listening on inet and/or inet6 socket\n"); } dprintf("sending on inet and/or inet6 socket\n"); } if ((fklog = open(_PATH_KLOG, O_RDONLY, 0)) >= 0) if (fcntl(fklog, F_SETFL, O_NONBLOCK) < 0) fklog = -1; if (fklog < 0) dprintf("can't open %s (%d)\n", _PATH_KLOG, errno); /* tuck my process id away */ pidfile_write(pfh); dprintf("off & running....\n"); init(0); /* prevent SIGHUP and SIGCHLD handlers from running in parallel */ sigemptyset(&mask); sigaddset(&mask, SIGCHLD); sact.sa_handler = init; sact.sa_mask = mask; sact.sa_flags = SA_RESTART; (void)sigaction(SIGHUP, &sact, NULL); tvp = &tv; tv.tv_sec = tv.tv_usec = 0; if (fklog != -1 && fklog > fdsrmax) fdsrmax = fklog; if (finet && !SecureMode) { for (i = 0; i < *finet; i++) { if (finet[i+1] != -1 && finet[i+1] > fdsrmax) fdsrmax = finet[i+1]; } } STAILQ_FOREACH(fx, &funixes, next) if (fx->s > fdsrmax) fdsrmax = fx->s; fdsr = (fd_set *)calloc(howmany(fdsrmax+1, NFDBITS), sizeof(fd_mask)); if (fdsr == NULL) errx(1, "calloc fd_set"); for (;;) { if (MarkSet) markit(); if (WantDie) die(WantDie); bzero(fdsr, howmany(fdsrmax+1, NFDBITS) * sizeof(fd_mask)); if (fklog != -1) FD_SET(fklog, fdsr); if (finet && !SecureMode) { for (i = 0; i < *finet; i++) { if (finet[i+1] != -1) FD_SET(finet[i+1], fdsr); } } STAILQ_FOREACH(fx, &funixes, next) FD_SET(fx->s, fdsr); i = select(fdsrmax+1, fdsr, NULL, NULL, needdofsync ? &tv : tvp); switch (i) { case 0: dofsync(); needdofsync = 0; if (tvp) { tvp = NULL; if (ppid != 1) kill(ppid, SIGALRM); } continue; case -1: if (errno != EINTR) logerror("select"); continue; } if (fklog != -1 && FD_ISSET(fklog, fdsr)) readklog(); if (finet && !SecureMode) { for (i = 0; i < *finet; i++) { if (FD_ISSET(finet[i+1], fdsr)) { len = sizeof(frominet); l = recvfrom(finet[i+1], line, MAXLINE, 0, (struct sockaddr *)&frominet, &len); if (l > 0) { line[l] = '\0'; hname = cvthname((struct sockaddr *)&frominet); unmapped((struct sockaddr *)&frominet); if (validate((struct sockaddr *)&frominet, hname)) printline(hname, line, RemoteAddDate ? ADDDATE : 0); } else if (l < 0 && errno != EINTR) logerror("recvfrom inet"); } } } STAILQ_FOREACH(fx, &funixes, next) { if (FD_ISSET(fx->s, fdsr)) { len = sizeof(fromunix); l = recvfrom(fx->s, line, MAXLINE, 0, (struct sockaddr *)&fromunix, &len); if (l > 0) { line[l] = '\0'; printline(LocalHostName, line, 0); } else if (l < 0 && errno != EINTR) logerror("recvfrom unix"); } } } if (fdsr) free(fdsr); } static void unmapped(struct sockaddr *sa) { struct sockaddr_in6 *sin6; struct sockaddr_in sin4; if (sa->sa_family != AF_INET6) return; if (sa->sa_len != sizeof(struct sockaddr_in6) || sizeof(sin4) > sa->sa_len) return; sin6 = (struct sockaddr_in6 *)sa; if (!IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) return; memset(&sin4, 0, sizeof(sin4)); sin4.sin_family = AF_INET; sin4.sin_len = sizeof(struct sockaddr_in); memcpy(&sin4.sin_addr, &sin6->sin6_addr.s6_addr[12], sizeof(sin4.sin_addr)); sin4.sin_port = sin6->sin6_port; memcpy(sa, &sin4, sin4.sin_len); } static void usage(void) { fprintf(stderr, "%s\n%s\n%s\n%s\n", - "usage: syslogd [-468ACcdknosTuv] [-a allowed_peer]", + "usage: syslogd [-468ACcdFknosTuv] [-a allowed_peer]", " [-b bind_address] [-f config_file]", " [-l [mode:]path] [-m mark_interval]", " [-P pid_file] [-p log_socket]"); exit(1); } /* * Take a raw input line, decode the message, and print the message * on the appropriate log files. */ static void printline(const char *hname, char *msg, int flags) { char *p, *q; long n; int c, pri; char line[MAXLINE + 1]; /* test for special codes */ p = msg; pri = DEFUPRI; if (*p == '<') { errno = 0; n = strtol(p + 1, &q, 10); if (*q == '>' && n >= 0 && n < INT_MAX && errno == 0) { p = q + 1; pri = n; } } if (pri &~ (LOG_FACMASK|LOG_PRIMASK)) pri = DEFUPRI; /* * Don't allow users to log kernel messages. * NOTE: since LOG_KERN == 0 this will also match * messages with no facility specified. */ if ((pri & LOG_FACMASK) == LOG_KERN && !KeepKernFac) pri = LOG_MAKEPRI(LOG_USER, LOG_PRI(pri)); q = line; while ((c = (unsigned char)*p++) != '\0' && q < &line[sizeof(line) - 4]) { if (mask_C1 && (c & 0x80) && c < 0xA0) { c &= 0x7F; *q++ = 'M'; *q++ = '-'; } if (isascii(c) && iscntrl(c)) { if (c == '\n') { *q++ = ' '; } else if (c == '\t') { *q++ = '\t'; } else { *q++ = '^'; *q++ = c ^ 0100; } } else { *q++ = c; } } *q = '\0'; logmsg(pri, line, hname, flags); } /* * Read /dev/klog while data are available, split into lines. */ static void readklog(void) { char *p, *q, line[MAXLINE + 1]; int len, i; len = 0; for (;;) { i = read(fklog, line + len, MAXLINE - 1 - len); if (i > 0) { line[i + len] = '\0'; } else { if (i < 0 && errno != EINTR && errno != EAGAIN) { logerror("klog"); fklog = -1; } break; } for (p = line; (q = strchr(p, '\n')) != NULL; p = q + 1) { *q = '\0'; printsys(p); } len = strlen(p); if (len >= MAXLINE - 1) { printsys(p); len = 0; } if (len > 0) memmove(line, p, len + 1); } if (len > 0) printsys(line); } /* * Take a raw input line from /dev/klog, format similar to syslog(). */ static void printsys(char *msg) { char *p, *q; long n; int flags, isprintf, pri; flags = ISKERNEL | SYNC_FILE | ADDDATE; /* fsync after write */ p = msg; pri = DEFSPRI; isprintf = 1; if (*p == '<') { errno = 0; n = strtol(p + 1, &q, 10); if (*q == '>' && n >= 0 && n < INT_MAX && errno == 0) { p = q + 1; pri = n; isprintf = 0; } } /* * Kernel printf's and LOG_CONSOLE messages have been displayed * on the console already. */ if (isprintf || (pri & LOG_FACMASK) == LOG_CONSOLE) flags |= IGN_CONS; if (pri &~ (LOG_FACMASK|LOG_PRIMASK)) pri = DEFSPRI; logmsg(pri, p, LocalHostName, flags); } static time_t now; /* * Match a program or host name against a specification. * Return a non-0 value if the message must be ignored * based on the specification. */ static int skip_message(const char *name, const char *spec, int checkcase) { const char *s; char prev, next; int exclude = 0; /* Behaviour on explicit match */ if (spec == NULL) return 0; switch (*spec) { case '-': exclude = 1; /*FALLTHROUGH*/ case '+': spec++; break; default: break; } if (checkcase) s = strstr (spec, name); else s = strcasestr (spec, name); if (s != NULL) { prev = (s == spec ? ',' : *(s - 1)); next = *(s + strlen (name)); if (prev == ',' && (next == '\0' || next == ',')) /* Explicit match: skip iff the spec is an exclusive one. */ return exclude; } /* No explicit match for this name: skip the message iff the spec is an inclusive one. */ return !exclude; } /* * Log a message to the appropriate log files, users, etc. based on * the priority. */ static void logmsg(int pri, const char *msg, const char *from, int flags) { struct filed *f; int i, fac, msglen, omask, prilev; const char *timestamp; char prog[NAME_MAX+1]; char buf[MAXLINE+1]; dprintf("logmsg: pri %o, flags %x, from %s, msg %s\n", pri, flags, from, msg); omask = sigblock(sigmask(SIGHUP)|sigmask(SIGALRM)); /* * Check to see if msg looks non-standard. */ msglen = strlen(msg); if (msglen < 16 || msg[3] != ' ' || msg[6] != ' ' || msg[9] != ':' || msg[12] != ':' || msg[15] != ' ') flags |= ADDDATE; (void)time(&now); if (flags & ADDDATE) { timestamp = ctime(&now) + 4; } else { timestamp = msg; msg += 16; msglen -= 16; } /* skip leading blanks */ while (isspace(*msg)) { msg++; msglen--; } /* extract facility and priority level */ if (flags & MARK) fac = LOG_NFACILITIES; else fac = LOG_FAC(pri); /* Check maximum facility number. */ if (fac > LOG_NFACILITIES) { (void)sigsetmask(omask); return; } prilev = LOG_PRI(pri); /* extract program name */ for (i = 0; i < NAME_MAX; i++) { if (!isprint(msg[i]) || msg[i] == ':' || msg[i] == '[' || msg[i] == '/' || isspace(msg[i])) break; prog[i] = msg[i]; } prog[i] = 0; /* add kernel prefix for kernel messages */ if (flags & ISKERNEL) { snprintf(buf, sizeof(buf), "%s: %s", use_bootfile ? bootfile : "kernel", msg); msg = buf; msglen = strlen(buf); } /* log the message to the particular outputs */ if (!Initialized) { f = &consfile; /* * Open in non-blocking mode to avoid hangs during open * and close(waiting for the port to drain). */ f->f_file = open(ctty, O_WRONLY | O_NONBLOCK, 0); if (f->f_file >= 0) { (void)strlcpy(f->f_lasttime, timestamp, sizeof(f->f_lasttime)); fprintlog(f, flags, msg); (void)close(f->f_file); } (void)sigsetmask(omask); return; } for (f = Files; f; f = f->f_next) { /* skip messages that are incorrect priority */ if (!(((f->f_pcmp[fac] & PRI_EQ) && (f->f_pmask[fac] == prilev)) ||((f->f_pcmp[fac] & PRI_LT) && (f->f_pmask[fac] < prilev)) ||((f->f_pcmp[fac] & PRI_GT) && (f->f_pmask[fac] > prilev)) ) || f->f_pmask[fac] == INTERNAL_NOPRI) continue; /* skip messages with the incorrect hostname */ if (skip_message(from, f->f_host, 0)) continue; /* skip messages with the incorrect program name */ if (skip_message(prog, f->f_program, 1)) continue; /* skip message to console if it has already been printed */ if (f->f_type == F_CONSOLE && (flags & IGN_CONS)) continue; /* don't output marks to recently written files */ if ((flags & MARK) && (now - f->f_time) < MarkInterval / 2) continue; /* * suppress duplicate lines to this file */ if (no_compress - (f->f_type != F_PIPE) < 1 && (flags & MARK) == 0 && msglen == f->f_prevlen && !strcmp(msg, f->f_prevline) && !strcasecmp(from, f->f_prevhost)) { (void)strlcpy(f->f_lasttime, timestamp, sizeof(f->f_lasttime)); f->f_prevcount++; dprintf("msg repeated %d times, %ld sec of %d\n", f->f_prevcount, (long)(now - f->f_time), repeatinterval[f->f_repeatcount]); /* * If domark would have logged this by now, * flush it now (so we don't hold isolated messages), * but back off so we'll flush less often * in the future. */ if (now > REPEATTIME(f)) { fprintlog(f, flags, (char *)NULL); BACKOFF(f); } } else { /* new line, save it */ if (f->f_prevcount) fprintlog(f, 0, (char *)NULL); f->f_repeatcount = 0; f->f_prevpri = pri; (void)strlcpy(f->f_lasttime, timestamp, sizeof(f->f_lasttime)); (void)strlcpy(f->f_prevhost, from, sizeof(f->f_prevhost)); if (msglen < MAXSVLINE) { f->f_prevlen = msglen; (void)strlcpy(f->f_prevline, msg, sizeof(f->f_prevline)); fprintlog(f, flags, (char *)NULL); } else { f->f_prevline[0] = 0; f->f_prevlen = 0; fprintlog(f, flags, msg); } } } (void)sigsetmask(omask); } static void dofsync(void) { struct filed *f; for (f = Files; f; f = f->f_next) { if ((f->f_type == F_FILE) && (f->f_flags & FFLAG_NEEDSYNC)) { f->f_flags &= ~FFLAG_NEEDSYNC; (void)fsync(f->f_file); } } } #define IOV_SIZE 7 static void fprintlog(struct filed *f, int flags, const char *msg) { struct iovec iov[IOV_SIZE]; struct iovec *v; struct addrinfo *r; int i, l, lsent = 0; char line[MAXLINE + 1], repbuf[80], greetings[200], *wmsg = NULL; char nul[] = "", space[] = " ", lf[] = "\n", crlf[] = "\r\n"; const char *msgret; v = iov; if (f->f_type == F_WALL) { v->iov_base = greetings; /* The time displayed is not synchornized with the other log * destinations (like messages). Following fragment was using * ctime(&now), which was updating the time every 30 sec. * With f_lasttime, time is synchronized correctly. */ v->iov_len = snprintf(greetings, sizeof greetings, "\r\n\7Message from syslogd@%s at %.24s ...\r\n", f->f_prevhost, f->f_lasttime); if (v->iov_len >= sizeof greetings) v->iov_len = sizeof greetings - 1; v++; v->iov_base = nul; v->iov_len = 0; v++; } else { v->iov_base = f->f_lasttime; v->iov_len = strlen(f->f_lasttime); v++; v->iov_base = space; v->iov_len = 1; v++; } if (LogFacPri) { static char fp_buf[30]; /* Hollow laugh */ int fac = f->f_prevpri & LOG_FACMASK; int pri = LOG_PRI(f->f_prevpri); const char *f_s = NULL; char f_n[5]; /* Hollow laugh */ const char *p_s = NULL; char p_n[5]; /* Hollow laugh */ if (LogFacPri > 1) { const CODE *c; for (c = facilitynames; c->c_name; c++) { if (c->c_val == fac) { f_s = c->c_name; break; } } for (c = prioritynames; c->c_name; c++) { if (c->c_val == pri) { p_s = c->c_name; break; } } } if (!f_s) { snprintf(f_n, sizeof f_n, "%d", LOG_FAC(fac)); f_s = f_n; } if (!p_s) { snprintf(p_n, sizeof p_n, "%d", pri); p_s = p_n; } snprintf(fp_buf, sizeof fp_buf, "<%s.%s> ", f_s, p_s); v->iov_base = fp_buf; v->iov_len = strlen(fp_buf); } else { v->iov_base = nul; v->iov_len = 0; } v++; v->iov_base = f->f_prevhost; v->iov_len = strlen(v->iov_base); v++; v->iov_base = space; v->iov_len = 1; v++; if (msg) { wmsg = strdup(msg); /* XXX iov_base needs a `const' sibling. */ if (wmsg == NULL) { logerror("strdup"); exit(1); } v->iov_base = wmsg; v->iov_len = strlen(msg); } else if (f->f_prevcount > 1) { v->iov_base = repbuf; v->iov_len = snprintf(repbuf, sizeof repbuf, "last message repeated %d times", f->f_prevcount); } else { v->iov_base = f->f_prevline; v->iov_len = f->f_prevlen; } v++; dprintf("Logging to %s", TypeNames[f->f_type]); f->f_time = now; switch (f->f_type) { int port; case F_UNUSED: dprintf("\n"); break; case F_FORW: port = (int)ntohs(((struct sockaddr_in *) (f->f_un.f_forw.f_addr->ai_addr))->sin_port); if (port != 514) { dprintf(" %s:%d\n", f->f_un.f_forw.f_hname, port); } else { dprintf(" %s\n", f->f_un.f_forw.f_hname); } /* check for local vs remote messages */ if (strcasecmp(f->f_prevhost, LocalHostName)) l = snprintf(line, sizeof line - 1, "<%d>%.15s Forwarded from %s: %s", f->f_prevpri, (char *)iov[0].iov_base, f->f_prevhost, (char *)iov[5].iov_base); else l = snprintf(line, sizeof line - 1, "<%d>%.15s %s", f->f_prevpri, (char *)iov[0].iov_base, (char *)iov[5].iov_base); if (l < 0) l = 0; else if (l > MAXLINE) l = MAXLINE; if (finet) { for (r = f->f_un.f_forw.f_addr; r; r = r->ai_next) { for (i = 0; i < *finet; i++) { #if 0 /* * should we check AF first, or just * trial and error? FWD */ if (r->ai_family == address_family_of(finet[i+1])) #endif lsent = sendto(finet[i+1], line, l, 0, r->ai_addr, r->ai_addrlen); if (lsent == l) break; } if (lsent == l && !send_to_all) break; } dprintf("lsent/l: %d/%d\n", lsent, l); if (lsent != l) { int e = errno; logerror("sendto"); errno = e; switch (errno) { case ENOBUFS: case ENETDOWN: case ENETUNREACH: case EHOSTUNREACH: case EHOSTDOWN: case EADDRNOTAVAIL: break; /* case EBADF: */ /* case EACCES: */ /* case ENOTSOCK: */ /* case EFAULT: */ /* case EMSGSIZE: */ /* case EAGAIN: */ /* case ENOBUFS: */ /* case ECONNREFUSED: */ default: dprintf("removing entry: errno=%d\n", e); f->f_type = F_UNUSED; break; } } } break; case F_FILE: dprintf(" %s\n", f->f_un.f_fname); v->iov_base = lf; v->iov_len = 1; if (writev(f->f_file, iov, IOV_SIZE) < 0) { /* * If writev(2) fails for potentially transient errors * like the filesystem being full, ignore it. * Otherwise remove this logfile from the list. */ if (errno != ENOSPC) { int e = errno; (void)close(f->f_file); f->f_type = F_UNUSED; errno = e; logerror(f->f_un.f_fname); } } else if ((flags & SYNC_FILE) && (f->f_flags & FFLAG_SYNC)) { f->f_flags |= FFLAG_NEEDSYNC; needdofsync = 1; } break; case F_PIPE: dprintf(" %s\n", f->f_un.f_pipe.f_pname); v->iov_base = lf; v->iov_len = 1; if (f->f_un.f_pipe.f_pid == 0) { if ((f->f_file = p_open(f->f_un.f_pipe.f_pname, &f->f_un.f_pipe.f_pid)) < 0) { f->f_type = F_UNUSED; logerror(f->f_un.f_pipe.f_pname); break; } } if (writev(f->f_file, iov, IOV_SIZE) < 0) { int e = errno; (void)close(f->f_file); if (f->f_un.f_pipe.f_pid > 0) deadq_enter(f->f_un.f_pipe.f_pid, f->f_un.f_pipe.f_pname); f->f_un.f_pipe.f_pid = 0; errno = e; logerror(f->f_un.f_pipe.f_pname); } break; case F_CONSOLE: if (flags & IGN_CONS) { dprintf(" (ignored)\n"); break; } /* FALLTHROUGH */ case F_TTY: dprintf(" %s%s\n", _PATH_DEV, f->f_un.f_fname); v->iov_base = crlf; v->iov_len = 2; errno = 0; /* ttymsg() only sometimes returns an errno */ if ((msgret = ttymsg(iov, IOV_SIZE, f->f_un.f_fname, 10))) { f->f_type = F_UNUSED; logerror(msgret); } break; case F_USERS: case F_WALL: dprintf("\n"); v->iov_base = crlf; v->iov_len = 2; wallmsg(f, iov, IOV_SIZE); break; } f->f_prevcount = 0; free(wmsg); } /* * WALLMSG -- Write a message to the world at large * * Write the specified message to either the entire * world, or a list of approved users. */ static void wallmsg(struct filed *f, struct iovec *iov, const int iovlen) { static int reenter; /* avoid calling ourselves */ struct utmpx *ut; int i; const char *p; if (reenter++) return; setutxent(); /* NOSTRICT */ while ((ut = getutxent()) != NULL) { if (ut->ut_type != USER_PROCESS) continue; if (f->f_type == F_WALL) { if ((p = ttymsg(iov, iovlen, ut->ut_line, TTYMSGTIME)) != NULL) { errno = 0; /* already in msg */ logerror(p); } continue; } /* should we send the message to this user? */ for (i = 0; i < MAXUNAMES; i++) { if (!f->f_un.f_uname[i][0]) break; if (!strcmp(f->f_un.f_uname[i], ut->ut_user)) { if ((p = ttymsg(iov, iovlen, ut->ut_line, TTYMSGTIME)) != NULL) { errno = 0; /* already in msg */ logerror(p); } break; } } } endutxent(); reenter = 0; } static void reapchild(int signo __unused) { int status; pid_t pid; struct filed *f; while ((pid = wait3(&status, WNOHANG, (struct rusage *)NULL)) > 0) { if (!Initialized) /* Don't tell while we are initting. */ continue; /* First, look if it's a process from the dead queue. */ if (deadq_remove(pid)) goto oncemore; /* Now, look in list of active processes. */ for (f = Files; f; f = f->f_next) if (f->f_type == F_PIPE && f->f_un.f_pipe.f_pid == pid) { (void)close(f->f_file); f->f_un.f_pipe.f_pid = 0; log_deadchild(pid, status, f->f_un.f_pipe.f_pname); break; } oncemore: continue; } } /* * Return a printable representation of a host address. */ static const char * cvthname(struct sockaddr *f) { int error, hl; sigset_t omask, nmask; static char hname[NI_MAXHOST], ip[NI_MAXHOST]; error = getnameinfo((struct sockaddr *)f, ((struct sockaddr *)f)->sa_len, ip, sizeof ip, NULL, 0, NI_NUMERICHOST); dprintf("cvthname(%s)\n", ip); if (error) { dprintf("Malformed from address %s\n", gai_strerror(error)); return ("???"); } if (!resolve) return (ip); sigemptyset(&nmask); sigaddset(&nmask, SIGHUP); sigprocmask(SIG_BLOCK, &nmask, &omask); error = getnameinfo((struct sockaddr *)f, ((struct sockaddr *)f)->sa_len, hname, sizeof hname, NULL, 0, NI_NAMEREQD); sigprocmask(SIG_SETMASK, &omask, NULL); if (error) { dprintf("Host name for your address (%s) unknown\n", ip); return (ip); } hl = strlen(hname); if (hl > 0 && hname[hl-1] == '.') hname[--hl] = '\0'; trimdomain(hname, hl); return (hname); } static void dodie(int signo) { WantDie = signo; } static void domark(int signo __unused) { MarkSet = 1; } /* * Print syslogd errors some place. */ static void logerror(const char *type) { char buf[512]; static int recursed = 0; /* If there's an error while trying to log an error, give up. */ if (recursed) return; recursed++; if (errno) (void)snprintf(buf, sizeof buf, "syslogd: %s: %s", type, strerror(errno)); else (void)snprintf(buf, sizeof buf, "syslogd: %s", type); errno = 0; dprintf("%s\n", buf); logmsg(LOG_SYSLOG|LOG_ERR, buf, LocalHostName, ADDDATE); recursed--; } static void die(int signo) { struct filed *f; struct funix *fx; int was_initialized; char buf[100]; was_initialized = Initialized; Initialized = 0; /* Don't log SIGCHLDs. */ for (f = Files; f != NULL; f = f->f_next) { /* flush any pending output */ if (f->f_prevcount) fprintlog(f, 0, (char *)NULL); if (f->f_type == F_PIPE && f->f_un.f_pipe.f_pid > 0) { (void)close(f->f_file); f->f_un.f_pipe.f_pid = 0; } } Initialized = was_initialized; if (signo) { dprintf("syslogd: exiting on signal %d\n", signo); (void)snprintf(buf, sizeof(buf), "exiting on signal %d", signo); errno = 0; logerror(buf); } STAILQ_FOREACH(fx, &funixes, next) (void)unlink(fx->name); pidfile_remove(pfh); exit(1); } /* * INIT -- Initialize syslogd from configuration table */ static void init(int signo) { int i; FILE *cf; struct filed *f, *next, **nextp; char *p; char cline[LINE_MAX]; char prog[LINE_MAX]; char host[MAXHOSTNAMELEN]; char oldLocalHostName[MAXHOSTNAMELEN]; char hostMsg[2*MAXHOSTNAMELEN+40]; char bootfileMsg[LINE_MAX]; dprintf("init\n"); /* * Load hostname (may have changed). */ if (signo != 0) (void)strlcpy(oldLocalHostName, LocalHostName, sizeof(oldLocalHostName)); if (gethostname(LocalHostName, sizeof(LocalHostName))) err(EX_OSERR, "gethostname() failed"); if ((p = strchr(LocalHostName, '.')) != NULL) { *p++ = '\0'; LocalDomain = p; } else { LocalDomain = ""; } /* * Close all open log files. */ Initialized = 0; for (f = Files; f != NULL; f = next) { /* flush any pending output */ if (f->f_prevcount) fprintlog(f, 0, (char *)NULL); switch (f->f_type) { case F_FILE: case F_FORW: case F_CONSOLE: case F_TTY: (void)close(f->f_file); break; case F_PIPE: if (f->f_un.f_pipe.f_pid > 0) { (void)close(f->f_file); deadq_enter(f->f_un.f_pipe.f_pid, f->f_un.f_pipe.f_pname); } f->f_un.f_pipe.f_pid = 0; break; } next = f->f_next; if (f->f_program) free(f->f_program); if (f->f_host) free(f->f_host); free((char *)f); } Files = NULL; nextp = &Files; /* open the configuration file */ if ((cf = fopen(ConfFile, "r")) == NULL) { dprintf("cannot open %s\n", ConfFile); *nextp = (struct filed *)calloc(1, sizeof(*f)); if (*nextp == NULL) { logerror("calloc"); exit(1); } cfline("*.ERR\t/dev/console", *nextp, "*", "*"); (*nextp)->f_next = (struct filed *)calloc(1, sizeof(*f)); if ((*nextp)->f_next == NULL) { logerror("calloc"); exit(1); } cfline("*.PANIC\t*", (*nextp)->f_next, "*", "*"); Initialized = 1; return; } /* * Foreach line in the conf table, open that file. */ f = NULL; (void)strlcpy(host, "*", sizeof(host)); (void)strlcpy(prog, "*", sizeof(prog)); while (fgets(cline, sizeof(cline), cf) != NULL) { /* * check for end-of-section, comments, strip off trailing * spaces and newline character. #!prog is treated specially: * following lines apply only to that program. */ for (p = cline; isspace(*p); ++p) continue; if (*p == 0) continue; if (*p == '#') { p++; if (*p != '!' && *p != '+' && *p != '-') continue; } if (*p == '+' || *p == '-') { host[0] = *p++; while (isspace(*p)) p++; if ((!*p) || (*p == '*')) { (void)strlcpy(host, "*", sizeof(host)); continue; } if (*p == '@') p = LocalHostName; for (i = 1; i < MAXHOSTNAMELEN - 1; i++) { if (!isalnum(*p) && *p != '.' && *p != '-' && *p != ',' && *p != ':' && *p != '%') break; host[i] = *p++; } host[i] = '\0'; continue; } if (*p == '!') { p++; while (isspace(*p)) p++; if ((!*p) || (*p == '*')) { (void)strlcpy(prog, "*", sizeof(prog)); continue; } for (i = 0; i < LINE_MAX - 1; i++) { if (!isprint(p[i]) || isspace(p[i])) break; prog[i] = p[i]; } prog[i] = 0; continue; } for (p = cline + 1; *p != '\0'; p++) { if (*p != '#') continue; if (*(p - 1) == '\\') { strcpy(p - 1, p); p--; continue; } *p = '\0'; break; } for (i = strlen(cline) - 1; i >= 0 && isspace(cline[i]); i--) cline[i] = '\0'; f = (struct filed *)calloc(1, sizeof(*f)); if (f == NULL) { logerror("calloc"); exit(1); } *nextp = f; nextp = &f->f_next; cfline(cline, f, prog, host); } /* close the configuration file */ (void)fclose(cf); Initialized = 1; if (Debug) { int port; for (f = Files; f; f = f->f_next) { for (i = 0; i <= LOG_NFACILITIES; i++) if (f->f_pmask[i] == INTERNAL_NOPRI) printf("X "); else printf("%d ", f->f_pmask[i]); printf("%s: ", TypeNames[f->f_type]); switch (f->f_type) { case F_FILE: printf("%s", f->f_un.f_fname); break; case F_CONSOLE: case F_TTY: printf("%s%s", _PATH_DEV, f->f_un.f_fname); break; case F_FORW: port = (int)ntohs(((struct sockaddr_in *) (f->f_un.f_forw.f_addr->ai_addr))->sin_port); if (port != 514) { printf("%s:%d", f->f_un.f_forw.f_hname, port); } else { printf("%s", f->f_un.f_forw.f_hname); } break; case F_PIPE: printf("%s", f->f_un.f_pipe.f_pname); break; case F_USERS: for (i = 0; i < MAXUNAMES && *f->f_un.f_uname[i]; i++) printf("%s, ", f->f_un.f_uname[i]); break; } if (f->f_program) printf(" (%s)", f->f_program); printf("\n"); } } logmsg(LOG_SYSLOG|LOG_INFO, "syslogd: restart", LocalHostName, ADDDATE); dprintf("syslogd: restarted\n"); /* * Log a change in hostname, but only on a restart. */ if (signo != 0 && strcmp(oldLocalHostName, LocalHostName) != 0) { (void)snprintf(hostMsg, sizeof(hostMsg), "syslogd: hostname changed, \"%s\" to \"%s\"", oldLocalHostName, LocalHostName); logmsg(LOG_SYSLOG|LOG_INFO, hostMsg, LocalHostName, ADDDATE); dprintf("%s\n", hostMsg); } /* * Log the kernel boot file if we aren't going to use it as * the prefix, and if this is *not* a restart. */ if (signo == 0 && !use_bootfile) { (void)snprintf(bootfileMsg, sizeof(bootfileMsg), "syslogd: kernel boot file is %s", bootfile); logmsg(LOG_KERN|LOG_INFO, bootfileMsg, LocalHostName, ADDDATE); dprintf("%s\n", bootfileMsg); } } /* * Crack a configuration file line */ static void cfline(const char *line, struct filed *f, const char *prog, const char *host) { struct addrinfo hints, *res; int error, i, pri, syncfile; const char *p, *q; char *bp; char buf[MAXLINE], ebuf[100]; dprintf("cfline(\"%s\", f, \"%s\", \"%s\")\n", line, prog, host); errno = 0; /* keep strerror() stuff out of logerror messages */ /* clear out file entry */ memset(f, 0, sizeof(*f)); for (i = 0; i <= LOG_NFACILITIES; i++) f->f_pmask[i] = INTERNAL_NOPRI; /* save hostname if any */ if (host && *host == '*') host = NULL; if (host) { int hl; f->f_host = strdup(host); if (f->f_host == NULL) { logerror("strdup"); exit(1); } hl = strlen(f->f_host); if (hl > 0 && f->f_host[hl-1] == '.') f->f_host[--hl] = '\0'; trimdomain(f->f_host, hl); } /* save program name if any */ if (prog && *prog == '*') prog = NULL; if (prog) { f->f_program = strdup(prog); if (f->f_program == NULL) { logerror("strdup"); exit(1); } } /* scan through the list of selectors */ for (p = line; *p && *p != '\t' && *p != ' ';) { int pri_done; int pri_cmp; int pri_invert; /* find the end of this facility name list */ for (q = p; *q && *q != '\t' && *q != ' ' && *q++ != '.'; ) continue; /* get the priority comparison */ pri_cmp = 0; pri_done = 0; pri_invert = 0; if (*q == '!') { pri_invert = 1; q++; } while (!pri_done) { switch (*q) { case '<': pri_cmp |= PRI_LT; q++; break; case '=': pri_cmp |= PRI_EQ; q++; break; case '>': pri_cmp |= PRI_GT; q++; break; default: pri_done++; break; } } /* collect priority name */ for (bp = buf; *q && !strchr("\t,; ", *q); ) *bp++ = *q++; *bp = '\0'; /* skip cruft */ while (strchr(",;", *q)) q++; /* decode priority name */ if (*buf == '*') { pri = LOG_PRIMASK; pri_cmp = PRI_LT | PRI_EQ | PRI_GT; } else { /* Ignore trailing spaces. */ for (i = strlen(buf) - 1; i >= 0 && buf[i] == ' '; i--) buf[i] = '\0'; pri = decode(buf, prioritynames); if (pri < 0) { errno = 0; (void)snprintf(ebuf, sizeof ebuf, "unknown priority name \"%s\"", buf); logerror(ebuf); return; } } if (!pri_cmp) pri_cmp = (UniquePriority) ? (PRI_EQ) : (PRI_EQ | PRI_GT) ; if (pri_invert) pri_cmp ^= PRI_LT | PRI_EQ | PRI_GT; /* scan facilities */ while (*p && !strchr("\t.; ", *p)) { for (bp = buf; *p && !strchr("\t,;. ", *p); ) *bp++ = *p++; *bp = '\0'; if (*buf == '*') { for (i = 0; i < LOG_NFACILITIES; i++) { f->f_pmask[i] = pri; f->f_pcmp[i] = pri_cmp; } } else { i = decode(buf, facilitynames); if (i < 0) { errno = 0; (void)snprintf(ebuf, sizeof ebuf, "unknown facility name \"%s\"", buf); logerror(ebuf); return; } f->f_pmask[i >> 3] = pri; f->f_pcmp[i >> 3] = pri_cmp; } while (*p == ',' || *p == ' ') p++; } p = q; } /* skip to action part */ while (*p == '\t' || *p == ' ') p++; if (*p == '-') { syncfile = 0; p++; } else syncfile = 1; switch (*p) { case '@': { char *tp; char endkey = ':'; /* * scan forward to see if there is a port defined. * so we can't use strlcpy.. */ i = sizeof(f->f_un.f_forw.f_hname); tp = f->f_un.f_forw.f_hname; p++; /* * an ipv6 address should start with a '[' in that case * we should scan for a ']' */ if (*p == '[') { p++; endkey = ']'; } while (*p && (*p != endkey) && (i-- > 0)) { *tp++ = *p++; } if (endkey == ']' && *p == endkey) p++; *tp = '\0'; } /* See if we copied a domain and have a port */ if (*p == ':') p++; else p = NULL; memset(&hints, 0, sizeof(hints)); hints.ai_family = family; hints.ai_socktype = SOCK_DGRAM; error = getaddrinfo(f->f_un.f_forw.f_hname, p ? p : "syslog", &hints, &res); if (error) { logerror(gai_strerror(error)); break; } f->f_un.f_forw.f_addr = res; f->f_type = F_FORW; break; case '/': if ((f->f_file = open(p, logflags, 0600)) < 0) { f->f_type = F_UNUSED; logerror(p); break; } if (syncfile) f->f_flags |= FFLAG_SYNC; if (isatty(f->f_file)) { if (strcmp(p, ctty) == 0) f->f_type = F_CONSOLE; else f->f_type = F_TTY; (void)strlcpy(f->f_un.f_fname, p + sizeof(_PATH_DEV) - 1, sizeof(f->f_un.f_fname)); } else { (void)strlcpy(f->f_un.f_fname, p, sizeof(f->f_un.f_fname)); f->f_type = F_FILE; } break; case '|': f->f_un.f_pipe.f_pid = 0; (void)strlcpy(f->f_un.f_pipe.f_pname, p + 1, sizeof(f->f_un.f_pipe.f_pname)); f->f_type = F_PIPE; break; case '*': f->f_type = F_WALL; break; default: for (i = 0; i < MAXUNAMES && *p; i++) { for (q = p; *q && *q != ','; ) q++; (void)strncpy(f->f_un.f_uname[i], p, MAXLOGNAME - 1); if ((q - p) >= MAXLOGNAME) f->f_un.f_uname[i][MAXLOGNAME - 1] = '\0'; else f->f_un.f_uname[i][q - p] = '\0'; while (*q == ',' || *q == ' ') q++; p = q; } f->f_type = F_USERS; break; } } /* * Decode a symbolic name to a numeric value */ static int decode(const char *name, const CODE *codetab) { const CODE *c; char *p, buf[40]; if (isdigit(*name)) return (atoi(name)); for (p = buf; *name && p < &buf[sizeof(buf) - 1]; p++, name++) { if (isupper(*name)) *p = tolower(*name); else *p = *name; } *p = '\0'; for (c = codetab; c->c_name; c++) if (!strcmp(buf, c->c_name)) return (c->c_val); return (-1); } static void markit(void) { struct filed *f; dq_t q, next; now = time((time_t *)NULL); MarkSeq += TIMERINTVL; if (MarkSeq >= MarkInterval) { logmsg(LOG_INFO, "-- MARK --", LocalHostName, ADDDATE|MARK); MarkSeq = 0; } for (f = Files; f; f = f->f_next) { if (f->f_prevcount && now >= REPEATTIME(f)) { dprintf("flush %s: repeated %d times, %d sec.\n", TypeNames[f->f_type], f->f_prevcount, repeatinterval[f->f_repeatcount]); fprintlog(f, 0, (char *)NULL); BACKOFF(f); } } /* Walk the dead queue, and see if we should signal somebody. */ for (q = TAILQ_FIRST(&deadq_head); q != NULL; q = next) { next = TAILQ_NEXT(q, dq_entries); switch (q->dq_timeout) { case 0: /* Already signalled once, try harder now. */ if (kill(q->dq_pid, SIGKILL) != 0) (void)deadq_remove(q->dq_pid); break; case 1: /* * Timed out on dead queue, send terminate * signal. Note that we leave the removal * from the dead queue to reapchild(), which * will also log the event (unless the process * didn't even really exist, in case we simply * drop it from the dead queue). */ if (kill(q->dq_pid, SIGTERM) != 0) (void)deadq_remove(q->dq_pid); /* FALLTHROUGH */ default: q->dq_timeout--; } } MarkSet = 0; (void)alarm(TIMERINTVL); } /* * fork off and become a daemon, but wait for the child to come online * before returing to the parent, or we get disk thrashing at boot etc. * Set a timer so we don't hang forever if it wedges. */ static int waitdaemon(int nochdir, int noclose, int maxwait) { int fd; int status; pid_t pid, childpid; switch (childpid = fork()) { case -1: return (-1); case 0: break; default: signal(SIGALRM, timedout); alarm(maxwait); while ((pid = wait3(&status, 0, NULL)) != -1) { if (WIFEXITED(status)) errx(1, "child pid %d exited with return code %d", pid, WEXITSTATUS(status)); if (WIFSIGNALED(status)) errx(1, "child pid %d exited on signal %d%s", pid, WTERMSIG(status), WCOREDUMP(status) ? " (core dumped)" : ""); if (pid == childpid) /* it's gone... */ break; } exit(0); } if (setsid() == -1) return (-1); if (!nochdir) (void)chdir("/"); if (!noclose && (fd = open(_PATH_DEVNULL, O_RDWR, 0)) != -1) { (void)dup2(fd, STDIN_FILENO); (void)dup2(fd, STDOUT_FILENO); (void)dup2(fd, STDERR_FILENO); if (fd > 2) (void)close (fd); } return (getppid()); } /* * We get a SIGALRM from the child when it's running and finished doing it's * fsync()'s or O_SYNC writes for all the boot messages. * * We also get a signal from the kernel if the timer expires, so check to * see what happened. */ static void timedout(int sig __unused) { int left; left = alarm(0); signal(SIGALRM, SIG_DFL); if (left == 0) errx(1, "timed out waiting for child"); else _exit(0); } /* * Add `s' to the list of allowable peer addresses to accept messages * from. * * `s' is a string in the form: * * [*]domainname[:{servicename|portnumber|*}] * * or * * netaddr/maskbits[:{servicename|portnumber|*}] * * Returns -1 on error, 0 if the argument was valid. */ static int allowaddr(char *s) { char *cp1, *cp2; struct allowedpeer ap; struct servent *se; int masklen = -1; struct addrinfo hints, *res; struct in_addr *addrp, *maskp; #ifdef INET6 int i; u_int32_t *addr6p, *mask6p; #endif char ip[NI_MAXHOST]; #ifdef INET6 if (*s != '[' || (cp1 = strchr(s + 1, ']')) == NULL) #endif cp1 = s; if ((cp1 = strrchr(cp1, ':'))) { /* service/port provided */ *cp1++ = '\0'; if (strlen(cp1) == 1 && *cp1 == '*') /* any port allowed */ ap.port = 0; else if ((se = getservbyname(cp1, "udp"))) { ap.port = ntohs(se->s_port); } else { ap.port = strtol(cp1, &cp2, 0); if (*cp2 != '\0') return (-1); /* port not numeric */ } } else { if ((se = getservbyname("syslog", "udp"))) ap.port = ntohs(se->s_port); else /* sanity, should not happen */ ap.port = 514; } if ((cp1 = strchr(s, '/')) != NULL && strspn(cp1 + 1, "0123456789") == strlen(cp1 + 1)) { *cp1 = '\0'; if ((masklen = atoi(cp1 + 1)) < 0) return (-1); } #ifdef INET6 if (*s == '[') { cp2 = s + strlen(s) - 1; if (*cp2 == ']') { ++s; *cp2 = '\0'; } else { cp2 = NULL; } } else { cp2 = NULL; } #endif memset(&hints, 0, sizeof(hints)); hints.ai_family = PF_UNSPEC; hints.ai_socktype = SOCK_DGRAM; hints.ai_flags = AI_PASSIVE | AI_NUMERICHOST; if (getaddrinfo(s, NULL, &hints, &res) == 0) { ap.isnumeric = 1; memcpy(&ap.a_addr, res->ai_addr, res->ai_addrlen); memset(&ap.a_mask, 0, sizeof(ap.a_mask)); ap.a_mask.ss_family = res->ai_family; if (res->ai_family == AF_INET) { ap.a_mask.ss_len = sizeof(struct sockaddr_in); maskp = &((struct sockaddr_in *)&ap.a_mask)->sin_addr; addrp = &((struct sockaddr_in *)&ap.a_addr)->sin_addr; if (masklen < 0) { /* use default netmask */ if (IN_CLASSA(ntohl(addrp->s_addr))) maskp->s_addr = htonl(IN_CLASSA_NET); else if (IN_CLASSB(ntohl(addrp->s_addr))) maskp->s_addr = htonl(IN_CLASSB_NET); else maskp->s_addr = htonl(IN_CLASSC_NET); } else if (masklen <= 32) { /* convert masklen to netmask */ if (masklen == 0) maskp->s_addr = 0; else maskp->s_addr = htonl(~((1 << (32 - masklen)) - 1)); } else { freeaddrinfo(res); return (-1); } /* Lose any host bits in the network number. */ addrp->s_addr &= maskp->s_addr; } #ifdef INET6 else if (res->ai_family == AF_INET6 && masklen <= 128) { ap.a_mask.ss_len = sizeof(struct sockaddr_in6); if (masklen < 0) masklen = 128; mask6p = (u_int32_t *)&((struct sockaddr_in6 *)&ap.a_mask)->sin6_addr; /* convert masklen to netmask */ while (masklen > 0) { if (masklen < 32) { *mask6p = htonl(~(0xffffffff >> masklen)); break; } *mask6p++ = 0xffffffff; masklen -= 32; } /* Lose any host bits in the network number. */ mask6p = (u_int32_t *)&((struct sockaddr_in6 *)&ap.a_mask)->sin6_addr; addr6p = (u_int32_t *)&((struct sockaddr_in6 *)&ap.a_addr)->sin6_addr; for (i = 0; i < 4; i++) addr6p[i] &= mask6p[i]; } #endif else { freeaddrinfo(res); return (-1); } freeaddrinfo(res); } else { /* arg `s' is domain name */ ap.isnumeric = 0; ap.a_name = s; if (cp1) *cp1 = '/'; #ifdef INET6 if (cp2) { *cp2 = ']'; --s; } #endif } if (Debug) { printf("allowaddr: rule %d: ", NumAllowed); if (ap.isnumeric) { printf("numeric, "); getnameinfo((struct sockaddr *)&ap.a_addr, ((struct sockaddr *)&ap.a_addr)->sa_len, ip, sizeof ip, NULL, 0, NI_NUMERICHOST); printf("addr = %s, ", ip); getnameinfo((struct sockaddr *)&ap.a_mask, ((struct sockaddr *)&ap.a_mask)->sa_len, ip, sizeof ip, NULL, 0, NI_NUMERICHOST); printf("mask = %s; ", ip); } else { printf("domainname = %s; ", ap.a_name); } printf("port = %d\n", ap.port); } if ((AllowedPeers = realloc(AllowedPeers, ++NumAllowed * sizeof(struct allowedpeer))) == NULL) { logerror("realloc"); exit(1); } memcpy(&AllowedPeers[NumAllowed - 1], &ap, sizeof(struct allowedpeer)); return (0); } /* * Validate that the remote peer has permission to log to us. */ static int validate(struct sockaddr *sa, const char *hname) { int i; size_t l1, l2; char *cp, name[NI_MAXHOST], ip[NI_MAXHOST], port[NI_MAXSERV]; struct allowedpeer *ap; struct sockaddr_in *sin4, *a4p = NULL, *m4p = NULL; #ifdef INET6 int j, reject; struct sockaddr_in6 *sin6, *a6p = NULL, *m6p = NULL; #endif struct addrinfo hints, *res; u_short sport; if (NumAllowed == 0) /* traditional behaviour, allow everything */ return (1); (void)strlcpy(name, hname, sizeof(name)); memset(&hints, 0, sizeof(hints)); hints.ai_family = PF_UNSPEC; hints.ai_socktype = SOCK_DGRAM; hints.ai_flags = AI_PASSIVE | AI_NUMERICHOST; if (getaddrinfo(name, NULL, &hints, &res) == 0) freeaddrinfo(res); else if (strchr(name, '.') == NULL) { strlcat(name, ".", sizeof name); strlcat(name, LocalDomain, sizeof name); } if (getnameinfo(sa, sa->sa_len, ip, sizeof ip, port, sizeof port, NI_NUMERICHOST | NI_NUMERICSERV) != 0) return (0); /* for safety, should not occur */ dprintf("validate: dgram from IP %s, port %s, name %s;\n", ip, port, name); sport = atoi(port); /* now, walk down the list */ for (i = 0, ap = AllowedPeers; i < NumAllowed; i++, ap++) { if (ap->port != 0 && ap->port != sport) { dprintf("rejected in rule %d due to port mismatch.\n", i); continue; } if (ap->isnumeric) { if (ap->a_addr.ss_family != sa->sa_family) { dprintf("rejected in rule %d due to address family mismatch.\n", i); continue; } if (ap->a_addr.ss_family == AF_INET) { sin4 = (struct sockaddr_in *)sa; a4p = (struct sockaddr_in *)&ap->a_addr; m4p = (struct sockaddr_in *)&ap->a_mask; if ((sin4->sin_addr.s_addr & m4p->sin_addr.s_addr) != a4p->sin_addr.s_addr) { dprintf("rejected in rule %d due to IP mismatch.\n", i); continue; } } #ifdef INET6 else if (ap->a_addr.ss_family == AF_INET6) { sin6 = (struct sockaddr_in6 *)sa; a6p = (struct sockaddr_in6 *)&ap->a_addr; m6p = (struct sockaddr_in6 *)&ap->a_mask; if (a6p->sin6_scope_id != 0 && sin6->sin6_scope_id != a6p->sin6_scope_id) { dprintf("rejected in rule %d due to scope mismatch.\n", i); continue; } reject = 0; for (j = 0; j < 16; j += 4) { if ((*(u_int32_t *)&sin6->sin6_addr.s6_addr[j] & *(u_int32_t *)&m6p->sin6_addr.s6_addr[j]) != *(u_int32_t *)&a6p->sin6_addr.s6_addr[j]) { ++reject; break; } } if (reject) { dprintf("rejected in rule %d due to IP mismatch.\n", i); continue; } } #endif else continue; } else { cp = ap->a_name; l1 = strlen(name); if (*cp == '*') { /* allow wildmatch */ cp++; l2 = strlen(cp); if (l2 > l1 || memcmp(cp, &name[l1 - l2], l2) != 0) { dprintf("rejected in rule %d due to name mismatch.\n", i); continue; } } else { /* exact match */ l2 = strlen(cp); if (l2 != l1 || memcmp(cp, name, l1) != 0) { dprintf("rejected in rule %d due to name mismatch.\n", i); continue; } } } dprintf("accepted in rule %d.\n", i); return (1); /* hooray! */ } return (0); } /* * Fairly similar to popen(3), but returns an open descriptor, as * opposed to a FILE *. */ static int p_open(const char *prog, pid_t *rpid) { int pfd[2], nulldesc; pid_t pid; sigset_t omask, mask; char *argv[4]; /* sh -c cmd NULL */ char errmsg[200]; if (pipe(pfd) == -1) return (-1); if ((nulldesc = open(_PATH_DEVNULL, O_RDWR)) == -1) /* we are royally screwed anyway */ return (-1); sigemptyset(&mask); sigaddset(&mask, SIGALRM); sigaddset(&mask, SIGHUP); sigprocmask(SIG_BLOCK, &mask, &omask); switch ((pid = fork())) { case -1: sigprocmask(SIG_SETMASK, &omask, 0); close(nulldesc); return (-1); case 0: argv[0] = strdup("sh"); argv[1] = strdup("-c"); argv[2] = strdup(prog); argv[3] = NULL; if (argv[0] == NULL || argv[1] == NULL || argv[2] == NULL) { logerror("strdup"); exit(1); } alarm(0); (void)setsid(); /* Avoid catching SIGHUPs. */ /* * Throw away pending signals, and reset signal * behaviour to standard values. */ signal(SIGALRM, SIG_IGN); signal(SIGHUP, SIG_IGN); sigprocmask(SIG_SETMASK, &omask, 0); signal(SIGPIPE, SIG_DFL); signal(SIGQUIT, SIG_DFL); signal(SIGALRM, SIG_DFL); signal(SIGHUP, SIG_DFL); dup2(pfd[0], STDIN_FILENO); dup2(nulldesc, STDOUT_FILENO); dup2(nulldesc, STDERR_FILENO); closefrom(3); (void)execvp(_PATH_BSHELL, argv); _exit(255); } sigprocmask(SIG_SETMASK, &omask, 0); close(nulldesc); close(pfd[0]); /* * Avoid blocking on a hung pipe. With O_NONBLOCK, we are * supposed to get an EWOULDBLOCK on writev(2), which is * caught by the logic above anyway, which will in turn close * the pipe, and fork a new logging subprocess if necessary. * The stale subprocess will be killed some time later unless * it terminated itself due to closing its input pipe (so we * get rid of really dead puppies). */ if (fcntl(pfd[1], F_SETFL, O_NONBLOCK) == -1) { /* This is bad. */ (void)snprintf(errmsg, sizeof errmsg, "Warning: cannot change pipe to PID %d to " "non-blocking behaviour.", (int)pid); logerror(errmsg); } *rpid = pid; return (pfd[1]); } static void deadq_enter(pid_t pid, const char *name) { dq_t p; int status; /* * Be paranoid, if we can't signal the process, don't enter it * into the dead queue (perhaps it's already dead). If possible, * we try to fetch and log the child's status. */ if (kill(pid, 0) != 0) { if (waitpid(pid, &status, WNOHANG) > 0) log_deadchild(pid, status, name); return; } p = malloc(sizeof(struct deadq_entry)); if (p == NULL) { logerror("malloc"); exit(1); } p->dq_pid = pid; p->dq_timeout = DQ_TIMO_INIT; TAILQ_INSERT_TAIL(&deadq_head, p, dq_entries); } static int deadq_remove(pid_t pid) { dq_t q; TAILQ_FOREACH(q, &deadq_head, dq_entries) { if (q->dq_pid == pid) { TAILQ_REMOVE(&deadq_head, q, dq_entries); free(q); return (1); } } return (0); } static void log_deadchild(pid_t pid, int status, const char *name) { int code; char buf[256]; const char *reason; errno = 0; /* Keep strerror() stuff out of logerror messages. */ if (WIFSIGNALED(status)) { reason = "due to signal"; code = WTERMSIG(status); } else { reason = "with status"; code = WEXITSTATUS(status); if (code == 0) return; } (void)snprintf(buf, sizeof buf, "Logging subprocess %d (%s) exited %s %d.", pid, name, reason, code); logerror(buf); } static int * socksetup(int af, char *bindhostname) { struct addrinfo hints, *res, *r; const char *bindservice; char *cp; int error, maxs, *s, *socks; /* * We have to handle this case for backwards compatibility: * If there are two (or more) colons but no '[' and ']', * assume this is an inet6 address without a service. */ bindservice = "syslog"; if (bindhostname != NULL) { #ifdef INET6 if (*bindhostname == '[' && (cp = strchr(bindhostname + 1, ']')) != NULL) { ++bindhostname; *cp = '\0'; if (cp[1] == ':' && cp[2] != '\0') bindservice = cp + 2; } else { #endif cp = strchr(bindhostname, ':'); if (cp != NULL && strchr(cp + 1, ':') == NULL) { *cp = '\0'; if (cp[1] != '\0') bindservice = cp + 1; if (cp == bindhostname) bindhostname = NULL; } #ifdef INET6 } #endif } memset(&hints, 0, sizeof(hints)); hints.ai_flags = AI_PASSIVE; hints.ai_family = af; hints.ai_socktype = SOCK_DGRAM; error = getaddrinfo(bindhostname, bindservice, &hints, &res); if (error) { logerror(gai_strerror(error)); errno = 0; die(0); } /* Count max number of sockets we may open */ for (maxs = 0, r = res; r; r = r->ai_next, maxs++); socks = malloc((maxs+1) * sizeof(int)); if (socks == NULL) { logerror("couldn't allocate memory for sockets"); die(0); } *socks = 0; /* num of sockets counter at start of array */ s = socks + 1; for (r = res; r; r = r->ai_next) { int on = 1; *s = socket(r->ai_family, r->ai_socktype, r->ai_protocol); if (*s < 0) { logerror("socket"); continue; } #ifdef INET6 if (r->ai_family == AF_INET6) { if (setsockopt(*s, IPPROTO_IPV6, IPV6_V6ONLY, (char *)&on, sizeof (on)) < 0) { logerror("setsockopt"); close(*s); continue; } } #endif if (setsockopt(*s, SOL_SOCKET, SO_REUSEADDR, (char *)&on, sizeof (on)) < 0) { logerror("setsockopt"); close(*s); continue; } /* * RFC 3164 recommends that client side message * should come from the privileged syslogd port. * * If the system administrator choose not to obey * this, we can skip the bind() step so that the * system will choose a port for us. */ if (!NoBind) { if (bind(*s, r->ai_addr, r->ai_addrlen) < 0) { logerror("bind"); close(*s); continue; } if (!SecureMode) increase_rcvbuf(*s); } (*socks)++; s++; } if (*socks == 0) { free(socks); if (Debug) return (NULL); else die(0); } if (res) freeaddrinfo(res); return (socks); } static void increase_rcvbuf(int fd) { socklen_t len, slen; slen = sizeof(len); if (getsockopt(fd, SOL_SOCKET, SO_RCVBUF, &len, &slen) == 0) { if (len < RCVBUF_MINSIZE) { len = RCVBUF_MINSIZE; setsockopt(fd, SOL_SOCKET, SO_RCVBUF, &len, sizeof(len)); } } } Index: projects/cxl_iscsi/usr.sbin/sysrc/sysrc =================================================================== --- projects/cxl_iscsi/usr.sbin/sysrc/sysrc (revision 279716) +++ projects/cxl_iscsi/usr.sbin/sysrc/sysrc (revision 279717) @@ -1,701 +1,743 @@ #!/bin/sh #- -# Copyright (c) 2010-2014 Devin Teske +# Copyright (c) 2010-2015 Devin Teske # All rights reserved. # # Redistribution and use in source and binary forms, with or without # modification, are permitted provided that the following conditions # are met: # 1. Redistributions of source code must retain the above copyright # notice, this list of conditions and the following disclaimer. # 2. Redistributions in binary form must reproduce the above copyright # notice, this list of conditions and the following disclaimer in the # documentation and/or other materials provided with the distribution. # # THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND # ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE # IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE # ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE # FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL # DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS # OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) # HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT # LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY # OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF # SUCH DAMAGE. # # $FreeBSD$ # ############################################################ INCLUDES # Prevent `-d' from being interpreted as a debug flag by common.subr DEBUG_SELF_INITIALIZE= BSDCFG_SHARE="/usr/share/bsdconfig" [ "$_COMMON_SUBR" ] || . $BSDCFG_SHARE/common.subr || exit 1 [ "$_SYSRC_SUBR" ] || f_include $BSDCFG_SHARE/sysrc.subr ############################################################ GLOBALS # # Version information # -SYSRC_VERSION="6.2 Nov-3,2014" +SYSRC_VERSION="6.3 Mar-4,2015" # # Options # CHECK_ONLY= DELETE= DESCRIBE= IGNORE_UNKNOWNS= JAIL= QUIET= ROOTDIR= SHOW_ALL= SHOW_EQUALS= SHOW_FILE= SHOW_NAME=1 SHOW_VALUE=1 SYSRC_VERBOSE= ############################################################ FUNCTIONS # die [ $fmt [ $opts ... ]] # # Optionally print a message to stderr before exiting with failure status. # die() { local fmt="$1" [ $# -gt 0 ] && shift 1 [ "$fmt" ] && f_err "$fmt\n" "$@" exit $FAILURE } # usage # # Prints a short syntax statement and exits. # usage() { f_err "Usage: %s [OPTIONS] name[[+]=value] ...\n" "$pgm" f_err "Try \`%s --help' for more information.\n" "$pgm" die } # help # # Prints a full syntax statement and exits. # help() { local optfmt="\t%-11s%s\n" local envfmt="\t%-17s%s\n" - f_err "Usage: %s [OPTIONS] name[[+]=value] ...\n" "$pgm" + f_err "Usage: %s [OPTIONS] name[[+|-]=value] ...\n" "$pgm" f_err "OPTIONS:\n" f_err "$optfmt" "-a" \ "Dump a list of all non-default configuration variables." f_err "$optfmt" "-A" \ "Dump a list of all configuration variables (incl. defaults)." f_err "$optfmt" "-c" \ "Check. Return success if set or no changes, else error." f_err "$optfmt" "-d" \ "Print a description of the given variable." f_err "$optfmt" "-D" \ "Show default value(s) only (this is the same as setting" f_err "$optfmt" "" \ "RC_CONFS to NULL or passing \`-f' with a NULL file-argument)." f_err "$optfmt" "-e" \ "Print query results as \`var=value' (useful for producing" f_err "$optfmt" "" \ "output to be fed back in). Ignored if \`-n' is specified." f_err "$optfmt" "-f file" \ "Operate on the specified file(s) instead of rc_conf_files." f_err "$optfmt" "" \ "Can be specified multiple times for additional files." f_err "$optfmt" "-F" \ "Show only the last rc.conf(5) file each directive is in." f_err "$optfmt" "-h" \ "Print a short usage statement to stderr and exit." f_err "$optfmt" "--help" \ "Print this message to stderr and exit." f_err "$optfmt" "-i" \ "Ignore unknown variables." f_err "$optfmt" "-j jail" \ "The jid or name of the jail to operate within (overrides" f_err "$optfmt" "" \ "\`-R dir'; requires jexec(8))." f_err "$optfmt" "-n" \ "Show only variable values, not their names." f_err "$optfmt" "-N" \ "Show only variable names, not their values." f_err "$optfmt" "-q" \ "Quiet. Disable verbose and hide certain errors." f_err "$optfmt" "-R dir" \ "Operate within the root directory \`dir' rather than \`/'." f_err "$optfmt" "-v" \ "Verbose. Print the pathname of the specific rc.conf(5)" f_err "$optfmt" "" \ "file where the directive was found." f_err "$optfmt" "--version" \ "Print version information to stdout and exit." f_err "$optfmt" "-x" \ "Remove variable(s) from specified file(s)." f_err "\n" f_err "ENVIRONMENT:\n" f_err "$envfmt" "RC_CONFS" \ "Override default rc_conf_files (even if set to NULL)." f_err "$envfmt" "RC_DEFAULTS" \ "Location of \`/etc/defaults/rc.conf' file." die } # jail_depend # # Dump dependencies such as language-file variables and include files to stdout # to be piped-into sh(1) running via jexec(8)/chroot(8). As a security measure, # this prevents existing language files and library files from being loaded in # the jail. This also relaxes the requirement to have these files in every jail # before sysrc can be used on said jail. # jail_depend() { # # Indicate that we are jailed # echo export _SYSRC_JAILED=1 # # Print i18n language variables (their current values are sanitized # and re-printed for interpretation so that the i18n language files # do not need to exist within the jail). # local var val for var in \ msg_cannot_create_permission_denied \ msg_permission_denied \ msg_previous_syntax_errors \ ; do val=$( eval echo \"\$$var\" | awk '{ gsub(/'\''/, "'\''\\'\'\''"); print }' ) echo $var="'$val'" done # # Print include dependencies # echo DEBUG_SELF_INITIALIZE= cat $BSDCFG_SHARE/common.subr cat $BSDCFG_SHARE/sysrc.subr } ############################################################ MAIN SOURCE # # Perform sanity checks # [ $# -gt 0 ] || usage # # Check for `--help' and `--version' command-line option # ( # Operate in sub-shell to protect $@ in parent while [ $# -gt 0 ]; do case "$1" in --help) help ;; --version) # see GLOBALS echo "$SYSRC_VERSION" exit 1 ;; -[fRj]) # These flags take an argument shift 1 ;; esac shift 1 done exit 0 ) || die # # Process command-line flags # while getopts aAcdDef:Fhij:nNqR:vxX flag; do case "$flag" in a) SHOW_ALL=${SHOW_ALL:-1};; A) SHOW_ALL=2;; c) CHECK_ONLY=1;; d) DESCRIBE=1;; D) RC_CONFS=;; e) SHOW_EQUALS=1;; f) RC_CONFS="$RC_CONFS${RC_CONFS:+ }$OPTARG";; F) SHOW_FILE=1;; h) usage;; i) IGNORE_UNKNOWNS=1;; j) [ "$OPTARG" ] || die \ "%s: Missing or null argument to \`-j' flag" "$pgm" JAIL="$OPTARG";; n) SHOW_NAME=;; N) SHOW_VALUE=;; q) QUIET=1 SYSRC_VERBOSE=;; R) [ "$OPTARG" ] || die \ "%s: Missing or null argument to \`-R' flag" "$pgm" ROOTDIR="$OPTARG";; v) SYSRC_VERBOSE=1 QUIET=;; x) DELETE=${DELETE:-1};; X) DELETE=2;; \?) usage;; esac done shift $(( $OPTIND - 1 )) # # [More] Sanity checks (e.g., "sysrc --") # [ $# -eq 0 -a ! "$SHOW_ALL" ] && usage # # Taint-check all rc.conf(5) files # errmsg="$pgm: Exiting due to previous syntax errors" if [ "${RC_CONFS+set}" ]; then ( for i in $RC_CONFS; do [ -e "$i" ] || continue /bin/sh -n "$i" || exit $FAILURE done exit $SUCCESS ) || die "$errmsg" else /bin/sh -n "$RC_DEFAULTS" || die "$errmsg" ( . "$RC_DEFAULTS" for i in $rc_conf_files; do [ -e "$i" ] || continue /bin/sh -n "$i" || exit $FAILURE done exit $SUCCESS ) || die "$errmsg" fi # # Process `-x' (and secret `-X') command-line options # errmsg="$pgm: \`-x' option incompatible with \`-a'/\`-A' options" errmsg="$errmsg (use \`-X' to override)" if [ "$DELETE" -a "$SHOW_ALL" ]; then [ "$DELETE" = "2" ] || die "$errmsg" fi # # Pre-flight for `-c' command-line option # [ "$CHECK_ONLY" -a "$SHOW_ALL" ] && die "$pgm: \`-c' option incompatible with \`-a'/\`-A' options" # # Process `-e', `-n', and `-N' command-line options # SEP=': ' [ "$SHOW_FILE" ] && SHOW_EQUALS= [ "$SHOW_NAME" ] || SHOW_EQUALS= [ "$SYSRC_VERBOSE" = "0" ] && SYSRC_VERBOSE= if [ ! "$SHOW_VALUE" ]; then SHOW_NAME=1 SHOW_EQUALS= fi [ "$SHOW_EQUALS" ] && SEP='="' # # Process `-j jail' and `-R dir' command-line options # if [ "$JAIL" -o "$ROOTDIR" ]; then # # Reconstruct the arguments that we want to carry-over # args=" ${SYSRC_VERBOSE:+-v} ${QUIET:+-q} $( [ "$DELETE" = "1" ] && echo \ -x ) $( [ "$DELETE" = "2" ] && echo \ -X ) $( [ "$SHOW_ALL" = "1" ] && echo \ -a ) $( [ "$SHOW_ALL" = "2" ] && echo \ -A ) ${CHECK_ONLY:+-c} ${DESCRIBE:+-d} ${SHOW_EQUALS:+-e} ${IGNORE_UNKNOWNS:+-i} $( [ "$SHOW_NAME" ] || echo \ -n ) $( [ "$SHOW_VALUE" ] || echo \ -N ) $( [ "$SHOW_FILE" ] && echo \ -F ) " if [ "${RC_CONFS+set}" ]; then args="$args -f '$RC_CONFS'" fi for arg in "$@"; do args="$args '$arg'" done # # If both are supplied, `-j jail' supercedes `-R dir' # if [ "$JAIL" ]; then # # Re-execute ourselves with sh(1) via jexec(8) # ( echo set -- $args jail_depend cat $0 ) | env - RC_DEFAULTS="$RC_DEFAULTS" \ /usr/sbin/jexec "$JAIL" /bin/sh exit $? elif [ "$ROOTDIR" ]; then # # Make sure that the root directory specified is not to any # running jails. # # NOTE: To maintain backward compatibility with older jails on # older systems, we will not perform this check if either the # jls(1) or jexec(8) utilities are missing. # if f_have jexec && f_have jls; then jid="`jls jid path | \ ( while read JID JROOT; do [ "$JROOT" = "$ROOTDIR" ] || continue echo $JID done )`" # # If multiple running jails match the specified root # directory, exit with error. # if [ "$jid" -a "${jid%[$IFS]*}" != "$jid" ]; then die "%s: %s: %s" "$pgm" "$ROOTDIR" \ "$( echo "Multiple jails claim this" \ "directory as their root." \ "(use \`-j jail' instead)" )" fi # # If only a single running jail matches the specified # root directory, implicitly use `-j jail'. # if [ "$jid" ]; then # # Re-execute outselves with sh(1) via jexec(8) # ( echo set -- $args jail_depend cat $0 ) | env - RC_DEFAULTS="$RC_DEFAULTS" \ /usr/sbin/jexec "$jid" /bin/sh exit $? fi # Otherwise, fall through and allow chroot(8) fi # # Re-execute ourselves with sh(1) via chroot(8) # ( echo set -- $args jail_depend cat $0 ) | env - RC_DEFAULTS="$RC_DEFAULTS" \ /usr/sbin/chroot "$ROOTDIR" /bin/sh exit $? fi fi # # Process `-a' or `-A' command-line options # if [ "$SHOW_ALL" ]; then # # Get a list of variables that are currently set in the rc.conf(5) # files (included `/etc/defaults/rc.conf') by performing a call to # source_rc_confs() in a clean environment. # ( # Operate in a sub-shell to protect the parent environment # # Set which variables we want to preserve in the environment. # Append the pipe-character (|) to the list of internal field # separation (IFS) characters, allowing us to use the below # list both as an extended grep (-E) pattern and argument list # (required to first get f_clean_env() to preserve these in the # environment and then later to prune them from the list of # variables produced by set(1)). # IFS="$IFS|" EXCEPT="IFS|EXCEPT|PATH|RC_DEFAULTS|OPTIND|DESCRIBE|SEP" EXCEPT="$EXCEPT|DELETE|SHOW_ALL|SHOW_EQUALS|SHOW_NAME" EXCEPT="$EXCEPT|SHOW_VALUE|SHOW_FILE|SYSRC_VERBOSE|RC_CONFS" EXCEPT="$EXCEPT|pgm|SUCCESS|FAILURE|CHECK_ONLY" EXCEPT="$EXCEPT|f_sysrc_desc_awk|f_sysrc_delete_awk" # # Clean the environment (except for our required variables) # and then source the required files. # f_clean_env --except $EXCEPT if [ -f "$RC_DEFAULTS" -a -r "$RC_DEFAULTS" ]; then . "$RC_DEFAULTS" # # If passed `-a' (rather than `-A'), re-purge the # environment, removing the rc.conf(5) defaults. # [ "$SHOW_ALL" = "1" ] \ && f_clean_env --except rc_conf_files $EXCEPT # # If `-f file' was passed, set $rc_conf_files to an # explicit value, modifying the default behavior of # source_rc_confs(). # [ "${RC_CONFS+set}" ] && rc_conf_files="$RC_CONFS" source_rc_confs # # If passed `-a' (rather than `-A'), remove # `rc_conf_files' unless it was defined somewhere # other than rc.conf(5) defaults. # [ "$SHOW_ALL" = "1" -a \ "$( f_sysrc_find rc_conf_files )" = "$RC_DEFAULTS" \ ] \ && unset rc_conf_files fi for NAME in $( set | awk -F= '/^[[:alpha:]_][[:alnum:]_]*=/ {print $1}' | grep -Ev "^($EXCEPT)$" ); do # # If enabled, describe rather than expand value # if [ "$DESCRIBE" ]; then echo "$NAME: $( f_sysrc_desc "$NAME" )" continue fi # # If `-F' is passed, find it and move on # if [ "$SHOW_FILE" ]; then [ "$SHOW_NAME" ] && echo -n "$NAME: " f_sysrc_find "$NAME" continue fi # # If `-X' is passed, delete the variables # if [ "$DELETE" = "2" ]; then f_sysrc_delete "$NAME" continue fi [ "$SYSRC_VERBOSE" ] && \ echo -n "$( f_sysrc_find "$NAME" ): " # # If `-N' is passed, simplify the output # if [ ! "$SHOW_VALUE" ]; then echo "$NAME" continue fi echo "${SHOW_NAME:+$NAME$SEP}$( f_sysrc_get "$NAME" )${SHOW_EQUALS:+\"}" done ) # # Ignore the remainder of positional arguments. # exit $SUCCESS fi # # Process command-line arguments # status=$SUCCESS while [ $# -gt 0 ]; do NAME="${1%%=*}" case "$NAME" in *+) mode=APPEND NAME="${NAME%+}" ;; + *-) mode=REMOVE NAME="${NAME%-}" ;; *) mode=ASSIGN esac [ "$DESCRIBE" ] && \ echo "$NAME: $( f_sysrc_desc "$NAME" )" case "$1" in *=*) # # Like sysctl(8), if both `-d' AND "name=value" is passed, # first describe (done above), then attempt to set # # If verbose, prefix line with where the directive lives if [ "$SYSRC_VERBOSE" -a ! "$CHECK_ONLY" ]; then file=$( f_sysrc_find "$NAME" ) [ "$file" = "$RC_DEFAULTS" -o ! "$file" ] && \ file=$( f_sysrc_get 'rc_conf_files%%[$IFS]*' ) if [ "$SHOW_EQUALS" ]; then echo -n ": $file; " else echo -n "$file: " fi fi # # If `-x' or `-X' is passed, delete the variable and ignore the # desire to set some value # if [ "$DELETE" ]; then f_sysrc_delete "$NAME" || status=$FAILURE shift 1 continue fi # # If `-c' is passed, simply compare and move on # if [ "$CHECK_ONLY" ]; then if ! IGNORED=$( f_sysrc_get "$NAME?" ); then status=$FAILURE [ "$SYSRC_VERBOSE" ] && echo "$NAME: not currently set" shift 1 continue fi value=$( f_sysrc_get "$NAME" ) if [ "$value" != "${1#*=}" ]; then status=$FAILURE if [ "$SYSRC_VERBOSE" ]; then echo -n "$( f_sysrc_find "$NAME" ): " echo -n "$NAME: would change from " echo "\`$value' to \`${1#*=}'" fi elif [ "$SYSRC_VERBOSE" ]; then echo -n "$( f_sysrc_find "$NAME" ): " echo "$NAME: already set to \`$value'" fi shift 1 continue fi # - # If `-N' is passed, simplify the output + # Determine both `before' value and appropriate `new' value # - if [ ! "$SHOW_VALUE" ]; then - echo "$NAME" - case "$mode" in - APPEND) - before=$( f_sysrc_get "$NAME" ) - f_sysrc_set "$NAME" "$before${1#*=}" - ;; - *) - f_sysrc_set "$NAME" "${1#*=}" + case "$mode" in + APPEND) + before=$( f_sysrc_get "$NAME" ) + add="${1#*=}" + delim="${add%"${add#?}"}" # first character + oldIFS="$IFS" + case "$delim" in + ""|[$IFS]|[a-zA-Z0-9]) delim=" " ;; + *) IFS="$delim" esac - else + new="$before" + for a in $add; do + [ "$a" ] || continue + skip= + for b in $before; do + [ "$b" = "$a" ] && skip=1 break + done + [ "$skip" ] || new="$new$delim$a" + done + new="${new#"$delim"}" IFS="$oldIFS" + unset add delim oldIFS a skip b + [ "$SHOW_FILE" ] && before=$( f_sysrc_find "$NAME" ) + ;; + REMOVE) + before=$( f_sysrc_get "$NAME" ) + remove="${1#*=}" + delim="${remove%"${remove#?}"}" # first character + oldIFS="$IFS" + case "$delim" in + ""|[$IFS]|[a-zA-Z0-9]) delim=" " ;; + *) IFS="$delim" + esac + new= + for b in $before; do + [ "$b" ] || continue + add=1 + for r in $remove; do + [ "$r" = "$b" ] && add= break + done + [ "$add" ] && new="$new$delim$b" + done + new="${new#"$delim"}" IFS="$oldIFS" + unset remove delim oldIFS b add r + [ "$SHOW_FILE" ] && before=$( f_sysrc_find "$NAME" ) + ;; + *) if [ "$SHOW_FILE" ]; then before=$( f_sysrc_find "$NAME" ) else before=$( f_sysrc_get "$NAME" ) fi - if case "$mode" in - APPEND) f_sysrc_set "$NAME" "$before${1#*=}" ;; - *) f_sysrc_set "$NAME" "${1#*=}" - esac - then + new="${1#*=}" + esac + + # + # If `-N' is passed, simplify the output + # + if [ ! "$SHOW_VALUE" ]; then + echo "$NAME" + f_sysrc_set "$NAME" "$new" + else + if f_sysrc_set "$NAME" "$new"; then if [ "$SHOW_FILE" ]; then after=$( f_sysrc_find "$NAME" ) else after=$( f_sysrc_get "$NAME" ) fi echo -n "${SHOW_NAME:+$NAME$SEP}" echo -n "$before${SHOW_EQUALS:+\" #}" echo -n " -> ${SHOW_EQUALS:+\"}$after" echo "${SHOW_EQUALS:+\"}" fi fi ;; *) if ! IGNORED=$( f_sysrc_get "$NAME?" ); then [ "$IGNORE_UNKNOWNS" -o "$QUIET" ] || echo "$pgm: unknown variable '$NAME'" shift 1 status=$FAILURE continue fi # The above check told us what we needed for `-c' if [ "$CHECK_ONLY" ]; then shift 1 continue fi # # Like sysctl(8), when `-d' is passed, desribe it # (already done above) rather than expanding it # if [ "$DESCRIBE" ]; then shift 1 continue fi # # If `-x' or `-X' is passed, delete the variable # if [ "$DELETE" ]; then f_sysrc_delete "$NAME" || status=$FAILURE shift 1 continue fi # # If `-F' is passed, find it and move on # if [ "$SHOW_FILE" ]; then [ "$SHOW_NAME" ] && echo -n "$NAME: " f_sysrc_find "$NAME" shift 1 continue fi if [ "$SYSRC_VERBOSE" ]; then if [ "$SHOW_EQUALS" ]; then echo -n ": $( f_sysrc_find "$NAME" ); " else echo -n "$( f_sysrc_find "$NAME" ): " fi fi # # If `-N' is passed, simplify the output # if [ ! "$SHOW_VALUE" ]; then echo "$NAME" else echo "${SHOW_NAME:+$NAME$SEP}$( f_sysrc_get "$NAME" )${SHOW_EQUALS:+\"}" fi esac shift 1 done exit $status # $SUCCESS unless error occurred with either `-c' or `-x' ################################################################################ # END ################################################################################ Index: projects/cxl_iscsi/usr.sbin/sysrc/sysrc.8 =================================================================== --- projects/cxl_iscsi/usr.sbin/sysrc/sysrc.8 (revision 279716) +++ projects/cxl_iscsi/usr.sbin/sysrc/sysrc.8 (revision 279717) @@ -1,319 +1,437 @@ -.\" Copyright (c) 2011-2014 Devin Teske +.\" Copyright (c) 2011-2015 Devin Teske .\" All rights reserved. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" $FreeBSD$ .\" -.Dd November 4, 2014 +.Dd March 4, 2015 .Dt SYSRC 8 .Os .Sh NAME .Nm sysrc .Nd safely edit system rc files .Sh SYNOPSIS .Nm .Op Fl cdDeFhinNqvx .Op Fl f Ar file .Op Fl j Ar jail | Fl R Ar dir -.Ar name Ns Op Ns Oo + Oc Ns = Ns Ar value +.Ar name Ns Op Ns Oo +|- Oc Ns = Ns Ar value .Ar ... .Nm .Op Fl cdDeFhinNqvx .Op Fl f Ar file .Op Fl j Ar jail | Fl R Ar dir .Fl a | A .Sh DESCRIPTION The .Nm utility retrieves .Xr rc.conf 5 variables from the collection of system rc files and allows processes with appropriate privilege to change values in a safe and effective manner. .Pp The following options are available: .Bl -tag -width indent+ .It Fl a Dump a list of all non-default configuration variables. .It Fl A Dump a list of all configuration variables .Pq incl. defaults . .It Fl c Check only. For querying, return success if all requested variables are set .Pq even if NULL , otherwise return error status. For assignments, return success if no changes are required, otherwise failure. If verbose .Pq see Dq Fl v prints a message stating whether variables are set and/or changes are required. .It Fl d Print a description of the given variable. .It Fl D Show default value(s) only (this is the same as setting RC_CONFS to NULL or passing `-f' with a NULL file-argument). .It Fl e Print query results as .Xr sh 1 compatible syntax .Pq for example, Ql var=value . Ignored if either .Ql Fl n or .Ql Fl F is specified. .It Fl f Ar file Operate on the specified file(s) instead of the files obtained by reading the .Sq rc_conf_files entry in the .Ev RC_DEFAULTS file. This option can be specified multiple times for additional files. .It Fl F Show only the last .Xr rc.conf 5 file each directive is in. .It Fl h Print a short usage message to stderr and exit. .It Fl -help Print a full usage statement to stderr and exit. .It Fl i Ignore unknown variables. .It Fl j Ar jail The .Ar jid or name of the .Ar jail to operate within .Pq overrides So Fl R Ar dir Sc ; requires Xr jexec 8 . .It Fl n Show only variable values, not their names. .It Fl N Show only variable names, not their values. .It Fl q Quiet. Disable verbose and hide certain errors. .It Fl R Ar dir Operate within the root directory .Sq Ar dir rather than .Sq / . .It Fl v Verbose. Print the pathname of the specific .Xr rc.conf 5 file where the directive was found. .It Fl -version Print version information to stdout and exit. .It Fl x Remove variable(s) from specified file(s). .El .Pp This utility has a similar syntax to .Xr sysctl 8 . It shares the `-e' and `-n' options .Pq detailed above and also has the same .Ql name[=value] syntax for making queries/assignments. In addition -.Pq unlike Xr sysctl 8 , +.Pq but unlike Xr sysctl 8 , .Ql name+=value -is supported for appending values. +is supported for adding items to values +.Pq see APPENDING VALUES +and +.Ql name-=value +is supported for removing items from values +.Pq see SUBTRACTING VALUES . .Pp However, while .Xr sysctl 8 serves to query/modify MIBs in the entrant kernel, .Nm instead works on values in the system .Xr rc.conf 5 configuration files. .Pp The list of system configuration files is configured in the file .Ql /etc/defaults/rc.conf within the variable .Ql rc_conf_files , which by-default contains a space-separated list of pathnames. On all FreeBSD systems, this defaults to the value "/etc/rc.conf /etc/rc.conf.local". Each pathname is sourced in-order upon startup. It is in the same fashion that .Nm sources the configuration files before returning the value of the given variable. .Pp When supplied a variable name, .Nm will return the value of the variable. If the variable does not appear in any of the configured .Ql rc_conf_files , an error is printed and error status is returned. .Pp When changing values of a given variable, it does not matter if the variable appears in any of the .Ql rc_conf_files or not. If the variable does not appear in any of the files, it is appended to the end of the first pathname in the .Ql rc_conf_files variable. Otherwise, .Nm will replace only the last-occurrence in the last-file found to contain the variable. This gets the value to take effect next boot without heavily modifying these integral files (yet taking care not to allow the file to grow unwieldy should .Nm be called repeatedly). +.Sh APPENDING VALUES +When using the +.Ql key+=value +syntax to add items to existing values, +the first character of the value is taken as the delimiter separating items +.Pq usually Qo \ Qc or Qo , Qc . +For example, in the following statement: +.Bl -tag -width indent+ +.It \ +.Nm +cloned_interfaces+=" gif0" +.El +.Pp +the first character is a space, informing +.Nm +that existing values are to be considered separated by whitespace. +If +.Ql gif0 +is not found in the existing value for +.Va cloned_interfaces , +it is added +.Pq with delimiter only if existing value is non-NULL . +.Pp +For convenience, if the first character is alpha-numeric +.Pq letters A-Z, a-z, or numbers 0-9 , +.Nm +uses the default setting of whitespace as separator. +For example, the above and below statements are equivalent since +.Dq gif0 +starts with an alpha-numeric character +.Pq the letter Li g : +.Pp +.Bl -tag -width indent+ +.It \ +.Nm +cloned_interfaces+=gif0 +.El +.Pp +Take the following sequence for example: +.Bl -tag -width indent+ +.It \ +.Nm +cloned_interfaces= # start with NULL +.It \ +.Nm +cloned_interfaces+=gif0 +.Dl # NULL -> `gif0' Pq NB: no preceding delimiter +.It \ +.Nm +cloned_interfaces+=gif0 # no change +.It \ +.Nm +cloned_interfaces+="tun0 gif0" +.Dl # `gif0' -> `gif0 tun0' Pq NB: no duplication +.El +.Pp +.Nm +prevents the same value from being added if already there. +.Sh SUBTRACTING VALUES +When using the +.Ql key-=value +syntax to remove items from existing values, +the first character of the value is taken as the delimiter separating items +.Pq usually Qo \ Qc or Qo , Qc . +For example, in the following statement: +.Pp +.Dl Nm cloned_interfaces-=" gif0" +.Pp +the first character is a space, informing +.Nm +that existing values are to be considered separated by whitespace. +If +.Ql gif0 +is found in the existing value for +.Va cloned_interfaces , +it is removed +.Pq extra delimiters removed . +.Pp +For convenience, if the first character is alpha-numeric +.Pq letters A-Z, a-z, or numbers 0-9 , +.Nm +uses the default setting of whitespace as separator. +For example, the above and below statements are equivalent since +.Dq gif0 +starts with an alpha-numeric character +.Pq the letter Li g : +.Pp +.Bl -tag -width indent+ +.It \ +.Nm +cloned_interfaces-=gif0 +.El +.Pp +Take the following sequence for example: +.Bl -tag -width indent+ +.It \ +.Nm +foo="bar baz" # start +.It \ +.Nm +foo-=bar # `bar baz' -> `baz' +.It \ +.Nm +foo-=baz # `baz' -> NULL +.El +.Pp +.Nm +removes all occurrences of all items provided +and collapses extra delimiters between items. .Sh ENVIRONMENT The following environment variables are referenced by .Nm : .Bl -tag -width ".Ev RC_DEFAULTS" .It Ev RC_CONFS Override default .Ql rc_conf_files .Pq even if set to NULL . .It Ev RC_DEFAULTS Location of .Ql /etc/defaults/rc.conf file. .El .Sh DEPENDENCIES The following standard commands are required by .Nm : .Pp .Xr awk 1 , .Xr cat 1 , .Xr chmod 1 , .Xr env 1 , .Xr grep 1 , .Xr jls 1 , .Xr mktemp 1 , .Xr mv 1 , .Xr rm 1 , .Xr sh 1 , .Xr stat 1 , .Xr tail 1 , .Xr chown 8 and .Xr jexec 8 . .Sh FILES .Bl -tag -width ".Pa /etc/defaults/rc.conf" -compact .It Pa /etc/defaults/rc.conf .It Pa /etc/rc.conf .It Pa /etc/rc.conf.local .El .Sh EXAMPLES Below are some simple examples of how .Nm can be used to query certain values from the .Xr rc.conf 5 collection of system configuration files: .Pp .Nm sshd_enable .Dl returns the value of $sshd_enable, usually YES or NO . .Pp .Nm defaultrouter .Dl returns IP address of default router Pq if configured . .Pp Working on other files, such as .Xr crontab 5 : .Pp .Nm -f /etc/crontab MAILTO .Dl returns the value of the MAILTO setting Pq if configured . .Pp Appending to existing values: .Pp .Nm -\&cloned_interfaces+=" gif0" -.Dl appends Qo \ gif0 Qc to $cloned_interfaces . +\&cloned_interfaces+=gif0 +.Dl appends Qo gif0 Qc to $cloned_interfaces Pq see APPENDING VALUES . +.Pp +.Nm +\&cloned_interfaces-=gif0 +.Dl removes Qo gif0 Qc from $cloned_interfaces Pq see SUBTRACTING VALUES . .Pp In addition to the above syntax, .Nm also supports inline .Xr sh 1 PARAMETER expansion for changing the way values are reported, shown below: .Pp .Nm \&'hostname%%.*' .Dl returns $hostname up to (but not including) first `.' . .Pp .Nm \&'network_interfaces%%[$IFS]*' .Dl returns first word of $network_interfaces . .Pp .Nm \&'ntpdate_flags##*[$IFS]' .Dl returns last word of $ntpdate_flags (time server address) . .Pp .Nm usbd_flags-"default" .Dl returns $usbd_flags or "default" if unset or NULL . .Pp .Nm cloned_interfaces+"alternate" .Dl returns "alternate" if $cloned_interfaces is set . .Pp .Nm \&'#kern_securelevel' .Dl returns length in characters of $kern_securelevel . .Pp .Nm \&'hostname?' .Dl returns NULL and error status 2 if $hostname is unset Pq or if set, returns the value of $hostname with no error status . .Pp .Nm \&'hostname:?' .Dl returns NULL and error status 2 if $hostname is unset or NULL Pq or if set and non-NULL, returns value without error status . .Sh LIMITATIONS The .Nm utility presently does not support the .Ql rc.conf.d collection of system configuration files .Pq which requires a service name to be known during execution . .Pp This will be corrected by a future enhancement. .Sh SEE ALSO .Xr jls 1 , .Xr rc.conf 5 , .Xr jail 8 , .Xr jexec 8 , .Xr rc 8 , .Xr sysctl 8 .Sh HISTORY A .Nm utility first appeared in .Fx 9.2 . .Sh AUTHORS .An Devin Teske Aq Mt dteske@FreeBSD.org .Sh THANKS TO Brandon Gooch, Garrett Cooper, Julian Elischer, Pawel Jakub Dawidek, Cyrille Lefevre, Ross West, Stefan Esser, Marco Steinbach, Jilles Tjoelker, Allan Jude, and Lars Engels for suggestions, help, and testing. Index: projects/cxl_iscsi =================================================================== --- projects/cxl_iscsi (revision 279716) +++ projects/cxl_iscsi (revision 279717) Property changes on: projects/cxl_iscsi ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head:r279555-279716