Index: projects/largeSMP/bin/sh/eval.c =================================================================== --- projects/largeSMP/bin/sh/eval.c (revision 222782) +++ projects/largeSMP/bin/sh/eval.c (revision 222783) @@ -1,1269 +1,1270 @@ /*- * Copyright (c) 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[] = "@(#)eval.c 8.9 (Berkeley) 6/8/95"; #endif #endif /* not lint */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include /* For WIFSIGNALED(status) */ #include /* * Evaluate a command. */ #include "shell.h" #include "nodes.h" #include "syntax.h" #include "expand.h" #include "parser.h" #include "jobs.h" #include "eval.h" #include "builtins.h" #include "options.h" #include "exec.h" #include "redir.h" #include "input.h" #include "output.h" #include "trap.h" #include "var.h" #include "memalloc.h" #include "error.h" #include "show.h" #include "mystring.h" #ifndef NO_HISTORY #include "myhistedit.h" #endif int evalskip; /* set if we are skipping commands */ static int skipcount; /* number of levels to skip */ MKINIT int loopnest; /* current loop nesting level */ int funcnest; /* depth of function calls */ static int builtin_flags; /* evalcommand flags for builtins */ char *commandname; struct strlist *cmdenviron; int exitstatus; /* exit status of last command */ int oexitstatus; /* saved exit status */ static void evalloop(union node *, int); static void evalfor(union node *, int); static void evalcase(union node *, int); static void evalsubshell(union node *, int); static void evalredir(union node *, int); static void expredir(union node *); static void evalpipe(union node *); static int is_valid_fast_cmdsubst(union node *n); static void evalcommand(union node *, int, struct backcmd *); static void prehash(union node *); /* * Called to reset things after an exception. */ #ifdef mkinit INCLUDE "eval.h" RESET { evalskip = 0; loopnest = 0; funcnest = 0; } #endif /* * The eval command. */ int evalcmd(int argc, char **argv) { char *p; char *concat; char **ap; if (argc > 1) { p = argv[1]; if (argc > 2) { STARTSTACKSTR(concat); ap = argv + 2; for (;;) { STPUTS(p, concat); if ((p = *ap++) == NULL) break; STPUTC(' ', concat); } STPUTC('\0', concat); p = grabstackstr(concat); } evalstring(p, builtin_flags & EV_TESTED); } else exitstatus = 0; return exitstatus; } /* * Execute a command or commands contained in a string. */ void evalstring(char *s, int flags) { union node *n; struct stackmark smark; int flags_exit; int any; flags_exit = flags & EV_EXIT; flags &= ~EV_EXIT; any = 0; setstackmark(&smark); setinputstring(s, 1); while ((n = parsecmd(0)) != NEOF) { if (n != NULL && !nflag) { if (flags_exit && preadateof()) evaltree(n, flags | EV_EXIT); else evaltree(n, flags); any = 1; } popstackmark(&smark); } popfile(); popstackmark(&smark); if (!any) exitstatus = 0; if (flags_exit) exraise(EXEXIT); } /* * Evaluate a parse tree. The value is left in the global variable * exitstatus. */ void evaltree(union node *n, int flags) { int do_etest; union node *next; do_etest = 0; if (n == NULL) { TRACE(("evaltree(NULL) called\n")); exitstatus = 0; goto out; } do { next = NULL; #ifndef NO_HISTORY displayhist = 1; /* show history substitutions done with fc */ #endif TRACE(("evaltree(%p: %d) called\n", (void *)n, n->type)); switch (n->type) { case NSEMI: evaltree(n->nbinary.ch1, flags & ~EV_EXIT); if (evalskip) goto out; next = n->nbinary.ch2; break; case NAND: evaltree(n->nbinary.ch1, EV_TESTED); if (evalskip || exitstatus != 0) { goto out; } next = n->nbinary.ch2; break; case NOR: evaltree(n->nbinary.ch1, EV_TESTED); if (evalskip || exitstatus == 0) goto out; next = n->nbinary.ch2; break; case NREDIR: evalredir(n, flags); break; case NSUBSHELL: evalsubshell(n, flags); do_etest = !(flags & EV_TESTED); break; case NBACKGND: evalsubshell(n, flags); break; case NIF: { evaltree(n->nif.test, EV_TESTED); if (evalskip) goto out; if (exitstatus == 0) next = n->nif.ifpart; else if (n->nif.elsepart) next = n->nif.elsepart; else exitstatus = 0; break; } case NWHILE: case NUNTIL: evalloop(n, flags & ~EV_EXIT); break; case NFOR: evalfor(n, flags & ~EV_EXIT); break; case NCASE: evalcase(n, flags); break; case NDEFUN: defun(n->narg.text, n->narg.next); exitstatus = 0; break; case NNOT: evaltree(n->nnot.com, EV_TESTED); exitstatus = !exitstatus; break; case NPIPE: evalpipe(n); do_etest = !(flags & EV_TESTED); break; case NCMD: evalcommand(n, flags, (struct backcmd *)NULL); do_etest = !(flags & EV_TESTED); break; default: out1fmt("Node type = %d\n", n->type); flushout(&output); break; } n = next; } while (n != NULL); out: if (pendingsigs) dotrap(); if (eflag && exitstatus != 0 && do_etest) exitshell(exitstatus); if (flags & EV_EXIT) exraise(EXEXIT); } static void evalloop(union node *n, int flags) { int status; loopnest++; status = 0; for (;;) { evaltree(n->nbinary.ch1, EV_TESTED); if (evalskip) { skipping: if (evalskip == SKIPCONT && --skipcount <= 0) { evalskip = 0; continue; } if (evalskip == SKIPBREAK && --skipcount <= 0) evalskip = 0; if (evalskip == SKIPFUNC || evalskip == SKIPFILE) status = exitstatus; break; } if (n->type == NWHILE) { if (exitstatus != 0) break; } else { if (exitstatus == 0) break; } evaltree(n->nbinary.ch2, flags); status = exitstatus; if (evalskip) goto skipping; } loopnest--; exitstatus = status; } static void evalfor(union node *n, int flags) { struct arglist arglist; union node *argp; struct strlist *sp; struct stackmark smark; setstackmark(&smark); arglist.lastp = &arglist.list; for (argp = n->nfor.args ; argp ; argp = argp->narg.next) { oexitstatus = exitstatus; expandarg(argp, &arglist, EXP_FULL | EXP_TILDE); if (evalskip) goto out; } *arglist.lastp = NULL; exitstatus = 0; loopnest++; for (sp = arglist.list ; sp ; sp = sp->next) { setvar(n->nfor.var, sp->text, 0); evaltree(n->nfor.body, flags); if (evalskip) { if (evalskip == SKIPCONT && --skipcount <= 0) { evalskip = 0; continue; } if (evalskip == SKIPBREAK && --skipcount <= 0) evalskip = 0; break; } } loopnest--; out: popstackmark(&smark); } static void evalcase(union node *n, int flags) { union node *cp; union node *patp; struct arglist arglist; struct stackmark smark; setstackmark(&smark); arglist.lastp = &arglist.list; oexitstatus = exitstatus; exitstatus = 0; expandarg(n->ncase.expr, &arglist, EXP_TILDE); for (cp = n->ncase.cases ; cp && evalskip == 0 ; cp = cp->nclist.next) { for (patp = cp->nclist.pattern ; patp ; patp = patp->narg.next) { if (casematch(patp, arglist.list->text)) { if (evalskip == 0) { evaltree(cp->nclist.body, flags); } goto out; } } } out: popstackmark(&smark); } /* * Kick off a subshell to evaluate a tree. */ static void evalsubshell(union node *n, int flags) { struct job *jp; int backgnd = (n->type == NBACKGND); + oexitstatus = exitstatus; expredir(n->nredir.redirect); if ((!backgnd && flags & EV_EXIT && !have_traps()) || forkshell(jp = makejob(n, 1), n, backgnd) == 0) { if (backgnd) flags &=~ EV_TESTED; redirect(n->nredir.redirect, 0); evaltree(n->nredir.n, flags | EV_EXIT); /* never returns */ } else if (! backgnd) { INTOFF; exitstatus = waitforjob(jp, (int *)NULL); INTON; } else exitstatus = 0; } /* * Evaluate a redirected compound command. */ static void evalredir(union node *n, int flags) { struct jmploc jmploc; struct jmploc *savehandler; volatile int in_redirect = 1; + oexitstatus = exitstatus; expredir(n->nredir.redirect); savehandler = handler; if (setjmp(jmploc.loc)) { int e; handler = savehandler; e = exception; popredir(); if (e == EXERROR || e == EXEXEC) { if (in_redirect) { exitstatus = 2; return; } } longjmp(handler->loc, 1); } else { INTOFF; handler = &jmploc; redirect(n->nredir.redirect, REDIR_PUSH); in_redirect = 0; INTON; evaltree(n->nredir.n, flags); } INTOFF; handler = savehandler; popredir(); INTON; } /* * Compute the names of the files in a redirection list. */ static void expredir(union node *n) { union node *redir; for (redir = n ; redir ; redir = redir->nfile.next) { struct arglist fn; fn.lastp = &fn.list; - oexitstatus = exitstatus; switch (redir->type) { case NFROM: case NTO: case NFROMTO: case NAPPEND: case NCLOBBER: expandarg(redir->nfile.fname, &fn, EXP_TILDE | EXP_REDIR); redir->nfile.expfname = fn.list->text; break; case NFROMFD: case NTOFD: if (redir->ndup.vname) { expandarg(redir->ndup.vname, &fn, EXP_TILDE | EXP_REDIR); fixredir(redir, fn.list->text, 1); } break; } } } /* * Evaluate a pipeline. All the processes in the pipeline are children * of the process creating the pipeline. (This differs from some versions * of the shell, which make the last process in a pipeline the parent * of all the rest.) */ static void evalpipe(union node *n) { struct job *jp; struct nodelist *lp; int pipelen; int prevfd; int pip[2]; TRACE(("evalpipe(%p) called\n", (void *)n)); pipelen = 0; for (lp = n->npipe.cmdlist ; lp ; lp = lp->next) pipelen++; INTOFF; jp = makejob(n, pipelen); prevfd = -1; for (lp = n->npipe.cmdlist ; lp ; lp = lp->next) { prehash(lp->n); pip[1] = -1; if (lp->next) { if (pipe(pip) < 0) { close(prevfd); error("Pipe call failed: %s", strerror(errno)); } } if (forkshell(jp, lp->n, n->npipe.backgnd) == 0) { INTON; if (prevfd > 0) { dup2(prevfd, 0); close(prevfd); } if (pip[1] >= 0) { if (!(prevfd >= 0 && pip[0] == 0)) close(pip[0]); if (pip[1] != 1) { dup2(pip[1], 1); close(pip[1]); } } evaltree(lp->n, EV_EXIT); } if (prevfd >= 0) close(prevfd); prevfd = pip[0]; if (pip[1] != -1) close(pip[1]); } INTON; if (n->npipe.backgnd == 0) { INTOFF; exitstatus = waitforjob(jp, (int *)NULL); TRACE(("evalpipe: job done exit status %d\n", exitstatus)); INTON; } else exitstatus = 0; } static int is_valid_fast_cmdsubst(union node *n) { union node *argp; if (n->type != NCMD) return 0; for (argp = n->ncmd.args ; argp ; argp = argp->narg.next) if (expandhassideeffects(argp->narg.text)) return 0; return 1; } /* * Execute a command inside back quotes. If it's a builtin command, we * want to save its output in a block obtained from malloc. Otherwise * we fork off a subprocess and get the output of the command via a pipe. * Should be called with interrupts off. */ void evalbackcmd(union node *n, struct backcmd *result) { int pip[2]; struct job *jp; struct stackmark smark; /* unnecessary */ struct jmploc jmploc; struct jmploc *savehandler; setstackmark(&smark); result->fd = -1; result->buf = NULL; result->nleft = 0; result->jp = NULL; if (n == NULL) { exitstatus = 0; goto out; } if (is_valid_fast_cmdsubst(n)) { exitstatus = oexitstatus; savehandler = handler; if (setjmp(jmploc.loc)) { if (exception == EXERROR || exception == EXEXEC) exitstatus = 2; else if (exception != 0) { handler = savehandler; longjmp(handler->loc, 1); } } else { handler = &jmploc; evalcommand(n, EV_BACKCMD, result); } handler = savehandler; } else { exitstatus = 0; if (pipe(pip) < 0) error("Pipe call failed: %s", strerror(errno)); jp = makejob(n, 1); if (forkshell(jp, n, FORK_NOJOB) == 0) { FORCEINTON; close(pip[0]); if (pip[1] != 1) { dup2(pip[1], 1); close(pip[1]); } evaltree(n, EV_EXIT); } close(pip[1]); result->fd = pip[0]; result->jp = jp; } out: popstackmark(&smark); TRACE(("evalbackcmd done: fd=%d buf=%p nleft=%d jp=%p\n", result->fd, result->buf, result->nleft, result->jp)); } /* * Check if a builtin can safely be executed in the same process, * even though it should be in a subshell (command substitution). * Note that jobid, jobs, times and trap can show information not * available in a child process; this is deliberate. * The arguments should already have been expanded. */ static int safe_builtin(int idx, int argc, char **argv) { if (idx == BLTINCMD || idx == COMMANDCMD || idx == ECHOCMD || idx == FALSECMD || idx == JOBIDCMD || idx == JOBSCMD || idx == KILLCMD || idx == PRINTFCMD || idx == PWDCMD || idx == TESTCMD || idx == TIMESCMD || idx == TRUECMD || idx == TYPECMD) return (1); if (idx == EXPORTCMD || idx == TRAPCMD || idx == ULIMITCMD || idx == UMASKCMD) return (argc <= 1 || (argc == 2 && argv[1][0] == '-')); if (idx == SETCMD) return (argc <= 1 || (argc == 2 && (argv[1][0] == '-' || argv[1][0] == '+') && argv[1][1] == 'o' && argv[1][2] == '\0')); return (0); } /* * Execute a simple command. * Note: This may or may not return if (flags & EV_EXIT). */ static void evalcommand(union node *cmd, int flags, struct backcmd *backcmd) { struct stackmark smark; union node *argp; struct arglist arglist; struct arglist varlist; char **argv; int argc; char **envp; int varflag; struct strlist *sp; int mode; int pip[2]; struct cmdentry cmdentry; struct job *jp; struct jmploc jmploc; struct jmploc *savehandler; char *savecmdname; struct shparam saveparam; struct localvar *savelocalvars; struct parsefile *savetopfile; volatile int e; char *lastarg; int realstatus; int do_clearcmdentry; const char *path = pathval(); /* First expand the arguments. */ TRACE(("evalcommand(%p, %d) called\n", (void *)cmd, flags)); setstackmark(&smark); arglist.lastp = &arglist.list; varlist.lastp = &varlist.list; varflag = 1; jp = NULL; do_clearcmdentry = 0; oexitstatus = exitstatus; exitstatus = 0; for (argp = cmd->ncmd.args ; argp ; argp = argp->narg.next) { if (varflag && isassignment(argp->narg.text)) { expandarg(argp, &varlist, EXP_VARTILDE); continue; } expandarg(argp, &arglist, EXP_FULL | EXP_TILDE); varflag = 0; } *arglist.lastp = NULL; *varlist.lastp = NULL; expredir(cmd->ncmd.redirect); argc = 0; for (sp = arglist.list ; sp ; sp = sp->next) argc++; /* Add one slot at the beginning for tryexec(). */ argv = stalloc(sizeof (char *) * (argc + 2)); argv++; for (sp = arglist.list ; sp ; sp = sp->next) { TRACE(("evalcommand arg: %s\n", sp->text)); *argv++ = sp->text; } *argv = NULL; lastarg = NULL; if (iflag && funcnest == 0 && argc > 0) lastarg = argv[-1]; argv -= argc; /* Print the command if xflag is set. */ if (xflag) { char sep = 0; const char *p; out2str(ps4val()); for (sp = varlist.list ; sp ; sp = sp->next) { if (sep != 0) out2c(' '); p = strchr(sp->text, '='); if (p != NULL) { p++; outbin(sp->text, p - sp->text, out2); out2qstr(p); } else out2qstr(sp->text); sep = ' '; } for (sp = arglist.list ; sp ; sp = sp->next) { if (sep != 0) out2c(' '); /* Disambiguate command looking like assignment. */ if (sp == arglist.list && strchr(sp->text, '=') != NULL && strchr(sp->text, '\'') == NULL) { out2c('\''); out2str(sp->text); out2c('\''); } else out2qstr(sp->text); sep = ' '; } out2c('\n'); flushout(&errout); } /* Now locate the command. */ if (argc == 0) { /* Variable assignment(s) without command */ cmdentry.cmdtype = CMDBUILTIN; cmdentry.u.index = BLTINCMD; cmdentry.special = 0; } else { static const char PATH[] = "PATH="; int cmd_flags = 0, bltinonly = 0; /* * Modify the command lookup path, if a PATH= assignment * is present */ for (sp = varlist.list ; sp ; sp = sp->next) if (strncmp(sp->text, PATH, sizeof(PATH) - 1) == 0) { path = sp->text + sizeof(PATH) - 1; /* * On `PATH=... command`, we need to make * sure that the command isn't using the * non-updated hash table of the outer PATH * setting and we need to make sure that * the hash table isn't filled with items * from the temporary setting. * * It would be better to forbit using and * updating the table while this command * runs, by the command finding mechanism * is heavily integrated with hash handling, * so we just delete the hash before and after * the command runs. Partly deleting like * changepatch() does doesn't seem worth the * bookinging effort, since most such runs add * directories in front of the new PATH. */ clearcmdentry(); do_clearcmdentry = 1; } for (;;) { if (bltinonly) { cmdentry.u.index = find_builtin(*argv, &cmdentry.special); if (cmdentry.u.index < 0) { cmdentry.u.index = BLTINCMD; argv--; argc++; break; } } else find_command(argv[0], &cmdentry, cmd_flags, path); /* implement the bltin and command builtins here */ if (cmdentry.cmdtype != CMDBUILTIN) break; if (cmdentry.u.index == BLTINCMD) { if (argc == 1) break; argv++; argc--; bltinonly = 1; } else if (cmdentry.u.index == COMMANDCMD) { if (argc == 1) break; if (!strcmp(argv[1], "-p")) { if (argc == 2) break; if (argv[2][0] == '-') { if (strcmp(argv[2], "--")) break; if (argc == 3) break; argv += 3; argc -= 3; } else { argv += 2; argc -= 2; } path = _PATH_STDPATH; clearcmdentry(); do_clearcmdentry = 1; } else if (!strcmp(argv[1], "--")) { if (argc == 2) break; argv += 2; argc -= 2; } else if (argv[1][0] == '-') break; else { argv++; argc--; } cmd_flags |= DO_NOFUNC; bltinonly = 0; } else break; } /* * Special builtins lose their special properties when * called via 'command'. */ if (cmd_flags & DO_NOFUNC) cmdentry.special = 0; } /* Fork off a child process if necessary. */ if (cmd->ncmd.backgnd || ((cmdentry.cmdtype == CMDNORMAL || cmdentry.cmdtype == CMDUNKNOWN) && ((flags & EV_EXIT) == 0 || have_traps())) || ((flags & EV_BACKCMD) != 0 && (cmdentry.cmdtype != CMDBUILTIN || !safe_builtin(cmdentry.u.index, argc, argv)))) { jp = makejob(cmd, 1); mode = cmd->ncmd.backgnd; if (flags & EV_BACKCMD) { mode = FORK_NOJOB; if (pipe(pip) < 0) error("Pipe call failed: %s", strerror(errno)); } if (forkshell(jp, cmd, mode) != 0) goto parent; /* at end of routine */ if (flags & EV_BACKCMD) { FORCEINTON; close(pip[0]); if (pip[1] != 1) { dup2(pip[1], 1); close(pip[1]); } } flags |= EV_EXIT; } /* This is the child process if a fork occurred. */ /* Execute the command. */ if (cmdentry.cmdtype == CMDFUNCTION) { #ifdef DEBUG trputs("Shell function: "); trargs(argv); #endif saveparam = shellparam; shellparam.malloc = 0; shellparam.reset = 1; shellparam.nparam = argc - 1; shellparam.p = argv + 1; shellparam.optnext = NULL; INTOFF; savelocalvars = localvars; localvars = NULL; reffunc(cmdentry.u.func); savehandler = handler; if (setjmp(jmploc.loc)) { freeparam(&shellparam); shellparam = saveparam; popredir(); unreffunc(cmdentry.u.func); poplocalvars(); localvars = savelocalvars; funcnest--; handler = savehandler; longjmp(handler->loc, 1); } handler = &jmploc; funcnest++; redirect(cmd->ncmd.redirect, REDIR_PUSH); INTON; for (sp = varlist.list ; sp ; sp = sp->next) mklocal(sp->text); exitstatus = oexitstatus; evaltree(getfuncnode(cmdentry.u.func), flags & (EV_TESTED | EV_EXIT)); INTOFF; unreffunc(cmdentry.u.func); poplocalvars(); localvars = savelocalvars; freeparam(&shellparam); shellparam = saveparam; handler = savehandler; funcnest--; popredir(); INTON; if (evalskip == SKIPFUNC) { evalskip = 0; skipcount = 0; } if (jp) exitshell(exitstatus); } else if (cmdentry.cmdtype == CMDBUILTIN) { #ifdef DEBUG trputs("builtin command: "); trargs(argv); #endif mode = (cmdentry.u.index == EXECCMD)? 0 : REDIR_PUSH; if (flags == EV_BACKCMD) { memout.nleft = 0; memout.nextc = memout.buf; memout.bufsize = 64; mode |= REDIR_BACKQ; cmdentry.special = 0; } savecmdname = commandname; savetopfile = getcurrentfile(); cmdenviron = varlist.list; e = -1; savehandler = handler; if (setjmp(jmploc.loc)) { e = exception; if (e == EXINT) exitstatus = SIGINT+128; else if (e != EXEXIT) exitstatus = 2; goto cmddone; } handler = &jmploc; redirect(cmd->ncmd.redirect, mode); /* * If there is no command word, redirection errors should * not be fatal but assignment errors should. */ if (argc == 0 && !(flags & EV_BACKCMD)) cmdentry.special = 1; listsetvar(cmdenviron, cmdentry.special ? 0 : VNOSET); if (argc > 0) bltinsetlocale(); commandname = argv[0]; argptr = argv + 1; nextopt_optptr = NULL; /* initialize nextopt */ builtin_flags = flags; exitstatus = (*builtinfunc[cmdentry.u.index])(argc, argv); flushall(); cmddone: if (argc > 0) bltinunsetlocale(); cmdenviron = NULL; out1 = &output; out2 = &errout; freestdout(); handler = savehandler; commandname = savecmdname; if (jp) exitshell(exitstatus); if (flags == EV_BACKCMD) { backcmd->buf = memout.buf; backcmd->nleft = memout.nextc - memout.buf; memout.buf = NULL; } if (cmdentry.u.index != EXECCMD) popredir(); if (e != -1) { if ((e != EXERROR && e != EXEXEC) || cmdentry.special) exraise(e); popfilesupto(savetopfile); if (flags != EV_BACKCMD) FORCEINTON; } } else { #ifdef DEBUG trputs("normal command: "); trargs(argv); #endif redirect(cmd->ncmd.redirect, 0); for (sp = varlist.list ; sp ; sp = sp->next) setvareq(sp->text, VEXPORT|VSTACK); envp = environment(); shellexec(argv, envp, path, cmdentry.u.index); /*NOTREACHED*/ } goto out; parent: /* parent process gets here (if we forked) */ if (mode == FORK_FG) { /* argument to fork */ INTOFF; exitstatus = waitforjob(jp, &realstatus); INTON; if (iflag && loopnest > 0 && WIFSIGNALED(realstatus)) { evalskip = SKIPBREAK; skipcount = loopnest; } } else if (mode == FORK_NOJOB) { backcmd->fd = pip[0]; close(pip[1]); backcmd->jp = jp; } else exitstatus = 0; out: if (lastarg) setvar("_", lastarg, 0); if (do_clearcmdentry) clearcmdentry(); popstackmark(&smark); } /* * Search for a command. This is called before we fork so that the * location of the command will be available in the parent as well as * the child. The check for "goodname" is an overly conservative * check that the name will not be subject to expansion. */ static void prehash(union node *n) { struct cmdentry entry; if (n && n->type == NCMD && n->ncmd.args) if (goodname(n->ncmd.args->narg.text)) find_command(n->ncmd.args->narg.text, &entry, 0, pathval()); } /* * Builtin commands. Builtin commands whose functions are closely * tied to evaluation are implemented here. */ /* * No command given, a bltin command with no arguments, or a bltin command * with an invalid name. */ int bltincmd(int argc, char **argv) { if (argc > 1) { out2fmt_flush("%s: not found\n", argv[1]); return 127; } /* * Preserve exitstatus of a previous possible redirection * as POSIX mandates */ return exitstatus; } /* * Handle break and continue commands. Break, continue, and return are * all handled by setting the evalskip flag. The evaluation routines * above all check this flag, and if it is set they start skipping * commands rather than executing them. The variable skipcount is * the number of loops to break/continue, or the number of function * levels to return. (The latter is always 1.) It should probably * be an error to break out of more loops than exist, but it isn't * in the standard shell so we don't make it one here. */ int breakcmd(int argc, char **argv) { int n = argc > 1 ? number(argv[1]) : 1; if (n > loopnest) n = loopnest; if (n > 0) { evalskip = (**argv == 'c')? SKIPCONT : SKIPBREAK; skipcount = n; } return 0; } /* * The `command' command. */ int commandcmd(int argc, char **argv) { const char *path; int ch; int cmd = -1; path = bltinlookup("PATH", 1); optind = optreset = 1; opterr = 0; while ((ch = getopt(argc, argv, "pvV")) != -1) { switch (ch) { case 'p': path = _PATH_STDPATH; break; case 'v': cmd = TYPECMD_SMALLV; break; case 'V': cmd = TYPECMD_BIGV; break; case '?': default: error("unknown option: -%c", optopt); } } argc -= optind; argv += optind; if (cmd != -1) { if (argc != 1) error("wrong number of arguments"); return typecmd_impl(2, argv - 1, cmd, path); } if (argc != 0) error("commandcmd bad call"); /* * Do nothing successfully if no command was specified; * ksh also does this. */ return 0; } /* * The return command. */ int returncmd(int argc, char **argv) { int ret = argc > 1 ? number(argv[1]) : oexitstatus; if (funcnest) { evalskip = SKIPFUNC; skipcount = 1; } else { /* skip the rest of the file */ evalskip = SKIPFILE; skipcount = 1; } return ret; } int falsecmd(int argc __unused, char **argv __unused) { return 1; } int truecmd(int argc __unused, char **argv __unused) { return 0; } int execcmd(int argc, char **argv) { /* * Because we have historically not supported any options, * only treat "--" specially. */ if (argc > 1 && strcmp(argv[1], "--") == 0) argc--, argv++; if (argc > 1) { struct strlist *sp; iflag = 0; /* exit on error */ mflag = 0; optschanged(); for (sp = cmdenviron; sp ; sp = sp->next) setvareq(sp->text, VEXPORT|VSTACK); shellexec(argv + 1, environment(), pathval(), 0); } return 0; } int timescmd(int argc __unused, char **argv __unused) { struct rusage ru; long shumins, shsmins, chumins, chsmins; double shusecs, shssecs, chusecs, chssecs; if (getrusage(RUSAGE_SELF, &ru) < 0) return 1; shumins = ru.ru_utime.tv_sec / 60; shusecs = ru.ru_utime.tv_sec % 60 + ru.ru_utime.tv_usec / 1000000.; shsmins = ru.ru_stime.tv_sec / 60; shssecs = ru.ru_stime.tv_sec % 60 + ru.ru_stime.tv_usec / 1000000.; if (getrusage(RUSAGE_CHILDREN, &ru) < 0) return 1; chumins = ru.ru_utime.tv_sec / 60; chusecs = ru.ru_utime.tv_sec % 60 + ru.ru_utime.tv_usec / 1000000.; chsmins = ru.ru_stime.tv_sec / 60; chssecs = ru.ru_stime.tv_sec % 60 + ru.ru_stime.tv_usec / 1000000.; out1fmt("%ldm%.3fs %ldm%.3fs\n%ldm%.3fs %ldm%.3fs\n", shumins, shusecs, shsmins, shssecs, chumins, chusecs, chsmins, chssecs); return 0; } Index: projects/largeSMP/bin/sh/options.c =================================================================== --- projects/largeSMP/bin/sh/options.c (revision 222782) +++ projects/largeSMP/bin/sh/options.c (revision 222783) @@ -1,570 +1,572 @@ /*- * 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[] = "@(#)options.c 8.2 (Berkeley) 5/4/95"; #endif #endif /* not lint */ #include __FBSDID("$FreeBSD$"); #include #include #include #include "shell.h" #define DEFINE_OPTIONS #include "options.h" #undef DEFINE_OPTIONS #include "nodes.h" /* for other header files */ #include "eval.h" #include "jobs.h" #include "input.h" #include "output.h" #include "trap.h" #include "var.h" #include "memalloc.h" #include "error.h" #include "mystring.h" #ifndef NO_HISTORY #include "myhistedit.h" #endif char *arg0; /* value of $0 */ struct shparam shellparam; /* current positional parameters */ char **argptr; /* argument list for builtin commands */ char *shoptarg; /* set by nextopt (like getopt) */ char *nextopt_optptr; /* used by nextopt */ char *minusc; /* argument to -c option */ static void options(int); static void minus_o(char *, int); static void setoption(int, int); static int getopts(char *, char *, char **, char ***, char **); /* * Process the shell command line arguments. */ void procargs(int argc, char **argv) { int i; + char *scriptname; argptr = argv; if (argc > 0) argptr++; for (i = 0; i < NOPTS; i++) optlist[i].val = 2; privileged = (getuid() != geteuid() || getgid() != getegid()); options(1); if (*argptr == NULL && minusc == NULL) sflag = 1; if (iflag != 0 && sflag == 1 && isatty(0) && isatty(1)) { iflag = 1; if (Eflag == 2) Eflag = 1; } if (mflag == 2) mflag = iflag; for (i = 0; i < NOPTS; i++) if (optlist[i].val == 2) optlist[i].val = 0; arg0 = argv[0]; if (sflag == 0 && minusc == NULL) { - commandname = arg0 = *argptr++; - setinputfile(commandname, 0); + scriptname = *argptr++; + setinputfile(scriptname, 0); + commandname = arg0 = scriptname; } /* POSIX 1003.2: first arg after -c cmd is $0, remainder $1... */ if (argptr && minusc && *argptr) arg0 = *argptr++; shellparam.p = argptr; shellparam.reset = 1; /* assert(shellparam.malloc == 0 && shellparam.nparam == 0); */ while (*argptr) { shellparam.nparam++; argptr++; } optschanged(); } void optschanged(void) { setinteractive(iflag); #ifndef NO_HISTORY histedit(); #endif setjobctl(mflag); } /* * Process shell options. The global variable argptr contains a pointer * to the argument list; we advance it past the options. */ static void options(int cmdline) { char *kp, *p; int val; int c; if (cmdline) minusc = NULL; while ((p = *argptr) != NULL) { argptr++; if ((c = *p++) == '-') { val = 1; /* A "-" or "--" terminates options */ if (p[0] == '\0') goto end_options1; if (p[0] == '-' && p[1] == '\0') goto end_options2; /** * For the benefit of `#!' lines in shell scripts, * treat a string of '-- *#.*' the same as '--'. * This is needed so that a script starting with: * #!/bin/sh -- # -*- perl -*- * will continue to work after a change is made to * kern/imgact_shell.c to NOT token-ize the options * specified on a '#!' line. A bit of a kludge, * but that trick is recommended in documentation * for some scripting languages, and we might as * well continue to support it. */ if (p[0] == '-') { kp = p + 1; while (*kp == ' ' || *kp == '\t') kp++; if (*kp == '#' || *kp == '\0') goto end_options2; } } else if (c == '+') { val = 0; } else { argptr--; break; } while ((c = *p++) != '\0') { if (c == 'c' && cmdline) { char *q; #ifdef NOHACK /* removing this code allows sh -ce 'foo' for compat */ if (*p == '\0') #endif q = *argptr++; if (q == NULL || minusc != NULL) error("Bad -c option"); minusc = q; #ifdef NOHACK break; #endif } else if (c == 'o') { minus_o(*argptr, val); if (*argptr) argptr++; } else setoption(c, val); } } return; /* When processing `set', a single "-" means turn off -x and -v */ end_options1: if (!cmdline) { xflag = vflag = 0; return; } /* * When processing `set', a "--" means the remaining arguments * replace the positional parameters in the active shell. If * there are no remaining options, then all the positional * parameters are cleared (equivalent to doing ``shift $#''). */ end_options2: if (!cmdline) { if (*argptr == NULL) setparam(argptr); return; } /* * At this point we are processing options given to 'sh' on a command * line. If an end-of-options marker ("-" or "--") is followed by an * arg of "#", then skip over all remaining arguments. Some scripting * languages (e.g.: perl) document that /bin/sh will implement this * behavior, and they recommend that users take advantage of it to * solve certain issues that can come up when writing a perl script. * Yes, this feature is in /bin/sh to help users write perl scripts. */ p = *argptr; if (p != NULL && p[0] == '#' && p[1] == '\0') { while (*argptr != NULL) argptr++; /* We need to keep the final argument */ argptr--; } } static void minus_o(char *name, int val) { int i; if (name == NULL) { if (val) { /* "Pretty" output. */ out1str("Current option settings\n"); for (i = 0; i < NOPTS; i++) out1fmt("%-16s%s\n", optlist[i].name, optlist[i].val ? "on" : "off"); } else { /* Output suitable for re-input to shell. */ for (i = 0; i < NOPTS; i++) out1fmt("%s %co %s%s", i % 6 == 0 ? "set" : "", optlist[i].val ? '-' : '+', optlist[i].name, i % 6 == 5 || i == NOPTS - 1 ? "\n" : ""); } } else { for (i = 0; i < NOPTS; i++) if (equal(name, optlist[i].name)) { setoption(optlist[i].letter, val); return; } error("Illegal option -o %s", name); } } static void setoption(int flag, int val) { int i; if (flag == 'p' && !val && privileged) { if (setgid(getgid()) == -1) error("setgid"); if (setuid(getuid()) == -1) error("setuid"); } for (i = 0; i < NOPTS; i++) if (optlist[i].letter == flag) { optlist[i].val = val; if (val) { /* #%$ hack for ksh semantics */ if (flag == 'V') Eflag = 0; else if (flag == 'E') Vflag = 0; } return; } error("Illegal option -%c", flag); } /* * Set the shell parameters. */ void setparam(char **argv) { char **newparam; char **ap; int nparam; for (nparam = 0 ; argv[nparam] ; nparam++); ap = newparam = ckmalloc((nparam + 1) * sizeof *ap); while (*argv) { *ap++ = savestr(*argv++); } *ap = NULL; freeparam(&shellparam); shellparam.malloc = 1; shellparam.nparam = nparam; shellparam.p = newparam; shellparam.reset = 1; shellparam.optnext = NULL; } /* * Free the list of positional parameters. */ void freeparam(struct shparam *param) { char **ap; if (param->malloc) { for (ap = param->p ; *ap ; ap++) ckfree(*ap); ckfree(param->p); } } /* * The shift builtin command. */ int shiftcmd(int argc, char **argv) { int n; char **ap1, **ap2; n = 1; if (argc > 1) n = number(argv[1]); if (n > shellparam.nparam) return 1; INTOFF; shellparam.nparam -= n; for (ap1 = shellparam.p ; --n >= 0 ; ap1++) { if (shellparam.malloc) ckfree(*ap1); } ap2 = shellparam.p; while ((*ap2++ = *ap1++) != NULL); shellparam.reset = 1; INTON; return 0; } /* * The set command builtin. */ int setcmd(int argc, char **argv) { if (argc == 1) return showvarscmd(argc, argv); INTOFF; options(0); optschanged(); if (*argptr != NULL) { setparam(argptr); } INTON; return 0; } void getoptsreset(const char *value) { if (number(value) == 1) { shellparam.reset = 1; } } /* * The getopts builtin. Shellparam.optnext points to the next argument * to be processed. Shellparam.optptr points to the next character to * be processed in the current argument. If shellparam.optnext is NULL, * then it's the first time getopts has been called. */ int getoptscmd(int argc, char **argv) { char **optbase = NULL; if (argc < 3) error("usage: getopts optstring var [arg]"); else if (argc == 3) optbase = shellparam.p; else optbase = &argv[3]; if (shellparam.reset == 1) { shellparam.optnext = optbase; shellparam.optptr = NULL; shellparam.reset = 0; } return getopts(argv[1], argv[2], optbase, &shellparam.optnext, &shellparam.optptr); } static int getopts(char *optstr, char *optvar, char **optfirst, char ***optnext, char **optptr) { char *p, *q; char c = '?'; int done = 0; int ind = 0; int err = 0; char s[10]; if ((p = *optptr) == NULL || *p == '\0') { /* Current word is done, advance */ if (*optnext == NULL) return 1; p = **optnext; if (p == NULL || *p != '-' || *++p == '\0') { atend: ind = *optnext - optfirst + 1; *optnext = NULL; p = NULL; done = 1; goto out; } (*optnext)++; if (p[0] == '-' && p[1] == '\0') /* check for "--" */ goto atend; } c = *p++; for (q = optstr; *q != c; ) { if (*q == '\0') { if (optstr[0] == ':') { s[0] = c; s[1] = '\0'; err |= setvarsafe("OPTARG", s, 0); } else { out1fmt("Illegal option -%c\n", c); (void) unsetvar("OPTARG"); } c = '?'; goto bad; } if (*++q == ':') q++; } if (*++q == ':') { if (*p == '\0' && (p = **optnext) == NULL) { if (optstr[0] == ':') { s[0] = c; s[1] = '\0'; err |= setvarsafe("OPTARG", s, 0); c = ':'; } else { out1fmt("No arg for -%c option\n", c); (void) unsetvar("OPTARG"); c = '?'; } goto bad; } if (p == **optnext) (*optnext)++; setvarsafe("OPTARG", p, 0); p = NULL; } else setvarsafe("OPTARG", "", 0); ind = *optnext - optfirst + 1; goto out; bad: ind = 1; *optnext = NULL; p = NULL; out: *optptr = p; fmtstr(s, sizeof(s), "%d", ind); err |= setvarsafe("OPTIND", s, VNOFUNC); s[0] = c; s[1] = '\0'; err |= setvarsafe(optvar, s, 0); if (err) { *optnext = NULL; *optptr = NULL; flushall(); exraise(EXERROR); } return done; } /* * XXX - should get rid of. have all builtins use getopt(3). the * library getopt must have the BSD extension static variable "optreset" * otherwise it can't be used within the shell safely. * * Standard option processing (a la getopt) for builtin routines. The * only argument that is passed to nextopt is the option string; the * other arguments are unnecessary. It return the character, or '\0' on * end of input. */ int nextopt(const char *optstring) { char *p; const char *q; char c; if ((p = nextopt_optptr) == NULL || *p == '\0') { p = *argptr; if (p == NULL || *p != '-' || *++p == '\0') return '\0'; argptr++; if (p[0] == '-' && p[1] == '\0') /* check for "--" */ return '\0'; } c = *p++; for (q = optstring ; *q != c ; ) { if (*q == '\0') error("Illegal option -%c", c); if (*++q == ':') q++; } if (*++q == ':') { if (*p == '\0' && (p = *argptr++) == NULL) error("No arg for -%c option", c); shoptarg = p; p = NULL; } nextopt_optptr = p; return c; } Index: projects/largeSMP/cddl/contrib/opensolaris =================================================================== --- projects/largeSMP/cddl/contrib/opensolaris (revision 222782) +++ projects/largeSMP/cddl/contrib/opensolaris (revision 222783) Property changes on: projects/largeSMP/cddl/contrib/opensolaris ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/cddl/contrib/opensolaris:r222698-222782 Index: projects/largeSMP/contrib/bind9 =================================================================== --- projects/largeSMP/contrib/bind9 (revision 222782) +++ projects/largeSMP/contrib/bind9 (revision 222783) Property changes on: projects/largeSMP/contrib/bind9 ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/bind9:r222698-222782 Index: projects/largeSMP/contrib/binutils =================================================================== --- projects/largeSMP/contrib/binutils (revision 222782) +++ projects/largeSMP/contrib/binutils (revision 222783) Property changes on: projects/largeSMP/contrib/binutils ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/binutils:r222698-222782 Index: projects/largeSMP/contrib/bzip2 =================================================================== --- projects/largeSMP/contrib/bzip2 (revision 222782) +++ projects/largeSMP/contrib/bzip2 (revision 222783) Property changes on: projects/largeSMP/contrib/bzip2 ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/bzip2:r222698-222782 Index: projects/largeSMP/contrib/compiler-rt =================================================================== --- projects/largeSMP/contrib/compiler-rt (revision 222782) +++ projects/largeSMP/contrib/compiler-rt (revision 222783) Property changes on: projects/largeSMP/contrib/compiler-rt ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/compiler-rt:r222698-222782 Index: projects/largeSMP/contrib/dialog =================================================================== --- projects/largeSMP/contrib/dialog (revision 222782) +++ projects/largeSMP/contrib/dialog (revision 222783) Property changes on: projects/largeSMP/contrib/dialog ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/dialog:r222698-222782 Index: projects/largeSMP/contrib/ee =================================================================== --- projects/largeSMP/contrib/ee (revision 222782) +++ projects/largeSMP/contrib/ee (revision 222783) Property changes on: projects/largeSMP/contrib/ee ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/ee:r222698-222782 Index: projects/largeSMP/contrib/expat =================================================================== --- projects/largeSMP/contrib/expat (revision 222782) +++ projects/largeSMP/contrib/expat (revision 222783) Property changes on: projects/largeSMP/contrib/expat ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/expat:r222698-222782 Index: projects/largeSMP/contrib/file =================================================================== --- projects/largeSMP/contrib/file (revision 222782) +++ projects/largeSMP/contrib/file (revision 222783) Property changes on: projects/largeSMP/contrib/file ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/file:r222698-222782 Index: projects/largeSMP/contrib/gcc =================================================================== --- projects/largeSMP/contrib/gcc (revision 222782) +++ projects/largeSMP/contrib/gcc (revision 222783) Property changes on: projects/largeSMP/contrib/gcc ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/gcc:r222698-222782 Index: projects/largeSMP/contrib/gdb =================================================================== --- projects/largeSMP/contrib/gdb (revision 222782) +++ projects/largeSMP/contrib/gdb (revision 222783) Property changes on: projects/largeSMP/contrib/gdb ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/gdb:r222698-222782 Index: projects/largeSMP/contrib/gdtoa =================================================================== --- projects/largeSMP/contrib/gdtoa (revision 222782) +++ projects/largeSMP/contrib/gdtoa (revision 222783) Property changes on: projects/largeSMP/contrib/gdtoa ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/gdtoa:r222698-222782 Index: projects/largeSMP/contrib/gnu-sort =================================================================== --- projects/largeSMP/contrib/gnu-sort (revision 222782) +++ projects/largeSMP/contrib/gnu-sort (revision 222783) Property changes on: projects/largeSMP/contrib/gnu-sort ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/gnu-sort:r222698-222782 Index: projects/largeSMP/contrib/groff =================================================================== --- projects/largeSMP/contrib/groff (revision 222782) +++ projects/largeSMP/contrib/groff (revision 222783) Property changes on: projects/largeSMP/contrib/groff ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/groff:r222698-222782 Index: projects/largeSMP/contrib/less =================================================================== --- projects/largeSMP/contrib/less (revision 222782) +++ projects/largeSMP/contrib/less (revision 222783) Property changes on: projects/largeSMP/contrib/less ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/less:r222698-222782 Index: projects/largeSMP/contrib/libpcap =================================================================== --- projects/largeSMP/contrib/libpcap (revision 222782) +++ projects/largeSMP/contrib/libpcap (revision 222783) Property changes on: projects/largeSMP/contrib/libpcap ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/libpcap:r222698-222782 Index: projects/largeSMP/contrib/libstdc++ =================================================================== --- projects/largeSMP/contrib/libstdc++ (revision 222782) +++ projects/largeSMP/contrib/libstdc++ (revision 222783) Property changes on: projects/largeSMP/contrib/libstdc++ ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/libstdc++:r222698-222782 Index: projects/largeSMP/contrib/llvm/tools/clang =================================================================== --- projects/largeSMP/contrib/llvm/tools/clang (revision 222782) +++ projects/largeSMP/contrib/llvm/tools/clang (revision 222783) Property changes on: projects/largeSMP/contrib/llvm/tools/clang ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/llvm/tools/clang:r222698-222782 Index: projects/largeSMP/contrib/llvm =================================================================== --- projects/largeSMP/contrib/llvm (revision 222782) +++ projects/largeSMP/contrib/llvm (revision 222783) Property changes on: projects/largeSMP/contrib/llvm ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/llvm:r222698-222782 Index: projects/largeSMP/contrib/ncurses =================================================================== --- projects/largeSMP/contrib/ncurses (revision 222782) +++ projects/largeSMP/contrib/ncurses (revision 222783) Property changes on: projects/largeSMP/contrib/ncurses ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/ncurses:r222698-222782 Index: projects/largeSMP/contrib/netcat =================================================================== --- projects/largeSMP/contrib/netcat (revision 222782) +++ projects/largeSMP/contrib/netcat (revision 222783) Property changes on: projects/largeSMP/contrib/netcat ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/netcat:r222698-222782 Index: projects/largeSMP/contrib/ntp =================================================================== --- projects/largeSMP/contrib/ntp (revision 222782) +++ projects/largeSMP/contrib/ntp (revision 222783) Property changes on: projects/largeSMP/contrib/ntp ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/ntp:r222698-222782 Index: projects/largeSMP/contrib/one-true-awk =================================================================== --- projects/largeSMP/contrib/one-true-awk (revision 222782) +++ projects/largeSMP/contrib/one-true-awk (revision 222783) Property changes on: projects/largeSMP/contrib/one-true-awk ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/one-true-awk:r222698-222782 Index: projects/largeSMP/contrib/openbsm =================================================================== --- projects/largeSMP/contrib/openbsm (revision 222782) +++ projects/largeSMP/contrib/openbsm (revision 222783) Property changes on: projects/largeSMP/contrib/openbsm ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/openbsm:r222698-222782 Index: projects/largeSMP/contrib/openpam =================================================================== --- projects/largeSMP/contrib/openpam (revision 222782) +++ projects/largeSMP/contrib/openpam (revision 222783) Property changes on: projects/largeSMP/contrib/openpam ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/openpam:r222698-222782 Index: projects/largeSMP/contrib/pf =================================================================== --- projects/largeSMP/contrib/pf (revision 222782) +++ projects/largeSMP/contrib/pf (revision 222783) Property changes on: projects/largeSMP/contrib/pf ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/pf:r222698-222782 Index: projects/largeSMP/contrib/sendmail =================================================================== --- projects/largeSMP/contrib/sendmail (revision 222782) +++ projects/largeSMP/contrib/sendmail (revision 222783) Property changes on: projects/largeSMP/contrib/sendmail ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/sendmail:r222698-222782 Index: projects/largeSMP/contrib/tcpdump =================================================================== --- projects/largeSMP/contrib/tcpdump (revision 222782) +++ projects/largeSMP/contrib/tcpdump (revision 222783) Property changes on: projects/largeSMP/contrib/tcpdump ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/tcpdump:r222698-222782 Index: projects/largeSMP/contrib/tcsh =================================================================== --- projects/largeSMP/contrib/tcsh (revision 222782) +++ projects/largeSMP/contrib/tcsh (revision 222783) Property changes on: projects/largeSMP/contrib/tcsh ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/tcsh:r222698-222782 Index: projects/largeSMP/contrib/top/install-sh =================================================================== --- projects/largeSMP/contrib/top/install-sh (revision 222782) +++ projects/largeSMP/contrib/top/install-sh (revision 222783) Property changes on: projects/largeSMP/contrib/top/install-sh ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/top/install-sh:r222698-222782 Index: projects/largeSMP/contrib/top =================================================================== --- projects/largeSMP/contrib/top (revision 222782) +++ projects/largeSMP/contrib/top (revision 222783) Property changes on: projects/largeSMP/contrib/top ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/top:r222698-222782 Index: projects/largeSMP/contrib/tzcode/stdtime =================================================================== --- projects/largeSMP/contrib/tzcode/stdtime (revision 222782) +++ projects/largeSMP/contrib/tzcode/stdtime (revision 222783) Property changes on: projects/largeSMP/contrib/tzcode/stdtime ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/tzcode/stdtime:r222698-222782 Index: projects/largeSMP/contrib/tzcode/zic =================================================================== --- projects/largeSMP/contrib/tzcode/zic (revision 222782) +++ projects/largeSMP/contrib/tzcode/zic (revision 222783) Property changes on: projects/largeSMP/contrib/tzcode/zic ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/tzcode/zic:r222698-222782 Index: projects/largeSMP/contrib/tzdata =================================================================== --- projects/largeSMP/contrib/tzdata (revision 222782) +++ projects/largeSMP/contrib/tzdata (revision 222783) Property changes on: projects/largeSMP/contrib/tzdata ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/tzdata:r222698-222782 Index: projects/largeSMP/contrib/wpa =================================================================== --- projects/largeSMP/contrib/wpa (revision 222782) +++ projects/largeSMP/contrib/wpa (revision 222783) Property changes on: projects/largeSMP/contrib/wpa ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/wpa:r222698-222782 Index: projects/largeSMP/contrib/xz =================================================================== --- projects/largeSMP/contrib/xz (revision 222782) +++ projects/largeSMP/contrib/xz (revision 222783) Property changes on: projects/largeSMP/contrib/xz ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/xz:r222698-222782 Index: projects/largeSMP/crypto/openssh =================================================================== --- projects/largeSMP/crypto/openssh (revision 222782) +++ projects/largeSMP/crypto/openssh (revision 222783) Property changes on: projects/largeSMP/crypto/openssh ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/crypto/openssh:r222698-222782 Index: projects/largeSMP/crypto/openssl =================================================================== --- projects/largeSMP/crypto/openssl (revision 222782) +++ projects/largeSMP/crypto/openssl (revision 222783) Property changes on: projects/largeSMP/crypto/openssl ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/crypto/openssl:r222698-222782 Index: projects/largeSMP/etc/network.subr =================================================================== --- projects/largeSMP/etc/network.subr (revision 222782) +++ projects/largeSMP/etc/network.subr (revision 222783) @@ -1,1395 +1,1389 @@ # # Copyright (c) 2003 The FreeBSD 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. # # 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. # # $FreeBSD$ # # # Subroutines commonly used from network startup scripts. # Requires that rc.conf be loaded first. # # ifn_start ifn # Bring up and configure an interface. If some configuration is # applied print the interface configuration. # ifn_start() { local ifn cfg ifn="$1" cfg=1 [ -z "$ifn" ] && err 1 "ifn_start called without an interface" ifscript_up ${ifn} && cfg=0 ifconfig_up ${ifn} && cfg=0 afexists inet && ipv4_up ${ifn} && cfg=0 afexists inet6 && ipv6_up ${ifn} && cfg=0 afexists ipx && ipx_up ${ifn} && cfg=0 childif_create ${ifn} && cfg=0 return $cfg } # ifn_stop ifn # Shutdown and de-configure an interface. If action is taken # print the interface name. # ifn_stop() { local ifn cfg ifn="$1" cfg=1 [ -z "$ifn" ] && err 1 "ifn_stop called without an interface" afexists ipx && ipx_down ${ifn} && cfg=0 afexists inet6 && ipv6_down ${ifn} && cfg=0 afexists inet && ipv4_down ${ifn} && cfg=0 ifconfig_down ${ifn} && cfg=0 ifscript_down ${ifn} && cfg=0 childif_destroy ${ifn} && cfg=0 return $cfg } # ifconfig_up if # Evaluate ifconfig(8) arguments for interface $if and # run ifconfig(8) with those arguments. It returns 0 if # arguments were found and executed or 1 if the interface # had no arguments. Pseudo arguments DHCP and WPA are handled # here. # ifconfig_up() { local _cfg _ipv6_opts ifconfig_args _cfg=1 # Make sure lo0 always comes up. if [ "$1" = "lo0" ]; then _cfg=0 fi # ifconfig_IF ifconfig_args=`ifconfig_getargs $1` if [ -n "${ifconfig_args}" ]; then ifconfig $1 ${ifconfig_args} _cfg=0 fi # inet6 specific if afexists inet6; then - if ipv6if $1; then - if checkyesno ipv6_gateway_enable; then - _ipv6_opts="-accept_rtadv" - fi - else - if checkyesno ipv6_activate_all_interfaces; then - _ipv6_opts="-ifdisabled" - else - _ipv6_opts="ifdisabled" - fi - - # backward compatibility: $ipv6_enable - case $ipv6_enable in - [Yy][Ee][Ss]|[Tt][Rr][Uu][Ee]|[Oo][Nn]|1) - _ipv6_opts="${_ipv6_opts} accept_rtadv" - ;; - esac + if checkyesno ipv6_activate_all_interfaces; then + _ipv6_opts="-ifdisabled" + elif [ "$1" != "lo0" ]; then + _ipv6_opts="ifdisabled" fi + + # backward compatibility: $ipv6_enable + case $ipv6_enable in + [Yy][Ee][Ss]|[Tt][Rr][Uu][Ee]|[Oo][Nn]|1) + _ipv6_opts="${_ipv6_opts} accept_rtadv" + ;; + esac if [ -n "${_ipv6_opts}" ]; then ifconfig $1 inet6 ${_ipv6_opts} fi # ifconfig_IF_ipv6 ifconfig_args=`ifconfig_getargs $1 ipv6` if [ -n "${ifconfig_args}" ]; then ifconfig $1 inet6 -ifdisabled ifconfig $1 ${ifconfig_args} _cfg=0 fi # backward compatiblity: $ipv6_ifconfig_IF ifconfig_args=`get_if_var $1 ipv6_ifconfig_IF` if [ -n "${ifconfig_args}" ]; then warn "\$ipv6_ifconfig_$1 is obsolete." \ " Use ifconfig_$1_ipv6 instead." ifconfig $1 inet6 -ifdisabled ifconfig $1 inet6 ${ifconfig_args} _cfg=0 fi fi if [ ${_cfg} -eq 0 ]; then ifconfig $1 up fi if wpaif $1; then /etc/rc.d/wpa_supplicant start $1 _cfg=0 # XXX: not sure this should count fi if dhcpif $1; then if [ $_cfg -ne 0 ] ; then ifconfig $1 up fi if syncdhcpif $1; then /etc/rc.d/dhclient start $1 fi _cfg=0 fi return $_cfg } # ifconfig_down if # returns 1 if wpa_supplicant or dhclient was stopped or # the interface exists. # ifconfig_down() { local _cfg _cfg=1 if wpaif $1; then /etc/rc.d/wpa_supplicant stop $1 _cfg=0 fi if dhcpif $1; then /etc/rc.d/dhclient stop $1 _cfg=0 fi if ifexists $1; then ifconfig $1 down _cfg=0 fi return $_cfg } # get_if_var if var [default] # Return the value of the pseudo-hash corresponding to $if where # $var is a string containg the sub-string "IF" which will be # replaced with $if after the characters defined in _punct are # replaced with '_'. If the variable is unset, replace it with # $default if given. get_if_var() { local _if _punct _punct_c _var _default prefix suffix if [ $# -ne 2 -a $# -ne 3 ]; then err 3 'USAGE: get_if_var name var [default]' fi _if=$1 _punct=". - / +" for _punct_c in $_punct; do _if=`ltr ${_if} ${_punct_c} '_'` done _var=$2 _default=$3 prefix=${_var%%IF*} suffix=${_var##*IF} eval echo \${${prefix}${_if}${suffix}-${_default}} } # _ifconfig_getargs if [af] # Echos the arguments for the supplied interface to stdout. # returns 1 if empty. In general, ifconfig_getargs should be used # outside this file. _ifconfig_getargs() { local _ifn _af _ifn=$1 _af=${2+_$2} if [ -z "$_ifn" ]; then return 1 fi get_if_var $_ifn ifconfig_IF$_af "$ifconfig_DEFAULT" } # ifconfig_getargs if [af] # Takes the result from _ifconfig_getargs and removes pseudo # args such as DHCP and WPA. ifconfig_getargs() { local _tmpargs _arg _args _tmpargs=`_ifconfig_getargs $1 $2` if [ $? -eq 1 ]; then return 1 fi _args= for _arg in $_tmpargs; do case $_arg in [Dd][Hh][Cc][Pp]) ;; [Nn][Oo][Aa][Uu][Tt][Oo]) ;; [Nn][Oo][Ss][Yy][Nn][Cc][Dd][Hh][Cc][Pp]) ;; [Ss][Yy][Nn][Cc][Dd][Hh][Cc][Pp]) ;; [Ww][Pp][Aa]) ;; *) _args="$_args $_arg" ;; esac done echo $_args } # autoif # Returns 0 if the interface should be automaticly configured at # boot time and 1 otherwise. autoif() { local _tmpargs _arg _tmpargs=`_ifconfig_getargs $1` for _arg in $_tmpargs; do case $_arg in [Nn][Oo][Aa][Uu][Tt][Oo]) return 1 ;; esac done return 0 } # dhcpif if # Returns 0 if the interface is a DHCP interface and 1 otherwise. dhcpif() { local _tmpargs _arg _tmpargs=`_ifconfig_getargs $1` for _arg in $_tmpargs; do case $_arg in [Dd][Hh][Cc][Pp]) return 0 ;; [Nn][Oo][Ss][Yy][Nn][Cc][Dd][Hh][Cc][Pp]) return 0 ;; [Ss][Yy][Nn][Cc][Dd][Hh][Cc][Pp]) return 0 ;; esac done return 1 } # syncdhcpif # Returns 0 if the interface should be configured synchronously and # 1 otherwise. syncdhcpif() { local _tmpargs _arg _tmpargs=`_ifconfig_getargs $1` for _arg in $_tmpargs; do case $_arg in [Nn][Oo][Ss][Yy][Nn][Cc][Dd][Hh][Cc][Pp]) return 1 ;; [Ss][Yy][Nn][Cc][Dd][Hh][Cc][Pp]) return 0 ;; esac done checkyesno synchronous_dhclient } # wpaif if # Returns 0 if the interface is a WPA interface and 1 otherwise. wpaif() { local _tmpargs _arg _tmpargs=`_ifconfig_getargs $1` for _arg in $_tmpargs; do case $_arg in [Ww][Pp][Aa]) return 0 ;; esac done return 1 } # afexists af # Returns 0 if the address family is enabled in the kernel # 1 otherwise. afexists() { local _af _af=$1 case ${_af} in inet) ${SYSCTL_N} kern.features.inet > /dev/null 2>&1 ;; inet6) ${SYSCTL_N} kern.features.inet6 > /dev/null 2>&1 ;; ipx) ${SYSCTL_N} net.ipx > /dev/null 2>&1 ;; atm) if [ -x /sbin/atmconfig ]; then /sbin/atmconfig diag list > /dev/null 2>&1 else return 1 fi ;; *) err 1 "afexists(): Unsupported address family: $_af" ;; esac } # noafif if # Returns 0 if the interface has no af configuration and 1 otherwise. noafif() { local _if _if=$1 case $_if in pflog[0-9]*|\ pfsync[0-9]*|\ an[0-9]*|\ ath[0-9]*|\ ipw[0-9]*|\ ipfw[0-9]*|\ iwi[0-9]*|\ iwn[0-9]*|\ ral[0-9]*|\ wi[0-9]*|\ wl[0-9]*|\ wpi[0-9]*) return 0 ;; esac return 1 } # ipv6if if # Returns 0 if the interface should be configured for IPv6 and # 1 otherwise. ipv6if() { local _if _tmpargs i _if=$1 if ! afexists inet6; then return 1 fi # lo0 is always IPv6-enabled case $_if in lo0) return 0 ;; esac case "${ipv6_network_interfaces}" in $_if|"$_if "*|*" $_if"|*" $_if "*|[Aa][Uu][Tt][Oo]) # True if $ifconfig_IF_ipv6 is defined. _tmpargs=`_ifconfig_getargs $_if ipv6` if [ -n "${_tmpargs}" ]; then return 0 fi # backward compatibility: True if $ipv6_ifconfig_IF is defined. _tmpargs=`get_if_var $_if ipv6_ifconfig_IF` if [ -n "${_tmpargs}" ]; then return 0 fi ;; esac return 1 } # ipv6_autoconfif if # Returns 0 if the interface should be configured for IPv6 with # Stateless Address Configuration, 1 otherwise. ipv6_autoconfif() { local _if _tmpargs _arg _if=$1 case $_if in lo0|\ stf[0-9]*|\ faith[0-9]*|\ lp[0-9]*|\ sl[0-9]*) return 1 ;; esac if noafif $_if; then return 1 fi if ! ipv6if $_if; then return 1 fi if checkyesno ipv6_gateway_enable; then return 1 fi _tmpargs=`get_if_var $_if ipv6_prefix_IF` if [ -n "${_tmpargs}" ]; then return 1 fi # backward compatibility: $ipv6_enable case $ipv6_enable in [Yy][Ee][Ss]|[Tt][Rr][Uu][Ee]|[Oo][Nn]|1) return 0 ;; esac _tmpargs=`_ifconfig_getargs $_if ipv6` for _arg in $_tmpargs; do case $_arg in accept_rtadv) return 0 ;; esac done # backward compatibility: $ipv6_ifconfig_IF _tmpargs=`get_if_var $_if ipv6_ifconfig_IF` for _arg in $_tmpargs; do case $_arg in accept_rtadv) return 0 ;; esac done return 1 } # ifexists if # Returns 0 if the interface exists and 1 otherwise. ifexists() { [ -z "$1" ] && return 1 ifconfig -n $1 > /dev/null 2>&1 } # ipv4_up if # add IPv4 addresses to the interface $if ipv4_up() { local _if _ret _if=$1 _ret=1 # Add 127.0.0.1/8 to lo0 unless otherwise specified. if [ "${_if}" = "lo0" ]; then ifconfig_args=`ifconfig_getargs ${_if}` if [ -z "${ifconfig_args}" ]; then ifconfig ${_if} inet 127.0.0.1/8 alias fi fi ifalias_up ${_if} inet && _ret=0 ipv4_addrs_common ${_if} alias && _ret=0 return $_ret } # ipv6_up if # add IPv6 addresses to the interface $if ipv6_up() { local _if _ret _if=$1 _ret=1 if ! ipv6if $_if; then return 0 fi ifalias_up ${_if} inet6 && _ret=0 ipv6_prefix_hostid_addr_up ${_if} && _ret=0 ipv6_accept_rtadv_up ${_if} && _ret=0 # wait for DAD sleep `${SYSCTL_N} net.inet6.ip6.dad_count` sleep 1 return $_ret } # ipv4_down if # remove IPv4 addresses from the interface $if ipv4_down() { local _if _ifs _ret inetList oldifs _inet _if=$1 _ifs="^" _ret=1 inetList="`ifconfig ${_if} | grep 'inet ' | tr "\n" "$_ifs"`" oldifs="$IFS" IFS="$_ifs" for _inet in $inetList ; do # get rid of extraneous line [ -z "$_inet" ] && break _inet=`expr "$_inet" : '.*\(inet \([0-9]\{1,3\}\.\)\{3\}[0-9]\{1,3\}\).*'` IFS="$oldifs" ifconfig ${_if} ${_inet} delete IFS="$_ifs" _ret=0 done IFS="$oldifs" ifalias_down ${_if} inet && _ret=0 ipv4_addrs_common ${_if} -alias && _ret=0 return $_ret } # ipv6_down if # remove IPv6 addresses from the interface $if ipv6_down() { local _if _ifs _ret inetList oldifs _inet6 _if=$1 _ifs="^" _ret=1 if ! ipv6if $_if; then return 0 fi ipv6_accept_rtadv_down ${_if} && _ret=0 ifalias_down ${_if} inet6 && _ret=0 inetList="`ifconfig ${_if} | grep 'inet6 ' | tr "\n" "$_ifs"`" oldifs="$IFS" IFS="$_ifs" for _inet6 in $inetList ; do # get rid of extraneous line [ -z "$_inet6" ] && break _inet6=`expr "$_inet6" : '.*\(inet6 \([0-9a-f:]*\)\).*'` IFS="$oldifs" ifconfig ${_if} ${_inet6} -alias IFS="$_ifs" _ret=0 done IFS="$oldifs" return $_ret } # ipv4_addrs_common if action # Evaluate the ifconfig_if_ipv4 arguments for interface $if and # use $action to add or remove IPv4 addresses from $if. ipv4_addrs_common() { local _ret _if _action _cidr _cidr_addr local _ipaddr _netmask _range _ipnet _iplow _iphigh _ipcount _ret=1 _if=$1 _action=$2 # get ipv4-addresses cidr_addr=`get_if_var $_if ipv4_addrs_IF` for _cidr in ${cidr_addr}; do _ipaddr=${_cidr%%/*} _netmask="/"${_cidr##*/} _range=${_ipaddr##*.} _ipnet=${_ipaddr%.*} _iplow=${_range%-*} _iphigh=${_range#*-} # clear netmask when removing aliases if [ "${_action}" = "-alias" ]; then _netmask="" fi _ipcount=${_iplow} while [ "${_ipcount}" -le "${_iphigh}" ]; do eval "ifconfig ${_if} ${_action} ${_ipnet}.${_ipcount}${_netmask}" _ipcount=$((${_ipcount}+1)) _ret=0 # only the first ipaddr in a subnet need the real netmask if [ "${_action}" != "-alias" ]; then _netmask="/32" fi done done return $_ret } # ifalias_up if af # Configure aliases for network interface $if. # It returns 0 if at least one alias was configured or # 1 if there were none. # ifalias_up() { local _ret _ret=1 case "$2" in inet) _ret=`ifalias_ipv4_up "$1"` ;; inet6) _ret=`ifalias_ipv6_up "$1"` ;; esac return $_ret } # ifalias_ipv4_up if # Helper function for ifalias_up(). Handles IPv4. # ifalias_ipv4_up() { local _ret alias ifconfig_args _ret=1 # ifconfig_IF_aliasN which starts with "inet" alias=0 while : ; do ifconfig_args=`get_if_var $1 ifconfig_IF_alias${alias}` case "${ifconfig_args}" in inet\ *) ifconfig $1 ${ifconfig_args} alias && _ret=0 ;; "") break ;; esac alias=$((${alias} + 1)) done return $_ret } # ifalias_ipv6_up if # Helper function for ifalias_up(). Handles IPv6. # ifalias_ipv6_up() { local _ret alias ifconfig_args _ret=1 # ifconfig_IF_aliasN which starts with "inet6" alias=0 while : ; do ifconfig_args=`get_if_var $1 ifconfig_IF_alias${alias}` case "${ifconfig_args}" in inet6\ *) ifconfig $1 ${ifconfig_args} alias && _ret=0 ;; "") break ;; esac alias=$((${alias} + 1)) done # backward compatibility: ipv6_ifconfig_IF_aliasN. alias=0 while : ; do ifconfig_args=`get_if_var $1 ipv6_ifconfig_IF_alias${alias}` case "${ifconfig_args}" in "") break ;; *) ifconfig $1 inet6 ${ifconfig_args} alias && _ret=0 warn "\$ipv6_ifconfig_$1_alias${alias} is obsolete." \ " Use ifconfig_$1_aliasN instead." ;; esac alias=$((${alias} + 1)) done return $_ret } # ifalias_down if af # Remove aliases for network interface $if. # It returns 0 if at least one alias was removed or # 1 if there were none. # ifalias_down() { local _ret _ret=1 case "$2" in inet) _ret=`ifalias_ipv4_down "$1"` ;; inet6) _ret=`ifalias_ipv6_down "$1"` ;; esac return $_ret } # ifalias_ipv4_down if # Helper function for ifalias_down(). Handles IPv4. # ifalias_ipv4_down() { local _ret alias ifconfig_args _ret=1 # ifconfig_IF_aliasN which starts with "inet" alias=0 while : ; do ifconfig_args=`get_if_var $1 ifconfig_IF_alias${alias}` case "${ifconfig_args}" in inet\ *) ifconfig $1 ${ifconfig_args} -alias && _ret=0 ;; "") break ;; esac alias=$((${alias} + 1)) done return $_ret } # ifalias_ipv6_down if # Helper function for ifalias_down(). Handles IPv6. # ifalias_ipv6_down() { local _ret alias ifconfig_args _ret=1 # ifconfig_IF_aliasN which starts with "inet6" alias=0 while : ; do ifconfig_args=`get_if_var $1 ifconfig_IF_alias${alias}` case "${ifconfig_args}" in inet6\ *) ifconfig $1 ${ifconfig_args} -alias && _ret=0 ;; "") break ;; esac alias=$((${alias} + 1)) done # backward compatibility: ipv6_ifconfig_IF_aliasN. alias=0 while : ; do ifconfig_args=`get_if_var $1 ipv6_ifconfig_IF_alias${alias}` case "${ifconfig_args}" in "") break ;; *) ifconfig $1 inet6 ${ifconfig_args} -alias && _ret=0 warn "\$ipv6_ifconfig_$1_alias${alias} is obsolete." \ " Use ifconfig_$1_aliasN instead." ;; esac alias=$((${alias} + 1)) done return $_ret } # ipv6_prefix_hostid_addr_up if # add IPv6 prefix + hostid addr to the interface $if ipv6_prefix_hostid_addr_up() { local _if prefix laddr hostid j address _if=$1 prefix=`get_if_var ${_if} ipv6_prefix_IF` if [ -n "${prefix}" ]; then laddr=`network6_getladdr ${_if}` hostid=${laddr#fe80::} hostid=${hostid%\%*} for j in ${prefix}; do address=$j\:${hostid} ifconfig ${_if} inet6 ${address} prefixlen 64 alias # if I am a router, add subnet router # anycast address (RFC 2373). if checkyesno ipv6_gateway_enable; then ifconfig ${_if} inet6 $j:: prefixlen 64 \ alias anycast fi done fi } # ipv6_accept_rtadv_up if # Enable accepting Router Advertisement and send Router # Solicitation message ipv6_accept_rtadv_up() { if ipv6_autoconfif $1; then ifconfig $1 inet6 accept_rtadv up if ! checkyesno rtsold_enable; then rtsol ${rtsol_flags} $1 fi fi } # ipv6_accept_rtadv_down if # Disable accepting Router Advertisement ipv6_accept_rtadv_down() { if ipv6_autoconfif $1; then ifconfig $1 inet6 -accept_rtadv fi } # ifscript_up if # Evaluate a startup script for the $if interface. # It returns 0 if a script was found and processed or # 1 if no script was found. # ifscript_up() { if [ -r /etc/start_if.$1 ]; then . /etc/start_if.$1 return 0 else return 1 fi } # ifscript_down if # Evaluate a shutdown script for the $if interface. # It returns 0 if a script was found and processed or # 1 if no script was found. # ifscript_down() { if [ -r /etc/stop_if.$1 ]; then . /etc/stop_if.$1 return 0 else return 1 fi } # clone_up # Create cloneable interfaces. # clone_up() { local _prefix _list ifn _prefix= _list= # create_args_IF for ifn in ${cloned_interfaces}; do ifconfig ${ifn} create `get_if_var ${ifn} create_args_IF` if [ $? -eq 0 ]; then _list="${_list}${_prefix}${ifn}" [ -z "$_prefix" ] && _prefix=' ' fi done debug "Cloned: ${_list}" } # clone_down # Destroy cloned interfaces. Destroyed interfaces are echoed to # standard output. # clone_down() { local _prefix _list ifn _prefix= _list= for ifn in ${cloned_interfaces}; do ifconfig -n ${ifn} destroy if [ $? -eq 0 ]; then _list="${_list}${_prefix}${ifn}" [ -z "$_prefix" ] && _prefix=' ' fi done debug "Destroyed clones: ${_list}" } # childif_create # Create and configure child interfaces. Return 0 if child # interfaces are created. # childif_create() { local cfg child child_vlans child_wlans create_args debug_flags ifn i cfg=1 ifn=$1 # Create wireless interfaces child_wlans=`get_if_var $ifn wlans_IF` for child in ${child_wlans}; do create_args="wlandev $ifn `get_if_var $child create_args_IF`" debug_flags="`get_if_var $child wlandebug_IF`" if expr $child : 'wlan[0-9][0-9]*$' >/dev/null 2>&1; then ifconfig $child create ${create_args} && cfg=0 if [ -n "${debug_flags}" ]; then wlandebug -i $child ${debug_flags} fi else i=`ifconfig wlan create ${create_args}` if [ -n "${debug_flags}" ]; then wlandebug -i $i ${debug_flags} fi ifconfig $i name $child && cfg=0 fi if autoif $child; then ifn_start $child fi done # Create vlan interfaces child_vlans=`get_if_var $ifn vlans_IF` if [ -n "${child_vlans}" ]; then load_kld if_vlan fi for child in ${child_vlans}; do if expr $child : '[1-9][0-9]*$' >/dev/null 2>&1; then child="${ifn}.${child}" create_args=`get_if_var $child create_args_IF` ifconfig $child create ${create_args} && cfg=0 else create_args="vlandev $ifn `get_if_var $child create_args_IF`" if expr $child : 'vlan[0-9][0-9]*$' >/dev/null 2>&1; then ifconfig $child create ${create_args} && cfg=0 else i=`ifconfig vlan create ${create_args}` ifconfig $i name $child && cfg=0 fi fi if autoif $child; then ifn_start $child fi done return ${cfg} } # childif_destroy # Destroy child interfaces. # childif_destroy() { local cfg child child_vlans child_wlans ifn cfg=1 child_wlans=`get_if_var $ifn wlans_IF` for child in ${child_wlans}; do if ! ifexists $child; then continue fi ifconfig -n $child destroy && cfg=0 done child_vlans=`get_if_var $ifn vlans_IF` for child in ${child_vlans}; do if expr $child : '[1-9][0-9]*$' >/dev/null 2>&1; then child="${ifn}.${child}" fi if ! ifexists $child; then continue fi ifconfig -n $child destroy && cfg=0 done return ${cfg} } # ng_mkpeer # Create netgraph nodes. # ng_mkpeer() { ngctl -f - 2> /dev/null </dev/null 2>&1; then ifconfig $i create >/dev/null 2>&1 else gif=`ifconfig gif create` ifconfig $gif name $i fi ifconfig $i tunnel ${peers} ifconfig $i up ;; esac done } # ng_fec_create ifn # Configure Fast EtherChannel for interface $ifn. Returns 0 if # FEC arguments were found and configured; returns !0 otherwise. ng_fec_create() { local req_iface iface bogus req_iface="$1" ngctl shutdown ${req_iface}: > /dev/null 2>&1 bogus="" while true; do iface=`ng_create_one fec dummy fec` if [ -z "${iface}" ]; then exit 2 fi if [ "${iface}" = "${req_iface}" ]; then break fi bogus="${bogus} ${iface}" done for iface in ${bogus}; do ngctl shutdown ${iface}: done } # fec_up # Create Fast EtherChannel interfaces. fec_up() { local i j for i in ${fec_interfaces}; do ng_fec_create $i for j in `get_if_var $i fecconfig_IF`; do case ${j} in '') continue ;; *) ngctl msg ${i}: add_iface "\"${j}\"" ;; esac done done } # ipx_up ifn # Configure any IPX addresses for interface $ifn. Returns 0 if # IPX arguments were found and configured; returns 1 otherwise. # ipx_up() { local ifn ifn="$1" # ifconfig_IF_ipx ifconfig_args=`_ifconfig_getargs $ifn ipx` if [ -n "${ifconfig_args}" ]; then ifconfig ${ifn} ${ifconfig_args} return 0 fi return 1 } # ipx_down ifn # Remove IPX addresses for interface $ifn. Returns 0 if IPX # addresses were found and unconfigured. It returns 1, otherwise. # ipx_down() { local _if _ifs _ret ipxList oldifs _ipx _if=$1 _ifs="^" _ret=1 ipxList="`ifconfig ${_if} | grep 'ipx ' | tr "\n" "$_ifs"`" oldifs="$IFS" IFS="$_ifs" for _ipx in $ipxList ; do # get rid of extraneous line [ -z "$_ipx" ] && break _ipx=`expr "$_ipx" : '.*\(ipx [0-9a-h]\{1,8\}H*\.[0-9a-h]\{1,12\}\).*'` IFS="$oldifs" ifconfig ${_if} ${_ipx} delete IFS="$_ifs" _ret=0 done IFS="$oldifs" return $_ret } # ifnet_rename # Rename all requested interfaces. # ifnet_rename() { local _if _ifname # ifconfig_IF_name for _if in `ifconfig -l`; do _ifname=`get_if_var $_if ifconfig_IF_name` if [ ! -z "$_ifname" ]; then ifconfig $_if name $_ifname fi done return 0 } # list_net_interfaces type # List all network interfaces. The type of interface returned # can be controlled by the type argument. The type # argument can be any of the following: # nodhcp - all interfaces, excluding DHCP configured interfaces # dhcp - list only DHCP configured interfaces # noautoconf - all interfaces, excluding IPv6 Stateless # Address Autoconf configured interfaces # autoconf - list only IPv6 Stateless Address Autoconf # configured interfaces # If no argument is specified all network interfaces are output. # Note that the list will include cloned interfaces if applicable. # Cloned interfaces must already exist to have a chance to appear # in the list if ${network_interfaces} is set to `auto'. # list_net_interfaces() { local type _tmplist _list _autolist _lo _if type=$1 # Get a list of ALL the interfaces and make lo0 first if it's there. # _tmplist= case ${network_interfaces} in [Aa][Uu][Tt][Oo]) _autolist="`ifconfig -l`" _lo= for _if in ${_autolist} ; do if autoif $_if; then if [ "$_if" = "lo0" ]; then _lo="lo0 " else _tmplist="${_tmplist} ${_if}" fi fi done _tmplist="${_lo}${_tmplist# }" ;; *) _tmplist="${network_interfaces} ${cloned_interfaces}" # lo0 is effectively mandatory, so help prevent foot-shooting # case "$_tmplist" in lo0|'lo0 '*|*' lo0'|*' lo0 '*) ;; # This is fine, do nothing *) _tmplist="lo0 ${_tmplist}" ;; esac ;; esac _list= case "$type" in nodhcp) for _if in ${_tmplist} ; do if ! dhcpif $_if && \ [ -n "`_ifconfig_getargs $_if`" ]; then _list="${_list# } ${_if}" fi done ;; dhcp) for _if in ${_tmplist} ; do if dhcpif $_if; then _list="${_list# } ${_if}" fi done ;; noautoconf) for _if in ${_tmplist} ; do if ! ipv6_autoconfif $_if && \ [ -n "`_ifconfig_getargs $_if ipv6`" ]; then _list="${_list# } ${_if}" fi done ;; autoconf) for _if in ${_tmplist} ; do if ipv6_autoconfif $_if; then _list="${_list# } ${_if}" fi done ;; *) _list=${_tmplist} ;; esac echo $_list return 0 } # get_default_if -address_family # Get the interface of the default route for the given address family. # The -address_family argument must be suitable passing to route(8). # get_default_if() { local routeget oldifs defif line defif= oldifs="$IFS" IFS=" " for line in `route -n get $1 default 2>/dev/null`; do case $line in *interface:*) defif=${line##*: } ;; esac done IFS=${oldifs} echo $defif } # hexdigit arg # Echo decimal number $arg (single digit) in hexadecimal format. hexdigit() { printf '%x\n' "$1" } # hexprint arg # Echo decimal number $arg (multiple digits) in hexadecimal format. hexprint() { printf '%x\n' "$1" } is_wired_interface() { local media case `ifconfig $1 2>/dev/null` in *media:?Ethernet*) media=Ethernet ;; esac test "$media" = "Ethernet" } # network6_getladdr if [flag] # Echo link-local address from $if if any. # If flag is defined, tentative ones will be excluded. network6_getladdr() { local proto addr rest ifconfig $1 2>/dev/null | while read proto addr rest; do case ${proto} in inet6) case ${addr} in fe80::*) if [ -z "$2" ]; then echo ${addr} return fi case ${rest} in *tentative*) continue ;; *) echo ${addr} return esac esac esac done } Index: projects/largeSMP/gnu/lib =================================================================== --- projects/largeSMP/gnu/lib (revision 222782) +++ projects/largeSMP/gnu/lib (revision 222783) Property changes on: projects/largeSMP/gnu/lib ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/gnu/lib:r222698-222782 Index: projects/largeSMP/gnu/usr.bin/binutils =================================================================== --- projects/largeSMP/gnu/usr.bin/binutils (revision 222782) +++ projects/largeSMP/gnu/usr.bin/binutils (revision 222783) Property changes on: projects/largeSMP/gnu/usr.bin/binutils ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/gnu/usr.bin/binutils:r222698-222782 Index: projects/largeSMP/gnu/usr.bin/cc/cc_tools =================================================================== --- projects/largeSMP/gnu/usr.bin/cc/cc_tools (revision 222782) +++ projects/largeSMP/gnu/usr.bin/cc/cc_tools (revision 222783) Property changes on: projects/largeSMP/gnu/usr.bin/cc/cc_tools ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/gnu/usr.bin/cc/cc_tools:r222698-222782 Index: projects/largeSMP/gnu/usr.bin/gdb =================================================================== --- projects/largeSMP/gnu/usr.bin/gdb (revision 222782) +++ projects/largeSMP/gnu/usr.bin/gdb (revision 222783) Property changes on: projects/largeSMP/gnu/usr.bin/gdb ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/gnu/usr.bin/gdb:r222698-222782 Index: projects/largeSMP/lib/libc/stdtime =================================================================== --- projects/largeSMP/lib/libc/stdtime (revision 222782) +++ projects/largeSMP/lib/libc/stdtime (revision 222783) Property changes on: projects/largeSMP/lib/libc/stdtime ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/lib/libc/stdtime:r222698-222782 Index: projects/largeSMP/lib/libc =================================================================== --- projects/largeSMP/lib/libc (revision 222782) +++ projects/largeSMP/lib/libc (revision 222783) Property changes on: projects/largeSMP/lib/libc ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/lib/libc:r222698-222782 Index: projects/largeSMP/lib/libiconv/Makefile =================================================================== --- projects/largeSMP/lib/libiconv/Makefile (revision 222782) +++ projects/largeSMP/lib/libiconv/Makefile (revision 222783) @@ -1,25 +1,24 @@ # $FreeBSD$ .PATH: ${.CURDIR}/../libc/iconv LIB= iconv SHLIB_MAJOR= 4 MAN= iconv.3 iconvctl.3 iconv_canonicalize.3 iconvlist.3 \ __iconv_get_list.3 MLNKS= iconv.3 iconv_open.3 \ iconv.3 iconv_open_into.3 \ iconv.3 iconv_close.3 \ iconv.3 __iconv.3 \ __iconv_get_list.3 __iconv_free_list.3 SRCS= citrus_bcs.c citrus_bcs_strtol.c citrus_bcs_strtoul.c \ citrus_csmapper.c citrus_db.c citrus_db_factory.c \ citrus_db_hash.c citrus_esdb.c citrus_hash.c \ citrus_iconv.c citrus_lookup.c citrus_lookup_factory.c \ citrus_mapper.c citrus_memstream.c citrus_mmap.c \ citrus_module.c citrus_none.c citrus_pivot_factory.c \ citrus_prop.c citrus_stdenc.c iconv.c -WARNS?= 6 CFLAGS+= --param max-inline-insns-single=128 -I ${.CURDIR}/../../include -I${.CURDIR}/../libc/include .include Index: projects/largeSMP/lib/libutil =================================================================== --- projects/largeSMP/lib/libutil (revision 222782) +++ projects/largeSMP/lib/libutil (revision 222783) Property changes on: projects/largeSMP/lib/libutil ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/lib/libutil:r222698-222782 Index: projects/largeSMP/lib/libz =================================================================== --- projects/largeSMP/lib/libz (revision 222782) +++ projects/largeSMP/lib/libz (revision 222783) Property changes on: projects/largeSMP/lib/libz ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/lib/libz:r222698-222782 Index: projects/largeSMP/release/ia64/mkisoimages.sh =================================================================== --- projects/largeSMP/release/ia64/mkisoimages.sh (revision 222782) +++ projects/largeSMP/release/ia64/mkisoimages.sh (revision 222783) @@ -1,75 +1,79 @@ #!/bin/sh # # Module: mkisoimages.sh # Author: Jordan K Hubbard # Date: 22 June 2001 # # $FreeBSD$ # # This script is used by release/Makefile to build the (optional) ISO images # for a FreeBSD release. It is considered architecture dependent since each # platform has a slightly unique way of making bootable CDs. This script # is also allowed to generate any number of images since that is more of # publishing decision than anything else. # # Usage: # # mkisoimages.sh [-b] image-label image-name base-bits-dir [extra-bits-dir] # # Where -b is passed if the ISO image should be made "bootable" by # whatever standards this architecture supports (may be unsupported), # image-label is the ISO image label, image-name is the filename of the # resulting ISO image, base-bits-dir contains the image contents and # extra-bits-dir, if provided, contains additional files to be merged # into base-bits-dir as part of making the image. set -e if [ "x$1" = "x-b" ]; then bootable=yes shift else bootable=no fi if [ $# -lt 3 ]; then echo usage: $0 '[-b] label iso-name base-dir [extra-dir]' exit 1 fi LABEL=$1; shift NAME=$1; shift BASE=$1; shift EFIPART=efipart.sys # To create a bootable CD under EFI, the boot image should be an EFI # system partition. if [ $bootable = yes ]; then EFISZ=65536 MNT=/mnt - dd if=/dev/zero of=$BASE/$EFIPART count=$EFISZ - md=`mdconfig -a -t vnode -f $BASE/$EFIPART` + dd if=/dev/zero of=$EFIPART count=$EFISZ + md=`mdconfig -a -t vnode -f $EFIPART` newfs_msdos -F 12 -S 512 -h 4 -o 0 -s $EFISZ -u 16 $md mount -t msdosfs /dev/$md $MNT mkdir -p $MNT/efi/boot $MNT/boot $MNT/boot/kernel cp -R $BASE/boot/defaults $MNT/boot cp $BASE/boot/kernel/kernel $MNT/boot/kernel - cp $BASE/boot/kernel/ispfw.ko $MNT/boot/kernel + if [ -s $BASE/boot/kernel/ispfw.ko ]; then + cp $BASE/boot/kernel/ispfw.ko $MNT/boot/kernel + fi cp $BASE/boot/device.hints $MNT/boot cp $BASE/boot/loader.* $MNT/boot - cp $BASE/boot/mfsroot.gz $MNT/boot + if [ -s $BASE/boot/mfsroot.gz ]; then + cp $BASE/boot/mfsroot.gz $MNT/boot + fi cp $BASE/boot/support.4th $MNT/boot mv $MNT/boot/loader.efi $MNT/efi/boot/bootia64.efi umount $MNT mdconfig -d -u $md - BOOTOPTS="-b bootimage=i386;$EFIPART -o no-emul-boot" + BOOTOPTS="-o bootimage=i386;$EFIPART -o no-emul-boot" else BOOTOPTS="" fi -echo "/dev/iso9660/$LABEL / cd9660 ro 0 0" > $1/etc/fstab +echo "/dev/iso9660/$LABEL / cd9660 ro 0 0" > $BASE/etc/fstab makefs -t cd9660 $BOOTOPTS -o rockridge -o label=$LABEL $NAME $BASE $* -rm -f $BASE/$EFIPART -rm $1/etc/fstab +rm $BASE/etc/fstab +rm -f $EFIPART exit 0 Index: projects/largeSMP/sbin/geom/class/part/gpart.8 =================================================================== --- projects/largeSMP/sbin/geom/class/part/gpart.8 (revision 222782) +++ projects/largeSMP/sbin/geom/class/part/gpart.8 (revision 222783) @@ -1,959 +1,959 @@ .\" Copyright (c) 2007, 2008 Marcel Moolenaar .\" 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 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 May 30, 2011 +.Dd June 6, 2011 .Dt GPART 8 .Os .Sh NAME .Nm gpart .Nd "control utility for the disk partitioning GEOM class" .Sh SYNOPSIS To add support for the disk partitioning GEOM class, place one or more of the following lines in your kernel configuration file: .Bd -ragged -offset indent .Cd "options GEOM_PART_APM" .Cd "options GEOM_PART_BSD" .Cd "options GEOM_PART_EBR" .Cd "options GEOM_PART_EBR_COMPAT" .Cd "options GEOM_PART_GPT" .Cd "options GEOM_PART_MBR" .Cd "options GEOM_PART_PC98" .Cd "options GEOM_PART_VTOC8" .Ed .Pp The .Dv GEOM_PART_APM option adds support for the Apple Partition Map (APM) found on Apple Macintosh computers. The .Dv GEOM_PART_BSD option adds support for the traditional .Bx disklabel. The .Dv GEOM_PART_EBR option adds support for the Extended Boot Record (EBR), which is used to define a logical partition. The .Dv GEOM_PART_EBR_COMPAT option enables backward compatibility for partition names in the EBR scheme. Also it makes impossible any types of actions with such partitions. The .Dv GEOM_PART_GPT option adds support for the GUID Partition Table (GPT) found on Intel Itanium computers and Intel-based Macintosh computers. The .Dv GEOM_PART_MBR option adds support for the Master Boot Record (MBR) found on PCs and used on many removable media. The .Dv GEOM_PART_PC98 option adds support for the MBR variant as used on NEC PC-98 computers. The .Dv GEOM_PART_VTOC8 option adds support for Sun's SMI VTOC8 label as found on computers based on .Tn SPARC64 and .Tn UltraSPARC. .Pp Usage of the .Ns Nm utility: .Pp .\" ==== ADD ==== .Nm .Cm add .Fl t Ar type .Op Fl a Ar alignment .Op Fl b Ar start .Op Fl s Ar size .Op Fl i Ar index .Op Fl l Ar label .Op Fl f Ar flags .Ar geom .\" ==== BACKUP ==== .Nm .Cm backup .Ar geom .\" ==== BOOTCODE ==== .Nm .Cm bootcode .Op Fl b Ar bootcode .Op Fl p Ar partcode Fl i Ar index .Op Fl f Ar flags .Ar geom .\" ==== COMMIT ==== .Nm .Cm commit .Ar geom .\" ==== CREATE ==== .Nm .Cm create .Fl s Ar scheme .Op Fl n Ar entries .Op Fl f Ar flags .Ar provider .\" ==== DELETE ==== .Nm .Cm delete .Fl i Ar index .Op Fl f Ar flags .Ar geom .\" ==== DESTROY ==== .Nm .Cm destroy .Op Fl F .Op Fl f Ar flags .Ar geom .\" ==== MODIFY ==== .Nm .Cm modify .Fl i Ar index .Op Fl l Ar label .Op Fl t Ar type .Op Fl f Ar flags .Ar geom .\" ==== RECOVER ==== .Nm .Cm recover .Op Fl f Ar flags .Ar geom .\" ==== RESIZE ==== .Nm .Cm resize .Fl i Ar index .Op Fl a Ar alignment .Op Fl s Ar size .Op Fl f Ar flags .Ar geom .\" ==== RESTORE ==== .Nm .Cm restore .Op Fl lF .Op Fl f Ar flags .Ar provider .Op Ar ... .\" ==== SET ==== .Nm .Cm set .Fl a Ar attrib .Fl i Ar index .Op Fl f Ar flags .Ar geom .\" ==== SHOW ==== .Nm .Cm show .Op Fl l | r .Op Fl p .Op Ar geom ... .\" ==== UNDO ==== .Nm .Cm undo .Ar geom .\" ==== UNSET ==== .Nm .Cm unset .Fl a Ar attrib .Fl i Ar index .Op Fl f Ar flags .Ar geom .\" .Sh DESCRIPTION The .Nm utility is used to partition GEOM providers, normally disks. The first argument of which is the action to be taken: .Bl -tag -width ".Cm bootcode" .\" ==== ADD ==== .It Cm add Add a new partition to the partitioning scheme given by .Ar geom . The partition begins on the logical block address given by the .Fl b Ar start option. Its size is given by the .Fl s Ar size option. SI unit suffixes are allowed. One or both .Fl b and .Fl s options can be omitted. If so they are automatically calculated. The type of the partition is given by the .Fl t Ar type option. Partition types are discussed below in the section entitled .Sx "PARTITION TYPES" . .Pp Additional options include: .Bl -tag -width 12n .It Fl a Ar alignment If specified, then .Nm utility tries to align .Ar start offset and partition .Ar size to be multiple of .Ar alignment value. .It Fl i Ar index The index in the partition table at which the new partition is to be placed. The index determines the name of the device special file used to represent the partition. .It Fl l Ar label The label attached to the partition. This option is only valid when used on partitioning schemes that support partition labels. .It Fl f Ar flags Additional operational flags. See the section entitled .Sx "OPERATIONAL FLAGS" below for a discussion about its use. .El .\" ==== BACKUP ==== .It Cm backup Dump a partition table to standard output in special format used by .Cm restore action. .\" ==== BOOTCODE ==== .It Cm bootcode Embed bootstrap code into the partitioning scheme's metadata on the .Ar geom (using .Fl b Ar bootcode ) or write bootstrap code into a partition (using .Fl p Ar partcode and .Fl i Ar index ) . Not all partitioning schemes have embedded bootstrap code, so the .Fl b Ar bootcode option is scheme-specific in nature. For the GPT scheme, embedded bootstrap code is supported. The bootstrap code is embedded in the protective MBR rather than the GPT. The .Fl b Ar bootcode option specifies a file that contains the bootstrap code. The contents and size of the file are determined by the partitioning scheme. For the MBR scheme, it is a 512 byte file of which the first 446 bytes are installed as bootstrap code. The .Fl p Ar partcode option specifies a file that contains the bootstrap code intended to be written to a partition. For the VTOC8 scheme, it is a 8192 byte file of which the last 7680 bytes are installed as bootstrap code. The partition is specified by the .Fl i Ar index option. For the VTOC8 scheme, if the .Fl i Ar index option is omitted, the bootstrap code is written to all sufficiently large partitions. The size of the file must be smaller than the size of the partition. .Pp Additional options include: .Bl -tag -width 10n .It Fl f Ar flags Additional operational flags. See the section entitled .Sx "OPERATIONAL FLAGS" below for a discussion about its use. .El .\" ==== COMMIT ==== .It Cm commit Commit any pending changes for geom .Ar geom . All actions are being committed by default and will not result in pending changes. Actions can be modified with the .Fl f Ar flags option so that they are not being committed by default. As such, they become pending. Pending changes are reflected by the geom and the .Nm utility, but they are not actually written to disk. The .Cm commit action will write any and all pending changes to disk. .\" ==== CREATE ==== .It Cm create Create a new partitioning scheme on a provider given by .Ar provider . The .Fl s Ar scheme option determines the scheme to use. The kernel needs to have support for a particular scheme before that scheme can be used to partition a disk. .Pp Additional options include: .Bl -tag -width 10n .It Fl n Ar entries The number of entries in the partition table. Every partitioning scheme has a minimum and a maximum number of entries and this option allows tables to be created with the number of entries that lies anywhere between the minimum and the maximum. Some schemes have a maximum equal to the minimum and some schemes have a maximum large enough to be considered unlimited. By default, partition tables are created with the minimum number of entries. .It Fl f Ar flags Additional operational flags. See the section entitled .Sx "OPERATIONAL FLAGS" below for a discussion about its use. .El .\" ==== DELETE ==== .It Cm delete Delete a partition from geom .Ar geom and further identified by the .Fl i Ar index option. The partition cannot be actively used by the kernel. .Pp Additional options include: .Bl -tag -width 10n .It Fl f Ar flags Additional operational flags. See the section entitled .Sx "OPERATIONAL FLAGS" below for a discussion about its use. .El .\" ==== DESTROY ==== .It Cm destroy Destroy the partitioning scheme as implemented by geom .Ar geom . .Pp Additional options include: .Bl -tag -width 10n .It Fl F Forced destroying of the partition table even if it is not empty. .It Fl f Ar flags Additional operational flags. See the section entitled .Sx "OPERATIONAL FLAGS" below for a discussion about its use. .El .\" ==== MODIFY ==== .It Cm modify Modify a partition from geom .Ar geom and further identified by the .Fl i Ar index option. Only the the type and/or label of the partition can be modified. To change the type of a partition, specify the new type with the .Fl t Ar type option. To change the label of a partition, specify the new label with the .Fl l Ar label option. Not all partitioning schemes support labels and it is invalid to try to change a partition label in such cases. .Pp Additional options include: .Bl -tag -width 10n .It Fl f Ar flags Additional operational flags. See the section entitled .Sx "OPERATIONAL FLAGS" below for a discussion about its use. .El .\" ==== RECOVER ==== .It Cm recover Recover corrupt partition's scheme metadata on the geom .Ar geom . See the section entitled .Sx RECOVERING below for the additional information. .Pp Additional options include: .Bl -tag -width 10n .It Fl f Ar flags Additional operational flags. See the section entitled .Sx "OPERATIONAL FLAGS" below for a discussion about its use. .El .\" ==== RESIZE ==== .It Cm resize Resize a partition from geom .Ar geom and further identified by the .Fl i Ar index option. New partition size is expressed in logical block numbers and can be given by the .Fl s Ar size option. If .Fl s option is omitted then new size is automatically calculated to maximum available from given geom .Ar geom . .Pp Additional options include: .Bl -tag -width 12n .It Fl a Ar alignment If specified, then .Nm utility tries to align partition .Ar size to be multiple of .Ar alignment value. .It Fl f Ar flags Additional operational flags. See the section entitled .Sx "OPERATIONAL FLAGS" below for a discussion about its use. .El .\" ==== RESTORE ==== .It Cm restore Restore the partition table from backup previously created by .Cm backup action and given from standard input. Only partition table may be restored. This action does not affect content of partitions. This mean that you should copy your data from backup after restoring partition table and write bootcode again if it is needed. .Pp Additional options include: .Bl -tag -width 10n .It Fl F Destroy partition table on the given .Ar provider before doing restore. .It Fl l Restore partition labels for partitioning schemes that support them. .It Fl f Ar flags Additional operational flags. See the section entitled .Sx "OPERATIONAL FLAGS" below for a discussion about its use. .El .\" ==== SET ==== .It Cm set Set the named attribute on the partition entry. See the section entitled .Sx ATTRIBUTES below for a list of available attributes. .Pp Additional options include: .Bl -tag -width 10n .It Fl f Ar flags Additional operational flags. See the section entitled .Sx "OPERATIONAL FLAGS" below for a discussion about its use. .El .\" ==== SHOW ==== .It Cm show Show the current partition information of the specified geoms or all geoms if none are specified. Additional options include: .Bl -tag -width 10n .It Fl l For partition schemes that support partition labels print them instead of partition type. .It Fl p Show provider names instead of partition indexes. .It Fl r Show raw partition type instead of symbolic name. .El .\" ==== UNDO ==== .It Cm undo Revert any pending changes for geom .Ar geom . This action is the opposite of the .Cm commit action and can be used to undo any changes that have not been committed. .\" ==== UNSET ==== .It Cm unset Clear the named attribute on the partition entry. See the section entitled .Sx ATTRIBUTES below for a list of available attributes. .Pp Additional options include: .Bl -tag -width 10n .It Fl f Ar flags Additional operational flags. See the section entitled .Sx "OPERATIONAL FLAGS" below for a discussion about its use. .El .El .\" .Sh PARTITION TYPES +Partition types are identified on disk by particular strings or magic +values. The .Nm -utility uses symbolic names for common partition types to avoid that the -user needs to know what the partitioning scheme in question is and what -the actual number or identification needs to be used for a particular -type. +utility uses symbolic names for common partition types to avoid the +user needing to know these values or other details of the partitioning +scheme in question. The .Nm utility also allows the user to specify scheme-specific partition types -for partition types that do not have symbol names. +for partition types that do not have symbolic names. The symbolic names currently understood are: .Bl -tag -width ".Cm freebsd-vinum" .It Cm bios-boot The system partition dedicated to second stage of the boot loader program. Usually it used by GRUB 2 loader when the partition table is GPT. The scheme-specific type is .Qq Li "!21686148-6449-6E6F-744E-656564454649" . .It Cm efi The system partition for computers that use the Extensible Firmware Interface (EFI). In such cases, the GPT partitioning scheme is being used and the actual partition type for the system partition can also be specified as .Qq Li "!c12a7328-f81f-11d2-ba4b-00a0c93ec93ab" . .It Cm freebsd A .Fx partition that uses the .Bx disklabel to sub-divide the partition into file systems. This is a legacy partition type and should not be used for the APM or GPT schemes. The scheme-specific types are .Qq Li "!165" for MBR, .Qq Li "!FreeBSD" for APM, and .Qq Li "!516e7cb4-6ecf-11d6-8ff8-00022d09712b" for GPT. .It Cm freebsd-boot A .Fx partition dedicated to bootstrap code. The scheme-specific type is .Qq Li "!83bd6b9d-7f41-11dc-be0b-001560b84f0f" for GPT. .It Cm freebsd-swap A .Fx partition dedicated to swap space. The scheme-specific types are .Qq Li "!FreeBSD-swap" for APM, .Qq Li "!516e7cb5-6ecf-11d6-8ff8-00022d09712b" for GPT, and tag 0x0901 for VTOC8. .It Cm freebsd-ufs A .Fx partition that contains a UFS or UFS2 file system. The scheme-specific types are .Qq Li "!FreeBSD-UFS" for APM, .Qq Li "!516e7cb6-6ecf-11d6-8ff8-00022d09712b" for GPT, and tag 0x0902 for VTOC8. .It Cm freebsd-vinum A .Fx partition that contains a Vinum volume. The scheme-specific types are .Qq Li "!FreeBSD-Vinum" for APM, .Qq Li "!516e7cb8-6ecf-11d6-8ff8-00022d09712b" for GPT, and tag 0x0903 for VTOC8. .It Cm freebsd-zfs A .Fx partition that contains a ZFS volume. The scheme-specific types are .Qq Li "!FreeBSD-ZFS" for APM, .Qq Li "!516e7cba-6ecf-11d6-8ff8-00022d09712b" for GPT, and 0x0904 for VTOC8. .It Cm mbr A partition that is sub-partitioned by a master boot record (MBR). This type is known as .Qq Li "!024dee41-33e7-11d3-9d69-0008c781f39f" by GPT. .El .Sh ATTRIBUTES The scheme-specific attributes for EBR: .Bl -tag -width ".Cm active" .It Cm active .El .Pp The scheme-specific attributes for GPT: .Bl -tag -width ".Cm bootfailed" .It Cm bootme When set, the .Nm gptboot stage 1 boot loader will try to boot the system from this partition. Multiple partitions might be marked with the .Cm bootme attribute. In such scenario the .Nm gptboot will try all .Cm bootme partitions one by one, until the next boot stage is successfully entered. .It Cm bootonce Setting this attribute automatically sets the .Cm bootme attribute. When set, the .Nm gptboot stage 1 boot loader will try to boot the system from this partition only once. Partitions with both .Cm bootonce and .Cm bootme attributes are tried before partitions with only the .Cm bootme attribute. Before .Cm bootonce partition is tried, the .Nm gptboot removes the .Cm bootme attribute and tries to execute the next boot stage. If it fails, the .Cm bootonce attribute that is now alone is replaced with the .Cm bootfailed attribute. If the execution of the next boot stage succeeds, but the system is not fully booted, the .Nm gptboot will look for .Cm bootonce attributes alone (without the .Cm bootme attribute) on the next system boot and will replace those with the .Cm bootfailed attribute. If the system is fully booted, the .Pa /etc/rc.d/gptboot start-up script will look for partition with the .Cm bootonce attribute alone, will remove the attribute and log that the system was successfully booted from this partition. There should be at most one .Cm bootonce partition when system is successfully booted. Multiple partitions might be marked with the .Cm bootonce and .Cm bootme attribute pairs. .It Cm bootfailed This attribute should not be manually managed. It is managed by the .Nm gptboot stage 1 boot loader and the .Pa /etc/rc.d/gptboot start-up script. This attribute is used to mark partitions that had the .Cm bootonce attribute set, but we failed to boot from them. Once we successfully boot, the .Pa /etc/rc.d/gptboot script will log all the partitions we failed to boot from and will remove the .Cm bootfailed attributes. .El .Pp The scheme-specific attributes for MBR: .Bl -tag -width ".Cm active" .It Cm active .El .Pp The scheme-specific attributes for PC98: .Bl -tag -width ".Cm bootable" .It Cm active .It Cm bootable .El .Sh OPERATIONAL FLAGS Actions other than the .Cm commit and .Cm undo actions take an optional .Fl f Ar flags option. This option is used to specify action-specific operational flags. By default, the .Nm utility defines the .Ql C flag so that the action is immediately committed. The user can specify .Dq Fl f Cm x to have the action result in a pending change that can later, with other pending changes, be committed as a single compound change with the .Cm commit action or reverted with the .Cm undo action. .Sh RECOVERING -The GEOM class PART supports recovering of partition tables only for GPT. +The GEOM PART class supports recovering of partition tables only for GPT. The GUID partition table has a primary and secondary (backup) copy of -metadata for redundance. -They are stored in the begining and in the end of device respectively. -Therefore it is acceptable to have some corruptions in the metadata that -are not fatal to work with GPT. -When kernel detects corrupt metadata it marks this table as corrupt and -reports about corruption. -Any changes in corrupt table are prohibited except +metadata for redundance, these are stored at the begining and the end +of the device respectively. +As a result of having two copies, it is acceptable to have some corruption +within the metadata that is not fatal to the working of GPT. +When the kernel detects corrupt metadata it marks this table as corrupt and +reports the corruption. +Any operations on corrupt tables are prohibited except for .Cm destroy and .Cm recover . .Pp -In case when only first sector is corrupt kernel can not detect GPT even -if partition table is not corrupt. -You can write protective MBR with +If the first sector of a provider is corrupt, the kernel can not detect GPT +even if partition table itself is not corrupt. +You can rewrite the protective MBR using the .Xr dd 1 -command to restore ability of GPT detection. -The copy of protective MBR is usually located in the +command, to restore the ability to detect the GPT. +The copy of the protective MBR is usually located in the .Pa /boot/pmbr file. .Pp -In case when some of metadata is corrupt you will get to know about this -from kernel's messages like these: +If one GPT header appears to be corrupt but the other copy remains intact, +the kernel will log the following: .Bd -literal -offset indent GEOM: provider: the primary GPT table is corrupt or invalid. GEOM: provider: using the secondary instead -- recovery strongly advised. .Ed .Pp or .Bd -literal -offset indent GEOM: provider: the secondary GPT table is corrupt or invalid. GEOM: provider: using the primary only -- recovery suggested. .Ed .Pp Also .Nm -commands like +commands such as .Cm show , status and .Cm list -will report about corrupt table. +will report about corrupt tables. .Pp -In case when the size of device has changed (e.g.\& volume expansion) the -secondary GPT header will become located not in the last sector. +If the size of the device has changed (e.g.\& volume expansion) the +secondary GPT header will no longer be located in the last sector. This is not a metadata corruption, but it is dangerous because any -corruption of the primary GPT will lead to lost of partition table. -Kernel reports about this problem with message: +corruption of the primary GPT will lead to loss of partition table. +This problem is reported by the kernel with the message: .Bd -literal -offset indent GEOM: provider: the secondary GPT header is not in the last LBA. .Ed .Pp -A corrupt table can be recovered with +This situation can be recovered with the .Cm recover command. -This command does reconstruction of corrupt metadata using -known valid metadata. -Also it can relocate secondary GPT to the end of device. +This command reconstructs the corrupt metadata using known valid +metadata and relocates the secondary GPT to the end of the device. .Pp .Em NOTE : -The GEOM class PART can detect the same partition table on different GEOM -providers and some of them will be marked as corrupt. -Be careful when choosing a provider for recovering. +The GEOM PART class can detect the same partition table visible through +different GEOM providers, and some of them will be marked as corrupt. +Be careful when choosing a provider for recovery. If you choose incorrectly you can destroy the metadata of another GEOM class, e.g.\& GEOM MIRROR or GEOM LABEL. .Sh SYSCTL VARIABLES The following .Xr sysctl 8 variables can be used to control the behavior of the .Nm PART GEOM class. The default value is shown next to each variable. .Bl -tag -width indent .It Va kern.geom.part.check_integrity : No 1 This variable controls the behaviour of metadata integrity checks. -When integrity checks are enabled +When integrity checks are enabled, the .Nm PART -GEOM class verifies all generic partition parameters that it gets from the +GEOM class verifies all generic partition parameters obtained from the disk metadata. -If some inconsistency is detected, partition table will be +If some inconsistency is detected, the partition table will be rejected with a diagnostic message: .Sy "GEOM_PART: Integrity check failed (provider, scheme)" . .El .Sh EXIT STATUS Exit status is 0 on success, and 1 if the command fails. .Sh EXAMPLES Create GPT scheme on .Pa ad0 : .Bd -literal -offset indent /sbin/gpart create -s GPT ad0 .Ed .Pp Embed GPT bootstrap code into protective MBR: .Bd -literal -offset indent /sbin/gpart bootcode -b /boot/pmbr ad0 .Ed .Pp Create a dedicated .Cm freebsd-boot partition that can boot .Fx from a .Cm freebsd-ufs partition, and install bootstrap code into it. This partition must be larger than .Pa /boot/gptboot , or the GPT boot you are planning to write, but smaller than 545 KB. A size of 15 blocks (7680 bytes) would be sufficient for booting from UFS but let's use 128 blocks (64 KB) here in this example, in order to reserve some space for potential future need (e.g.\& from a ZFS partition). .Bd -literal -offset indent /sbin/gpart add -b 34 -s 128 -t freebsd-boot ad0 /sbin/gpart bootcode -p /boot/gptboot -i 1 ad0 .Ed .Pp Create a 512MB-sized .Cm freebsd-ufs partition that would contain UFS where the system boots from: .Bd -literal -offset indent /sbin/gpart add -b 162 -s 1048576 -t freebsd-ufs ad0 .Ed .Pp Create MBR scheme on .Pa ada0 , then create 30GB-sized .Fx slice, mark it active and install .Nm boot0 boot manager: .Bd -literal -offset indent /sbin/gpart create -s MBR ada0 /sbin/gpart add -t freebsd -s 30G ada0 /sbin/gpart set -a active -i 1 ada0 /sbin/gpart bootcode -b /boot/boot0 ada0 .Ed .Pp Now create .Bx scheme .Pf ( Bx label) with ability to have up to 20 partitions: .Bd -literal -offset indent /sbin/gpart create -s BSD -n 20 ada0s1 .Ed .Pp Create 1GB-sized UFS partition and 4GB-sized swap partition: .Bd -literal -offset indent /sbin/gpart add -t freebsd-ufs -s 1G ada0s1 /sbin/gpart add -t freebsd-swap -s 4G ada0s1 .Ed .Pp Install bootstrap code for the .Bx label: .Bd -literal -offset indent /sbin/gpart bootcode -b /boot/boot ada0s1 .Ed .Pp Create VTOC8 scheme on .Pa da0 : .Bd -literal -offset indent /sbin/gpart create -s VTOC8 da0 .Ed .Pp Create a 512MB-sized .Cm freebsd-ufs partition that would contain UFS where the system boots from: .Bd -literal -offset indent /sbin/gpart add -s 512M -t freebsd-ufs da0 .Ed .Pp Create a 15GB-sized .Cm freebsd-ufs partition that would contain UFS and aligned on 4KB boundaries: .Bd -literal -offset indent /sbin/gpart add -s 15G -t freebsd-ufs -a 4k da0 .Ed .Pp After having created all required partitions, embed bootstrap code into them: .Bd -literal -offset indent /sbin/gpart bootcode -p /boot/boot1 da0 .Ed .Pp Create backup of partition table from .Pa da0 : .Bd -literal -offset indent /sbin/gpart backup da0 > da0.backup .Ed .Pp Restore partition table from backup to .Pa da0 : .Bd -literal -offset indent /sbin/gpart restore -l da0 < /mnt/da0.backup .Ed .Pp Clone partition table from .Pa ada0 to .Pa ada1 and .Pa ada2 : .Bd -literal -offset indent /sbin/gpart backup ada0 | /sbin/gpart restore -F ada1 ada2 .Ed .Sh SEE ALSO .Xr dd 1 , .Xr geom 4 , .Xr geom 8 .Sh HISTORY The .Nm utility appeared in .Fx 7.0 . .Sh AUTHORS .An Marcel Moolenaar Aq marcel@FreeBSD.org Index: projects/largeSMP/sbin/geom/class/sched/Makefile =================================================================== --- projects/largeSMP/sbin/geom/class/sched/Makefile (revision 222782) +++ projects/largeSMP/sbin/geom/class/sched/Makefile (revision 222783) @@ -1,10 +1,8 @@ # GEOM_LIBRARY_PATH # $FreeBSD$ .PATH: ${.CURDIR}/../../misc GEOM_CLASS= sched -WARNS?= 6 - .include Index: projects/largeSMP/sbin/ifconfig/af_inet6.c =================================================================== --- projects/largeSMP/sbin/ifconfig/af_inet6.c (revision 222782) +++ projects/largeSMP/sbin/ifconfig/af_inet6.c (revision 222783) @@ -1,552 +1,556 @@ /* * Copyright (c) 1983, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef lint static const char rcsid[] = "$FreeBSD$"; #endif /* not lint */ #include #include #include #include #include #include #include #include #include #include #include #include #include /* 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 void in6_fillscopeid(struct sockaddr_in6 *sin6); 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) { time_t newval, t; char *ep; t = time(NULL); 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 = t + newval; in6_addreq.ifra_lifetime.ia6t_vltime = newval; } else if (strcmp(cmd, "pltime") == 0) { in6_addreq.ifra_lifetime.ia6t_preferred = t + 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_fillscopeid(struct sockaddr_in6 *sin6) { #if defined(__KAME__) && defined(KAME_SCOPEID) if (IN6_IS_ADDR_LINKLOCAL(&sin6->sin6_addr)) { sin6->sin6_scope_id = ntohs(*(u_int16_t *)&sin6->sin6_addr.s6_addr[2]); sin6->sin6_addr.s6_addr[2] = sin6->sin6_addr.s6_addr[3] = 0; } #endif } 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; time_t t = time(NULL); int error; u_int32_t scopeid; 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); /* XXX: embedded link local addr check */ if (IN6_IS_ADDR_LINKLOCAL(&sin->sin6_addr) && *(u_short *)&sin->sin6_addr.s6_addr[2] != 0) { u_short index; index = *(u_short *)&sin->sin6_addr.s6_addr[2]; *(u_short *)&sin->sin6_addr.s6_addr[2] = 0; if (sin->sin6_scope_id == 0) sin->sin6_scope_id = ntohs(index); } scopeid = sin->sin6_scope_id; 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; /* XXX: embedded link local addr check */ if (IN6_IS_ADDR_LINKLOCAL(&sin->sin6_addr) && *(u_short *)&sin->sin6_addr.s6_addr[2] != 0) { u_short index; index = *(u_short *)&sin->sin6_addr.s6_addr[2]; *(u_short *)&sin->sin6_addr.s6_addr[2] = 0; if (sin->sin6_scope_id == 0) sin->sin6_scope_id = ntohs(index); } 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 (scopeid) printf("scopeid 0x%x ", scopeid); if (ip6lifetime && (lifetime.ia6t_preferred || lifetime.ia6t_expire)) { printf("pltime "); if (lifetime.ia6t_preferred) { printf("%s ", lifetime.ia6t_preferred < t ? "0" : sec2str(lifetime.ia6t_preferred - t)); } else printf("infty "); printf("vltime "); if (lifetime.ia6t_expire) { printf("%s ", lifetime.ia6t_expire < t ? "0" : sec2str(lifetime.ia6t_expire - t)); } else printf("infty "); } 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; in6_fillscopeid(&in6_ifr.ifr_addr); 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; in6_fillscopeid(&in6_ifr.ifr_addr); 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("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("prefer_source",ND6_IFF_PREFER_SOURCE, setnd6flags), DEF_CMD("-prefer_source",-ND6_IFF_PREFER_SOURCE,setnd6flags), DEF_CMD("auto_linklocal",ND6_IFF_AUTO_LINKLOCAL,setnd6flags), DEF_CMD("-auto_linklocal",-ND6_IFF_AUTO_LINKLOCAL,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; if (!feature_present("inet6")) return; 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/largeSMP/sbin/ifconfig/af_nd6.c =================================================================== --- projects/largeSMP/sbin/ifconfig/af_nd6.c (revision 222782) +++ projects/largeSMP/sbin/ifconfig/af_nd6.c (revision 222783) @@ -1,233 +1,167 @@ /* * 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" \ - "\020DEFAULTIF" + "\007NO_RADR\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); } -static void +void nd6_status(int s) { struct in6_ndireq nd; - struct rt_msghdr *rtm; - size_t needed; - char *buf, *next; - int mib[6], ntry; int s6; int error; - int isinet6, isdefif; + int isdefif; - /* Check if the interface has at least one IPv6 address. */ - mib[0] = CTL_NET; - mib[1] = PF_ROUTE; - mib[2] = 0; - mib[3] = AF_INET6; - mib[4] = NET_RT_IFLIST; - mib[5] = if_nametoindex(ifr.ifr_name); - - /* Try to prevent a race between two sysctls. */ - ntry = 0; - do { - error = sysctl(mib, 6, NULL, &needed, NULL, 0); - if (error) { - warn("sysctl(NET_RT_IFLIST)/estimate"); - return; - } - buf = malloc(needed); - if (buf == NULL) { - warn("malloc for sysctl(NET_RT_IFLIST) failed"); - return; - } - if ((error = sysctl(mib, 6, buf, &needed, NULL, 0)) < 0) { - if (errno != ENOMEM || ++ntry >= MAX_SYSCTL_TRY) { - warn("sysctl(NET_RT_IFLIST)/get"); - free(buf); - return; - } - free(buf); - buf = NULL; - } - } while (buf == NULL); - - isinet6 = 0; - for (next = buf; next < buf + needed; next += rtm->rtm_msglen) { - rtm = (struct rt_msghdr *)next; - - if (rtm->rtm_version != RTM_VERSION) - continue; - if (rtm->rtm_type == RTM_NEWADDR) { - isinet6 = 1; - break; - } - } - free(buf); - if (!isinet6) - return; - memset(&nd, 0, sizeof(nd)); strncpy(nd.ifname, ifr.ifr_name, sizeof(nd.ifname)); if ((s6 = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) { warn("socket(AF_INET6, SOCK_DGRAM)"); return; } error = ioctl(s6, SIOCGIFINFO_IN6, &nd); if (error) { 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'); -} - -static struct afswtch af_nd6 = { - .af_name = "nd6", - .af_af = AF_LOCAL, - .af_other_status= nd6_status, -}; - -static __constructor void -nd6_ctor(void) -{ - - if (!feature_present("inet6")) - return; - - af_register(&af_nd6); } Index: projects/largeSMP/sbin/ipfw/main.c =================================================================== --- projects/largeSMP/sbin/ipfw/main.c (revision 222782) +++ projects/largeSMP/sbin/ipfw/main.c (revision 222783) @@ -1,615 +1,616 @@ /* * Copyright (c) 2002-2003,2010 Luigi Rizzo * Copyright (c) 1996 Alex Nash, Paul Traina, Poul-Henning Kamp * Copyright (c) 1994 Ugen J.S.Antsilevich * * Idea and grammar partially left from: * Copyright (c) 1993 Daniel Boulet * * Redistribution and use in source forms, with and without modification, * are permitted provided that this entire comment appears intact. * * Redistribution in binary form may occur without any restrictions. * Obviously, it would be nice if you gave credit where credit is due * but requiring it would be too onerous. * * This software is provided ``AS IS'' without any warranties of any kind. * * Command line interface for IP firewall facility * * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include #include "ipfw2.h" static void help(void) { fprintf(stderr, "ipfw syntax summary (but please do read the ipfw(8) manpage):\n\n" "\tipfw [-abcdefhnNqStTv] \n\n" "where is one of the following:\n\n" "add [num] [set N] [prob x] RULE-BODY\n" "{pipe|queue} N config PIPE-BODY\n" "[pipe|queue] {zero|delete|show} [N{,N}]\n" "nat N config {ip IPADDR|if IFNAME|log|deny_in|same_ports|unreg_only|reset|\n" " reverse|proxy_only|redirect_addr linkspec|\n" " redirect_port linkspec|redirect_proto linkspec}\n" "set [disable N... enable N...] | move [rule] X to Y | swap X Y | show\n" "set N {show|list|zero|resetlog|delete} [N{,N}] | flush\n" "table N {add ip[/bits] [value] | delete ip[/bits] | flush | list}\n" "table all {flush | list}\n" "\n" "RULE-BODY: check-state [PARAMS] | ACTION [PARAMS] ADDR [OPTION_LIST]\n" "ACTION: check-state | allow | count | deny | unreach{,6} CODE |\n" " skipto N | {divert|tee} PORT | forward ADDR |\n" " pipe N | queue N | nat N | setfib FIB | reass\n" "PARAMS: [log [logamount LOGLIMIT]] [altq QUEUE_NAME]\n" "ADDR: [ MAC dst src ether_type ] \n" " [ ip from IPADDR [ PORT ] to IPADDR [ PORTLIST ] ]\n" " [ ipv6|ip6 from IP6ADDR [ PORT ] to IP6ADDR [ PORTLIST ] ]\n" "IPADDR: [not] { any | me | ip/bits{x,y,z} | table(t[,v]) | IPLIST }\n" "IP6ADDR: [not] { any | me | me6 | ip6/bits | IP6LIST }\n" "IP6LIST: { ip6 | ip6/bits }[,IP6LIST]\n" "IPLIST: { ip | ip/bits | ip:mask }[,IPLIST]\n" "OPTION_LIST: OPTION [OPTION_LIST]\n" "OPTION: bridged | diverted | diverted-loopback | diverted-output |\n" " {dst-ip|src-ip} IPADDR | {dst-ip6|src-ip6|dst-ipv6|src-ipv6} IP6ADDR |\n" " {dst-port|src-port} LIST |\n" " estab | frag | {gid|uid} N | icmptypes LIST | in | out | ipid LIST |\n" " iplen LIST | ipoptions SPEC | ipprecedence | ipsec | iptos SPEC |\n" " ipttl LIST | ipversion VER | keep-state | layer2 | limit ... |\n" " icmp6types LIST | ext6hdr LIST | flow-id N[,N] | fib FIB |\n" " mac ... | mac-type LIST | proto LIST | {recv|xmit|via} {IF|IPADDR} |\n" " setup | {tcpack|tcpseq|tcpwin} NN | tcpflags SPEC | tcpoptions SPEC |\n" " tcpdatalen LIST | verrevpath | versrcreach | antispoof\n" ); exit(0); } /* * Called with the arguments, including program name because getopt * wants it to be present. * Returns 0 if successful, 1 if empty command, errx() in case of errors. * First thing we do is process parameters creating an argv[] array * which includes the program name and a NULL entry at the end. * If we are called with a single string, we split it on whitespace. * Also, arguments with a trailing ',' are joined to the next one. * The pointers (av[]) and data are in a a single chunk of memory. * av[0] points to the original program name, all other entries * point into the allocated chunk. */ static int ipfw_main(int oldac, char **oldav) { int ch, ac; const char *errstr; char **av, **save_av; int do_acct = 0; /* Show packet/byte count */ int try_next = 0; /* set if pipe cmd not found */ int av_size; /* compute the av size */ char *av_p; /* used to build the av list */ #define WHITESP " \t\f\v\n\r" if (oldac < 2) return 1; /* need at least one argument */ if (oldac == 2) { /* * If we are called with one argument, try to split it into * words for subsequent parsing. Spaces after a ',' are * removed by copying the string in-place. */ char *arg = oldav[1]; /* The string is the first arg. */ int l = strlen(arg); int copy = 0; /* 1 if we need to copy, 0 otherwise */ int i, j; for (i = j = 0; i < l; i++) { if (arg[i] == '#') /* comment marker */ break; if (copy) { arg[j++] = arg[i]; copy = !index("," WHITESP, arg[i]); } else { copy = !index(WHITESP, arg[i]); if (copy) arg[j++] = arg[i]; } } if (!copy && j > 0) /* last char was a 'blank', remove it */ j--; l = j; /* the new argument length */ arg[j++] = '\0'; if (l == 0) /* empty string! */ return 1; /* * First, count number of arguments. Because of the previous * processing, this is just the number of blanks plus 1. */ for (i = 0, ac = 1; i < l; i++) if (index(WHITESP, arg[i]) != NULL) ac++; /* * Allocate the argument list structure as a single block * of memory, containing pointers and the argument * strings. We include one entry for the program name * because getopt expects it, and a NULL at the end * to simplify further parsing. */ ac++; /* add 1 for the program name */ av_size = (ac+1) * sizeof(char *) + l + 1; av = safe_calloc(av_size, 1); /* * Init the argument pointer to the end of the array * and copy arguments from arg[] to av[]. For each one, * j is the initial character, i is the one past the end. */ av_p = (char *)&av[ac+1]; for (ac = 1, i = j = 0; i < l; i++) { if (index(WHITESP, arg[i]) != NULL || i == l-1) { if (i == l-1) i++; bcopy(arg+j, av_p, i-j); av[ac] = av_p; av_p += i-j; /* the lenght of the string */ *av_p++ = '\0'; ac++; j = i + 1; } } } else { /* * If an argument ends with ',' join with the next one. */ int first, i, l=0; /* * Allocate the argument list structure as a single block * of memory, containing both pointers and the argument * strings. We include some space for the program name * because getopt expects it. * We add an extra pointer to the end of the array, * to make simpler further parsing. */ for (i=0; i= 2 && !strcmp(av[1], "sysctl")) { char *s; int i; if (ac != 3) { printf( "sysctl emulation usage:\n" " ipfw sysctl name[=value]\n" " ipfw sysctl -a\n"); return 0; } s = index(av[2], '='); if (s == NULL) { s = !strcmp(av[2], "-a") ? NULL : av[2]; sysctlbyname(s, NULL, NULL, NULL, 0); } else { /* ipfw sysctl x.y.z=value */ /* assume an INT value, will extend later */ if (s[1] == '\0') { printf("ipfw sysctl: missing value\n\n"); return 0; } *s = '\0'; i = strtol(s+1, NULL, 0); sysctlbyname(av[2], NULL, NULL, &i, sizeof(int)); } return 0; } #endif /* Save arguments for final freeing of memory. */ save_av = av; optind = optreset = 1; /* restart getopt() */ while ((ch = getopt(ac, av, "abcdefhinNqs:STtv")) != -1) switch (ch) { case 'a': do_acct = 1; break; case 'b': co.comment_only = 1; co.do_compact = 1; break; case 'c': co.do_compact = 1; break; case 'd': co.do_dynamic = 1; break; case 'e': co.do_expired = 1; break; case 'f': co.do_force = 1; break; case 'h': /* help */ free(save_av); help(); break; /* NOTREACHED */ case 'i': co.do_value_as_ip = 1; break; case 'n': co.test_only = 1; break; case 'N': co.do_resolv = 1; break; case 'q': co.do_quiet = 1; break; case 's': /* sort */ co.do_sort = atoi(optarg); break; case 'S': co.show_sets = 1; break; case 't': co.do_time = 1; break; case 'T': co.do_time = 2; /* numeric timestamp */ break; case 'v': /* verbose */ co.verbose = 1; break; default: free(save_av); return 1; } ac -= optind; av += optind; NEED1("bad arguments, for usage summary ``ipfw''"); /* * An undocumented behaviour of ipfw1 was to allow rule numbers first, * e.g. "100 add allow ..." instead of "add 100 allow ...". * In case, swap first and second argument to get the normal form. */ if (ac > 1 && isdigit(*av[0])) { char *p = av[0]; av[0] = av[1]; av[1] = p; } /* * Optional: pipe, queue or nat. */ co.do_nat = 0; co.do_pipe = 0; + co.use_set = 0; if (!strncmp(*av, "nat", strlen(*av))) co.do_nat = 1; else if (!strncmp(*av, "pipe", strlen(*av))) co.do_pipe = 1; else if (_substrcmp(*av, "queue") == 0) co.do_pipe = 2; else if (_substrcmp(*av, "flowset") == 0) co.do_pipe = 2; else if (_substrcmp(*av, "sched") == 0) co.do_pipe = 3; else if (!strncmp(*av, "set", strlen(*av))) { if (ac > 1 && isdigit(av[1][0])) { co.use_set = strtonum(av[1], 0, resvd_set_number, &errstr); if (errstr) errx(EX_DATAERR, "invalid set number %s\n", av[1]); ac -= 2; av += 2; co.use_set++; } } if (co.do_pipe || co.do_nat) { ac--; av++; } NEED1("missing command"); /* * For pipes, queues and nats we normally say 'nat|pipe NN config' * but the code is easier to parse as 'nat|pipe config NN' * so we swap the two arguments. */ if ((co.do_pipe || co.do_nat) && ac > 1 && isdigit(*av[0])) { char *p = av[0]; av[0] = av[1]; av[1] = p; } if (co.use_set == 0) { if (_substrcmp(*av, "add") == 0) ipfw_add(av); else if (co.do_nat && _substrcmp(*av, "show") == 0) ipfw_show_nat(ac, av); else if (co.do_pipe && _substrcmp(*av, "config") == 0) ipfw_config_pipe(ac, av); else if (co.do_nat && _substrcmp(*av, "config") == 0) ipfw_config_nat(ac, av); else if (_substrcmp(*av, "set") == 0) ipfw_sets_handler(av); else if (_substrcmp(*av, "table") == 0) ipfw_table_handler(ac, av); else if (_substrcmp(*av, "enable") == 0) ipfw_sysctl_handler(av, 1); else if (_substrcmp(*av, "disable") == 0) ipfw_sysctl_handler(av, 0); else try_next = 1; } if (co.use_set || try_next) { if (_substrcmp(*av, "delete") == 0) ipfw_delete(av); else if (_substrcmp(*av, "flush") == 0) ipfw_flush(co.do_force); else if (_substrcmp(*av, "zero") == 0) ipfw_zero(ac, av, 0 /* IP_FW_ZERO */); else if (_substrcmp(*av, "resetlog") == 0) ipfw_zero(ac, av, 1 /* IP_FW_RESETLOG */); else if (_substrcmp(*av, "print") == 0 || _substrcmp(*av, "list") == 0) ipfw_list(ac, av, do_acct); else if (_substrcmp(*av, "show") == 0) ipfw_list(ac, av, 1 /* show counters */); else errx(EX_USAGE, "bad command `%s'", *av); } /* Free memory allocated in the argument parsing. */ free(save_av); return 0; } static void ipfw_readfile(int ac, char *av[]) { #define MAX_ARGS 32 - char buf[BUFSIZ]; + char buf[4096]; char *progname = av[0]; /* original program name */ const char *cmd = NULL; /* preprocessor name, if any */ const char *filename = av[ac-1]; /* file to read */ int c, lineno=0; FILE *f = NULL; pid_t preproc = 0; while ((c = getopt(ac, av, "cfNnp:qS")) != -1) { switch(c) { case 'c': co.do_compact = 1; break; case 'f': co.do_force = 1; break; case 'N': co.do_resolv = 1; break; case 'n': co.test_only = 1; break; case 'p': /* * ipfw -p cmd [args] filename * * We are done with getopt(). All arguments * except the filename go to the preprocessor, * so we need to do the following: * - check that a filename is actually present; * - advance av by optind-1 to skip arguments * already processed; * - decrease ac by optind, to remove the args * already processed and the final filename; * - set the last entry in av[] to NULL so * popen() can detect the end of the array; * - set optind=ac to let getopt() terminate. */ if (optind == ac) errx(EX_USAGE, "no filename argument"); cmd = optarg; av[ac-1] = NULL; av += optind - 1; ac -= optind; optind = ac; break; case 'q': co.do_quiet = 1; break; case 'S': co.show_sets = 1; break; default: errx(EX_USAGE, "bad arguments, for usage" " summary ``ipfw''"); } } if (cmd == NULL && ac != optind + 1) errx(EX_USAGE, "extraneous filename arguments %s", av[ac-1]); if ((f = fopen(filename, "r")) == NULL) err(EX_UNAVAILABLE, "fopen: %s", filename); if (cmd != NULL) { /* pipe through preprocessor */ int pipedes[2]; if (pipe(pipedes) == -1) err(EX_OSERR, "cannot create pipe"); preproc = fork(); if (preproc == -1) err(EX_OSERR, "cannot fork"); if (preproc == 0) { /* * Child, will run the preprocessor with the * file on stdin and the pipe on stdout. */ if (dup2(fileno(f), 0) == -1 || dup2(pipedes[1], 1) == -1) err(EX_OSERR, "dup2()"); fclose(f); close(pipedes[1]); close(pipedes[0]); execvp(cmd, av); err(EX_OSERR, "execvp(%s) failed", cmd); } else { /* parent, will reopen f as the pipe */ fclose(f); close(pipedes[1]); if ((f = fdopen(pipedes[0], "r")) == NULL) { int savederrno = errno; (void)kill(preproc, SIGTERM); errno = savederrno; err(EX_OSERR, "fdopen()"); } } } - while (fgets(buf, BUFSIZ, f)) { /* read commands */ + while (fgets(buf, sizeof(buf), f)) { /* read commands */ char linename[20]; char *args[2]; lineno++; snprintf(linename, sizeof(linename), "Line %d", lineno); setprogname(linename); /* XXX */ args[0] = progname; args[1] = buf; ipfw_main(2, args); } fclose(f); if (cmd != NULL) { int status; if (waitpid(preproc, &status, 0) == -1) errx(EX_OSERR, "waitpid()"); if (WIFEXITED(status) && WEXITSTATUS(status) != EX_OK) errx(EX_UNAVAILABLE, "preprocessor exited with status %d", WEXITSTATUS(status)); else if (WIFSIGNALED(status)) errx(EX_UNAVAILABLE, "preprocessor exited with signal %d", WTERMSIG(status)); } } int main(int ac, char *av[]) { #if defined(_WIN32) && defined(TCC) { WSADATA wsaData; int ret=0; unsigned short wVersionRequested = MAKEWORD(2, 2); ret = WSAStartup(wVersionRequested, &wsaData); if (ret != 0) { /* Tell the user that we could not find a usable */ /* Winsock DLL. */ printf("WSAStartup failed with error: %d\n", ret); return 1; } } #endif /* * If the last argument is an absolute pathname, interpret it * as a file to be preprocessed. */ if (ac > 1 && av[ac - 1][0] == '/' && access(av[ac - 1], R_OK) == 0) ipfw_readfile(ac, av); else { if (ipfw_main(ac, av)) { errx(EX_USAGE, "usage: ipfw [options]\n" "do \"ipfw -h\" or \"man ipfw\" for details"); } } return EX_OK; } Index: projects/largeSMP/sbin/ipfw =================================================================== --- projects/largeSMP/sbin/ipfw (revision 222782) +++ projects/largeSMP/sbin/ipfw (revision 222783) Property changes on: projects/largeSMP/sbin/ipfw ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sbin/ipfw:r222698-222782 Index: projects/largeSMP/sbin/mount/mount.8 =================================================================== --- projects/largeSMP/sbin/mount/mount.8 (revision 222782) +++ projects/largeSMP/sbin/mount/mount.8 (revision 222783) @@ -1,571 +1,580 @@ .\" Copyright (c) 1980, 1989, 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. .\" .\" @(#)mount.8 8.8 (Berkeley) 6/16/94 .\" $FreeBSD$ .\" -.Dd April 28, 2011 +.Dd June 6, 2011 .Dt MOUNT 8 .Os .Sh NAME .Nm mount .Nd mount file systems .Sh SYNOPSIS .Nm .Op Fl adflpruvw .Op Fl F Ar fstab .Op Fl o Ar options .Op Fl t Cm ufs | Ar external_type .Nm .Op Fl dfpruvw .Ar special | node .Nm .Op Fl dfpruvw .Op Fl o Ar options .Op Fl t Cm ufs | Ar external_type .Ar special node .Sh DESCRIPTION The .Nm utility calls the .Xr nmount 2 system call to prepare and graft a .Ar special device or the remote node (rhost:path) on to the file system tree at the point .Ar node . If either .Ar special or .Ar node are not provided, the appropriate information is taken from the .Xr fstab 5 file. .Pp The system maintains a list of currently mounted file systems. If no arguments are given to .Nm , this list is printed. .Pp The options are as follows: .Bl -tag -width indent .It Fl a All the file systems described in .Xr fstab 5 are mounted. Exceptions are those marked as .Dq Li noauto , those marked as .Dq Li late (unless the .Fl l option was specified), those excluded by the .Fl t flag (see below), or if they are already mounted (except the root file system which is always remounted to preserve traditional single user mode behavior). .It Fl d Causes everything to be done except for the actual system call. This option is useful in conjunction with the .Fl v flag to determine what the .Nm command is trying to do. .It Fl F Ar fstab Specify the .Pa fstab file to use. .It Fl f Forces the revocation of write access when trying to downgrade a file system mount status from read-write to read-only. Also forces the R/W mount of an unclean file system (dangerous; use with caution). .It Fl l When used in conjunction with the .Fl a option, also mount those file systems which are marked as .Dq Li late . .It Fl o Options are specified with a .Fl o flag followed by a comma separated string of options. In case of conflicting options being specified, the rightmost option takes effect. The following options are available: .Bl -tag -width indent .It Cm acls Enable POSIX.1e Access Control Lists, or ACLs, which can be customized via the .Xr setfacl 1 and .Xr getfacl 1 commands. This flag is mutually exclusive with .Cm nfsv4acls flag. .It Cm async All .Tn I/O to the file system should be done asynchronously. This is a .Em dangerous flag to set, since it does not guarantee that the file system structure on the disk will remain consistent. For this reason, the .Cm async flag should be used sparingly, and only when some data recovery mechanism is present. .It Cm current When used with the .Fl u flag, this is the same as specifying the options currently in effect for the mounted file system. .It Cm force The same as .Fl f ; forces the revocation of write access when trying to downgrade a file system mount status from read-write to read-only. Also forces the R/W mount of an unclean file system (dangerous; use with caution). .It Cm fstab When used with the .Fl u flag, this is the same as specifying all the options listed in the .Xr fstab 5 file for the file system. .It Cm late This file system should be skipped when .Nm is run with the .Fl a flag but without the .Fl l flag. .It Cm mountprog Ns = Ns Aq Ar program Force .Nm to use the specified program to mount the file system, instead of calling .Xr nmount 2 directly. For example: .Bd -literal mount -t foofs -o mountprog=/mydir/fooprog /dev/acd0 /mnt .Ed .It Cm multilabel Enable multi-label Mandatory Access Control, or MAC, on the specified file system. If the file system supports multilabel operation, individual labels will be maintained for each object in the file system, rather than using a single label for all objects. An alternative to the .Fl l flag in .Xr tunefs 8 . See .Xr mac 4 for more information, which cause the multilabel mount flag to be set automatically at mount-time. .It Cm nfsv4acls Enable NFSv4 ACLs, which can be customized via the .Xr setfacl 1 and .Xr getfacl 1 commands. This flag is mutually exclusive with .Cm acls flag. .It Cm noasync Metadata I/O should be done synchronously, while data I/O should be done asynchronously. This is the default. .It Cm noatime Do not update the file access time when reading from a file. This option is useful on file systems where there are large numbers of files and performance is more critical than updating the file access time (which is rarely ever important). This option is currently only supported on local file systems. .It Cm noauto This file system should be skipped when .Nm is run with the .Fl a flag. .It Cm noclusterr Disable read clustering. .It Cm noclusterw Disable write clustering. .It Cm noexec Do not allow execution of any binaries on the mounted file system. This option is useful for a server that has file systems containing binaries for architectures other than its own. Note: This option was not designed as a security feature and no guarantee is made that it will prevent malicious code execution; for example, it is still possible to execute scripts which reside on a .Cm noexec mounted partition. .It Cm nosuid Do not allow set-user-identifier or set-group-identifier bits to take effect. Note: this option is worthless if a public available suid or sgid wrapper like .Xr suidperl 1 is installed on your system. It is set automatically when the user does not have super-user privileges. .It Cm nosymfollow Do not follow symlinks on the mounted file system. .It Cm ro The same as .Fl r ; mount the file system read-only (even the super-user may not write it). .It Cm snapshot This option allows a snapshot of the specified file system to be taken. The .Fl u flag is required with this option. Note that snapshot files must be created in the file system that is being snapshotted. You may create up to 20 snapshots per file system. Active snapshots are recorded in the superblock, so they persist across unmount and remount operations and across system reboots. When you are done with a snapshot, it can be removed with the .Xr rm 1 command. Snapshots may be removed in any order, however you may not get back all the space contained in the snapshot as another snapshot may claim some of the blocks that it is releasing. Note that the schg flag is set on snapshots to ensure that not even the root user can write to them. The unlink command makes an exception for snapshot files in that it allows them to be removed even though they have the schg flag set, so it is not necessary to clear the schg flag before removing a snapshot file. .Pp Once you have taken a snapshot, there are three interesting things that you can do with it: .Pp .Bl -enum -compact .It Run .Xr fsck 8 on the snapshot file. Assuming that the file system was clean when it was mounted, you should always get a clean (and unchanging) result from running fsck on the snapshot. This is essentially what the background fsck process does. .Pp .It Run .Xr dump 8 on the snapshot. You will get a dump that is consistent with the file system as of the timestamp of the snapshot. .Pp .It Mount the snapshot as a frozen image of the file system. To mount the snapshot .Pa /var/snapshot/snap1 : .Bd -literal mdconfig -a -t vnode -f /var/snapshot/snap1 -u 4 mount -r /dev/md4 /mnt .Ed .Pp You can now cruise around your frozen .Pa /var file system at .Pa /mnt . Everything will be in the same state that it was at the time the snapshot was taken. The one exception is that any earlier snapshots will appear as zero length files. When you are done with the mounted snapshot: .Bd -literal umount /mnt mdconfig -d -u 4 .Ed .El .It Cm suiddir A directory on the mounted file system will respond to the SUID bit being set, by setting the owner of any new files to be the same as the owner of the directory. New directories will inherit the bit from their parents. Execute bits are removed from the file, and it will not be given to root. .Pp This feature is designed for use on fileservers serving PC users via ftp, SAMBA, or netatalk. It provides security holes for shell users and as such should not be used on shell machines, especially on home directories. This option requires the SUIDDIR option in the kernel to work. Only UFS file systems support this option. See .Xr chmod 2 for more information. .It Cm sync All .Tn I/O to the file system should be done synchronously. .It Cm update The same as .Fl u ; indicate that the status of an already mounted file system should be changed. .It Cm union Causes the namespace at the mount point to appear as the union of the mounted file system root and the existing directory. Lookups will be done in the mounted file system first. If those operations fail due to a non-existent file the underlying directory is then accessed. All creates are done in the mounted file system. .El .Pp Any additional options specific to a file system type that is not one of the internally known types (see the .Fl t option) may be passed as a comma separated list; these options are distinguished by a leading .Dq \&- (dash). -Options that take a value are specified using the syntax -option=value. For example, the .Nm command: .Bd -literal -offset indent mount -t cd9660 -o -e /dev/cd0 /cdrom .Ed .Pp causes .Nm to execute the equivalent of: .Bd -literal -offset indent /sbin/mount_cd9660 -e /dev/cd0 /cdrom +.Ed +.Pp +Options that take a value are specified using the -option=value syntax: +.Bd -literal -offset indent +mount -t msdosfs -o -u=fred,-g=wheel /dev/da0s1 /mnt +.Ed +.Pp +is equivalent to +.Bd -literal -offset indent +/sbin/mount_msdosfs -u fred -g wheel /dev/da0s1 /mnt .Ed .Pp Additional options specific to file system types which are not internally known (see the description of the .Fl t option below) may be described in the manual pages for the associated .Pa /sbin/mount_ Ns Sy XXX utilities. .It Fl p Print mount information in .Xr fstab 5 format. Implies also the .Fl v option. .It Fl r The file system is to be mounted read-only. Mount the file system read-only (even the super-user may not write it). The same as the .Cm ro argument to the .Fl o option. .It Fl t Cm ufs | Ar external_type The argument following the .Fl t is used to indicate the file system type. The type .Cm ufs is the default. The .Fl t option can be used to indicate that the actions should only be taken on file systems of the specified type. More than one type may be specified in a comma separated list. The list of file system types can be prefixed with .Dq Li no to specify the file system types for which action should .Em not be taken. For example, the .Nm command: .Bd -literal -offset indent mount -a -t nonfs,nullfs .Ed .Pp mounts all file systems except those of type .Tn NFS and .Tn NULLFS . .Pp The default behavior of .Nm is to pass the .Fl t option directly to the .Xr nmount 2 system call in the .Li fstype option. .Pp However, for the following file system types: .Cm cd9660 , .Cm mfs , .Cm msdosfs , .Cm nfs , .Cm ntfs , .Cm nwfs , .Cm nullfs , .Cm oldnfs , .Cm portalfs , .Cm smbfs , .Cm udf , and .Cm unionfs , .Nm will not call .Xr nmount 2 directly and will instead attempt to execute a program in .Pa /sbin/mount_ Ns Sy XXX where .Sy XXX is replaced by the file system type name. For example, nfs file systems are mounted by the program .Pa /sbin/mount_nfs . .Pp Most file systems will be dynamically loaded by the kernel if not already present, and if the kernel module is available. .It Fl u The .Fl u flag indicates that the status of an already mounted file system should be changed. Any of the options discussed above (the .Fl o option) may be changed; also a file system can be changed from read-only to read-write or vice versa. An attempt to change from read-write to read-only will fail if any files on the file system are currently open for writing unless the .Fl f flag is also specified. The set of options is determined by applying the options specified in the argument to .Fl o and finally applying the .Fl r or .Fl w option. .It Fl v Verbose mode. If the .Fl v is used alone, show all file systems, including those that were mounted with the .Dv MNT_IGNORE flag and show additional information about each file system (including fsid when run by root). .It Fl w The file system object is to be read and write. .El .Sh ENVIRONMENT .Bl -tag -width ".Ev PATH_FSTAB" .It Ev PATH_FSTAB If the environment variable .Ev PATH_FSTAB is set, all operations are performed against the specified file. .Ev PATH_FSTAB will not be honored if the process environment or memory address space is considered .Dq tainted . (See .Xr issetugid 2 for more information.) .El .Sh FILES .Bl -tag -width /etc/fstab -compact .It Pa /etc/fstab file system table .El .Sh DIAGNOSTICS Various, most of them are self-explanatory. .Pp .Dl XXXXX file system is not available .Pp The kernel does not support the respective file system type. Note that support for a particular file system might be provided either on a static (kernel compile-time), or dynamic basis (loaded as a kernel module by .Xr kldload 8 ) . .Sh SEE ALSO .Xr getfacl 1 , .Xr setfacl 1 , .Xr nmount 2 , .Xr acl 3 , .Xr mac 4 , .Xr devfs 5 , .Xr ext2fs 5 , .Xr fstab 5 , .Xr procfs 5 , .Xr kldload 8 , .Xr mount_cd9660 8 , .Xr mount_msdosfs 8 , .Xr mount_nfs 8 , .Xr mount_ntfs 8 , .Xr mount_nullfs 8 , .Xr mount_nwfs 8 , .Xr mount_portalfs 8 , .Xr mount_smbfs 8 , .Xr mount_udf 8 , .Xr mount_unionfs 8 , .Xr umount 8 , .Xr zfs 8 , .Xr zpool 8 .Sh HISTORY A .Nm utility appeared in .At v1 . .Sh CAVEATS After a successful .Nm , the permissions on the original mount point determine if .Pa ..\& is accessible from the mounted file system. The minimum permissions for the mount point for traversal across the mount point in both directions to be possible for all users is 0111 (execute for all). .Pp Use of the .Nm is preferred over the use of the file system specific .Pa mount_ Ns Sy XXX commands. In particular, .Xr mountd 8 gets a .Dv SIGHUP signal (that causes an update of the export list) only when the file system is mounted via .Nm . .Sh BUGS It is possible for a corrupted file system to cause a crash. Index: projects/largeSMP/sbin/rcorder/rcorder.8 =================================================================== --- projects/largeSMP/sbin/rcorder/rcorder.8 (revision 222782) +++ projects/largeSMP/sbin/rcorder/rcorder.8 (revision 222783) @@ -1,186 +1,192 @@ .\" $NetBSD: rcorder.8,v 1.3 2000/07/17 14:16:22 mrg Exp $ .\" .\" Copyright (c) 1998 .\" Perry E. Metzger. 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 acknowledgment: .\" This product includes software developed for the NetBSD Project .\" by Perry E. Metzger. .\" 4. 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. .\" .\" $FreeBSD$ .\" -.Dd June 9, 2008 +.Dd June 6, 2011 .Dt RCORDER 8 .Os .Sh NAME .Nm rcorder .Nd print a dependency ordering of interdependent files .Sh SYNOPSIS .Nm .Op Fl k Ar keep .Op Fl s Ar skip .Ar .Sh DESCRIPTION The .Nm utility is designed to print out a dependency ordering of a set of interdependent files. Typically it is used to find an execution sequence for a set of shell scripts in which certain files must be executed before others. .Pp Each file passed to .Nm must be annotated with special lines (which look like comments to the shell) which indicate the dependencies the files have upon certain points in the sequence, known as .Dq conditions , and which indicate, for each file, which .Dq conditions may be expected to be filled by that file. .Pp Within each file, a block containing a series of .Dq Li REQUIRE , .Dq Li PROVIDE , .Dq Li BEFORE and .Dq Li KEYWORD lines must appear. The format of the lines is rigid. Each line must begin with a single .Ql # , followed by a single space, followed by .Dq Li PROVIDE: , .Dq Li REQUIRE: , .Dq Li BEFORE: , or .Dq Li KEYWORD: . No deviation is permitted. Each dependency line is then followed by a series of conditions, separated by whitespace. Multiple .Dq Li PROVIDE , .Dq Li REQUIRE , .Dq Li BEFORE and .Dq Li KEYWORD lines may appear, but all such lines must appear in a sequence without any intervening lines, as once a line that does not follow the format is reached, parsing stops. +Note that for historical reasons, +.Dq Li REQUIRES , +.Dq Li PROVIDES , +and +.Dq Li KEYWORDS +are also accepted in addition to the above. .Pp The options are as follows: .Bl -tag -width indent .It Fl k Add the specified keyword to the .Dq "keep list" . If any .Fl k option is given, only those files containing the matching keyword are listed. .It Fl s Add the specified keyword to the .Dq "skip list" . If any .Fl s option is given, files containing the matching keyword are not listed. .El .Pp An example block follows: .Bd -literal -offset indent # REQUIRE: networking syslog # REQUIRE: usr # PROVIDE: dns nscd .Ed .Pp This block states that the file in which it appears depends upon the .Dq Li networking , .Dq Li syslog , and .Dq Li usr conditions, and provides the .Dq Li dns and .Dq Li nscd conditions. .Pp A file may contain zero .Dq Li PROVIDE lines, in which case it provides no conditions, and may contain zero .Dq Li REQUIRE lines, in which case it has no dependencies. There must be at least one file with no dependencies in the set of arguments passed to .Nm in order for it to find a starting place in the dependency ordering. .Sh DIAGNOSTICS The .Nm utility may print one of the following error messages and exit with a non-zero status if it encounters an error while processing the file list. .Bl -diag .It "Requirement %s has no providers, aborting." No file has a .Dq Li PROVIDE line corresponding to a condition present in a .Dq Li REQUIRE line in another file. .It "Circular dependency on provision %s, aborting." A set of files has a circular dependency which was detected while processing the stated condition. .It "Circular dependency on file %s, aborting." A set of files has a circular dependency which was detected while processing the stated file. .El .Sh SEE ALSO .Xr rc 8 .Sh HISTORY The .Nm utility first appeared in .Nx 1.5 . .Sh AUTHORS .An -nosplit Written by .An Perry E. Metzger Aq perry@piermont.com and .An Matthew R. Green Aq mrg@eterna.com.au . .Sh BUGS The .Dq Li REQUIRE keyword is misleading: It doesn't describe which daemons have to be running before a script will be started. It describes which scripts must be placed before it in the dependency ordering. For example, if your script has a .Dq Li REQUIRE on .Dq Li named , it means the script must be placed after the .Dq Li named script in the dependency ordering, not necessarily that it requires .Xr named 8 to be started or enabled. Index: projects/largeSMP/sbin =================================================================== --- projects/largeSMP/sbin (revision 222782) +++ projects/largeSMP/sbin (revision 222783) Property changes on: projects/largeSMP/sbin ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sbin:r222698-222782 Index: projects/largeSMP/share/man/man4/amdsbwd.4 =================================================================== --- projects/largeSMP/share/man/man4/amdsbwd.4 (revision 222782) +++ projects/largeSMP/share/man/man4/amdsbwd.4 (revision 222783) @@ -1,72 +1,74 @@ .\"- .\" Copyright (c) 2009 Andriy Gapon .\" 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 30, 2009 .Dt AMDSBWD 4 .Os .Sh NAME .Nm amdsbwd .Nd device driver for the AMD SB600/SB700/SB710/SB750 watchdog timer .Sh SYNOPSIS To compile this driver into the kernel, place the following line in your kernel configuration file: .Bd -ragged -offset indent .Cd "device amdsbwd" .Ed .Pp Alternatively, to load the driver as a module at boot time, place the following line in .Xr loader.conf 5 : .Bd -literal -offset indent amdsbwd_load="YES" .Ed .Sh DESCRIPTION The .Nm driver provides .Xr watchdog 4 support for the watchdog timers present on AMD SB600 and SB7xx south bridge chips. .Sh SEE ALSO .Xr watchdog 4 , .Xr watchdog 8 , .Xr watchdogd 8 , .Xr watchdog 9 .Sh HISTORY The .Nm driver first appeared in -.Fx 9.0 . +.Fx 7.3 +and +.Fx 8.1 . .Sh AUTHORS .An -nosplit The .Nm driver was written by -.An Andiry Gapon Aq avg@FreeBSD.org . +.An Andriy Gapon Aq avg@FreeBSD.org . This manual page was written by -.An Andiry Gapon Aq avg@FreeBSD.org . +.An Andriy Gapon Aq avg@FreeBSD.org . Index: projects/largeSMP/share/mk/bsd.arch.inc.mk =================================================================== --- projects/largeSMP/share/mk/bsd.arch.inc.mk (revision 222782) +++ projects/largeSMP/share/mk/bsd.arch.inc.mk (revision 222783) Property changes on: projects/largeSMP/share/mk/bsd.arch.inc.mk ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/share/mk/bsd.arch.inc.mk:r222698-222782 Index: projects/largeSMP/share/zoneinfo =================================================================== --- projects/largeSMP/share/zoneinfo (revision 222782) +++ projects/largeSMP/share/zoneinfo (revision 222783) Property changes on: projects/largeSMP/share/zoneinfo ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/share/zoneinfo:r222698-222782 Index: projects/largeSMP/sys/amd64/amd64/mp_machdep.c =================================================================== --- projects/largeSMP/sys/amd64/amd64/mp_machdep.c (revision 222782) +++ projects/largeSMP/sys/amd64/amd64/mp_machdep.c (revision 222783) @@ -1,1669 +1,1672 @@ /*- * 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_kstack_pages.h" #include "opt_mp_watchdog.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 #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; static 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 pcb **susppcbs = NULL; /* 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]; static u_long *ipi_hardclock_counts[MAXCPU]; #endif extern inthand_t IDTVEC(fast_syscall), IDTVEC(fast_syscall32); /* * 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 */ static 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_all_aps(void); static int start_ap(int apic_id); static void release_aps(void *dummy); static int hlt_logical_cpus; static u_int hyperthreading_cpus; /* logical cpus sharing L1 cache */ static cpuset_t hyperthreading_cpus_mask; static int hyperthreading_allowed = 1; static struct sysctl_ctx_list logical_cpu_clist; 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. */ - cpuid_count(0x04, 0, p); - max_cores = ((p[0] >> 26) & 0x3f) + 1; + 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 */ 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 */ 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) { cpuset_t tcpuset, tallcpus; struct pcpu *pc; struct nmi_pcpu *np; u_int64_t msr, cr0; 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) + IOPAGES * PAGE_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); /* Disable local APIC just to be sure. */ lapic_disable(); /* signal our startup to the BSP. */ mp_naps++; /* Spin until the BSP releases the AP's. */ while (!aps_ready) ia32_pause(); /* Initialize the PAT MSR. */ pmap_init_pat(); /* set up CPU registers and state */ cpu_setregs(); /* set up SSE/NX registers */ initializecpu(); /* set up FPU state on the AP */ fpuinit(); /* A quick check from sanity claus */ if (PCPU_GET(apic_id) != lapic_id()) { printf("SMP: cpuid = %d\n", PCPU_GET(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", PCPU_GET(cpuid)); printf("SMP: AP CPU #%d Launched!\n", PCPU_GET(cpuid)); tcpuset = PCPU_GET(cpumask); /* 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_OR(&logical_cpus_mask, &tcpuset); /* Determine if we are a hyperthread. */ if (hyperthreading_cpus > 1 && PCPU_GET(apic_id) % hyperthreading_cpus != 0) CPU_OR(&hyperthreading_cpus_mask, &tcpuset); /* Build our map of 'other' CPUs. */ tallcpus = all_cpus; CPU_NAND(&tallcpus, &tcpuset); PCPU_SET(other_cpus, tallcpus); if (bootverbose) lapic_dump("AP"); if (smp_cpus == mp_ncpus) { /* enable IPI's, tlb shootdown, freezes etc */ atomic_store_rel_int(&smp_started, 1); smp_active = 1; /* historic */ } /* * Enable global pages TLB extension * This also implicitly flushes the TLB */ load_cr4(rcr4() | CR4_PGE); 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) 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; #if defined(SCHED_ULE) /* * Don't use HT CPU if it has been disabled by a * tunable. */ if (hyperthreading_allowed == 0) { cpu_info[i].cpu_disabled = 1; continue; } #endif } /* 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; } } /* * 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 */ static int start_all_aps(void) { cpuset_t tallcpus, tcpuset; 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_alloc(kernel_map, KSTACK_PAGES * PAGE_SIZE); doublefault_stack = (char *)kmem_alloc(kernel_map, PAGE_SIZE); nmi_stack = (char *)kmem_alloc(kernel_map, PAGE_SIZE); dpcpu = (void *)kmem_alloc(kernel_map, DPCPU_SIZE); 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 */ } /* build our map of 'other' CPUs */ tallcpus = all_cpus; tcpuset = PCPU_GET(cpumask); CPU_NAND(&tallcpus, &tcpuset); PCPU_SET(other_cpus, tallcpus); /* 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; /* * first we do an INIT/RESET IPI this INIT IPI might be run, reseting * 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. */ /* do an INIT IPI: assert RESET */ lapic_ipi_raw(APIC_DEST_DESTFLD | APIC_TRIGMOD_EDGE | APIC_LEVEL_ASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_INIT, apic_id); /* wait for pending status end */ lapic_ipi_wait(-1); /* do an INIT IPI: deassert RESET */ lapic_ipi_raw(APIC_DEST_ALLESELF | APIC_TRIGMOD_LEVEL | APIC_LEVEL_DEASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_INIT, 0); /* wait for pending status end */ DELAY(10000); /* wait ~10mS */ lapic_ipi_wait(-1); /* * 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. */ /* do a STARTUP IPI */ lapic_ipi_raw(APIC_DEST_DESTFLD | APIC_TRIGMOD_EDGE | APIC_LEVEL_DEASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_STARTUP | vector, apic_id); lapic_ipi_wait(-1); 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_DEASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_STARTUP | vector, apic_id); lapic_ipi_wait(-1); DELAY(200); /* wait ~200uS */ /* 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]; 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 */ /* * 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_rflags() & 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 { sched_pin(); CPU_NAND(&mask, PCPU_PTR(cpumask)); sched_unpin(); if (CPU_EMPTY(&mask)) return; } if (!(read_rflags() & 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 = cpusetobj_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 = cpusetobj_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) { sched_pin(); if (IPI_IS_BITMAPED(ipi)) { ipi_selected(PCPU_GET(other_cpus), ipi); sched_unpin(); 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, PCPU_PTR(other_cpus)); sched_unpin(); CTR2(KTR_SMP, "%s: ipi: %x", __func__, ipi); lapic_ipi_vectored(ipi, APIC_IPI_DEST_OTHERS); } int ipi_nmi_handler() { cpuset_t cpumask; /* * 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. */ sched_pin(); cpumask = PCPU_GET(cpumask); sched_unpin(); if (!CPU_OVERLAP(&ipi_nmi_pending, &cpumask)) return (1); CPU_NAND_ATOMIC(&ipi_nmi_pending, &cpumask); cpustop_handler(); return (0); } /* * Handle an IPI_STOP by saving our current context and spinning until we * are resumed. */ void cpustop_handler(void) { cpuset_t cpumask; u_int cpu; sched_pin(); cpu = PCPU_GET(cpuid); cpumask = PCPU_GET(cpumask); sched_unpin(); savectx(&stoppcbs[cpu]); /* Indicate that we are stopped */ CPU_OR_ATOMIC(&stopped_cpus, &cpumask); /* Wait for restart */ while (!CPU_OVERLAP(&started_cpus, &cpumask)) ia32_pause(); CPU_NAND_ATOMIC(&started_cpus, &cpumask); CPU_NAND_ATOMIC(&stopped_cpus, &cpumask); 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) { cpuset_t cpumask; register_t cr3, rf; u_int cpu; cpu = PCPU_GET(cpuid); cpumask = PCPU_GET(cpumask); rf = intr_disable(); cr3 = rcr3(); if (savectx(susppcbs[cpu])) { wbinvd(); CPU_OR_ATOMIC(&stopped_cpus, &cpumask); } else { pmap_init_pat(); PCPU_SET(switchtime, 0); PCPU_SET(switchticks, ticks); } /* Wait for resume */ while (!CPU_OVERLAP(&started_cpus, &cpumask)) ia32_pause(); CPU_NAND_ATOMIC(&started_cpus, &cpumask); CPU_NAND_ATOMIC(&stopped_cpus, &cpumask); /* Restore CR3 and enable interrupts */ load_cr3(cr3); mca_resume(); lapic_setup(0); intr_restore(rf); } /* * 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); static int sysctl_hlt_cpus(SYSCTL_HANDLER_ARGS) { cpuset_t mask; int error; mask = hlt_cpus_mask; error = sysctl_handle_opaque(oidp, &mask, sizeof(mask), req); if (error || !req->newptr) return (error); if (!CPU_EMPTY(&logical_cpus_mask) && CPU_SUBSET(&mask, &logical_cpus_mask)) hlt_logical_cpus = 1; else hlt_logical_cpus = 0; if (! hyperthreading_allowed) CPU_OR(&mask, &hyperthreading_cpus_mask); if (CPU_SUBSET(&mask, &all_cpus)) CPU_CLR(0, &mask); hlt_cpus_mask = mask; return (error); } SYSCTL_PROC(_machdep, OID_AUTO, hlt_cpus, CTLTYPE_STRUCT | CTLFLAG_RW | CTLFLAG_MPSAFE, 0, 0, sysctl_hlt_cpus, "S", "Bitmap of CPUs to halt. 101 (binary) will halt CPUs 0 and 2."); static int sysctl_hlt_logical_cpus(SYSCTL_HANDLER_ARGS) { int disable, error; disable = hlt_logical_cpus; error = sysctl_handle_int(oidp, &disable, 0, req); if (error || !req->newptr) return (error); if (disable) CPU_OR(&hlt_cpus_mask, &logical_cpus_mask); else CPU_NAND(&hlt_cpus_mask, &logical_cpus_mask); if (! hyperthreading_allowed) CPU_OR(&hlt_cpus_mask, &hyperthreading_cpus_mask); if (CPU_SUBSET(&hlt_cpus_mask, &all_cpus)) CPU_CLR(0, &hlt_cpus_mask); hlt_logical_cpus = disable; return (error); } static int sysctl_hyperthreading_allowed(SYSCTL_HANDLER_ARGS) { int allowed, error; allowed = hyperthreading_allowed; error = sysctl_handle_int(oidp, &allowed, 0, req); if (error || !req->newptr) return (error); #ifdef SCHED_ULE /* * SCHED_ULE doesn't allow enabling/disabling HT cores at * run-time. */ if (allowed != hyperthreading_allowed) return (ENOTSUP); return (error); #endif if (allowed) CPU_NAND(&hlt_cpus_mask, &hyperthreading_cpus_mask); else CPU_OR(&hlt_cpus_mask, &hyperthreading_cpus_mask); if (!CPU_EMPTY(&logical_cpus_mask) && CPU_SUBSET(&hlt_cpus_mask, &logical_cpus_mask)) hlt_logical_cpus = 1; else hlt_logical_cpus = 0; if (CPU_SUBSET(&hlt_cpus_mask, &all_cpus)) CPU_CLR(0, &hlt_cpus_mask); hyperthreading_allowed = allowed; return (error); } static void cpu_hlt_setup(void *dummy __unused) { if (!CPU_EMPTY(&logical_cpus_mask)) { TUNABLE_INT_FETCH("machdep.hlt_logical_cpus", &hlt_logical_cpus); sysctl_ctx_init(&logical_cpu_clist); SYSCTL_ADD_PROC(&logical_cpu_clist, SYSCTL_STATIC_CHILDREN(_machdep), OID_AUTO, "hlt_logical_cpus", CTLTYPE_INT|CTLFLAG_RW, 0, 0, sysctl_hlt_logical_cpus, "IU", ""); SYSCTL_ADD_UINT(&logical_cpu_clist, SYSCTL_STATIC_CHILDREN(_machdep), OID_AUTO, "logical_cpus_mask", CTLTYPE_INT|CTLFLAG_RD, &logical_cpus_mask, 0, ""); if (hlt_logical_cpus) CPU_OR(&hlt_cpus_mask, &logical_cpus_mask); /* * If necessary for security purposes, force * hyperthreading off, regardless of the value * of hlt_logical_cpus. */ if (!CPU_EMPTY(&hyperthreading_cpus_mask)) { SYSCTL_ADD_PROC(&logical_cpu_clist, SYSCTL_STATIC_CHILDREN(_machdep), OID_AUTO, "hyperthreading_allowed", CTLTYPE_INT|CTLFLAG_RW, 0, 0, sysctl_hyperthreading_allowed, "IU", ""); if (! hyperthreading_allowed) CPU_OR(&hlt_cpus_mask, &hyperthreading_cpus_mask); } } } SYSINIT(cpu_hlt, SI_SUB_SMP, SI_ORDER_ANY, cpu_hlt_setup, NULL); int mp_grab_cpu_hlt(void) { cpuset_t mask; #ifdef MP_WATCHDOG u_int cpuid; #endif int retval; mask = PCPU_GET(cpumask); #ifdef MP_WATCHDOG cpuid = PCPU_GET(cpuid); ap_watchdog(cpuid); #endif retval = 0; while (CPU_OVERLAP(&mask, &hlt_cpus_mask)) { retval = 1; __asm __volatile("sti; hlt" : : : "memory"); } return (retval); } #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: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/largeSMP/sys/amd64/include/xen =================================================================== --- projects/largeSMP/sys/amd64/include/xen (revision 222782) +++ projects/largeSMP/sys/amd64/include/xen (revision 222783) Property changes on: projects/largeSMP/sys/amd64/include/xen ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/amd64/include/xen:r222698-222782 Index: projects/largeSMP/sys/boot/i386/efi =================================================================== --- projects/largeSMP/sys/boot/i386/efi (revision 222782) +++ projects/largeSMP/sys/boot/i386/efi (revision 222783) Property changes on: projects/largeSMP/sys/boot/i386/efi ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/boot/i386/efi:r222698-222782 Index: projects/largeSMP/sys/boot/ia64/efi =================================================================== --- projects/largeSMP/sys/boot/ia64/efi (revision 222782) +++ projects/largeSMP/sys/boot/ia64/efi (revision 222783) Property changes on: projects/largeSMP/sys/boot/ia64/efi ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/boot/ia64/efi:r222698-222782 Index: projects/largeSMP/sys/boot/ia64/ski =================================================================== --- projects/largeSMP/sys/boot/ia64/ski (revision 222782) +++ projects/largeSMP/sys/boot/ia64/ski (revision 222783) Property changes on: projects/largeSMP/sys/boot/ia64/ski ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/boot/ia64/ski:r222698-222782 Index: projects/largeSMP/sys/boot/powerpc/boot1.chrp =================================================================== --- projects/largeSMP/sys/boot/powerpc/boot1.chrp (revision 222782) +++ projects/largeSMP/sys/boot/powerpc/boot1.chrp (revision 222783) Property changes on: projects/largeSMP/sys/boot/powerpc/boot1.chrp ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/boot/powerpc/boot1.chrp:r222698-222782 Index: projects/largeSMP/sys/boot/powerpc/ofw =================================================================== --- projects/largeSMP/sys/boot/powerpc/ofw (revision 222782) +++ projects/largeSMP/sys/boot/powerpc/ofw (revision 222783) Property changes on: projects/largeSMP/sys/boot/powerpc/ofw ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/boot/powerpc/ofw:r222698-222782 Index: projects/largeSMP/sys/boot =================================================================== --- projects/largeSMP/sys/boot (revision 222782) +++ projects/largeSMP/sys/boot (revision 222783) Property changes on: projects/largeSMP/sys/boot ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/boot:r222698-222782 Index: projects/largeSMP/sys/cddl/compat/opensolaris/sys/atomic.h =================================================================== --- projects/largeSMP/sys/cddl/compat/opensolaris/sys/atomic.h (revision 222782) +++ projects/largeSMP/sys/cddl/compat/opensolaris/sys/atomic.h (revision 222783) @@ -1,138 +1,136 @@ /*- * Copyright (c) 2007 Pawel Jakub Dawidek * 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 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. * * $FreeBSD$ */ #ifndef _OPENSOLARIS_SYS_ATOMIC_H_ #define _OPENSOLARIS_SYS_ATOMIC_H_ #include #include #define casptr(_a, _b, _c) \ atomic_cmpset_ptr((volatile uintptr_t *)(_a), (uintptr_t)(_b), (uintptr_t) (_c)) #define cas32 atomic_cmpset_32 #if !defined(__LP64__) && !defined(__mips_n32) extern void atomic_add_64(volatile uint64_t *target, int64_t delta); extern void atomic_dec_64(volatile uint64_t *target); #endif -#ifndef __LP64__ -#endif #ifndef __sparc64__ extern uint32_t atomic_cas_32(volatile uint32_t *target, uint32_t cmp, uint32_t newval); extern uint64_t atomic_cas_64(volatile uint64_t *target, uint64_t cmp, uint64_t newval); #endif extern uint64_t atomic_add_64_nv(volatile uint64_t *target, int64_t delta); extern uint8_t atomic_or_8_nv(volatile uint8_t *target, uint8_t value); extern void membar_producer(void); #if defined(__sparc64__) || defined(__powerpc__) || defined(__arm__) || \ defined(__mips__) extern void atomic_or_8(volatile uint8_t *target, uint8_t value); #else static __inline void atomic_or_8(volatile uint8_t *target, uint8_t value) { atomic_set_8(target, value); } #endif static __inline uint32_t atomic_add_32_nv(volatile uint32_t *target, int32_t delta) { return (atomic_fetchadd_32(target, delta) + delta); } static __inline u_int atomic_add_int_nv(volatile u_int *target, int delta) { return (atomic_add_32_nv(target, delta)); } static __inline void atomic_dec_32(volatile uint32_t *target) { atomic_subtract_32(target, 1); } static __inline uint32_t atomic_dec_32_nv(volatile uint32_t *target) { return (atomic_fetchadd_32(target, -1) - 1); } #if defined(__LP64__) || defined(__mips_n32) static __inline void atomic_dec_64(volatile uint64_t *target) { atomic_subtract_64(target, 1); } #endif static __inline void atomic_inc_32(volatile uint32_t *target) { atomic_add_32(target, 1); } static __inline uint32_t atomic_inc_32_nv(volatile uint32_t *target) { return (atomic_add_32_nv(target, 1)); } static __inline void atomic_inc_64(volatile uint64_t *target) { atomic_add_64(target, 1); } static __inline uint64_t atomic_inc_64_nv(volatile uint64_t *target) { return (atomic_add_64_nv(target, 1)); } #if !defined(COMPAT_32BIT) && defined(__LP64__) static __inline void * atomic_cas_ptr(volatile void *target, void *cmp, void *newval) { return ((void *)atomic_cas_64((volatile uint64_t *)target, (uint64_t)cmp, (uint64_t)newval)); } #else static __inline void * atomic_cas_ptr(volatile void *target, void *cmp, void *newval) { return ((void *)atomic_cas_32((volatile uint32_t *)target, (uint32_t)cmp, (uint32_t)newval)); } #endif /* !defined(COMPAT_32BIT) && defined(__LP64__) */ #endif /* !_OPENSOLARIS_SYS_ATOMIC_H_ */ Index: projects/largeSMP/sys/cddl/contrib/opensolaris =================================================================== --- projects/largeSMP/sys/cddl/contrib/opensolaris (revision 222782) +++ projects/largeSMP/sys/cddl/contrib/opensolaris (revision 222783) Property changes on: projects/largeSMP/sys/cddl/contrib/opensolaris ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/cddl/contrib/opensolaris:r222698-222782 Index: projects/largeSMP/sys/conf/files =================================================================== --- projects/largeSMP/sys/conf/files (revision 222782) +++ projects/largeSMP/sys/conf/files (revision 222783) @@ -1,3488 +1,3490 @@ # $FreeBSD$ # # The long compile-with and dependency lines are required because of # limitations in config: backslash-newline doesn't work in strings, and # dependency lines other than the first are silently ignored. # acpi_quirks.h optional acpi \ dependency "$S/tools/acpi_quirks2h.awk $S/dev/acpica/acpi_quirks" \ compile-with "${AWK} -f $S/tools/acpi_quirks2h.awk $S/dev/acpica/acpi_quirks" \ no-obj no-implicit-rule before-depend \ clean "acpi_quirks.h" aicasm optional ahc | ahd \ dependency "$S/dev/aic7xxx/aicasm/*.[chyl]" \ compile-with "CC='${CC}' ${MAKE} -f $S/dev/aic7xxx/aicasm/Makefile MAKESRCPATH=$S/dev/aic7xxx/aicasm" \ no-obj no-implicit-rule \ clean "aicasm* y.tab.h" aic7xxx_seq.h optional ahc \ compile-with "./aicasm ${INCLUDES} -I$S/cam/scsi -I$S/dev/aic7xxx -o aic7xxx_seq.h -r aic7xxx_reg.h -p aic7xxx_reg_print.c -i $S/dev/aic7xxx/aic7xxx_osm.h $S/dev/aic7xxx/aic7xxx.seq" \ no-obj no-implicit-rule before-depend local \ clean "aic7xxx_seq.h" \ dependency "$S/dev/aic7xxx/aic7xxx.{reg,seq} $S/cam/scsi/scsi_message.h aicasm" aic7xxx_reg.h optional ahc \ compile-with "./aicasm ${INCLUDES} -I$S/cam/scsi -I$S/dev/aic7xxx -o aic7xxx_seq.h -r aic7xxx_reg.h -p aic7xxx_reg_print.c -i $S/dev/aic7xxx/aic7xxx_osm.h $S/dev/aic7xxx/aic7xxx.seq" \ no-obj no-implicit-rule before-depend local \ clean "aic7xxx_reg.h" \ dependency "$S/dev/aic7xxx/aic7xxx.{reg,seq} $S/cam/scsi/scsi_message.h aicasm" aic7xxx_reg_print.c optional ahc \ compile-with "./aicasm ${INCLUDES} -I$S/cam/scsi -I$S/dev/aic7xxx -o aic7xxx_seq.h -r aic7xxx_reg.h -p aic7xxx_reg_print.c -i $S/dev/aic7xxx/aic7xxx_osm.h $S/dev/aic7xxx/aic7xxx.seq" \ no-obj no-implicit-rule local \ clean "aic7xxx_reg_print.c" \ dependency "$S/dev/aic7xxx/aic7xxx.{reg,seq} $S/cam/scsi/scsi_message.h aicasm" aic7xxx_reg_print.o optional ahc ahc_reg_pretty_print \ compile-with "${NORMAL_C}" \ no-implicit-rule local aic79xx_seq.h optional ahd pci \ compile-with "./aicasm ${INCLUDES} -I$S/cam/scsi -I$S/dev/aic7xxx -o aic79xx_seq.h -r aic79xx_reg.h -p aic79xx_reg_print.c -i $S/dev/aic7xxx/aic79xx_osm.h $S/dev/aic7xxx/aic79xx.seq" \ no-obj no-implicit-rule before-depend local \ clean "aic79xx_seq.h" \ dependency "$S/dev/aic7xxx/aic79xx.{reg,seq} $S/cam/scsi/scsi_message.h aicasm" aic79xx_reg.h optional ahd pci \ compile-with "./aicasm ${INCLUDES} -I$S/cam/scsi -I$S/dev/aic7xxx -o aic79xx_seq.h -r aic79xx_reg.h -p aic79xx_reg_print.c -i $S/dev/aic7xxx/aic79xx_osm.h $S/dev/aic7xxx/aic79xx.seq" \ no-obj no-implicit-rule before-depend local \ clean "aic79xx_reg.h" \ dependency "$S/dev/aic7xxx/aic79xx.{reg,seq} $S/cam/scsi/scsi_message.h aicasm" aic79xx_reg_print.c optional ahd pci \ compile-with "./aicasm ${INCLUDES} -I$S/cam/scsi -I$S/dev/aic7xxx -o aic79xx_seq.h -r aic79xx_reg.h -p aic79xx_reg_print.c -i $S/dev/aic7xxx/aic79xx_osm.h $S/dev/aic7xxx/aic79xx.seq" \ no-obj no-implicit-rule local \ clean "aic79xx_reg_print.c" \ dependency "$S/dev/aic7xxx/aic79xx.{reg,seq} $S/cam/scsi/scsi_message.h aicasm" aic79xx_reg_print.o optional ahd pci ahd_reg_pretty_print \ compile-with "${NORMAL_C}" \ no-implicit-rule local emu10k1-alsa%diked.h optional snd_emu10k1 | snd_emu10kx \ dependency "$S/tools/sound/emu10k1-mkalsa.sh $S/gnu/dev/sound/pci/emu10k1-alsa.h" \ compile-with "CC='${CC}' AWK=${AWK} sh $S/tools/sound/emu10k1-mkalsa.sh $S/gnu/dev/sound/pci/emu10k1-alsa.h emu10k1-alsa%diked.h" \ no-obj no-implicit-rule before-depend \ clean "emu10k1-alsa%diked.h" # # The 'fdt_dtb_file' target covers an actual DTB file name, which is derived # from the specified source (DTS) file: .dts -> .dtb # fdt_dtb_file optional fdt \ compile-with "if [ -f $S/boot/fdt/dts/${FDT_DTS_FILE} ]; then dtc -O dtb -o `echo ${FDT_DTS_FILE} | cut -d. -f1`.dtb -b 0 -p 1024 $S/boot/fdt/dts/${FDT_DTS_FILE}; fi" \ no-obj no-implicit-rule before-depend \ clean "`echo ${FDT_DTS_FILE} | cut -d. -f1`.dtb" fdt_static_dtb.h optional fdt fdt_dtb_static \ compile-with "sh $S/tools/fdt/make_dtbh.sh ${FDT_DTS_FILE} ." \ no-obj no-implicit-rule before-depend \ clean "fdt_static_dtb.h" p16v-alsa%diked.h optional snd_emu10kx pci \ dependency "$S/tools/sound/emu10k1-mkalsa.sh $S/gnu/dev/sound/pci/p16v-alsa.h" \ compile-with "CC='${CC}' AWK=${AWK} sh $S/tools/sound/emu10k1-mkalsa.sh $S/gnu/dev/sound/pci/p16v-alsa.h p16v-alsa%diked.h" \ no-obj no-implicit-rule before-depend \ clean "p16v-alsa%diked.h" p17v-alsa%diked.h optional snd_emu10kx pci \ dependency "$S/tools/sound/emu10k1-mkalsa.sh $S/gnu/dev/sound/pci/p17v-alsa.h" \ compile-with "CC='${CC}' AWK=${AWK} sh $S/tools/sound/emu10k1-mkalsa.sh $S/gnu/dev/sound/pci/p17v-alsa.h p17v-alsa%diked.h" \ no-obj no-implicit-rule before-depend \ clean "p17v-alsa%diked.h" feeder_eq_gen.h optional sound \ dependency "$S/tools/sound/feeder_eq_mkfilter.awk" \ compile-with "${AWK} -f $S/tools/sound/feeder_eq_mkfilter.awk -- ${FEEDER_EQ_PRESETS} > feeder_eq_gen.h" \ no-obj no-implicit-rule before-depend \ clean "feeder_eq_gen.h" feeder_rate_gen.h optional sound \ dependency "$S/tools/sound/feeder_rate_mkfilter.awk" \ compile-with "${AWK} -f $S/tools/sound/feeder_rate_mkfilter.awk -- ${FEEDER_RATE_PRESETS} > feeder_rate_gen.h" \ no-obj no-implicit-rule before-depend \ clean "feeder_rate_gen.h" snd_fxdiv_gen.h optional sound \ dependency "$S/tools/sound/snd_fxdiv_gen.awk" \ compile-with "${AWK} -f $S/tools/sound/snd_fxdiv_gen.awk -- > snd_fxdiv_gen.h" \ no-obj no-implicit-rule before-depend \ clean "snd_fxdiv_gen.h" miidevs.h optional miibus | mii \ dependency "$S/tools/miidevs2h.awk $S/dev/mii/miidevs" \ compile-with "${AWK} -f $S/tools/miidevs2h.awk $S/dev/mii/miidevs" \ no-obj no-implicit-rule before-depend \ clean "miidevs.h" pccarddevs.h standard \ dependency "$S/tools/pccarddevs2h.awk $S/dev/pccard/pccarddevs" \ compile-with "${AWK} -f $S/tools/pccarddevs2h.awk $S/dev/pccard/pccarddevs" \ no-obj no-implicit-rule before-depend \ clean "pccarddevs.h" teken_state.h optional sc \ dependency "$S/teken/gensequences $S/teken/sequences" \ compile-with "${AWK} -f $S/teken/gensequences $S/teken/sequences > teken_state.h" \ no-obj no-implicit-rule before-depend \ clean "teken_state.h" usbdevs.h optional usb \ dependency "$S/tools/usbdevs2h.awk $S/dev/usb/usbdevs" \ compile-with "${AWK} -f $S/tools/usbdevs2h.awk $S/dev/usb/usbdevs -h" \ no-obj no-implicit-rule before-depend \ clean "usbdevs.h" usbdevs_data.h optional usb \ dependency "$S/tools/usbdevs2h.awk $S/dev/usb/usbdevs" \ compile-with "${AWK} -f $S/tools/usbdevs2h.awk $S/dev/usb/usbdevs -d" \ no-obj no-implicit-rule before-depend \ clean "usbdevs_data.h" cam/cam.c optional scbus cam/cam_periph.c optional scbus cam/cam_queue.c optional scbus cam/cam_sim.c optional scbus cam/cam_xpt.c optional scbus cam/ata/ata_all.c optional scbus cam/ata/ata_xpt.c optional scbus cam/ata/ata_pmp.c optional scbus cam/scsi/scsi_xpt.c optional scbus cam/scsi/scsi_all.c optional scbus cam/scsi/scsi_cd.c optional cd cam/scsi/scsi_ch.c optional ch cam/ata/ata_da.c optional ada | da cam/scsi/scsi_da.c optional da cam/scsi/scsi_low.c optional ct | ncv | nsp | stg cam/scsi/scsi_low_pisa.c optional ct | ncv | nsp | stg cam/scsi/scsi_pass.c optional pass cam/scsi/scsi_pt.c optional pt cam/scsi/scsi_sa.c optional sa cam/scsi/scsi_ses.c optional ses cam/scsi/scsi_sg.c optional sg cam/scsi/scsi_targ_bh.c optional targbh cam/scsi/scsi_target.c optional targ cam/scsi/smp_all.c optional scbus contrib/altq/altq/altq_cbq.c optional altq \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/altq/altq/altq_cdnr.c optional altq contrib/altq/altq/altq_hfsc.c optional altq \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/altq/altq/altq_priq.c optional altq \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/altq/altq/altq_red.c optional altq \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/altq/altq/altq_rio.c optional altq \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/altq/altq/altq_rmclass.c optional altq contrib/altq/altq/altq_subr.c optional altq \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/dev/acpica/debugger/dbcmds.c optional acpi acpi_debug contrib/dev/acpica/debugger/dbdisply.c optional acpi acpi_debug contrib/dev/acpica/debugger/dbexec.c optional acpi acpi_debug contrib/dev/acpica/debugger/dbfileio.c optional acpi acpi_debug contrib/dev/acpica/debugger/dbhistry.c optional acpi acpi_debug contrib/dev/acpica/debugger/dbinput.c optional acpi acpi_debug contrib/dev/acpica/debugger/dbmethod.c optional acpi acpi_debug contrib/dev/acpica/debugger/dbnames.c optional acpi acpi_debug contrib/dev/acpica/debugger/dbstats.c optional acpi acpi_debug contrib/dev/acpica/debugger/dbutils.c optional acpi acpi_debug contrib/dev/acpica/debugger/dbxface.c optional acpi acpi_debug contrib/dev/acpica/disassembler/dmbuffer.c optional acpi acpi_debug contrib/dev/acpica/disassembler/dmnames.c optional acpi acpi_debug contrib/dev/acpica/disassembler/dmopcode.c optional acpi acpi_debug contrib/dev/acpica/disassembler/dmobject.c optional acpi acpi_debug contrib/dev/acpica/disassembler/dmresrc.c optional acpi acpi_debug contrib/dev/acpica/disassembler/dmresrcl.c optional acpi acpi_debug contrib/dev/acpica/disassembler/dmresrcs.c optional acpi acpi_debug contrib/dev/acpica/disassembler/dmutils.c optional acpi acpi_debug contrib/dev/acpica/disassembler/dmwalk.c optional acpi acpi_debug contrib/dev/acpica/dispatcher/dsargs.c optional acpi contrib/dev/acpica/dispatcher/dscontrol.c optional acpi contrib/dev/acpica/dispatcher/dsfield.c optional acpi contrib/dev/acpica/dispatcher/dsinit.c optional acpi contrib/dev/acpica/dispatcher/dsmethod.c optional acpi contrib/dev/acpica/dispatcher/dsmthdat.c optional acpi contrib/dev/acpica/dispatcher/dsobject.c optional acpi contrib/dev/acpica/dispatcher/dsopcode.c optional acpi contrib/dev/acpica/dispatcher/dsutils.c optional acpi contrib/dev/acpica/dispatcher/dswexec.c optional acpi contrib/dev/acpica/dispatcher/dswload.c optional acpi contrib/dev/acpica/dispatcher/dswload2.c optional acpi contrib/dev/acpica/dispatcher/dswscope.c optional acpi contrib/dev/acpica/dispatcher/dswstate.c optional acpi contrib/dev/acpica/events/evevent.c optional acpi contrib/dev/acpica/events/evglock.c optional acpi contrib/dev/acpica/events/evgpe.c optional acpi contrib/dev/acpica/events/evgpeblk.c optional acpi contrib/dev/acpica/events/evgpeinit.c optional acpi contrib/dev/acpica/events/evgpeutil.c optional acpi contrib/dev/acpica/events/evmisc.c optional acpi contrib/dev/acpica/events/evregion.c optional acpi contrib/dev/acpica/events/evrgnini.c optional acpi contrib/dev/acpica/events/evsci.c optional acpi contrib/dev/acpica/events/evxface.c optional acpi contrib/dev/acpica/events/evxfevnt.c optional acpi contrib/dev/acpica/events/evxfgpe.c optional acpi contrib/dev/acpica/events/evxfregn.c optional acpi contrib/dev/acpica/executer/exconfig.c optional acpi contrib/dev/acpica/executer/exconvrt.c optional acpi contrib/dev/acpica/executer/excreate.c optional acpi contrib/dev/acpica/executer/exdebug.c optional acpi contrib/dev/acpica/executer/exdump.c optional acpi contrib/dev/acpica/executer/exfield.c optional acpi contrib/dev/acpica/executer/exfldio.c optional acpi contrib/dev/acpica/executer/exmisc.c optional acpi contrib/dev/acpica/executer/exmutex.c optional acpi contrib/dev/acpica/executer/exnames.c optional acpi contrib/dev/acpica/executer/exoparg1.c optional acpi contrib/dev/acpica/executer/exoparg2.c optional acpi contrib/dev/acpica/executer/exoparg3.c optional acpi contrib/dev/acpica/executer/exoparg6.c optional acpi contrib/dev/acpica/executer/exprep.c optional acpi contrib/dev/acpica/executer/exregion.c optional acpi contrib/dev/acpica/executer/exresnte.c optional acpi contrib/dev/acpica/executer/exresolv.c optional acpi contrib/dev/acpica/executer/exresop.c optional acpi contrib/dev/acpica/executer/exstore.c optional acpi contrib/dev/acpica/executer/exstoren.c optional acpi contrib/dev/acpica/executer/exstorob.c optional acpi contrib/dev/acpica/executer/exsystem.c optional acpi contrib/dev/acpica/executer/exutils.c optional acpi contrib/dev/acpica/hardware/hwacpi.c optional acpi contrib/dev/acpica/hardware/hwgpe.c optional acpi contrib/dev/acpica/hardware/hwpci.c optional acpi contrib/dev/acpica/hardware/hwregs.c optional acpi contrib/dev/acpica/hardware/hwsleep.c optional acpi contrib/dev/acpica/hardware/hwtimer.c optional acpi contrib/dev/acpica/hardware/hwvalid.c optional acpi contrib/dev/acpica/hardware/hwxface.c optional acpi contrib/dev/acpica/namespace/nsaccess.c optional acpi contrib/dev/acpica/namespace/nsalloc.c optional acpi contrib/dev/acpica/namespace/nsdump.c optional acpi contrib/dev/acpica/namespace/nseval.c optional acpi contrib/dev/acpica/namespace/nsinit.c optional acpi contrib/dev/acpica/namespace/nsload.c optional acpi contrib/dev/acpica/namespace/nsnames.c optional acpi contrib/dev/acpica/namespace/nsobject.c optional acpi contrib/dev/acpica/namespace/nsparse.c optional acpi contrib/dev/acpica/namespace/nspredef.c optional acpi contrib/dev/acpica/namespace/nsrepair.c optional acpi contrib/dev/acpica/namespace/nsrepair2.c optional acpi contrib/dev/acpica/namespace/nssearch.c optional acpi contrib/dev/acpica/namespace/nsutils.c optional acpi contrib/dev/acpica/namespace/nswalk.c optional acpi contrib/dev/acpica/namespace/nsxfeval.c optional acpi contrib/dev/acpica/namespace/nsxfname.c optional acpi contrib/dev/acpica/namespace/nsxfobj.c optional acpi contrib/dev/acpica/parser/psargs.c optional acpi contrib/dev/acpica/parser/psloop.c optional acpi contrib/dev/acpica/parser/psopcode.c optional acpi contrib/dev/acpica/parser/psparse.c optional acpi contrib/dev/acpica/parser/psscope.c optional acpi contrib/dev/acpica/parser/pstree.c optional acpi contrib/dev/acpica/parser/psutils.c optional acpi contrib/dev/acpica/parser/pswalk.c optional acpi contrib/dev/acpica/parser/psxface.c optional acpi contrib/dev/acpica/resources/rsaddr.c optional acpi contrib/dev/acpica/resources/rscalc.c optional acpi contrib/dev/acpica/resources/rscreate.c optional acpi contrib/dev/acpica/resources/rsdump.c optional acpi contrib/dev/acpica/resources/rsinfo.c optional acpi contrib/dev/acpica/resources/rsio.c optional acpi contrib/dev/acpica/resources/rsirq.c optional acpi contrib/dev/acpica/resources/rslist.c optional acpi contrib/dev/acpica/resources/rsmemory.c optional acpi contrib/dev/acpica/resources/rsmisc.c optional acpi contrib/dev/acpica/resources/rsutils.c optional acpi contrib/dev/acpica/resources/rsxface.c optional acpi contrib/dev/acpica/tables/tbfadt.c optional acpi contrib/dev/acpica/tables/tbfind.c optional acpi contrib/dev/acpica/tables/tbinstal.c optional acpi contrib/dev/acpica/tables/tbutils.c optional acpi contrib/dev/acpica/tables/tbxface.c optional acpi contrib/dev/acpica/tables/tbxfroot.c optional acpi contrib/dev/acpica/utilities/utalloc.c optional acpi contrib/dev/acpica/utilities/utcache.c optional acpi contrib/dev/acpica/utilities/utcopy.c optional acpi contrib/dev/acpica/utilities/utdebug.c optional acpi contrib/dev/acpica/utilities/utdecode.c optional acpi contrib/dev/acpica/utilities/utdelete.c optional acpi contrib/dev/acpica/utilities/uteval.c optional acpi contrib/dev/acpica/utilities/utglobal.c optional acpi contrib/dev/acpica/utilities/utids.c optional acpi contrib/dev/acpica/utilities/utinit.c optional acpi contrib/dev/acpica/utilities/utlock.c optional acpi contrib/dev/acpica/utilities/utmath.c optional acpi contrib/dev/acpica/utilities/utmisc.c optional acpi contrib/dev/acpica/utilities/utmutex.c optional acpi contrib/dev/acpica/utilities/utobject.c optional acpi contrib/dev/acpica/utilities/utosi.c optional acpi contrib/dev/acpica/utilities/utresrc.c optional acpi contrib/dev/acpica/utilities/utstate.c optional acpi contrib/dev/acpica/utilities/utxface.c optional acpi contrib/dev/acpica/utilities/utxferror.c optional acpi contrib/ipfilter/netinet/fil.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_auth.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_fil_freebsd.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_frag.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_log.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_nat.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_proxy.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_state.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_lookup.c optional ipfilter inet \ compile-with "${NORMAL_C} -Wno-error -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_pool.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_htable.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_sync.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/ipfilter/netinet/mlfk_ipl.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/libfdt/fdt.c optional fdt contrib/libfdt/fdt_ro.c optional fdt contrib/libfdt/fdt_rw.c optional fdt contrib/libfdt/fdt_strerror.c optional fdt contrib/libfdt/fdt_sw.c optional fdt contrib/libfdt/fdt_wip.c optional fdt contrib/ngatm/netnatm/api/cc_conn.c optional ngatm_ccatm \ compile-with "${NORMAL_C_NOWERROR} -I$S/contrib/ngatm" contrib/ngatm/netnatm/api/cc_data.c optional ngatm_ccatm \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/api/cc_dump.c optional ngatm_ccatm \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/api/cc_port.c optional ngatm_ccatm \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/api/cc_sig.c optional ngatm_ccatm \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/api/cc_user.c optional ngatm_ccatm \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/api/unisap.c optional ngatm_ccatm \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/misc/straddr.c optional ngatm_atmbase \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/misc/unimsg_common.c optional ngatm_atmbase \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/msg/traffic.c optional ngatm_atmbase \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/msg/uni_ie.c optional ngatm_atmbase \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/msg/uni_msg.c optional ngatm_atmbase \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/saal/saal_sscfu.c optional ngatm_sscfu \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/saal/saal_sscop.c optional ngatm_sscop \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/sig/sig_call.c optional ngatm_uni \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/sig/sig_coord.c optional ngatm_uni \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/sig/sig_party.c optional ngatm_uni \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/sig/sig_print.c optional ngatm_uni \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/sig/sig_reset.c optional ngatm_uni \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/sig/sig_uni.c optional ngatm_uni \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/sig/sig_unimsgcpy.c optional ngatm_uni \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/ngatm/netnatm/sig/sig_verify.c optional ngatm_uni \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" contrib/pf/net/if_pflog.c optional pflog pf inet \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/pf/net/if_pfsync.c optional pfsync pf inet \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/pf/net/pf.c optional pf inet \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/pf/net/pf_if.c optional pf inet \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/pf/net/pf_ioctl.c optional pf inet \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/pf/net/pf_norm.c optional pf inet \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/pf/net/pf_osfp.c optional pf inet \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/pf/net/pf_ruleset.c optional pf inet \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/pf/net/pf_subr.c optional pf inet \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/pf/net/pf_table.c optional pf inet \ compile-with "${NORMAL_C} -I$S/contrib/pf" contrib/pf/netinet/in4_cksum.c optional pf inet crypto/blowfish/bf_ecb.c optional ipsec crypto/blowfish/bf_skey.c optional crypto | ipsec crypto/camellia/camellia.c optional crypto | ipsec crypto/camellia/camellia-api.c optional crypto | ipsec crypto/des/des_ecb.c optional crypto | ipsec | netsmb crypto/des/des_setkey.c optional crypto | ipsec | netsmb crypto/rc4/rc4.c optional netgraph_mppc_encryption | kgssapi crypto/rijndael/rijndael-alg-fst.c optional crypto | geom_bde | \ ipsec | random | wlan_ccmp crypto/rijndael/rijndael-api-fst.c optional geom_bde | random crypto/rijndael/rijndael-api.c optional crypto | ipsec | wlan_ccmp crypto/sha1.c optional carp | crypto | ipsec | \ netgraph_mppc_encryption | sctp crypto/sha2/sha2.c optional crypto | geom_bde | ipsec | random | \ sctp ddb/db_access.c optional ddb ddb/db_break.c optional ddb ddb/db_capture.c optional ddb ddb/db_command.c optional ddb ddb/db_examine.c optional ddb ddb/db_expr.c optional ddb ddb/db_input.c optional ddb ddb/db_lex.c optional ddb ddb/db_main.c optional ddb ddb/db_output.c optional ddb ddb/db_print.c optional ddb ddb/db_ps.c optional ddb ddb/db_run.c optional ddb ddb/db_script.c optional ddb ddb/db_sym.c optional ddb ddb/db_thread.c optional ddb ddb/db_textdump.c optional ddb ddb/db_variables.c optional ddb ddb/db_watch.c optional ddb ddb/db_write_cmd.c optional ddb #dev/dpt/dpt_control.c optional dpt dev/aac/aac.c optional aac dev/aac/aac_cam.c optional aacp aac dev/aac/aac_debug.c optional aac dev/aac/aac_disk.c optional aac dev/aac/aac_linux.c optional aac compat_linux dev/aac/aac_pci.c optional aac pci dev/acpi_support/acpi_wmi.c optional acpi_wmi acpi dev/acpi_support/acpi_asus.c optional acpi_asus acpi dev/acpi_support/acpi_fujitsu.c optional acpi_fujitsu acpi dev/acpi_support/acpi_hp.c optional acpi_hp acpi dev/acpi_support/acpi_ibm.c optional acpi_ibm acpi dev/acpi_support/acpi_panasonic.c optional acpi_panasonic acpi dev/acpi_support/acpi_sony.c optional acpi_sony acpi dev/acpi_support/acpi_toshiba.c optional acpi_toshiba acpi dev/acpi_support/atk0110.c optional aibs acpi dev/acpica/Osd/OsdDebug.c optional acpi dev/acpica/Osd/OsdHardware.c optional acpi dev/acpica/Osd/OsdInterrupt.c optional acpi dev/acpica/Osd/OsdMemory.c optional acpi dev/acpica/Osd/OsdSchedule.c optional acpi dev/acpica/Osd/OsdStream.c optional acpi dev/acpica/Osd/OsdSynch.c optional acpi dev/acpica/Osd/OsdTable.c optional acpi dev/acpica/acpi.c optional acpi dev/acpica/acpi_acad.c optional acpi dev/acpica/acpi_battery.c optional acpi dev/acpica/acpi_button.c optional acpi dev/acpica/acpi_cmbat.c optional acpi dev/acpica/acpi_cpu.c optional acpi dev/acpica/acpi_ec.c optional acpi dev/acpica/acpi_hpet.c optional acpi dev/acpica/acpi_isab.c optional acpi isa dev/acpica/acpi_lid.c optional acpi dev/acpica/acpi_package.c optional acpi dev/acpica/acpi_pci.c optional acpi pci dev/acpica/acpi_pci_link.c optional acpi pci dev/acpica/acpi_pcib.c optional acpi pci dev/acpica/acpi_pcib_acpi.c optional acpi pci dev/acpica/acpi_pcib_pci.c optional acpi pci dev/acpica/acpi_perf.c optional acpi dev/acpica/acpi_powerres.c optional acpi dev/acpica/acpi_quirk.c optional acpi dev/acpica/acpi_resource.c optional acpi dev/acpica/acpi_smbat.c optional acpi dev/acpica/acpi_thermal.c optional acpi dev/acpica/acpi_throttle.c optional acpi dev/acpica/acpi_timer.c optional acpi dev/acpica/acpi_video.c optional acpi_video acpi dev/acpica/acpi_dock.c optional acpi_dock acpi dev/adlink/adlink.c optional adlink dev/advansys/adv_eisa.c optional adv eisa dev/advansys/adv_pci.c optional adv pci dev/advansys/advansys.c optional adv dev/advansys/advlib.c optional adv dev/advansys/advmcode.c optional adv dev/advansys/adw_pci.c optional adw pci dev/advansys/adwcam.c optional adw dev/advansys/adwlib.c optional adw dev/advansys/adwmcode.c optional adw dev/ae/if_ae.c optional ae pci dev/age/if_age.c optional age pci dev/agp/agp.c optional agp pci dev/agp/agp_if.m optional agp pci dev/aha/aha.c optional aha dev/aha/aha_isa.c optional aha isa dev/aha/aha_mca.c optional aha mca dev/ahb/ahb.c optional ahb eisa dev/ahci/ahci.c optional ahci pci dev/aic/aic.c optional aic dev/aic/aic_pccard.c optional aic pccard dev/aic7xxx/ahc_eisa.c optional ahc eisa dev/aic7xxx/ahc_isa.c optional ahc isa dev/aic7xxx/ahc_pci.c optional ahc pci dev/aic7xxx/ahd_pci.c optional ahd pci dev/aic7xxx/aic7770.c optional ahc dev/aic7xxx/aic79xx.c optional ahd pci dev/aic7xxx/aic79xx_osm.c optional ahd pci dev/aic7xxx/aic79xx_pci.c optional ahd pci dev/aic7xxx/aic7xxx.c optional ahc dev/aic7xxx/aic7xxx_93cx6.c optional ahc dev/aic7xxx/aic7xxx_osm.c optional ahc dev/aic7xxx/aic7xxx_pci.c optional ahc pci dev/alc/if_alc.c optional alc pci dev/ale/if_ale.c optional ale pci dev/amd/amd.c optional amd dev/amr/amr.c optional amr dev/amr/amr_cam.c optional amrp amr dev/amr/amr_disk.c optional amr dev/amr/amr_linux.c optional amr compat_linux dev/amr/amr_pci.c optional amr pci dev/an/if_an.c optional an dev/an/if_an_isa.c optional an isa dev/an/if_an_pccard.c optional an pccard dev/an/if_an_pci.c optional an pci dev/asr/asr.c optional asr pci # dev/ata/ata_if.m optional ata | atacore dev/ata/ata-all.c optional ata | atacore dev/ata/ata-dma.c optional ata | atacore dev/ata/ata-lowlevel.c optional ata | atacore dev/ata/ata-queue.c optional ata | atacore dev/ata/ata-sata.c optional ata | atacore dev/ata/ata-card.c optional ata pccard | atapccard dev/ata/ata-cbus.c optional ata pc98 | atapc98 dev/ata/ata-isa.c optional ata isa | ataisa dev/ata/ata-pci.c optional ata pci | atapci dev/ata/chipsets/ata-ahci.c optional ata pci | ataahci | ataacerlabs | \ ataati | ataintel | atajmicron | \ atavia | atanvidia dev/ata/chipsets/ata-acard.c optional ata pci | ataacard dev/ata/chipsets/ata-acerlabs.c optional ata pci | ataacerlabs dev/ata/chipsets/ata-adaptec.c optional ata pci | ataadaptec dev/ata/chipsets/ata-amd.c optional ata pci | ataamd dev/ata/chipsets/ata-ati.c optional ata pci | ataati dev/ata/chipsets/ata-cenatek.c optional ata pci | atacenatek dev/ata/chipsets/ata-cypress.c optional ata pci | atacypress dev/ata/chipsets/ata-cyrix.c optional ata pci | atacyrix dev/ata/chipsets/ata-highpoint.c optional ata pci | atahighpoint dev/ata/chipsets/ata-intel.c optional ata pci | ataintel dev/ata/chipsets/ata-ite.c optional ata pci | ataite dev/ata/chipsets/ata-jmicron.c optional ata pci | atajmicron dev/ata/chipsets/ata-marvell.c optional ata pci | atamarvell | ataadaptec dev/ata/chipsets/ata-micron.c optional ata pci | atamicron dev/ata/chipsets/ata-national.c optional ata pci | atanational dev/ata/chipsets/ata-netcell.c optional ata pci | atanetcell dev/ata/chipsets/ata-nvidia.c optional ata pci | atanvidia dev/ata/chipsets/ata-promise.c optional ata pci | atapromise dev/ata/chipsets/ata-serverworks.c optional ata pci | ataserverworks dev/ata/chipsets/ata-siliconimage.c optional ata pci | atasiliconimage | ataati dev/ata/chipsets/ata-sis.c optional ata pci | atasis dev/ata/chipsets/ata-via.c optional ata pci | atavia dev/ata/ata-disk.c optional atadisk dev/ata/ata-raid.c optional ataraid dev/ata/atapi-cd.c optional atapicd dev/ata/atapi-fd.c optional atapifd dev/ata/atapi-tape.c optional atapist dev/ata/atapi-cam.c optional atapicam # dev/ath/if_ath_pci.c optional ath_pci pci \ compile-with "${NORMAL_C} -I$S/dev/ath" # dev/ath/if_ath_ahb.c optional ath_ahb \ compile-with "${NORMAL_C} -I$S/dev/ath" # dev/ath/if_ath.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_debug.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_keycache.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_tx.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_tx_ht.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_sysctl.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/ah_osdep.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" # dev/ath/ath_hal/ah.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/ath_hal/ah_eeprom_v1.c optional ath_hal | ath_ar5210 \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/ath_hal/ah_eeprom_v3.c optional ath_hal | ath_ar5211 | ath_ar5212 \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/ath_hal/ah_eeprom_v14.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/ath_hal/ah_eeprom_v4k.c \ optional ath_hal | ath_ar9285 \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/ath_hal/ah_eeprom_9287.c \ optional ath_hal | ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/ath_hal/ah_regdomain.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" # ar5210 dev/ath/ath_hal/ar5210/ar5210_attach.c optional ath_hal | ath_ar5210 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5210/ar5210_beacon.c optional ath_hal | ath_ar5210 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5210/ar5210_interrupts.c optional ath_hal | ath_ar5210 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5210/ar5210_keycache.c optional ath_hal | ath_ar5210 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5210/ar5210_misc.c optional ath_hal | ath_ar5210 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5210/ar5210_phy.c optional ath_hal | ath_ar5210 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5210/ar5210_power.c optional ath_hal | ath_ar5210 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5210/ar5210_recv.c optional ath_hal | ath_ar5210 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5210/ar5210_reset.c optional ath_hal | ath_ar5210 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5210/ar5210_xmit.c optional ath_hal | ath_ar5210 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" # ar5211 dev/ath/ath_hal/ar5211/ar5211_attach.c optional ath_hal | ath_ar5211 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5211/ar5211_beacon.c optional ath_hal | ath_ar5211 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5211/ar5211_interrupts.c optional ath_hal | ath_ar5211 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5211/ar5211_keycache.c optional ath_hal | ath_ar5211 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5211/ar5211_misc.c optional ath_hal | ath_ar5211 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5211/ar5211_phy.c optional ath_hal | ath_ar5211 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5211/ar5211_power.c optional ath_hal | ath_ar5211 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5211/ar5211_recv.c optional ath_hal | ath_ar5211 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5211/ar5211_reset.c optional ath_hal | ath_ar5211 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5211/ar5211_xmit.c optional ath_hal | ath_ar5211 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" # ar5212 dev/ath/ath_hal/ar5212/ar5212_ani.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5212_attach.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5212_beacon.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5212_eeprom.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5212_gpio.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5212_interrupts.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5212_keycache.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5212_misc.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5212_phy.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5212_power.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5212_recv.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5212_reset.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5212_rfgain.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5212_xmit.c \ optional ath_hal | ath_ar5212 | ath_ar5416 | ath_ar9160 | ath_ar9280 | \ ath_ar9285 ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" # ar5416 (depends on ar5212) dev/ath/ath_hal/ar5416/ar5416_ani.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_attach.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_beacon.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_cal.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_cal_iq.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_cal_adcgain.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_cal_adcdc.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_eeprom.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_gpio.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_interrupts.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_keycache.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_misc.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_phy.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_power.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_recv.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_reset.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_xmit.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" # ar9130 (depends upon ar5416) - also requires AH_SUPPORT_AR9130 dev/ath/ath_hal/ar9001/ar9130_attach.c optional ath_hal | ath_ar9130 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9001/ar9130_phy.c optional ath_hal | ath_ar9130 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9001/ar9130_eeprom.c optional ath_hal | ath_ar9130 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" # ar9160 (depends on ar5416) dev/ath/ath_hal/ar9001/ar9160_attach.c optional ath_hal | ath_ar9160 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" # ar9280 (depends on ar5416) dev/ath/ath_hal/ar9002/ar9280_attach.c optional ath_hal | ath_ar9280 | \ ath_ar9285 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9002/ar9280_olc.c optional ath_hal | ath_ar9280 | \ ath_ar9285 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" # ar9285 (depends on ar5416 and ar9280) dev/ath/ath_hal/ar9002/ar9285_attach.c optional ath_hal | ath_ar9285 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9002/ar9285_reset.c optional ath_hal | ath_ar9285 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9002/ar9285_cal.c optional ath_hal | ath_ar9285 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9002/ar9285_phy.c optional ath_hal | ath_ar9285 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9002/ar9285_diversity.c optional ath_hal | ath_ar9285 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" # ar9287 (depends on ar5416) dev/ath/ath_hal/ar9002/ar9287_attach.c optional ath_hal | ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9002/ar9287_reset.c optional ath_hal | ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9002/ar9287_cal.c optional ath_hal | ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9002/ar9287_olc.c optional ath_hal | ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" # rf backends dev/ath/ath_hal/ar5212/ar2316.c optional ath_rf2316 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar2317.c optional ath_rf2317 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar2413.c optional ath_hal | ath_rf2413 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar2425.c optional ath_hal | ath_rf2425 | ath_rf2417 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5111.c optional ath_hal | ath_rf5111 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5112.c optional ath_hal | ath_rf5112 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5212/ar5413.c optional ath_hal | ath_rf5413 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar2133.c optional ath_hal | ath_ar5416 | ath_ar9160 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9002/ar9280.c optional ath_hal | ath_ar9280 | ath_ar9285 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9002/ar9285.c optional ath_hal | ath_ar9285 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9002/ar9287.c optional ath_hal | ath_ar9287 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" # ath rate control algorithms dev/ath/ath_rate/amrr/amrr.c optional ath_rate_amrr \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/ath_rate/onoe/onoe.c optional ath_rate_onoe \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/ath_rate/sample/sample.c optional ath_rate_sample \ compile-with "${NORMAL_C} -I$S/dev/ath" # ath DFS modules dev/ath/ath_dfs/null/dfs_null.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" # dev/bce/if_bce.c optional bce dev/bfe/if_bfe.c optional bfe dev/bge/if_bge.c optional bge dev/bktr/bktr_audio.c optional bktr pci dev/bktr/bktr_card.c optional bktr pci dev/bktr/bktr_core.c optional bktr pci dev/bktr/bktr_i2c.c optional bktr pci smbus dev/bktr/bktr_os.c optional bktr pci dev/bktr/bktr_tuner.c optional bktr pci dev/bktr/msp34xx.c optional bktr pci dev/buslogic/bt.c optional bt dev/buslogic/bt_eisa.c optional bt eisa dev/buslogic/bt_isa.c optional bt isa dev/buslogic/bt_mca.c optional bt mca dev/buslogic/bt_pci.c optional bt pci dev/bwi/bwimac.c optional bwi dev/bwi/bwiphy.c optional bwi dev/bwi/bwirf.c optional bwi dev/bwi/if_bwi.c optional bwi dev/bwi/if_bwi_pci.c optional bwi pci dev/bwn/if_bwn.c optional bwn siba_bwn dev/bxe/if_bxe.c optional bxe dev/bxe/bxe_link.c optional bxe dev/cardbus/cardbus.c optional cardbus dev/cardbus/cardbus_cis.c optional cardbus dev/cardbus/cardbus_device.c optional cardbus dev/cas/if_cas.c optional cas dev/cfi/cfi_core.c optional cfi dev/cfi/cfi_dev.c optional cfi dev/cfi/cfi_disk.c optional cfid dev/ciss/ciss.c optional ciss dev/cm/smc90cx6.c optional cm dev/cmx/cmx.c optional cmx dev/cmx/cmx_pccard.c optional cmx pccard dev/cpufreq/ichss.c optional cpufreq dev/cs/if_cs.c optional cs dev/cs/if_cs_isa.c optional cs isa dev/cs/if_cs_pccard.c optional cs pccard dev/cxgb/cxgb_main.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/cxgb_offload.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/cxgb_sge.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/common/cxgb_mc5.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/common/cxgb_vsc7323.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/common/cxgb_vsc8211.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/common/cxgb_ael1002.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/common/cxgb_aq100x.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/common/cxgb_mv88e1xxx.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/common/cxgb_xgmac.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/common/cxgb_t3_hw.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/common/cxgb_tn1010.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/sys/uipc_mvec.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/cxgb_t3fw.c optional cxgb cxgb_t3fw \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgbe/t4_main.c optional cxgbe pci \ compile-with "${NORMAL_C} -I$S/dev/cxgbe" dev/cxgbe/t4_sge.c optional cxgbe pci \ compile-with "${NORMAL_C} -I$S/dev/cxgbe" dev/cxgbe/t4_l2t.c optional cxgbe pci \ compile-with "${NORMAL_C} -I$S/dev/cxgbe" dev/cxgbe/common/t4_hw.c optional cxgbe pci \ compile-with "${NORMAL_C} -I$S/dev/cxgbe" dev/cy/cy.c optional cy dev/cy/cy_isa.c optional cy isa dev/cy/cy_pci.c optional cy pci dev/dc/if_dc.c optional dc pci dev/dc/dcphy.c optional dc pci dev/dc/pnphy.c optional dc pci dev/dcons/dcons.c optional dcons dev/dcons/dcons_crom.c optional dcons_crom dev/dcons/dcons_os.c optional dcons dev/de/if_de.c optional de pci dev/digi/CX.c optional digi_CX dev/digi/CX_PCI.c optional digi_CX_PCI dev/digi/EPCX.c optional digi_EPCX dev/digi/EPCX_PCI.c optional digi_EPCX_PCI dev/digi/Xe.c optional digi_Xe dev/digi/Xem.c optional digi_Xem dev/digi/Xr.c optional digi_Xr dev/digi/digi.c optional digi dev/digi/digi_isa.c optional digi isa dev/digi/digi_pci.c optional digi pci dev/dpt/dpt_eisa.c optional dpt eisa dev/dpt/dpt_pci.c optional dpt pci dev/dpt/dpt_scsi.c optional dpt dev/drm/ati_pcigart.c optional drm dev/drm/drm_agpsupport.c optional drm dev/drm/drm_auth.c optional drm dev/drm/drm_bufs.c optional drm dev/drm/drm_context.c optional drm dev/drm/drm_dma.c optional drm dev/drm/drm_drawable.c optional drm dev/drm/drm_drv.c optional drm dev/drm/drm_fops.c optional drm dev/drm/drm_hashtab.c optional drm dev/drm/drm_ioctl.c optional drm dev/drm/drm_irq.c optional drm dev/drm/drm_lock.c optional drm dev/drm/drm_memory.c optional drm dev/drm/drm_mm.c optional drm dev/drm/drm_pci.c optional drm dev/drm/drm_scatter.c optional drm dev/drm/drm_sman.c optional drm dev/drm/drm_sysctl.c optional drm dev/drm/drm_vm.c optional drm dev/drm/i915_dma.c optional i915drm dev/drm/i915_drv.c optional i915drm dev/drm/i915_irq.c optional i915drm dev/drm/i915_mem.c optional i915drm dev/drm/i915_suspend.c optional i915drm dev/drm/mach64_dma.c optional mach64drm dev/drm/mach64_drv.c optional mach64drm dev/drm/mach64_irq.c optional mach64drm dev/drm/mach64_state.c optional mach64drm dev/drm/mga_dma.c optional mgadrm dev/drm/mga_drv.c optional mgadrm dev/drm/mga_irq.c optional mgadrm dev/drm/mga_state.c optional mgadrm \ compile-with "${NORMAL_C} -finline-limit=13500" dev/drm/mga_warp.c optional mgadrm dev/drm/r128_cce.c optional r128drm dev/drm/r128_drv.c optional r128drm dev/drm/r128_irq.c optional r128drm dev/drm/r128_state.c optional r128drm \ compile-with "${NORMAL_C} -finline-limit=13500" dev/drm/r300_cmdbuf.c optional radeondrm dev/drm/r600_blit.c optional radeondrm dev/drm/r600_cp.c optional radeondrm dev/drm/radeon_cp.c optional radeondrm dev/drm/radeon_cs.c optional radeondrm dev/drm/radeon_drv.c optional radeondrm dev/drm/radeon_irq.c optional radeondrm dev/drm/radeon_mem.c optional radeondrm dev/drm/radeon_state.c optional radeondrm dev/drm/savage_bci.c optional savagedrm dev/drm/savage_drv.c optional savagedrm dev/drm/savage_state.c optional savagedrm dev/drm/sis_drv.c optional sisdrm dev/drm/sis_ds.c optional sisdrm dev/drm/sis_mm.c optional sisdrm dev/drm/tdfx_drv.c optional tdfxdrm dev/drm/via_dma.c optional viadrm dev/drm/via_dmablit.c optional viadrm dev/drm/via_drv.c optional viadrm dev/drm/via_irq.c optional viadrm dev/drm/via_map.c optional viadrm dev/drm/via_mm.c optional viadrm dev/drm/via_verifier.c optional viadrm dev/drm/via_video.c optional viadrm dev/ed/if_ed.c optional ed dev/ed/if_ed_novell.c optional ed dev/ed/if_ed_rtl80x9.c optional ed dev/ed/if_ed_pccard.c optional ed pccard dev/ed/if_ed_pci.c optional ed pci dev/eisa/eisa_if.m standard dev/eisa/eisaconf.c optional eisa dev/e1000/if_em.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/if_lem.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/if_igb.c optional igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_80003es2lan.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_82540.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_82541.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_82542.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_82543.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_82571.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_82575.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_ich8lan.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_api.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_mac.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_manage.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_nvm.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_phy.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_vf.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_mbx.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/e1000/e1000_osdep.c optional em | igb \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/et/if_et.c optional et dev/en/if_en_pci.c optional en pci dev/en/midway.c optional en dev/ep/if_ep.c optional ep dev/ep/if_ep_eisa.c optional ep eisa dev/ep/if_ep_isa.c optional ep isa dev/ep/if_ep_mca.c optional ep mca dev/ep/if_ep_pccard.c optional ep pccard dev/esp/ncr53c9x.c optional esp dev/ex/if_ex.c optional ex dev/ex/if_ex_isa.c optional ex isa dev/ex/if_ex_pccard.c optional ex pccard dev/exca/exca.c optional cbb dev/fatm/if_fatm.c optional fatm pci dev/fb/splash.c optional splash dev/fdt/fdt_common.c optional fdt dev/fdt/fdt_pci.c optional fdt pci dev/fdt/fdt_static_dtb.S optional fdt fdt_dtb_static dev/fdt/fdtbus.c optional fdt dev/fdt/simplebus.c optional fdt dev/fe/if_fe.c optional fe dev/fe/if_fe_pccard.c optional fe pccard dev/firewire/firewire.c optional firewire dev/firewire/fwcrom.c optional firewire dev/firewire/fwdev.c optional firewire dev/firewire/fwdma.c optional firewire dev/firewire/fwmem.c optional firewire dev/firewire/fwohci.c optional firewire dev/firewire/fwohci_pci.c optional firewire pci dev/firewire/if_fwe.c optional fwe dev/firewire/if_fwip.c optional fwip dev/firewire/sbp.c optional sbp dev/firewire/sbp_targ.c optional sbp_targ dev/flash/at45d.c optional at45d dev/flash/mx25l.c optional mx25l dev/fxp/if_fxp.c optional fxp dev/gem/if_gem.c optional gem dev/gem/if_gem_pci.c optional gem pci dev/gem/if_gem_sbus.c optional gem sbus dev/gpio/gpiobus.c optional gpio \ dependency "gpiobus_if.h" dev/gpio/gpioc.c optional gpio \ dependency "gpio_if.h" dev/gpio/gpioiic.c optional gpioiic dev/gpio/gpioled.c optional gpioled dev/gpio/gpio_if.m optional gpio dev/gpio/gpiobus_if.m optional gpio dev/hatm/if_hatm.c optional hatm pci dev/hatm/if_hatm_intr.c optional hatm pci dev/hatm/if_hatm_ioctl.c optional hatm pci dev/hatm/if_hatm_rx.c optional hatm pci dev/hatm/if_hatm_tx.c optional hatm pci dev/hifn/hifn7751.c optional hifn dev/hme/if_hme.c optional hme dev/hme/if_hme_pci.c optional hme pci dev/hme/if_hme_sbus.c optional hme sbus dev/hptiop/hptiop.c optional hptiop scbus dev/hwpmc/hwpmc_logging.c optional hwpmc dev/hwpmc/hwpmc_mod.c optional hwpmc dev/ichsmb/ichsmb.c optional ichsmb dev/ichsmb/ichsmb_pci.c optional ichsmb pci dev/ida/ida.c optional ida dev/ida/ida_disk.c optional ida dev/ida/ida_eisa.c optional ida eisa dev/ida/ida_pci.c optional ida pci dev/ie/if_ie.c optional ie isa nowerror dev/ie/if_ie_isa.c optional ie isa dev/ieee488/ibfoo.c optional pcii | tnt4882 dev/ieee488/pcii.c optional pcii dev/ieee488/tnt4882.c optional tnt4882 dev/ieee488/upd7210.c optional pcii | tnt4882 dev/iicbus/ad7418.c optional ad7418 dev/iicbus/ds133x.c optional ds133x dev/iicbus/ds1672.c optional ds1672 dev/iicbus/icee.c optional icee dev/iicbus/if_ic.c optional ic dev/iicbus/iic.c optional iic dev/iicbus/iicbb.c optional iicbb dev/iicbus/iicbb_if.m optional iicbb dev/iicbus/iicbus.c optional iicbus dev/iicbus/iicbus_if.m optional iicbus dev/iicbus/iiconf.c optional iicbus dev/iicbus/iicsmb.c optional iicsmb \ dependency "iicbus_if.h" dev/iir/iir.c optional iir dev/iir/iir_ctrl.c optional iir dev/iir/iir_pci.c optional iir pci dev/ips/ips.c optional ips dev/ips/ips_commands.c optional ips dev/ips/ips_disk.c optional ips dev/ips/ips_ioctl.c optional ips dev/ips/ips_pci.c optional ips pci dev/ipw/if_ipw.c optional ipw ipwbssfw.c optional ipwbssfw | ipwfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk ipw_bss.fw:ipw_bss:130 -lintel_ipw -mipw_bss -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "ipwbssfw.c" ipw_bss.fwo optional ipwbssfw | ipwfw \ dependency "ipw_bss.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} ipw_bss.fw" \ no-implicit-rule \ clean "ipw_bss.fwo" ipw_bss.fw optional ipwbssfw | ipwfw \ dependency "$S/contrib/dev/ipw/ipw2100-1.3.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/ipw/ipw2100-1.3.fw.uu" \ no-obj no-implicit-rule \ clean "ipw_bss.fw" ipwibssfw.c optional ipwibssfw | ipwfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk ipw_ibss.fw:ipw_ibss:130 -lintel_ipw -mipw_ibss -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "ipwibssfw.c" ipw_ibss.fwo optional ipwibssfw | ipwfw \ dependency "ipw_ibss.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} ipw_ibss.fw" \ no-implicit-rule \ clean "ipw_ibss.fwo" ipw_ibss.fw optional ipwibssfw | ipwfw \ dependency "$S/contrib/dev/ipw/ipw2100-1.3-i.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/ipw/ipw2100-1.3-i.fw.uu" \ no-obj no-implicit-rule \ clean "ipw_ibss.fw" ipwmonitorfw.c optional ipwmonitorfw | ipwfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk ipw_monitor.fw:ipw_monitor:130 -lintel_ipw -mipw_monitor -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "ipwmonitorfw.c" ipw_monitor.fwo optional ipwmonitorfw | ipwfw \ dependency "ipw_monitor.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} ipw_monitor.fw" \ no-implicit-rule \ clean "ipw_monitor.fwo" ipw_monitor.fw optional ipwmonitorfw | ipwfw \ dependency "$S/contrib/dev/ipw/ipw2100-1.3-p.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/ipw/ipw2100-1.3-p.fw.uu" \ no-obj no-implicit-rule \ clean "ipw_monitor.fw" dev/iscsi/initiator/iscsi.c optional iscsi_initiator scbus dev/iscsi/initiator/iscsi_subr.c optional iscsi_initiator scbus dev/iscsi/initiator/isc_cam.c optional iscsi_initiator scbus dev/iscsi/initiator/isc_soc.c optional iscsi_initiator scbus dev/iscsi/initiator/isc_sm.c optional iscsi_initiator scbus dev/iscsi/initiator/isc_subr.c optional iscsi_initiator scbus dev/isp/isp.c optional isp dev/isp/isp_freebsd.c optional isp dev/isp/isp_library.c optional isp dev/isp/isp_pci.c optional isp pci dev/isp/isp_sbus.c optional isp sbus dev/isp/isp_target.c optional isp dev/ispfw/ispfw.c optional ispfw dev/iwi/if_iwi.c optional iwi iwibssfw.c optional iwibssfw | iwifw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwi_bss.fw:iwi_bss:300 -lintel_iwi -miwi_bss -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwibssfw.c" iwi_bss.fwo optional iwibssfw | iwifw \ dependency "iwi_bss.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} iwi_bss.fw" \ no-implicit-rule \ clean "iwi_bss.fwo" iwi_bss.fw optional iwibssfw | iwifw \ dependency "$S/contrib/dev/iwi/ipw2200-bss.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/iwi/ipw2200-bss.fw.uu" \ no-obj no-implicit-rule \ clean "iwi_bss.fw" iwiibssfw.c optional iwiibssfw | iwifw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwi_ibss.fw:iwi_ibss:300 -lintel_iwi -miwi_ibss -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwiibssfw.c" iwi_ibss.fwo optional iwiibssfw | iwifw \ dependency "iwi_ibss.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} iwi_ibss.fw" \ no-implicit-rule \ clean "iwi_ibss.fwo" iwi_ibss.fw optional iwiibssfw | iwifw \ dependency "$S/contrib/dev/iwi/ipw2200-ibss.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/iwi/ipw2200-ibss.fw.uu" \ no-obj no-implicit-rule \ clean "iwi_ibss.fw" iwimonitorfw.c optional iwimonitorfw | iwifw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwi_monitor.fw:iwi_monitor:300 -lintel_iwi -miwi_monitor -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwimonitorfw.c" iwi_monitor.fwo optional iwimonitorfw | iwifw \ dependency "iwi_monitor.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} iwi_monitor.fw" \ no-implicit-rule \ clean "iwi_monitor.fwo" iwi_monitor.fw optional iwimonitorfw | iwifw \ dependency "$S/contrib/dev/iwi/ipw2200-sniffer.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/iwi/ipw2200-sniffer.fw.uu" \ no-obj no-implicit-rule \ clean "iwi_monitor.fw" dev/iwn/if_iwn.c optional iwn iwn1000fw.c optional iwn1000fw | iwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwn1000.fw:iwn1000fw -miwn1000fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwn1000fw.c" iwn1000fw.fwo optional iwn1000fw | iwnfw \ dependency "iwn1000.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} iwn1000.fw" \ no-implicit-rule \ clean "iwn1000fw.fwo" iwn1000.fw optional iwn1000fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-1000-39.31.5.1.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/iwn/iwlwifi-1000-39.31.5.1.fw.uu" \ no-obj no-implicit-rule \ clean "iwn1000.fw" iwn4965fw.c optional iwn4965fw | iwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwn4965.fw:iwn4965fw -miwn4965fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwn4965fw.c" iwn4965fw.fwo optional iwn4965fw | iwnfw \ dependency "iwn4965.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} iwn4965.fw" \ no-implicit-rule \ clean "iwn4965fw.fwo" iwn4965.fw optional iwn4965fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-4965-228.61.2.24.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/iwn/iwlwifi-4965-228.61.2.24.fw.uu" \ no-obj no-implicit-rule \ clean "iwn4965.fw" iwn5000fw.c optional iwn5000fw | iwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwn5000.fw:iwn5000fw -miwn5000fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwn5000fw.c" iwn5000fw.fwo optional iwn5000fw | iwnfw \ dependency "iwn5000.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} iwn5000.fw" \ no-implicit-rule \ clean "iwn5000fw.fwo" iwn5000.fw optional iwn5000fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-5000-8.83.5.1.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/iwn/iwlwifi-5000-8.83.5.1.fw.uu" \ no-obj no-implicit-rule \ clean "iwn5000.fw" iwn5150fw.c optional iwn5150fw | iwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwn5150.fw:iwn5150fw -miwn5150fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwn5150fw.c" iwn5150fw.fwo optional iwn5150fw | iwnfw \ dependency "iwn5150.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} iwn5150.fw" \ no-implicit-rule \ clean "iwn5150fw.fwo" iwn5150.fw optional iwn5150fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-5150-8.24.2.2.fw.uu"\ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/iwn/iwlwifi-5150-8.24.2.2.fw.uu" \ no-obj no-implicit-rule \ clean "iwn5150.fw" iwn6000fw.c optional iwn6000fw | iwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwn6000.fw:iwn6000fw -miwn6000fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwn6000fw.c" iwn6000fw.fwo optional iwn6000fw | iwnfw \ dependency "iwn6000.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} iwn6000.fw" \ no-implicit-rule \ clean "iwn6000fw.fwo" iwn6000.fw optional iwn6000fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-6000-9.221.4.1.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/iwn/iwlwifi-6000-9.221.4.1.fw.uu" \ no-obj no-implicit-rule \ clean "iwn6000.fw" iwn6000g2afw.c optional iwn6000g2afw | iwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwn6000g2a.fw:iwn6000g2afw -miwn6000g2afw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwn6000g2afw.c" iwn6000g2afw.fwo optional iwn6000g2afw | iwnfw \ dependency "iwn6000g2a.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} iwn6000g2a.fw" \ no-implicit-rule \ clean "iwn6000g2afw.fwo" iwn6000g2a.fw optional iwn6000g2afw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-6000g2a-17.168.5.2.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/iwn/iwlwifi-6000g2a-17.168.5.2.fw.uu" \ no-obj no-implicit-rule \ clean "iwn6000g2a.fw" iwn6000g2bfw.c optional iwn6000g2bfw | iwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwn6000g2b.fw:iwn6000g2bfw -miwn6000g2bfw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwn6000g2bfw.c" iwn6000g2bfw.fwo optional iwn6000g2bfw | iwnfw \ dependency "iwn6000g2b.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} iwn6000g2b.fw" \ no-implicit-rule \ clean "iwn6000g2bfw.fwo" iwn6000g2b.fw optional iwn6000g2bfw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-6000g2b-17.168.5.2.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/iwn/iwlwifi-6000g2b-17.168.5.2.fw.uu" \ no-obj no-implicit-rule \ clean "iwn6000g2b.fw" iwn6050fw.c optional iwn6050fw | iwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwn6050.fw:iwn6050fw -miwn6050fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwn6050fw.c" iwn6050fw.fwo optional iwn6050fw | iwnfw \ dependency "iwn6050.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} iwn6050.fw" \ no-implicit-rule \ clean "iwn6050fw.fwo" iwn6050.fw optional iwn6050fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-6050-41.28.5.1.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/iwn/iwlwifi-6050-41.28.5.1.fw.uu" \ no-obj no-implicit-rule \ clean "iwn6050.fw" dev/ixgb/if_ixgb.c optional ixgb dev/ixgb/ixgb_ee.c optional ixgb dev/ixgb/ixgb_hw.c optional ixgb dev/ixgbe/ixgbe.c optional ixgbe inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixv.c optional ixgbe inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_phy.c optional ixgbe inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_api.c optional ixgbe inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_common.c optional ixgbe inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_mbx.c optional ixgbe inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_vf.c optional ixgbe inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_82598.c optional ixgbe inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_82599.c optional ixgbe inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/jme/if_jme.c optional jme pci dev/joy/joy.c optional joy dev/joy/joy_isa.c optional joy isa dev/joy/joy_pccard.c optional joy pccard dev/kbdmux/kbdmux.c optional kbdmux dev/ksyms/ksyms.c optional ksyms dev/le/am7990.c optional le dev/le/am79900.c optional le dev/le/if_le_pci.c optional le pci dev/le/lance.c optional le dev/led/led.c standard dev/lge/if_lge.c optional lge dev/lmc/if_lmc.c optional lmc dev/malo/if_malo.c optional malo dev/malo/if_malohal.c optional malo dev/malo/if_malo_pci.c optional malo pci dev/mc146818/mc146818.c optional mc146818 dev/mca/mca_bus.c optional mca dev/mcd/mcd.c optional mcd isa nowerror dev/mcd/mcd_isa.c optional mcd isa nowerror dev/md/md.c optional md dev/mem/memdev.c optional mem dev/mem/memutil.c optional mem dev/mfi/mfi.c optional mfi dev/mfi/mfi_debug.c optional mfi dev/mfi/mfi_pci.c optional mfi pci dev/mfi/mfi_disk.c optional mfi dev/mfi/mfi_linux.c optional mfi compat_linux dev/mfi/mfi_cam.c optional mfip scbus dev/mii/acphy.c optional miibus | acphy dev/mii/amphy.c optional miibus | amphy dev/mii/atphy.c optional miibus | atphy dev/mii/axphy.c optional miibus | axphy dev/mii/bmtphy.c optional miibus | bmtphy dev/mii/brgphy.c optional miibus | brgphy dev/mii/ciphy.c optional miibus | ciphy dev/mii/e1000phy.c optional miibus | e1000phy # XXX only xl cards? dev/mii/exphy.c optional miibus | exphy dev/mii/gentbi.c optional miibus | gentbi dev/mii/icsphy.c optional miibus | icsphy # XXX only fxp cards? dev/mii/inphy.c optional miibus | inphy dev/mii/ip1000phy.c optional miibus | ip1000phy dev/mii/jmphy.c optional miibus | jmphy dev/mii/lxtphy.c optional miibus | lxtphy dev/mii/mii.c optional miibus | mii dev/mii/mii_physubr.c optional miibus | mii dev/mii/miibus_if.m optional miibus | mii dev/mii/mlphy.c optional miibus | mlphy dev/mii/nsgphy.c optional miibus | nsgphy dev/mii/nsphy.c optional miibus | nsphy dev/mii/nsphyter.c optional miibus | nsphyter dev/mii/pnaphy.c optional miibus | pnaphy dev/mii/qsphy.c optional miibus | qsphy dev/mii/rdcphy.c optional miibus | rdcphy dev/mii/rgephy.c optional miibus | rgephy dev/mii/rlphy.c optional miibus | rlphy dev/mii/rlswitch.c optional rlswitch # XXX rue only? dev/mii/ruephy.c optional miibus | ruephy dev/mii/smcphy.c optional miibus | smcphy dev/mii/tdkphy.c optional miibus | tdkphy dev/mii/tlphy.c optional miibus | tlphy dev/mii/truephy.c optional miibus | truephy dev/mii/ukphy.c optional miibus | mii dev/mii/ukphy_subr.c optional miibus | mii dev/mii/xmphy.c optional miibus | xmphy dev/mk48txx/mk48txx.c optional mk48txx dev/mlx/mlx.c optional mlx dev/mlx/mlx_disk.c optional mlx dev/mlx/mlx_pci.c optional mlx pci dev/mly/mly.c optional mly dev/mmc/mmc.c optional mmc dev/mmc/mmcbr_if.m standard dev/mmc/mmcbus_if.m standard dev/mmc/mmcsd.c optional mmcsd dev/mn/if_mn.c optional mn pci dev/mps/mps.c optional mps dev/mps/mps_pci.c optional mps pci dev/mps/mps_sas.c optional mps dev/mps/mps_table.c optional mps dev/mps/mps_user.c optional mps dev/mpt/mpt.c optional mpt dev/mpt/mpt_cam.c optional mpt dev/mpt/mpt_debug.c optional mpt dev/mpt/mpt_pci.c optional mpt pci dev/mpt/mpt_raid.c optional mpt dev/mpt/mpt_user.c optional mpt dev/msk/if_msk.c optional msk dev/mvs/mvs.c optional mvs dev/mvs/mvs_if.m optional mvs dev/mvs/mvs_pci.c optional mvs pci dev/mwl/if_mwl.c optional mwl dev/mwl/if_mwl_pci.c optional mwl pci dev/mwl/mwlhal.c optional mwl mwlfw.c optional mwlfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk mw88W8363.fw:mw88W8363fw mwlboot.fw:mwlboot -mmwl -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "mwlfw.c" mw88W8363.fwo optional mwlfw \ dependency "mw88W8363.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} mw88W8363.fw" \ no-implicit-rule \ clean "mw88W8363.fwo" mw88W8363.fw optional mwlfw \ dependency "$S/contrib/dev/mwl/mw88W8363.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/mwl/mw88W8363.fw.uu" \ no-obj no-implicit-rule \ clean "mw88W8363.fw" mwlboot.fwo optional mwlfw \ dependency "mwlboot.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} mwlboot.fw" \ no-implicit-rule \ clean "mwlboot.fwo" mwlboot.fw optional mwlfw \ dependency "$S/contrib/dev/mwl/mwlboot.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/mwl/mwlboot.fw.uu" \ no-obj no-implicit-rule \ clean "mwlboot.fw" dev/mxge/if_mxge.c optional mxge pci dev/mxge/mxge_lro.c optional mxge pci dev/mxge/mxge_eth_z8e.c optional mxge pci dev/mxge/mxge_ethp_z8e.c optional mxge pci dev/mxge/mxge_rss_eth_z8e.c optional mxge pci dev/mxge/mxge_rss_ethp_z8e.c optional mxge pci dev/my/if_my.c optional my dev/ncv/ncr53c500.c optional ncv dev/ncv/ncr53c500_pccard.c optional ncv pccard dev/nge/if_nge.c optional nge dev/nxge/if_nxge.c optional nxge dev/nxge/xgehal/xgehal-device.c optional nxge dev/nxge/xgehal/xgehal-mm.c optional nxge dev/nxge/xgehal/xge-queue.c optional nxge dev/nxge/xgehal/xgehal-driver.c optional nxge dev/nxge/xgehal/xgehal-ring.c optional nxge dev/nxge/xgehal/xgehal-channel.c optional nxge dev/nxge/xgehal/xgehal-fifo.c optional nxge dev/nxge/xgehal/xgehal-stats.c optional nxge dev/nxge/xgehal/xgehal-config.c optional nxge dev/nxge/xgehal/xgehal-mgmt.c optional nxge dev/nmdm/nmdm.c optional nmdm dev/nsp/nsp.c optional nsp dev/nsp/nsp_pccard.c optional nsp pccard dev/null/null.c standard dev/patm/if_patm.c optional patm pci dev/patm/if_patm_attach.c optional patm pci dev/patm/if_patm_intr.c optional patm pci dev/patm/if_patm_ioctl.c optional patm pci dev/patm/if_patm_rtables.c optional patm pci dev/patm/if_patm_rx.c optional patm pci dev/patm/if_patm_tx.c optional patm pci dev/pbio/pbio.c optional pbio isa dev/pccard/card_if.m standard dev/pccard/pccard.c optional pccard dev/pccard/pccard_cis.c optional pccard dev/pccard/pccard_cis_quirks.c optional pccard dev/pccard/pccard_device.c optional pccard dev/pccard/power_if.m standard dev/pccbb/pccbb.c optional cbb dev/pccbb/pccbb_isa.c optional cbb isa dev/pccbb/pccbb_pci.c optional cbb pci dev/pcf/pcf.c optional pcf dev/pci/eisa_pci.c optional pci eisa dev/pci/fixup_pci.c optional pci dev/pci/hostb_pci.c optional pci dev/pci/ignore_pci.c optional pci dev/pci/isa_pci.c optional pci isa dev/pci/pci.c optional pci dev/pci/pci_if.m standard dev/pci/pci_pci.c optional pci dev/pci/pci_user.c optional pci dev/pci/pcib_if.m standard dev/pci/vga_pci.c optional pci dev/pcn/if_pcn.c optional pcn pci dev/pdq/if_fea.c optional fea eisa dev/pdq/if_fpa.c optional fpa pci dev/pdq/pdq.c optional nowerror fea eisa | fpa pci dev/pdq/pdq_ifsubr.c optional nowerror fea eisa | fpa pci dev/ppbus/if_plip.c optional plip dev/ppbus/immio.c optional vpo dev/ppbus/lpbb.c optional lpbb dev/ppbus/lpt.c optional lpt dev/ppbus/pcfclock.c optional pcfclock dev/ppbus/ppb_1284.c optional ppbus dev/ppbus/ppb_base.c optional ppbus dev/ppbus/ppb_msq.c optional ppbus dev/ppbus/ppbconf.c optional ppbus dev/ppbus/ppbus_if.m optional ppbus dev/ppbus/ppi.c optional ppi dev/ppbus/pps.c optional pps dev/ppbus/vpo.c optional vpo dev/ppbus/vpoio.c optional vpo dev/ppc/ppc.c optional ppc dev/ppc/ppc_acpi.c optional ppc acpi dev/ppc/ppc_isa.c optional ppc isa dev/ppc/ppc_pci.c optional ppc pci dev/ppc/ppc_puc.c optional ppc puc dev/pst/pst-iop.c optional pst dev/pst/pst-pci.c optional pst pci dev/pst/pst-raid.c optional pst dev/pty/pty.c optional pty dev/puc/puc.c optional puc dev/puc/puc_cfg.c optional puc dev/puc/puc_pccard.c optional puc pccard dev/puc/puc_pci.c optional puc pci dev/puc/pucdata.c optional puc pci dev/quicc/quicc_core.c optional quicc dev/ral/rt2560.c optional ral dev/ral/rt2661.c optional ral dev/ral/if_ral_pci.c optional ral pci rt2561fw.c optional rt2561fw | ralfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk rt2561.fw:rt2561fw -mrt2561 -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "rt2561fw.c" rt2561fw.fwo optional rt2561fw | ralfw \ dependency "rt2561.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} rt2561.fw" \ no-implicit-rule \ clean "rt2561fw.fwo" rt2561.fw optional rt2561fw | ralfw \ dependency "$S/contrib/dev/ral/rt2561.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/ral/rt2561.fw.uu" \ no-obj no-implicit-rule \ clean "rt2561.fw" rt2561sfw.c optional rt2561sfw | ralfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk rt2561s.fw:rt2561sfw -mrt2561s -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "rt2561sfw.c" rt2561sfw.fwo optional rt2561sfw | ralfw \ dependency "rt2561s.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} rt2561s.fw" \ no-implicit-rule \ clean "rt2561sfw.fwo" rt2561s.fw optional rt2561sfw | ralfw \ dependency "$S/contrib/dev/ral/rt2561s.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/ral/rt2561s.fw.uu" \ no-obj no-implicit-rule \ clean "rt2561s.fw" rt2661fw.c optional rt2661fw | ralfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk rt2661.fw:rt2661fw -mrt2661 -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "rt2661fw.c" rt2661fw.fwo optional rt2661fw | ralfw \ dependency "rt2661.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} rt2661.fw" \ no-implicit-rule \ clean "rt2661fw.fwo" rt2661.fw optional rt2661fw | ralfw \ dependency "$S/contrib/dev/ral/rt2661.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/ral/rt2661.fw.uu" \ no-obj no-implicit-rule \ clean "rt2661.fw" rt2860fw.c optional rt2860fw | ralfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk rt2860.fw:rt2860fw -mrt2860 -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "rt2860fw.c" rt2860fw.fwo optional rt2860fw | ralfw \ dependency "rt2860.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} rt2860.fw" \ no-implicit-rule \ clean "rt2860fw.fwo" rt2860.fw optional rt2860fw | ralfw \ dependency "$S/contrib/dev/ral/rt2860.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/ral/rt2860.fw.uu" \ no-obj no-implicit-rule \ clean "rt2860.fw" dev/random/harvest.c standard dev/random/hash.c optional random dev/random/probe.c optional random dev/random/randomdev.c optional random dev/random/randomdev_soft.c optional random dev/random/yarrow.c optional random dev/rc/rc.c optional rc dev/re/if_re.c optional re dev/rndtest/rndtest.c optional rndtest dev/rp/rp.c optional rp dev/rp/rp_isa.c optional rp isa dev/rp/rp_pci.c optional rp pci dev/safe/safe.c optional safe dev/scc/scc_if.m optional scc dev/scc/scc_bfe_ebus.c optional scc ebus dev/scc/scc_bfe_quicc.c optional scc quicc dev/scc/scc_bfe_sbus.c optional scc fhc | scc sbus dev/scc/scc_core.c optional scc dev/scc/scc_dev_quicc.c optional scc quicc dev/scc/scc_dev_sab82532.c optional scc dev/scc/scc_dev_z8530.c optional scc dev/scd/scd.c optional scd isa dev/scd/scd_isa.c optional scd isa dev/sdhci/sdhci.c optional sdhci pci dev/sf/if_sf.c optional sf pci dev/sge/if_sge.c optional sge pci dev/si/si.c optional si dev/si/si2_z280.c optional si dev/si/si3_t225.c optional si dev/si/si_eisa.c optional si eisa dev/si/si_isa.c optional si isa dev/si/si_pci.c optional si pci dev/siba/siba_bwn.c optional siba_bwn pci dev/siba/siba_core.c optional siba_bwn pci dev/siis/siis.c optional siis pci dev/sis/if_sis.c optional sis pci dev/sk/if_sk.c optional sk pci dev/smbus/smb.c optional smb dev/smbus/smbconf.c optional smbus dev/smbus/smbus.c optional smbus dev/smbus/smbus_if.m optional smbus dev/smc/if_smc.c optional smc dev/sn/if_sn.c optional sn dev/sn/if_sn_isa.c optional sn isa dev/sn/if_sn_pccard.c optional sn pccard dev/snp/snp.c optional snp dev/sound/clone.c optional sound dev/sound/unit.c optional sound dev/sound/isa/ad1816.c optional snd_ad1816 isa dev/sound/isa/ess.c optional snd_ess isa dev/sound/isa/gusc.c optional snd_gusc isa dev/sound/isa/mss.c optional snd_mss isa dev/sound/isa/sb16.c optional snd_sb16 isa dev/sound/isa/sb8.c optional snd_sb8 isa dev/sound/isa/sbc.c optional snd_sbc isa dev/sound/isa/sndbuf_dma.c optional sound isa dev/sound/pci/als4000.c optional snd_als4000 pci dev/sound/pci/atiixp.c optional snd_atiixp pci dev/sound/pci/cmi.c optional snd_cmi pci dev/sound/pci/cs4281.c optional snd_cs4281 pci dev/sound/pci/csa.c optional snd_csa pci \ warning "kernel contains GPL contaminated csaimg.h header" dev/sound/pci/csapcm.c optional snd_csa pci dev/sound/pci/ds1.c optional snd_ds1 pci dev/sound/pci/emu10k1.c optional snd_emu10k1 pci \ dependency "emu10k1-alsa%diked.h" \ warning "kernel contains GPL contaminated emu10k1 headers" dev/sound/pci/emu10kx.c optional snd_emu10kx pci \ dependency "emu10k1-alsa%diked.h" \ dependency "p16v-alsa%diked.h" \ dependency "p17v-alsa%diked.h" \ warning "kernel contains GPL contaminated emu10kx headers" dev/sound/pci/emu10kx-pcm.c optional snd_emu10kx pci \ dependency "emu10k1-alsa%diked.h" \ dependency "p16v-alsa%diked.h" \ dependency "p17v-alsa%diked.h" \ warning "kernel contains GPL contaminated emu10kx headers" dev/sound/pci/emu10kx-midi.c optional snd_emu10kx pci \ dependency "emu10k1-alsa%diked.h" \ warning "kernel contains GPL contaminated emu10kx headers" dev/sound/pci/envy24.c optional snd_envy24 pci dev/sound/pci/envy24ht.c optional snd_envy24ht pci dev/sound/pci/es137x.c optional snd_es137x pci dev/sound/pci/fm801.c optional snd_fm801 pci dev/sound/pci/ich.c optional snd_ich pci dev/sound/pci/maestro.c optional snd_maestro pci dev/sound/pci/maestro3.c optional snd_maestro3 pci \ warning "kernel contains GPL contaminated maestro3 headers" dev/sound/pci/neomagic.c optional snd_neomagic pci dev/sound/pci/solo.c optional snd_solo pci dev/sound/pci/spicds.c optional snd_spicds pci dev/sound/pci/t4dwave.c optional snd_t4dwave pci dev/sound/pci/via8233.c optional snd_via8233 pci dev/sound/pci/via82c686.c optional snd_via82c686 pci dev/sound/pci/vibes.c optional snd_vibes pci dev/sound/pci/hda/hdac.c optional snd_hda pci dev/sound/pcm/ac97.c optional sound dev/sound/pcm/ac97_if.m optional sound dev/sound/pcm/ac97_patch.c optional sound dev/sound/pcm/buffer.c optional sound \ dependency "snd_fxdiv_gen.h" dev/sound/pcm/channel.c optional sound dev/sound/pcm/channel_if.m optional sound dev/sound/pcm/dsp.c optional sound dev/sound/pcm/feeder.c optional sound dev/sound/pcm/feeder_chain.c optional sound dev/sound/pcm/feeder_eq.c optional sound \ dependency "feeder_eq_gen.h" \ dependency "snd_fxdiv_gen.h" dev/sound/pcm/feeder_if.m optional sound dev/sound/pcm/feeder_format.c optional sound \ dependency "snd_fxdiv_gen.h" dev/sound/pcm/feeder_matrix.c optional sound \ dependency "snd_fxdiv_gen.h" dev/sound/pcm/feeder_mixer.c optional sound \ dependency "snd_fxdiv_gen.h" dev/sound/pcm/feeder_rate.c optional sound \ dependency "feeder_rate_gen.h" \ dependency "snd_fxdiv_gen.h" dev/sound/pcm/feeder_volume.c optional sound \ dependency "snd_fxdiv_gen.h" dev/sound/pcm/mixer.c optional sound dev/sound/pcm/mixer_if.m optional sound dev/sound/pcm/sndstat.c optional sound dev/sound/pcm/sound.c optional sound dev/sound/pcm/vchan.c optional sound dev/sound/usb/uaudio.c optional snd_uaudio usb dev/sound/usb/uaudio_pcm.c optional snd_uaudio usb dev/sound/midi/midi.c optional sound dev/sound/midi/mpu401.c optional sound dev/sound/midi/mpu_if.m optional sound dev/sound/midi/mpufoi_if.m optional sound dev/sound/midi/sequencer.c optional sound dev/sound/midi/synth_if.m optional sound dev/spibus/spibus.c optional spibus \ dependency "spibus_if.h" dev/spibus/spibus_if.m optional spibus dev/ste/if_ste.c optional ste pci dev/stg/tmc18c30.c optional stg dev/stg/tmc18c30_isa.c optional stg isa dev/stg/tmc18c30_pccard.c optional stg pccard dev/stg/tmc18c30_pci.c optional stg pci dev/stg/tmc18c30_subr.c optional stg dev/stge/if_stge.c optional stge dev/streams/streams.c optional streams dev/sym/sym_hipd.c optional sym \ dependency "$S/dev/sym/sym_{conf,defs}.h" dev/syscons/blank/blank_saver.c optional blank_saver dev/syscons/daemon/daemon_saver.c optional daemon_saver dev/syscons/dragon/dragon_saver.c optional dragon_saver dev/syscons/fade/fade_saver.c optional fade_saver dev/syscons/fire/fire_saver.c optional fire_saver dev/syscons/green/green_saver.c optional green_saver dev/syscons/logo/logo.c optional logo_saver dev/syscons/logo/logo_saver.c optional logo_saver dev/syscons/rain/rain_saver.c optional rain_saver dev/syscons/schistory.c optional sc dev/syscons/scmouse.c optional sc dev/syscons/scterm.c optional sc dev/syscons/scvidctl.c optional sc dev/syscons/snake/snake_saver.c optional snake_saver dev/syscons/star/star_saver.c optional star_saver dev/syscons/syscons.c optional sc dev/syscons/sysmouse.c optional sc dev/syscons/warp/warp_saver.c optional warp_saver dev/tdfx/tdfx_linux.c optional tdfx_linux tdfx compat_linux dev/tdfx/tdfx_pci.c optional tdfx pci dev/ti/if_ti.c optional ti pci dev/tl/if_tl.c optional tl pci dev/trm/trm.c optional trm dev/twa/tw_cl_init.c optional twa \ compile-with "${NORMAL_C} -I$S/dev/twa" dev/twa/tw_cl_intr.c optional twa \ compile-with "${NORMAL_C} -I$S/dev/twa" dev/twa/tw_cl_io.c optional twa \ compile-with "${NORMAL_C} -I$S/dev/twa" dev/twa/tw_cl_misc.c optional twa \ compile-with "${NORMAL_C} -I$S/dev/twa" dev/twa/tw_osl_cam.c optional twa \ compile-with "${NORMAL_C} -I$S/dev/twa" dev/twa/tw_osl_freebsd.c optional twa \ compile-with "${NORMAL_C} -I$S/dev/twa" dev/twe/twe.c optional twe dev/twe/twe_freebsd.c optional twe dev/tx/if_tx.c optional tx dev/txp/if_txp.c optional txp dev/uart/uart_bus_acpi.c optional uart acpi #dev/uart/uart_bus_cbus.c optional uart cbus dev/uart/uart_bus_ebus.c optional uart ebus dev/uart/uart_bus_fdt.c optional uart fdt dev/uart/uart_bus_isa.c optional uart isa dev/uart/uart_bus_pccard.c optional uart pccard dev/uart/uart_bus_pci.c optional uart pci dev/uart/uart_bus_puc.c optional uart puc dev/uart/uart_bus_scc.c optional uart scc dev/uart/uart_core.c optional uart dev/uart/uart_dbg.c optional uart gdb dev/uart/uart_dev_ns8250.c optional uart uart_ns8250 dev/uart/uart_dev_quicc.c optional uart quicc dev/uart/uart_dev_sab82532.c optional uart uart_sab82532 dev/uart/uart_dev_sab82532.c optional uart scc dev/uart/uart_dev_z8530.c optional uart uart_z8530 dev/uart/uart_dev_z8530.c optional uart scc dev/uart/uart_if.m optional uart dev/uart/uart_subr.c optional uart dev/uart/uart_tty.c optional uart dev/ubsec/ubsec.c optional ubsec # # USB controller drivers # dev/usb/controller/at91dci.c optional at91dci dev/usb/controller/at91dci_atmelarm.c optional at91dci at91rm9200 dev/usb/controller/musb_otg.c optional musb dev/usb/controller/musb_otg_atmelarm.c optional musb at91rm9200 dev/usb/controller/ehci.c optional ehci dev/usb/controller/ehci_pci.c optional ehci pci dev/usb/controller/ohci.c optional ohci dev/usb/controller/ohci_atmelarm.c optional ohci at91rm9200 dev/usb/controller/ohci_pci.c optional ohci pci dev/usb/controller/uhci.c optional uhci dev/usb/controller/uhci_pci.c optional uhci pci dev/usb/controller/xhci.c optional xhci dev/usb/controller/xhci_pci.c optional xhci pci dev/usb/controller/uss820dci.c optional uss820dci dev/usb/controller/uss820dci_atmelarm.c optional uss820dci at91rm9200 dev/usb/controller/usb_controller.c optional usb # # USB storage drivers # dev/usb/storage/umass.c optional umass dev/usb/storage/urio.c optional urio dev/usb/storage/ustorage_fs.c optional usfs # # USB core # dev/usb/usb_busdma.c optional usb dev/usb/usb_compat_linux.c optional usb dev/usb/usb_core.c optional usb dev/usb/usb_debug.c optional usb dev/usb/usb_dev.c optional usb dev/usb/usb_device.c optional usb dev/usb/usb_dynamic.c optional usb dev/usb/usb_error.c optional usb dev/usb/usb_generic.c optional usb dev/usb/usb_handle_request.c optional usb dev/usb/usb_hid.c optional usb dev/usb/usb_hub.c optional usb dev/usb/usb_if.m optional usb dev/usb/usb_lookup.c optional usb dev/usb/usb_mbuf.c optional usb dev/usb/usb_msctest.c optional usb dev/usb/usb_parse.c optional usb dev/usb/usb_pf.c optional usb dev/usb/usb_process.c optional usb dev/usb/usb_request.c optional usb dev/usb/usb_transfer.c optional usb dev/usb/usb_util.c optional usb # # USB network drivers # dev/usb/net/if_aue.c optional aue dev/usb/net/if_axe.c optional axe dev/usb/net/if_cdce.c optional cdce dev/usb/net/if_cue.c optional cue dev/usb/net/if_ipheth.c optional ipheth dev/usb/net/if_kue.c optional kue dev/usb/net/if_mos.c optional mos dev/usb/net/if_rue.c optional rue dev/usb/net/if_udav.c optional udav dev/usb/net/usb_ethernet.c optional aue | axe | cdce | cue | kue | mos | \ rue | udav dev/usb/net/uhso.c optional uhso # # USB WLAN drivers # dev/usb/wlan/if_rum.c optional rum dev/usb/wlan/if_run.c optional run runfw.c optional runfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk runfw:runfw -mrunfw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "runfw.c" runfw.fwo optional runfw \ dependency "runfw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} runfw" \ no-implicit-rule \ clean "runfw.fwo" runfw optional runfw \ dependency "$S/contrib/dev/run/rt2870.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/run/rt2870.fw.uu" \ no-obj no-implicit-rule \ clean "runfw" dev/usb/wlan/if_uath.c optional uath dev/usb/wlan/if_upgt.c optional upgt dev/usb/wlan/if_ural.c optional ural dev/usb/wlan/if_urtw.c optional urtw dev/usb/wlan/if_zyd.c optional zyd # # USB serial and parallel port drivers # dev/usb/serial/u3g.c optional u3g dev/usb/serial/uark.c optional uark dev/usb/serial/ubsa.c optional ubsa dev/usb/serial/ubser.c optional ubser dev/usb/serial/uchcom.c optional uchcom dev/usb/serial/ucycom.c optional ucycom dev/usb/serial/ufoma.c optional ufoma dev/usb/serial/uftdi.c optional uftdi dev/usb/serial/ugensa.c optional ugensa dev/usb/serial/uipaq.c optional uipaq dev/usb/serial/ulpt.c optional ulpt dev/usb/serial/umcs.c optional umcs dev/usb/serial/umct.c optional umct dev/usb/serial/umodem.c optional umodem dev/usb/serial/umoscom.c optional umoscom dev/usb/serial/uplcom.c optional uplcom dev/usb/serial/uslcom.c optional uslcom dev/usb/serial/uvisor.c optional uvisor dev/usb/serial/uvscom.c optional uvscom dev/usb/serial/usb_serial.c optional ucom | u3g | uark | ubsa | ubser | \ uchcom | ucycom | ufoma | uftdi | \ ugensa | uipaq | umcs | umct | \ umodem | umoscom | uplcom | uslcom | \ uvisor | uvscom # # USB misc drivers # dev/usb/misc/ufm.c optional ufm dev/usb/misc/udbp.c optional udbp # # USB input drivers # dev/usb/input/atp.c optional atp dev/usb/input/uep.c optional uep dev/usb/input/uhid.c optional uhid dev/usb/input/ukbd.c optional ukbd dev/usb/input/ums.c optional ums # # USB quirks # dev/usb/quirk/usb_quirk.c optional usb # # USB templates # dev/usb/template/usb_template.c optional usb_template dev/usb/template/usb_template_cdce.c optional usb_template dev/usb/template/usb_template_msc.c optional usb_template dev/usb/template/usb_template_mtp.c optional usb_template # # USB END # dev/utopia/idtphy.c optional utopia dev/utopia/suni.c optional utopia dev/utopia/utopia.c optional utopia dev/vge/if_vge.c optional vge dev/vkbd/vkbd.c optional vkbd dev/vr/if_vr.c optional vr pci dev/vte/if_vte.c optional vte pci dev/vx/if_vx.c optional vx dev/vx/if_vx_eisa.c optional vx eisa dev/vx/if_vx_pci.c optional vx pci dev/vxge/vxge.c optional vxge dev/vxge/vxgehal/vxgehal-ifmsg.c optional vxge dev/vxge/vxgehal/vxgehal-mrpcim.c optional vxge dev/vxge/vxgehal/vxge-queue.c optional vxge dev/vxge/vxgehal/vxgehal-ring.c optional vxge dev/vxge/vxgehal/vxgehal-swapper.c optional vxge dev/vxge/vxgehal/vxgehal-mgmt.c optional vxge dev/vxge/vxgehal/vxgehal-srpcim.c optional vxge dev/vxge/vxgehal/vxgehal-config.c optional vxge dev/vxge/vxgehal/vxgehal-blockpool.c optional vxge dev/vxge/vxgehal/vxgehal-doorbells.c optional vxge dev/vxge/vxgehal/vxgehal-mgmtaux.c optional vxge dev/vxge/vxgehal/vxgehal-device.c optional vxge dev/vxge/vxgehal/vxgehal-mm.c optional vxge dev/vxge/vxgehal/vxgehal-driver.c optional vxge dev/vxge/vxgehal/vxgehal-virtualpath.c optional vxge dev/vxge/vxgehal/vxgehal-channel.c optional vxge dev/vxge/vxgehal/vxgehal-fifo.c optional vxge dev/watchdog/watchdog.c standard dev/wb/if_wb.c optional wb pci dev/wds/wd7000.c optional wds isa dev/wi/if_wi.c optional wi dev/wi/if_wi_pccard.c optional wi pccard dev/wi/if_wi_pci.c optional wi pci dev/wl/if_wl.c optional wl isa dev/wpi/if_wpi.c optional wpi pci wpifw.c optional wpifw \ compile-with "${AWK} -f $S/tools/fw_stub.awk wpi.fw:wpifw:153229 -mwpi -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "wpifw.c" wpifw.fwo optional wpifw \ dependency "wpi.fw" \ compile-with "${LD} -b binary -d -warn-common -r -d -o ${.TARGET} wpi.fw" \ no-implicit-rule \ clean "wpifw.fwo" wpi.fw optional wpifw \ dependency "$S/contrib/dev/wpi/iwlwifi-3945-15.32.2.9.fw.uu" \ compile-with "uudecode -o ${.TARGET} $S/contrib/dev/wpi/iwlwifi-3945-15.32.2.9.fw.uu" \ no-obj no-implicit-rule \ clean "wpi.fw" dev/xe/if_xe.c optional xe dev/xe/if_xe_pccard.c optional xe pccard dev/xl/if_xl.c optional xl pci fs/coda/coda_fbsd.c optional vcoda fs/coda/coda_psdev.c optional vcoda fs/coda/coda_subr.c optional vcoda fs/coda/coda_venus.c optional vcoda fs/coda/coda_vfsops.c optional vcoda fs/coda/coda_vnops.c optional vcoda fs/deadfs/dead_vnops.c standard fs/devfs/devfs_devs.c standard fs/devfs/devfs_dir.c standard fs/devfs/devfs_rule.c standard fs/devfs/devfs_vfsops.c standard fs/devfs/devfs_vnops.c standard fs/fdescfs/fdesc_vfsops.c optional fdescfs fs/fdescfs/fdesc_vnops.c optional fdescfs fs/fifofs/fifo_vnops.c standard fs/hpfs/hpfs_alsubr.c optional hpfs fs/hpfs/hpfs_lookup.c optional hpfs fs/hpfs/hpfs_subr.c optional hpfs fs/hpfs/hpfs_vfsops.c optional hpfs fs/hpfs/hpfs_vnops.c optional hpfs fs/msdosfs/msdosfs_conv.c optional msdosfs fs/msdosfs/msdosfs_denode.c optional msdosfs fs/msdosfs/msdosfs_fat.c optional msdosfs fs/msdosfs/msdosfs_fileno.c optional msdosfs fs/msdosfs/msdosfs_iconv.c optional msdosfs_iconv fs/msdosfs/msdosfs_lookup.c optional msdosfs fs/msdosfs/msdosfs_vfsops.c optional msdosfs fs/msdosfs/msdosfs_vnops.c optional msdosfs fs/nfs/nfs_commonkrpc.c optional nfscl | nfsd fs/nfs/nfs_commonsubs.c optional nfscl | nfsd fs/nfs/nfs_commonport.c optional nfscl | nfsd fs/nfs/nfs_commonacl.c optional nfscl | nfsd fs/nfsclient/nfs_clcomsubs.c optional nfscl fs/nfsclient/nfs_clsubs.c optional nfscl fs/nfsclient/nfs_clstate.c optional nfscl fs/nfsclient/nfs_clkrpc.c optional nfscl fs/nfsclient/nfs_clrpcops.c optional nfscl fs/nfsclient/nfs_clvnops.c optional nfscl fs/nfsclient/nfs_clnode.c optional nfscl fs/nfsclient/nfs_clvfsops.c optional nfscl fs/nfsclient/nfs_clport.c optional nfscl fs/nfsclient/nfs_clbio.c optional nfscl fs/nfsclient/nfs_clnfsiod.c optional nfscl fs/nfsserver/nfs_nfsdsocket.c optional nfsd inet fs/nfsserver/nfs_nfsdsubs.c optional nfsd inet fs/nfsserver/nfs_nfsdstate.c optional nfsd inet fs/nfsserver/nfs_nfsdkrpc.c optional nfsd inet fs/nfsserver/nfs_nfsdserv.c optional nfsd inet fs/nfsserver/nfs_nfsdport.c optional nfsd inet fs/nfsserver/nfs_nfsdcache.c optional nfsd inet fs/ntfs/ntfs_compr.c optional ntfs fs/ntfs/ntfs_iconv.c optional ntfs_iconv fs/ntfs/ntfs_ihash.c optional ntfs fs/ntfs/ntfs_subr.c optional ntfs fs/ntfs/ntfs_vfsops.c optional ntfs fs/ntfs/ntfs_vnops.c optional ntfs fs/nullfs/null_subr.c optional nullfs fs/nullfs/null_vfsops.c optional nullfs fs/nullfs/null_vnops.c optional nullfs fs/nwfs/nwfs_io.c optional nwfs fs/nwfs/nwfs_ioctl.c optional nwfs fs/nwfs/nwfs_node.c optional nwfs fs/nwfs/nwfs_subr.c optional nwfs fs/nwfs/nwfs_vfsops.c optional nwfs fs/nwfs/nwfs_vnops.c optional nwfs fs/portalfs/portal_vfsops.c optional portalfs fs/portalfs/portal_vnops.c optional portalfs fs/procfs/procfs.c optional procfs fs/procfs/procfs_ctl.c optional procfs fs/procfs/procfs_dbregs.c optional procfs fs/procfs/procfs_fpregs.c optional procfs fs/procfs/procfs_ioctl.c optional procfs fs/procfs/procfs_map.c optional procfs fs/procfs/procfs_mem.c optional procfs fs/procfs/procfs_note.c optional procfs fs/procfs/procfs_osrel.c optional procfs fs/procfs/procfs_regs.c optional procfs fs/procfs/procfs_rlimit.c optional procfs fs/procfs/procfs_status.c optional procfs fs/procfs/procfs_type.c optional procfs fs/pseudofs/pseudofs.c optional pseudofs fs/pseudofs/pseudofs_fileno.c optional pseudofs fs/pseudofs/pseudofs_vncache.c optional pseudofs fs/pseudofs/pseudofs_vnops.c optional pseudofs fs/smbfs/smbfs_io.c optional smbfs fs/smbfs/smbfs_node.c optional smbfs fs/smbfs/smbfs_smb.c optional smbfs fs/smbfs/smbfs_subr.c optional smbfs fs/smbfs/smbfs_vfsops.c optional smbfs fs/smbfs/smbfs_vnops.c optional smbfs fs/udf/osta.c optional udf fs/udf/udf_iconv.c optional udf_iconv fs/udf/udf_vfsops.c optional udf fs/udf/udf_vnops.c optional udf fs/unionfs/union_subr.c optional unionfs fs/unionfs/union_vfsops.c optional unionfs fs/unionfs/union_vnops.c optional unionfs fs/tmpfs/tmpfs_vnops.c optional tmpfs fs/tmpfs/tmpfs_fifoops.c optional tmpfs fs/tmpfs/tmpfs_vfsops.c optional tmpfs fs/tmpfs/tmpfs_subr.c optional tmpfs gdb/gdb_cons.c optional gdb gdb/gdb_main.c optional gdb gdb/gdb_packet.c optional gdb geom/bde/g_bde.c optional geom_bde geom/bde/g_bde_crypt.c optional geom_bde geom/bde/g_bde_lock.c optional geom_bde geom/bde/g_bde_work.c optional geom_bde geom/cache/g_cache.c optional geom_cache geom/concat/g_concat.c optional geom_concat geom/eli/g_eli.c optional geom_eli geom/eli/g_eli_crypto.c optional geom_eli geom/eli/g_eli_ctl.c optional geom_eli geom/eli/g_eli_integrity.c optional geom_eli geom/eli/g_eli_key.c optional geom_eli geom/eli/g_eli_key_cache.c optional geom_eli geom/eli/g_eli_privacy.c optional geom_eli geom/eli/pkcs5v2.c optional geom_eli geom/gate/g_gate.c optional geom_gate geom/geom_aes.c optional geom_aes geom/geom_bsd.c optional geom_bsd geom/geom_bsd_enc.c optional geom_bsd geom/geom_ccd.c optional ccd | geom_ccd geom/geom_ctl.c standard geom/geom_dev.c standard geom/geom_disk.c standard geom/geom_dump.c standard geom/geom_event.c standard geom/geom_fox.c optional geom_fox geom/geom_io.c standard geom/geom_kern.c standard geom/geom_map.c optional geom_map geom/geom_mbr.c optional geom_mbr geom/geom_mbr_enc.c optional geom_mbr geom/geom_pc98.c optional geom_pc98 geom/geom_pc98_enc.c optional geom_pc98 geom/geom_redboot.c optional geom_redboot geom/geom_slice.c standard geom/geom_subr.c standard geom/geom_sunlabel.c optional geom_sunlabel geom/geom_sunlabel_enc.c optional geom_sunlabel geom/geom_vfs.c standard geom/geom_vol_ffs.c optional geom_vol geom/journal/g_journal.c optional geom_journal geom/journal/g_journal_ufs.c optional geom_journal geom/label/g_label.c optional geom_label geom/label/g_label_ext2fs.c optional geom_label geom/label/g_label_iso9660.c optional geom_label geom/label/g_label_msdosfs.c optional geom_label geom/label/g_label_ntfs.c optional geom_label geom/label/g_label_reiserfs.c optional geom_label geom/label/g_label_ufs.c optional geom_label geom/label/g_label_gpt.c optional geom_label geom/linux_lvm/g_linux_lvm.c optional geom_linux_lvm geom/mirror/g_mirror.c optional geom_mirror geom/mirror/g_mirror_ctl.c optional geom_mirror geom/mountver/g_mountver.c optional geom_mountver geom/multipath/g_multipath.c optional geom_multipath geom/nop/g_nop.c optional geom_nop geom/part/g_part.c standard geom/part/g_part_if.m standard geom/part/g_part_apm.c optional geom_part_apm geom/part/g_part_bsd.c optional geom_part_bsd geom/part/g_part_ebr.c optional geom_part_ebr geom/part/g_part_gpt.c optional geom_part_gpt geom/part/g_part_mbr.c optional geom_part_mbr geom/part/g_part_pc98.c optional geom_part_pc98 geom/part/g_part_vtoc8.c optional geom_part_vtoc8 geom/raid/g_raid.c optional geom_raid geom/raid/g_raid_ctl.c optional geom_raid geom/raid/g_raid_md_if.m optional geom_raid geom/raid/g_raid_tr_if.m optional geom_raid geom/raid/md_intel.c optional geom_raid geom/raid/md_jmicron.c optional geom_raid geom/raid/md_nvidia.c optional geom_raid geom/raid/md_promise.c optional geom_raid geom/raid/md_sii.c optional geom_raid geom/raid/tr_concat.c optional geom_raid geom/raid/tr_raid0.c optional geom_raid geom/raid/tr_raid1.c optional geom_raid geom/raid/tr_raid1e.c optional geom_raid geom/raid3/g_raid3.c optional geom_raid3 geom/raid3/g_raid3_ctl.c optional geom_raid3 geom/shsec/g_shsec.c optional geom_shsec geom/stripe/g_stripe.c optional geom_stripe geom/uzip/g_uzip.c optional geom_uzip geom/virstor/binstream.c optional geom_virstor geom/virstor/g_virstor.c optional geom_virstor geom/virstor/g_virstor_md.c optional geom_virstor geom/zero/g_zero.c optional geom_zero fs/ext2fs/ext2_alloc.c optional ext2fs fs/ext2fs/ext2_balloc.c optional ext2fs fs/ext2fs/ext2_bmap.c optional ext2fs fs/ext2fs/ext2_inode.c optional ext2fs fs/ext2fs/ext2_inode_cnv.c optional ext2fs fs/ext2fs/ext2_lookup.c optional ext2fs fs/ext2fs/ext2_subr.c optional ext2fs fs/ext2fs/ext2_vfsops.c optional ext2fs fs/ext2fs/ext2_vnops.c optional ext2fs gnu/fs/reiserfs/reiserfs_hashes.c optional reiserfs \ warning "kernel contains GPL contaminated ReiserFS filesystem" gnu/fs/reiserfs/reiserfs_inode.c optional reiserfs gnu/fs/reiserfs/reiserfs_item_ops.c optional reiserfs gnu/fs/reiserfs/reiserfs_namei.c optional reiserfs gnu/fs/reiserfs/reiserfs_prints.c optional reiserfs gnu/fs/reiserfs/reiserfs_stree.c optional reiserfs gnu/fs/reiserfs/reiserfs_vfsops.c optional reiserfs gnu/fs/reiserfs/reiserfs_vnops.c optional reiserfs # isa/isa_if.m standard isa/isa_common.c optional isa isa/isahint.c optional isa isa/pnp.c optional isa isapnp isa/pnpparse.c optional isa isapnp fs/cd9660/cd9660_bmap.c optional cd9660 fs/cd9660/cd9660_lookup.c optional cd9660 fs/cd9660/cd9660_node.c optional cd9660 fs/cd9660/cd9660_rrip.c optional cd9660 fs/cd9660/cd9660_util.c optional cd9660 fs/cd9660/cd9660_vfsops.c optional cd9660 fs/cd9660/cd9660_vnops.c optional cd9660 fs/cd9660/cd9660_iconv.c optional cd9660_iconv kern/bus_if.m standard kern/clock_if.m standard kern/cpufreq_if.m standard kern/device_if.m standard kern/imgact_elf.c standard kern/imgact_shell.c standard kern/inflate.c optional gzip kern/init_main.c standard kern/init_sysent.c standard kern/ksched.c optional _kposix_priority_scheduling kern/kern_acct.c standard kern/kern_alq.c optional alq kern/kern_clock.c standard kern/kern_condvar.c standard kern/kern_conf.c standard kern/kern_cons.c standard kern/kern_cpu.c standard kern/kern_cpuset.c standard kern/kern_context.c standard kern/kern_descrip.c standard kern/kern_dtrace.c optional kdtrace_hooks kern/kern_environment.c standard kern/kern_et.c standard kern/kern_event.c standard kern/kern_exec.c standard kern/kern_exit.c standard kern/kern_fail.c standard kern/kern_fork.c standard kern/kern_gzio.c optional gzio kern/kern_hhook.c standard kern/kern_idle.c standard kern/kern_intr.c standard kern/kern_jail.c standard kern/kern_khelp.c standard kern/kern_kthread.c standard kern/kern_ktr.c optional ktr kern/kern_ktrace.c standard kern/kern_linker.c standard kern/kern_lock.c standard kern/kern_lockf.c standard kern/kern_lockstat.c optional kdtrace_hooks kern/kern_loginclass.c standard kern/kern_malloc.c standard kern/kern_mbuf.c standard kern/kern_mib.c standard kern/kern_module.c standard kern/kern_mtxpool.c standard kern/kern_mutex.c standard kern/kern_ntptime.c standard kern/kern_osd.c standard kern/kern_physio.c standard kern/kern_pmc.c standard kern/kern_poll.c optional device_polling kern/kern_priv.c standard kern/kern_proc.c standard kern/kern_prot.c standard kern/kern_racct.c standard kern/kern_rctl.c standard kern/kern_resource.c standard kern/kern_rmlock.c standard kern/kern_rwlock.c standard kern/kern_sdt.c optional kdtrace_hooks kern/kern_sema.c standard kern/kern_shutdown.c standard kern/kern_sig.c standard kern/kern_switch.c standard kern/kern_sx.c standard kern/kern_synch.c standard kern/kern_syscalls.c standard kern/kern_sysctl.c standard kern/kern_tc.c standard kern/kern_thr.c standard kern/kern_thread.c standard kern/kern_time.c standard kern/kern_timeout.c standard kern/kern_umtx.c standard kern/kern_uuid.c standard kern/kern_xxx.c standard kern/link_elf.c standard kern/linker_if.m standard kern/md4c.c optional netsmb kern/md5c.c standard kern/p1003_1b.c standard kern/posix4_mib.c standard kern/sched_4bsd.c optional sched_4bsd kern/sched_ule.c optional sched_ule kern/serdev_if.m standard kern/stack_protector.c standard \ compile-with "${NORMAL_C:N-fstack-protector*}" kern/subr_acl_nfs4.c standard kern/subr_acl_posix1e.c standard kern/subr_autoconf.c standard kern/subr_blist.c standard kern/subr_bus.c standard kern/subr_bufring.c standard kern/subr_clock.c standard kern/subr_devstat.c standard kern/subr_disk.c standard kern/subr_eventhandler.c standard kern/subr_fattime.c standard kern/subr_firmware.c optional firmware kern/subr_hash.c standard kern/subr_hints.c standard kern/subr_kdb.c standard kern/subr_kobj.c standard kern/subr_lock.c standard kern/subr_log.c standard kern/subr_mbpool.c optional libmbpool kern/subr_mchain.c optional libmchain kern/subr_module.c standard kern/subr_msgbuf.c standard kern/subr_param.c standard kern/subr_pcpu.c standard kern/subr_power.c standard kern/subr_prf.c standard kern/subr_prof.c standard kern/subr_rman.c standard kern/subr_rtc.c standard kern/subr_sbuf.c standard kern/subr_scanf.c standard kern/subr_sglist.c standard kern/subr_sleepqueue.c standard kern/subr_smp.c standard kern/subr_stack.c optional ddb | stack | ktr kern/subr_taskqueue.c standard kern/subr_trap.c standard kern/subr_turnstile.c standard kern/subr_uio.c standard kern/subr_unit.c standard kern/subr_witness.c optional witness kern/sys_capability.c standard kern/sys_generic.c standard kern/sys_pipe.c standard kern/sys_process.c standard kern/sys_socket.c standard kern/syscalls.c standard kern/sysv_ipc.c standard kern/sysv_msg.c optional sysvmsg kern/sysv_sem.c optional sysvsem kern/sysv_shm.c optional sysvshm kern/tty.c standard kern/tty_compat.c optional compat_43tty kern/tty_info.c standard kern/tty_inq.c standard kern/tty_outq.c standard kern/tty_pts.c standard kern/tty_tty.c standard kern/tty_ttydisc.c standard kern/uipc_accf.c optional inet kern/uipc_cow.c optional zero_copy_sockets kern/uipc_debug.c optional ddb kern/uipc_domain.c standard kern/uipc_mbuf.c standard kern/uipc_mbuf2.c standard kern/uipc_mqueue.c optional p1003_1b_mqueue kern/uipc_sem.c optional p1003_1b_semaphores kern/uipc_shm.c standard kern/uipc_sockbuf.c standard kern/uipc_socket.c standard kern/uipc_syscalls.c standard kern/uipc_usrreq.c standard kern/vfs_acl.c standard kern/vfs_aio.c optional vfs_aio kern/vfs_bio.c standard kern/vfs_cache.c standard kern/vfs_cluster.c standard kern/vfs_default.c standard kern/vfs_export.c standard kern/vfs_extattr.c standard kern/vfs_hash.c standard kern/vfs_init.c standard kern/vfs_lookup.c standard kern/vfs_mount.c standard kern/vfs_mountroot.c standard kern/vfs_subr.c standard kern/vfs_syscalls.c standard kern/vfs_vnops.c standard # # Kernel GSS-API # gssd.h optional kgssapi \ dependency "$S/kgssapi/gssd.x" \ compile-with "rpcgen -hM $S/kgssapi/gssd.x | grep -v pthread.h > gssd.h" \ no-obj no-implicit-rule before-depend local \ clean "gssd.h" gssd_xdr.c optional kgssapi \ dependency "$S/kgssapi/gssd.x gssd.h" \ compile-with "rpcgen -c $S/kgssapi/gssd.x -o gssd_xdr.c" \ no-implicit-rule before-depend local \ clean "gssd_xdr.c" gssd_clnt.c optional kgssapi \ dependency "$S/kgssapi/gssd.x gssd.h" \ compile-with "rpcgen -lM $S/kgssapi/gssd.x | grep -v string.h > gssd_clnt.c" \ no-implicit-rule before-depend local \ clean "gssd_clnt.c" kgssapi/gss_accept_sec_context.c optional kgssapi kgssapi/gss_add_oid_set_member.c optional kgssapi kgssapi/gss_acquire_cred.c optional kgssapi kgssapi/gss_canonicalize_name.c optional kgssapi kgssapi/gss_create_empty_oid_set.c optional kgssapi kgssapi/gss_delete_sec_context.c optional kgssapi kgssapi/gss_display_status.c optional kgssapi kgssapi/gss_export_name.c optional kgssapi kgssapi/gss_get_mic.c optional kgssapi kgssapi/gss_init_sec_context.c optional kgssapi kgssapi/gss_impl.c optional kgssapi kgssapi/gss_import_name.c optional kgssapi kgssapi/gss_names.c optional kgssapi kgssapi/gss_pname_to_uid.c optional kgssapi kgssapi/gss_release_buffer.c optional kgssapi kgssapi/gss_release_cred.c optional kgssapi kgssapi/gss_release_name.c optional kgssapi kgssapi/gss_release_oid_set.c optional kgssapi kgssapi/gss_set_cred_option.c optional kgssapi kgssapi/gss_test_oid_set_member.c optional kgssapi kgssapi/gss_unwrap.c optional kgssapi kgssapi/gss_verify_mic.c optional kgssapi kgssapi/gss_wrap.c optional kgssapi kgssapi/gss_wrap_size_limit.c optional kgssapi kgssapi/gssd_prot.c optional kgssapi kgssapi/krb5/krb5_mech.c optional kgssapi kgssapi/krb5/kcrypto.c optional kgssapi kgssapi/krb5/kcrypto_aes.c optional kgssapi kgssapi/krb5/kcrypto_arcfour.c optional kgssapi kgssapi/krb5/kcrypto_des.c optional kgssapi kgssapi/krb5/kcrypto_des3.c optional kgssapi kgssapi/kgss_if.m optional kgssapi kgssapi/gsstest.c optional kgssapi_debug # These files in libkern/ are those needed by all architectures. Some # of the files in libkern/ are only needed on some architectures, e.g., # libkern/divdi3.c is needed by i386 but not alpha. Also, some of these # routines may be optimized for a particular platform. In either case, # the file should be moved to conf/files. from here. # libkern/arc4random.c standard libkern/bcd.c standard libkern/bsearch.c standard libkern/crc32.c standard libkern/fnmatch.c standard libkern/gets.c standard libkern/iconv.c optional libiconv libkern/iconv_converter_if.m optional libiconv libkern/iconv_xlat.c optional libiconv libkern/iconv_xlat16.c optional libiconv libkern/index.c standard libkern/inet_aton.c standard libkern/inet_ntoa.c standard libkern/inet_ntop.c standard libkern/inet_pton.c standard libkern/mcount.c optional profiling-routine libkern/memcmp.c standard libkern/qsort.c standard libkern/qsort_r.c standard libkern/random.c standard libkern/rindex.c standard libkern/scanc.c standard libkern/skpc.c standard libkern/strcasecmp.c standard libkern/strcat.c standard libkern/strcmp.c standard libkern/strcpy.c standard libkern/strcspn.c standard libkern/strdup.c standard libkern/strlcat.c standard libkern/strlcpy.c standard libkern/strlen.c standard libkern/strncmp.c standard libkern/strncpy.c standard libkern/strsep.c standard libkern/strspn.c standard libkern/strstr.c standard libkern/strtol.c standard libkern/strtoq.c standard libkern/strtoul.c standard libkern/strtouq.c standard libkern/strvalid.c standard net/bpf.c standard net/bpf_buffer.c optional bpf net/bpf_jitter.c optional bpf_jitter net/bpf_filter.c optional bpf | netgraph_bpf net/bpf_zerocopy.c optional bpf net/bridgestp.c optional bridge | if_bridge net/flowtable.c optional flowtable inet | flowtable inet6 net/ieee8023ad_lacp.c optional lagg net/if.c standard net/if_arcsubr.c optional arcnet net/if_atmsubr.c optional atm net/if_bridge.c optional bridge inet | if_bridge inet net/if_clone.c standard net/if_dead.c standard net/if_debug.c optional ddb net/if_disc.c optional disc net/if_edsc.c optional edsc net/if_ef.c optional ef net/if_enc.c optional enc ipsec inet | enc ipsec inet6 net/if_epair.c optional epair net/if_ethersubr.c optional ether \ compile-with "${NORMAL_C} -I$S/contrib/pf" net/if_faith.c optional faith net/if_fddisubr.c optional fddi net/if_fwsubr.c optional fwip net/if_gif.c optional gif | netgraph_gif net/if_gre.c optional gre inet net/if_iso88025subr.c optional token net/if_lagg.c optional lagg net/if_loop.c optional loop net/if_llatbl.c standard net/if_media.c standard net/if_mib.c standard net/if_spppfr.c optional sppp | netgraph_sppp net/if_spppsubr.c optional sppp | netgraph_sppp net/if_stf.c optional stf inet inet6 net/if_tun.c optional tun net/if_tap.c optional tap net/if_vlan.c optional vlan net/mppcc.c optional netgraph_mppc_compression net/mppcd.c optional netgraph_mppc_compression net/netisr.c standard net/pfil.c optional ether | inet net/radix.c standard net/radix_mpath.c standard net/raw_cb.c standard net/raw_usrreq.c standard net/route.c standard net/rtsock.c standard net/slcompress.c optional netgraph_vjc | sppp | \ netgraph_sppp net/vnet.c optional vimage net/zlib.c optional crypto | geom_uzip | ipsec | \ mxge | netgraph_deflate | \ ddb_ctf | gzio net80211/ieee80211.c optional wlan net80211/ieee80211_acl.c optional wlan wlan_acl net80211/ieee80211_action.c optional wlan net80211/ieee80211_ageq.c optional wlan net80211/ieee80211_adhoc.c optional wlan net80211/ieee80211_ageq.c optional wlan net80211/ieee80211_amrr.c optional wlan | wlan_amrr net80211/ieee80211_crypto.c optional wlan net80211/ieee80211_crypto_ccmp.c optional wlan wlan_ccmp net80211/ieee80211_crypto_none.c optional wlan net80211/ieee80211_crypto_tkip.c optional wlan wlan_tkip net80211/ieee80211_crypto_wep.c optional wlan wlan_wep net80211/ieee80211_ddb.c optional wlan ddb net80211/ieee80211_dfs.c optional wlan net80211/ieee80211_freebsd.c optional wlan net80211/ieee80211_hostap.c optional wlan net80211/ieee80211_ht.c optional wlan net80211/ieee80211_hwmp.c optional wlan ieee80211_support_mesh net80211/ieee80211_input.c optional wlan net80211/ieee80211_ioctl.c optional wlan net80211/ieee80211_mesh.c optional wlan ieee80211_support_mesh net80211/ieee80211_monitor.c optional wlan net80211/ieee80211_node.c optional wlan net80211/ieee80211_output.c optional wlan net80211/ieee80211_phy.c optional wlan net80211/ieee80211_power.c optional wlan net80211/ieee80211_proto.c optional wlan net80211/ieee80211_radiotap.c optional wlan net80211/ieee80211_ratectl.c optional wlan net80211/ieee80211_ratectl_none.c optional wlan net80211/ieee80211_regdomain.c optional wlan net80211/ieee80211_rssadapt.c optional wlan wlan_rssadapt net80211/ieee80211_scan.c optional wlan net80211/ieee80211_scan_sta.c optional wlan net80211/ieee80211_sta.c optional wlan net80211/ieee80211_superg.c optional wlan ieee80211_support_superg net80211/ieee80211_tdma.c optional wlan ieee80211_support_tdma net80211/ieee80211_wds.c optional wlan net80211/ieee80211_xauth.c optional wlan wlan_xauth net80211/ieee80211_alq.c optional wlan ieee80211_alq netatalk/aarp.c optional netatalk netatalk/at_control.c optional netatalk netatalk/at_proto.c optional netatalk netatalk/at_rmx.c optional netatalk netatalk/ddp_input.c optional netatalk netatalk/ddp_output.c optional netatalk netatalk/ddp_pcb.c optional netatalk netatalk/ddp_usrreq.c optional netatalk netgraph/atm/ccatm/ng_ccatm.c optional ngatm_ccatm \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" netgraph/atm/ng_atm.c optional ngatm_atm netgraph/atm/ngatmbase.c optional ngatm_atmbase \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" netgraph/atm/sscfu/ng_sscfu.c optional ngatm_sscfu \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" netgraph/atm/sscop/ng_sscop.c optional ngatm_sscop \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" netgraph/atm/uni/ng_uni.c optional ngatm_uni \ compile-with "${NORMAL_C} -I$S/contrib/ngatm" netgraph/bluetooth/common/ng_bluetooth.c optional netgraph_bluetooth netgraph/bluetooth/drivers/bt3c/ng_bt3c_pccard.c optional netgraph_bluetooth_bt3c netgraph/bluetooth/drivers/h4/ng_h4.c optional netgraph_bluetooth_h4 netgraph/bluetooth/drivers/ubt/ng_ubt.c optional netgraph_bluetooth_ubt usb netgraph/bluetooth/drivers/ubtbcmfw/ubtbcmfw.c optional netgraph_bluetooth_ubtbcmfw usb netgraph/bluetooth/hci/ng_hci_cmds.c optional netgraph_bluetooth_hci netgraph/bluetooth/hci/ng_hci_evnt.c optional netgraph_bluetooth_hci netgraph/bluetooth/hci/ng_hci_main.c optional netgraph_bluetooth_hci netgraph/bluetooth/hci/ng_hci_misc.c optional netgraph_bluetooth_hci netgraph/bluetooth/hci/ng_hci_ulpi.c optional netgraph_bluetooth_hci netgraph/bluetooth/l2cap/ng_l2cap_cmds.c optional netgraph_bluetooth_l2cap netgraph/bluetooth/l2cap/ng_l2cap_evnt.c optional netgraph_bluetooth_l2cap netgraph/bluetooth/l2cap/ng_l2cap_llpi.c optional netgraph_bluetooth_l2cap netgraph/bluetooth/l2cap/ng_l2cap_main.c optional netgraph_bluetooth_l2cap netgraph/bluetooth/l2cap/ng_l2cap_misc.c optional netgraph_bluetooth_l2cap netgraph/bluetooth/l2cap/ng_l2cap_ulpi.c optional netgraph_bluetooth_l2cap netgraph/bluetooth/socket/ng_btsocket.c optional netgraph_bluetooth_socket netgraph/bluetooth/socket/ng_btsocket_hci_raw.c optional netgraph_bluetooth_socket netgraph/bluetooth/socket/ng_btsocket_l2cap.c optional netgraph_bluetooth_socket netgraph/bluetooth/socket/ng_btsocket_l2cap_raw.c optional netgraph_bluetooth_socket netgraph/bluetooth/socket/ng_btsocket_rfcomm.c optional netgraph_bluetooth_socket netgraph/bluetooth/socket/ng_btsocket_sco.c optional netgraph_bluetooth_socket netgraph/netflow/netflow.c optional netgraph_netflow netgraph/netflow/netflow_v9.c optional netgraph_netflow netgraph/netflow/ng_netflow.c optional netgraph_netflow netgraph/ng_UI.c optional netgraph_UI netgraph/ng_async.c optional netgraph_async netgraph/ng_atmllc.c optional netgraph_atmllc netgraph/ng_base.c optional netgraph netgraph/ng_bpf.c optional netgraph_bpf netgraph/ng_bridge.c optional netgraph_bridge netgraph/ng_car.c optional netgraph_car netgraph/ng_cisco.c optional netgraph_cisco netgraph/ng_deflate.c optional netgraph_deflate netgraph/ng_device.c optional netgraph_device netgraph/ng_echo.c optional netgraph_echo netgraph/ng_eiface.c optional netgraph_eiface netgraph/ng_ether.c optional netgraph_ether netgraph/ng_ether_echo.c optional netgraph_ether_echo netgraph/ng_fec.c optional netgraph_fec netgraph/ng_frame_relay.c optional netgraph_frame_relay netgraph/ng_gif.c optional netgraph_gif netgraph/ng_gif_demux.c optional netgraph_gif_demux netgraph/ng_hole.c optional netgraph_hole netgraph/ng_iface.c optional netgraph_iface netgraph/ng_ip_input.c optional netgraph_ip_input netgraph/ng_ipfw.c optional netgraph_ipfw inet ipfirewall netgraph/ng_ksocket.c optional netgraph_ksocket netgraph/ng_l2tp.c optional netgraph_l2tp netgraph/ng_lmi.c optional netgraph_lmi netgraph/ng_mppc.c optional netgraph_mppc_compression | \ netgraph_mppc_encryption netgraph/ng_nat.c optional netgraph_nat inet libalias netgraph/ng_one2many.c optional netgraph_one2many netgraph/ng_parse.c optional netgraph netgraph/ng_patch.c optional netgraph_patch netgraph/ng_pipe.c optional netgraph_pipe netgraph/ng_ppp.c optional netgraph_ppp netgraph/ng_pppoe.c optional netgraph_pppoe netgraph/ng_pptpgre.c optional netgraph_pptpgre netgraph/ng_pred1.c optional netgraph_pred1 netgraph/ng_rfc1490.c optional netgraph_rfc1490 netgraph/ng_socket.c optional netgraph_socket netgraph/ng_split.c optional netgraph_split netgraph/ng_sppp.c optional netgraph_sppp netgraph/ng_tag.c optional netgraph_tag netgraph/ng_tcpmss.c optional netgraph_tcpmss netgraph/ng_tee.c optional netgraph_tee netgraph/ng_tty.c optional netgraph_tty netgraph/ng_vjc.c optional netgraph_vjc netgraph/ng_vlan.c optional netgraph_vlan netinet/accf_data.c optional accept_filter_data inet netinet/accf_dns.c optional accept_filter_dns inet netinet/accf_http.c optional accept_filter_http inet netinet/if_atm.c optional atm netinet/if_ether.c optional inet ether netinet/igmp.c optional inet netinet/in.c optional inet netinet/in_debug.c optional inet ddb netinet/ip_carp.c optional inet carp | inet6 carp netinet/in_gif.c optional gif inet | netgraph_gif inet netinet/ip_gre.c optional gre inet netinet/ip_id.c optional inet netinet/in_mcast.c optional inet netinet/in_pcb.c optional inet | inet6 +netinet/in_pcbgroup.c optional inet pcbgroup | inet6 pcbgroup netinet/in_proto.c optional inet | inet6 \ compile-with "${NORMAL_C} -I$S/contrib/pf" netinet/in_rmx.c optional inet netinet/ip_divert.c optional inet ipdivert ipfirewall netinet/ipfw/dn_heap.c optional inet dummynet netinet/ipfw/dn_sched_fifo.c optional inet dummynet netinet/ipfw/dn_sched_prio.c optional inet dummynet netinet/ipfw/dn_sched_qfq.c optional inet dummynet netinet/ipfw/dn_sched_rr.c optional inet dummynet netinet/ipfw/dn_sched_wf2q.c optional inet dummynet netinet/ipfw/ip_dummynet.c optional inet dummynet netinet/ipfw/ip_dn_io.c optional inet dummynet netinet/ipfw/ip_dn_glue.c optional inet dummynet netinet/ip_ecn.c optional inet | inet6 netinet/ip_encap.c optional inet | inet6 netinet/ip_fastfwd.c optional inet netinet/ipfw/ip_fw2.c optional inet ipfirewall \ compile-with "${NORMAL_C} -I$S/contrib/pf" netinet/ipfw/ip_fw_dynamic.c optional inet ipfirewall netinet/ipfw/ip_fw_log.c optional inet ipfirewall netinet/ipfw/ip_fw_pfil.c optional inet ipfirewall netinet/ipfw/ip_fw_sockopt.c optional inet ipfirewall netinet/ipfw/ip_fw_table.c optional inet ipfirewall netinet/ipfw/ip_fw_nat.c optional inet ipfirewall_nat netinet/ip_icmp.c optional inet | inet6 netinet/ip_input.c optional inet netinet/ip_ipsec.c optional inet ipsec netinet/ip_mroute.c optional mrouting inet netinet/ip_options.c optional inet netinet/ip_output.c optional inet netinet/raw_ip.c optional inet | inet6 netinet/cc/cc.c optional inet | inet6 netinet/cc/cc_newreno.c optional inet | inet6 netinet/sctp_asconf.c optional inet sctp | inet6 sctp netinet/sctp_auth.c optional inet sctp | inet6 sctp netinet/sctp_bsd_addr.c optional inet sctp | inet6 sctp netinet/sctp_cc_functions.c optional inet sctp | inet6 sctp netinet/sctp_crc32.c optional inet sctp | inet6 sctp netinet/sctp_indata.c optional inet sctp | inet6 sctp netinet/sctp_input.c optional inet sctp | inet6 sctp netinet/sctp_output.c optional inet sctp | inet6 sctp netinet/sctp_pcb.c optional inet sctp | inet6 sctp netinet/sctp_peeloff.c optional inet sctp | inet6 sctp netinet/sctp_ss_functions.c optional inet sctp | inet6 sctp netinet/sctp_sysctl.c optional inet sctp | inet6 sctp netinet/sctp_timer.c optional inet sctp | inet6 sctp netinet/sctp_usrreq.c optional inet sctp | inet6 sctp netinet/sctputil.c optional inet sctp | inet6 sctp netinet/tcp_debug.c optional tcpdebug netinet/tcp_hostcache.c optional inet | inet6 netinet/tcp_input.c optional inet | inet6 netinet/tcp_lro.c optional inet | inet6 netinet/tcp_output.c optional inet | inet6 netinet/tcp_offload.c optional inet | inet6 netinet/tcp_reass.c optional inet | inet6 netinet/tcp_sack.c optional inet | inet6 netinet/tcp_subr.c optional inet | inet6 netinet/tcp_syncache.c optional inet | inet6 netinet/tcp_timer.c optional inet | inet6 netinet/tcp_timewait.c optional inet | inet6 netinet/tcp_usrreq.c optional inet | inet6 netinet/udp_usrreq.c optional inet | inet6 netinet/libalias/alias.c optional libalias inet | netgraph_nat inet netinet/libalias/alias_db.c optional libalias inet | netgraph_nat inet netinet/libalias/alias_mod.c optional libalias | netgraph_nat netinet/libalias/alias_proxy.c optional libalias inet | netgraph_nat inet netinet/libalias/alias_util.c optional libalias inet | netgraph_nat inet netinet/libalias/alias_sctp.c optional libalias inet | netgraph_nat inet netinet6/dest6.c optional inet6 netinet6/frag6.c optional inet6 netinet6/icmp6.c optional inet6 netinet6/in6.c optional inet6 netinet6/in6_cksum.c optional inet6 netinet6/in6_gif.c optional gif inet6 | netgraph_gif inet6 netinet6/in6_ifattach.c optional inet6 netinet6/in6_mcast.c optional inet6 netinet6/in6_pcb.c optional inet6 +netinet6/in6_pcbgroup.c optional inet6 pcbgroup netinet6/in6_proto.c optional inet6 netinet6/in6_rmx.c optional inet6 netinet6/in6_src.c optional inet6 netinet6/ip6_forward.c optional inet6 netinet6/ip6_id.c optional inet6 netinet6/ip6_input.c optional inet6 netinet6/ip6_mroute.c optional mrouting inet6 netinet6/ip6_output.c optional inet6 netinet6/ip6_ipsec.c optional inet6 ipsec netinet6/mld6.c optional inet6 netinet6/nd6.c optional inet6 netinet6/nd6_nbr.c optional inet6 netinet6/nd6_rtr.c optional inet6 netinet6/raw_ip6.c optional inet6 netinet6/route6.c optional inet6 netinet6/scope6.c optional inet6 netinet6/sctp6_usrreq.c optional inet6 sctp netinet6/udp6_usrreq.c optional inet6 netipsec/ipsec.c optional ipsec inet | ipsec inet6 netipsec/ipsec_input.c optional ipsec inet | ipsec inet6 netipsec/ipsec_mbuf.c optional ipsec inet | ipsec inet6 netipsec/ipsec_output.c optional ipsec inet | ipsec inet6 netipsec/key.c optional ipsec inet | ipsec inet6 netipsec/key_debug.c optional ipsec inet | ipsec inet6 netipsec/keysock.c optional ipsec inet | ipsec inet6 netipsec/xform_ah.c optional ipsec inet | ipsec inet6 netipsec/xform_esp.c optional ipsec inet | ipsec inet6 netipsec/xform_ipcomp.c optional ipsec inet | ipsec inet6 netipsec/xform_ipip.c optional ipsec inet | ipsec inet6 netipsec/xform_tcp.c optional ipsec inet tcp_signature | \ ipsec inet6 tcp_signature netipx/ipx.c optional ipx netipx/ipx_cksum.c optional ipx netipx/ipx_input.c optional ipx netipx/ipx_outputfl.c optional ipx netipx/ipx_pcb.c optional ipx netipx/ipx_proto.c optional ipx netipx/ipx_usrreq.c optional ipx netipx/spx_debug.c optional ipx netipx/spx_reass.c optional ipx netipx/spx_usrreq.c optional ipx netnatm/natm.c optional natm netnatm/natm_pcb.c optional natm netnatm/natm_proto.c optional natm netncp/ncp_conn.c optional ncp netncp/ncp_crypt.c optional ncp netncp/ncp_login.c optional ncp netncp/ncp_mod.c optional ncp netncp/ncp_ncp.c optional ncp netncp/ncp_nls.c optional ncp netncp/ncp_rq.c optional ncp netncp/ncp_sock.c optional ncp netncp/ncp_subr.c optional ncp netsmb/smb_conn.c optional netsmb netsmb/smb_crypt.c optional netsmb netsmb/smb_dev.c optional netsmb netsmb/smb_iod.c optional netsmb netsmb/smb_rq.c optional netsmb netsmb/smb_smb.c optional netsmb netsmb/smb_subr.c optional netsmb netsmb/smb_trantcp.c optional netsmb netsmb/smb_usr.c optional netsmb nfs/bootp_subr.c optional bootp nfsclient | bootp nfscl nfs/krpc_subr.c optional bootp nfsclient | bootp nfscl nfs/nfs_common.c optional nfsclient | nfsserver nfs/nfs_diskless.c optional nfsclient nfs_root | nfscl nfs_root nfs/nfs_lock.c optional nfsclient | nfscl | nfslockd | nfsd nfsclient/nfs_bio.c optional nfsclient nfsclient/nfs_node.c optional nfsclient nfsclient/nfs_krpc.c optional nfsclient nfsclient/nfs_subs.c optional nfsclient nfsclient/nfs_nfsiod.c optional nfsclient nfsclient/nfs_vfsops.c optional nfsclient nfsclient/nfs_vnops.c optional nfsclient nfsserver/nfs_fha.c optional nfsserver nfsserver/nfs_serv.c optional nfsserver nfsserver/nfs_srvkrpc.c optional nfsserver nfsserver/nfs_srvsubs.c optional nfsserver nfs/nfs_nfssvc.c optional nfsserver | nfscl | nfsd nlm/nlm_advlock.c optional nfslockd | nfsd nlm/nlm_prot_clnt.c optional nfslockd | nfsd nlm/nlm_prot_impl.c optional nfslockd | nfsd nlm/nlm_prot_server.c optional nfslockd | nfsd nlm/nlm_prot_svc.c optional nfslockd | nfsd nlm/nlm_prot_xdr.c optional nfslockd | nfsd nlm/sm_inter_xdr.c optional nfslockd | nfsd # OpenFabrics Enterprise Distribution (Infiniband) ofed/include/linux/linux_compat.c optional ofed \ no-depend compile-with "${OFED_C}" ofed/include/linux/linux_idr.c optional ofed \ no-depend compile-with "${OFED_C}" ofed/include/linux/linux_radix.c optional ofed \ no-depend compile-with "${OFED_C}" ofed/drivers/infiniband/core/addr.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/agent.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/cache.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" # XXX Mad.c must be ordered before cm.c for sysinit sets to occur in # the correct order. ofed/drivers/infiniband/core/mad.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/cm.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/cma.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/device.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/fmr_pool.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/iwcm.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/local_sa.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/mad_rmpp.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/multicast.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/notice.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/packer.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/sa_query.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/smi.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/sysfs.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/ucm.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/ucma.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/ud_header.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/umem.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/user_mad.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/uverbs_cmd.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/uverbs_main.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/uverbs_marshall.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/verbs.c optional ofed \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/ulp/ipoib/ipoib_cm.c optional ipoib \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" #ofed/drivers/infiniband/ulp/ipoib/ipoib_fs.c optional ipoib \ # no-depend \ # compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" ofed/drivers/infiniband/ulp/ipoib/ipoib_ib.c optional ipoib \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" ofed/drivers/infiniband/ulp/ipoib/ipoib_main.c optional ipoib \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" ofed/drivers/infiniband/ulp/ipoib/ipoib_multicast.c optional ipoib \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" ofed/drivers/infiniband/ulp/ipoib/ipoib_verbs.c optional ipoib \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" #ofed/drivers/infiniband/ulp/ipoib/ipoib_vlan.c optional ipoib \ # no-depend \ # compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" ofed/drivers/infiniband/ulp/sdp/sdp_bcopy.c optional sdp \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/sdp/" ofed/drivers/infiniband/ulp/sdp/sdp_main.c optional sdp \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/sdp/" ofed/drivers/infiniband/ulp/sdp/sdp_rx.c optional sdp \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/sdp/" ofed/drivers/infiniband/ulp/sdp/sdp_cma.c optional sdp \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/sdp/" ofed/drivers/infiniband/ulp/sdp/sdp_tx.c optional sdp \ no-depend \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/sdp/" ofed/drivers/infiniband/hw/mlx4/ah.c optional mlx4ib \ no-depend obj-prefix "mlx4ib_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/infiniband/hw/mlx4/" ofed/drivers/infiniband/hw/mlx4/cq.c optional mlx4ib \ no-depend obj-prefix "mlx4ib_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/infiniband/hw/mlx4/" ofed/drivers/infiniband/hw/mlx4/doorbell.c optional mlx4ib \ no-depend obj-prefix "mlx4ib_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/infiniband/hw/mlx4/" ofed/drivers/infiniband/hw/mlx4/mad.c optional mlx4ib \ no-depend obj-prefix "mlx4ib_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/infiniband/hw/mlx4/" ofed/drivers/infiniband/hw/mlx4/main.c optional mlx4ib \ no-depend obj-prefix "mlx4ib_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/infiniband/hw/mlx4/" ofed/drivers/infiniband/hw/mlx4/mr.c optional mlx4ib \ no-depend obj-prefix "mlx4ib_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/infiniband/hw/mlx4/" ofed/drivers/infiniband/hw/mlx4/qp.c optional mlx4ib \ no-depend obj-prefix "mlx4ib_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/infiniband/hw/mlx4/" ofed/drivers/infiniband/hw/mlx4/srq.c optional mlx4ib \ no-depend obj-prefix "mlx4ib_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/infiniband/hw/mlx4/" ofed/drivers/infiniband/hw/mlx4/wc.c optional mlx4ib \ no-depend obj-prefix "mlx4ib_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/infiniband/hw/mlx4/" ofed/drivers/net/mlx4/alloc.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/catas.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/cmd.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/cq.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/eq.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/fw.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/icm.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/intf.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/main.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/mcg.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/mr.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/pd.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/port.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/profile.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/qp.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/reset.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/sense.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/srq.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/xrcd.c optional mlx4ib | mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/en_cq.c optional mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/en_frag.c optional mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/en_main.c optional mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/en_netdev.c optional mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/en_port.c optional mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/en_resources.c optional mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/en_rx.c optional mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/net/mlx4/en_tx.c optional mlxen \ no-depend obj-prefix "mlx4_" \ compile-with "${OFED_C_NOIMP} -I$S/ofed/drivers/net/mlx4/" ofed/drivers/infiniband/hw/mthca/mthca_allocator.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_av.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_catas.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_cmd.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_cq.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_eq.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_mad.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_main.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_mcg.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_memfree.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_mr.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_pd.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_profile.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_provider.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_qp.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_reset.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_srq.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" ofed/drivers/infiniband/hw/mthca/mthca_uar.c optional mthca \ no-depend compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/mthca/" # crypto support opencrypto/cast.c optional crypto | ipsec opencrypto/criov.c optional crypto opencrypto/crypto.c optional crypto opencrypto/cryptodev.c optional cryptodev opencrypto/cryptodev_if.m optional crypto opencrypto/cryptosoft.c optional crypto opencrypto/deflate.c optional crypto opencrypto/rmd160.c optional crypto | ipsec opencrypto/skipjack.c optional crypto opencrypto/xform.c optional crypto pci/alpm.c optional alpm pci pci/amdpm.c optional amdpm pci | nfpm pci pci/amdsmb.c optional amdsmb pci pci/if_rl.c optional rl pci pci/intpm.c optional intpm pci pci/ncr.c optional ncr pci pci/nfsmb.c optional nfsmb pci pci/viapm.c optional viapm pci rpc/auth_none.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd rpc/auth_unix.c optional krpc | nfslockd | nfsclient | nfscl | nfsd rpc/authunix_prot.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd rpc/clnt_dg.c optional krpc | nfslockd | nfsclient | nfscl | nfsd rpc/clnt_rc.c optional krpc | nfslockd | nfsclient | nfscl | nfsd rpc/clnt_vc.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd rpc/getnetconfig.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd rpc/replay.c optional krpc | nfslockd | nfsserver | nfscl | nfsd rpc/rpc_callmsg.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd rpc/rpc_generic.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd rpc/rpc_prot.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd rpc/rpcb_clnt.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd rpc/rpcb_prot.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd rpc/svc.c optional krpc | nfslockd | nfsserver | nfscl | nfsd rpc/svc_auth.c optional krpc | nfslockd | nfsserver | nfscl | nfsd rpc/svc_auth_unix.c optional krpc | nfslockd | nfsserver | nfscl | nfsd rpc/svc_dg.c optional krpc | nfslockd | nfsserver | nfscl | nfsd rpc/svc_generic.c optional krpc | nfslockd | nfsserver | nfscl | nfsd rpc/svc_vc.c optional krpc | nfslockd | nfsserver | nfscl | nfsd rpc/rpcsec_gss/rpcsec_gss.c optional krpc kgssapi | nfslockd kgssapi | nfscl kgssapi | nfsd kgssapi rpc/rpcsec_gss/rpcsec_gss_conf.c optional krpc kgssapi | nfslockd kgssapi | nfscl kgssapi | nfsd kgssapi rpc/rpcsec_gss/rpcsec_gss_misc.c optional krpc kgssapi | nfslockd kgssapi | nfscl kgssapi | nfsd kgssapi rpc/rpcsec_gss/rpcsec_gss_prot.c optional krpc kgssapi | nfslockd kgssapi | nfscl kgssapi | nfsd kgssapi rpc/rpcsec_gss/svc_rpcsec_gss.c optional krpc kgssapi | nfslockd kgssapi | nfscl kgssapi | nfsd kgssapi security/audit/audit.c optional audit security/audit/audit_arg.c optional audit security/audit/audit_bsm.c optional audit security/audit/audit_bsm_domain.c optional audit security/audit/audit_bsm_errno.c optional audit security/audit/audit_bsm_fcntl.c optional audit security/audit/audit_bsm_klib.c optional audit security/audit/audit_bsm_socket_type.c optional audit security/audit/audit_bsm_token.c optional audit security/audit/audit_pipe.c optional audit security/audit/audit_syscalls.c standard security/audit/audit_trigger.c optional audit security/audit/audit_worker.c optional audit security/mac/mac_atalk.c optional mac netatalk security/mac/mac_audit.c optional mac audit security/mac/mac_cred.c optional mac security/mac/mac_framework.c optional mac security/mac/mac_inet.c optional mac inet | mac inet6 security/mac/mac_inet6.c optional mac inet6 security/mac/mac_label.c optional mac security/mac/mac_net.c optional mac security/mac/mac_pipe.c optional mac security/mac/mac_posix_sem.c optional mac security/mac/mac_posix_shm.c optional mac security/mac/mac_priv.c optional mac security/mac/mac_process.c optional mac security/mac/mac_socket.c optional mac security/mac/mac_syscalls.c standard security/mac/mac_system.c optional mac security/mac/mac_sysv_msg.c optional mac security/mac/mac_sysv_sem.c optional mac security/mac/mac_sysv_shm.c optional mac security/mac/mac_vfs.c optional mac security/mac_biba/mac_biba.c optional mac_biba security/mac_bsdextended/mac_bsdextended.c optional mac_bsdextended security/mac_bsdextended/ugidfw_system.c optional mac_bsdextended security/mac_bsdextended/ugidfw_vnode.c optional mac_bsdextended security/mac_ifoff/mac_ifoff.c optional mac_ifoff security/mac_lomac/mac_lomac.c optional mac_lomac security/mac_mls/mac_mls.c optional mac_mls security/mac_none/mac_none.c optional mac_none security/mac_partition/mac_partition.c optional mac_partition security/mac_portacl/mac_portacl.c optional mac_portacl security/mac_seeotheruids/mac_seeotheruids.c optional mac_seeotheruids security/mac_stub/mac_stub.c optional mac_stub security/mac_test/mac_test.c optional mac_test teken/teken.c optional sc ufs/ffs/ffs_alloc.c optional ffs ufs/ffs/ffs_balloc.c optional ffs ufs/ffs/ffs_inode.c optional ffs ufs/ffs/ffs_snapshot.c optional ffs ufs/ffs/ffs_softdep.c optional ffs ufs/ffs/ffs_subr.c optional ffs ufs/ffs/ffs_tables.c optional ffs ufs/ffs/ffs_vfsops.c optional ffs ufs/ffs/ffs_vnops.c optional ffs ufs/ffs/ffs_rawread.c optional directio ufs/ufs/ufs_acl.c optional ffs ufs/ufs/ufs_bmap.c optional ffs ufs/ufs/ufs_dirhash.c optional ffs ufs/ufs/ufs_extattr.c optional ffs ufs/ufs/ufs_gjournal.c optional ffs UFS_GJOURNAL ufs/ufs/ufs_inode.c optional ffs ufs/ufs/ufs_lookup.c optional ffs ufs/ufs/ufs_quota.c optional ffs ufs/ufs/ufs_vfsops.c optional ffs ufs/ufs/ufs_vnops.c optional ffs vm/default_pager.c standard vm/device_pager.c standard vm/phys_pager.c standard vm/redzone.c optional DEBUG_REDZONE vm/sg_pager.c standard vm/swap_pager.c standard vm/uma_core.c standard vm/uma_dbg.c standard vm/vm_contig.c standard vm/memguard.c optional DEBUG_MEMGUARD vm/vm_fault.c standard vm/vm_glue.c standard vm/vm_init.c standard vm/vm_kern.c standard vm/vm_map.c standard vm/vm_meter.c standard vm/vm_mmap.c standard vm/vm_object.c standard vm/vm_page.c standard vm/vm_pageout.c standard vm/vm_pager.c standard vm/vm_phys.c standard vm/vm_reserv.c standard vm/vm_unix.c standard vm/vm_zeroidle.c standard vm/vnode_pager.c standard xdr/xdr.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd xdr/xdr_array.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd xdr/xdr_mbuf.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd xdr/xdr_mem.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd xdr/xdr_reference.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd xdr/xdr_sizeof.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd # gnu/fs/xfs/xfs_alloc.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" \ warning "kernel contains GPL contaminated xfs filesystem" gnu/fs/xfs/xfs_alloc_btree.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_bit.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_bmap.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_bmap_btree.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_btree.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_buf_item.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_da_btree.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_dir.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_dir2.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_dir2_block.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_dir2_data.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_dir2_leaf.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_dir2_node.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_dir2_sf.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_dir2_trace.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_dir_leaf.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_error.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_extfree_item.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_fsops.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_ialloc.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_ialloc_btree.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_inode.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_inode_item.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_iocore.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_itable.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_dfrag.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_log.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_log_recover.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_mount.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_rename.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_trans.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_trans_ail.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_trans_buf.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_trans_extfree.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_trans_inode.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_trans_item.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_utils.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_vfsops.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_vnodeops.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_rw.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_attr_leaf.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_attr.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_dmops.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_qmops.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_iget.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_freebsd_iget.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_mountops.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_vnops.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_frw.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_buf.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_globals.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_dmistubs.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_super.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_stats.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_vfs.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_vnode.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_sysctl.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_fs_subr.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/xfs_ioctl.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/support/debug.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/support/ktrace.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/support/mrlock.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/support/uuid.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/FreeBSD/support/kmem.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_iomap.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" gnu/fs/xfs/xfs_behavior.c optional xfs \ compile-with "${NORMAL_C} -I$S/gnu/fs/xfs/FreeBSD -I$S/gnu/fs/xfs/FreeBSD/support -I$S/gnu/fs/xfs" xen/gnttab.c optional xen | xenhvm xen/features.c optional xen | xenhvm xen/evtchn/evtchn.c optional xen xen/evtchn/evtchn_dev.c optional xen | xenhvm xen/xenbus/xenbus_if.m optional xen | xenhvm xen/xenbus/xenbus.c optional xen | xenhvm xen/xenbus/xenbusb_if.m optional xen | xenhvm xen/xenbus/xenbusb.c optional xen | xenhvm xen/xenbus/xenbusb_front.c optional xen | xenhvm xen/xenbus/xenbusb_back.c optional xen | xenhvm xen/xenstore/xenstore.c optional xen | xenhvm xen/xenstore/xenstore_dev.c optional xen | xenhvm dev/xen/balloon/balloon.c optional xen | xenhvm dev/xen/blkfront/blkfront.c optional xen | xenhvm dev/xen/blkback/blkback.c optional xen | xenhvm dev/xen/console/console.c optional xen dev/xen/console/xencons_ring.c optional xen dev/xen/control/control.c optional xen | xenhvm dev/xen/netfront/netfront.c optional xen | xenhvm dev/xen/xenpci/xenpci.c optional xenpci dev/xen/xenpci/evtchn.c optional xenpci Index: projects/largeSMP/sys/conf/options =================================================================== --- projects/largeSMP/sys/conf/options (revision 222782) +++ projects/largeSMP/sys/conf/options (revision 222783) @@ -1,898 +1,899 @@ # $FreeBSD$ # # On the handling of kernel options # # All kernel options should be listed in NOTES, with suitable # descriptions. Negative options (options that make some code not # compile) should be commented out; LINT (generated from NOTES) should # compile as much code as possible. Try to structure option-using # code so that a single option only switch code on, or only switch # code off, to make it possible to have a full compile-test. If # necessary, you can check for COMPILING_LINT to get maximum code # coverage. # # All new options shall also be listed in either "conf/options" or # "conf/options.". Options that affect a single source-file # .[c|s] should be directed into "opt_.h", while options # that affect multiple files should either go in "opt_global.h" if # this is a kernel-wide option (used just about everywhere), or in # "opt_.h" if it affects only some files. # Note that the effect of listing only an option without a # header-file-name in conf/options (and cousins) is that the last # convention is followed. # # This handling scheme is not yet fully implemented. # # # Format of this file: # Option name filename # # If filename is missing, the default is # opt_.h AAC_DEBUG opt_aac.h AHC_ALLOW_MEMIO opt_aic7xxx.h AHC_TMODE_ENABLE opt_aic7xxx.h AHC_DUMP_EEPROM opt_aic7xxx.h AHC_DEBUG opt_aic7xxx.h AHC_DEBUG_OPTS opt_aic7xxx.h AHC_REG_PRETTY_PRINT opt_aic7xxx.h AHD_DEBUG opt_aic79xx.h AHD_DEBUG_OPTS opt_aic79xx.h AHD_TMODE_ENABLE opt_aic79xx.h AHD_REG_PRETTY_PRINT opt_aic79xx.h ADW_ALLOW_MEMIO opt_adw.h TWA_DEBUG opt_twa.h TWA_FLASH_FIRMWARE opt_twa.h # Debugging options. DDB DDB_BUFR_SIZE opt_ddb.h DDB_CAPTURE_DEFAULTBUFSIZE opt_ddb.h DDB_CAPTURE_MAXBUFSIZE opt_ddb.h DDB_CTF opt_ddb.h DDB_NUMSYM opt_ddb.h GDB KDB opt_global.h KDB_TRACE opt_kdb.h KDB_UNATTENDED opt_kdb.h SYSCTL_DEBUG opt_sysctl.h # Miscellaneous options. ADAPTIVE_LOCKMGRS ALQ AUDIT opt_global.h CAPABILITIES opt_capabilities.h CODA_COMPAT_5 opt_coda.h COMPAT_43 opt_compat.h COMPAT_43TTY opt_compat.h COMPAT_FREEBSD4 opt_compat.h COMPAT_FREEBSD5 opt_compat.h COMPAT_FREEBSD6 opt_compat.h COMPAT_FREEBSD7 opt_compat.h COMPILING_LINT opt_global.h COMPRESS_USER_CORES opt_core.h CY_PCI_FASTINTR DEADLKRES opt_watchdog.h DIRECTIO FULL_PREEMPTION opt_sched.h IPI_PREEMPTION opt_sched.h GEOM_AES opt_geom.h GEOM_BDE opt_geom.h GEOM_BSD opt_geom.h GEOM_CACHE opt_geom.h GEOM_CONCAT opt_geom.h GEOM_ELI opt_geom.h GEOM_FOX opt_geom.h GEOM_GATE opt_geom.h GEOM_JOURNAL opt_geom.h GEOM_LABEL opt_geom.h GEOM_LINUX_LVM opt_geom.h GEOM_MBR opt_geom.h GEOM_MIRROR opt_geom.h GEOM_MULTIPATH opt_geom.h GEOM_NOP opt_geom.h GEOM_PART_APM opt_geom.h GEOM_PART_BSD opt_geom.h GEOM_PART_EBR opt_geom.h GEOM_PART_EBR_COMPAT opt_geom.h GEOM_PART_GPT opt_geom.h GEOM_PART_MBR opt_geom.h GEOM_PART_PC98 opt_geom.h GEOM_PART_VTOC8 opt_geom.h GEOM_PC98 opt_geom.h GEOM_RAID opt_geom.h GEOM_RAID3 opt_geom.h GEOM_SHSEC opt_geom.h GEOM_STRIPE opt_geom.h GEOM_SUNLABEL opt_geom.h GEOM_UZIP opt_geom.h GEOM_VIRSTOR opt_geom.h GEOM_VOL opt_geom.h GEOM_ZERO opt_geom.h KDTRACE_HOOKS opt_kdtrace.h KSTACK_MAX_PAGES KSTACK_PAGES KTRACE KTRACE_REQUEST_POOL opt_ktrace.h LIBICONV MAC opt_global.h MAC_BIBA opt_dontuse.h MAC_BSDEXTENDED opt_dontuse.h MAC_IFOFF opt_dontuse.h MAC_LOMAC opt_dontuse.h MAC_MLS opt_dontuse.h MAC_NONE opt_dontuse.h MAC_PARTITION opt_dontuse.h MAC_PORTACL opt_dontuse.h MAC_SEEOTHERUIDS opt_dontuse.h MAC_STATIC opt_mac.h MAC_STUB opt_dontuse.h MAC_TEST opt_dontuse.h MD_ROOT opt_md.h MD_ROOT_SIZE opt_md.h MFI_DEBUG opt_mfi.h MFI_DECODE_LOG opt_mfi.h MPROF_BUFFERS opt_mprof.h MPROF_HASH_SIZE opt_mprof.h NEW_PCIB opt_global.h NO_ADAPTIVE_MUTEXES opt_adaptive_mutexes.h NO_ADAPTIVE_RWLOCKS NO_ADAPTIVE_SX NO_SYSCTL_DESCR opt_global.h NSWBUF_MIN opt_swap.h MBUF_PACKET_ZONE_DISABLE opt_global.h PANIC_REBOOT_WAIT_TIME opt_panic.h PPC_DEBUG opt_ppc.h PPC_PROBE_CHIPSET opt_ppc.h PPS_SYNC opt_ntp.h PREEMPTION opt_sched.h QUOTA SCHED_4BSD opt_sched.h SCHED_STATS opt_sched.h SCHED_ULE opt_sched.h SHOW_BUSYBUFS SLEEPQUEUE_PROFILING SLHCI_DEBUG opt_slhci.h SPX_HACK STACK opt_stack.h SUIDDIR MSGMNB opt_sysvipc.h MSGMNI opt_sysvipc.h MSGSEG opt_sysvipc.h MSGSSZ opt_sysvipc.h MSGTQL opt_sysvipc.h SEMMAP opt_sysvipc.h SEMMNI opt_sysvipc.h SEMMNS opt_sysvipc.h SEMMNU opt_sysvipc.h SEMMSL opt_sysvipc.h SEMOPM opt_sysvipc.h SEMUME opt_sysvipc.h SHMALL opt_sysvipc.h SHMMAX opt_sysvipc.h SHMMAXPGS opt_sysvipc.h SHMMIN opt_sysvipc.h SHMMNI opt_sysvipc.h SHMSEG opt_sysvipc.h SYSVMSG opt_sysvipc.h SYSVSEM opt_sysvipc.h SYSVSHM opt_sysvipc.h SW_WATCHDOG opt_watchdog.h TURNSTILE_PROFILING VFS_AIO VERBOSE_SYSINIT opt_global.h WLCACHE opt_wavelan.h WLDEBUG opt_wavelan.h # POSIX kernel options P1003_1B_MQUEUE opt_posix.h P1003_1B_SEMAPHORES opt_posix.h _KPOSIX_PRIORITY_SCHEDULING opt_posix.h # Do we want the config file compiled into the kernel? INCLUDE_CONFIG_FILE opt_config.h # Options for static filesystems. These should only be used at config # time, since the corresponding lkms cannot work if there are any static # dependencies. Unusability is enforced by hiding the defines for the # options in a never-included header. CD9660 opt_dontuse.h CODA opt_dontuse.h EXT2FS opt_dontuse.h FDESCFS opt_dontuse.h FFS opt_dontuse.h HPFS opt_dontuse.h MSDOSFS opt_dontuse.h NTFS opt_dontuse.h NULLFS opt_dontuse.h NWFS opt_dontuse.h PORTALFS opt_dontuse.h PROCFS opt_dontuse.h PSEUDOFS opt_dontuse.h REISERFS opt_dontuse.h SMBFS opt_dontuse.h TMPFS opt_dontuse.h UDF opt_dontuse.h UNIONFS opt_dontuse.h # Pseudofs debugging PSEUDOFS_TRACE opt_pseudofs.h # In-kernel GSS-API KGSSAPI opt_kgssapi.h KGSSAPI_DEBUG opt_kgssapi.h # These static filesystems have one slightly bogus static dependency in # sys/i386/i386/autoconf.c. If any of these filesystems are # statically compiled into the kernel, code for mounting them as root # filesystems will be enabled - but look below. NFSCLIENT opt_nfs.h NFSSERVER opt_nfs.h # Use these options to compile the experimental nfs client and/or # server that supports NFSv4 into a kernel. # NFSCL - client # NFSD - server NFSCL opt_nfs.h NFSD opt_nfs.h # filesystems and libiconv bridge CD9660_ICONV opt_dontuse.h MSDOSFS_ICONV opt_dontuse.h NTFS_ICONV opt_dontuse.h UDF_ICONV opt_dontuse.h # If you are following the conditions in the copyright, # you can enable soft-updates which will speed up a lot of thigs # and make the system safer from crashes at the same time. # otherwise a STUB module will be compiled in. SOFTUPDATES opt_ffs.h # On small, embedded systems, it can be useful to turn off support for # snapshots. It saves about 30-40k for a feature that would be lightly # used, if it is used at all. NO_FFS_SNAPSHOT opt_ffs.h # Enabling this option turns on support for Access Control Lists in UFS, # which can be used to support high security configurations. Depends on # UFS_EXTATTR. UFS_ACL opt_ufs.h # Enabling this option turns on support for extended attributes in UFS-based # filesystems, which can be used to support high security configurations # as well as new filesystem features. UFS_EXTATTR opt_ufs.h UFS_EXTATTR_AUTOSTART opt_ufs.h # Enable fast hash lookups for large directories on UFS-based filesystems. UFS_DIRHASH opt_ufs.h # Enable gjournal-based UFS journal. UFS_GJOURNAL opt_ufs.h # The below sentence is not in English, and neither is this one. # We plan to remove the static dependences above, with a # _ROOT option to control if it usable as root. This list # allows these options to be present in config files already (though # they won't make any difference yet). NFS_ROOT opt_nfsroot.h # SMB/CIFS requester NETSMB opt_netsmb.h # Options used only in subr_param.c. HZ opt_param.h MAXFILES opt_param.h NBUF opt_param.h NSFBUFS opt_param.h VM_BCACHE_SIZE_MAX opt_param.h VM_SWZONE_SIZE_MAX opt_param.h MAXUSERS DFLDSIZ opt_param.h MAXDSIZ opt_param.h MAXSSIZ opt_param.h # Generic SCSI options. CAM_MAX_HIGHPOWER opt_cam.h CAMDEBUG opt_cam.h CAM_DEBUG_DELAY opt_cam.h CAM_DEBUG_BUS opt_cam.h CAM_DEBUG_TARGET opt_cam.h CAM_DEBUG_LUN opt_cam.h CAM_DEBUG_FLAGS opt_cam.h SCSI_DELAY opt_scsi.h SCSI_NO_SENSE_STRINGS opt_scsi.h SCSI_NO_OP_STRINGS opt_scsi.h # Options used only in cam/ata/ata_da.c ADA_TEST_FAILURE opt_ada.h # Options used only in cam/scsi/scsi_cd.c CHANGER_MIN_BUSY_SECONDS opt_cd.h CHANGER_MAX_BUSY_SECONDS opt_cd.h # Options used only in cam/scsi/scsi_sa.c. SA_IO_TIMEOUT opt_sa.h SA_SPACE_TIMEOUT opt_sa.h SA_REWIND_TIMEOUT opt_sa.h SA_ERASE_TIMEOUT opt_sa.h SA_1FM_AT_EOD opt_sa.h # Options used only in cam/scsi/scsi_pt.c SCSI_PT_DEFAULT_TIMEOUT opt_pt.h # Options used only in cam/scsi/scsi_ses.c SES_ENABLE_PASSTHROUGH opt_ses.h # Options used in dev/sym/ (Symbios SCSI driver). SYM_SETUP_LP_PROBE_MAP opt_sym.h #-Low Priority Probe Map (bits) # Allows the ncr to take precedence # 1 (1<<0) -> 810a, 860 # 2 (1<<1) -> 825a, 875, 885, 895 # 4 (1<<2) -> 895a, 896, 1510d SYM_SETUP_SCSI_DIFF opt_sym.h #-HVD support for 825a, 875, 885 # disabled:0 (default), enabled:1 SYM_SETUP_PCI_PARITY opt_sym.h #-PCI parity checking # disabled:0, enabled:1 (default) SYM_SETUP_MAX_LUN opt_sym.h #-Number of LUNs supported # default:8, range:[1..64] # Options used only in pci/ncr.c SCSI_NCR_DEBUG opt_ncr.h SCSI_NCR_MAX_SYNC opt_ncr.h SCSI_NCR_MAX_WIDE opt_ncr.h SCSI_NCR_MYADDR opt_ncr.h # Options used only in dev/isp/* ISP_TARGET_MODE opt_isp.h ISP_FW_CRASH_DUMP opt_isp.h ISP_DEFAULT_ROLES opt_isp.h ISP_INTERNAL_TARGET opt_isp.h # Options used only in dev/iscsi ISCSI_INITIATOR_DEBUG opt_iscsi_initiator.h # Options used in the 'ata' ATA/ATAPI driver ATA_STATIC_ID opt_ata.h ATA_NOPCI opt_ata.h ATA_REQUEST_TIMEOUT opt_ata.h ATA_CAM opt_ata.h # Net stuff. ACCEPT_FILTER_DATA ACCEPT_FILTER_DNS ACCEPT_FILTER_HTTP ALTQ opt_global.h ALTQ_CBQ opt_altq.h ALTQ_CDNR opt_altq.h ALTQ_DEBUG opt_altq.h ALTQ_HFSC opt_altq.h ALTQ_NOPCC opt_altq.h ALTQ_PRIQ opt_altq.h ALTQ_RED opt_altq.h ALTQ_RIO opt_altq.h BOOTP opt_bootp.h BOOTP_BLOCKSIZE opt_bootp.h BOOTP_COMPAT opt_bootp.h BOOTP_NFSROOT opt_bootp.h BOOTP_NFSV3 opt_bootp.h BOOTP_WIRED_TO opt_bootp.h DEVICE_POLLING DEV_ENC opt_enc.h DEV_PF opt_pf.h DEV_PFLOG opt_pf.h DEV_PFSYNC opt_pf.h DEV_VLAN opt_vlan.h DUMMYNET opt_ipdn.h ETHER_8022 opt_ef.h ETHER_8023 opt_ef.h ETHER_II opt_ef.h ETHER_SNAP opt_ef.h INET opt_inet.h INET6 opt_inet6.h IPDIVERT IPFILTER opt_ipfilter.h IPFILTER_DEFAULT_BLOCK opt_ipfilter.h IPFILTER_LOG opt_ipfilter.h IPFILTER_LOOKUP opt_ipfilter.h IPFIREWALL opt_ipfw.h IPFIREWALL_DEFAULT_TO_ACCEPT opt_ipfw.h IPFIREWALL_FORWARD opt_ipfw.h IPFIREWALL_NAT opt_ipfw.h IPFIREWALL_VERBOSE opt_ipfw.h IPFIREWALL_VERBOSE_LIMIT opt_ipfw.h IPSEC opt_ipsec.h IPSEC_DEBUG opt_ipsec.h IPSEC_FILTERTUNNEL opt_ipsec.h IPSEC_NAT_T opt_ipsec.h IPSTEALTH IPX KRPC LIBALIAS LIBMBPOOL LIBMCHAIN MBUF_PROFILING MBUF_STRESS_TEST MROUTING opt_mrouting.h NCP NETATALK opt_atalk.h NFSLOCKD +PCBGROUP opt_pcbgroup.h RADIX_MPATH opt_mpath.h ROUTETABLES opt_route.h SLIP_IFF_OPTS opt_slip.h TCPDEBUG TCP_OFFLOAD_DISABLE opt_inet.h #Disable code to dispatch tcp offloading TCP_SIGNATURE opt_inet.h TCP_SORECEIVE_STREAM opt_inet.h VLAN_ARRAY opt_vlan.h XBONEHACK FLOWTABLE opt_route.h # # SCTP # SCTP opt_sctp.h SCTP_DEBUG opt_sctp.h # Enable debug printfs SCTP_WITH_NO_CSUM opt_sctp.h # Use this at your peril SCTP_LOCK_LOGGING opt_sctp.h # Log to KTR lock activity SCTP_MBUF_LOGGING opt_sctp.h # Log to KTR general mbuf aloc/free SCTP_MBCNT_LOGGING opt_sctp.h # Log to KTR mbcnt activity SCTP_PACKET_LOGGING opt_sctp.h # Log to a packet buffer last N packets SCTP_LTRACE_CHUNKS opt_sctp.h # Log to KTR chunks processed SCTP_LTRACE_ERRORS opt_sctp.h # Log to KTR error returns. SCTP_USE_PERCPU_STAT opt_sctp.h # Use per cpu stats. SCTP_MCORE_INPUT opt_sctp.h # Have multiple input threads for input mbufs # # # # Netgraph(4). Use option NETGRAPH to enable the base netgraph code. # Each netgraph node type can be either be compiled into the kernel # or loaded dynamically. To get the former, include the corresponding # option below. Each type has its own man page, e.g. ng_async(4). NETGRAPH NETGRAPH_DEBUG opt_netgraph.h NETGRAPH_ASYNC opt_netgraph.h NETGRAPH_ATMLLC opt_netgraph.h NETGRAPH_ATM_ATMPIF opt_netgraph.h NETGRAPH_BLUETOOTH opt_netgraph.h NETGRAPH_BLUETOOTH_BT3C opt_netgraph.h NETGRAPH_BLUETOOTH_H4 opt_netgraph.h NETGRAPH_BLUETOOTH_HCI opt_netgraph.h NETGRAPH_BLUETOOTH_L2CAP opt_netgraph.h NETGRAPH_BLUETOOTH_SOCKET opt_netgraph.h NETGRAPH_BLUETOOTH_UBT opt_netgraph.h NETGRAPH_BLUETOOTH_UBTBCMFW opt_netgraph.h NETGRAPH_BPF opt_netgraph.h NETGRAPH_BRIDGE opt_netgraph.h NETGRAPH_CAR opt_netgraph.h NETGRAPH_CISCO opt_netgraph.h NETGRAPH_DEFLATE opt_netgraph.h NETGRAPH_DEVICE opt_netgraph.h NETGRAPH_ECHO opt_netgraph.h NETGRAPH_EIFACE opt_netgraph.h NETGRAPH_ETHER opt_netgraph.h NETGRAPH_ETHER_ECHO opt_netgraph.h NETGRAPH_FEC opt_netgraph.h NETGRAPH_FRAME_RELAY opt_netgraph.h NETGRAPH_GIF opt_netgraph.h NETGRAPH_GIF_DEMUX opt_netgraph.h NETGRAPH_HOLE opt_netgraph.h NETGRAPH_IFACE opt_netgraph.h NETGRAPH_IP_INPUT opt_netgraph.h NETGRAPH_IPFW opt_netgraph.h NETGRAPH_KSOCKET opt_netgraph.h NETGRAPH_L2TP opt_netgraph.h NETGRAPH_LMI opt_netgraph.h # MPPC compression requires proprietary files (not included) NETGRAPH_MPPC_COMPRESSION opt_netgraph.h NETGRAPH_MPPC_ENCRYPTION opt_netgraph.h NETGRAPH_NAT opt_netgraph.h NETGRAPH_NETFLOW opt_netgraph.h NETGRAPH_ONE2MANY opt_netgraph.h NETGRAPH_PATCH opt_netgraph.h NETGRAPH_PIPE opt_netgraph.h NETGRAPH_PPP opt_netgraph.h NETGRAPH_PPPOE opt_netgraph.h NETGRAPH_PPTPGRE opt_netgraph.h NETGRAPH_PRED1 opt_netgraph.h NETGRAPH_RFC1490 opt_netgraph.h NETGRAPH_SOCKET opt_netgraph.h NETGRAPH_SPLIT opt_netgraph.h NETGRAPH_SPPP opt_netgraph.h NETGRAPH_TAG opt_netgraph.h NETGRAPH_TCPMSS opt_netgraph.h NETGRAPH_TEE opt_netgraph.h NETGRAPH_TTY opt_netgraph.h NETGRAPH_UI opt_netgraph.h NETGRAPH_VJC opt_netgraph.h NETGRAPH_VLAN opt_netgraph.h # NgATM options NGATM_ATM opt_netgraph.h NGATM_ATMBASE opt_netgraph.h NGATM_SSCOP opt_netgraph.h NGATM_SSCFU opt_netgraph.h NGATM_UNI opt_netgraph.h NGATM_CCATM opt_netgraph.h # DRM options DRM_DEBUG opt_drm.h ZERO_COPY_SOCKETS opt_zero.h TI_PRIVATE_JUMBOS opt_ti.h TI_JUMBO_HDRSPLIT opt_ti.h BCE_JUMBO_HDRSPLIT opt_bce.h # XXX Conflict: # of devices vs network protocol (Native ATM). # This makes "atm.h" unusable. NATM # DPT driver debug flags DPT_MEASURE_PERFORMANCE opt_dpt.h DPT_HANDLE_TIMEOUTS opt_dpt.h DPT_TIMEOUT_FACTOR opt_dpt.h DPT_LOST_IRQ opt_dpt.h DPT_RESET_HBA opt_dpt.h # Misc debug flags. Most of these should probably be replaced with # 'DEBUG', and then let people recompile just the interesting modules # with 'make CC="cc -DDEBUG"'. CLUSTERDEBUG opt_debug_cluster.h DEBUG_1284 opt_ppb_1284.h VP0_DEBUG opt_vpo.h LPT_DEBUG opt_lpt.h PLIP_DEBUG opt_plip.h LOCKF_DEBUG opt_debug_lockf.h NETATALKDEBUG opt_atalk.h SI_DEBUG opt_debug_si.h # Fb options FB_DEBUG opt_fb.h FB_INSTALL_CDEV opt_fb.h # ppbus related options PERIPH_1284 opt_ppb_1284.h DONTPROBE_1284 opt_ppb_1284.h # smbus related options ENABLE_ALART opt_intpm.h # These cause changes all over the kernel BLKDEV_IOSIZE opt_global.h BURN_BRIDGES opt_global.h DEBUG opt_global.h DEBUG_LOCKS opt_global.h DEBUG_VFS_LOCKS opt_global.h DFLTPHYS opt_global.h DIAGNOSTIC opt_global.h INVARIANT_SUPPORT opt_global.h INVARIANTS opt_global.h MAXPHYS opt_global.h MCLSHIFT opt_global.h MUTEX_DEBUG opt_global.h MUTEX_NOINLINE opt_global.h LOCK_PROFILING opt_global.h LOCK_PROFILING_FAST opt_global.h MSIZE opt_global.h REGRESSION opt_global.h RESTARTABLE_PANICS opt_global.h RWLOCK_NOINLINE opt_global.h SX_NOINLINE opt_global.h VFS_BIO_DEBUG opt_global.h # These are VM related options VM_KMEM_SIZE opt_vm.h VM_KMEM_SIZE_SCALE opt_vm.h VM_KMEM_SIZE_MAX opt_vm.h VM_NRESERVLEVEL opt_vm.h VM_LEVEL_0_ORDER opt_vm.h NO_SWAPPING opt_vm.h MALLOC_MAKE_FAILURES opt_vm.h MALLOC_PROFILE opt_vm.h MALLOC_DEBUG_MAXZONES opt_vm.h # The MemGuard replacement allocator used for tamper-after-free detection DEBUG_MEMGUARD opt_vm.h # The RedZone malloc(9) protection DEBUG_REDZONE opt_vm.h # Standard SMP options SMP opt_global.h # Size of the kernel message buffer MSGBUF_SIZE opt_msgbuf.h # NFS options NFS_MINATTRTIMO opt_nfs.h NFS_MAXATTRTIMO opt_nfs.h NFS_MINDIRATTRTIMO opt_nfs.h NFS_MAXDIRATTRTIMO opt_nfs.h NFS_GATHERDELAY opt_nfs.h NFS_WDELAYHASHSIZ opt_nfs.h NFS_DEBUG opt_nfs.h # For the Bt848/Bt848A/Bt849/Bt878/Bt879 driver OVERRIDE_CARD opt_bktr.h OVERRIDE_TUNER opt_bktr.h OVERRIDE_DBX opt_bktr.h OVERRIDE_MSP opt_bktr.h BROOKTREE_SYSTEM_DEFAULT opt_bktr.h BROOKTREE_ALLOC_PAGES opt_bktr.h BKTR_OVERRIDE_CARD opt_bktr.h BKTR_OVERRIDE_TUNER opt_bktr.h BKTR_OVERRIDE_DBX opt_bktr.h BKTR_OVERRIDE_MSP opt_bktr.h BKTR_SYSTEM_DEFAULT opt_bktr.h BKTR_ALLOC_PAGES opt_bktr.h BKTR_USE_PLL opt_bktr.h BKTR_GPIO_ACCESS opt_bktr.h BKTR_NO_MSP_RESET opt_bktr.h BKTR_430_FX_MODE opt_bktr.h BKTR_SIS_VIA_MODE opt_bktr.h BKTR_USE_FREEBSD_SMBUS opt_bktr.h BKTR_NEW_MSP34XX_DRIVER opt_bktr.h BREAK_TO_DEBUGGER opt_comconsole.h ALT_BREAK_TO_DEBUGGER opt_comconsole.h # Options to support PPS UART_PPS_ON_CTS opt_uart.h # options for bus/device framework BUS_DEBUG opt_bus.h # options for USB support USB_DEBUG opt_usb.h USB_HOST_ALIGN opt_usb.h USB_REQ_DEBUG opt_usb.h USB_VERBOSE opt_usb.h USB_EHCI_BIG_ENDIAN_DESC opt_usb.h U3G_DEBUG opt_u3g.h UKBD_DFLT_KEYMAP opt_ukbd.h UPLCOM_INTR_INTERVAL opt_uplcom.h UVSCOM_DEFAULT_OPKTSIZE opt_uvscom.h UVSCOM_INTR_INTERVAL opt_uvscom.h # Embedded system options INIT_PATH ROOTDEVNAME FDC_DEBUG opt_fdc.h PCFCLOCK_VERBOSE opt_pcfclock.h PCFCLOCK_MAX_RETRIES opt_pcfclock.h KTR opt_global.h KTR_ALQ opt_ktr.h KTR_MASK opt_ktr.h KTR_CPUMASK opt_ktr.h KTR_COMPILE opt_global.h KTR_ENTRIES opt_global.h KTR_VERBOSE opt_ktr.h WITNESS opt_global.h WITNESS_KDB opt_witness.h WITNESS_SKIPSPIN opt_witness.h # options for ACPI support ACPI_DEBUG opt_acpi.h ACPI_MAX_TASKS opt_acpi.h ACPI_MAX_THREADS opt_acpi.h # ISA support DEV_ISA opt_isa.h ISAPNP opt_isa.h # various 'device presence' options. DEV_BPF opt_bpf.h DEV_MCA opt_mca.h DEV_CARP opt_carp.h DEV_SPLASH opt_splash.h # EISA support DEV_EISA opt_eisa.h EISA_SLOTS opt_eisa.h # ed driver ED_HPP opt_ed.h ED_3C503 opt_ed.h ED_SIC opt_ed.h # bce driver BCE_DEBUG opt_bce.h BCE_NVRAM_WRITE_SUPPORT opt_bce.h # bxe driver BXE_DEBUG opt_bxe.h BXE_NVRAM_WRITE_SUPPORT opt_bxe.h SOCKBUF_DEBUG opt_global.h # options for ubsec driver UBSEC_DEBUG opt_ubsec.h UBSEC_RNDTEST opt_ubsec.h UBSEC_NO_RNG opt_ubsec.h # options for hifn driver HIFN_DEBUG opt_hifn.h HIFN_RNDTEST opt_hifn.h # options for safenet driver SAFE_DEBUG opt_safe.h SAFE_NO_RNG opt_safe.h SAFE_RNDTEST opt_safe.h # syscons options MAXCONS opt_syscons.h SC_ALT_MOUSE_IMAGE opt_syscons.h SC_CUT_SPACES2TABS opt_syscons.h SC_CUT_SEPCHARS opt_syscons.h SC_DEBUG_LEVEL opt_syscons.h SC_DFLT_FONT opt_syscons.h SC_DISABLE_KDBKEY opt_syscons.h SC_DISABLE_REBOOT opt_syscons.h SC_HISTORY_SIZE opt_syscons.h SC_KERNEL_CONS_ATTR opt_syscons.h SC_KERNEL_CONS_REV_ATTR opt_syscons.h SC_MOUSE_CHAR opt_syscons.h SC_NO_CUTPASTE opt_syscons.h SC_NO_FONT_LOADING opt_syscons.h SC_NO_HISTORY opt_syscons.h SC_NO_MODE_CHANGE opt_syscons.h SC_NO_SUSPEND_VTYSWITCH opt_syscons.h SC_NO_SYSMOUSE opt_syscons.h SC_NORM_ATTR opt_syscons.h SC_NORM_REV_ATTR opt_syscons.h SC_PIXEL_MODE opt_syscons.h SC_RENDER_DEBUG opt_syscons.h SC_TWOBUTTON_MOUSE opt_syscons.h # teken terminal emulator options TEKEN_CONS25 opt_teken.h TEKEN_UTF8 opt_teken.h # options for printf PRINTF_BUFR_SIZE opt_printf.h # kbd options KBD_DISABLE_KEYMAP_LOAD opt_kbd.h KBD_INSTALL_CDEV opt_kbd.h KBD_MAXRETRY opt_kbd.h KBD_MAXWAIT opt_kbd.h KBD_RESETDELAY opt_kbd.h KBDIO_DEBUG opt_kbd.h # options for the Atheros driver ATH_DEBUG opt_ath.h ATH_TXBUF opt_ath.h ATH_RXBUF opt_ath.h ATH_DIAGAPI opt_ath.h ATH_TX99_DIAG opt_ath.h ATH_ENABLE_11N opt_ah.h # options for the Atheros hal AH_SUPPORT_AR5416 opt_ah.h # XXX For now, this breaks non-AR9130 chipsets, so only use it # XXX when actually targetting AR9130. AH_SUPPORT_AR9130 opt_ah.h AH_DEBUG opt_ah.h AH_ASSERT opt_ah.h AH_DEBUG_ALQ opt_ah.h AH_REGOPS_FUNC opt_ah.h AH_WRITE_REGDOMAIN opt_ah.h AH_DEBUG_COUNTRY opt_ah.h AH_WRITE_EEPROM opt_ah.h AH_PRIVATE_DIAG opt_ah.h AH_NEED_DESC_SWAP opt_ah.h AH_USE_INIPDGAIN opt_ah.h AH_MAXCHAN opt_ah.h AH_RXCFG_SDMAMW_4BYTES opt_ah.h # AR5416 and later interrupt mitigation # XXX do not use this for AR9130 AH_AR5416_INTERRUPT_MITIGATION opt_ah.h # options for the Broadcom BCM43xx driver (bwi) BWI_DEBUG opt_bwi.h BWI_DEBUG_VERBOSE opt_bwi.h # options for the Marvell 8335 wireless driver MALO_DEBUG opt_malo.h MALO_TXBUF opt_malo.h MALO_RXBUF opt_malo.h # options for the Marvell wireless driver MWL_DEBUG opt_mwl.h MWL_TXBUF opt_mwl.h MWL_RXBUF opt_mwl.h MWL_DIAGAPI opt_mwl.h MWL_AGGR_SIZE opt_mwl.h MWL_TX_NODROP opt_mwl.h # dcons options DCONS_BUF_SIZE opt_dcons.h DCONS_POLL_HZ opt_dcons.h DCONS_FORCE_CONSOLE opt_dcons.h DCONS_FORCE_GDB opt_dcons.h # HWPMC options HWPMC_HOOKS # XBOX options for FreeBSD/i386, but some files are MI XBOX opt_xbox.h # XFS XFS # Interrupt filtering INTR_FILTER # 802.11 support layer IEEE80211_DEBUG opt_wlan.h IEEE80211_DEBUG_REFCNT opt_wlan.h IEEE80211_AMPDU_AGE opt_wlan.h IEEE80211_SUPPORT_MESH opt_wlan.h IEEE80211_SUPPORT_SUPERG opt_wlan.h IEEE80211_SUPPORT_TDMA opt_wlan.h IEEE80211_ALQ opt_wlan.h # 802.11 TDMA support TDMA_SLOTLEN_DEFAULT opt_tdma.h TDMA_SLOTCNT_DEFAULT opt_tdma.h TDMA_BINTVAL_DEFAULT opt_tdma.h TDMA_TXRATE_11B_DEFAULT opt_tdma.h TDMA_TXRATE_11G_DEFAULT opt_tdma.h TDMA_TXRATE_11A_DEFAULT opt_tdma.h TDMA_TXRATE_TURBO_DEFAULT opt_tdma.h TDMA_TXRATE_HALF_DEFAULT opt_tdma.h TDMA_TXRATE_QUARTER_DEFAULT opt_tdma.h TDMA_TXRATE_11NA_DEFAULT opt_tdma.h TDMA_TXRATE_11NG_DEFAULT opt_tdma.h # Network stack virtualization options VIMAGE opt_global.h VNET_DEBUG opt_global.h # Common Flash Interface (CFI) options CFI_SUPPORT_STRATAFLASH opt_cfi.h CFI_ARMEDANDDANGEROUS opt_cfi.h # Sound options SND_DEBUG opt_snd.h SND_DIAGNOSTIC opt_snd.h SND_FEEDER_MULTIFORMAT opt_snd.h SND_FEEDER_FULL_MULTIFORMAT opt_snd.h SND_FEEDER_RATE_HP opt_snd.h SND_PCM_64 opt_snd.h SND_OLDSTEREO opt_snd.h X86BIOS # Flattened device tree options FDT opt_platform.h FDT_DTB_STATIC opt_platform.h # OFED Infiniband stack OFED opt_ofed.h OFED_DEBUG_INIT opt_ofed.h SDP opt_ofed.h SDP_DEBUG opt_ofed.h IPOIB_DEBUG opt_ofed.h IPOIB_CM opt_ofed.h # Resource Accounting RACCT opt_global.h # Resource Limits RCTL opt_global.h # At least one of the AR71XX ubiquiti boards has a Redboot configuration # that "lies" about the amount of RAM it has. Until a cleaner method is # defined, this option will suffice in overriding what Redboot says. AR71XX_REALMEM opt_global.h Index: projects/largeSMP/sys/conf =================================================================== --- projects/largeSMP/sys/conf (revision 222782) +++ projects/largeSMP/sys/conf (revision 222783) Property changes on: projects/largeSMP/sys/conf ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/conf:r222698-222782 Index: projects/largeSMP/sys/contrib/dev/acpica =================================================================== --- projects/largeSMP/sys/contrib/dev/acpica (revision 222782) +++ projects/largeSMP/sys/contrib/dev/acpica (revision 222783) Property changes on: projects/largeSMP/sys/contrib/dev/acpica ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/contrib/dev/acpica:r222698-222782 Index: projects/largeSMP/sys/contrib/octeon-sdk =================================================================== --- projects/largeSMP/sys/contrib/octeon-sdk (revision 222782) +++ projects/largeSMP/sys/contrib/octeon-sdk (revision 222783) Property changes on: projects/largeSMP/sys/contrib/octeon-sdk ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/contrib/octeon-sdk:r222698-222782 Index: projects/largeSMP/sys/contrib/pf =================================================================== --- projects/largeSMP/sys/contrib/pf (revision 222782) +++ projects/largeSMP/sys/contrib/pf (revision 222783) Property changes on: projects/largeSMP/sys/contrib/pf ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/contrib/pf:r222698-222782 Index: projects/largeSMP/sys/contrib/x86emu =================================================================== --- projects/largeSMP/sys/contrib/x86emu (revision 222782) +++ projects/largeSMP/sys/contrib/x86emu (revision 222783) Property changes on: projects/largeSMP/sys/contrib/x86emu ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/contrib/x86emu:r222698-222782 Index: projects/largeSMP/sys/dev/ath/if_ath.c =================================================================== --- projects/largeSMP/sys/dev/ath/if_ath.c (revision 222782) +++ projects/largeSMP/sys/dev/ath/if_ath.c (revision 222783) @@ -1,5728 +1,5736 @@ /*- * Copyright (c) 2002-2009 Sam Leffler, Errno Consulting * 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 * 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 NONINFRINGEMENT, 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("$FreeBSD$"); /* * Driver for the Atheros Wireless LAN controller. * * This software is derived from work of Atsushi Onoe; his contribution * is greatly appreciated. */ #include "opt_inet.h" #include "opt_ath.h" #include "opt_wlan.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 #ifdef IEEE80211_SUPPORT_SUPERG #include #endif #ifdef IEEE80211_SUPPORT_TDMA #include #endif #include #ifdef INET #include #include #endif #include #include /* XXX for softled */ #include #include #include #include #include #include #include #ifdef ATH_TX99_DIAG #include #endif /* * ATH_BCBUF determines the number of vap's that can transmit * beacons and also (currently) the number of vap's that can * have unique mac addresses/bssid. When staggering beacons * 4 is probably a good max as otherwise the beacons become * very closely spaced and there is limited time for cab q traffic * to go out. You can burst beacons instead but that is not good * for stations in power save and at some point you really want * another radio (and channel). * * The limit on the number of mac addresses is tied to our use of * the U/L bit and tracking addresses in a byte; it would be * worthwhile to allow more for applications like proxy sta. */ CTASSERT(ATH_BCBUF <= 8); static struct ieee80211vap *ath_vap_create(struct ieee80211com *, const char name[IFNAMSIZ], int unit, int opmode, int flags, const uint8_t bssid[IEEE80211_ADDR_LEN], const uint8_t mac[IEEE80211_ADDR_LEN]); static void ath_vap_delete(struct ieee80211vap *); static void ath_init(void *); static void ath_stop_locked(struct ifnet *); static void ath_stop(struct ifnet *); static void ath_start(struct ifnet *); static int ath_reset_vap(struct ieee80211vap *, u_long); static int ath_media_change(struct ifnet *); static void ath_watchdog(void *); static int ath_ioctl(struct ifnet *, u_long, caddr_t); static void ath_fatal_proc(void *, int); static void ath_bmiss_vap(struct ieee80211vap *); static void ath_bmiss_proc(void *, int); static void ath_key_update_begin(struct ieee80211vap *); static void ath_key_update_end(struct ieee80211vap *); static void ath_update_mcast(struct ifnet *); static void ath_update_promisc(struct ifnet *); static void ath_mode_init(struct ath_softc *); static void ath_setslottime(struct ath_softc *); static void ath_updateslot(struct ifnet *); static int ath_beaconq_setup(struct ath_hal *); static int ath_beacon_alloc(struct ath_softc *, struct ieee80211_node *); static void ath_beacon_update(struct ieee80211vap *, int item); static void ath_beacon_setup(struct ath_softc *, struct ath_buf *); static void ath_beacon_proc(void *, int); static struct ath_buf *ath_beacon_generate(struct ath_softc *, struct ieee80211vap *); static void ath_bstuck_proc(void *, int); static void ath_beacon_return(struct ath_softc *, struct ath_buf *); static void ath_beacon_free(struct ath_softc *); static void ath_beacon_config(struct ath_softc *, struct ieee80211vap *); static void ath_descdma_cleanup(struct ath_softc *sc, struct ath_descdma *, ath_bufhead *); static int ath_desc_alloc(struct ath_softc *); static void ath_desc_free(struct ath_softc *); static struct ieee80211_node *ath_node_alloc(struct ieee80211vap *, const uint8_t [IEEE80211_ADDR_LEN]); static void ath_node_free(struct ieee80211_node *); static void ath_node_getsignal(const struct ieee80211_node *, int8_t *, int8_t *); static int ath_rxbuf_init(struct ath_softc *, struct ath_buf *); static void ath_recv_mgmt(struct ieee80211_node *ni, struct mbuf *m, int subtype, int rssi, int nf); static void ath_setdefantenna(struct ath_softc *, u_int); static void ath_rx_proc(void *, int); static void ath_txq_init(struct ath_softc *sc, struct ath_txq *, int); static struct ath_txq *ath_txq_setup(struct ath_softc*, int qtype, int subtype); static int ath_tx_setup(struct ath_softc *, int, int); static int ath_wme_update(struct ieee80211com *); static void ath_tx_cleanupq(struct ath_softc *, struct ath_txq *); static void ath_tx_cleanup(struct ath_softc *); static void ath_tx_proc_q0(void *, int); static void ath_tx_proc_q0123(void *, int); static void ath_tx_proc(void *, int); static void ath_tx_draintxq(struct ath_softc *, struct ath_txq *); static int ath_chan_set(struct ath_softc *, struct ieee80211_channel *); static void ath_draintxq(struct ath_softc *); static void ath_stoprecv(struct ath_softc *); static int ath_startrecv(struct ath_softc *); static void ath_chan_change(struct ath_softc *, struct ieee80211_channel *); static void ath_scan_start(struct ieee80211com *); static void ath_scan_end(struct ieee80211com *); static void ath_set_channel(struct ieee80211com *); static void ath_calibrate(void *); static int ath_newstate(struct ieee80211vap *, enum ieee80211_state, int); static void ath_setup_stationkey(struct ieee80211_node *); static void ath_newassoc(struct ieee80211_node *, int); static int ath_setregdomain(struct ieee80211com *, struct ieee80211_regdomain *, int, struct ieee80211_channel []); static void ath_getradiocaps(struct ieee80211com *, int, int *, struct ieee80211_channel []); static int ath_getchannels(struct ath_softc *); static void ath_led_event(struct ath_softc *, int); static int ath_rate_setup(struct ath_softc *, u_int mode); static void ath_setcurmode(struct ath_softc *, enum ieee80211_phymode); static void ath_announce(struct ath_softc *); static void ath_dfs_tasklet(void *, int); #ifdef IEEE80211_SUPPORT_TDMA static void ath_tdma_settimers(struct ath_softc *sc, u_int32_t nexttbtt, u_int32_t bintval); static void ath_tdma_bintvalsetup(struct ath_softc *sc, const struct ieee80211_tdma_state *tdma); static void ath_tdma_config(struct ath_softc *sc, struct ieee80211vap *vap); static void ath_tdma_update(struct ieee80211_node *ni, const struct ieee80211_tdma_param *tdma, int); static void ath_tdma_beacon_send(struct ath_softc *sc, struct ieee80211vap *vap); static __inline void ath_hal_setcca(struct ath_hal *ah, int ena) { /* * NB: fill me in; this is not provided by default because disabling * CCA in most locales violates regulatory. */ } static __inline int ath_hal_getcca(struct ath_hal *ah) { u_int32_t diag; if (ath_hal_getcapability(ah, HAL_CAP_DIAG, 0, &diag) != HAL_OK) return 1; return ((diag & 0x500000) == 0); } #define TDMA_EP_MULTIPLIER (1<<10) /* pow2 to optimize out * and / */ #define TDMA_LPF_LEN 6 #define TDMA_DUMMY_MARKER 0x127 #define TDMA_EP_MUL(x, mul) ((x) * (mul)) #define TDMA_IN(x) (TDMA_EP_MUL((x), TDMA_EP_MULTIPLIER)) #define TDMA_LPF(x, y, len) \ ((x != TDMA_DUMMY_MARKER) ? (((x) * ((len)-1) + (y)) / (len)) : (y)) #define TDMA_SAMPLE(x, y) do { \ x = TDMA_LPF((x), TDMA_IN(y), TDMA_LPF_LEN); \ } while (0) #define TDMA_EP_RND(x,mul) \ ((((x)%(mul)) >= ((mul)/2)) ? ((x) + ((mul) - 1)) / (mul) : (x)/(mul)) #define TDMA_AVG(x) TDMA_EP_RND(x, TDMA_EP_MULTIPLIER) #endif /* IEEE80211_SUPPORT_TDMA */ SYSCTL_DECL(_hw_ath); /* XXX validate sysctl values */ static int ath_longcalinterval = 30; /* long cals every 30 secs */ SYSCTL_INT(_hw_ath, OID_AUTO, longcal, CTLFLAG_RW, &ath_longcalinterval, 0, "long chip calibration interval (secs)"); static int ath_shortcalinterval = 100; /* short cals every 100 ms */ SYSCTL_INT(_hw_ath, OID_AUTO, shortcal, CTLFLAG_RW, &ath_shortcalinterval, 0, "short chip calibration interval (msecs)"); static int ath_resetcalinterval = 20*60; /* reset cal state 20 mins */ SYSCTL_INT(_hw_ath, OID_AUTO, resetcal, CTLFLAG_RW, &ath_resetcalinterval, 0, "reset chip calibration results (secs)"); static int ath_anicalinterval = 100; /* ANI calibration - 100 msec */ SYSCTL_INT(_hw_ath, OID_AUTO, anical, CTLFLAG_RW, &ath_anicalinterval, 0, "ANI calibration (msecs)"); static int ath_rxbuf = ATH_RXBUF; /* # rx buffers to allocate */ SYSCTL_INT(_hw_ath, OID_AUTO, rxbuf, CTLFLAG_RW, &ath_rxbuf, 0, "rx buffers allocated"); TUNABLE_INT("hw.ath.rxbuf", &ath_rxbuf); static int ath_txbuf = ATH_TXBUF; /* # tx buffers to allocate */ SYSCTL_INT(_hw_ath, OID_AUTO, txbuf, CTLFLAG_RW, &ath_txbuf, 0, "tx buffers allocated"); TUNABLE_INT("hw.ath.txbuf", &ath_txbuf); static int ath_bstuck_threshold = 4; /* max missed beacons */ SYSCTL_INT(_hw_ath, OID_AUTO, bstuck, CTLFLAG_RW, &ath_bstuck_threshold, 0, "max missed beacon xmits before chip reset"); MALLOC_DEFINE(M_ATHDEV, "athdev", "ath driver dma buffers"); #define HAL_MODE_HT20 (HAL_MODE_11NG_HT20 | HAL_MODE_11NA_HT20) #define HAL_MODE_HT40 \ (HAL_MODE_11NG_HT40PLUS | HAL_MODE_11NG_HT40MINUS | \ HAL_MODE_11NA_HT40PLUS | HAL_MODE_11NA_HT40MINUS) int ath_attach(u_int16_t devid, struct ath_softc *sc) { struct ifnet *ifp; struct ieee80211com *ic; struct ath_hal *ah = NULL; HAL_STATUS status; int error = 0, i; u_int wmodes; uint8_t macaddr[IEEE80211_ADDR_LEN]; DPRINTF(sc, ATH_DEBUG_ANY, "%s: devid 0x%x\n", __func__, devid); ifp = sc->sc_ifp = if_alloc(IFT_IEEE80211); if (ifp == NULL) { device_printf(sc->sc_dev, "can not if_alloc()\n"); error = ENOSPC; goto bad; } ic = ifp->if_l2com; /* set these up early for if_printf use */ if_initname(ifp, device_get_name(sc->sc_dev), device_get_unit(sc->sc_dev)); ah = ath_hal_attach(devid, sc, sc->sc_st, sc->sc_sh, sc->sc_eepromdata, &status); if (ah == NULL) { if_printf(ifp, "unable to attach hardware; HAL status %u\n", status); error = ENXIO; goto bad; } sc->sc_ah = ah; sc->sc_invalid = 0; /* ready to go, enable interrupt handling */ #ifdef ATH_DEBUG sc->sc_debug = ath_debug; #endif /* * Check if the MAC has multi-rate retry support. * We do this by trying to setup a fake extended * descriptor. MAC's that don't have support will * return false w/o doing anything. MAC's that do * support it will return true w/o doing anything. */ sc->sc_mrretry = ath_hal_setupxtxdesc(ah, NULL, 0,0, 0,0, 0,0); /* * Check if the device has hardware counters for PHY * errors. If so we need to enable the MIB interrupt * so we can act on stat triggers. */ if (ath_hal_hwphycounters(ah)) sc->sc_needmib = 1; /* * Get the hardware key cache size. */ sc->sc_keymax = ath_hal_keycachesize(ah); if (sc->sc_keymax > ATH_KEYMAX) { if_printf(ifp, "Warning, using only %u of %u key cache slots\n", ATH_KEYMAX, sc->sc_keymax); sc->sc_keymax = ATH_KEYMAX; } /* * Reset the key cache since some parts do not * reset the contents on initial power up. */ for (i = 0; i < sc->sc_keymax; i++) ath_hal_keyreset(ah, i); /* * Collect the default channel list. */ error = ath_getchannels(sc); if (error != 0) goto bad; /* * Setup rate tables for all potential media types. */ ath_rate_setup(sc, IEEE80211_MODE_11A); ath_rate_setup(sc, IEEE80211_MODE_11B); ath_rate_setup(sc, IEEE80211_MODE_11G); ath_rate_setup(sc, IEEE80211_MODE_TURBO_A); ath_rate_setup(sc, IEEE80211_MODE_TURBO_G); ath_rate_setup(sc, IEEE80211_MODE_STURBO_A); ath_rate_setup(sc, IEEE80211_MODE_11NA); ath_rate_setup(sc, IEEE80211_MODE_11NG); ath_rate_setup(sc, IEEE80211_MODE_HALF); ath_rate_setup(sc, IEEE80211_MODE_QUARTER); /* NB: setup here so ath_rate_update is happy */ ath_setcurmode(sc, IEEE80211_MODE_11A); /* * Allocate tx+rx descriptors and populate the lists. */ error = ath_desc_alloc(sc); if (error != 0) { if_printf(ifp, "failed to allocate descriptors: %d\n", error); goto bad; } callout_init_mtx(&sc->sc_cal_ch, &sc->sc_mtx, 0); callout_init_mtx(&sc->sc_wd_ch, &sc->sc_mtx, 0); ATH_TXBUF_LOCK_INIT(sc); sc->sc_tq = taskqueue_create("ath_taskq", M_NOWAIT, taskqueue_thread_enqueue, &sc->sc_tq); taskqueue_start_threads(&sc->sc_tq, 1, PI_NET, "%s taskq", ifp->if_xname); TASK_INIT(&sc->sc_rxtask, 0, ath_rx_proc, sc); TASK_INIT(&sc->sc_bmisstask, 0, ath_bmiss_proc, sc); TASK_INIT(&sc->sc_bstucktask,0, ath_bstuck_proc, sc); /* * Allocate hardware transmit queues: one queue for * beacon frames and one data queue for each QoS * priority. Note that the hal handles resetting * these queues at the needed time. * * XXX PS-Poll */ sc->sc_bhalq = ath_beaconq_setup(ah); if (sc->sc_bhalq == (u_int) -1) { if_printf(ifp, "unable to setup a beacon xmit queue!\n"); error = EIO; goto bad2; } sc->sc_cabq = ath_txq_setup(sc, HAL_TX_QUEUE_CAB, 0); if (sc->sc_cabq == NULL) { if_printf(ifp, "unable to setup CAB xmit queue!\n"); error = EIO; goto bad2; } /* NB: insure BK queue is the lowest priority h/w queue */ if (!ath_tx_setup(sc, WME_AC_BK, HAL_WME_AC_BK)) { if_printf(ifp, "unable to setup xmit queue for %s traffic!\n", ieee80211_wme_acnames[WME_AC_BK]); error = EIO; goto bad2; } if (!ath_tx_setup(sc, WME_AC_BE, HAL_WME_AC_BE) || !ath_tx_setup(sc, WME_AC_VI, HAL_WME_AC_VI) || !ath_tx_setup(sc, WME_AC_VO, HAL_WME_AC_VO)) { /* * Not enough hardware tx queues to properly do WME; * just punt and assign them all to the same h/w queue. * We could do a better job of this if, for example, * we allocate queues when we switch from station to * AP mode. */ if (sc->sc_ac2q[WME_AC_VI] != NULL) ath_tx_cleanupq(sc, sc->sc_ac2q[WME_AC_VI]); if (sc->sc_ac2q[WME_AC_BE] != NULL) ath_tx_cleanupq(sc, sc->sc_ac2q[WME_AC_BE]); sc->sc_ac2q[WME_AC_BE] = sc->sc_ac2q[WME_AC_BK]; sc->sc_ac2q[WME_AC_VI] = sc->sc_ac2q[WME_AC_BK]; sc->sc_ac2q[WME_AC_VO] = sc->sc_ac2q[WME_AC_BK]; } /* * Special case certain configurations. Note the * CAB queue is handled by these specially so don't * include them when checking the txq setup mask. */ switch (sc->sc_txqsetup &~ (1<sc_cabq->axq_qnum)) { case 0x01: TASK_INIT(&sc->sc_txtask, 0, ath_tx_proc_q0, sc); break; case 0x0f: TASK_INIT(&sc->sc_txtask, 0, ath_tx_proc_q0123, sc); break; default: TASK_INIT(&sc->sc_txtask, 0, ath_tx_proc, sc); break; } /* * Setup rate control. Some rate control modules * call back to change the anntena state so expose * the necessary entry points. * XXX maybe belongs in struct ath_ratectrl? */ sc->sc_setdefantenna = ath_setdefantenna; sc->sc_rc = ath_rate_attach(sc); if (sc->sc_rc == NULL) { error = EIO; goto bad2; } /* Attach DFS module */ if (! ath_dfs_attach(sc)) { device_printf(sc->sc_dev, "%s: unable to attach DFS\n", __func__); error = EIO; goto bad2; } /* Start DFS processing tasklet */ TASK_INIT(&sc->sc_dfstask, 0, ath_dfs_tasklet, sc); sc->sc_blinking = 0; sc->sc_ledstate = 1; sc->sc_ledon = 0; /* low true */ sc->sc_ledidle = (2700*hz)/1000; /* 2.7sec */ callout_init(&sc->sc_ledtimer, CALLOUT_MPSAFE); /* * Auto-enable soft led processing for IBM cards and for * 5211 minipci cards. Users can also manually enable/disable * support with a sysctl. */ sc->sc_softled = (devid == AR5212_DEVID_IBM || devid == AR5211_DEVID); if (sc->sc_softled) { ath_hal_gpioCfgOutput(ah, sc->sc_ledpin, HAL_GPIO_MUX_MAC_NETWORK_LED); ath_hal_gpioset(ah, sc->sc_ledpin, !sc->sc_ledon); } ifp->if_softc = sc; ifp->if_flags = IFF_SIMPLEX | IFF_BROADCAST | IFF_MULTICAST; ifp->if_start = ath_start; ifp->if_ioctl = ath_ioctl; ifp->if_init = ath_init; IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen); ifp->if_snd.ifq_drv_maxlen = ifqmaxlen; IFQ_SET_READY(&ifp->if_snd); ic->ic_ifp = ifp; /* XXX not right but it's not used anywhere important */ ic->ic_phytype = IEEE80211_T_OFDM; ic->ic_opmode = IEEE80211_M_STA; ic->ic_caps = IEEE80211_C_STA /* station mode */ | IEEE80211_C_IBSS /* ibss, nee adhoc, mode */ | IEEE80211_C_HOSTAP /* hostap mode */ | IEEE80211_C_MONITOR /* monitor mode */ | IEEE80211_C_AHDEMO /* adhoc demo mode */ | IEEE80211_C_WDS /* 4-address traffic works */ | IEEE80211_C_MBSS /* mesh point link mode */ | IEEE80211_C_SHPREAMBLE /* short preamble supported */ | IEEE80211_C_SHSLOT /* short slot time supported */ | IEEE80211_C_WPA /* capable of WPA1+WPA2 */ | IEEE80211_C_BGSCAN /* capable of bg scanning */ | IEEE80211_C_TXFRAG /* handle tx frags */ ; /* * Query the hal to figure out h/w crypto support. */ if (ath_hal_ciphersupported(ah, HAL_CIPHER_WEP)) ic->ic_cryptocaps |= IEEE80211_CRYPTO_WEP; if (ath_hal_ciphersupported(ah, HAL_CIPHER_AES_OCB)) ic->ic_cryptocaps |= IEEE80211_CRYPTO_AES_OCB; if (ath_hal_ciphersupported(ah, HAL_CIPHER_AES_CCM)) ic->ic_cryptocaps |= IEEE80211_CRYPTO_AES_CCM; if (ath_hal_ciphersupported(ah, HAL_CIPHER_CKIP)) ic->ic_cryptocaps |= IEEE80211_CRYPTO_CKIP; if (ath_hal_ciphersupported(ah, HAL_CIPHER_TKIP)) { ic->ic_cryptocaps |= IEEE80211_CRYPTO_TKIP; /* * Check if h/w does the MIC and/or whether the * separate key cache entries are required to * handle both tx+rx MIC keys. */ if (ath_hal_ciphersupported(ah, HAL_CIPHER_MIC)) ic->ic_cryptocaps |= IEEE80211_CRYPTO_TKIPMIC; /* * If the h/w supports storing tx+rx MIC keys * in one cache slot automatically enable use. */ if (ath_hal_hastkipsplit(ah) || !ath_hal_settkipsplit(ah, AH_FALSE)) sc->sc_splitmic = 1; /* * If the h/w can do TKIP MIC together with WME then * we use it; otherwise we force the MIC to be done * in software by the net80211 layer. */ if (ath_hal_haswmetkipmic(ah)) sc->sc_wmetkipmic = 1; } sc->sc_hasclrkey = ath_hal_ciphersupported(ah, HAL_CIPHER_CLR); /* * Check for multicast key search support. */ if (ath_hal_hasmcastkeysearch(sc->sc_ah) && !ath_hal_getmcastkeysearch(sc->sc_ah)) { ath_hal_setmcastkeysearch(sc->sc_ah, 1); } sc->sc_mcastkey = ath_hal_getmcastkeysearch(ah); /* * Mark key cache slots associated with global keys * as in use. If we knew TKIP was not to be used we * could leave the +32, +64, and +32+64 slots free. */ for (i = 0; i < IEEE80211_WEP_NKID; i++) { setbit(sc->sc_keymap, i); setbit(sc->sc_keymap, i+64); if (sc->sc_splitmic) { setbit(sc->sc_keymap, i+32); setbit(sc->sc_keymap, i+32+64); } } /* * TPC support can be done either with a global cap or * per-packet support. The latter is not available on * all parts. We're a bit pedantic here as all parts * support a global cap. */ if (ath_hal_hastpc(ah) || ath_hal_hastxpowlimit(ah)) ic->ic_caps |= IEEE80211_C_TXPMGT; /* * Mark WME capability only if we have sufficient * hardware queues to do proper priority scheduling. */ if (sc->sc_ac2q[WME_AC_BE] != sc->sc_ac2q[WME_AC_BK]) ic->ic_caps |= IEEE80211_C_WME; /* * Check for misc other capabilities. */ if (ath_hal_hasbursting(ah)) ic->ic_caps |= IEEE80211_C_BURST; sc->sc_hasbmask = ath_hal_hasbssidmask(ah); sc->sc_hasbmatch = ath_hal_hasbssidmatch(ah); sc->sc_hastsfadd = ath_hal_hastsfadjust(ah); sc->sc_rxslink = ath_hal_self_linked_final_rxdesc(ah); if (ath_hal_hasfastframes(ah)) ic->ic_caps |= IEEE80211_C_FF; wmodes = ath_hal_getwirelessmodes(ah); if (wmodes & (HAL_MODE_108G|HAL_MODE_TURBO)) ic->ic_caps |= IEEE80211_C_TURBOP; #ifdef IEEE80211_SUPPORT_TDMA if (ath_hal_macversion(ah) > 0x78) { ic->ic_caps |= IEEE80211_C_TDMA; /* capable of TDMA */ ic->ic_tdma_update = ath_tdma_update; } #endif /* * The if_ath 11n support is completely not ready for normal use. * Enabling this option will likely break everything and everything. * Don't think of doing that unless you know what you're doing. */ #ifdef ATH_ENABLE_11N /* * Query HT capabilities */ if (ath_hal_getcapability(ah, HAL_CAP_HT, 0, NULL) == HAL_OK && (wmodes & (HAL_MODE_HT20 | HAL_MODE_HT40))) { int rxs, txs; device_printf(sc->sc_dev, "[HT] enabling HT modes\n"); ic->ic_htcaps = IEEE80211_HTC_HT /* HT operation */ | IEEE80211_HTC_AMPDU /* A-MPDU tx/rx */ | IEEE80211_HTC_AMSDU /* A-MSDU tx/rx */ | IEEE80211_HTCAP_MAXAMSDU_3839 /* max A-MSDU length */ | IEEE80211_HTCAP_SMPS_OFF; /* SM power save off */ ; /* * Enable short-GI for HT20 only if the hardware * advertises support. * Notably, anything earlier than the AR9287 doesn't. */ if ((ath_hal_getcapability(ah, HAL_CAP_HT20_SGI, 0, NULL) == HAL_OK) && (wmodes & HAL_MODE_HT20)) { device_printf(sc->sc_dev, "[HT] enabling short-GI in 20MHz mode\n"); ic->ic_htcaps |= IEEE80211_HTCAP_SHORTGI20; } if (wmodes & HAL_MODE_HT40) ic->ic_htcaps |= IEEE80211_HTCAP_CHWIDTH40 | IEEE80211_HTCAP_SHORTGI40; /* * rx/tx stream is not currently used anywhere; it needs to be taken * into account when negotiating which MCS rates it'll receive and * what MCS rates are available for TX. */ (void) ath_hal_getcapability(ah, HAL_CAP_STREAMS, 0, &rxs); (void) ath_hal_getcapability(ah, HAL_CAP_STREAMS, 1, &txs); ath_hal_getrxchainmask(ah, &sc->sc_rxchainmask); ath_hal_gettxchainmask(ah, &sc->sc_txchainmask); ic->ic_txstream = txs; ic->ic_rxstream = rxs; device_printf(sc->sc_dev, "[HT] %d RX streams; %d TX streams\n", rxs, txs); } #endif /* * Indicate we need the 802.11 header padded to a * 32-bit boundary for 4-address and QoS frames. */ ic->ic_flags |= IEEE80211_F_DATAPAD; /* * Query the hal about antenna support. */ sc->sc_defant = ath_hal_getdefantenna(ah); /* * Not all chips have the VEOL support we want to * use with IBSS beacons; check here for it. */ sc->sc_hasveol = ath_hal_hasveol(ah); /* get mac address from hardware */ ath_hal_getmac(ah, macaddr); if (sc->sc_hasbmask) ath_hal_getbssidmask(ah, sc->sc_hwbssidmask); /* NB: used to size node table key mapping array */ ic->ic_max_keyix = sc->sc_keymax; /* call MI attach routine. */ ieee80211_ifattach(ic, macaddr); ic->ic_setregdomain = ath_setregdomain; ic->ic_getradiocaps = ath_getradiocaps; sc->sc_opmode = HAL_M_STA; /* override default methods */ ic->ic_newassoc = ath_newassoc; ic->ic_updateslot = ath_updateslot; ic->ic_wme.wme_update = ath_wme_update; ic->ic_vap_create = ath_vap_create; ic->ic_vap_delete = ath_vap_delete; ic->ic_raw_xmit = ath_raw_xmit; ic->ic_update_mcast = ath_update_mcast; ic->ic_update_promisc = ath_update_promisc; ic->ic_node_alloc = ath_node_alloc; sc->sc_node_free = ic->ic_node_free; ic->ic_node_free = ath_node_free; ic->ic_node_getsignal = ath_node_getsignal; ic->ic_scan_start = ath_scan_start; ic->ic_scan_end = ath_scan_end; ic->ic_set_channel = ath_set_channel; ieee80211_radiotap_attach(ic, &sc->sc_tx_th.wt_ihdr, sizeof(sc->sc_tx_th), ATH_TX_RADIOTAP_PRESENT, &sc->sc_rx_th.wr_ihdr, sizeof(sc->sc_rx_th), ATH_RX_RADIOTAP_PRESENT); /* * Setup dynamic sysctl's now that country code and * regdomain are available from the hal. */ ath_sysctlattach(sc); ath_sysctl_stats_attach(sc); if (bootverbose) ieee80211_announce(ic); ath_announce(sc); return 0; bad2: ath_tx_cleanup(sc); ath_desc_free(sc); bad: if (ah) ath_hal_detach(ah); if (ifp != NULL) if_free(ifp); sc->sc_invalid = 1; return error; } int ath_detach(struct ath_softc *sc) { struct ifnet *ifp = sc->sc_ifp; DPRINTF(sc, ATH_DEBUG_ANY, "%s: if_flags %x\n", __func__, ifp->if_flags); /* * NB: the order of these is important: * o stop the chip so no more interrupts will fire * o call the 802.11 layer before detaching the hal to * insure callbacks into the driver to delete global * key cache entries can be handled * o free the taskqueue which drains any pending tasks * o reclaim the tx queue data structures after calling * the 802.11 layer as we'll get called back to reclaim * node state and potentially want to use them * o to cleanup the tx queues the hal is called, so detach * it last * Other than that, it's straightforward... */ ath_stop(ifp); ieee80211_ifdetach(ifp->if_l2com); taskqueue_free(sc->sc_tq); #ifdef ATH_TX99_DIAG if (sc->sc_tx99 != NULL) sc->sc_tx99->detach(sc->sc_tx99); #endif ath_rate_detach(sc->sc_rc); ath_dfs_detach(sc); ath_desc_free(sc); ath_tx_cleanup(sc); ath_hal_detach(sc->sc_ah); /* NB: sets chip in full sleep */ if_free(ifp); return 0; } /* * MAC address handling for multiple BSS on the same radio. * The first vap uses the MAC address from the EEPROM. For * subsequent vap's we set the U/L bit (bit 1) in the MAC * address and use the next six bits as an index. */ static void assign_address(struct ath_softc *sc, uint8_t mac[IEEE80211_ADDR_LEN], int clone) { int i; if (clone && sc->sc_hasbmask) { /* NB: we only do this if h/w supports multiple bssid */ for (i = 0; i < 8; i++) if ((sc->sc_bssidmask & (1<sc_bssidmask |= 1<sc_hwbssidmask[0] &= ~mac[0]; if (i == 0) sc->sc_nbssid0++; } static void reclaim_address(struct ath_softc *sc, const uint8_t mac[IEEE80211_ADDR_LEN]) { int i = mac[0] >> 2; uint8_t mask; if (i != 0 || --sc->sc_nbssid0 == 0) { sc->sc_bssidmask &= ~(1<sc_bssidmask & (1<sc_hwbssidmask[0] |= mask; } } /* * Assign a beacon xmit slot. We try to space out * assignments so when beacons are staggered the * traffic coming out of the cab q has maximal time * to go out before the next beacon is scheduled. */ static int assign_bslot(struct ath_softc *sc) { u_int slot, free; free = 0; for (slot = 0; slot < ATH_BCBUF; slot++) if (sc->sc_bslot[slot] == NULL) { if (sc->sc_bslot[(slot+1)%ATH_BCBUF] == NULL && sc->sc_bslot[(slot-1)%ATH_BCBUF] == NULL) return slot; free = slot; /* NB: keep looking for a double slot */ } return free; } static struct ieee80211vap * ath_vap_create(struct ieee80211com *ic, const char name[IFNAMSIZ], int unit, int opmode, int flags, const uint8_t bssid[IEEE80211_ADDR_LEN], const uint8_t mac0[IEEE80211_ADDR_LEN]) { struct ath_softc *sc = ic->ic_ifp->if_softc; struct ath_vap *avp; struct ieee80211vap *vap; uint8_t mac[IEEE80211_ADDR_LEN]; int ic_opmode, needbeacon, error; avp = (struct ath_vap *) malloc(sizeof(struct ath_vap), M_80211_VAP, M_WAITOK | M_ZERO); needbeacon = 0; IEEE80211_ADDR_COPY(mac, mac0); ATH_LOCK(sc); ic_opmode = opmode; /* default to opmode of new vap */ switch (opmode) { case IEEE80211_M_STA: if (sc->sc_nstavaps != 0) { /* XXX only 1 for now */ device_printf(sc->sc_dev, "only 1 sta vap supported\n"); goto bad; } if (sc->sc_nvaps) { /* * With multiple vaps we must fall back * to s/w beacon miss handling. */ flags |= IEEE80211_CLONE_NOBEACONS; } if (flags & IEEE80211_CLONE_NOBEACONS) { /* * Station mode w/o beacons are implemented w/ AP mode. */ ic_opmode = IEEE80211_M_HOSTAP; } break; case IEEE80211_M_IBSS: if (sc->sc_nvaps != 0) { /* XXX only 1 for now */ device_printf(sc->sc_dev, "only 1 ibss vap supported\n"); goto bad; } needbeacon = 1; break; case IEEE80211_M_AHDEMO: #ifdef IEEE80211_SUPPORT_TDMA if (flags & IEEE80211_CLONE_TDMA) { if (sc->sc_nvaps != 0) { device_printf(sc->sc_dev, "only 1 tdma vap supported\n"); goto bad; } needbeacon = 1; flags |= IEEE80211_CLONE_NOBEACONS; } /* fall thru... */ #endif case IEEE80211_M_MONITOR: if (sc->sc_nvaps != 0 && ic->ic_opmode != opmode) { /* * Adopt existing mode. Adding a monitor or ahdemo * vap to an existing configuration is of dubious * value but should be ok. */ /* XXX not right for monitor mode */ ic_opmode = ic->ic_opmode; } break; case IEEE80211_M_HOSTAP: case IEEE80211_M_MBSS: needbeacon = 1; break; case IEEE80211_M_WDS: if (sc->sc_nvaps != 0 && ic->ic_opmode == IEEE80211_M_STA) { device_printf(sc->sc_dev, "wds not supported in sta mode\n"); goto bad; } /* * Silently remove any request for a unique * bssid; WDS vap's always share the local * mac address. */ flags &= ~IEEE80211_CLONE_BSSID; if (sc->sc_nvaps == 0) ic_opmode = IEEE80211_M_HOSTAP; else ic_opmode = ic->ic_opmode; break; default: device_printf(sc->sc_dev, "unknown opmode %d\n", opmode); goto bad; } /* * Check that a beacon buffer is available; the code below assumes it. */ if (needbeacon & STAILQ_EMPTY(&sc->sc_bbuf)) { device_printf(sc->sc_dev, "no beacon buffer available\n"); goto bad; } /* STA, AHDEMO? */ if (opmode == IEEE80211_M_HOSTAP || opmode == IEEE80211_M_MBSS) { assign_address(sc, mac, flags & IEEE80211_CLONE_BSSID); ath_hal_setbssidmask(sc->sc_ah, sc->sc_hwbssidmask); } vap = &avp->av_vap; /* XXX can't hold mutex across if_alloc */ ATH_UNLOCK(sc); error = ieee80211_vap_setup(ic, vap, name, unit, opmode, flags, bssid, mac); ATH_LOCK(sc); if (error != 0) { device_printf(sc->sc_dev, "%s: error %d creating vap\n", __func__, error); goto bad2; } /* h/w crypto support */ vap->iv_key_alloc = ath_key_alloc; vap->iv_key_delete = ath_key_delete; vap->iv_key_set = ath_key_set; vap->iv_key_update_begin = ath_key_update_begin; vap->iv_key_update_end = ath_key_update_end; /* override various methods */ avp->av_recv_mgmt = vap->iv_recv_mgmt; vap->iv_recv_mgmt = ath_recv_mgmt; vap->iv_reset = ath_reset_vap; vap->iv_update_beacon = ath_beacon_update; avp->av_newstate = vap->iv_newstate; vap->iv_newstate = ath_newstate; avp->av_bmiss = vap->iv_bmiss; vap->iv_bmiss = ath_bmiss_vap; /* Set default parameters */ /* * Anything earlier than some AR9300 series MACs don't * support a smaller MPDU density. */ vap->iv_ampdu_density = IEEE80211_HTCAP_MPDUDENSITY_8; /* * All NICs can handle the maximum size, however * AR5416 based MACs can only TX aggregates w/ RTS * protection when the total aggregate size is <= 8k. * However, for now that's enforced by the TX path. */ vap->iv_ampdu_rxmax = IEEE80211_HTCAP_MAXRXAMPDU_64K; avp->av_bslot = -1; if (needbeacon) { /* * Allocate beacon state and setup the q for buffered * multicast frames. We know a beacon buffer is * available because we checked above. */ avp->av_bcbuf = STAILQ_FIRST(&sc->sc_bbuf); STAILQ_REMOVE_HEAD(&sc->sc_bbuf, bf_list); if (opmode != IEEE80211_M_IBSS || !sc->sc_hasveol) { /* * Assign the vap to a beacon xmit slot. As above * this cannot fail to find a free one. */ avp->av_bslot = assign_bslot(sc); KASSERT(sc->sc_bslot[avp->av_bslot] == NULL, ("beacon slot %u not empty", avp->av_bslot)); sc->sc_bslot[avp->av_bslot] = vap; sc->sc_nbcnvaps++; } if (sc->sc_hastsfadd && sc->sc_nbcnvaps > 0) { /* * Multple vaps are to transmit beacons and we * have h/w support for TSF adjusting; enable * use of staggered beacons. */ sc->sc_stagbeacons = 1; } ath_txq_init(sc, &avp->av_mcastq, ATH_TXQ_SWQ); } ic->ic_opmode = ic_opmode; if (opmode != IEEE80211_M_WDS) { sc->sc_nvaps++; if (opmode == IEEE80211_M_STA) sc->sc_nstavaps++; if (opmode == IEEE80211_M_MBSS) sc->sc_nmeshvaps++; } switch (ic_opmode) { case IEEE80211_M_IBSS: sc->sc_opmode = HAL_M_IBSS; break; case IEEE80211_M_STA: sc->sc_opmode = HAL_M_STA; break; case IEEE80211_M_AHDEMO: #ifdef IEEE80211_SUPPORT_TDMA if (vap->iv_caps & IEEE80211_C_TDMA) { sc->sc_tdma = 1; /* NB: disable tsf adjust */ sc->sc_stagbeacons = 0; } /* * NB: adhoc demo mode is a pseudo mode; to the hal it's * just ap mode. */ /* fall thru... */ #endif case IEEE80211_M_HOSTAP: case IEEE80211_M_MBSS: sc->sc_opmode = HAL_M_HOSTAP; break; case IEEE80211_M_MONITOR: sc->sc_opmode = HAL_M_MONITOR; break; default: /* XXX should not happen */ break; } if (sc->sc_hastsfadd) { /* * Configure whether or not TSF adjust should be done. */ ath_hal_settsfadjust(sc->sc_ah, sc->sc_stagbeacons); } if (flags & IEEE80211_CLONE_NOBEACONS) { /* * Enable s/w beacon miss handling. */ sc->sc_swbmiss = 1; } ATH_UNLOCK(sc); /* complete setup */ ieee80211_vap_attach(vap, ath_media_change, ieee80211_media_status); return vap; bad2: reclaim_address(sc, mac); ath_hal_setbssidmask(sc->sc_ah, sc->sc_hwbssidmask); bad: free(avp, M_80211_VAP); ATH_UNLOCK(sc); return NULL; } static void ath_vap_delete(struct ieee80211vap *vap) { struct ieee80211com *ic = vap->iv_ic; struct ifnet *ifp = ic->ic_ifp; struct ath_softc *sc = ifp->if_softc; struct ath_hal *ah = sc->sc_ah; struct ath_vap *avp = ATH_VAP(vap); if (ifp->if_drv_flags & IFF_DRV_RUNNING) { /* * Quiesce the hardware while we remove the vap. In * particular we need to reclaim all references to * the vap state by any frames pending on the tx queues. */ ath_hal_intrset(ah, 0); /* disable interrupts */ ath_draintxq(sc); /* stop xmit side */ ath_stoprecv(sc); /* stop recv side */ } ieee80211_vap_detach(vap); ATH_LOCK(sc); /* * Reclaim beacon state. Note this must be done before * the vap instance is reclaimed as we may have a reference * to it in the buffer for the beacon frame. */ if (avp->av_bcbuf != NULL) { if (avp->av_bslot != -1) { sc->sc_bslot[avp->av_bslot] = NULL; sc->sc_nbcnvaps--; } ath_beacon_return(sc, avp->av_bcbuf); avp->av_bcbuf = NULL; if (sc->sc_nbcnvaps == 0) { sc->sc_stagbeacons = 0; if (sc->sc_hastsfadd) ath_hal_settsfadjust(sc->sc_ah, 0); } /* * Reclaim any pending mcast frames for the vap. */ ath_tx_draintxq(sc, &avp->av_mcastq); ATH_TXQ_LOCK_DESTROY(&avp->av_mcastq); } /* * Update bookkeeping. */ if (vap->iv_opmode == IEEE80211_M_STA) { sc->sc_nstavaps--; if (sc->sc_nstavaps == 0 && sc->sc_swbmiss) sc->sc_swbmiss = 0; } else if (vap->iv_opmode == IEEE80211_M_HOSTAP || vap->iv_opmode == IEEE80211_M_MBSS) { reclaim_address(sc, vap->iv_myaddr); ath_hal_setbssidmask(ah, sc->sc_hwbssidmask); if (vap->iv_opmode == IEEE80211_M_MBSS) sc->sc_nmeshvaps--; } if (vap->iv_opmode != IEEE80211_M_WDS) sc->sc_nvaps--; #ifdef IEEE80211_SUPPORT_TDMA /* TDMA operation ceases when the last vap is destroyed */ if (sc->sc_tdma && sc->sc_nvaps == 0) { sc->sc_tdma = 0; sc->sc_swbmiss = 0; } #endif ATH_UNLOCK(sc); free(avp, M_80211_VAP); if (ifp->if_drv_flags & IFF_DRV_RUNNING) { /* * Restart rx+tx machines if still running (RUNNING will * be reset if we just destroyed the last vap). */ if (ath_startrecv(sc) != 0) if_printf(ifp, "%s: unable to restart recv logic\n", __func__); if (sc->sc_beacons) { /* restart beacons */ #ifdef IEEE80211_SUPPORT_TDMA if (sc->sc_tdma) ath_tdma_config(sc, NULL); else #endif ath_beacon_config(sc, NULL); } ath_hal_intrset(ah, sc->sc_imask); } } void ath_suspend(struct ath_softc *sc) { struct ifnet *ifp = sc->sc_ifp; struct ieee80211com *ic = ifp->if_l2com; DPRINTF(sc, ATH_DEBUG_ANY, "%s: if_flags %x\n", __func__, ifp->if_flags); sc->sc_resume_up = (ifp->if_flags & IFF_UP) != 0; if (ic->ic_opmode == IEEE80211_M_STA) ath_stop(ifp); else ieee80211_suspend_all(ic); /* * NB: don't worry about putting the chip in low power * mode; pci will power off our socket on suspend and * CardBus detaches the device. */ } /* * Reset the key cache since some parts do not reset the * contents on resume. First we clear all entries, then * re-load keys that the 802.11 layer assumes are setup * in h/w. */ static void ath_reset_keycache(struct ath_softc *sc) { struct ifnet *ifp = sc->sc_ifp; struct ieee80211com *ic = ifp->if_l2com; struct ath_hal *ah = sc->sc_ah; int i; for (i = 0; i < sc->sc_keymax; i++) ath_hal_keyreset(ah, i); ieee80211_crypto_reload_keys(ic); } void ath_resume(struct ath_softc *sc) { struct ifnet *ifp = sc->sc_ifp; struct ieee80211com *ic = ifp->if_l2com; struct ath_hal *ah = sc->sc_ah; HAL_STATUS status; DPRINTF(sc, ATH_DEBUG_ANY, "%s: if_flags %x\n", __func__, ifp->if_flags); /* * Must reset the chip before we reload the * keycache as we were powered down on suspend. */ ath_hal_reset(ah, sc->sc_opmode, sc->sc_curchan != NULL ? sc->sc_curchan : ic->ic_curchan, AH_FALSE, &status); ath_reset_keycache(sc); /* Let DFS at it in case it's a DFS channel */ ath_dfs_radar_enable(sc, ic->ic_curchan); if (sc->sc_resume_up) { if (ic->ic_opmode == IEEE80211_M_STA) { ath_init(sc); /* * Program the beacon registers using the last rx'd * beacon frame and enable sync on the next beacon * we see. This should handle the case where we * wakeup and find the same AP and also the case where * we wakeup and need to roam. For the latter we * should get bmiss events that trigger a roam. */ ath_beacon_config(sc, NULL); sc->sc_syncbeacon = 1; } else ieee80211_resume_all(ic); } if (sc->sc_softled) { ath_hal_gpioCfgOutput(ah, sc->sc_ledpin, HAL_GPIO_MUX_MAC_NETWORK_LED); ath_hal_gpioset(ah, sc->sc_ledpin, !sc->sc_ledon); } } void ath_shutdown(struct ath_softc *sc) { struct ifnet *ifp = sc->sc_ifp; DPRINTF(sc, ATH_DEBUG_ANY, "%s: if_flags %x\n", __func__, ifp->if_flags); ath_stop(ifp); /* NB: no point powering down chip as we're about to reboot */ } /* * Interrupt handler. Most of the actual processing is deferred. */ void ath_intr(void *arg) { struct ath_softc *sc = arg; struct ifnet *ifp = sc->sc_ifp; struct ath_hal *ah = sc->sc_ah; HAL_INT status = 0; if (sc->sc_invalid) { /* * The hardware is not ready/present, don't touch anything. * Note this can happen early on if the IRQ is shared. */ DPRINTF(sc, ATH_DEBUG_ANY, "%s: invalid; ignored\n", __func__); return; } if (!ath_hal_intrpend(ah)) /* shared irq, not for us */ return; if ((ifp->if_flags & IFF_UP) == 0 || (ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { HAL_INT status; DPRINTF(sc, ATH_DEBUG_ANY, "%s: if_flags 0x%x\n", __func__, ifp->if_flags); ath_hal_getisr(ah, &status); /* clear ISR */ ath_hal_intrset(ah, 0); /* disable further intr's */ return; } /* * Figure out the reason(s) for the interrupt. Note * that the hal returns a pseudo-ISR that may include * bits we haven't explicitly enabled so we mask the * value to insure we only process bits we requested. */ ath_hal_getisr(ah, &status); /* NB: clears ISR too */ DPRINTF(sc, ATH_DEBUG_INTR, "%s: status 0x%x\n", __func__, status); status &= sc->sc_imask; /* discard unasked for bits */ /* Short-circuit un-handled interrupts */ if (status == 0x0) return; if (status & HAL_INT_FATAL) { sc->sc_stats.ast_hardware++; ath_hal_intrset(ah, 0); /* disable intr's until reset */ ath_fatal_proc(sc, 0); } else { if (status & HAL_INT_SWBA) { /* * Software beacon alert--time to send a beacon. * Handle beacon transmission directly; deferring * this is too slow to meet timing constraints * under load. */ #ifdef IEEE80211_SUPPORT_TDMA if (sc->sc_tdma) { if (sc->sc_tdmaswba == 0) { struct ieee80211com *ic = ifp->if_l2com; struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps); ath_tdma_beacon_send(sc, vap); sc->sc_tdmaswba = vap->iv_tdma->tdma_bintval; } else sc->sc_tdmaswba--; } else #endif { ath_beacon_proc(sc, 0); #ifdef IEEE80211_SUPPORT_SUPERG /* * Schedule the rx taskq in case there's no * traffic so any frames held on the staging * queue are aged and potentially flushed. */ taskqueue_enqueue(sc->sc_tq, &sc->sc_rxtask); #endif } } if (status & HAL_INT_RXEOL) { /* * NB: the hardware should re-read the link when * RXE bit is written, but it doesn't work at * least on older hardware revs. */ sc->sc_stats.ast_rxeol++; sc->sc_rxlink = NULL; } if (status & HAL_INT_TXURN) { sc->sc_stats.ast_txurn++; /* bump tx trigger level */ ath_hal_updatetxtriglevel(ah, AH_TRUE); } if (status & HAL_INT_RX) taskqueue_enqueue(sc->sc_tq, &sc->sc_rxtask); if (status & HAL_INT_TX) taskqueue_enqueue(sc->sc_tq, &sc->sc_txtask); if (status & HAL_INT_BMISS) { sc->sc_stats.ast_bmiss++; taskqueue_enqueue(sc->sc_tq, &sc->sc_bmisstask); } if (status & HAL_INT_GTT) sc->sc_stats.ast_tx_timeout++; if (status & HAL_INT_CST) sc->sc_stats.ast_tx_cst++; if (status & HAL_INT_MIB) { sc->sc_stats.ast_mib++; /* * Disable interrupts until we service the MIB * interrupt; otherwise it will continue to fire. */ ath_hal_intrset(ah, 0); /* * Let the hal handle the event. We assume it will * clear whatever condition caused the interrupt. */ ath_hal_mibevent(ah, &sc->sc_halstats); ath_hal_intrset(ah, sc->sc_imask); } if (status & HAL_INT_RXORN) { /* NB: hal marks HAL_INT_FATAL when RXORN is fatal */ sc->sc_stats.ast_rxorn++; } } } static void ath_fatal_proc(void *arg, int pending) { struct ath_softc *sc = arg; struct ifnet *ifp = sc->sc_ifp; u_int32_t *state; u_int32_t len; void *sp; if_printf(ifp, "hardware error; resetting\n"); /* * Fatal errors are unrecoverable. Typically these * are caused by DMA errors. Collect h/w state from * the hal so we can diagnose what's going on. */ if (ath_hal_getfatalstate(sc->sc_ah, &sp, &len)) { KASSERT(len >= 6*sizeof(u_int32_t), ("len %u bytes", len)); state = sp; if_printf(ifp, "0x%08x 0x%08x 0x%08x, 0x%08x 0x%08x 0x%08x\n", state[0], state[1] , state[2], state[3], state[4], state[5]); } ath_reset(ifp); } static void ath_bmiss_vap(struct ieee80211vap *vap) { /* * Workaround phantom bmiss interrupts by sanity-checking * the time of our last rx'd frame. If it is within the * beacon miss interval then ignore the interrupt. If it's * truly a bmiss we'll get another interrupt soon and that'll * be dispatched up for processing. Note this applies only * for h/w beacon miss events. */ if ((vap->iv_flags_ext & IEEE80211_FEXT_SWBMISS) == 0) { struct ifnet *ifp = vap->iv_ic->ic_ifp; struct ath_softc *sc = ifp->if_softc; u_int64_t lastrx = sc->sc_lastrx; u_int64_t tsf = ath_hal_gettsf64(sc->sc_ah); u_int bmisstimeout = vap->iv_bmissthreshold * vap->iv_bss->ni_intval * 1024; DPRINTF(sc, ATH_DEBUG_BEACON, "%s: tsf %llu lastrx %lld (%llu) bmiss %u\n", __func__, (unsigned long long) tsf, (unsigned long long)(tsf - lastrx), (unsigned long long) lastrx, bmisstimeout); if (tsf - lastrx <= bmisstimeout) { sc->sc_stats.ast_bmiss_phantom++; return; } } ATH_VAP(vap)->av_bmiss(vap); } static int ath_hal_gethangstate(struct ath_hal *ah, uint32_t mask, uint32_t *hangs) { uint32_t rsize; void *sp; if (!ath_hal_getdiagstate(ah, HAL_DIAG_CHECK_HANGS, &mask, sizeof(mask), &sp, &rsize)) return 0; KASSERT(rsize == sizeof(uint32_t), ("resultsize %u", rsize)); *hangs = *(uint32_t *)sp; return 1; } static void ath_bmiss_proc(void *arg, int pending) { struct ath_softc *sc = arg; struct ifnet *ifp = sc->sc_ifp; uint32_t hangs; DPRINTF(sc, ATH_DEBUG_ANY, "%s: pending %u\n", __func__, pending); if (ath_hal_gethangstate(sc->sc_ah, 0xff, &hangs) && hangs != 0) { if_printf(ifp, "bb hang detected (0x%x), resetting\n", hangs); ath_reset(ifp); } else ieee80211_beacon_miss(ifp->if_l2com); } /* * Handle TKIP MIC setup to deal hardware that doesn't do MIC * calcs together with WME. If necessary disable the crypto * hardware and mark the 802.11 state so keys will be setup * with the MIC work done in software. */ static void ath_settkipmic(struct ath_softc *sc) { struct ifnet *ifp = sc->sc_ifp; struct ieee80211com *ic = ifp->if_l2com; if ((ic->ic_cryptocaps & IEEE80211_CRYPTO_TKIP) && !sc->sc_wmetkipmic) { if (ic->ic_flags & IEEE80211_F_WME) { ath_hal_settkipmic(sc->sc_ah, AH_FALSE); ic->ic_cryptocaps &= ~IEEE80211_CRYPTO_TKIPMIC; } else { ath_hal_settkipmic(sc->sc_ah, AH_TRUE); ic->ic_cryptocaps |= IEEE80211_CRYPTO_TKIPMIC; } } } static void ath_init(void *arg) { struct ath_softc *sc = (struct ath_softc *) arg; struct ifnet *ifp = sc->sc_ifp; struct ieee80211com *ic = ifp->if_l2com; struct ath_hal *ah = sc->sc_ah; HAL_STATUS status; DPRINTF(sc, ATH_DEBUG_ANY, "%s: if_flags 0x%x\n", __func__, ifp->if_flags); ATH_LOCK(sc); /* * Stop anything previously setup. This is safe * whether this is the first time through or not. */ ath_stop_locked(ifp); /* * The basic interface to setting the hardware in a good * state is ``reset''. On return the hardware is known to * be powered up and with interrupts disabled. This must * be followed by initialization of the appropriate bits * and then setup of the interrupt mask. */ ath_settkipmic(sc); if (!ath_hal_reset(ah, sc->sc_opmode, ic->ic_curchan, AH_FALSE, &status)) { if_printf(ifp, "unable to reset hardware; hal status %u\n", status); ATH_UNLOCK(sc); return; } ath_chan_change(sc, ic->ic_curchan); /* Let DFS at it in case it's a DFS channel */ ath_dfs_radar_enable(sc, ic->ic_curchan); /* * Likewise this is set during reset so update * state cached in the driver. */ sc->sc_diversity = ath_hal_getdiversity(ah); sc->sc_lastlongcal = 0; sc->sc_resetcal = 1; sc->sc_lastcalreset = 0; sc->sc_lastani = 0; sc->sc_lastshortcal = 0; sc->sc_doresetcal = AH_FALSE; /* * Setup the hardware after reset: the key cache * is filled as needed and the receive engine is * set going. Frame transmit is handled entirely * in the frame output path; there's nothing to do * here except setup the interrupt mask. */ if (ath_startrecv(sc) != 0) { if_printf(ifp, "unable to start recv logic\n"); ATH_UNLOCK(sc); return; } /* * Enable interrupts. */ sc->sc_imask = HAL_INT_RX | HAL_INT_TX | HAL_INT_RXEOL | HAL_INT_RXORN | HAL_INT_FATAL | HAL_INT_GLOBAL; /* * Enable MIB interrupts when there are hardware phy counters. * Note we only do this (at the moment) for station mode. */ if (sc->sc_needmib && ic->ic_opmode == IEEE80211_M_STA) sc->sc_imask |= HAL_INT_MIB; /* Enable global TX timeout and carrier sense timeout if available */ if (ath_hal_gtxto_supported(ah)) sc->sc_imask |= HAL_INT_GTT; DPRINTF(sc, ATH_DEBUG_RESET, "%s: imask=0x%x\n", __func__, sc->sc_imask); ifp->if_drv_flags |= IFF_DRV_RUNNING; callout_reset(&sc->sc_wd_ch, hz, ath_watchdog, sc); ath_hal_intrset(ah, sc->sc_imask); ATH_UNLOCK(sc); #ifdef ATH_TX99_DIAG if (sc->sc_tx99 != NULL) sc->sc_tx99->start(sc->sc_tx99); else #endif ieee80211_start_all(ic); /* start all vap's */ } static void ath_stop_locked(struct ifnet *ifp) { struct ath_softc *sc = ifp->if_softc; struct ath_hal *ah = sc->sc_ah; DPRINTF(sc, ATH_DEBUG_ANY, "%s: invalid %u if_flags 0x%x\n", __func__, sc->sc_invalid, ifp->if_flags); ATH_LOCK_ASSERT(sc); if (ifp->if_drv_flags & IFF_DRV_RUNNING) { /* * Shutdown the hardware and driver: * reset 802.11 state machine * turn off timers * disable interrupts * turn off the radio * clear transmit machinery * clear receive machinery * drain and release tx queues * reclaim beacon resources * power down hardware * * Note that some of this work is not possible if the * hardware is gone (invalid). */ #ifdef ATH_TX99_DIAG if (sc->sc_tx99 != NULL) sc->sc_tx99->stop(sc->sc_tx99); #endif callout_stop(&sc->sc_wd_ch); sc->sc_wd_timer = 0; ifp->if_drv_flags &= ~IFF_DRV_RUNNING; if (!sc->sc_invalid) { if (sc->sc_softled) { callout_stop(&sc->sc_ledtimer); ath_hal_gpioset(ah, sc->sc_ledpin, !sc->sc_ledon); sc->sc_blinking = 0; } ath_hal_intrset(ah, 0); } ath_draintxq(sc); if (!sc->sc_invalid) { ath_stoprecv(sc); ath_hal_phydisable(ah); } else sc->sc_rxlink = NULL; ath_beacon_free(sc); /* XXX not needed */ } } static void ath_stop(struct ifnet *ifp) { struct ath_softc *sc = ifp->if_softc; ATH_LOCK(sc); ath_stop_locked(ifp); ATH_UNLOCK(sc); } /* * Reset the hardware w/o losing operational state. This is * basically a more efficient way of doing ath_stop, ath_init, * followed by state transitions to the current 802.11 * operational state. Used to recover from various errors and * to reset or reload hardware state. */ int ath_reset(struct ifnet *ifp) { struct ath_softc *sc = ifp->if_softc; struct ieee80211com *ic = ifp->if_l2com; struct ath_hal *ah = sc->sc_ah; HAL_STATUS status; ath_hal_intrset(ah, 0); /* disable interrupts */ ath_draintxq(sc); /* stop xmit side */ ath_stoprecv(sc); /* stop recv side */ ath_settkipmic(sc); /* configure TKIP MIC handling */ /* NB: indicate channel change so we do a full reset */ if (!ath_hal_reset(ah, sc->sc_opmode, ic->ic_curchan, AH_TRUE, &status)) if_printf(ifp, "%s: unable to reset hardware; hal status %u\n", __func__, status); sc->sc_diversity = ath_hal_getdiversity(ah); /* Let DFS at it in case it's a DFS channel */ ath_dfs_radar_enable(sc, ic->ic_curchan); if (ath_startrecv(sc) != 0) /* restart recv */ if_printf(ifp, "%s: unable to start recv logic\n", __func__); /* * We may be doing a reset in response to an ioctl * that changes the channel so update any state that * might change as a result. */ ath_chan_change(sc, ic->ic_curchan); if (sc->sc_beacons) { /* restart beacons */ #ifdef IEEE80211_SUPPORT_TDMA if (sc->sc_tdma) ath_tdma_config(sc, NULL); else #endif ath_beacon_config(sc, NULL); } ath_hal_intrset(ah, sc->sc_imask); ath_start(ifp); /* restart xmit */ return 0; } static int ath_reset_vap(struct ieee80211vap *vap, u_long cmd) { struct ieee80211com *ic = vap->iv_ic; struct ifnet *ifp = ic->ic_ifp; struct ath_softc *sc = ifp->if_softc; struct ath_hal *ah = sc->sc_ah; switch (cmd) { case IEEE80211_IOC_TXPOWER: /* * If per-packet TPC is enabled, then we have nothing * to do; otherwise we need to force the global limit. * All this can happen directly; no need to reset. */ if (!ath_hal_gettpc(ah)) ath_hal_settxpowlimit(ah, ic->ic_txpowlimit); return 0; } return ath_reset(ifp); } struct ath_buf * _ath_getbuf_locked(struct ath_softc *sc) { struct ath_buf *bf; ATH_TXBUF_LOCK_ASSERT(sc); bf = STAILQ_FIRST(&sc->sc_txbuf); if (bf != NULL && (bf->bf_flags & ATH_BUF_BUSY) == 0) STAILQ_REMOVE_HEAD(&sc->sc_txbuf, bf_list); else bf = NULL; if (bf == NULL) { DPRINTF(sc, ATH_DEBUG_XMIT, "%s: %s\n", __func__, STAILQ_FIRST(&sc->sc_txbuf) == NULL ? "out of xmit buffers" : "xmit buffer busy"); } return bf; } struct ath_buf * ath_getbuf(struct ath_softc *sc) { struct ath_buf *bf; ATH_TXBUF_LOCK(sc); bf = _ath_getbuf_locked(sc); if (bf == NULL) { struct ifnet *ifp = sc->sc_ifp; DPRINTF(sc, ATH_DEBUG_XMIT, "%s: stop queue\n", __func__); sc->sc_stats.ast_tx_qstop++; ifp->if_drv_flags |= IFF_DRV_OACTIVE; } ATH_TXBUF_UNLOCK(sc); return bf; } static void ath_start(struct ifnet *ifp) { struct ath_softc *sc = ifp->if_softc; struct ieee80211_node *ni; struct ath_buf *bf; struct mbuf *m, *next; ath_bufhead frags; if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0 || sc->sc_invalid) return; for (;;) { /* * Grab a TX buffer and associated resources. */ bf = ath_getbuf(sc); if (bf == NULL) break; IFQ_DEQUEUE(&ifp->if_snd, m); if (m == NULL) { ATH_TXBUF_LOCK(sc); STAILQ_INSERT_HEAD(&sc->sc_txbuf, bf, bf_list); ATH_TXBUF_UNLOCK(sc); break; } ni = (struct ieee80211_node *) m->m_pkthdr.rcvif; /* * Check for fragmentation. If this frame * has been broken up verify we have enough * buffers to send all the fragments so all * go out or none... */ STAILQ_INIT(&frags); if ((m->m_flags & M_FRAG) && !ath_txfrag_setup(sc, &frags, m, ni)) { DPRINTF(sc, ATH_DEBUG_XMIT, "%s: out of txfrag buffers\n", __func__); sc->sc_stats.ast_tx_nofrag++; ifp->if_oerrors++; ath_freetx(m); goto bad; } ifp->if_opackets++; nextfrag: /* * Pass the frame to the h/w for transmission. * Fragmented frames have each frag chained together * with m_nextpkt. We know there are sufficient ath_buf's * to send all the frags because of work done by * ath_txfrag_setup. We leave m_nextpkt set while * calling ath_tx_start so it can use it to extend the * the tx duration to cover the subsequent frag and * so it can reclaim all the mbufs in case of an error; * ath_tx_start clears m_nextpkt once it commits to * handing the frame to the hardware. */ next = m->m_nextpkt; if (ath_tx_start(sc, ni, bf, m)) { bad: ifp->if_oerrors++; reclaim: bf->bf_m = NULL; bf->bf_node = NULL; ATH_TXBUF_LOCK(sc); STAILQ_INSERT_HEAD(&sc->sc_txbuf, bf, bf_list); ath_txfrag_cleanup(sc, &frags, ni); ATH_TXBUF_UNLOCK(sc); if (ni != NULL) ieee80211_free_node(ni); continue; } if (next != NULL) { /* * Beware of state changing between frags. * XXX check sta power-save state? */ if (ni->ni_vap->iv_state != IEEE80211_S_RUN) { DPRINTF(sc, ATH_DEBUG_XMIT, "%s: flush fragmented packet, state %s\n", __func__, ieee80211_state_name[ni->ni_vap->iv_state]); ath_freetx(next); goto reclaim; } m = next; bf = STAILQ_FIRST(&frags); KASSERT(bf != NULL, ("no buf for txfrag")); STAILQ_REMOVE_HEAD(&frags, bf_list); goto nextfrag; } sc->sc_wd_timer = 5; } } static int ath_media_change(struct ifnet *ifp) { int error = ieee80211_media_change(ifp); /* NB: only the fixed rate can change and that doesn't need a reset */ return (error == ENETRESET ? 0 : error); } /* * Block/unblock tx+rx processing while a key change is done. * We assume the caller serializes key management operations * so we only need to worry about synchronization with other * uses that originate in the driver. */ static void ath_key_update_begin(struct ieee80211vap *vap) { struct ifnet *ifp = vap->iv_ic->ic_ifp; struct ath_softc *sc = ifp->if_softc; DPRINTF(sc, ATH_DEBUG_KEYCACHE, "%s:\n", __func__); taskqueue_block(sc->sc_tq); IF_LOCK(&ifp->if_snd); /* NB: doesn't block mgmt frames */ } static void ath_key_update_end(struct ieee80211vap *vap) { struct ifnet *ifp = vap->iv_ic->ic_ifp; struct ath_softc *sc = ifp->if_softc; DPRINTF(sc, ATH_DEBUG_KEYCACHE, "%s:\n", __func__); IF_UNLOCK(&ifp->if_snd); taskqueue_unblock(sc->sc_tq); } /* * Calculate the receive filter according to the * operating mode and state: * * o always accept unicast, broadcast, and multicast traffic * o accept PHY error frames when hardware doesn't have MIB support * to count and we need them for ANI (sta mode only until recently) * and we are not scanning (ANI is disabled) * NB: older hal's add rx filter bits out of sight and we need to * blindly preserve them * o probe request frames are accepted only when operating in * hostap, adhoc, mesh, or monitor modes * o enable promiscuous mode * - when in monitor mode * - if interface marked PROMISC (assumes bridge setting is filtered) * o accept beacons: * - when operating in station mode for collecting rssi data when * the station is otherwise quiet, or * - when operating in adhoc mode so the 802.11 layer creates * node table entries for peers, * - when scanning * - when doing s/w beacon miss (e.g. for ap+sta) * - when operating in ap mode in 11g to detect overlapping bss that * require protection * - when operating in mesh mode to detect neighbors * o accept control frames: * - when in monitor mode * XXX HT protection for 11n */ static u_int32_t ath_calcrxfilter(struct ath_softc *sc) { struct ifnet *ifp = sc->sc_ifp; struct ieee80211com *ic = ifp->if_l2com; u_int32_t rfilt; rfilt = HAL_RX_FILTER_UCAST | HAL_RX_FILTER_BCAST | HAL_RX_FILTER_MCAST; if (!sc->sc_needmib && !sc->sc_scanning) rfilt |= HAL_RX_FILTER_PHYERR; if (ic->ic_opmode != IEEE80211_M_STA) rfilt |= HAL_RX_FILTER_PROBEREQ; /* XXX ic->ic_monvaps != 0? */ if (ic->ic_opmode == IEEE80211_M_MONITOR || (ifp->if_flags & IFF_PROMISC)) rfilt |= HAL_RX_FILTER_PROM; if (ic->ic_opmode == IEEE80211_M_STA || ic->ic_opmode == IEEE80211_M_IBSS || sc->sc_swbmiss || sc->sc_scanning) rfilt |= HAL_RX_FILTER_BEACON; /* * NB: We don't recalculate the rx filter when * ic_protmode changes; otherwise we could do * this only when ic_protmode != NONE. */ if (ic->ic_opmode == IEEE80211_M_HOSTAP && IEEE80211_IS_CHAN_ANYG(ic->ic_curchan)) rfilt |= HAL_RX_FILTER_BEACON; /* * Enable hardware PS-POLL RX only for hostap mode; * STA mode sends PS-POLL frames but never * receives them. */ if (ath_hal_getcapability(sc->sc_ah, HAL_CAP_PSPOLL, 0, NULL) == HAL_OK && ic->ic_opmode == IEEE80211_M_HOSTAP) rfilt |= HAL_RX_FILTER_PSPOLL; if (sc->sc_nmeshvaps) { rfilt |= HAL_RX_FILTER_BEACON; if (sc->sc_hasbmatch) rfilt |= HAL_RX_FILTER_BSSID; else rfilt |= HAL_RX_FILTER_PROM; } if (ic->ic_opmode == IEEE80211_M_MONITOR) rfilt |= HAL_RX_FILTER_CONTROL; if (sc->sc_dodfs) { rfilt |= HAL_RX_FILTER_PHYRADAR; } /* * Enable RX of compressed BAR frames only when doing * 802.11n. Required for A-MPDU. */ if (IEEE80211_IS_CHAN_HT(ic->ic_curchan)) rfilt |= HAL_RX_FILTER_COMPBAR; DPRINTF(sc, ATH_DEBUG_MODE, "%s: RX filter 0x%x, %s if_flags 0x%x\n", __func__, rfilt, ieee80211_opmode_name[ic->ic_opmode], ifp->if_flags); return rfilt; } static void ath_update_promisc(struct ifnet *ifp) { struct ath_softc *sc = ifp->if_softc; u_int32_t rfilt; /* configure rx filter */ rfilt = ath_calcrxfilter(sc); ath_hal_setrxfilter(sc->sc_ah, rfilt); DPRINTF(sc, ATH_DEBUG_MODE, "%s: RX filter 0x%x\n", __func__, rfilt); } static void ath_update_mcast(struct ifnet *ifp) { struct ath_softc *sc = ifp->if_softc; u_int32_t mfilt[2]; /* calculate and install multicast filter */ if ((ifp->if_flags & IFF_ALLMULTI) == 0) { struct ifmultiaddr *ifma; /* * Merge multicast addresses to form the hardware filter. */ mfilt[0] = mfilt[1] = 0; if_maddr_rlock(ifp); /* XXX need some fiddling to remove? */ TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) { caddr_t dl; u_int32_t val; u_int8_t pos; /* calculate XOR of eight 6bit values */ dl = LLADDR((struct sockaddr_dl *) ifma->ifma_addr); val = LE_READ_4(dl + 0); pos = (val >> 18) ^ (val >> 12) ^ (val >> 6) ^ val; val = LE_READ_4(dl + 3); pos ^= (val >> 18) ^ (val >> 12) ^ (val >> 6) ^ val; pos &= 0x3f; mfilt[pos / 32] |= (1 << (pos % 32)); } if_maddr_runlock(ifp); } else mfilt[0] = mfilt[1] = ~0; ath_hal_setmcastfilter(sc->sc_ah, mfilt[0], mfilt[1]); DPRINTF(sc, ATH_DEBUG_MODE, "%s: MC filter %08x:%08x\n", __func__, mfilt[0], mfilt[1]); } static void ath_mode_init(struct ath_softc *sc) { struct ifnet *ifp = sc->sc_ifp; struct ath_hal *ah = sc->sc_ah; u_int32_t rfilt; /* configure rx filter */ rfilt = ath_calcrxfilter(sc); ath_hal_setrxfilter(ah, rfilt); /* configure operational mode */ ath_hal_setopmode(ah); /* handle any link-level address change */ ath_hal_setmac(ah, IF_LLADDR(ifp)); /* calculate and install multicast filter */ ath_update_mcast(ifp); } /* * Set the slot time based on the current setting. */ static void ath_setslottime(struct ath_softc *sc) { struct ieee80211com *ic = sc->sc_ifp->if_l2com; struct ath_hal *ah = sc->sc_ah; u_int usec; if (IEEE80211_IS_CHAN_HALF(ic->ic_curchan)) usec = 13; else if (IEEE80211_IS_CHAN_QUARTER(ic->ic_curchan)) usec = 21; else if (IEEE80211_IS_CHAN_ANYG(ic->ic_curchan)) { /* honor short/long slot time only in 11g */ /* XXX shouldn't honor on pure g or turbo g channel */ if (ic->ic_flags & IEEE80211_F_SHSLOT) usec = HAL_SLOT_TIME_9; else usec = HAL_SLOT_TIME_20; } else usec = HAL_SLOT_TIME_9; DPRINTF(sc, ATH_DEBUG_RESET, "%s: chan %u MHz flags 0x%x %s slot, %u usec\n", __func__, ic->ic_curchan->ic_freq, ic->ic_curchan->ic_flags, ic->ic_flags & IEEE80211_F_SHSLOT ? "short" : "long", usec); ath_hal_setslottime(ah, usec); sc->sc_updateslot = OK; } /* * Callback from the 802.11 layer to update the * slot time based on the current setting. */ static void ath_updateslot(struct ifnet *ifp) { struct ath_softc *sc = ifp->if_softc; struct ieee80211com *ic = ifp->if_l2com; /* * When not coordinating the BSS, change the hardware * immediately. For other operation we defer the change * until beacon updates have propagated to the stations. */ if (ic->ic_opmode == IEEE80211_M_HOSTAP || ic->ic_opmode == IEEE80211_M_MBSS) sc->sc_updateslot = UPDATE; else ath_setslottime(sc); } /* * Setup a h/w transmit queue for beacons. */ static int ath_beaconq_setup(struct ath_hal *ah) { HAL_TXQ_INFO qi; memset(&qi, 0, sizeof(qi)); qi.tqi_aifs = HAL_TXQ_USEDEFAULT; qi.tqi_cwmin = HAL_TXQ_USEDEFAULT; qi.tqi_cwmax = HAL_TXQ_USEDEFAULT; /* NB: for dynamic turbo, don't enable any other interrupts */ qi.tqi_qflags = HAL_TXQ_TXDESCINT_ENABLE; return ath_hal_setuptxqueue(ah, HAL_TX_QUEUE_BEACON, &qi); } /* * Setup the transmit queue parameters for the beacon queue. */ static int ath_beaconq_config(struct ath_softc *sc) { #define ATH_EXPONENT_TO_VALUE(v) ((1<<(v))-1) struct ieee80211com *ic = sc->sc_ifp->if_l2com; struct ath_hal *ah = sc->sc_ah; HAL_TXQ_INFO qi; ath_hal_gettxqueueprops(ah, sc->sc_bhalq, &qi); if (ic->ic_opmode == IEEE80211_M_HOSTAP || ic->ic_opmode == IEEE80211_M_MBSS) { /* * Always burst out beacon and CAB traffic. */ qi.tqi_aifs = ATH_BEACON_AIFS_DEFAULT; qi.tqi_cwmin = ATH_BEACON_CWMIN_DEFAULT; qi.tqi_cwmax = ATH_BEACON_CWMAX_DEFAULT; } else { struct wmeParams *wmep = &ic->ic_wme.wme_chanParams.cap_wmeParams[WME_AC_BE]; /* * Adhoc mode; important thing is to use 2x cwmin. */ qi.tqi_aifs = wmep->wmep_aifsn; qi.tqi_cwmin = 2*ATH_EXPONENT_TO_VALUE(wmep->wmep_logcwmin); qi.tqi_cwmax = ATH_EXPONENT_TO_VALUE(wmep->wmep_logcwmax); } if (!ath_hal_settxqueueprops(ah, sc->sc_bhalq, &qi)) { device_printf(sc->sc_dev, "unable to update parameters for " "beacon hardware queue!\n"); return 0; } else { ath_hal_resettxqueue(ah, sc->sc_bhalq); /* push to h/w */ return 1; } #undef ATH_EXPONENT_TO_VALUE } /* * Allocate and setup an initial beacon frame. */ static int ath_beacon_alloc(struct ath_softc *sc, struct ieee80211_node *ni) { struct ieee80211vap *vap = ni->ni_vap; struct ath_vap *avp = ATH_VAP(vap); struct ath_buf *bf; struct mbuf *m; int error; bf = avp->av_bcbuf; if (bf->bf_m != NULL) { bus_dmamap_unload(sc->sc_dmat, bf->bf_dmamap); m_freem(bf->bf_m); bf->bf_m = NULL; } if (bf->bf_node != NULL) { ieee80211_free_node(bf->bf_node); bf->bf_node = NULL; } /* * NB: the beacon data buffer must be 32-bit aligned; * we assume the mbuf routines will return us something * with this alignment (perhaps should assert). */ m = ieee80211_beacon_alloc(ni, &avp->av_boff); if (m == NULL) { device_printf(sc->sc_dev, "%s: cannot get mbuf\n", __func__); sc->sc_stats.ast_be_nombuf++; return ENOMEM; } error = bus_dmamap_load_mbuf_sg(sc->sc_dmat, bf->bf_dmamap, m, bf->bf_segs, &bf->bf_nseg, BUS_DMA_NOWAIT); if (error != 0) { device_printf(sc->sc_dev, "%s: cannot map mbuf, bus_dmamap_load_mbuf_sg returns %d\n", __func__, error); m_freem(m); return error; } /* * Calculate a TSF adjustment factor required for staggered * beacons. Note that we assume the format of the beacon * frame leaves the tstamp field immediately following the * header. */ if (sc->sc_stagbeacons && avp->av_bslot > 0) { uint64_t tsfadjust; struct ieee80211_frame *wh; /* * The beacon interval is in TU's; the TSF is in usecs. * We figure out how many TU's to add to align the timestamp * then convert to TSF units and handle byte swapping before * inserting it in the frame. The hardware will then add this * each time a beacon frame is sent. Note that we align vap's * 1..N and leave vap 0 untouched. This means vap 0 has a * timestamp in one beacon interval while the others get a * timstamp aligned to the next interval. */ tsfadjust = ni->ni_intval * (ATH_BCBUF - avp->av_bslot) / ATH_BCBUF; tsfadjust = htole64(tsfadjust << 10); /* TU -> TSF */ DPRINTF(sc, ATH_DEBUG_BEACON, "%s: %s beacons bslot %d intval %u tsfadjust %llu\n", __func__, sc->sc_stagbeacons ? "stagger" : "burst", avp->av_bslot, ni->ni_intval, (long long unsigned) le64toh(tsfadjust)); wh = mtod(m, struct ieee80211_frame *); memcpy(&wh[1], &tsfadjust, sizeof(tsfadjust)); } bf->bf_m = m; bf->bf_node = ieee80211_ref_node(ni); return 0; } /* * Setup the beacon frame for transmit. */ static void ath_beacon_setup(struct ath_softc *sc, struct ath_buf *bf) { #define USE_SHPREAMBLE(_ic) \ (((_ic)->ic_flags & (IEEE80211_F_SHPREAMBLE | IEEE80211_F_USEBARKER))\ == IEEE80211_F_SHPREAMBLE) struct ieee80211_node *ni = bf->bf_node; struct ieee80211com *ic = ni->ni_ic; struct mbuf *m = bf->bf_m; struct ath_hal *ah = sc->sc_ah; struct ath_desc *ds; int flags, antenna; const HAL_RATE_TABLE *rt; u_int8_t rix, rate; DPRINTF(sc, ATH_DEBUG_BEACON_PROC, "%s: m %p len %u\n", __func__, m, m->m_len); /* setup descriptors */ ds = bf->bf_desc; flags = HAL_TXDESC_NOACK; if (ic->ic_opmode == IEEE80211_M_IBSS && sc->sc_hasveol) { ds->ds_link = bf->bf_daddr; /* self-linked */ flags |= HAL_TXDESC_VEOL; /* * Let hardware handle antenna switching. */ antenna = sc->sc_txantenna; } else { ds->ds_link = 0; /* * Switch antenna every 4 beacons. * XXX assumes two antenna */ if (sc->sc_txantenna != 0) antenna = sc->sc_txantenna; else if (sc->sc_stagbeacons && sc->sc_nbcnvaps != 0) antenna = ((sc->sc_stats.ast_be_xmit / sc->sc_nbcnvaps) & 4 ? 2 : 1); else antenna = (sc->sc_stats.ast_be_xmit & 4 ? 2 : 1); } KASSERT(bf->bf_nseg == 1, ("multi-segment beacon frame; nseg %u", bf->bf_nseg)); ds->ds_data = bf->bf_segs[0].ds_addr; /* * Calculate rate code. * XXX everything at min xmit rate */ rix = 0; rt = sc->sc_currates; rate = rt->info[rix].rateCode; if (USE_SHPREAMBLE(ic)) rate |= rt->info[rix].shortPreamble; ath_hal_setuptxdesc(ah, ds , m->m_len + IEEE80211_CRC_LEN /* frame length */ , sizeof(struct ieee80211_frame)/* header length */ , HAL_PKT_TYPE_BEACON /* Atheros packet type */ , ni->ni_txpower /* txpower XXX */ , rate, 1 /* series 0 rate/tries */ , HAL_TXKEYIX_INVALID /* no encryption */ , antenna /* antenna mode */ , flags /* no ack, veol for beacons */ , 0 /* rts/cts rate */ , 0 /* rts/cts duration */ ); /* NB: beacon's BufLen must be a multiple of 4 bytes */ ath_hal_filltxdesc(ah, ds , roundup(m->m_len, 4) /* buffer length */ , AH_TRUE /* first segment */ , AH_TRUE /* last segment */ , ds /* first descriptor */ ); #if 0 ath_desc_swap(ds); #endif #undef USE_SHPREAMBLE } static void ath_beacon_update(struct ieee80211vap *vap, int item) { struct ieee80211_beacon_offsets *bo = &ATH_VAP(vap)->av_boff; setbit(bo->bo_flags, item); } /* * Append the contents of src to dst; both queues * are assumed to be locked. */ static void ath_txqmove(struct ath_txq *dst, struct ath_txq *src) { STAILQ_CONCAT(&dst->axq_q, &src->axq_q); dst->axq_link = src->axq_link; src->axq_link = NULL; dst->axq_depth += src->axq_depth; src->axq_depth = 0; } /* * Transmit a beacon frame at SWBA. Dynamic updates to the * frame contents are done as needed and the slot time is * also adjusted based on current state. */ static void ath_beacon_proc(void *arg, int pending) { struct ath_softc *sc = arg; struct ath_hal *ah = sc->sc_ah; struct ieee80211vap *vap; struct ath_buf *bf; int slot, otherant; uint32_t bfaddr; DPRINTF(sc, ATH_DEBUG_BEACON_PROC, "%s: pending %u\n", __func__, pending); /* * Check if the previous beacon has gone out. If * not don't try to post another, skip this period * and wait for the next. Missed beacons indicate * a problem and should not occur. If we miss too * many consecutive beacons reset the device. */ if (ath_hal_numtxpending(ah, sc->sc_bhalq) != 0) { sc->sc_bmisscount++; sc->sc_stats.ast_be_missed++; DPRINTF(sc, ATH_DEBUG_BEACON, "%s: missed %u consecutive beacons\n", __func__, sc->sc_bmisscount); if (sc->sc_bmisscount >= ath_bstuck_threshold) taskqueue_enqueue(sc->sc_tq, &sc->sc_bstucktask); return; } if (sc->sc_bmisscount != 0) { DPRINTF(sc, ATH_DEBUG_BEACON, "%s: resume beacon xmit after %u misses\n", __func__, sc->sc_bmisscount); sc->sc_bmisscount = 0; } if (sc->sc_stagbeacons) { /* staggered beacons */ struct ieee80211com *ic = sc->sc_ifp->if_l2com; uint32_t tsftu; tsftu = ath_hal_gettsf32(ah) >> 10; /* XXX lintval */ slot = ((tsftu % ic->ic_lintval) * ATH_BCBUF) / ic->ic_lintval; vap = sc->sc_bslot[(slot+1) % ATH_BCBUF]; bfaddr = 0; if (vap != NULL && vap->iv_state >= IEEE80211_S_RUN) { bf = ath_beacon_generate(sc, vap); if (bf != NULL) bfaddr = bf->bf_daddr; } } else { /* burst'd beacons */ uint32_t *bflink = &bfaddr; for (slot = 0; slot < ATH_BCBUF; slot++) { vap = sc->sc_bslot[slot]; if (vap != NULL && vap->iv_state >= IEEE80211_S_RUN) { bf = ath_beacon_generate(sc, vap); if (bf != NULL) { *bflink = bf->bf_daddr; bflink = &bf->bf_desc->ds_link; } } } *bflink = 0; /* terminate list */ } /* * Handle slot time change when a non-ERP station joins/leaves * an 11g network. The 802.11 layer notifies us via callback, * we mark updateslot, then wait one beacon before effecting * the change. This gives associated stations at least one * beacon interval to note the state change. */ /* XXX locking */ if (sc->sc_updateslot == UPDATE) { sc->sc_updateslot = COMMIT; /* commit next beacon */ sc->sc_slotupdate = slot; } else if (sc->sc_updateslot == COMMIT && sc->sc_slotupdate == slot) ath_setslottime(sc); /* commit change to h/w */ /* * Check recent per-antenna transmit statistics and flip * the default antenna if noticeably more frames went out * on the non-default antenna. * XXX assumes 2 anntenae */ if (!sc->sc_diversity && (!sc->sc_stagbeacons || slot == 0)) { otherant = sc->sc_defant & 1 ? 2 : 1; if (sc->sc_ant_tx[otherant] > sc->sc_ant_tx[sc->sc_defant] + 2) ath_setdefantenna(sc, otherant); sc->sc_ant_tx[1] = sc->sc_ant_tx[2] = 0; } if (bfaddr != 0) { /* * Stop any current dma and put the new frame on the queue. * This should never fail since we check above that no frames * are still pending on the queue. */ if (!ath_hal_stoptxdma(ah, sc->sc_bhalq)) { DPRINTF(sc, ATH_DEBUG_ANY, "%s: beacon queue %u did not stop?\n", __func__, sc->sc_bhalq); } /* NB: cabq traffic should already be queued and primed */ ath_hal_puttxbuf(ah, sc->sc_bhalq, bfaddr); ath_hal_txstart(ah, sc->sc_bhalq); sc->sc_stats.ast_be_xmit++; } } static struct ath_buf * ath_beacon_generate(struct ath_softc *sc, struct ieee80211vap *vap) { struct ath_vap *avp = ATH_VAP(vap); struct ath_txq *cabq = sc->sc_cabq; struct ath_buf *bf; struct mbuf *m; int nmcastq, error; KASSERT(vap->iv_state >= IEEE80211_S_RUN, ("not running, state %d", vap->iv_state)); KASSERT(avp->av_bcbuf != NULL, ("no beacon buffer")); /* * Update dynamic beacon contents. If this returns * non-zero then we need to remap the memory because * the beacon frame changed size (probably because * of the TIM bitmap). */ bf = avp->av_bcbuf; m = bf->bf_m; nmcastq = avp->av_mcastq.axq_depth; if (ieee80211_beacon_update(bf->bf_node, &avp->av_boff, m, nmcastq)) { /* XXX too conservative? */ bus_dmamap_unload(sc->sc_dmat, bf->bf_dmamap); error = bus_dmamap_load_mbuf_sg(sc->sc_dmat, bf->bf_dmamap, m, bf->bf_segs, &bf->bf_nseg, BUS_DMA_NOWAIT); if (error != 0) { if_printf(vap->iv_ifp, "%s: bus_dmamap_load_mbuf_sg failed, error %u\n", __func__, error); return NULL; } } if ((avp->av_boff.bo_tim[4] & 1) && cabq->axq_depth) { DPRINTF(sc, ATH_DEBUG_BEACON, "%s: cabq did not drain, mcastq %u cabq %u\n", __func__, nmcastq, cabq->axq_depth); sc->sc_stats.ast_cabq_busy++; if (sc->sc_nvaps > 1 && sc->sc_stagbeacons) { /* * CABQ traffic from a previous vap is still pending. * We must drain the q before this beacon frame goes * out as otherwise this vap's stations will get cab * frames from a different vap. * XXX could be slow causing us to miss DBA */ ath_tx_draintxq(sc, cabq); } } ath_beacon_setup(sc, bf); bus_dmamap_sync(sc->sc_dmat, bf->bf_dmamap, BUS_DMASYNC_PREWRITE); /* * Enable the CAB queue before the beacon queue to * insure cab frames are triggered by this beacon. */ if (avp->av_boff.bo_tim[4] & 1) { struct ath_hal *ah = sc->sc_ah; /* NB: only at DTIM */ ATH_TXQ_LOCK(cabq); ATH_TXQ_LOCK(&avp->av_mcastq); if (nmcastq) { struct ath_buf *bfm; /* * Move frames from the s/w mcast q to the h/w cab q. * XXX MORE_DATA bit */ bfm = STAILQ_FIRST(&avp->av_mcastq.axq_q); if (cabq->axq_link != NULL) { *cabq->axq_link = bfm->bf_daddr; } else ath_hal_puttxbuf(ah, cabq->axq_qnum, bfm->bf_daddr); ath_txqmove(cabq, &avp->av_mcastq); sc->sc_stats.ast_cabq_xmit += nmcastq; } /* NB: gated by beacon so safe to start here */ ath_hal_txstart(ah, cabq->axq_qnum); ATH_TXQ_UNLOCK(cabq); ATH_TXQ_UNLOCK(&avp->av_mcastq); } return bf; } static void ath_beacon_start_adhoc(struct ath_softc *sc, struct ieee80211vap *vap) { struct ath_vap *avp = ATH_VAP(vap); struct ath_hal *ah = sc->sc_ah; struct ath_buf *bf; struct mbuf *m; int error; KASSERT(avp->av_bcbuf != NULL, ("no beacon buffer")); /* * Update dynamic beacon contents. If this returns * non-zero then we need to remap the memory because * the beacon frame changed size (probably because * of the TIM bitmap). */ bf = avp->av_bcbuf; m = bf->bf_m; if (ieee80211_beacon_update(bf->bf_node, &avp->av_boff, m, 0)) { /* XXX too conservative? */ bus_dmamap_unload(sc->sc_dmat, bf->bf_dmamap); error = bus_dmamap_load_mbuf_sg(sc->sc_dmat, bf->bf_dmamap, m, bf->bf_segs, &bf->bf_nseg, BUS_DMA_NOWAIT); if (error != 0) { if_printf(vap->iv_ifp, "%s: bus_dmamap_load_mbuf_sg failed, error %u\n", __func__, error); return; } } ath_beacon_setup(sc, bf); bus_dmamap_sync(sc->sc_dmat, bf->bf_dmamap, BUS_DMASYNC_PREWRITE); /* NB: caller is known to have already stopped tx dma */ ath_hal_puttxbuf(ah, sc->sc_bhalq, bf->bf_daddr); ath_hal_txstart(ah, sc->sc_bhalq); } /* * Reset the hardware after detecting beacons have stopped. */ static void ath_bstuck_proc(void *arg, int pending) { struct ath_softc *sc = arg; struct ifnet *ifp = sc->sc_ifp; if_printf(ifp, "stuck beacon; resetting (bmiss count %u)\n", sc->sc_bmisscount); sc->sc_stats.ast_bstuck++; ath_reset(ifp); } /* * Reclaim beacon resources and return buffer to the pool. */ static void ath_beacon_return(struct ath_softc *sc, struct ath_buf *bf) { if (bf->bf_m != NULL) { bus_dmamap_unload(sc->sc_dmat, bf->bf_dmamap); m_freem(bf->bf_m); bf->bf_m = NULL; } if (bf->bf_node != NULL) { ieee80211_free_node(bf->bf_node); bf->bf_node = NULL; } STAILQ_INSERT_TAIL(&sc->sc_bbuf, bf, bf_list); } /* * Reclaim beacon resources. */ static void ath_beacon_free(struct ath_softc *sc) { struct ath_buf *bf; STAILQ_FOREACH(bf, &sc->sc_bbuf, bf_list) { if (bf->bf_m != NULL) { bus_dmamap_unload(sc->sc_dmat, bf->bf_dmamap); m_freem(bf->bf_m); bf->bf_m = NULL; } if (bf->bf_node != NULL) { ieee80211_free_node(bf->bf_node); bf->bf_node = NULL; } } } /* * Configure the beacon and sleep timers. * * When operating as an AP this resets the TSF and sets * up the hardware to notify us when we need to issue beacons. * * When operating in station mode this sets up the beacon * timers according to the timestamp of the last received * beacon and the current TSF, configures PCF and DTIM * handling, programs the sleep registers so the hardware * will wakeup in time to receive beacons, and configures * the beacon miss handling so we'll receive a BMISS * interrupt when we stop seeing beacons from the AP * we've associated with. */ static void ath_beacon_config(struct ath_softc *sc, struct ieee80211vap *vap) { #define TSF_TO_TU(_h,_l) \ ((((u_int32_t)(_h)) << 22) | (((u_int32_t)(_l)) >> 10)) #define FUDGE 2 struct ath_hal *ah = sc->sc_ah; struct ieee80211com *ic = sc->sc_ifp->if_l2com; struct ieee80211_node *ni; u_int32_t nexttbtt, intval, tsftu; u_int64_t tsf; if (vap == NULL) vap = TAILQ_FIRST(&ic->ic_vaps); /* XXX */ ni = vap->iv_bss; /* extract tstamp from last beacon and convert to TU */ nexttbtt = TSF_TO_TU(LE_READ_4(ni->ni_tstamp.data + 4), LE_READ_4(ni->ni_tstamp.data)); if (ic->ic_opmode == IEEE80211_M_HOSTAP || ic->ic_opmode == IEEE80211_M_MBSS) { /* * For multi-bss ap/mesh support beacons are either staggered * evenly over N slots or burst together. For the former * arrange for the SWBA to be delivered for each slot. * Slots that are not occupied will generate nothing. */ /* NB: the beacon interval is kept internally in TU's */ intval = ni->ni_intval & HAL_BEACON_PERIOD; if (sc->sc_stagbeacons) intval /= ATH_BCBUF; } else { /* NB: the beacon interval is kept internally in TU's */ intval = ni->ni_intval & HAL_BEACON_PERIOD; } if (nexttbtt == 0) /* e.g. for ap mode */ nexttbtt = intval; else if (intval) /* NB: can be 0 for monitor mode */ nexttbtt = roundup(nexttbtt, intval); DPRINTF(sc, ATH_DEBUG_BEACON, "%s: nexttbtt %u intval %u (%u)\n", __func__, nexttbtt, intval, ni->ni_intval); if (ic->ic_opmode == IEEE80211_M_STA && !sc->sc_swbmiss) { HAL_BEACON_STATE bs; int dtimperiod, dtimcount; int cfpperiod, cfpcount; /* * Setup dtim and cfp parameters according to * last beacon we received (which may be none). */ dtimperiod = ni->ni_dtim_period; if (dtimperiod <= 0) /* NB: 0 if not known */ dtimperiod = 1; dtimcount = ni->ni_dtim_count; if (dtimcount >= dtimperiod) /* NB: sanity check */ dtimcount = 0; /* XXX? */ cfpperiod = 1; /* NB: no PCF support yet */ cfpcount = 0; /* * Pull nexttbtt forward to reflect the current * TSF and calculate dtim+cfp state for the result. */ tsf = ath_hal_gettsf64(ah); tsftu = TSF_TO_TU(tsf>>32, tsf) + FUDGE; do { nexttbtt += intval; if (--dtimcount < 0) { dtimcount = dtimperiod - 1; if (--cfpcount < 0) cfpcount = cfpperiod - 1; } } while (nexttbtt < tsftu); memset(&bs, 0, sizeof(bs)); bs.bs_intval = intval; bs.bs_nexttbtt = nexttbtt; bs.bs_dtimperiod = dtimperiod*intval; bs.bs_nextdtim = bs.bs_nexttbtt + dtimcount*intval; bs.bs_cfpperiod = cfpperiod*bs.bs_dtimperiod; bs.bs_cfpnext = bs.bs_nextdtim + cfpcount*bs.bs_dtimperiod; bs.bs_cfpmaxduration = 0; #if 0 /* * The 802.11 layer records the offset to the DTIM * bitmap while receiving beacons; use it here to * enable h/w detection of our AID being marked in * the bitmap vector (to indicate frames for us are * pending at the AP). * XXX do DTIM handling in s/w to WAR old h/w bugs * XXX enable based on h/w rev for newer chips */ bs.bs_timoffset = ni->ni_timoff; #endif /* * Calculate the number of consecutive beacons to miss * before taking a BMISS interrupt. * Note that we clamp the result to at most 10 beacons. */ bs.bs_bmissthreshold = vap->iv_bmissthreshold; if (bs.bs_bmissthreshold > 10) bs.bs_bmissthreshold = 10; else if (bs.bs_bmissthreshold <= 0) bs.bs_bmissthreshold = 1; /* * Calculate sleep duration. The configuration is * given in ms. We insure a multiple of the beacon * period is used. Also, if the sleep duration is * greater than the DTIM period then it makes senses * to make it a multiple of that. * * XXX fixed at 100ms */ bs.bs_sleepduration = roundup(IEEE80211_MS_TO_TU(100), bs.bs_intval); if (bs.bs_sleepduration > bs.bs_dtimperiod) bs.bs_sleepduration = roundup(bs.bs_sleepduration, bs.bs_dtimperiod); DPRINTF(sc, ATH_DEBUG_BEACON, "%s: tsf %ju tsf:tu %u intval %u nexttbtt %u dtim %u nextdtim %u bmiss %u sleep %u cfp:period %u maxdur %u next %u timoffset %u\n" , __func__ , tsf, tsftu , bs.bs_intval , bs.bs_nexttbtt , bs.bs_dtimperiod , bs.bs_nextdtim , bs.bs_bmissthreshold , bs.bs_sleepduration , bs.bs_cfpperiod , bs.bs_cfpmaxduration , bs.bs_cfpnext , bs.bs_timoffset ); ath_hal_intrset(ah, 0); ath_hal_beacontimers(ah, &bs); sc->sc_imask |= HAL_INT_BMISS; ath_hal_intrset(ah, sc->sc_imask); } else { ath_hal_intrset(ah, 0); if (nexttbtt == intval) intval |= HAL_BEACON_RESET_TSF; if (ic->ic_opmode == IEEE80211_M_IBSS) { /* * In IBSS mode enable the beacon timers but only * enable SWBA interrupts if we need to manually * prepare beacon frames. Otherwise we use a * self-linked tx descriptor and let the hardware * deal with things. */ intval |= HAL_BEACON_ENA; if (!sc->sc_hasveol) sc->sc_imask |= HAL_INT_SWBA; if ((intval & HAL_BEACON_RESET_TSF) == 0) { /* * Pull nexttbtt forward to reflect * the current TSF. */ tsf = ath_hal_gettsf64(ah); tsftu = TSF_TO_TU(tsf>>32, tsf) + FUDGE; do { nexttbtt += intval; } while (nexttbtt < tsftu); } ath_beaconq_config(sc); } else if (ic->ic_opmode == IEEE80211_M_HOSTAP || ic->ic_opmode == IEEE80211_M_MBSS) { /* * In AP/mesh mode we enable the beacon timers * and SWBA interrupts to prepare beacon frames. */ intval |= HAL_BEACON_ENA; sc->sc_imask |= HAL_INT_SWBA; /* beacon prepare */ ath_beaconq_config(sc); } ath_hal_beaconinit(ah, nexttbtt, intval); sc->sc_bmisscount = 0; ath_hal_intrset(ah, sc->sc_imask); /* * When using a self-linked beacon descriptor in * ibss mode load it once here. */ if (ic->ic_opmode == IEEE80211_M_IBSS && sc->sc_hasveol) ath_beacon_start_adhoc(sc, vap); } sc->sc_syncbeacon = 0; #undef FUDGE #undef TSF_TO_TU } static void ath_load_cb(void *arg, bus_dma_segment_t *segs, int nsegs, int error) { bus_addr_t *paddr = (bus_addr_t*) arg; KASSERT(error == 0, ("error %u on bus_dma callback", error)); *paddr = segs->ds_addr; } static int ath_descdma_setup(struct ath_softc *sc, struct ath_descdma *dd, ath_bufhead *head, const char *name, int nbuf, int ndesc) { #define DS2PHYS(_dd, _ds) \ ((_dd)->dd_desc_paddr + ((caddr_t)(_ds) - (caddr_t)(_dd)->dd_desc)) struct ifnet *ifp = sc->sc_ifp; struct ath_desc *ds; struct ath_buf *bf; int i, bsize, error; DPRINTF(sc, ATH_DEBUG_RESET, "%s: %s DMA: %u buffers %u desc/buf\n", __func__, name, nbuf, ndesc); dd->dd_name = name; dd->dd_desc_len = sizeof(struct ath_desc) * nbuf * ndesc; /* * Setup DMA descriptor area. */ error = bus_dma_tag_create(bus_get_dma_tag(sc->sc_dev), /* parent */ PAGE_SIZE, 0, /* alignment, bounds */ BUS_SPACE_MAXADDR_32BIT, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ dd->dd_desc_len, /* maxsize */ 1, /* nsegments */ dd->dd_desc_len, /* maxsegsize */ BUS_DMA_ALLOCNOW, /* flags */ NULL, /* lockfunc */ NULL, /* lockarg */ &dd->dd_dmat); if (error != 0) { if_printf(ifp, "cannot allocate %s DMA tag\n", dd->dd_name); return error; } /* allocate descriptors */ error = bus_dmamap_create(dd->dd_dmat, BUS_DMA_NOWAIT, &dd->dd_dmamap); if (error != 0) { if_printf(ifp, "unable to create dmamap for %s descriptors, " "error %u\n", dd->dd_name, error); goto fail0; } error = bus_dmamem_alloc(dd->dd_dmat, (void**) &dd->dd_desc, BUS_DMA_NOWAIT | BUS_DMA_COHERENT, &dd->dd_dmamap); if (error != 0) { if_printf(ifp, "unable to alloc memory for %u %s descriptors, " "error %u\n", nbuf * ndesc, dd->dd_name, error); goto fail1; } error = bus_dmamap_load(dd->dd_dmat, dd->dd_dmamap, dd->dd_desc, dd->dd_desc_len, ath_load_cb, &dd->dd_desc_paddr, BUS_DMA_NOWAIT); if (error != 0) { if_printf(ifp, "unable to map %s descriptors, error %u\n", dd->dd_name, error); goto fail2; } ds = dd->dd_desc; DPRINTF(sc, ATH_DEBUG_RESET, "%s: %s DMA map: %p (%lu) -> %p (%lu)\n", __func__, dd->dd_name, ds, (u_long) dd->dd_desc_len, (caddr_t) dd->dd_desc_paddr, /*XXX*/ (u_long) dd->dd_desc_len); /* allocate rx buffers */ bsize = sizeof(struct ath_buf) * nbuf; bf = malloc(bsize, M_ATHDEV, M_NOWAIT | M_ZERO); if (bf == NULL) { if_printf(ifp, "malloc of %s buffers failed, size %u\n", dd->dd_name, bsize); goto fail3; } dd->dd_bufptr = bf; STAILQ_INIT(head); for (i = 0; i < nbuf; i++, bf++, ds += ndesc) { bf->bf_desc = ds; bf->bf_daddr = DS2PHYS(dd, ds); error = bus_dmamap_create(sc->sc_dmat, BUS_DMA_NOWAIT, &bf->bf_dmamap); if (error != 0) { if_printf(ifp, "unable to create dmamap for %s " "buffer %u, error %u\n", dd->dd_name, i, error); ath_descdma_cleanup(sc, dd, head); return error; } STAILQ_INSERT_TAIL(head, bf, bf_list); } return 0; fail3: bus_dmamap_unload(dd->dd_dmat, dd->dd_dmamap); fail2: bus_dmamem_free(dd->dd_dmat, dd->dd_desc, dd->dd_dmamap); fail1: bus_dmamap_destroy(dd->dd_dmat, dd->dd_dmamap); fail0: bus_dma_tag_destroy(dd->dd_dmat); memset(dd, 0, sizeof(*dd)); return error; #undef DS2PHYS } static void ath_descdma_cleanup(struct ath_softc *sc, struct ath_descdma *dd, ath_bufhead *head) { struct ath_buf *bf; struct ieee80211_node *ni; bus_dmamap_unload(dd->dd_dmat, dd->dd_dmamap); bus_dmamem_free(dd->dd_dmat, dd->dd_desc, dd->dd_dmamap); bus_dmamap_destroy(dd->dd_dmat, dd->dd_dmamap); bus_dma_tag_destroy(dd->dd_dmat); STAILQ_FOREACH(bf, head, bf_list) { if (bf->bf_m) { m_freem(bf->bf_m); bf->bf_m = NULL; } if (bf->bf_dmamap != NULL) { bus_dmamap_destroy(sc->sc_dmat, bf->bf_dmamap); bf->bf_dmamap = NULL; } ni = bf->bf_node; bf->bf_node = NULL; if (ni != NULL) { /* * Reclaim node reference. */ ieee80211_free_node(ni); } } STAILQ_INIT(head); free(dd->dd_bufptr, M_ATHDEV); memset(dd, 0, sizeof(*dd)); } static int ath_desc_alloc(struct ath_softc *sc) { int error; error = ath_descdma_setup(sc, &sc->sc_rxdma, &sc->sc_rxbuf, "rx", ath_rxbuf, 1); if (error != 0) return error; error = ath_descdma_setup(sc, &sc->sc_txdma, &sc->sc_txbuf, "tx", ath_txbuf, ATH_TXDESC); if (error != 0) { ath_descdma_cleanup(sc, &sc->sc_rxdma, &sc->sc_rxbuf); return error; } error = ath_descdma_setup(sc, &sc->sc_bdma, &sc->sc_bbuf, "beacon", ATH_BCBUF, 1); if (error != 0) { ath_descdma_cleanup(sc, &sc->sc_txdma, &sc->sc_txbuf); ath_descdma_cleanup(sc, &sc->sc_rxdma, &sc->sc_rxbuf); return error; } return 0; } static void ath_desc_free(struct ath_softc *sc) { if (sc->sc_bdma.dd_desc_len != 0) ath_descdma_cleanup(sc, &sc->sc_bdma, &sc->sc_bbuf); if (sc->sc_txdma.dd_desc_len != 0) ath_descdma_cleanup(sc, &sc->sc_txdma, &sc->sc_txbuf); if (sc->sc_rxdma.dd_desc_len != 0) ath_descdma_cleanup(sc, &sc->sc_rxdma, &sc->sc_rxbuf); } static struct ieee80211_node * ath_node_alloc(struct ieee80211vap *vap, const uint8_t mac[IEEE80211_ADDR_LEN]) { struct ieee80211com *ic = vap->iv_ic; struct ath_softc *sc = ic->ic_ifp->if_softc; const size_t space = sizeof(struct ath_node) + sc->sc_rc->arc_space; struct ath_node *an; an = malloc(space, M_80211_NODE, M_NOWAIT|M_ZERO); if (an == NULL) { /* XXX stat+msg */ return NULL; } ath_rate_node_init(sc, an); DPRINTF(sc, ATH_DEBUG_NODE, "%s: an %p\n", __func__, an); return &an->an_node; } static void ath_node_free(struct ieee80211_node *ni) { struct ieee80211com *ic = ni->ni_ic; struct ath_softc *sc = ic->ic_ifp->if_softc; DPRINTF(sc, ATH_DEBUG_NODE, "%s: ni %p\n", __func__, ni); ath_rate_node_cleanup(sc, ATH_NODE(ni)); sc->sc_node_free(ni); } static void ath_node_getsignal(const struct ieee80211_node *ni, int8_t *rssi, int8_t *noise) { struct ieee80211com *ic = ni->ni_ic; struct ath_softc *sc = ic->ic_ifp->if_softc; struct ath_hal *ah = sc->sc_ah; *rssi = ic->ic_node_getrssi(ni); if (ni->ni_chan != IEEE80211_CHAN_ANYC) *noise = ath_hal_getchannoise(ah, ni->ni_chan); else *noise = -95; /* nominally correct */ } static int ath_rxbuf_init(struct ath_softc *sc, struct ath_buf *bf) { struct ath_hal *ah = sc->sc_ah; int error; struct mbuf *m; struct ath_desc *ds; m = bf->bf_m; if (m == NULL) { /* * NB: by assigning a page to the rx dma buffer we * implicitly satisfy the Atheros requirement that * this buffer be cache-line-aligned and sized to be * multiple of the cache line size. Not doing this * causes weird stuff to happen (for the 5210 at least). */ m = m_getcl(M_DONTWAIT, MT_DATA, M_PKTHDR); if (m == NULL) { DPRINTF(sc, ATH_DEBUG_ANY, "%s: no mbuf/cluster\n", __func__); sc->sc_stats.ast_rx_nombuf++; return ENOMEM; } m->m_pkthdr.len = m->m_len = m->m_ext.ext_size; error = bus_dmamap_load_mbuf_sg(sc->sc_dmat, bf->bf_dmamap, m, bf->bf_segs, &bf->bf_nseg, BUS_DMA_NOWAIT); if (error != 0) { DPRINTF(sc, ATH_DEBUG_ANY, "%s: bus_dmamap_load_mbuf_sg failed; error %d\n", __func__, error); sc->sc_stats.ast_rx_busdma++; m_freem(m); return error; } KASSERT(bf->bf_nseg == 1, ("multi-segment packet; nseg %u", bf->bf_nseg)); bf->bf_m = m; } bus_dmamap_sync(sc->sc_dmat, bf->bf_dmamap, BUS_DMASYNC_PREREAD); /* * Setup descriptors. For receive we always terminate * the descriptor list with a self-linked entry so we'll * not get overrun under high load (as can happen with a * 5212 when ANI processing enables PHY error frames). * * To insure the last descriptor is self-linked we create * each descriptor as self-linked and add it to the end. As * each additional descriptor is added the previous self-linked * entry is ``fixed'' naturally. This should be safe even * if DMA is happening. When processing RX interrupts we * never remove/process the last, self-linked, entry on the * descriptor list. This insures the hardware always has * someplace to write a new frame. */ /* * 11N: we can no longer afford to self link the last descriptor. * MAC acknowledges BA status as long as it copies frames to host * buffer (or rx fifo). This can incorrectly acknowledge packets * to a sender if last desc is self-linked. */ ds = bf->bf_desc; if (sc->sc_rxslink) ds->ds_link = bf->bf_daddr; /* link to self */ else ds->ds_link = 0; /* terminate the list */ ds->ds_data = bf->bf_segs[0].ds_addr; ath_hal_setuprxdesc(ah, ds , m->m_len /* buffer size */ , 0 ); if (sc->sc_rxlink != NULL) *sc->sc_rxlink = bf->bf_daddr; sc->sc_rxlink = &ds->ds_link; return 0; } /* * Extend 15-bit time stamp from rx descriptor to * a full 64-bit TSF using the specified TSF. */ static __inline u_int64_t ath_extend_tsf(u_int32_t rstamp, u_int64_t tsf) { if ((tsf & 0x7fff) < rstamp) tsf -= 0x8000; return ((tsf &~ 0x7fff) | rstamp); } /* * Intercept management frames to collect beacon rssi data * and to do ibss merges. */ static void ath_recv_mgmt(struct ieee80211_node *ni, struct mbuf *m, int subtype, int rssi, int nf) { struct ieee80211vap *vap = ni->ni_vap; struct ath_softc *sc = vap->iv_ic->ic_ifp->if_softc; /* * Call up first so subsequent work can use information * potentially stored in the node (e.g. for ibss merge). */ ATH_VAP(vap)->av_recv_mgmt(ni, m, subtype, rssi, nf); switch (subtype) { case IEEE80211_FC0_SUBTYPE_BEACON: /* update rssi statistics for use by the hal */ ATH_RSSI_LPF(sc->sc_halstats.ns_avgbrssi, rssi); if (sc->sc_syncbeacon && ni == vap->iv_bss && vap->iv_state == IEEE80211_S_RUN) { /* * Resync beacon timers using the tsf of the beacon * frame we just received. */ ath_beacon_config(sc, vap); } /* fall thru... */ case IEEE80211_FC0_SUBTYPE_PROBE_RESP: if (vap->iv_opmode == IEEE80211_M_IBSS && vap->iv_state == IEEE80211_S_RUN) { uint32_t rstamp = sc->sc_lastrs->rs_tstamp; uint64_t tsf = ath_extend_tsf(rstamp, ath_hal_gettsf64(sc->sc_ah)); /* * Handle ibss merge as needed; check the tsf on the * frame before attempting the merge. The 802.11 spec * says the station should change it's bssid to match * the oldest station with the same ssid, where oldest * is determined by the tsf. Note that hardware * reconfiguration happens through callback to * ath_newstate as the state machine will go from * RUN -> RUN when this happens. */ if (le64toh(ni->ni_tstamp.tsf) >= tsf) { DPRINTF(sc, ATH_DEBUG_STATE, "ibss merge, rstamp %u tsf %ju " "tstamp %ju\n", rstamp, (uintmax_t)tsf, (uintmax_t)ni->ni_tstamp.tsf); (void) ieee80211_ibss_merge(ni); } } break; } } /* * Set the default antenna. */ static void ath_setdefantenna(struct ath_softc *sc, u_int antenna) { struct ath_hal *ah = sc->sc_ah; /* XXX block beacon interrupts */ ath_hal_setdefantenna(ah, antenna); if (sc->sc_defant != antenna) sc->sc_stats.ast_ant_defswitch++; sc->sc_defant = antenna; sc->sc_rxotherant = 0; } static void ath_rx_tap(struct ifnet *ifp, struct mbuf *m, const struct ath_rx_status *rs, u_int64_t tsf, int16_t nf) { #define CHAN_HT20 htole32(IEEE80211_CHAN_HT20) #define CHAN_HT40U htole32(IEEE80211_CHAN_HT40U) #define CHAN_HT40D htole32(IEEE80211_CHAN_HT40D) #define CHAN_HT (CHAN_HT20|CHAN_HT40U|CHAN_HT40D) struct ath_softc *sc = ifp->if_softc; const HAL_RATE_TABLE *rt; uint8_t rix; rt = sc->sc_currates; KASSERT(rt != NULL, ("no rate table, mode %u", sc->sc_curmode)); rix = rt->rateCodeToIndex[rs->rs_rate]; sc->sc_rx_th.wr_rate = sc->sc_hwmap[rix].ieeerate; sc->sc_rx_th.wr_flags = sc->sc_hwmap[rix].rxflags; #ifdef AH_SUPPORT_AR5416 sc->sc_rx_th.wr_chan_flags &= ~CHAN_HT; if (sc->sc_rx_th.wr_rate & IEEE80211_RATE_MCS) { /* HT rate */ struct ieee80211com *ic = ifp->if_l2com; if ((rs->rs_flags & HAL_RX_2040) == 0) sc->sc_rx_th.wr_chan_flags |= CHAN_HT20; else if (IEEE80211_IS_CHAN_HT40U(ic->ic_curchan)) sc->sc_rx_th.wr_chan_flags |= CHAN_HT40U; else sc->sc_rx_th.wr_chan_flags |= CHAN_HT40D; if ((rs->rs_flags & HAL_RX_GI) == 0) sc->sc_rx_th.wr_flags |= IEEE80211_RADIOTAP_F_SHORTGI; } #endif sc->sc_rx_th.wr_tsf = htole64(ath_extend_tsf(rs->rs_tstamp, tsf)); if (rs->rs_status & HAL_RXERR_CRC) sc->sc_rx_th.wr_flags |= IEEE80211_RADIOTAP_F_BADFCS; /* XXX propagate other error flags from descriptor */ sc->sc_rx_th.wr_antnoise = nf; sc->sc_rx_th.wr_antsignal = nf + rs->rs_rssi; sc->sc_rx_th.wr_antenna = rs->rs_antenna; #undef CHAN_HT #undef CHAN_HT20 #undef CHAN_HT40U #undef CHAN_HT40D } static void ath_handle_micerror(struct ieee80211com *ic, struct ieee80211_frame *wh, int keyix) { struct ieee80211_node *ni; /* XXX recheck MIC to deal w/ chips that lie */ /* XXX discard MIC errors on !data frames */ ni = ieee80211_find_rxnode(ic, (const struct ieee80211_frame_min *) wh); if (ni != NULL) { ieee80211_notify_michael_failure(ni->ni_vap, wh, keyix); ieee80211_free_node(ni); } } static void ath_rx_proc(void *arg, int npending) { #define PA2DESC(_sc, _pa) \ ((struct ath_desc *)((caddr_t)(_sc)->sc_rxdma.dd_desc + \ ((_pa) - (_sc)->sc_rxdma.dd_desc_paddr))) struct ath_softc *sc = arg; struct ath_buf *bf; struct ifnet *ifp = sc->sc_ifp; struct ieee80211com *ic = ifp->if_l2com; struct ath_hal *ah = sc->sc_ah; struct ath_desc *ds; struct ath_rx_status *rs; struct mbuf *m; struct ieee80211_node *ni; int len, type, ngood; HAL_STATUS status; int16_t nf; u_int64_t tsf; DPRINTF(sc, ATH_DEBUG_RX_PROC, "%s: pending %u\n", __func__, npending); ngood = 0; nf = ath_hal_getchannoise(ah, sc->sc_curchan); sc->sc_stats.ast_rx_noise = nf; tsf = ath_hal_gettsf64(ah); do { bf = STAILQ_FIRST(&sc->sc_rxbuf); if (sc->sc_rxslink && bf == NULL) { /* NB: shouldn't happen */ if_printf(ifp, "%s: no buffer!\n", __func__); break; } else if (bf == NULL) { /* * End of List: * this can happen for non-self-linked RX chains */ sc->sc_stats.ast_rx_hitqueueend++; break; } m = bf->bf_m; if (m == NULL) { /* NB: shouldn't happen */ /* * If mbuf allocation failed previously there * will be no mbuf; try again to re-populate it. */ /* XXX make debug msg */ if_printf(ifp, "%s: no mbuf!\n", __func__); STAILQ_REMOVE_HEAD(&sc->sc_rxbuf, bf_list); goto rx_next; } ds = bf->bf_desc; if (ds->ds_link == bf->bf_daddr) { /* NB: never process the self-linked entry at the end */ sc->sc_stats.ast_rx_hitqueueend++; break; } /* XXX sync descriptor memory */ /* * Must provide the virtual address of the current * descriptor, the physical address, and the virtual * address of the next descriptor in the h/w chain. * This allows the HAL to look ahead to see if the * hardware is done with a descriptor by checking the * done bit in the following descriptor and the address * of the current descriptor the DMA engine is working * on. All this is necessary because of our use of * a self-linked list to avoid rx overruns. */ rs = &bf->bf_status.ds_rxstat; status = ath_hal_rxprocdesc(ah, ds, bf->bf_daddr, PA2DESC(sc, ds->ds_link), rs); #ifdef ATH_DEBUG if (sc->sc_debug & ATH_DEBUG_RECV_DESC) ath_printrxbuf(sc, bf, 0, status == HAL_OK); #endif if (status == HAL_EINPROGRESS) break; STAILQ_REMOVE_HEAD(&sc->sc_rxbuf, bf_list); /* These aren't specifically errors */ if (rs->rs_flags & HAL_RX_GI) sc->sc_stats.ast_rx_halfgi++; if (rs->rs_flags & HAL_RX_2040) sc->sc_stats.ast_rx_2040++; if (rs->rs_flags & HAL_RX_DELIM_CRC_PRE) sc->sc_stats.ast_rx_pre_crc_err++; if (rs->rs_flags & HAL_RX_DELIM_CRC_POST) sc->sc_stats.ast_rx_post_crc_err++; if (rs->rs_flags & HAL_RX_DECRYPT_BUSY) sc->sc_stats.ast_rx_decrypt_busy_err++; if (rs->rs_flags & HAL_RX_HI_RX_CHAIN) sc->sc_stats.ast_rx_hi_rx_chain++; if (rs->rs_status != 0) { if (rs->rs_status & HAL_RXERR_CRC) sc->sc_stats.ast_rx_crcerr++; if (rs->rs_status & HAL_RXERR_FIFO) sc->sc_stats.ast_rx_fifoerr++; if (rs->rs_status & HAL_RXERR_PHY) { sc->sc_stats.ast_rx_phyerr++; /* Process DFS radar events */ - ath_dfs_process_phy_err(sc, mtod(m, char *), tsf, rs); + if ((rs->rs_phyerr == HAL_PHYERR_RADAR) || + (rs->rs_phyerr == HAL_PHYERR_FALSE_RADAR_EXT)) { + /* Since we're touching the frame data, sync it */ + bus_dmamap_sync(sc->sc_dmat, + bf->bf_dmamap, + BUS_DMASYNC_POSTREAD); + /* Now pass it to the radar processing code */ + ath_dfs_process_phy_err(sc, mtod(m, char *), tsf, rs); + } /* Be suitably paranoid about receiving phy errors out of the stats array bounds */ if (rs->rs_phyerr < 64) sc->sc_stats.ast_rx_phy[rs->rs_phyerr]++; goto rx_error; /* NB: don't count in ierrors */ } if (rs->rs_status & HAL_RXERR_DECRYPT) { /* * Decrypt error. If the error occurred * because there was no hardware key, then * let the frame through so the upper layers * can process it. This is necessary for 5210 * parts which have no way to setup a ``clear'' * key cache entry. * * XXX do key cache faulting */ if (rs->rs_keyix == HAL_RXKEYIX_INVALID) goto rx_accept; sc->sc_stats.ast_rx_badcrypt++; } if (rs->rs_status & HAL_RXERR_MIC) { sc->sc_stats.ast_rx_badmic++; /* * Do minimal work required to hand off * the 802.11 header for notification. */ /* XXX frag's and qos frames */ len = rs->rs_datalen; if (len >= sizeof (struct ieee80211_frame)) { bus_dmamap_sync(sc->sc_dmat, bf->bf_dmamap, BUS_DMASYNC_POSTREAD); ath_handle_micerror(ic, mtod(m, struct ieee80211_frame *), sc->sc_splitmic ? rs->rs_keyix-32 : rs->rs_keyix); } } ifp->if_ierrors++; rx_error: /* * Cleanup any pending partial frame. */ if (sc->sc_rxpending != NULL) { m_freem(sc->sc_rxpending); sc->sc_rxpending = NULL; } /* * When a tap is present pass error frames * that have been requested. By default we * pass decrypt+mic errors but others may be * interesting (e.g. crc). */ if (ieee80211_radiotap_active(ic) && (rs->rs_status & sc->sc_monpass)) { bus_dmamap_sync(sc->sc_dmat, bf->bf_dmamap, BUS_DMASYNC_POSTREAD); /* NB: bpf needs the mbuf length setup */ len = rs->rs_datalen; m->m_pkthdr.len = m->m_len = len; ath_rx_tap(ifp, m, rs, tsf, nf); ieee80211_radiotap_rx_all(ic, m); } /* XXX pass MIC errors up for s/w reclaculation */ goto rx_next; } rx_accept: /* * Sync and unmap the frame. At this point we're * committed to passing the mbuf somewhere so clear * bf_m; this means a new mbuf must be allocated * when the rx descriptor is setup again to receive * another frame. */ bus_dmamap_sync(sc->sc_dmat, bf->bf_dmamap, BUS_DMASYNC_POSTREAD); bus_dmamap_unload(sc->sc_dmat, bf->bf_dmamap); bf->bf_m = NULL; len = rs->rs_datalen; m->m_len = len; if (rs->rs_more) { /* * Frame spans multiple descriptors; save * it for the next completed descriptor, it * will be used to construct a jumbogram. */ if (sc->sc_rxpending != NULL) { /* NB: max frame size is currently 2 clusters */ sc->sc_stats.ast_rx_toobig++; m_freem(sc->sc_rxpending); } m->m_pkthdr.rcvif = ifp; m->m_pkthdr.len = len; sc->sc_rxpending = m; goto rx_next; } else if (sc->sc_rxpending != NULL) { /* * This is the second part of a jumbogram, * chain it to the first mbuf, adjust the * frame length, and clear the rxpending state. */ sc->sc_rxpending->m_next = m; sc->sc_rxpending->m_pkthdr.len += len; m = sc->sc_rxpending; sc->sc_rxpending = NULL; } else { /* * Normal single-descriptor receive; setup * the rcvif and packet length. */ m->m_pkthdr.rcvif = ifp; m->m_pkthdr.len = len; } ifp->if_ipackets++; sc->sc_stats.ast_ant_rx[rs->rs_antenna]++; /* * Populate the rx status block. When there are bpf * listeners we do the additional work to provide * complete status. Otherwise we fill in only the * material required by ieee80211_input. Note that * noise setting is filled in above. */ if (ieee80211_radiotap_active(ic)) ath_rx_tap(ifp, m, rs, tsf, nf); /* * From this point on we assume the frame is at least * as large as ieee80211_frame_min; verify that. */ if (len < IEEE80211_MIN_LEN) { if (!ieee80211_radiotap_active(ic)) { DPRINTF(sc, ATH_DEBUG_RECV, "%s: short packet %d\n", __func__, len); sc->sc_stats.ast_rx_tooshort++; } else { /* NB: in particular this captures ack's */ ieee80211_radiotap_rx_all(ic, m); } m_freem(m); goto rx_next; } if (IFF_DUMPPKTS(sc, ATH_DEBUG_RECV)) { const HAL_RATE_TABLE *rt = sc->sc_currates; uint8_t rix = rt->rateCodeToIndex[rs->rs_rate]; ieee80211_dump_pkt(ic, mtod(m, caddr_t), len, sc->sc_hwmap[rix].ieeerate, rs->rs_rssi); } m_adj(m, -IEEE80211_CRC_LEN); /* * Locate the node for sender, track state, and then * pass the (referenced) node up to the 802.11 layer * for its use. */ ni = ieee80211_find_rxnode_withkey(ic, mtod(m, const struct ieee80211_frame_min *), rs->rs_keyix == HAL_RXKEYIX_INVALID ? IEEE80211_KEYIX_NONE : rs->rs_keyix); sc->sc_lastrs = rs; if (rs->rs_isaggr) sc->sc_stats.ast_rx_agg++; if (ni != NULL) { /* * Only punt packets for ampdu reorder processing for * 11n nodes; net80211 enforces that M_AMPDU is only * set for 11n nodes. */ if (ni->ni_flags & IEEE80211_NODE_HT) m->m_flags |= M_AMPDU; /* * Sending station is known, dispatch directly. */ type = ieee80211_input(ni, m, rs->rs_rssi, nf); ieee80211_free_node(ni); /* * Arrange to update the last rx timestamp only for * frames from our ap when operating in station mode. * This assumes the rx key is always setup when * associated. */ if (ic->ic_opmode == IEEE80211_M_STA && rs->rs_keyix != HAL_RXKEYIX_INVALID) ngood++; } else { type = ieee80211_input_all(ic, m, rs->rs_rssi, nf); } /* * Track rx rssi and do any rx antenna management. */ ATH_RSSI_LPF(sc->sc_halstats.ns_avgrssi, rs->rs_rssi); if (sc->sc_diversity) { /* * When using fast diversity, change the default rx * antenna if diversity chooses the other antenna 3 * times in a row. */ if (sc->sc_defant != rs->rs_antenna) { if (++sc->sc_rxotherant >= 3) ath_setdefantenna(sc, rs->rs_antenna); } else sc->sc_rxotherant = 0; } /* Newer school diversity - kite specific for now */ /* XXX perhaps migrate the normal diversity code to this? */ if ((ah)->ah_rxAntCombDiversity) (*(ah)->ah_rxAntCombDiversity)(ah, rs, ticks, hz); if (sc->sc_softled) { /* * Blink for any data frame. Otherwise do a * heartbeat-style blink when idle. The latter * is mainly for station mode where we depend on * periodic beacon frames to trigger the poll event. */ if (type == IEEE80211_FC0_TYPE_DATA) { const HAL_RATE_TABLE *rt = sc->sc_currates; ath_led_event(sc, rt->rateCodeToIndex[rs->rs_rate]); } else if (ticks - sc->sc_ledevent >= sc->sc_ledidle) ath_led_event(sc, 0); } rx_next: STAILQ_INSERT_TAIL(&sc->sc_rxbuf, bf, bf_list); } while (ath_rxbuf_init(sc, bf) == 0); /* rx signal state monitoring */ ath_hal_rxmonitor(ah, &sc->sc_halstats, sc->sc_curchan); if (ngood) sc->sc_lastrx = tsf; /* Queue DFS tasklet if needed */ if (ath_dfs_tasklet_needed(sc, sc->sc_curchan)) taskqueue_enqueue(sc->sc_tq, &sc->sc_dfstask); if ((ifp->if_drv_flags & IFF_DRV_OACTIVE) == 0) { #ifdef IEEE80211_SUPPORT_SUPERG ieee80211_ff_age_all(ic, 100); #endif if (!IFQ_IS_EMPTY(&ifp->if_snd)) ath_start(ifp); } #undef PA2DESC } static void ath_txq_init(struct ath_softc *sc, struct ath_txq *txq, int qnum) { txq->axq_qnum = qnum; txq->axq_ac = 0; txq->axq_depth = 0; txq->axq_intrcnt = 0; txq->axq_link = NULL; STAILQ_INIT(&txq->axq_q); ATH_TXQ_LOCK_INIT(sc, txq); } /* * Setup a h/w transmit queue. */ static struct ath_txq * ath_txq_setup(struct ath_softc *sc, int qtype, int subtype) { #define N(a) (sizeof(a)/sizeof(a[0])) struct ath_hal *ah = sc->sc_ah; HAL_TXQ_INFO qi; int qnum; memset(&qi, 0, sizeof(qi)); qi.tqi_subtype = subtype; qi.tqi_aifs = HAL_TXQ_USEDEFAULT; qi.tqi_cwmin = HAL_TXQ_USEDEFAULT; qi.tqi_cwmax = HAL_TXQ_USEDEFAULT; /* * Enable interrupts only for EOL and DESC conditions. * We mark tx descriptors to receive a DESC interrupt * when a tx queue gets deep; otherwise waiting for the * EOL to reap descriptors. Note that this is done to * reduce interrupt load and this only defers reaping * descriptors, never transmitting frames. Aside from * reducing interrupts this also permits more concurrency. * The only potential downside is if the tx queue backs * up in which case the top half of the kernel may backup * due to a lack of tx descriptors. */ qi.tqi_qflags = HAL_TXQ_TXEOLINT_ENABLE | HAL_TXQ_TXDESCINT_ENABLE; qnum = ath_hal_setuptxqueue(ah, qtype, &qi); if (qnum == -1) { /* * NB: don't print a message, this happens * normally on parts with too few tx queues */ return NULL; } if (qnum >= N(sc->sc_txq)) { device_printf(sc->sc_dev, "hal qnum %u out of range, max %zu!\n", qnum, N(sc->sc_txq)); ath_hal_releasetxqueue(ah, qnum); return NULL; } if (!ATH_TXQ_SETUP(sc, qnum)) { ath_txq_init(sc, &sc->sc_txq[qnum], qnum); sc->sc_txqsetup |= 1<sc_txq[qnum]; #undef N } /* * Setup a hardware data transmit queue for the specified * access control. The hal may not support all requested * queues in which case it will return a reference to a * previously setup queue. We record the mapping from ac's * to h/w queues for use by ath_tx_start and also track * the set of h/w queues being used to optimize work in the * transmit interrupt handler and related routines. */ static int ath_tx_setup(struct ath_softc *sc, int ac, int haltype) { #define N(a) (sizeof(a)/sizeof(a[0])) struct ath_txq *txq; if (ac >= N(sc->sc_ac2q)) { device_printf(sc->sc_dev, "AC %u out of range, max %zu!\n", ac, N(sc->sc_ac2q)); return 0; } txq = ath_txq_setup(sc, HAL_TX_QUEUE_DATA, haltype); if (txq != NULL) { txq->axq_ac = ac; sc->sc_ac2q[ac] = txq; return 1; } else return 0; #undef N } /* * Update WME parameters for a transmit queue. */ static int ath_txq_update(struct ath_softc *sc, int ac) { #define ATH_EXPONENT_TO_VALUE(v) ((1<sc_ifp; struct ieee80211com *ic = ifp->if_l2com; struct ath_txq *txq = sc->sc_ac2q[ac]; struct wmeParams *wmep = &ic->ic_wme.wme_chanParams.cap_wmeParams[ac]; struct ath_hal *ah = sc->sc_ah; HAL_TXQ_INFO qi; ath_hal_gettxqueueprops(ah, txq->axq_qnum, &qi); #ifdef IEEE80211_SUPPORT_TDMA if (sc->sc_tdma) { /* * AIFS is zero so there's no pre-transmit wait. The * burst time defines the slot duration and is configured * through net80211. The QCU is setup to not do post-xmit * back off, lockout all lower-priority QCU's, and fire * off the DMA beacon alert timer which is setup based * on the slot configuration. */ qi.tqi_qflags = HAL_TXQ_TXOKINT_ENABLE | HAL_TXQ_TXERRINT_ENABLE | HAL_TXQ_TXURNINT_ENABLE | HAL_TXQ_TXEOLINT_ENABLE | HAL_TXQ_DBA_GATED | HAL_TXQ_BACKOFF_DISABLE | HAL_TXQ_ARB_LOCKOUT_GLOBAL ; qi.tqi_aifs = 0; /* XXX +dbaprep? */ qi.tqi_readyTime = sc->sc_tdmaslotlen; qi.tqi_burstTime = qi.tqi_readyTime; } else { #endif qi.tqi_qflags = HAL_TXQ_TXOKINT_ENABLE | HAL_TXQ_TXERRINT_ENABLE | HAL_TXQ_TXDESCINT_ENABLE | HAL_TXQ_TXURNINT_ENABLE ; qi.tqi_aifs = wmep->wmep_aifsn; qi.tqi_cwmin = ATH_EXPONENT_TO_VALUE(wmep->wmep_logcwmin); qi.tqi_cwmax = ATH_EXPONENT_TO_VALUE(wmep->wmep_logcwmax); qi.tqi_readyTime = 0; qi.tqi_burstTime = ATH_TXOP_TO_US(wmep->wmep_txopLimit); #ifdef IEEE80211_SUPPORT_TDMA } #endif DPRINTF(sc, ATH_DEBUG_RESET, "%s: Q%u qflags 0x%x aifs %u cwmin %u cwmax %u burstTime %u\n", __func__, txq->axq_qnum, qi.tqi_qflags, qi.tqi_aifs, qi.tqi_cwmin, qi.tqi_cwmax, qi.tqi_burstTime); if (!ath_hal_settxqueueprops(ah, txq->axq_qnum, &qi)) { if_printf(ifp, "unable to update hardware queue " "parameters for %s traffic!\n", ieee80211_wme_acnames[ac]); return 0; } else { ath_hal_resettxqueue(ah, txq->axq_qnum); /* push to h/w */ return 1; } #undef ATH_TXOP_TO_US #undef ATH_EXPONENT_TO_VALUE } /* * Callback from the 802.11 layer to update WME parameters. */ static int ath_wme_update(struct ieee80211com *ic) { struct ath_softc *sc = ic->ic_ifp->if_softc; return !ath_txq_update(sc, WME_AC_BE) || !ath_txq_update(sc, WME_AC_BK) || !ath_txq_update(sc, WME_AC_VI) || !ath_txq_update(sc, WME_AC_VO) ? EIO : 0; } /* * Reclaim resources for a setup queue. */ static void ath_tx_cleanupq(struct ath_softc *sc, struct ath_txq *txq) { ath_hal_releasetxqueue(sc->sc_ah, txq->axq_qnum); ATH_TXQ_LOCK_DESTROY(txq); sc->sc_txqsetup &= ~(1<axq_qnum); } /* * Reclaim all tx queue resources. */ static void ath_tx_cleanup(struct ath_softc *sc) { int i; ATH_TXBUF_LOCK_DESTROY(sc); for (i = 0; i < HAL_NUM_TX_QUEUES; i++) if (ATH_TXQ_SETUP(sc, i)) ath_tx_cleanupq(sc, &sc->sc_txq[i]); } /* * Return h/w rate index for an IEEE rate (w/o basic rate bit) * using the current rates in sc_rixmap. */ int ath_tx_findrix(const struct ath_softc *sc, uint8_t rate) { int rix = sc->sc_rixmap[rate]; /* NB: return lowest rix for invalid rate */ return (rix == 0xff ? 0 : rix); } /* * Process completed xmit descriptors from the specified queue. */ static int ath_tx_processq(struct ath_softc *sc, struct ath_txq *txq) { struct ath_hal *ah = sc->sc_ah; struct ifnet *ifp = sc->sc_ifp; struct ieee80211com *ic = ifp->if_l2com; struct ath_buf *bf, *last; struct ath_desc *ds, *ds0; struct ath_tx_status *ts; struct ieee80211_node *ni; struct ath_node *an; int sr, lr, pri, nacked; HAL_STATUS status; DPRINTF(sc, ATH_DEBUG_TX_PROC, "%s: tx queue %u head %p link %p\n", __func__, txq->axq_qnum, (caddr_t)(uintptr_t) ath_hal_gettxbuf(sc->sc_ah, txq->axq_qnum), txq->axq_link); nacked = 0; for (;;) { ATH_TXQ_LOCK(txq); txq->axq_intrcnt = 0; /* reset periodic desc intr count */ bf = STAILQ_FIRST(&txq->axq_q); if (bf == NULL) { ATH_TXQ_UNLOCK(txq); break; } ds0 = &bf->bf_desc[0]; ds = &bf->bf_desc[bf->bf_nseg - 1]; ts = &bf->bf_status.ds_txstat; status = ath_hal_txprocdesc(ah, ds, ts); #ifdef ATH_DEBUG if (sc->sc_debug & ATH_DEBUG_XMIT_DESC) ath_printtxbuf(sc, bf, txq->axq_qnum, 0, status == HAL_OK); #endif if (status == HAL_EINPROGRESS) { ATH_TXQ_UNLOCK(txq); break; } ATH_TXQ_REMOVE_HEAD(txq, bf_list); #ifdef IEEE80211_SUPPORT_TDMA if (txq->axq_depth > 0) { /* * More frames follow. Mark the buffer busy * so it's not re-used while the hardware may * still re-read the link field in the descriptor. */ bf->bf_flags |= ATH_BUF_BUSY; } else #else if (txq->axq_depth == 0) #endif txq->axq_link = NULL; ATH_TXQ_UNLOCK(txq); ni = bf->bf_node; if (ni != NULL) { an = ATH_NODE(ni); if (ts->ts_status == 0) { u_int8_t txant = ts->ts_antenna; sc->sc_stats.ast_ant_tx[txant]++; sc->sc_ant_tx[txant]++; if (ts->ts_finaltsi != 0) sc->sc_stats.ast_tx_altrate++; pri = M_WME_GETAC(bf->bf_m); if (pri >= WME_AC_VO) ic->ic_wme.wme_hipri_traffic++; if ((bf->bf_txflags & HAL_TXDESC_NOACK) == 0) ni->ni_inact = ni->ni_inact_reload; } else { if (ts->ts_status & HAL_TXERR_XRETRY) sc->sc_stats.ast_tx_xretries++; if (ts->ts_status & HAL_TXERR_FIFO) sc->sc_stats.ast_tx_fifoerr++; if (ts->ts_status & HAL_TXERR_FILT) sc->sc_stats.ast_tx_filtered++; if (ts->ts_status & HAL_TXERR_XTXOP) sc->sc_stats.ast_tx_xtxop++; if (ts->ts_status & HAL_TXERR_TIMER_EXPIRED) sc->sc_stats.ast_tx_timerexpired++; /* XXX HAL_TX_DATA_UNDERRUN */ /* XXX HAL_TX_DELIM_UNDERRUN */ if (bf->bf_m->m_flags & M_FF) sc->sc_stats.ast_ff_txerr++; } /* XXX when is this valid? */ if (ts->ts_status & HAL_TX_DESC_CFG_ERR) sc->sc_stats.ast_tx_desccfgerr++; sr = ts->ts_shortretry; lr = ts->ts_longretry; sc->sc_stats.ast_tx_shortretry += sr; sc->sc_stats.ast_tx_longretry += lr; /* * Hand the descriptor to the rate control algorithm. */ if ((ts->ts_status & HAL_TXERR_FILT) == 0 && (bf->bf_txflags & HAL_TXDESC_NOACK) == 0) { /* * If frame was ack'd update statistics, * including the last rx time used to * workaround phantom bmiss interrupts. */ if (ts->ts_status == 0) { nacked++; sc->sc_stats.ast_tx_rssi = ts->ts_rssi; ATH_RSSI_LPF(sc->sc_halstats.ns_avgtxrssi, ts->ts_rssi); } ath_rate_tx_complete(sc, an, bf); } /* * Do any tx complete callback. Note this must * be done before releasing the node reference. */ if (bf->bf_m->m_flags & M_TXCB) ieee80211_process_callback(ni, bf->bf_m, (bf->bf_txflags & HAL_TXDESC_NOACK) == 0 ? ts->ts_status : HAL_TXERR_XRETRY); ieee80211_free_node(ni); } bus_dmamap_sync(sc->sc_dmat, bf->bf_dmamap, BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(sc->sc_dmat, bf->bf_dmamap); m_freem(bf->bf_m); bf->bf_m = NULL; bf->bf_node = NULL; ATH_TXBUF_LOCK(sc); last = STAILQ_LAST(&sc->sc_txbuf, ath_buf, bf_list); if (last != NULL) last->bf_flags &= ~ATH_BUF_BUSY; STAILQ_INSERT_TAIL(&sc->sc_txbuf, bf, bf_list); ATH_TXBUF_UNLOCK(sc); } #ifdef IEEE80211_SUPPORT_SUPERG /* * Flush fast-frame staging queue when traffic slows. */ if (txq->axq_depth <= 1) ieee80211_ff_flush(ic, txq->axq_ac); #endif return nacked; } static __inline int txqactive(struct ath_hal *ah, int qnum) { u_int32_t txqs = 1<sc_ifp; if (txqactive(sc->sc_ah, 0) && ath_tx_processq(sc, &sc->sc_txq[0])) sc->sc_lastrx = ath_hal_gettsf64(sc->sc_ah); if (txqactive(sc->sc_ah, sc->sc_cabq->axq_qnum)) ath_tx_processq(sc, sc->sc_cabq); ifp->if_drv_flags &= ~IFF_DRV_OACTIVE; sc->sc_wd_timer = 0; if (sc->sc_softled) ath_led_event(sc, sc->sc_txrix); ath_start(ifp); } /* * Deferred processing of transmit interrupt; special-cased * for four hardware queues, 0-3 (e.g. 5212 w/ WME support). */ static void ath_tx_proc_q0123(void *arg, int npending) { struct ath_softc *sc = arg; struct ifnet *ifp = sc->sc_ifp; int nacked; /* * Process each active queue. */ nacked = 0; if (txqactive(sc->sc_ah, 0)) nacked += ath_tx_processq(sc, &sc->sc_txq[0]); if (txqactive(sc->sc_ah, 1)) nacked += ath_tx_processq(sc, &sc->sc_txq[1]); if (txqactive(sc->sc_ah, 2)) nacked += ath_tx_processq(sc, &sc->sc_txq[2]); if (txqactive(sc->sc_ah, 3)) nacked += ath_tx_processq(sc, &sc->sc_txq[3]); if (txqactive(sc->sc_ah, sc->sc_cabq->axq_qnum)) ath_tx_processq(sc, sc->sc_cabq); if (nacked) sc->sc_lastrx = ath_hal_gettsf64(sc->sc_ah); ifp->if_drv_flags &= ~IFF_DRV_OACTIVE; sc->sc_wd_timer = 0; if (sc->sc_softled) ath_led_event(sc, sc->sc_txrix); ath_start(ifp); } /* * Deferred processing of transmit interrupt. */ static void ath_tx_proc(void *arg, int npending) { struct ath_softc *sc = arg; struct ifnet *ifp = sc->sc_ifp; int i, nacked; /* * Process each active queue. */ nacked = 0; for (i = 0; i < HAL_NUM_TX_QUEUES; i++) if (ATH_TXQ_SETUP(sc, i) && txqactive(sc->sc_ah, i)) nacked += ath_tx_processq(sc, &sc->sc_txq[i]); if (nacked) sc->sc_lastrx = ath_hal_gettsf64(sc->sc_ah); ifp->if_drv_flags &= ~IFF_DRV_OACTIVE; sc->sc_wd_timer = 0; if (sc->sc_softled) ath_led_event(sc, sc->sc_txrix); ath_start(ifp); } static void ath_tx_draintxq(struct ath_softc *sc, struct ath_txq *txq) { #ifdef ATH_DEBUG struct ath_hal *ah = sc->sc_ah; #endif struct ieee80211_node *ni; struct ath_buf *bf; u_int ix; /* * NB: this assumes output has been stopped and * we do not need to block ath_tx_proc */ ATH_TXBUF_LOCK(sc); bf = STAILQ_LAST(&sc->sc_txbuf, ath_buf, bf_list); if (bf != NULL) bf->bf_flags &= ~ATH_BUF_BUSY; ATH_TXBUF_UNLOCK(sc); for (ix = 0;; ix++) { ATH_TXQ_LOCK(txq); bf = STAILQ_FIRST(&txq->axq_q); if (bf == NULL) { txq->axq_link = NULL; ATH_TXQ_UNLOCK(txq); break; } ATH_TXQ_REMOVE_HEAD(txq, bf_list); ATH_TXQ_UNLOCK(txq); #ifdef ATH_DEBUG if (sc->sc_debug & ATH_DEBUG_RESET) { struct ieee80211com *ic = sc->sc_ifp->if_l2com; ath_printtxbuf(sc, bf, txq->axq_qnum, ix, ath_hal_txprocdesc(ah, bf->bf_desc, &bf->bf_status.ds_txstat) == HAL_OK); ieee80211_dump_pkt(ic, mtod(bf->bf_m, const uint8_t *), bf->bf_m->m_len, 0, -1); } #endif /* ATH_DEBUG */ bus_dmamap_unload(sc->sc_dmat, bf->bf_dmamap); ni = bf->bf_node; bf->bf_node = NULL; if (ni != NULL) { /* * Do any callback and reclaim the node reference. */ if (bf->bf_m->m_flags & M_TXCB) ieee80211_process_callback(ni, bf->bf_m, -1); ieee80211_free_node(ni); } m_freem(bf->bf_m); bf->bf_m = NULL; bf->bf_flags &= ~ATH_BUF_BUSY; ATH_TXBUF_LOCK(sc); STAILQ_INSERT_TAIL(&sc->sc_txbuf, bf, bf_list); ATH_TXBUF_UNLOCK(sc); } } static void ath_tx_stopdma(struct ath_softc *sc, struct ath_txq *txq) { struct ath_hal *ah = sc->sc_ah; DPRINTF(sc, ATH_DEBUG_RESET, "%s: tx queue [%u] %p, link %p\n", __func__, txq->axq_qnum, (caddr_t)(uintptr_t) ath_hal_gettxbuf(ah, txq->axq_qnum), txq->axq_link); (void) ath_hal_stoptxdma(ah, txq->axq_qnum); } /* * Drain the transmit queues and reclaim resources. */ static void ath_draintxq(struct ath_softc *sc) { struct ath_hal *ah = sc->sc_ah; struct ifnet *ifp = sc->sc_ifp; int i; /* XXX return value */ if (!sc->sc_invalid) { /* don't touch the hardware if marked invalid */ DPRINTF(sc, ATH_DEBUG_RESET, "%s: tx queue [%u] %p, link %p\n", __func__, sc->sc_bhalq, (caddr_t)(uintptr_t) ath_hal_gettxbuf(ah, sc->sc_bhalq), NULL); (void) ath_hal_stoptxdma(ah, sc->sc_bhalq); for (i = 0; i < HAL_NUM_TX_QUEUES; i++) if (ATH_TXQ_SETUP(sc, i)) ath_tx_stopdma(sc, &sc->sc_txq[i]); } for (i = 0; i < HAL_NUM_TX_QUEUES; i++) if (ATH_TXQ_SETUP(sc, i)) ath_tx_draintxq(sc, &sc->sc_txq[i]); #ifdef ATH_DEBUG if (sc->sc_debug & ATH_DEBUG_RESET) { struct ath_buf *bf = STAILQ_FIRST(&sc->sc_bbuf); if (bf != NULL && bf->bf_m != NULL) { ath_printtxbuf(sc, bf, sc->sc_bhalq, 0, ath_hal_txprocdesc(ah, bf->bf_desc, &bf->bf_status.ds_txstat) == HAL_OK); ieee80211_dump_pkt(ifp->if_l2com, mtod(bf->bf_m, const uint8_t *), bf->bf_m->m_len, 0, -1); } } #endif /* ATH_DEBUG */ ifp->if_drv_flags &= ~IFF_DRV_OACTIVE; sc->sc_wd_timer = 0; } /* * Disable the receive h/w in preparation for a reset. */ static void ath_stoprecv(struct ath_softc *sc) { #define PA2DESC(_sc, _pa) \ ((struct ath_desc *)((caddr_t)(_sc)->sc_rxdma.dd_desc + \ ((_pa) - (_sc)->sc_rxdma.dd_desc_paddr))) struct ath_hal *ah = sc->sc_ah; ath_hal_stoppcurecv(ah); /* disable PCU */ ath_hal_setrxfilter(ah, 0); /* clear recv filter */ ath_hal_stopdmarecv(ah); /* disable DMA engine */ DELAY(3000); /* 3ms is long enough for 1 frame */ #ifdef ATH_DEBUG if (sc->sc_debug & (ATH_DEBUG_RESET | ATH_DEBUG_FATAL)) { struct ath_buf *bf; u_int ix; printf("%s: rx queue %p, link %p\n", __func__, (caddr_t)(uintptr_t) ath_hal_getrxbuf(ah), sc->sc_rxlink); ix = 0; STAILQ_FOREACH(bf, &sc->sc_rxbuf, bf_list) { struct ath_desc *ds = bf->bf_desc; struct ath_rx_status *rs = &bf->bf_status.ds_rxstat; HAL_STATUS status = ath_hal_rxprocdesc(ah, ds, bf->bf_daddr, PA2DESC(sc, ds->ds_link), rs); if (status == HAL_OK || (sc->sc_debug & ATH_DEBUG_FATAL)) ath_printrxbuf(sc, bf, ix, status == HAL_OK); ix++; } } #endif if (sc->sc_rxpending != NULL) { m_freem(sc->sc_rxpending); sc->sc_rxpending = NULL; } sc->sc_rxlink = NULL; /* just in case */ #undef PA2DESC } /* * Enable the receive h/w following a reset. */ static int ath_startrecv(struct ath_softc *sc) { struct ath_hal *ah = sc->sc_ah; struct ath_buf *bf; sc->sc_rxlink = NULL; sc->sc_rxpending = NULL; STAILQ_FOREACH(bf, &sc->sc_rxbuf, bf_list) { int error = ath_rxbuf_init(sc, bf); if (error != 0) { DPRINTF(sc, ATH_DEBUG_RECV, "%s: ath_rxbuf_init failed %d\n", __func__, error); return error; } } bf = STAILQ_FIRST(&sc->sc_rxbuf); ath_hal_putrxbuf(ah, bf->bf_daddr); ath_hal_rxena(ah); /* enable recv descriptors */ ath_mode_init(sc); /* set filters, etc. */ ath_hal_startpcurecv(ah); /* re-enable PCU/DMA engine */ return 0; } /* * Update internal state after a channel change. */ static void ath_chan_change(struct ath_softc *sc, struct ieee80211_channel *chan) { enum ieee80211_phymode mode; /* * Change channels and update the h/w rate map * if we're switching; e.g. 11a to 11b/g. */ mode = ieee80211_chan2mode(chan); if (mode != sc->sc_curmode) ath_setcurmode(sc, mode); sc->sc_curchan = chan; } /* * Set/change channels. If the channel is really being changed, * it's done by resetting the chip. To accomplish this we must * first cleanup any pending DMA, then restart stuff after a la * ath_init. */ static int ath_chan_set(struct ath_softc *sc, struct ieee80211_channel *chan) { struct ifnet *ifp = sc->sc_ifp; struct ieee80211com *ic = ifp->if_l2com; struct ath_hal *ah = sc->sc_ah; DPRINTF(sc, ATH_DEBUG_RESET, "%s: %u (%u MHz, flags 0x%x)\n", __func__, ieee80211_chan2ieee(ic, chan), chan->ic_freq, chan->ic_flags); if (chan != sc->sc_curchan) { HAL_STATUS status; /* * To switch channels clear any pending DMA operations; * wait long enough for the RX fifo to drain, reset the * hardware at the new frequency, and then re-enable * the relevant bits of the h/w. */ ath_hal_intrset(ah, 0); /* disable interrupts */ ath_draintxq(sc); /* clear pending tx frames */ ath_stoprecv(sc); /* turn off frame recv */ if (!ath_hal_reset(ah, sc->sc_opmode, chan, AH_TRUE, &status)) { if_printf(ifp, "%s: unable to reset " "channel %u (%u MHz, flags 0x%x), hal status %u\n", __func__, ieee80211_chan2ieee(ic, chan), chan->ic_freq, chan->ic_flags, status); return EIO; } sc->sc_diversity = ath_hal_getdiversity(ah); /* Let DFS at it in case it's a DFS channel */ ath_dfs_radar_enable(sc, ic->ic_curchan); /* * Re-enable rx framework. */ if (ath_startrecv(sc) != 0) { if_printf(ifp, "%s: unable to restart recv logic\n", __func__); return EIO; } /* * Change channels and update the h/w rate map * if we're switching; e.g. 11a to 11b/g. */ ath_chan_change(sc, chan); /* * Re-enable interrupts. */ ath_hal_intrset(ah, sc->sc_imask); } return 0; } /* * Periodically recalibrate the PHY to account * for temperature/environment changes. */ static void ath_calibrate(void *arg) { struct ath_softc *sc = arg; struct ath_hal *ah = sc->sc_ah; struct ifnet *ifp = sc->sc_ifp; struct ieee80211com *ic = ifp->if_l2com; HAL_BOOL longCal, isCalDone; HAL_BOOL aniCal, shortCal = AH_FALSE; int nextcal; if (ic->ic_flags & IEEE80211_F_SCAN) /* defer, off channel */ goto restart; longCal = (ticks - sc->sc_lastlongcal >= ath_longcalinterval*hz); aniCal = (ticks - sc->sc_lastani >= ath_anicalinterval*hz/1000); if (sc->sc_doresetcal) shortCal = (ticks - sc->sc_lastshortcal >= ath_shortcalinterval*hz/1000); DPRINTF(sc, ATH_DEBUG_CALIBRATE, "%s: shortCal=%d; longCal=%d; aniCal=%d\n", __func__, shortCal, longCal, aniCal); if (aniCal) { sc->sc_stats.ast_ani_cal++; sc->sc_lastani = ticks; ath_hal_ani_poll(ah, sc->sc_curchan); } if (longCal) { sc->sc_stats.ast_per_cal++; sc->sc_lastlongcal = ticks; if (ath_hal_getrfgain(ah) == HAL_RFGAIN_NEED_CHANGE) { /* * Rfgain is out of bounds, reset the chip * to load new gain values. */ DPRINTF(sc, ATH_DEBUG_CALIBRATE, "%s: rfgain change\n", __func__); sc->sc_stats.ast_per_rfgain++; ath_reset(ifp); } /* * If this long cal is after an idle period, then * reset the data collection state so we start fresh. */ if (sc->sc_resetcal) { (void) ath_hal_calreset(ah, sc->sc_curchan); sc->sc_lastcalreset = ticks; sc->sc_lastshortcal = ticks; sc->sc_resetcal = 0; sc->sc_doresetcal = AH_TRUE; } } /* Only call if we're doing a short/long cal, not for ANI calibration */ if (shortCal || longCal) { if (ath_hal_calibrateN(ah, sc->sc_curchan, longCal, &isCalDone)) { if (longCal) { /* * Calibrate noise floor data again in case of change. */ ath_hal_process_noisefloor(ah); } } else { DPRINTF(sc, ATH_DEBUG_ANY, "%s: calibration of channel %u failed\n", __func__, sc->sc_curchan->ic_freq); sc->sc_stats.ast_per_calfail++; } if (shortCal) sc->sc_lastshortcal = ticks; } if (!isCalDone) { restart: /* * Use a shorter interval to potentially collect multiple * data samples required to complete calibration. Once * we're told the work is done we drop back to a longer * interval between requests. We're more aggressive doing * work when operating as an AP to improve operation right * after startup. */ sc->sc_lastshortcal = ticks; nextcal = ath_shortcalinterval*hz/1000; if (sc->sc_opmode != HAL_M_HOSTAP) nextcal *= 10; sc->sc_doresetcal = AH_TRUE; } else { /* nextcal should be the shortest time for next event */ nextcal = ath_longcalinterval*hz; if (sc->sc_lastcalreset == 0) sc->sc_lastcalreset = sc->sc_lastlongcal; else if (ticks - sc->sc_lastcalreset >= ath_resetcalinterval*hz) sc->sc_resetcal = 1; /* setup reset next trip */ sc->sc_doresetcal = AH_FALSE; } /* ANI calibration may occur more often than short/long/resetcal */ if (ath_anicalinterval > 0) nextcal = MIN(nextcal, ath_anicalinterval*hz/1000); if (nextcal != 0) { DPRINTF(sc, ATH_DEBUG_CALIBRATE, "%s: next +%u (%sisCalDone)\n", __func__, nextcal, isCalDone ? "" : "!"); callout_reset(&sc->sc_cal_ch, nextcal, ath_calibrate, sc); } else { DPRINTF(sc, ATH_DEBUG_CALIBRATE, "%s: calibration disabled\n", __func__); /* NB: don't rearm timer */ } } static void ath_scan_start(struct ieee80211com *ic) { struct ifnet *ifp = ic->ic_ifp; struct ath_softc *sc = ifp->if_softc; struct ath_hal *ah = sc->sc_ah; u_int32_t rfilt; /* XXX calibration timer? */ sc->sc_scanning = 1; sc->sc_syncbeacon = 0; rfilt = ath_calcrxfilter(sc); ath_hal_setrxfilter(ah, rfilt); ath_hal_setassocid(ah, ifp->if_broadcastaddr, 0); DPRINTF(sc, ATH_DEBUG_STATE, "%s: RX filter 0x%x bssid %s aid 0\n", __func__, rfilt, ether_sprintf(ifp->if_broadcastaddr)); } static void ath_scan_end(struct ieee80211com *ic) { struct ifnet *ifp = ic->ic_ifp; struct ath_softc *sc = ifp->if_softc; struct ath_hal *ah = sc->sc_ah; u_int32_t rfilt; sc->sc_scanning = 0; rfilt = ath_calcrxfilter(sc); ath_hal_setrxfilter(ah, rfilt); ath_hal_setassocid(ah, sc->sc_curbssid, sc->sc_curaid); ath_hal_process_noisefloor(ah); DPRINTF(sc, ATH_DEBUG_STATE, "%s: RX filter 0x%x bssid %s aid 0x%x\n", __func__, rfilt, ether_sprintf(sc->sc_curbssid), sc->sc_curaid); } static void ath_set_channel(struct ieee80211com *ic) { struct ifnet *ifp = ic->ic_ifp; struct ath_softc *sc = ifp->if_softc; (void) ath_chan_set(sc, ic->ic_curchan); /* * If we are returning to our bss channel then mark state * so the next recv'd beacon's tsf will be used to sync the * beacon timers. Note that since we only hear beacons in * sta/ibss mode this has no effect in other operating modes. */ if (!sc->sc_scanning && ic->ic_curchan == ic->ic_bsschan) sc->sc_syncbeacon = 1; } /* * Walk the vap list and check if there any vap's in RUN state. */ static int ath_isanyrunningvaps(struct ieee80211vap *this) { struct ieee80211com *ic = this->iv_ic; struct ieee80211vap *vap; IEEE80211_LOCK_ASSERT(ic); TAILQ_FOREACH(vap, &ic->ic_vaps, iv_next) { if (vap != this && vap->iv_state >= IEEE80211_S_RUN) return 1; } return 0; } static int ath_newstate(struct ieee80211vap *vap, enum ieee80211_state nstate, int arg) { struct ieee80211com *ic = vap->iv_ic; struct ath_softc *sc = ic->ic_ifp->if_softc; struct ath_vap *avp = ATH_VAP(vap); struct ath_hal *ah = sc->sc_ah; struct ieee80211_node *ni = NULL; int i, error, stamode; u_int32_t rfilt; static const HAL_LED_STATE leds[] = { HAL_LED_INIT, /* IEEE80211_S_INIT */ HAL_LED_SCAN, /* IEEE80211_S_SCAN */ HAL_LED_AUTH, /* IEEE80211_S_AUTH */ HAL_LED_ASSOC, /* IEEE80211_S_ASSOC */ HAL_LED_RUN, /* IEEE80211_S_CAC */ HAL_LED_RUN, /* IEEE80211_S_RUN */ HAL_LED_RUN, /* IEEE80211_S_CSA */ HAL_LED_RUN, /* IEEE80211_S_SLEEP */ }; DPRINTF(sc, ATH_DEBUG_STATE, "%s: %s -> %s\n", __func__, ieee80211_state_name[vap->iv_state], ieee80211_state_name[nstate]); callout_drain(&sc->sc_cal_ch); ath_hal_setledstate(ah, leds[nstate]); /* set LED */ if (nstate == IEEE80211_S_SCAN) { /* * Scanning: turn off beacon miss and don't beacon. * Mark beacon state so when we reach RUN state we'll * [re]setup beacons. Unblock the task q thread so * deferred interrupt processing is done. */ ath_hal_intrset(ah, sc->sc_imask &~ (HAL_INT_SWBA | HAL_INT_BMISS)); sc->sc_imask &= ~(HAL_INT_SWBA | HAL_INT_BMISS); sc->sc_beacons = 0; taskqueue_unblock(sc->sc_tq); } ni = vap->iv_bss; rfilt = ath_calcrxfilter(sc); stamode = (vap->iv_opmode == IEEE80211_M_STA || vap->iv_opmode == IEEE80211_M_AHDEMO || vap->iv_opmode == IEEE80211_M_IBSS); if (stamode && nstate == IEEE80211_S_RUN) { sc->sc_curaid = ni->ni_associd; IEEE80211_ADDR_COPY(sc->sc_curbssid, ni->ni_bssid); ath_hal_setassocid(ah, sc->sc_curbssid, sc->sc_curaid); } DPRINTF(sc, ATH_DEBUG_STATE, "%s: RX filter 0x%x bssid %s aid 0x%x\n", __func__, rfilt, ether_sprintf(sc->sc_curbssid), sc->sc_curaid); ath_hal_setrxfilter(ah, rfilt); /* XXX is this to restore keycache on resume? */ if (vap->iv_opmode != IEEE80211_M_STA && (vap->iv_flags & IEEE80211_F_PRIVACY)) { for (i = 0; i < IEEE80211_WEP_NKID; i++) if (ath_hal_keyisvalid(ah, i)) ath_hal_keysetmac(ah, i, ni->ni_bssid); } /* * Invoke the parent method to do net80211 work. */ error = avp->av_newstate(vap, nstate, arg); if (error != 0) goto bad; if (nstate == IEEE80211_S_RUN) { /* NB: collect bss node again, it may have changed */ ni = vap->iv_bss; DPRINTF(sc, ATH_DEBUG_STATE, "%s(RUN): iv_flags 0x%08x bintvl %d bssid %s " "capinfo 0x%04x chan %d\n", __func__, vap->iv_flags, ni->ni_intval, ether_sprintf(ni->ni_bssid), ni->ni_capinfo, ieee80211_chan2ieee(ic, ic->ic_curchan)); switch (vap->iv_opmode) { #ifdef IEEE80211_SUPPORT_TDMA case IEEE80211_M_AHDEMO: if ((vap->iv_caps & IEEE80211_C_TDMA) == 0) break; /* fall thru... */ #endif case IEEE80211_M_HOSTAP: case IEEE80211_M_IBSS: case IEEE80211_M_MBSS: /* * Allocate and setup the beacon frame. * * Stop any previous beacon DMA. This may be * necessary, for example, when an ibss merge * causes reconfiguration; there will be a state * transition from RUN->RUN that means we may * be called with beacon transmission active. */ ath_hal_stoptxdma(ah, sc->sc_bhalq); error = ath_beacon_alloc(sc, ni); if (error != 0) goto bad; /* * If joining an adhoc network defer beacon timer * configuration to the next beacon frame so we * have a current TSF to use. Otherwise we're * starting an ibss/bss so there's no need to delay; * if this is the first vap moving to RUN state, then * beacon state needs to be [re]configured. */ if (vap->iv_opmode == IEEE80211_M_IBSS && ni->ni_tstamp.tsf != 0) { sc->sc_syncbeacon = 1; } else if (!sc->sc_beacons) { #ifdef IEEE80211_SUPPORT_TDMA if (vap->iv_caps & IEEE80211_C_TDMA) ath_tdma_config(sc, vap); else #endif ath_beacon_config(sc, vap); sc->sc_beacons = 1; } break; case IEEE80211_M_STA: /* * Defer beacon timer configuration to the next * beacon frame so we have a current TSF to use * (any TSF collected when scanning is likely old). */ sc->sc_syncbeacon = 1; break; case IEEE80211_M_MONITOR: /* * Monitor mode vaps have only INIT->RUN and RUN->RUN * transitions so we must re-enable interrupts here to * handle the case of a single monitor mode vap. */ ath_hal_intrset(ah, sc->sc_imask); break; case IEEE80211_M_WDS: break; default: break; } /* * Let the hal process statistics collected during a * scan so it can provide calibrated noise floor data. */ ath_hal_process_noisefloor(ah); /* * Reset rssi stats; maybe not the best place... */ sc->sc_halstats.ns_avgbrssi = ATH_RSSI_DUMMY_MARKER; sc->sc_halstats.ns_avgrssi = ATH_RSSI_DUMMY_MARKER; sc->sc_halstats.ns_avgtxrssi = ATH_RSSI_DUMMY_MARKER; /* * Finally, start any timers and the task q thread * (in case we didn't go through SCAN state). */ if (ath_longcalinterval != 0) { /* start periodic recalibration timer */ callout_reset(&sc->sc_cal_ch, 1, ath_calibrate, sc); } else { DPRINTF(sc, ATH_DEBUG_CALIBRATE, "%s: calibration disabled\n", __func__); } taskqueue_unblock(sc->sc_tq); } else if (nstate == IEEE80211_S_INIT) { /* * If there are no vaps left in RUN state then * shutdown host/driver operation: * o disable interrupts * o disable the task queue thread * o mark beacon processing as stopped */ if (!ath_isanyrunningvaps(vap)) { sc->sc_imask &= ~(HAL_INT_SWBA | HAL_INT_BMISS); /* disable interrupts */ ath_hal_intrset(ah, sc->sc_imask &~ HAL_INT_GLOBAL); taskqueue_block(sc->sc_tq); sc->sc_beacons = 0; } #ifdef IEEE80211_SUPPORT_TDMA ath_hal_setcca(ah, AH_TRUE); #endif } bad: return error; } /* * Allocate a key cache slot to the station so we can * setup a mapping from key index to node. The key cache * slot is needed for managing antenna state and for * compression when stations do not use crypto. We do * it uniliaterally here; if crypto is employed this slot * will be reassigned. */ static void ath_setup_stationkey(struct ieee80211_node *ni) { struct ieee80211vap *vap = ni->ni_vap; struct ath_softc *sc = vap->iv_ic->ic_ifp->if_softc; ieee80211_keyix keyix, rxkeyix; if (!ath_key_alloc(vap, &ni->ni_ucastkey, &keyix, &rxkeyix)) { /* * Key cache is full; we'll fall back to doing * the more expensive lookup in software. Note * this also means no h/w compression. */ /* XXX msg+statistic */ } else { /* XXX locking? */ ni->ni_ucastkey.wk_keyix = keyix; ni->ni_ucastkey.wk_rxkeyix = rxkeyix; /* NB: must mark device key to get called back on delete */ ni->ni_ucastkey.wk_flags |= IEEE80211_KEY_DEVKEY; IEEE80211_ADDR_COPY(ni->ni_ucastkey.wk_macaddr, ni->ni_macaddr); /* NB: this will create a pass-thru key entry */ ath_keyset(sc, &ni->ni_ucastkey, vap->iv_bss); } } /* * Setup driver-specific state for a newly associated node. * Note that we're called also on a re-associate, the isnew * param tells us if this is the first time or not. */ static void ath_newassoc(struct ieee80211_node *ni, int isnew) { struct ath_node *an = ATH_NODE(ni); struct ieee80211vap *vap = ni->ni_vap; struct ath_softc *sc = vap->iv_ic->ic_ifp->if_softc; const struct ieee80211_txparam *tp = ni->ni_txparms; an->an_mcastrix = ath_tx_findrix(sc, tp->mcastrate); an->an_mgmtrix = ath_tx_findrix(sc, tp->mgmtrate); ath_rate_newassoc(sc, an, isnew); if (isnew && (vap->iv_flags & IEEE80211_F_PRIVACY) == 0 && sc->sc_hasclrkey && ni->ni_ucastkey.wk_keyix == IEEE80211_KEYIX_NONE) ath_setup_stationkey(ni); } static int ath_setregdomain(struct ieee80211com *ic, struct ieee80211_regdomain *reg, int nchans, struct ieee80211_channel chans[]) { struct ath_softc *sc = ic->ic_ifp->if_softc; struct ath_hal *ah = sc->sc_ah; HAL_STATUS status; DPRINTF(sc, ATH_DEBUG_REGDOMAIN, "%s: rd %u cc %u location %c%s\n", __func__, reg->regdomain, reg->country, reg->location, reg->ecm ? " ecm" : ""); status = ath_hal_set_channels(ah, chans, nchans, reg->country, reg->regdomain); if (status != HAL_OK) { DPRINTF(sc, ATH_DEBUG_REGDOMAIN, "%s: failed, status %u\n", __func__, status); return EINVAL; /* XXX */ } return 0; } static void ath_getradiocaps(struct ieee80211com *ic, int maxchans, int *nchans, struct ieee80211_channel chans[]) { struct ath_softc *sc = ic->ic_ifp->if_softc; struct ath_hal *ah = sc->sc_ah; DPRINTF(sc, ATH_DEBUG_REGDOMAIN, "%s: use rd %u cc %d\n", __func__, SKU_DEBUG, CTRY_DEFAULT); /* XXX check return */ (void) ath_hal_getchannels(ah, chans, maxchans, nchans, HAL_MODE_ALL, CTRY_DEFAULT, SKU_DEBUG, AH_TRUE); } static int ath_getchannels(struct ath_softc *sc) { struct ifnet *ifp = sc->sc_ifp; struct ieee80211com *ic = ifp->if_l2com; struct ath_hal *ah = sc->sc_ah; HAL_STATUS status; /* * Collect channel set based on EEPROM contents. */ status = ath_hal_init_channels(ah, ic->ic_channels, IEEE80211_CHAN_MAX, &ic->ic_nchans, HAL_MODE_ALL, CTRY_DEFAULT, SKU_NONE, AH_TRUE); if (status != HAL_OK) { if_printf(ifp, "%s: unable to collect channel list from hal, " "status %d\n", __func__, status); return EINVAL; } (void) ath_hal_getregdomain(ah, &sc->sc_eerd); ath_hal_getcountrycode(ah, &sc->sc_eecc); /* NB: cannot fail */ /* XXX map Atheros sku's to net80211 SKU's */ /* XXX net80211 types too small */ ic->ic_regdomain.regdomain = (uint16_t) sc->sc_eerd; ic->ic_regdomain.country = (uint16_t) sc->sc_eecc; ic->ic_regdomain.isocc[0] = ' '; /* XXX don't know */ ic->ic_regdomain.isocc[1] = ' '; ic->ic_regdomain.ecm = 1; ic->ic_regdomain.location = 'I'; DPRINTF(sc, ATH_DEBUG_REGDOMAIN, "%s: eeprom rd %u cc %u (mapped rd %u cc %u) location %c%s\n", __func__, sc->sc_eerd, sc->sc_eecc, ic->ic_regdomain.regdomain, ic->ic_regdomain.country, ic->ic_regdomain.location, ic->ic_regdomain.ecm ? " ecm" : ""); return 0; } static void ath_led_done(void *arg) { struct ath_softc *sc = arg; sc->sc_blinking = 0; } /* * Turn the LED off: flip the pin and then set a timer so no * update will happen for the specified duration. */ static void ath_led_off(void *arg) { struct ath_softc *sc = arg; ath_hal_gpioset(sc->sc_ah, sc->sc_ledpin, !sc->sc_ledon); callout_reset(&sc->sc_ledtimer, sc->sc_ledoff, ath_led_done, sc); } /* * Blink the LED according to the specified on/off times. */ static void ath_led_blink(struct ath_softc *sc, int on, int off) { DPRINTF(sc, ATH_DEBUG_LED, "%s: on %u off %u\n", __func__, on, off); ath_hal_gpioset(sc->sc_ah, sc->sc_ledpin, sc->sc_ledon); sc->sc_blinking = 1; sc->sc_ledoff = off; callout_reset(&sc->sc_ledtimer, on, ath_led_off, sc); } static void ath_led_event(struct ath_softc *sc, int rix) { sc->sc_ledevent = ticks; /* time of last event */ if (sc->sc_blinking) /* don't interrupt active blink */ return; ath_led_blink(sc, sc->sc_hwmap[rix].ledon, sc->sc_hwmap[rix].ledoff); } static int ath_rate_setup(struct ath_softc *sc, u_int mode) { struct ath_hal *ah = sc->sc_ah; const HAL_RATE_TABLE *rt; switch (mode) { case IEEE80211_MODE_11A: rt = ath_hal_getratetable(ah, HAL_MODE_11A); break; case IEEE80211_MODE_HALF: rt = ath_hal_getratetable(ah, HAL_MODE_11A_HALF_RATE); break; case IEEE80211_MODE_QUARTER: rt = ath_hal_getratetable(ah, HAL_MODE_11A_QUARTER_RATE); break; case IEEE80211_MODE_11B: rt = ath_hal_getratetable(ah, HAL_MODE_11B); break; case IEEE80211_MODE_11G: rt = ath_hal_getratetable(ah, HAL_MODE_11G); break; case IEEE80211_MODE_TURBO_A: rt = ath_hal_getratetable(ah, HAL_MODE_108A); break; case IEEE80211_MODE_TURBO_G: rt = ath_hal_getratetable(ah, HAL_MODE_108G); break; case IEEE80211_MODE_STURBO_A: rt = ath_hal_getratetable(ah, HAL_MODE_TURBO); break; case IEEE80211_MODE_11NA: rt = ath_hal_getratetable(ah, HAL_MODE_11NA_HT20); break; case IEEE80211_MODE_11NG: rt = ath_hal_getratetable(ah, HAL_MODE_11NG_HT20); break; default: DPRINTF(sc, ATH_DEBUG_ANY, "%s: invalid mode %u\n", __func__, mode); return 0; } sc->sc_rates[mode] = rt; return (rt != NULL); } static void ath_setcurmode(struct ath_softc *sc, enum ieee80211_phymode mode) { #define N(a) (sizeof(a)/sizeof(a[0])) /* NB: on/off times from the Atheros NDIS driver, w/ permission */ static const struct { u_int rate; /* tx/rx 802.11 rate */ u_int16_t timeOn; /* LED on time (ms) */ u_int16_t timeOff; /* LED off time (ms) */ } blinkrates[] = { { 108, 40, 10 }, { 96, 44, 11 }, { 72, 50, 13 }, { 48, 57, 14 }, { 36, 67, 16 }, { 24, 80, 20 }, { 22, 100, 25 }, { 18, 133, 34 }, { 12, 160, 40 }, { 10, 200, 50 }, { 6, 240, 58 }, { 4, 267, 66 }, { 2, 400, 100 }, { 0, 500, 130 }, /* XXX half/quarter rates */ }; const HAL_RATE_TABLE *rt; int i, j; memset(sc->sc_rixmap, 0xff, sizeof(sc->sc_rixmap)); rt = sc->sc_rates[mode]; KASSERT(rt != NULL, ("no h/w rate set for phy mode %u", mode)); for (i = 0; i < rt->rateCount; i++) { uint8_t ieeerate = rt->info[i].dot11Rate & IEEE80211_RATE_VAL; if (rt->info[i].phy != IEEE80211_T_HT) sc->sc_rixmap[ieeerate] = i; else sc->sc_rixmap[ieeerate | IEEE80211_RATE_MCS] = i; } memset(sc->sc_hwmap, 0, sizeof(sc->sc_hwmap)); for (i = 0; i < N(sc->sc_hwmap); i++) { if (i >= rt->rateCount) { sc->sc_hwmap[i].ledon = (500 * hz) / 1000; sc->sc_hwmap[i].ledoff = (130 * hz) / 1000; continue; } sc->sc_hwmap[i].ieeerate = rt->info[i].dot11Rate & IEEE80211_RATE_VAL; if (rt->info[i].phy == IEEE80211_T_HT) sc->sc_hwmap[i].ieeerate |= IEEE80211_RATE_MCS; sc->sc_hwmap[i].txflags = IEEE80211_RADIOTAP_F_DATAPAD; if (rt->info[i].shortPreamble || rt->info[i].phy == IEEE80211_T_OFDM) sc->sc_hwmap[i].txflags |= IEEE80211_RADIOTAP_F_SHORTPRE; sc->sc_hwmap[i].rxflags = sc->sc_hwmap[i].txflags; for (j = 0; j < N(blinkrates)-1; j++) if (blinkrates[j].rate == sc->sc_hwmap[i].ieeerate) break; /* NB: this uses the last entry if the rate isn't found */ /* XXX beware of overlow */ sc->sc_hwmap[i].ledon = (blinkrates[j].timeOn * hz) / 1000; sc->sc_hwmap[i].ledoff = (blinkrates[j].timeOff * hz) / 1000; } sc->sc_currates = rt; sc->sc_curmode = mode; /* * All protection frames are transmited at 2Mb/s for * 11g, otherwise at 1Mb/s. */ if (mode == IEEE80211_MODE_11G) sc->sc_protrix = ath_tx_findrix(sc, 2*2); else sc->sc_protrix = ath_tx_findrix(sc, 2*1); /* NB: caller is responsible for resetting rate control state */ #undef N } static void ath_watchdog(void *arg) { struct ath_softc *sc = arg; if (sc->sc_wd_timer != 0 && --sc->sc_wd_timer == 0) { struct ifnet *ifp = sc->sc_ifp; uint32_t hangs; if (ath_hal_gethangstate(sc->sc_ah, 0xffff, &hangs) && hangs != 0) { if_printf(ifp, "%s hang detected (0x%x)\n", hangs & 0xff ? "bb" : "mac", hangs); } else if_printf(ifp, "device timeout\n"); ath_reset(ifp); ifp->if_oerrors++; sc->sc_stats.ast_watchdog++; } callout_schedule(&sc->sc_wd_ch, hz); } #ifdef ATH_DIAGAPI /* * Diagnostic interface to the HAL. This is used by various * tools to do things like retrieve register contents for * debugging. The mechanism is intentionally opaque so that * it can change frequently w/o concern for compatiblity. */ static int ath_ioctl_diag(struct ath_softc *sc, struct ath_diag *ad) { struct ath_hal *ah = sc->sc_ah; u_int id = ad->ad_id & ATH_DIAG_ID; void *indata = NULL; void *outdata = NULL; u_int32_t insize = ad->ad_in_size; u_int32_t outsize = ad->ad_out_size; int error = 0; if (ad->ad_id & ATH_DIAG_IN) { /* * Copy in data. */ indata = malloc(insize, M_TEMP, M_NOWAIT); if (indata == NULL) { error = ENOMEM; goto bad; } error = copyin(ad->ad_in_data, indata, insize); if (error) goto bad; } if (ad->ad_id & ATH_DIAG_DYN) { /* * Allocate a buffer for the results (otherwise the HAL * returns a pointer to a buffer where we can read the * results). Note that we depend on the HAL leaving this * pointer for us to use below in reclaiming the buffer; * may want to be more defensive. */ outdata = malloc(outsize, M_TEMP, M_NOWAIT); if (outdata == NULL) { error = ENOMEM; goto bad; } } if (ath_hal_getdiagstate(ah, id, indata, insize, &outdata, &outsize)) { if (outsize < ad->ad_out_size) ad->ad_out_size = outsize; if (outdata != NULL) error = copyout(outdata, ad->ad_out_data, ad->ad_out_size); } else { error = EINVAL; } bad: if ((ad->ad_id & ATH_DIAG_IN) && indata != NULL) free(indata, M_TEMP); if ((ad->ad_id & ATH_DIAG_DYN) && outdata != NULL) free(outdata, M_TEMP); return error; } #endif /* ATH_DIAGAPI */ static int ath_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data) { #define IS_RUNNING(ifp) \ ((ifp->if_flags & IFF_UP) && (ifp->if_drv_flags & IFF_DRV_RUNNING)) struct ath_softc *sc = ifp->if_softc; struct ieee80211com *ic = ifp->if_l2com; struct ifreq *ifr = (struct ifreq *)data; const HAL_RATE_TABLE *rt; int error = 0; switch (cmd) { case SIOCSIFFLAGS: ATH_LOCK(sc); if (IS_RUNNING(ifp)) { /* * To avoid rescanning another access point, * do not call ath_init() here. Instead, * only reflect promisc mode settings. */ ath_mode_init(sc); } else if (ifp->if_flags & IFF_UP) { /* * Beware of being called during attach/detach * to reset promiscuous mode. In that case we * will still be marked UP but not RUNNING. * However trying to re-init the interface * is the wrong thing to do as we've already * torn down much of our state. There's * probably a better way to deal with this. */ if (!sc->sc_invalid) ath_init(sc); /* XXX lose error */ } else { ath_stop_locked(ifp); #ifdef notyet /* XXX must wakeup in places like ath_vap_delete */ if (!sc->sc_invalid) ath_hal_setpower(sc->sc_ah, HAL_PM_FULL_SLEEP); #endif } ATH_UNLOCK(sc); break; case SIOCGIFMEDIA: case SIOCSIFMEDIA: error = ifmedia_ioctl(ifp, ifr, &ic->ic_media, cmd); break; case SIOCGATHSTATS: /* NB: embed these numbers to get a consistent view */ sc->sc_stats.ast_tx_packets = ifp->if_opackets; sc->sc_stats.ast_rx_packets = ifp->if_ipackets; sc->sc_stats.ast_tx_rssi = ATH_RSSI(sc->sc_halstats.ns_avgtxrssi); sc->sc_stats.ast_rx_rssi = ATH_RSSI(sc->sc_halstats.ns_avgrssi); #ifdef IEEE80211_SUPPORT_TDMA sc->sc_stats.ast_tdma_tsfadjp = TDMA_AVG(sc->sc_avgtsfdeltap); sc->sc_stats.ast_tdma_tsfadjm = TDMA_AVG(sc->sc_avgtsfdeltam); #endif rt = sc->sc_currates; sc->sc_stats.ast_tx_rate = rt->info[sc->sc_txrix].dot11Rate &~ IEEE80211_RATE_BASIC; if (rt->info[sc->sc_txrix].phy & IEEE80211_T_HT) sc->sc_stats.ast_tx_rate |= IEEE80211_RATE_MCS; return copyout(&sc->sc_stats, ifr->ifr_data, sizeof (sc->sc_stats)); case SIOCZATHSTATS: error = priv_check(curthread, PRIV_DRIVER); if (error == 0) memset(&sc->sc_stats, 0, sizeof(sc->sc_stats)); break; #ifdef ATH_DIAGAPI case SIOCGATHDIAG: error = ath_ioctl_diag(sc, (struct ath_diag *) ifr); break; #endif case SIOCGIFADDR: error = ether_ioctl(ifp, cmd, data); break; default: error = EINVAL; break; } return error; #undef IS_RUNNING } /* * Announce various information on device/driver attach. */ static void ath_announce(struct ath_softc *sc) { struct ifnet *ifp = sc->sc_ifp; struct ath_hal *ah = sc->sc_ah; if_printf(ifp, "AR%s mac %d.%d RF%s phy %d.%d\n", ath_hal_mac_name(ah), ah->ah_macVersion, ah->ah_macRev, ath_hal_rf_name(ah), ah->ah_phyRev >> 4, ah->ah_phyRev & 0xf); if (bootverbose) { int i; for (i = 0; i <= WME_AC_VO; i++) { struct ath_txq *txq = sc->sc_ac2q[i]; if_printf(ifp, "Use hw queue %u for %s traffic\n", txq->axq_qnum, ieee80211_wme_acnames[i]); } if_printf(ifp, "Use hw queue %u for CAB traffic\n", sc->sc_cabq->axq_qnum); if_printf(ifp, "Use hw queue %u for beacons\n", sc->sc_bhalq); } if (ath_rxbuf != ATH_RXBUF) if_printf(ifp, "using %u rx buffers\n", ath_rxbuf); if (ath_txbuf != ATH_TXBUF) if_printf(ifp, "using %u tx buffers\n", ath_txbuf); if (sc->sc_mcastkey && bootverbose) if_printf(ifp, "using multicast key search\n"); } #ifdef IEEE80211_SUPPORT_TDMA static __inline uint32_t ath_hal_getnexttbtt(struct ath_hal *ah) { #define AR_TIMER0 0x8028 return OS_REG_READ(ah, AR_TIMER0); } static __inline void ath_hal_adjusttsf(struct ath_hal *ah, int32_t tsfdelta) { /* XXX handle wrap/overflow */ OS_REG_WRITE(ah, AR_TSF_L32, OS_REG_READ(ah, AR_TSF_L32) + tsfdelta); } static void ath_tdma_settimers(struct ath_softc *sc, u_int32_t nexttbtt, u_int32_t bintval) { struct ath_hal *ah = sc->sc_ah; HAL_BEACON_TIMERS bt; bt.bt_intval = bintval | HAL_BEACON_ENA; bt.bt_nexttbtt = nexttbtt; bt.bt_nextdba = (nexttbtt<<3) - sc->sc_tdmadbaprep; bt.bt_nextswba = (nexttbtt<<3) - sc->sc_tdmaswbaprep; bt.bt_nextatim = nexttbtt+1; ath_hal_beaconsettimers(ah, &bt); } /* * Calculate the beacon interval. This is periodic in the * superframe for the bss. We assume each station is configured * identically wrt transmit rate so the guard time we calculate * above will be the same on all stations. Note we need to * factor in the xmit time because the hardware will schedule * a frame for transmit if the start of the frame is within * the burst time. When we get hardware that properly kills * frames in the PCU we can reduce/eliminate the guard time. * * Roundup to 1024 is so we have 1 TU buffer in the guard time * to deal with the granularity of the nexttbtt timer. 11n MAC's * with 1us timer granularity should allow us to reduce/eliminate * this. */ static void ath_tdma_bintvalsetup(struct ath_softc *sc, const struct ieee80211_tdma_state *tdma) { /* copy from vap state (XXX check all vaps have same value?) */ sc->sc_tdmaslotlen = tdma->tdma_slotlen; sc->sc_tdmabintval = roundup((sc->sc_tdmaslotlen+sc->sc_tdmaguard) * tdma->tdma_slotcnt, 1024); sc->sc_tdmabintval >>= 10; /* TSF -> TU */ if (sc->sc_tdmabintval & 1) sc->sc_tdmabintval++; if (tdma->tdma_slot == 0) { /* * Only slot 0 beacons; other slots respond. */ sc->sc_imask |= HAL_INT_SWBA; sc->sc_tdmaswba = 0; /* beacon immediately */ } else { /* XXX all vaps must be slot 0 or slot !0 */ sc->sc_imask &= ~HAL_INT_SWBA; } } /* * Max 802.11 overhead. This assumes no 4-address frames and * the encapsulation done by ieee80211_encap (llc). We also * include potential crypto overhead. */ #define IEEE80211_MAXOVERHEAD \ (sizeof(struct ieee80211_qosframe) \ + sizeof(struct llc) \ + IEEE80211_ADDR_LEN \ + IEEE80211_WEP_IVLEN \ + IEEE80211_WEP_KIDLEN \ + IEEE80211_WEP_CRCLEN \ + IEEE80211_WEP_MICLEN \ + IEEE80211_CRC_LEN) /* * Setup initially for tdma operation. Start the beacon * timers and enable SWBA if we are slot 0. Otherwise * we wait for slot 0 to arrive so we can sync up before * starting to transmit. */ static void ath_tdma_config(struct ath_softc *sc, struct ieee80211vap *vap) { struct ath_hal *ah = sc->sc_ah; struct ifnet *ifp = sc->sc_ifp; struct ieee80211com *ic = ifp->if_l2com; const struct ieee80211_txparam *tp; const struct ieee80211_tdma_state *tdma = NULL; int rix; if (vap == NULL) { vap = TAILQ_FIRST(&ic->ic_vaps); /* XXX */ if (vap == NULL) { if_printf(ifp, "%s: no vaps?\n", __func__); return; } } tp = vap->iv_bss->ni_txparms; /* * Calculate the guard time for each slot. This is the * time to send a maximal-size frame according to the * fixed/lowest transmit rate. Note that the interface * mtu does not include the 802.11 overhead so we must * tack that on (ath_hal_computetxtime includes the * preamble and plcp in it's calculation). */ tdma = vap->iv_tdma; if (tp->ucastrate != IEEE80211_FIXED_RATE_NONE) rix = ath_tx_findrix(sc, tp->ucastrate); else rix = ath_tx_findrix(sc, tp->mcastrate); /* XXX short preamble assumed */ sc->sc_tdmaguard = ath_hal_computetxtime(ah, sc->sc_currates, ifp->if_mtu + IEEE80211_MAXOVERHEAD, rix, AH_TRUE); ath_hal_intrset(ah, 0); ath_beaconq_config(sc); /* setup h/w beacon q */ if (sc->sc_setcca) ath_hal_setcca(ah, AH_FALSE); /* disable CCA */ ath_tdma_bintvalsetup(sc, tdma); /* calculate beacon interval */ ath_tdma_settimers(sc, sc->sc_tdmabintval, sc->sc_tdmabintval | HAL_BEACON_RESET_TSF); sc->sc_syncbeacon = 0; sc->sc_avgtsfdeltap = TDMA_DUMMY_MARKER; sc->sc_avgtsfdeltam = TDMA_DUMMY_MARKER; ath_hal_intrset(ah, sc->sc_imask); DPRINTF(sc, ATH_DEBUG_TDMA, "%s: slot %u len %uus cnt %u " "bsched %u guard %uus bintval %u TU dba prep %u\n", __func__, tdma->tdma_slot, tdma->tdma_slotlen, tdma->tdma_slotcnt, tdma->tdma_bintval, sc->sc_tdmaguard, sc->sc_tdmabintval, sc->sc_tdmadbaprep); } /* * Update tdma operation. Called from the 802.11 layer * when a beacon is received from the TDMA station operating * in the slot immediately preceding us in the bss. Use * the rx timestamp for the beacon frame to update our * beacon timers so we follow their schedule. Note that * by using the rx timestamp we implicitly include the * propagation delay in our schedule. */ static void ath_tdma_update(struct ieee80211_node *ni, const struct ieee80211_tdma_param *tdma, int changed) { #define TSF_TO_TU(_h,_l) \ ((((u_int32_t)(_h)) << 22) | (((u_int32_t)(_l)) >> 10)) #define TU_TO_TSF(_tu) (((u_int64_t)(_tu)) << 10) struct ieee80211vap *vap = ni->ni_vap; struct ieee80211com *ic = ni->ni_ic; struct ath_softc *sc = ic->ic_ifp->if_softc; struct ath_hal *ah = sc->sc_ah; const HAL_RATE_TABLE *rt = sc->sc_currates; u_int64_t tsf, rstamp, nextslot; u_int32_t txtime, nextslottu, timer0; int32_t tudelta, tsfdelta; const struct ath_rx_status *rs; int rix; sc->sc_stats.ast_tdma_update++; /* * Check for and adopt configuration changes. */ if (changed != 0) { const struct ieee80211_tdma_state *ts = vap->iv_tdma; ath_tdma_bintvalsetup(sc, ts); if (changed & TDMA_UPDATE_SLOTLEN) ath_wme_update(ic); DPRINTF(sc, ATH_DEBUG_TDMA, "%s: adopt slot %u slotcnt %u slotlen %u us " "bintval %u TU\n", __func__, ts->tdma_slot, ts->tdma_slotcnt, ts->tdma_slotlen, sc->sc_tdmabintval); /* XXX right? */ ath_hal_intrset(ah, sc->sc_imask); /* NB: beacon timers programmed below */ } /* extend rx timestamp to 64 bits */ rs = sc->sc_lastrs; tsf = ath_hal_gettsf64(ah); rstamp = ath_extend_tsf(rs->rs_tstamp, tsf); /* * The rx timestamp is set by the hardware on completing * reception (at the point where the rx descriptor is DMA'd * to the host). To find the start of our next slot we * must adjust this time by the time required to send * the packet just received. */ rix = rt->rateCodeToIndex[rs->rs_rate]; txtime = ath_hal_computetxtime(ah, rt, rs->rs_datalen, rix, rt->info[rix].shortPreamble); /* NB: << 9 is to cvt to TU and /2 */ nextslot = (rstamp - txtime) + (sc->sc_tdmabintval << 9); nextslottu = TSF_TO_TU(nextslot>>32, nextslot) & HAL_BEACON_PERIOD; /* * TIMER0 is the h/w's idea of NextTBTT (in TU's). Convert * to usecs and calculate the difference between what the * other station thinks and what we have programmed. This * lets us figure how to adjust our timers to match. The * adjustments are done by pulling the TSF forward and possibly * rewriting the beacon timers. */ timer0 = ath_hal_getnexttbtt(ah); tsfdelta = (int32_t)((nextslot % TU_TO_TSF(HAL_BEACON_PERIOD+1)) - TU_TO_TSF(timer0)); DPRINTF(sc, ATH_DEBUG_TDMA_TIMER, "tsfdelta %d avg +%d/-%d\n", tsfdelta, TDMA_AVG(sc->sc_avgtsfdeltap), TDMA_AVG(sc->sc_avgtsfdeltam)); if (tsfdelta < 0) { TDMA_SAMPLE(sc->sc_avgtsfdeltap, 0); TDMA_SAMPLE(sc->sc_avgtsfdeltam, -tsfdelta); tsfdelta = -tsfdelta % 1024; nextslottu++; } else if (tsfdelta > 0) { TDMA_SAMPLE(sc->sc_avgtsfdeltap, tsfdelta); TDMA_SAMPLE(sc->sc_avgtsfdeltam, 0); tsfdelta = 1024 - (tsfdelta % 1024); nextslottu++; } else { TDMA_SAMPLE(sc->sc_avgtsfdeltap, 0); TDMA_SAMPLE(sc->sc_avgtsfdeltam, 0); } tudelta = nextslottu - timer0; /* * Copy sender's timetstamp into tdma ie so they can * calculate roundtrip time. We submit a beacon frame * below after any timer adjustment. The frame goes out * at the next TBTT so the sender can calculate the * roundtrip by inspecting the tdma ie in our beacon frame. * * NB: This tstamp is subtlely preserved when * IEEE80211_BEACON_TDMA is marked (e.g. when the * slot position changes) because ieee80211_add_tdma * skips over the data. */ memcpy(ATH_VAP(vap)->av_boff.bo_tdma + __offsetof(struct ieee80211_tdma_param, tdma_tstamp), &ni->ni_tstamp.data, 8); #if 0 DPRINTF(sc, ATH_DEBUG_TDMA_TIMER, "tsf %llu nextslot %llu (%d, %d) nextslottu %u timer0 %u (%d)\n", (unsigned long long) tsf, (unsigned long long) nextslot, (int)(nextslot - tsf), tsfdelta, nextslottu, timer0, tudelta); #endif /* * Adjust the beacon timers only when pulling them forward * or when going back by less than the beacon interval. * Negative jumps larger than the beacon interval seem to * cause the timers to stop and generally cause instability. * This basically filters out jumps due to missed beacons. */ if (tudelta != 0 && (tudelta > 0 || -tudelta < sc->sc_tdmabintval)) { ath_tdma_settimers(sc, nextslottu, sc->sc_tdmabintval); sc->sc_stats.ast_tdma_timers++; } if (tsfdelta > 0) { ath_hal_adjusttsf(ah, tsfdelta); sc->sc_stats.ast_tdma_tsf++; } ath_tdma_beacon_send(sc, vap); /* prepare response */ #undef TU_TO_TSF #undef TSF_TO_TU } /* * Transmit a beacon frame at SWBA. Dynamic updates * to the frame contents are done as needed. */ static void ath_tdma_beacon_send(struct ath_softc *sc, struct ieee80211vap *vap) { struct ath_hal *ah = sc->sc_ah; struct ath_buf *bf; int otherant; /* * Check if the previous beacon has gone out. If * not don't try to post another, skip this period * and wait for the next. Missed beacons indicate * a problem and should not occur. If we miss too * many consecutive beacons reset the device. */ if (ath_hal_numtxpending(ah, sc->sc_bhalq) != 0) { sc->sc_bmisscount++; DPRINTF(sc, ATH_DEBUG_BEACON, "%s: missed %u consecutive beacons\n", __func__, sc->sc_bmisscount); if (sc->sc_bmisscount >= ath_bstuck_threshold) taskqueue_enqueue(sc->sc_tq, &sc->sc_bstucktask); return; } if (sc->sc_bmisscount != 0) { DPRINTF(sc, ATH_DEBUG_BEACON, "%s: resume beacon xmit after %u misses\n", __func__, sc->sc_bmisscount); sc->sc_bmisscount = 0; } /* * Check recent per-antenna transmit statistics and flip * the default antenna if noticeably more frames went out * on the non-default antenna. * XXX assumes 2 anntenae */ if (!sc->sc_diversity) { otherant = sc->sc_defant & 1 ? 2 : 1; if (sc->sc_ant_tx[otherant] > sc->sc_ant_tx[sc->sc_defant] + 2) ath_setdefantenna(sc, otherant); sc->sc_ant_tx[1] = sc->sc_ant_tx[2] = 0; } bf = ath_beacon_generate(sc, vap); if (bf != NULL) { /* * Stop any current dma and put the new frame on the queue. * This should never fail since we check above that no frames * are still pending on the queue. */ if (!ath_hal_stoptxdma(ah, sc->sc_bhalq)) { DPRINTF(sc, ATH_DEBUG_ANY, "%s: beacon queue %u did not stop?\n", __func__, sc->sc_bhalq); /* NB: the HAL still stops DMA, so proceed */ } ath_hal_puttxbuf(ah, sc->sc_bhalq, bf->bf_daddr); ath_hal_txstart(ah, sc->sc_bhalq); sc->sc_stats.ast_be_xmit++; /* XXX per-vap? */ /* * Record local TSF for our last send for use * in arbitrating slot collisions. */ vap->iv_bss->ni_tstamp.tsf = ath_hal_gettsf64(ah); } } #endif /* IEEE80211_SUPPORT_TDMA */ static void ath_dfs_tasklet(void *p, int npending) { struct ath_softc *sc = (struct ath_softc *) p; struct ifnet *ifp = sc->sc_ifp; struct ieee80211com *ic = ifp->if_l2com; /* * If previous processing has found a radar event, * signal this to the net80211 layer to begin DFS * processing. */ if (ath_dfs_process_radar_event(sc, sc->sc_curchan)) { /* DFS event found, initiate channel change */ ieee80211_dfs_notify_radar(ic, sc->sc_curchan); } } MODULE_VERSION(if_ath, 1); MODULE_DEPEND(if_ath, wlan, 1, 1, 1); /* 802.11 media layer */ Index: projects/largeSMP/sys/dev/cardbus/cardbus_cis.c =================================================================== --- projects/largeSMP/sys/dev/cardbus/cardbus_cis.c (revision 222782) +++ projects/largeSMP/sys/dev/cardbus/cardbus_cis.c (revision 222783) @@ -1,641 +1,640 @@ /*- * Copyright (c) 2005-2008, M. Warner Losh * Copyright (c) 2000,2001 Jonathan Chen. * 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("$FreeBSD$"); /* * CIS Handling for the Cardbus Bus */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include extern int cardbus_cis_debug; #define DPRINTF(a) if (cardbus_cis_debug) printf a #define DEVPRINTF(x) if (cardbus_cis_debug) device_printf x #define CIS_CONFIG_SPACE (struct resource *)~0UL static int decode_tuple_generic(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *); static int decode_tuple_linktarget(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *); static int decode_tuple_vers_1(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *); static int decode_tuple_funcid(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *); static int decode_tuple_manfid(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *); static int decode_tuple_funce(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *); static int decode_tuple_bar(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *); static int decode_tuple_unhandled(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *); static int decode_tuple_end(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *); static int cardbus_read_tuple_conf(device_t cbdev, device_t child, uint32_t start, uint32_t *off, int *tupleid, int *len, uint8_t *tupledata); static int cardbus_read_tuple_mem(device_t cbdev, struct resource *res, uint32_t start, uint32_t *off, int *tupleid, int *len, uint8_t *tupledata); static int cardbus_read_tuple(device_t cbdev, device_t child, struct resource *res, uint32_t start, uint32_t *off, int *tupleid, int *len, uint8_t *tupledata); static void cardbus_read_tuple_finish(device_t cbdev, device_t child, int rid, struct resource *res); static struct resource *cardbus_read_tuple_init(device_t cbdev, device_t child, uint32_t *start, int *rid); static int decode_tuple(device_t cbdev, device_t child, int tupleid, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *callbacks, void *); #define MAKETUPLE(NAME,FUNC) { CISTPL_ ## NAME, #NAME, decode_tuple_ ## FUNC } static char *funcnames[] = { "Multi-Functioned", "Memory", "Serial Port", "Parallel Port", "Fixed Disk", "Video Adaptor", "Network Adaptor", "AIMS", "SCSI", "Security" }; /* * Handler functions for various CIS tuples */ static int decode_tuple_generic(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *argp) { int i; if (cardbus_cis_debug) { if (info) printf("TUPLE: %s [%d]:", info->name, len); else printf("TUPLE: Unknown(0x%02x) [%d]:", id, len); for (i = 0; i < len; i++) { if (i % 0x10 == 0 && len > 0x10) printf("\n 0x%02x:", i); printf(" %02x", tupledata[i]); } printf("\n"); } return (0); } static int decode_tuple_linktarget(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *argp) { int i; if (cardbus_cis_debug) { printf("TUPLE: %s [%d]:", info->name, len); for (i = 0; i < len; i++) { if (i % 0x10 == 0 && len > 0x10) printf("\n 0x%02x:", i); printf(" %02x", tupledata[i]); } printf("\n"); } if (len != 3 || tupledata[0] != 'C' || tupledata[1] != 'I' || tupledata[2] != 'S') { printf("Invalid data for CIS Link Target!\n"); decode_tuple_generic(cbdev, child, id, len, tupledata, start, off, info, argp); return (EINVAL); } return (0); } static int decode_tuple_vers_1(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *argp) { int i; if (cardbus_cis_debug) { printf("Product version: %d.%d\n", tupledata[0], tupledata[1]); printf("Product name: "); for (i = 2; i < len; i++) { if (tupledata[i] == '\0') printf(" | "); else if (tupledata[i] == 0xff) break; else printf("%c", tupledata[i]); } printf("\n"); } return (0); } static int decode_tuple_funcid(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *argp) { struct cardbus_devinfo *dinfo = device_get_ivars(child); int numnames = sizeof(funcnames) / sizeof(funcnames[0]); int i; if (cardbus_cis_debug) { printf("Functions: "); for (i = 0; i < len; i++) { if (tupledata[i] < numnames) printf("%s", funcnames[tupledata[i]]); else printf("Unknown(%d)", tupledata[i]); if (i < len - 1) printf(", "); } printf("\n"); } if (len > 0) dinfo->funcid = tupledata[0]; /* use first in list */ return (0); } static int decode_tuple_manfid(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *argp) { struct cardbus_devinfo *dinfo = device_get_ivars(child); int i; if (cardbus_cis_debug) { printf("Manufacturer ID: "); for (i = 0; i < len; i++) printf("%02x", tupledata[i]); printf("\n"); } if (len == 5) { dinfo->mfrid = tupledata[1] | (tupledata[2] << 8); dinfo->prodid = tupledata[3] | (tupledata[4] << 8); } return (0); } static int decode_tuple_funce(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *argp) { struct cardbus_devinfo *dinfo = device_get_ivars(child); int type, i; if (cardbus_cis_debug) { printf("Function Extension: "); for (i = 0; i < len; i++) printf("%02x", tupledata[i]); printf("\n"); } if (len < 2) /* too short */ return (0); type = tupledata[0]; /* XXX <32 always? */ switch (dinfo->funcid) { case PCCARD_FUNCTION_NETWORK: switch (type) { case PCCARD_TPLFE_TYPE_LAN_NID: if (tupledata[1] > sizeof(dinfo->funce.lan.nid)) { /* ignore, warning? */ return (0); } bcopy(tupledata + 2, dinfo->funce.lan.nid, tupledata[1]); break; } dinfo->fepresent |= 1<mprefetchable |= (1 << PCI_RID2BAR(bar)); /* * The PC Card spec says we're only supposed to honor this * hint when the cardbus bridge is a child of pci0 (the main * bus). The PC Card spec seems to indicate that this should * only be done on x86 based machines, which suggests that on - * non-x86 machines the adddresses can be anywhere. Since the + * non-x86 machines the addresses can be anywhere. Since the * hardware can do it on non-x86 machines, it should be able * to do it on x86 machines too. Therefore, we can and should * ignore this hint. Furthermore, the PC Card spec recommends * always allocating memory above 1MB, contradicting the other * part of the PC Card spec, it seems. We make note of it, * but otherwise don't use this information. * * Some Realtek cards have this set in their CIS, but fail * to actually work when mapped this way, and experience * has shown ignoring this big to be a wise choice. * * XXX We should cite chapter and verse for standard refs. */ if (reg & TPL_BAR_REG_BELOW1MB) dinfo->mbelow1mb |= (1 << PCI_RID2BAR(bar)); } return (0); } static int decode_tuple_unhandled(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *argp) { /* Make this message suck less XXX */ printf("TUPLE: %s [%d] is unhandled! Bailing...", info->name, len); return (EINVAL); } static int decode_tuple_end(device_t cbdev, device_t child, int id, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *info, void *argp) { if (cardbus_cis_debug) printf("CIS reading done\n"); return (0); } /* * Functions to read the a tuple from the card */ static int cardbus_read_tuple_conf(device_t cbdev, device_t child, uint32_t start, uint32_t *off, int *tupleid, int *len, uint8_t *tupledata) { int i, j; uint32_t e; uint32_t loc; loc = start + *off; e = pci_read_config(child, loc - loc % 4, 4); for (j = loc % 4; j > 0; j--) e >>= 8; *len = 0; for (i = loc, j = -2; j < *len; j++, i++) { if (i % 4 == 0) e = pci_read_config(child, i, 4); if (j == -2) *tupleid = 0xff & e; else if (j == -1) *len = 0xff & e; else tupledata[j] = 0xff & e; e >>= 8; } *off += *len + 2; return (0); } static int cardbus_read_tuple_mem(device_t cbdev, struct resource *res, uint32_t start, uint32_t *off, int *tupleid, int *len, uint8_t *tupledata) { int ret; *tupleid = bus_read_1(res, start + *off); *len = bus_read_1(res, start + *off + 1); bus_read_region_1(res, *off + start + 2, tupledata, *len); ret = 0; *off += *len + 2; return (ret); } static int cardbus_read_tuple(device_t cbdev, device_t child, struct resource *res, uint32_t start, uint32_t *off, int *tupleid, int *len, uint8_t *tupledata) { if (res == CIS_CONFIG_SPACE) return (cardbus_read_tuple_conf(cbdev, child, start, off, tupleid, len, tupledata)); return (cardbus_read_tuple_mem(cbdev, res, start, off, tupleid, len, tupledata)); } static void cardbus_read_tuple_finish(device_t cbdev, device_t child, int rid, struct resource *res) { if (res != CIS_CONFIG_SPACE) { bus_release_resource(child, SYS_RES_MEMORY, rid, res); - bus_delete_resource(child, SYS_RES_MEMORY, rid); } } static struct resource * cardbus_read_tuple_init(device_t cbdev, device_t child, uint32_t *start, int *rid) { struct resource *res; uint32_t space; space = *start & PCIM_CIS_ASI_MASK; switch (space) { case PCIM_CIS_ASI_CONFIG: DEVPRINTF((cbdev, "CIS in PCI config space\n")); /* CIS in PCI config space need no initialization */ return (CIS_CONFIG_SPACE); case PCIM_CIS_ASI_BAR0: case PCIM_CIS_ASI_BAR1: case PCIM_CIS_ASI_BAR2: case PCIM_CIS_ASI_BAR3: case PCIM_CIS_ASI_BAR4: case PCIM_CIS_ASI_BAR5: *rid = PCIR_BAR(space - PCIM_CIS_ASI_BAR0); DEVPRINTF((cbdev, "CIS in BAR %#x\n", *rid)); break; case PCIM_CIS_ASI_ROM: *rid = PCIR_BIOS; DEVPRINTF((cbdev, "CIS in option rom\n")); break; default: device_printf(cbdev, "Unable to read CIS: Unknown space: %d\n", space); return (NULL); } /* allocate the memory space to read CIS */ - res = bus_alloc_resource(child, SYS_RES_MEMORY, rid, 0, ~0, 1, + res = bus_alloc_resource_any(child, SYS_RES_MEMORY, rid, rman_make_alignment_flags(4096) | RF_ACTIVE); if (res == NULL) { device_printf(cbdev, "Unable to allocate resource " "to read CIS.\n"); return (NULL); } DEVPRINTF((cbdev, "CIS Mapped to %#lx\n", rman_get_start(res))); /* Flip to the right ROM image if CIS is in ROM */ if (space == PCIM_CIS_ASI_ROM) { uint32_t imagesize; uint32_t imagebase = 0; uint32_t pcidata; uint16_t romsig; int romnum = 0; int imagenum; imagenum = (*start & PCIM_CIS_ROM_MASK) >> 28; for (romnum = 0;; romnum++) { romsig = bus_read_2(res, imagebase + CARDBUS_EXROM_SIGNATURE); if (romsig != 0xaa55) { device_printf(cbdev, "Bad header in rom %d: " "[%x] %04x\n", romnum, imagebase + CARDBUS_EXROM_SIGNATURE, romsig); cardbus_read_tuple_finish(cbdev, child, *rid, res); *rid = 0; return (NULL); } /* * If this was the Option ROM image that we were * looking for, then we are done. */ if (romnum == imagenum) break; /* Find out where the next Option ROM image is */ pcidata = imagebase + bus_read_2(res, imagebase + CARDBUS_EXROM_DATA_PTR); imagesize = bus_read_2(res, pcidata + CARDBUS_EXROM_DATA_IMAGE_LENGTH); if (imagesize == 0) { /* * XXX some ROMs seem to have this as zero, * can we assume this means 1 block? */ device_printf(cbdev, "Warning, size of Option " "ROM image %d is 0 bytes, assuming 512 " "bytes.\n", romnum); imagesize = 1; } /* Image size is in 512 byte units */ imagesize <<= 9; if ((bus_read_1(res, pcidata + CARDBUS_EXROM_DATA_INDICATOR) & 0x80) != 0) { device_printf(cbdev, "Cannot find CIS in " "Option ROM\n"); cardbus_read_tuple_finish(cbdev, child, *rid, res); *rid = 0; return (NULL); } imagebase += imagesize; } *start = imagebase + (*start & PCIM_CIS_ADDR_MASK); } else { *start = *start & PCIM_CIS_ADDR_MASK; } DEVPRINTF((cbdev, "CIS offset is %#x\n", *start)); return (res); } /* * Dispatch the right handler function per tuple */ static int decode_tuple(device_t cbdev, device_t child, int tupleid, int len, uint8_t *tupledata, uint32_t start, uint32_t *off, struct tuple_callbacks *callbacks, void *argp) { int i; for (i = 0; callbacks[i].id != CISTPL_GENERIC; i++) { if (tupleid == callbacks[i].id) return (callbacks[i].func(cbdev, child, tupleid, len, tupledata, start, off, &callbacks[i], argp)); } return (callbacks[i].func(cbdev, child, tupleid, len, tupledata, start, off, NULL, argp)); } int cardbus_parse_cis(device_t cbdev, device_t child, struct tuple_callbacks *callbacks, void *argp) { uint8_t tupledata[MAXTUPLESIZE]; int tupleid = CISTPL_NULL; int len; int expect_linktarget; uint32_t start, off; struct resource *res; int rid; bzero(tupledata, MAXTUPLESIZE); expect_linktarget = TRUE; if ((start = pci_read_config(child, PCIR_CIS, 4)) == 0) { DEVPRINTF((cbdev, "Warning: CIS pointer is 0: (no CIS)\n")); return (ENXIO); } DEVPRINTF((cbdev, "CIS pointer is %#x\n", start)); off = 0; res = cardbus_read_tuple_init(cbdev, child, &start, &rid); if (res == NULL) { device_printf(cbdev, "Unable to allocate resources for CIS\n"); return (ENXIO); } do { if (cardbus_read_tuple(cbdev, child, res, start, &off, &tupleid, &len, tupledata) != 0) { device_printf(cbdev, "Failed to read CIS.\n"); cardbus_read_tuple_finish(cbdev, child, rid, res); return (ENXIO); } if (expect_linktarget && tupleid != CISTPL_LINKTARGET) { device_printf(cbdev, "Expecting link target, got 0x%x\n", tupleid); cardbus_read_tuple_finish(cbdev, child, rid, res); return (EINVAL); } expect_linktarget = decode_tuple(cbdev, child, tupleid, len, tupledata, start, &off, callbacks, argp); if (expect_linktarget != 0) { device_printf(cbdev, "Parsing failed with %d\n", expect_linktarget); cardbus_read_tuple_finish(cbdev, child, rid, res); return (expect_linktarget); } } while (tupleid != CISTPL_END); cardbus_read_tuple_finish(cbdev, child, rid, res); return (0); } int cardbus_do_cis(device_t cbdev, device_t child) { struct tuple_callbacks init_callbacks[] = { MAKETUPLE(LONGLINK_CB, unhandled), MAKETUPLE(INDIRECT, unhandled), MAKETUPLE(LONGLINK_MFC, unhandled), MAKETUPLE(BAR, bar), MAKETUPLE(LONGLINK_A, unhandled), MAKETUPLE(LONGLINK_C, unhandled), MAKETUPLE(LINKTARGET, linktarget), MAKETUPLE(VERS_1, vers_1), MAKETUPLE(MANFID, manfid), MAKETUPLE(FUNCID, funcid), MAKETUPLE(FUNCE, funce), MAKETUPLE(END, end), MAKETUPLE(GENERIC, generic), }; return (cardbus_parse_cis(cbdev, child, init_callbacks, NULL)); } Index: projects/largeSMP/sys/dev/cxgbe/adapter.h =================================================================== --- projects/largeSMP/sys/dev/cxgbe/adapter.h (revision 222782) +++ projects/largeSMP/sys/dev/cxgbe/adapter.h (revision 222783) @@ -1,616 +1,617 @@ /*- * Copyright (c) 2011 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. * * $FreeBSD$ * */ #ifndef __T4_ADAPTER_H__ #define __T4_ADAPTER_H__ #include #include #include #include #include #include #include #include #include #include #include #include #include #include "offload.h" #include "common/t4fw_interface.h" #define T4_FWNAME "t4fw" MALLOC_DECLARE(M_CXGBE); #define CXGBE_UNIMPLEMENTED(s) \ panic("%s (%s, line %d) not implemented yet.", s, __FILE__, __LINE__) #if defined(__i386__) || defined(__amd64__) static __inline void prefetch(void *x) { __asm volatile("prefetcht0 %0" :: "m" (*(unsigned long *)x)); } #else #define prefetch(x) #endif #ifdef __amd64__ /* XXX: need systemwide bus_space_read_8/bus_space_write_8 */ static __inline uint64_t t4_bus_space_read_8(bus_space_tag_t tag, bus_space_handle_t handle, bus_size_t offset) { KASSERT(tag == X86_BUS_SPACE_MEM, ("%s: can only handle mem space", __func__)); return (*(volatile uint64_t *)(handle + offset)); } static __inline void t4_bus_space_write_8(bus_space_tag_t tag, bus_space_handle_t bsh, bus_size_t offset, uint64_t value) { KASSERT(tag == X86_BUS_SPACE_MEM, ("%s: can only handle mem space", __func__)); *(volatile uint64_t *)(bsh + offset) = value; } #else static __inline uint64_t t4_bus_space_read_8(bus_space_tag_t tag, bus_space_handle_t handle, bus_size_t offset) { return (uint64_t)bus_space_read_4(tag, handle, offset) + ((uint64_t)bus_space_read_4(tag, handle, offset + 4) << 32); } static __inline void t4_bus_space_write_8(bus_space_tag_t tag, bus_space_handle_t bsh, bus_size_t offset, uint64_t value) { bus_space_write_4(tag, bsh, offset, value); bus_space_write_4(tag, bsh, offset + 4, value >> 32); } #endif struct adapter; typedef struct adapter adapter_t; enum { FW_IQ_QSIZE = 256, FW_IQ_ESIZE = 64, /* At least 64 mandated by the firmware spec */ INTR_IQ_QSIZE = 64, INTR_IQ_ESIZE = 64, /* Handles some CPLs too, do not reduce */ CTRL_EQ_QSIZE = 128, CTRL_EQ_ESIZE = 64, RX_IQ_QSIZE = 1024, RX_IQ_ESIZE = 64, /* At least 64 so CPL_RX_PKT will fit */ RX_FL_ESIZE = 64, /* 8 64bit addresses */ #if MJUMPAGESIZE != MCLBYTES FL_BUF_SIZES = 4, /* cluster, jumbop, jumbo9k, jumbo16k */ #else FL_BUF_SIZES = 3, /* cluster, jumbo9k, jumbo16k */ #endif TX_EQ_QSIZE = 1024, TX_EQ_ESIZE = 64, TX_SGL_SEGS = 36, TX_WR_FLITS = SGE_MAX_WR_LEN / 8 }; enum { /* adapter intr_type */ INTR_INTX = (1 << 0), INTR_MSI = (1 << 1), INTR_MSIX = (1 << 2) }; enum { /* adapter flags */ FULL_INIT_DONE = (1 << 0), FW_OK = (1 << 1), INTR_SHARED = (1 << 2), /* one set of intrq's for all ports */ CXGBE_BUSY = (1 << 9), /* port flags */ DOOMED = (1 << 0), VI_ENABLED = (1 << 1), }; #define IS_DOOMED(pi) (pi->flags & DOOMED) #define SET_DOOMED(pi) do {pi->flags |= DOOMED;} while (0) #define IS_BUSY(sc) (sc->flags & CXGBE_BUSY) #define SET_BUSY(sc) do {sc->flags |= CXGBE_BUSY;} while (0) #define CLR_BUSY(sc) do {sc->flags &= ~CXGBE_BUSY;} while (0) struct port_info { device_t dev; struct adapter *adapter; struct ifnet *ifp; struct ifmedia media; struct mtx pi_lock; char lockname[16]; unsigned long flags; int if_flags; uint16_t viid; int16_t xact_addr_filt;/* index of exact MAC address filter */ uint16_t rss_size; /* size of VI's RSS table slice */ uint8_t lport; /* associated offload logical port */ int8_t mdio_addr; uint8_t port_type; uint8_t mod_type; uint8_t port_id; uint8_t tx_chan; /* These need to be int as they are used in sysctl */ int ntxq; /* # of tx queues */ int first_txq; /* index of first tx queue */ int nrxq; /* # of rx queues */ int first_rxq; /* index of first rx queue */ int tmr_idx; int pktc_idx; int qsize_rxq; int qsize_txq; struct link_config link_cfg; struct port_stats stats; struct taskqueue *tq; struct callout tick; struct sysctl_ctx_list ctx; /* lives from ifconfig up to down */ struct sysctl_oid *oid_rxq; struct sysctl_oid *oid_txq; uint8_t hw_addr[ETHER_ADDR_LEN]; /* factory MAC address, won't change */ }; struct fl_sdesc { struct mbuf *m; bus_dmamap_t map; caddr_t cl; uint8_t tag_idx; /* the sc->fl_tag this map comes from */ #ifdef INVARIANTS __be64 ba_tag; #endif }; struct tx_desc { __be64 flit[8]; }; struct tx_map { struct mbuf *m; bus_dmamap_t map; }; struct tx_sdesc { uint8_t desc_used; /* # of hardware descriptors used by the WR */ uint8_t credits; /* NIC txq: # of frames sent out in the WR */ }; typedef void (iq_intr_handler_t)(void *); enum { /* iq flags */ IQ_ALLOCATED = (1 << 1), /* firmware resources allocated */ IQ_STARTED = (1 << 2), /* started */ /* iq state */ IQS_DISABLED = 0, IQS_BUSY = 1, IQS_IDLE = 2, }; /* * Ingress Queue: T4 is producer, driver is consumer. */ struct sge_iq { bus_dma_tag_t desc_tag; bus_dmamap_t desc_map; bus_addr_t ba; /* bus address of descriptor ring */ char lockname[16]; uint32_t flags; uint16_t abs_id; /* absolute SGE id for the iq */ int8_t intr_pktc_idx; /* packet count threshold index */ int8_t pad0; iq_intr_handler_t *handler; __be64 *desc; /* KVA of descriptor ring */ volatile uint32_t state; struct adapter *adapter; const __be64 *cdesc; /* current descriptor */ uint8_t gen; /* generation bit */ uint8_t intr_params; /* interrupt holdoff parameters */ uint8_t intr_next; /* holdoff for next interrupt */ uint8_t esize; /* size (bytes) of each entry in the queue */ uint16_t qsize; /* size (# of entries) of the queue */ uint16_t cidx; /* consumer index */ uint16_t cntxt_id; /* SGE context id for the iq */ }; enum { /* eq flags */ EQ_ALLOCATED = (1 << 1), /* firmware resources allocated */ EQ_STARTED = (1 << 2), /* started */ EQ_CRFLUSHED = (1 << 3), /* expecting an update from SGE */ }; /* * Egress Queue: driver is producer, T4 is consumer. * * Note: A free list is an egress queue (driver produces the buffers and T4 * consumes them) but it's special enough to have its own struct (see sge_fl). */ struct sge_eq { bus_dma_tag_t desc_tag; bus_dmamap_t desc_map; char lockname[16]; unsigned int flags; struct mtx eq_lock; struct tx_desc *desc; /* KVA of descriptor ring */ bus_addr_t ba; /* bus address of descriptor ring */ struct sge_qstat *spg; /* status page, for convenience */ uint16_t cap; /* max # of desc, for convenience */ uint16_t avail; /* available descriptors, for convenience */ uint16_t qsize; /* size (# of entries) of the queue */ uint16_t cidx; /* consumer idx (desc idx) */ uint16_t pidx; /* producer idx (desc idx) */ uint16_t pending; /* # of descriptors used since last doorbell */ uint16_t iqid; /* iq that gets egr_update for the eq */ uint32_t cntxt_id; /* SGE context id for the eq */ }; struct sge_fl { bus_dma_tag_t desc_tag; bus_dmamap_t desc_map; bus_dma_tag_t tag[FL_BUF_SIZES]; uint8_t tag_idx; struct mtx fl_lock; char lockname[16]; __be64 *desc; /* KVA of descriptor ring, ptr to addresses */ bus_addr_t ba; /* bus address of descriptor ring */ struct fl_sdesc *sdesc; /* KVA of software descriptor ring */ uint32_t cap; /* max # of buffers, for convenience */ uint16_t qsize; /* size (# of entries) of the queue */ uint16_t cntxt_id; /* SGE context id for the freelist */ uint32_t cidx; /* consumer idx (buffer idx, NOT hw desc idx) */ uint32_t pidx; /* producer idx (buffer idx, NOT hw desc idx) */ uint32_t needed; /* # of buffers needed to fill up fl. */ uint32_t pending; /* # of bufs allocated since last doorbell */ unsigned int dmamap_failed; }; /* txq: SGE egress queue + what's needed for Ethernet NIC */ struct sge_txq { struct sge_eq eq; /* MUST be first */ struct ifnet *ifp; /* the interface this txq belongs to */ bus_dma_tag_t tx_tag; /* tag for transmit buffers */ struct buf_ring *br; /* tx buffer ring */ struct tx_sdesc *sdesc; /* KVA of software descriptor ring */ struct mbuf *m; /* held up due to temporary resource shortage */ struct task resume_tx; /* DMA maps used for tx */ struct tx_map *maps; uint32_t map_total; /* # of DMA maps */ uint32_t map_pidx; /* next map to be used */ uint32_t map_cidx; /* reclaimed up to this index */ uint32_t map_avail; /* # of available maps */ /* stats for common events first */ uint64_t txcsum; /* # of times hardware assisted with checksum */ uint64_t tso_wrs; /* # of IPv4 TSO work requests */ uint64_t vlan_insertion;/* # of times VLAN tag was inserted */ uint64_t imm_wrs; /* # of work requests with immediate data */ uint64_t sgl_wrs; /* # of work requests with direct SGL */ uint64_t txpkt_wrs; /* # of txpkt work requests (not coalesced) */ uint64_t txpkts_wrs; /* # of coalesced tx work requests */ uint64_t txpkts_pkts; /* # of frames in coalesced tx work requests */ /* stats for not-that-common events */ uint32_t no_dmamap; /* no DMA map to load the mbuf */ uint32_t no_desc; /* out of hardware descriptors */ uint32_t egr_update; /* # of SGE_EGR_UPDATE notifications for txq */ } __aligned(CACHE_LINE_SIZE); enum { RXQ_LRO_ENABLED = (1 << 0) }; /* rxq: SGE ingress queue + SGE free list + miscellaneous items */ struct sge_rxq { struct sge_iq iq; /* MUST be first */ struct sge_fl fl; struct ifnet *ifp; /* the interface this rxq belongs to */ unsigned int flags; #ifdef INET struct lro_ctrl lro; /* LRO state */ #endif /* stats for common events first */ uint64_t rxcsum; /* # of times hardware assisted with checksum */ uint64_t vlan_extraction;/* # of times VLAN tag was extracted */ /* stats for not-that-common events */ } __aligned(CACHE_LINE_SIZE); /* ctrlq: SGE egress queue + stats for control queue */ struct sge_ctrlq { struct sge_eq eq; /* MUST be first */ /* stats for common events first */ /* stats for not-that-common events */ uint32_t no_desc; /* out of hardware descriptors */ } __aligned(CACHE_LINE_SIZE); struct sge { uint16_t timer_val[SGE_NTIMERS]; uint8_t counter_val[SGE_NCOUNTERS]; + int fl_starve_threshold; int nrxq; /* total rx queues (all ports and the rest) */ int ntxq; /* total tx queues (all ports and the rest) */ int niq; /* total ingress queues */ int neq; /* total egress queues */ struct sge_iq fwq; /* Firmware event queue */ struct sge_ctrlq *ctrlq;/* Control queues */ struct sge_iq *intrq; /* Interrupt queues */ struct sge_txq *txq; /* NIC tx queues */ struct sge_rxq *rxq; /* NIC rx queues */ uint16_t iq_start; int eq_start; struct sge_iq **iqmap; /* iq->cntxt_id to iq mapping */ struct sge_eq **eqmap; /* eq->cntxt_id to eq mapping */ }; struct adapter { device_t dev; struct cdev *cdev; /* PCIe register resources */ int regs_rid; struct resource *regs_res; int msix_rid; struct resource *msix_res; bus_space_handle_t bh; bus_space_tag_t bt; bus_size_t mmio_len; unsigned int pf; unsigned int mbox; /* Interrupt information */ int intr_type; int intr_count; struct irq { struct resource *res; int rid; void *tag; } *irq; bus_dma_tag_t dmat; /* Parent DMA tag */ struct sge sge; struct port_info *port[MAX_NPORTS]; uint8_t chan_map[NCHAN]; struct l2t_data *l2t; /* L2 table */ struct tid_info tids; int registered_device_map; int open_device_map; int flags; char fw_version[32]; struct adapter_params params; struct t4_virt_res vres; struct sysctl_ctx_list ctx; /* from first_port_up to last_port_down */ struct sysctl_oid *oid_fwq; struct sysctl_oid *oid_ctrlq; struct sysctl_oid *oid_intrq; struct mtx sc_lock; char lockname[16]; }; #define ADAPTER_LOCK(sc) mtx_lock(&(sc)->sc_lock) #define ADAPTER_UNLOCK(sc) mtx_unlock(&(sc)->sc_lock) #define ADAPTER_LOCK_ASSERT_OWNED(sc) mtx_assert(&(sc)->sc_lock, MA_OWNED) #define ADAPTER_LOCK_ASSERT_NOTOWNED(sc) mtx_assert(&(sc)->sc_lock, MA_NOTOWNED) #define PORT_LOCK(pi) mtx_lock(&(pi)->pi_lock) #define PORT_UNLOCK(pi) mtx_unlock(&(pi)->pi_lock) #define PORT_LOCK_ASSERT_OWNED(pi) mtx_assert(&(pi)->pi_lock, MA_OWNED) #define PORT_LOCK_ASSERT_NOTOWNED(pi) mtx_assert(&(pi)->pi_lock, MA_NOTOWNED) #define FL_LOCK(fl) mtx_lock(&(fl)->fl_lock) #define FL_TRYLOCK(fl) mtx_trylock(&(fl)->fl_lock) #define FL_UNLOCK(fl) mtx_unlock(&(fl)->fl_lock) #define FL_LOCK_ASSERT_OWNED(fl) mtx_assert(&(fl)->fl_lock, MA_OWNED) #define FL_LOCK_ASSERT_NOTOWNED(fl) mtx_assert(&(fl)->fl_lock, MA_NOTOWNED) #define RXQ_FL_LOCK(rxq) FL_LOCK(&(rxq)->fl) #define RXQ_FL_UNLOCK(rxq) FL_UNLOCK(&(rxq)->fl) #define RXQ_FL_LOCK_ASSERT_OWNED(rxq) FL_LOCK_ASSERT_OWNED(&(rxq)->fl) #define RXQ_FL_LOCK_ASSERT_NOTOWNED(rxq) FL_LOCK_ASSERT_NOTOWNED(&(rxq)->fl) #define EQ_LOCK(eq) mtx_lock(&(eq)->eq_lock) #define EQ_TRYLOCK(eq) mtx_trylock(&(eq)->eq_lock) #define EQ_UNLOCK(eq) mtx_unlock(&(eq)->eq_lock) #define EQ_LOCK_ASSERT_OWNED(eq) mtx_assert(&(eq)->eq_lock, MA_OWNED) #define EQ_LOCK_ASSERT_NOTOWNED(eq) mtx_assert(&(eq)->eq_lock, MA_NOTOWNED) #define TXQ_LOCK(txq) EQ_LOCK(&(txq)->eq) #define TXQ_TRYLOCK(txq) EQ_TRYLOCK(&(txq)->eq) #define TXQ_UNLOCK(txq) EQ_UNLOCK(&(txq)->eq) #define TXQ_LOCK_ASSERT_OWNED(txq) EQ_LOCK_ASSERT_OWNED(&(txq)->eq) #define TXQ_LOCK_ASSERT_NOTOWNED(txq) EQ_LOCK_ASSERT_NOTOWNED(&(txq)->eq) #define for_each_txq(pi, iter, txq) \ txq = &pi->adapter->sge.txq[pi->first_txq]; \ for (iter = 0; iter < pi->ntxq; ++iter, ++txq) #define for_each_rxq(pi, iter, rxq) \ rxq = &pi->adapter->sge.rxq[pi->first_rxq]; \ for (iter = 0; iter < pi->nrxq; ++iter, ++rxq) /* One for errors, one for firmware events */ #define T4_EXTRA_INTR 2 #define NINTRQ(sc) ((sc)->intr_count > T4_EXTRA_INTR ? \ (sc)->intr_count - T4_EXTRA_INTR : 1) static inline uint32_t t4_read_reg(struct adapter *sc, uint32_t reg) { return bus_space_read_4(sc->bt, sc->bh, reg); } static inline void t4_write_reg(struct adapter *sc, uint32_t reg, uint32_t val) { bus_space_write_4(sc->bt, sc->bh, reg, val); } static inline uint64_t t4_read_reg64(struct adapter *sc, uint32_t reg) { return t4_bus_space_read_8(sc->bt, sc->bh, reg); } static inline void t4_write_reg64(struct adapter *sc, uint32_t reg, uint64_t val) { t4_bus_space_write_8(sc->bt, sc->bh, reg, val); } static inline void t4_os_pci_read_cfg1(struct adapter *sc, int reg, uint8_t *val) { *val = pci_read_config(sc->dev, reg, 1); } static inline void t4_os_pci_write_cfg1(struct adapter *sc, int reg, uint8_t val) { pci_write_config(sc->dev, reg, val, 1); } static inline void t4_os_pci_read_cfg2(struct adapter *sc, int reg, uint16_t *val) { *val = pci_read_config(sc->dev, reg, 2); } static inline void t4_os_pci_write_cfg2(struct adapter *sc, int reg, uint16_t val) { pci_write_config(sc->dev, reg, val, 2); } static inline void t4_os_pci_read_cfg4(struct adapter *sc, int reg, uint32_t *val) { *val = pci_read_config(sc->dev, reg, 4); } static inline void t4_os_pci_write_cfg4(struct adapter *sc, int reg, uint32_t val) { pci_write_config(sc->dev, reg, val, 4); } static inline struct port_info * adap2pinfo(struct adapter *sc, int idx) { return (sc->port[idx]); } static inline void t4_os_set_hw_addr(struct adapter *sc, int idx, uint8_t hw_addr[]) { bcopy(hw_addr, sc->port[idx]->hw_addr, ETHER_ADDR_LEN); } static inline bool is_10G_port(const struct port_info *pi) { return ((pi->link_cfg.supported & FW_PORT_CAP_SPEED_10G) != 0); } /* t4_main.c */ void cxgbe_txq_start(void *, int); int t4_os_find_pci_capability(struct adapter *, int); int t4_os_pci_save_state(struct adapter *); int t4_os_pci_restore_state(struct adapter *); void t4_os_portmod_changed(const struct adapter *, int); void t4_os_link_changed(struct adapter *, int, int); /* t4_sge.c */ void t4_sge_modload(void); void t4_sge_init(struct adapter *); int t4_create_dma_tag(struct adapter *); int t4_destroy_dma_tag(struct adapter *); int t4_setup_adapter_queues(struct adapter *); int t4_teardown_adapter_queues(struct adapter *); int t4_setup_eth_queues(struct port_info *); int t4_teardown_eth_queues(struct port_info *); void t4_intr_all(void *); void t4_intr(void *); void t4_intr_err(void *); void t4_intr_evt(void *); int t4_mgmt_tx(struct adapter *, struct mbuf *); int t4_eth_tx(struct ifnet *, struct sge_txq *, struct mbuf *); void t4_update_fl_bufsize(struct ifnet *); #endif Index: projects/largeSMP/sys/dev/cxgbe/t4_main.c =================================================================== --- projects/largeSMP/sys/dev/cxgbe/t4_main.c (revision 222782) +++ projects/largeSMP/sys/dev/cxgbe/t4_main.c (revision 222783) @@ -1,3605 +1,3607 @@ /*- * Copyright (c) 2011 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 #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "common/t4_hw.h" #include "common/common.h" #include "common/t4_msg.h" #include "common/t4_regs.h" #include "common/t4_regs_values.h" #include "common/t4fw_interface.h" #include "t4_ioctl.h" #include "t4_l2t.h" /* T4 bus driver interface */ static int t4_probe(device_t); static int t4_attach(device_t); static int t4_detach(device_t); static device_method_t t4_methods[] = { DEVMETHOD(device_probe, t4_probe), DEVMETHOD(device_attach, t4_attach), DEVMETHOD(device_detach, t4_detach), /* bus interface */ DEVMETHOD(bus_print_child, bus_generic_print_child), DEVMETHOD(bus_driver_added, bus_generic_driver_added), { 0, 0 } }; static driver_t t4_driver = { "t4nex", t4_methods, sizeof(struct adapter) }; /* T4 port (cxgbe) interface */ static int cxgbe_probe(device_t); static int cxgbe_attach(device_t); static int cxgbe_detach(device_t); static device_method_t cxgbe_methods[] = { DEVMETHOD(device_probe, cxgbe_probe), DEVMETHOD(device_attach, cxgbe_attach), DEVMETHOD(device_detach, cxgbe_detach), { 0, 0 } }; static driver_t cxgbe_driver = { "cxgbe", cxgbe_methods, sizeof(struct port_info) }; static d_ioctl_t t4_ioctl; static d_open_t t4_open; static d_close_t t4_close; static struct cdevsw t4_cdevsw = { .d_version = D_VERSION, .d_flags = 0, .d_open = t4_open, .d_close = t4_close, .d_ioctl = t4_ioctl, .d_name = "t4nex", }; /* ifnet + media interface */ static void cxgbe_init(void *); static int cxgbe_ioctl(struct ifnet *, unsigned long, caddr_t); static void cxgbe_start(struct ifnet *); static int cxgbe_transmit(struct ifnet *, struct mbuf *); static void cxgbe_qflush(struct ifnet *); static int cxgbe_media_change(struct ifnet *); static void cxgbe_media_status(struct ifnet *, struct ifmediareq *); MALLOC_DEFINE(M_CXGBE, "cxgbe", "Chelsio T4 Ethernet driver and services"); /* * Tunables. */ SYSCTL_NODE(_hw, OID_AUTO, cxgbe, CTLFLAG_RD, 0, "cxgbe driver parameters"); static int force_firmware_install = 0; TUNABLE_INT("hw.cxgbe.force_firmware_install", &force_firmware_install); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, force_firmware_install, CTLFLAG_RDTUN, &force_firmware_install, 0, "install firmware on every attach."); /* * Holdoff timer and packet counter values. */ static unsigned int intr_timer[SGE_NTIMERS] = {1, 5, 10, 50, 100, 200}; static unsigned int intr_pktcount[SGE_NCOUNTERS] = {1, 8, 16, 32}; /* 63 max */ /* * Max # of tx and rx queues to use for each 10G and 1G port. */ static unsigned int max_ntxq_10g = 8; TUNABLE_INT("hw.cxgbe.max_ntxq_10G_port", &max_ntxq_10g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, max_ntxq_10G_port, CTLFLAG_RDTUN, &max_ntxq_10g, 0, "maximum number of tx queues per 10G port."); static unsigned int max_nrxq_10g = 8; TUNABLE_INT("hw.cxgbe.max_nrxq_10G_port", &max_nrxq_10g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, max_nrxq_10G_port, CTLFLAG_RDTUN, &max_nrxq_10g, 0, "maximum number of rxq's (per 10G port)."); static unsigned int max_ntxq_1g = 2; TUNABLE_INT("hw.cxgbe.max_ntxq_1G_port", &max_ntxq_1g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, max_ntxq_1G_port, CTLFLAG_RDTUN, &max_ntxq_1g, 0, "maximum number of tx queues per 1G port."); static unsigned int max_nrxq_1g = 2; TUNABLE_INT("hw.cxgbe.max_nrxq_1G_port", &max_nrxq_1g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, max_nrxq_1G_port, CTLFLAG_RDTUN, &max_nrxq_1g, 0, "maximum number of rxq's (per 1G port)."); /* * Holdoff parameters for 10G and 1G ports. */ static unsigned int tmr_idx_10g = 1; TUNABLE_INT("hw.cxgbe.holdoff_timer_idx_10G", &tmr_idx_10g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, holdoff_timer_idx_10G, CTLFLAG_RDTUN, &tmr_idx_10g, 0, "default timer index for interrupt holdoff (10G ports)."); static int pktc_idx_10g = 2; TUNABLE_INT("hw.cxgbe.holdoff_pktc_idx_10G", &pktc_idx_10g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, holdoff_pktc_idx_10G, CTLFLAG_RDTUN, &pktc_idx_10g, 0, "default pkt counter index for interrupt holdoff (10G ports)."); static unsigned int tmr_idx_1g = 1; TUNABLE_INT("hw.cxgbe.holdoff_timer_idx_1G", &tmr_idx_1g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, holdoff_timer_idx_1G, CTLFLAG_RDTUN, &tmr_idx_1g, 0, "default timer index for interrupt holdoff (1G ports)."); static int pktc_idx_1g = 2; TUNABLE_INT("hw.cxgbe.holdoff_pktc_idx_1G", &pktc_idx_1g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, holdoff_pktc_idx_1G, CTLFLAG_RDTUN, &pktc_idx_1g, 0, "default pkt counter index for interrupt holdoff (1G ports)."); /* * Size (# of entries) of each tx and rx queue. */ static unsigned int qsize_txq = TX_EQ_QSIZE; TUNABLE_INT("hw.cxgbe.qsize_txq", &qsize_txq); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, qsize_txq, CTLFLAG_RDTUN, &qsize_txq, 0, "default queue size of NIC tx queues."); static unsigned int qsize_rxq = RX_IQ_QSIZE; TUNABLE_INT("hw.cxgbe.qsize_rxq", &qsize_rxq); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, qsize_rxq, CTLFLAG_RDTUN, &qsize_rxq, 0, "default queue size of NIC rx queues."); /* * Interrupt types allowed. */ static int intr_types = INTR_MSIX | INTR_MSI | INTR_INTX; TUNABLE_INT("hw.cxgbe.interrupt_types", &intr_types); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, interrupt_types, CTLFLAG_RDTUN, &intr_types, 0, "interrupt types allowed (bits 0, 1, 2 = INTx, MSI, MSI-X respectively)"); /* * Force the driver to use the same set of interrupts for all ports. */ static int intr_shared = 0; TUNABLE_INT("hw.cxgbe.interrupts_shared", &intr_shared); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, interrupts_shared, CTLFLAG_RDTUN, &intr_shared, 0, "interrupts shared between all ports"); static unsigned int filter_mode = HW_TPL_FR_MT_PR_IV_P_FC; TUNABLE_INT("hw.cxgbe.filter_mode", &filter_mode); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, filter_mode, CTLFLAG_RDTUN, &filter_mode, 0, "default global filter mode."); struct intrs_and_queues { int intr_type; /* INTx, MSI, or MSI-X */ int nirq; /* Number of vectors */ int intr_shared; /* Interrupts shared between all ports */ int ntxq10g; /* # of NIC txq's for each 10G port */ int nrxq10g; /* # of NIC rxq's for each 10G port */ int ntxq1g; /* # of NIC txq's for each 1G port */ int nrxq1g; /* # of NIC rxq's for each 1G port */ }; struct filter_entry { uint32_t valid:1; /* filter allocated and valid */ uint32_t locked:1; /* filter is administratively locked */ uint32_t pending:1; /* filter action is pending firmware reply */ uint32_t smtidx:8; /* Source MAC Table index for smac */ struct l2t_entry *l2t; /* Layer Two Table entry for dmac */ struct t4_filter_specification fs; }; enum { MEMWIN0_APERTURE = 2048, MEMWIN0_BASE = 0x1b800, MEMWIN1_APERTURE = 32768, MEMWIN1_BASE = 0x28000, MEMWIN2_APERTURE = 65536, MEMWIN2_BASE = 0x30000, }; enum { XGMAC_MTU = (1 << 0), XGMAC_PROMISC = (1 << 1), XGMAC_ALLMULTI = (1 << 2), XGMAC_VLANEX = (1 << 3), XGMAC_UCADDR = (1 << 4), XGMAC_MCADDRS = (1 << 5), XGMAC_ALL = 0xffff }; static int map_bars(struct adapter *); static void setup_memwin(struct adapter *); static int cfg_itype_and_nqueues(struct adapter *, int, int, struct intrs_and_queues *); static int prep_firmware(struct adapter *); static int get_devlog_params(struct adapter *, struct devlog_params *); static int get_capabilities(struct adapter *, struct fw_caps_config_cmd *); static int get_params(struct adapter *, struct fw_caps_config_cmd *); static void t4_set_desc(struct adapter *); static void build_medialist(struct port_info *); static int update_mac_settings(struct port_info *, int); static int cxgbe_init_locked(struct port_info *); static int cxgbe_init_synchronized(struct port_info *); static int cxgbe_uninit_locked(struct port_info *); static int cxgbe_uninit_synchronized(struct port_info *); static int first_port_up(struct adapter *); static int last_port_down(struct adapter *); static int t4_alloc_irq(struct adapter *, struct irq *, int rid, iq_intr_handler_t *, void *, char *); static int t4_free_irq(struct adapter *, struct irq *); static void reg_block_dump(struct adapter *, uint8_t *, unsigned int, unsigned int); static void t4_get_regs(struct adapter *, struct t4_regdump *, uint8_t *); static void cxgbe_tick(void *); static int t4_sysctls(struct adapter *); static int cxgbe_sysctls(struct port_info *); static int sysctl_int_array(SYSCTL_HANDLER_ARGS); static int sysctl_holdoff_tmr_idx(SYSCTL_HANDLER_ARGS); static int sysctl_holdoff_pktc_idx(SYSCTL_HANDLER_ARGS); static int sysctl_qsize_rxq(SYSCTL_HANDLER_ARGS); static int sysctl_qsize_txq(SYSCTL_HANDLER_ARGS); static int sysctl_handle_t4_reg64(SYSCTL_HANDLER_ARGS); static int sysctl_devlog(SYSCTL_HANDLER_ARGS); static inline void txq_start(struct ifnet *, struct sge_txq *); static uint32_t fconf_to_mode(uint32_t); static uint32_t mode_to_fconf(uint32_t); static uint32_t fspec_to_fconf(struct t4_filter_specification *); static int get_filter_mode(struct adapter *, uint32_t *); static int set_filter_mode(struct adapter *, uint32_t); static inline uint64_t get_filter_hits(struct adapter *, uint32_t); static int get_filter(struct adapter *, struct t4_filter *); static int set_filter(struct adapter *, struct t4_filter *); static int del_filter(struct adapter *, struct t4_filter *); static void clear_filter(struct filter_entry *); static int set_filter_wr(struct adapter *, int); static int del_filter_wr(struct adapter *, int); void filter_rpl(struct adapter *, const struct cpl_set_tcb_rpl *); static int t4_mod_event(module_t, int, void *); struct t4_pciids { uint16_t device; uint8_t mpf; char *desc; } t4_pciids[] = { {0xa000, 0, "Chelsio Terminator 4 FPGA"}, {0x4400, 4, "Chelsio T440-dbg"}, {0x4401, 4, "Chelsio T420-CR"}, {0x4402, 4, "Chelsio T422-CR"}, {0x4403, 4, "Chelsio T440-CR"}, {0x4404, 4, "Chelsio T420-BCH"}, {0x4405, 4, "Chelsio T440-BCH"}, {0x4406, 4, "Chelsio T440-CH"}, {0x4407, 4, "Chelsio T420-SO"}, {0x4408, 4, "Chelsio T420-CX"}, {0x4409, 4, "Chelsio T420-BT"}, {0x440a, 4, "Chelsio T404-BT"}, }; static int t4_probe(device_t dev) { int i; uint16_t v = pci_get_vendor(dev); uint16_t d = pci_get_device(dev); if (v != PCI_VENDOR_ID_CHELSIO) return (ENXIO); for (i = 0; i < ARRAY_SIZE(t4_pciids); i++) { if (d == t4_pciids[i].device && pci_get_function(dev) == t4_pciids[i].mpf) { device_set_desc(dev, t4_pciids[i].desc); return (BUS_PROBE_DEFAULT); } } return (ENXIO); } static int t4_attach(device_t dev) { struct adapter *sc; int rc = 0, i, n10g, n1g, rqidx, tqidx; struct fw_caps_config_cmd caps; uint32_t p, v; struct intrs_and_queues iaq; struct sge *s; sc = device_get_softc(dev); sc->dev = dev; sc->pf = pci_get_function(dev); sc->mbox = sc->pf; pci_enable_busmaster(dev); if (pci_find_cap(dev, PCIY_EXPRESS, &i) == 0) { pci_set_max_read_req(dev, 4096); v = pci_read_config(dev, i + PCIR_EXPRESS_DEVICE_CTL, 2); v |= PCIM_EXP_CTL_RELAXED_ORD_ENABLE; pci_write_config(dev, i + PCIR_EXPRESS_DEVICE_CTL, v, 2); } snprintf(sc->lockname, sizeof(sc->lockname), "%s", device_get_nameunit(dev)); mtx_init(&sc->sc_lock, sc->lockname, 0, MTX_DEF); rc = map_bars(sc); if (rc != 0) goto done; /* error message displayed already */ memset(sc->chan_map, 0xff, sizeof(sc->chan_map)); /* Prepare the adapter for operation */ rc = -t4_prep_adapter(sc); if (rc != 0) { device_printf(dev, "failed to prepare adapter: %d.\n", rc); goto done; } /* Do this really early */ sc->cdev = make_dev(&t4_cdevsw, device_get_unit(dev), UID_ROOT, GID_WHEEL, 0600, "%s", device_get_nameunit(dev)); sc->cdev->si_drv1 = sc; /* Prepare the firmware for operation */ rc = prep_firmware(sc); if (rc != 0) goto done; /* error message displayed already */ /* Read firmware devlog parameters */ (void) get_devlog_params(sc, &sc->params.devlog); /* Get device capabilities and select which ones we'll use */ rc = get_capabilities(sc, &caps); if (rc != 0) { device_printf(dev, "failed to initialize adapter capabilities: %d.\n", rc); goto done; } /* Choose the global RSS mode. */ rc = -t4_config_glbl_rss(sc, sc->mbox, FW_RSS_GLB_CONFIG_CMD_MODE_BASICVIRTUAL, F_FW_RSS_GLB_CONFIG_CMD_TNLMAPEN | F_FW_RSS_GLB_CONFIG_CMD_HASHTOEPLITZ | F_FW_RSS_GLB_CONFIG_CMD_TNLALLLKP); if (rc != 0) { device_printf(dev, "failed to select global RSS mode: %d.\n", rc); goto done; } /* These are total (sum of all ports) limits for a bus driver */ rc = -t4_cfg_pfvf(sc, sc->mbox, sc->pf, 0, 128, /* max # of egress queues */ 64, /* max # of egress Ethernet or control queues */ 64, /* max # of ingress queues with fl/interrupt */ 0, /* max # of ingress queues without interrupt */ 0, /* PCIe traffic class */ 4, /* max # of virtual interfaces */ M_FW_PFVF_CMD_CMASK, M_FW_PFVF_CMD_PMASK, 16, FW_CMD_CAP_PF, FW_CMD_CAP_PF); if (rc != 0) { device_printf(dev, "failed to configure pf/vf resources: %d.\n", rc); goto done; } /* Need this before sge_init */ for (i = 0; i < SGE_NTIMERS; i++) sc->sge.timer_val[i] = min(intr_timer[i], 200U); for (i = 0; i < SGE_NCOUNTERS; i++) sc->sge.counter_val[i] = min(intr_pktcount[i], M_THRESHOLD_0); /* Also need the cooked value of cclk before sge_init */ p = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_DEV) | V_FW_PARAMS_PARAM_X(FW_PARAMS_PARAM_DEV_CCLK)); rc = -t4_query_params(sc, sc->mbox, sc->pf, 0, 1, &p, &v); if (rc != 0) { device_printf(sc->dev, "failed to obtain core clock value: %d.\n", rc); goto done; } sc->params.vpd.cclk = v; t4_sge_init(sc); t4_set_filter_mode(sc, filter_mode); t4_set_reg_field(sc, A_TP_GLOBAL_CONFIG, V_FIVETUPLELOOKUP(M_FIVETUPLELOOKUP), V_FIVETUPLELOOKUP(M_FIVETUPLELOOKUP)); t4_tp_wr_bits_indirect(sc, A_TP_INGRESS_CONFIG, F_CSUM_HAS_PSEUDO_HDR, F_LOOKUPEVERYPKT); /* get basic stuff going */ rc = -t4_early_init(sc, sc->mbox); if (rc != 0) { device_printf(dev, "early init failed: %d.\n", rc); goto done; } rc = get_params(sc, &caps); if (rc != 0) goto done; /* error message displayed already */ /* These are finalized by FW initialization, load their values now */ v = t4_read_reg(sc, A_TP_TIMER_RESOLUTION); sc->params.tp.tre = G_TIMERRESOLUTION(v); sc->params.tp.dack_re = G_DELAYEDACKRESOLUTION(v); t4_read_mtu_tbl(sc, sc->params.mtus, NULL); /* tweak some settings */ t4_write_reg(sc, A_TP_SHIFT_CNT, V_SYNSHIFTMAX(6) | V_RXTSHIFTMAXR1(4) | V_RXTSHIFTMAXR2(15) | V_PERSHIFTBACKOFFMAX(8) | V_PERSHIFTMAX(8) | V_KEEPALIVEMAXR1(4) | V_KEEPALIVEMAXR2(9)); t4_write_reg(sc, A_ULP_RX_TDDP_PSZ, V_HPZ0(PAGE_SHIFT - 12)); + t4_set_reg_field(sc, A_TP_PARA_REG3, F_TUNNELCNGDROP0 | + F_TUNNELCNGDROP1 | F_TUNNELCNGDROP2 | F_TUNNELCNGDROP3, 0); setup_memwin(sc); rc = t4_create_dma_tag(sc); if (rc != 0) goto done; /* error message displayed already */ /* * First pass over all the ports - allocate VIs and initialize some * basic parameters like mac address, port type, etc. We also figure * out whether a port is 10G or 1G and use that information when * calculating how many interrupts to attempt to allocate. */ n10g = n1g = 0; for_each_port(sc, i) { struct port_info *pi; pi = malloc(sizeof(*pi), M_CXGBE, M_ZERO | M_WAITOK); sc->port[i] = pi; /* These must be set before t4_port_init */ pi->adapter = sc; pi->port_id = i; /* Allocate the vi and initialize parameters like mac addr */ rc = -t4_port_init(pi, sc->mbox, sc->pf, 0); if (rc != 0) { device_printf(dev, "unable to initialize port %d: %d\n", i, rc); free(pi, M_CXGBE); sc->port[i] = NULL; goto done; } snprintf(pi->lockname, sizeof(pi->lockname), "%sp%d", device_get_nameunit(dev), i); mtx_init(&pi->pi_lock, pi->lockname, 0, MTX_DEF); if (is_10G_port(pi)) { n10g++; pi->tmr_idx = tmr_idx_10g; pi->pktc_idx = pktc_idx_10g; } else { n1g++; pi->tmr_idx = tmr_idx_1g; pi->pktc_idx = pktc_idx_1g; } pi->xact_addr_filt = -1; pi->qsize_rxq = max(qsize_rxq, 128); while (pi->qsize_rxq & 7) pi->qsize_rxq++; pi->qsize_txq = max(qsize_txq, 128); if (pi->qsize_rxq != qsize_rxq) { device_printf(dev, "using %d instead of %d as the rx queue size.\n", pi->qsize_rxq, qsize_rxq); } if (pi->qsize_txq != qsize_txq) { device_printf(dev, "using %d instead of %d as the tx queue size.\n", pi->qsize_txq, qsize_txq); } pi->dev = device_add_child(dev, "cxgbe", -1); if (pi->dev == NULL) { device_printf(dev, "failed to add device for port %d.\n", i); rc = ENXIO; goto done; } device_set_softc(pi->dev, pi); setbit(&sc->registered_device_map, i); } if (sc->registered_device_map == 0) { device_printf(dev, "no usable ports\n"); rc = ENXIO; goto done; } /* * Interrupt type, # of interrupts, # of rx/tx queues, etc. */ rc = cfg_itype_and_nqueues(sc, n10g, n1g, &iaq); if (rc != 0) goto done; /* error message displayed already */ sc->intr_type = iaq.intr_type; sc->intr_count = iaq.nirq; s = &sc->sge; s->nrxq = n10g * iaq.nrxq10g + n1g * iaq.nrxq1g; s->ntxq = n10g * iaq.ntxq10g + n1g * iaq.ntxq1g; s->neq = s->ntxq + s->nrxq; /* the free list in an rxq is an eq */ s->neq += sc->params.nports; /* control queues, 1 per port */ s->niq = s->nrxq + 1; /* 1 extra for firmware event queue */ if (iaq.intr_shared) sc->flags |= INTR_SHARED; s->niq += NINTRQ(sc); /* interrupt queues */ s->intrq = malloc(NINTRQ(sc) * sizeof(struct sge_iq), M_CXGBE, M_ZERO | M_WAITOK); s->ctrlq = malloc(sc->params.nports * sizeof(struct sge_ctrlq), M_CXGBE, M_ZERO | M_WAITOK); s->rxq = malloc(s->nrxq * sizeof(struct sge_rxq), M_CXGBE, M_ZERO | M_WAITOK); s->txq = malloc(s->ntxq * sizeof(struct sge_txq), M_CXGBE, M_ZERO | M_WAITOK); s->iqmap = malloc(s->niq * sizeof(struct sge_iq *), M_CXGBE, M_ZERO | M_WAITOK); s->eqmap = malloc(s->neq * sizeof(struct sge_eq *), M_CXGBE, M_ZERO | M_WAITOK); sc->irq = malloc(sc->intr_count * sizeof(struct irq), M_CXGBE, M_ZERO | M_WAITOK); sc->l2t = t4_init_l2t(M_WAITOK); t4_sysctls(sc); /* * Second pass over the ports. This time we know the number of rx and * tx queues that each port should get. */ rqidx = tqidx = 0; for_each_port(sc, i) { struct port_info *pi = sc->port[i]; if (pi == NULL) continue; pi->first_rxq = rqidx; pi->nrxq = is_10G_port(pi) ? iaq.nrxq10g : iaq.nrxq1g; pi->first_txq = tqidx; pi->ntxq = is_10G_port(pi) ? iaq.ntxq10g : iaq.ntxq1g; rqidx += pi->nrxq; tqidx += pi->ntxq; } rc = bus_generic_attach(dev); if (rc != 0) { device_printf(dev, "failed to attach all child ports: %d\n", rc); goto done; } #ifdef INVARIANTS device_printf(dev, "%p, %d ports (0x%x), %d intr_type, %d intr_count\n", sc, sc->params.nports, sc->params.portvec, sc->intr_type, sc->intr_count); #endif t4_set_desc(sc); done: if (rc != 0) t4_detach(dev); return (rc); } /* * Idempotent */ static int t4_detach(device_t dev) { struct adapter *sc; struct port_info *pi; int i; sc = device_get_softc(dev); if (sc->cdev) destroy_dev(sc->cdev); bus_generic_detach(dev); for (i = 0; i < MAX_NPORTS; i++) { pi = sc->port[i]; if (pi) { t4_free_vi(pi->adapter, sc->mbox, sc->pf, 0, pi->viid); if (pi->dev) device_delete_child(dev, pi->dev); mtx_destroy(&pi->pi_lock); free(pi, M_CXGBE); } } if (sc->flags & FW_OK) t4_fw_bye(sc, sc->mbox); if (sc->intr_type == INTR_MSI || sc->intr_type == INTR_MSIX) pci_release_msi(dev); if (sc->regs_res) bus_release_resource(dev, SYS_RES_MEMORY, sc->regs_rid, sc->regs_res); if (sc->msix_res) bus_release_resource(dev, SYS_RES_MEMORY, sc->msix_rid, sc->msix_res); if (sc->l2t) t4_free_l2t(sc->l2t); free(sc->irq, M_CXGBE); free(sc->sge.rxq, M_CXGBE); free(sc->sge.txq, M_CXGBE); free(sc->sge.ctrlq, M_CXGBE); free(sc->sge.intrq, M_CXGBE); free(sc->sge.iqmap, M_CXGBE); free(sc->sge.eqmap, M_CXGBE); free(sc->tids.ftid_tab, M_CXGBE); t4_destroy_dma_tag(sc); mtx_destroy(&sc->sc_lock); bzero(sc, sizeof(*sc)); return (0); } static int cxgbe_probe(device_t dev) { char buf[128]; struct port_info *pi = device_get_softc(dev); snprintf(buf, sizeof(buf), "Port %d", pi->port_id); device_set_desc_copy(dev, buf); return (BUS_PROBE_DEFAULT); } #define T4_CAP (IFCAP_VLAN_HWTAGGING | IFCAP_VLAN_MTU | IFCAP_HWCSUM | \ IFCAP_VLAN_HWCSUM | IFCAP_TSO | IFCAP_JUMBO_MTU | IFCAP_LRO | \ IFCAP_VLAN_HWTSO) #define T4_CAP_ENABLE (T4_CAP & ~IFCAP_TSO6) static int cxgbe_attach(device_t dev) { struct port_info *pi = device_get_softc(dev); struct ifnet *ifp; /* Allocate an ifnet and set it up */ ifp = if_alloc(IFT_ETHER); if (ifp == NULL) { device_printf(dev, "Cannot allocate ifnet\n"); return (ENOMEM); } pi->ifp = ifp; ifp->if_softc = pi; callout_init(&pi->tick, CALLOUT_MPSAFE); pi->tq = taskqueue_create("cxgbe_taskq", M_NOWAIT, taskqueue_thread_enqueue, &pi->tq); if (pi->tq == NULL) { device_printf(dev, "failed to allocate port task queue\n"); if_free(pi->ifp); return (ENOMEM); } taskqueue_start_threads(&pi->tq, 1, PI_NET, "%s taskq", device_get_nameunit(dev)); if_initname(ifp, device_get_name(dev), device_get_unit(dev)); ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST; ifp->if_init = cxgbe_init; ifp->if_ioctl = cxgbe_ioctl; ifp->if_start = cxgbe_start; ifp->if_transmit = cxgbe_transmit; ifp->if_qflush = cxgbe_qflush; ifp->if_snd.ifq_drv_maxlen = 1024; IFQ_SET_MAXLEN(&ifp->if_snd, ifp->if_snd.ifq_drv_maxlen); IFQ_SET_READY(&ifp->if_snd); ifp->if_capabilities = T4_CAP; ifp->if_capenable = T4_CAP_ENABLE; ifp->if_hwassist = CSUM_TCP | CSUM_UDP | CSUM_IP | CSUM_TSO; /* Initialize ifmedia for this port */ ifmedia_init(&pi->media, IFM_IMASK, cxgbe_media_change, cxgbe_media_status); build_medialist(pi); ether_ifattach(ifp, pi->hw_addr); #ifdef INVARIANTS device_printf(dev, "%p, %d txq, %d rxq\n", pi, pi->ntxq, pi->nrxq); #endif cxgbe_sysctls(pi); return (0); } static int cxgbe_detach(device_t dev) { struct port_info *pi = device_get_softc(dev); struct adapter *sc = pi->adapter; int rc; /* Tell if_ioctl and if_init that the port is going away */ ADAPTER_LOCK(sc); SET_DOOMED(pi); wakeup(&sc->flags); while (IS_BUSY(sc)) mtx_sleep(&sc->flags, &sc->sc_lock, 0, "t4detach", 0); SET_BUSY(sc); ADAPTER_UNLOCK(sc); rc = cxgbe_uninit_synchronized(pi); if (rc != 0) device_printf(dev, "port uninit failed: %d.\n", rc); taskqueue_free(pi->tq); ifmedia_removeall(&pi->media); ether_ifdetach(pi->ifp); if_free(pi->ifp); ADAPTER_LOCK(sc); CLR_BUSY(sc); wakeup_one(&sc->flags); ADAPTER_UNLOCK(sc); return (0); } static void cxgbe_init(void *arg) { struct port_info *pi = arg; struct adapter *sc = pi->adapter; ADAPTER_LOCK(sc); cxgbe_init_locked(pi); /* releases adapter lock */ ADAPTER_LOCK_ASSERT_NOTOWNED(sc); } static int cxgbe_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; switch (cmd) { case SIOCSIFMTU: ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc) { fail: ADAPTER_UNLOCK(sc); return (rc); } mtu = ifr->ifr_mtu; if ((mtu < ETHERMIN) || (mtu > ETHERMTU_JUMBO)) { rc = EINVAL; } else { ifp->if_mtu = mtu; if (ifp->if_drv_flags & IFF_DRV_RUNNING) { t4_update_fl_bufsize(ifp); PORT_LOCK(pi); rc = update_mac_settings(pi, XGMAC_MTU); PORT_UNLOCK(pi); } } ADAPTER_UNLOCK(sc); break; case SIOCSIFFLAGS: ADAPTER_LOCK(sc); if (IS_DOOMED(pi)) { rc = ENXIO; goto fail; } if (ifp->if_flags & IFF_UP) { if (ifp->if_drv_flags & IFF_DRV_RUNNING) { flags = pi->if_flags; if ((ifp->if_flags ^ flags) & (IFF_PROMISC | IFF_ALLMULTI)) { if (IS_BUSY(sc)) { rc = EBUSY; goto fail; } PORT_LOCK(pi); rc = update_mac_settings(pi, XGMAC_PROMISC | XGMAC_ALLMULTI); PORT_UNLOCK(pi); } ADAPTER_UNLOCK(sc); } else rc = cxgbe_init_locked(pi); pi->if_flags = ifp->if_flags; } else if (ifp->if_drv_flags & IFF_DRV_RUNNING) rc = cxgbe_uninit_locked(pi); else ADAPTER_UNLOCK(sc); ADAPTER_LOCK_ASSERT_NOTOWNED(sc); break; case SIOCADDMULTI: case SIOCDELMULTI: /* these two can be called with a mutex held :-( */ ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc) goto fail; if (ifp->if_drv_flags & IFF_DRV_RUNNING) { PORT_LOCK(pi); rc = update_mac_settings(pi, XGMAC_MCADDRS); PORT_UNLOCK(pi); } ADAPTER_UNLOCK(sc); break; case SIOCSIFCAP: ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc) goto fail; 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 (IFCAP_TSO & ifp->if_capenable && !(IFCAP_TXCSUM & ifp->if_capenable)) { ifp->if_capenable &= ~IFCAP_TSO; ifp->if_hwassist &= ~CSUM_TSO; if_printf(ifp, "tso disabled due to -txcsum.\n"); } } if (mask & IFCAP_RXCSUM) ifp->if_capenable ^= IFCAP_RXCSUM; if (mask & IFCAP_TSO4) { ifp->if_capenable ^= IFCAP_TSO4; if (IFCAP_TSO & ifp->if_capenable) { if (IFCAP_TXCSUM & ifp->if_capenable) ifp->if_hwassist |= CSUM_TSO; else { ifp->if_capenable &= ~IFCAP_TSO; ifp->if_hwassist &= ~CSUM_TSO; if_printf(ifp, "enable txcsum first.\n"); rc = EAGAIN; } } else ifp->if_hwassist &= ~CSUM_TSO; } if (mask & IFCAP_LRO) { #ifdef INET int i; struct sge_rxq *rxq; ifp->if_capenable ^= IFCAP_LRO; for_each_rxq(pi, i, rxq) { if (ifp->if_capenable & IFCAP_LRO) rxq->flags |= RXQ_LRO_ENABLED; else rxq->flags &= ~RXQ_LRO_ENABLED; } #endif } #ifndef TCP_OFFLOAD_DISABLE if (mask & IFCAP_TOE4) { rc = EOPNOTSUPP; } #endif if (mask & IFCAP_VLAN_HWTAGGING) { ifp->if_capenable ^= IFCAP_VLAN_HWTAGGING; if (ifp->if_drv_flags & IFF_DRV_RUNNING) { PORT_LOCK(pi); rc = update_mac_settings(pi, XGMAC_VLANEX); PORT_UNLOCK(pi); } } if (mask & IFCAP_VLAN_MTU) { ifp->if_capenable ^= IFCAP_VLAN_MTU; /* Need to find out how to disable auto-mtu-inflation */ } if (mask & IFCAP_VLAN_HWTSO) ifp->if_capenable ^= IFCAP_VLAN_HWTSO; if (mask & IFCAP_VLAN_HWCSUM) ifp->if_capenable ^= IFCAP_VLAN_HWCSUM; #ifdef VLAN_CAPABILITIES VLAN_CAPABILITIES(ifp); #endif ADAPTER_UNLOCK(sc); break; case SIOCSIFMEDIA: case SIOCGIFMEDIA: ifmedia_ioctl(ifp, ifr, &pi->media, cmd); break; default: rc = ether_ioctl(ifp, cmd, data); } return (rc); } static void cxgbe_start(struct ifnet *ifp) { struct port_info *pi = ifp->if_softc; struct sge_txq *txq; int i; for_each_txq(pi, i, txq) { if (TXQ_TRYLOCK(txq)) { txq_start(ifp, txq); TXQ_UNLOCK(txq); } } } static int cxgbe_transmit(struct ifnet *ifp, struct mbuf *m) { struct port_info *pi = ifp->if_softc; struct adapter *sc = pi->adapter; struct sge_txq *txq = &sc->sge.txq[pi->first_txq]; struct buf_ring *br; int rc; M_ASSERTPKTHDR(m); if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { m_freem(m); return (0); } if (m->m_flags & M_FLOWID) txq += (m->m_pkthdr.flowid % pi->ntxq); br = txq->br; if (TXQ_TRYLOCK(txq) == 0) { /* * XXX: make sure that this packet really is sent out. There is * a small race where t4_eth_tx may stop draining the drbr and * goes away, just before we enqueued this mbuf. */ return (drbr_enqueue(ifp, br, m)); } /* * txq->m is the mbuf that is held up due to a temporary shortage of * resources and it should be put on the wire first. Then what's in * drbr and finally the mbuf that was just passed in to us. * * Return code should indicate the fate of the mbuf that was passed in * this time. */ TXQ_LOCK_ASSERT_OWNED(txq); if (drbr_needs_enqueue(ifp, br) || txq->m) { /* Queued for transmission. */ rc = drbr_enqueue(ifp, br, m); m = txq->m ? txq->m : drbr_dequeue(ifp, br); (void) t4_eth_tx(ifp, txq, m); TXQ_UNLOCK(txq); return (rc); } /* Direct transmission. */ rc = t4_eth_tx(ifp, txq, m); if (rc != 0 && txq->m) rc = 0; /* held, will be transmitted soon (hopefully) */ TXQ_UNLOCK(txq); return (rc); } static void cxgbe_qflush(struct ifnet *ifp) { struct port_info *pi = ifp->if_softc; struct sge_txq *txq; int i; struct mbuf *m; /* queues do not exist if !IFF_DRV_RUNNING. */ if (ifp->if_drv_flags & IFF_DRV_RUNNING) { for_each_txq(pi, i, txq) { TXQ_LOCK(txq); m_freem(txq->m); while ((m = buf_ring_dequeue_sc(txq->br)) != NULL) m_freem(m); TXQ_UNLOCK(txq); } } if_qflush(ifp); } static int cxgbe_media_change(struct ifnet *ifp) { struct port_info *pi = ifp->if_softc; device_printf(pi->dev, "%s unimplemented.\n", __func__); return (EOPNOTSUPP); } static void cxgbe_media_status(struct ifnet *ifp, struct ifmediareq *ifmr) { struct port_info *pi = ifp->if_softc; struct ifmedia_entry *cur = pi->media.ifm_cur; int speed = pi->link_cfg.speed; int data = (pi->port_type << 8) | pi->mod_type; if (cur->ifm_data != data) { build_medialist(pi); cur = pi->media.ifm_cur; } ifmr->ifm_status = IFM_AVALID; if (!pi->link_cfg.link_ok) return; ifmr->ifm_status |= IFM_ACTIVE; /* active and current will differ iff current media is autoselect. */ if (IFM_SUBTYPE(cur->ifm_media) != IFM_AUTO) return; ifmr->ifm_active = IFM_ETHER | IFM_FDX; if (speed == SPEED_10000) ifmr->ifm_active |= IFM_10G_T; else if (speed == SPEED_1000) ifmr->ifm_active |= IFM_1000_T; else if (speed == SPEED_100) ifmr->ifm_active |= IFM_100_TX; else if (speed == SPEED_10) ifmr->ifm_active |= IFM_10_T; else KASSERT(0, ("%s: link up but speed unknown (%u)", __func__, speed)); } void t4_fatal_err(struct adapter *sc) { t4_set_reg_field(sc, A_SGE_CONTROL, F_GLOBALENABLE, 0); t4_intr_disable(sc); log(LOG_EMERG, "%s: encountered fatal error, adapter stopped.\n", device_get_nameunit(sc->dev)); } static int map_bars(struct adapter *sc) { sc->regs_rid = PCIR_BAR(0); sc->regs_res = bus_alloc_resource_any(sc->dev, SYS_RES_MEMORY, &sc->regs_rid, RF_ACTIVE); if (sc->regs_res == NULL) { device_printf(sc->dev, "cannot map registers.\n"); return (ENXIO); } sc->bt = rman_get_bustag(sc->regs_res); sc->bh = rman_get_bushandle(sc->regs_res); sc->mmio_len = rman_get_size(sc->regs_res); sc->msix_rid = PCIR_BAR(4); sc->msix_res = bus_alloc_resource_any(sc->dev, SYS_RES_MEMORY, &sc->msix_rid, RF_ACTIVE); if (sc->msix_res == NULL) { device_printf(sc->dev, "cannot map MSI-X BAR.\n"); return (ENXIO); } return (0); } static void setup_memwin(struct adapter *sc) { u_long bar0; bar0 = rman_get_start(sc->regs_res); t4_write_reg(sc, PCIE_MEM_ACCESS_REG(A_PCIE_MEM_ACCESS_BASE_WIN, 0), (bar0 + MEMWIN0_BASE) | V_BIR(0) | V_WINDOW(ilog2(MEMWIN0_APERTURE) - 10)); t4_write_reg(sc, PCIE_MEM_ACCESS_REG(A_PCIE_MEM_ACCESS_BASE_WIN, 1), (bar0 + MEMWIN1_BASE) | V_BIR(0) | V_WINDOW(ilog2(MEMWIN1_APERTURE) - 10)); t4_write_reg(sc, PCIE_MEM_ACCESS_REG(A_PCIE_MEM_ACCESS_BASE_WIN, 2), (bar0 + MEMWIN2_BASE) | V_BIR(0) | V_WINDOW(ilog2(MEMWIN2_APERTURE) - 10)); } static int cfg_itype_and_nqueues(struct adapter *sc, int n10g, int n1g, struct intrs_and_queues *iaq) { int rc, itype, navail, nc, nrxq10g, nrxq1g; bzero(iaq, sizeof(*iaq)); nc = mp_ncpus; /* our snapshot of the number of CPUs */ for (itype = INTR_MSIX; itype; itype >>= 1) { if ((itype & intr_types) == 0) continue; /* not allowed */ if (itype == INTR_MSIX) navail = pci_msix_count(sc->dev); else if (itype == INTR_MSI) navail = pci_msi_count(sc->dev); else navail = 1; if (navail == 0) continue; iaq->intr_type = itype; iaq->ntxq10g = min(nc, max_ntxq_10g); iaq->ntxq1g = min(nc, max_ntxq_1g); nrxq10g = min(nc, max_nrxq_10g); nrxq1g = min(nc, max_nrxq_1g); iaq->nirq = n10g * nrxq10g + n1g * nrxq1g + T4_EXTRA_INTR; if (iaq->nirq <= navail && intr_shared == 0) { if (itype == INTR_MSI && !powerof2(iaq->nirq)) goto share; /* One for err, one for fwq, and one for each rxq */ iaq->intr_shared = 0; iaq->nrxq10g = nrxq10g; iaq->nrxq1g = nrxq1g; } else { share: iaq->intr_shared = 1; if (navail >= nc + T4_EXTRA_INTR) { if (itype == INTR_MSIX) navail = nc + T4_EXTRA_INTR; /* navail is and must remain a pow2 for MSI */ if (itype == INTR_MSI) { KASSERT(powerof2(navail), ("%d not power of 2", navail)); while (navail / 2 >= nc + T4_EXTRA_INTR) navail /= 2; } } iaq->nirq = navail; /* total # of interrupts */ /* * If we have multiple vectors available reserve one * exclusively for errors. The rest will be shared by * the fwq and data. */ if (navail > 1) navail--; iaq->nrxq10g = min(nrxq10g, navail); iaq->nrxq1g = min(nrxq1g, navail); } navail = iaq->nirq; rc = 0; if (itype == INTR_MSIX) rc = pci_alloc_msix(sc->dev, &navail); else if (itype == INTR_MSI) rc = pci_alloc_msi(sc->dev, &navail); if (rc == 0) { if (navail == iaq->nirq) return (0); /* * Didn't get the number requested. Use whatever number * the kernel is willing to allocate (it's in navail). */ pci_release_msi(sc->dev); goto share; } device_printf(sc->dev, "failed to allocate vectors:%d, type=%d, req=%d, rcvd=%d\n", itype, rc, iaq->nirq, navail); } device_printf(sc->dev, "failed to find a usable interrupt type. " "allowed=%d, msi-x=%d, msi=%d, intx=1", intr_types, pci_msix_count(sc->dev), pci_msi_count(sc->dev)); return (ENXIO); } /* * Install a compatible firmware (if required), establish contact with it, * become the master, and reset the device. */ static int prep_firmware(struct adapter *sc) { const struct firmware *fw; int rc; enum dev_state state; /* Check firmware version and install a different one if necessary */ rc = t4_check_fw_version(sc); if (rc != 0 || force_firmware_install) { uint32_t v = 0; fw = firmware_get(T4_FWNAME); if (fw != NULL) { const struct fw_hdr *hdr = (const void *)fw->data; v = ntohl(hdr->fw_ver); /* * The firmware module will not be used if it isn't the * same major version as what the driver was compiled * with. This check trumps force_firmware_install. */ if (G_FW_HDR_FW_VER_MAJOR(v) != FW_VERSION_MAJOR) { device_printf(sc->dev, "Found firmware image but version %d " "can not be used with this driver (%d)\n", G_FW_HDR_FW_VER_MAJOR(v), FW_VERSION_MAJOR); firmware_put(fw, FIRMWARE_UNLOAD); fw = NULL; } } if (fw == NULL && (rc < 0 || force_firmware_install)) { device_printf(sc->dev, "No usable firmware. " "card has %d.%d.%d, driver compiled with %d.%d.%d, " "force_firmware_install%s set", G_FW_HDR_FW_VER_MAJOR(sc->params.fw_vers), G_FW_HDR_FW_VER_MINOR(sc->params.fw_vers), G_FW_HDR_FW_VER_MICRO(sc->params.fw_vers), FW_VERSION_MAJOR, FW_VERSION_MINOR, FW_VERSION_MICRO, force_firmware_install ? "" : " not"); return (EAGAIN); } /* * Always upgrade, even for minor/micro/build mismatches. * Downgrade only for a major version mismatch or if * force_firmware_install was specified. */ if (fw != NULL && (rc < 0 || force_firmware_install || v > sc->params.fw_vers)) { device_printf(sc->dev, "installing firmware %d.%d.%d.%d on card.\n", G_FW_HDR_FW_VER_MAJOR(v), G_FW_HDR_FW_VER_MINOR(v), G_FW_HDR_FW_VER_MICRO(v), G_FW_HDR_FW_VER_BUILD(v)); rc = -t4_load_fw(sc, fw->data, fw->datasize); if (rc != 0) { device_printf(sc->dev, "failed to install firmware: %d\n", rc); firmware_put(fw, FIRMWARE_UNLOAD); return (rc); } else { /* refresh */ (void) t4_check_fw_version(sc); } } if (fw != NULL) firmware_put(fw, FIRMWARE_UNLOAD); } /* Contact firmware, request master */ rc = t4_fw_hello(sc, sc->mbox, sc->mbox, MASTER_MUST, &state); if (rc < 0) { rc = -rc; device_printf(sc->dev, "failed to connect to the firmware: %d.\n", rc); return (rc); } /* Reset device */ rc = -t4_fw_reset(sc, sc->mbox, F_PIORSTMODE | F_PIORST); if (rc != 0) { device_printf(sc->dev, "firmware reset failed: %d.\n", rc); if (rc != ETIMEDOUT && rc != EIO) t4_fw_bye(sc, sc->mbox); return (rc); } snprintf(sc->fw_version, sizeof(sc->fw_version), "%u.%u.%u.%u", G_FW_HDR_FW_VER_MAJOR(sc->params.fw_vers), G_FW_HDR_FW_VER_MINOR(sc->params.fw_vers), G_FW_HDR_FW_VER_MICRO(sc->params.fw_vers), G_FW_HDR_FW_VER_BUILD(sc->params.fw_vers)); sc->flags |= FW_OK; return (0); } static int get_devlog_params(struct adapter *sc, struct devlog_params *dlog) { struct fw_devlog_cmd devlog_cmd; uint32_t meminfo; int rc; bzero(&devlog_cmd, sizeof(devlog_cmd)); devlog_cmd.op_to_write = htobe32(V_FW_CMD_OP(FW_DEVLOG_CMD) | F_FW_CMD_REQUEST | F_FW_CMD_READ); devlog_cmd.retval_len16 = htobe32(FW_LEN16(devlog_cmd)); rc = -t4_wr_mbox(sc, sc->mbox, &devlog_cmd, sizeof(devlog_cmd), &devlog_cmd); if (rc != 0) { device_printf(sc->dev, "failed to get devlog parameters: %d.\n", rc); bzero(dlog, sizeof (*dlog)); return (rc); } meminfo = be32toh(devlog_cmd.memtype_devlog_memaddr16_devlog); dlog->memtype = G_FW_DEVLOG_CMD_MEMTYPE_DEVLOG(meminfo); dlog->start = G_FW_DEVLOG_CMD_MEMADDR16_DEVLOG(meminfo) << 4; dlog->size = be32toh(devlog_cmd.memsize_devlog); return (0); } static int get_capabilities(struct adapter *sc, struct fw_caps_config_cmd *caps) { int rc; bzero(caps, sizeof(*caps)); caps->op_to_write = htobe32(V_FW_CMD_OP(FW_CAPS_CONFIG_CMD) | F_FW_CMD_REQUEST | F_FW_CMD_READ); caps->retval_len16 = htobe32(FW_LEN16(*caps)); rc = -t4_wr_mbox(sc, sc->mbox, caps, sizeof(*caps), caps); if (rc != 0) return (rc); if (caps->niccaps & htobe16(FW_CAPS_CONFIG_NIC_VM)) caps->niccaps ^= htobe16(FW_CAPS_CONFIG_NIC_VM); caps->op_to_write = htobe32(V_FW_CMD_OP(FW_CAPS_CONFIG_CMD) | F_FW_CMD_REQUEST | F_FW_CMD_WRITE); rc = -t4_wr_mbox(sc, sc->mbox, caps, sizeof(*caps), NULL); return (rc); } static int get_params(struct adapter *sc, struct fw_caps_config_cmd *caps) { int rc; uint32_t params[7], val[7]; #define FW_PARAM_DEV(param) \ (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_DEV) | \ V_FW_PARAMS_PARAM_X(FW_PARAMS_PARAM_DEV_##param)) #define FW_PARAM_PFVF(param) \ (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_PFVF) | \ V_FW_PARAMS_PARAM_X(FW_PARAMS_PARAM_PFVF_##param)) params[0] = FW_PARAM_DEV(PORTVEC); params[1] = FW_PARAM_PFVF(IQFLINT_START); params[2] = FW_PARAM_PFVF(EQ_START); params[3] = FW_PARAM_PFVF(FILTER_START); params[4] = FW_PARAM_PFVF(FILTER_END); rc = -t4_query_params(sc, sc->mbox, sc->pf, 0, 5, params, val); if (rc != 0) { device_printf(sc->dev, "failed to query parameters: %d.\n", rc); goto done; } sc->params.portvec = val[0]; sc->params.nports = 0; while (val[0]) { sc->params.nports++; val[0] &= val[0] - 1; } sc->sge.iq_start = val[1]; sc->sge.eq_start = val[2]; sc->tids.ftid_base = val[3]; sc->tids.nftids = val[4] - val[3] + 1; if (caps->toecaps) { /* query offload-related parameters */ params[0] = FW_PARAM_DEV(NTID); params[1] = FW_PARAM_PFVF(SERVER_START); params[2] = FW_PARAM_PFVF(SERVER_END); params[3] = FW_PARAM_PFVF(TDDP_START); params[4] = FW_PARAM_PFVF(TDDP_END); params[5] = FW_PARAM_DEV(FLOWC_BUFFIFO_SZ); rc = -t4_query_params(sc, sc->mbox, sc->pf, 0, 6, params, val); if (rc != 0) { device_printf(sc->dev, "failed to query TOE parameters: %d.\n", rc); goto done; } sc->tids.ntids = val[0]; sc->tids.natids = min(sc->tids.ntids / 2, MAX_ATIDS); sc->tids.stid_base = val[1]; sc->tids.nstids = val[2] - val[1] + 1; sc->vres.ddp.start = val[3]; sc->vres.ddp.size = val[4] - val[3] + 1; sc->params.ofldq_wr_cred = val[5]; sc->params.offload = 1; } if (caps->rdmacaps) { params[0] = FW_PARAM_PFVF(STAG_START); params[1] = FW_PARAM_PFVF(STAG_END); params[2] = FW_PARAM_PFVF(RQ_START); params[3] = FW_PARAM_PFVF(RQ_END); params[4] = FW_PARAM_PFVF(PBL_START); params[5] = FW_PARAM_PFVF(PBL_END); rc = -t4_query_params(sc, sc->mbox, sc->pf, 0, 6, params, val); if (rc != 0) { device_printf(sc->dev, "failed to query RDMA parameters: %d.\n", rc); goto done; } sc->vres.stag.start = val[0]; sc->vres.stag.size = val[1] - val[0] + 1; sc->vres.rq.start = val[2]; sc->vres.rq.size = val[3] - val[2] + 1; sc->vres.pbl.start = val[4]; sc->vres.pbl.size = val[5] - val[4] + 1; } if (caps->iscsicaps) { params[0] = FW_PARAM_PFVF(ISCSI_START); params[1] = FW_PARAM_PFVF(ISCSI_END); rc = -t4_query_params(sc, sc->mbox, sc->pf, 0, 2, params, val); if (rc != 0) { device_printf(sc->dev, "failed to query iSCSI parameters: %d.\n", rc); goto done; } sc->vres.iscsi.start = val[0]; sc->vres.iscsi.size = val[1] - val[0] + 1; } #undef FW_PARAM_PFVF #undef FW_PARAM_DEV done: return (rc); } static void t4_set_desc(struct adapter *sc) { char buf[128]; struct adapter_params *p = &sc->params; snprintf(buf, sizeof(buf), "Chelsio %s (rev %d) %d port %sNIC PCIe-x%d %d %s, S/N:%s, E/C:%s", p->vpd.id, p->rev, p->nports, is_offload(sc) ? "R" : "", p->pci.width, sc->intr_count, sc->intr_type == INTR_MSIX ? "MSI-X" : (sc->intr_type == INTR_MSI ? "MSI" : "INTx"), p->vpd.sn, p->vpd.ec); device_set_desc_copy(sc->dev, buf); } static void build_medialist(struct port_info *pi) { struct ifmedia *media = &pi->media; int data, m; PORT_LOCK(pi); ifmedia_removeall(media); m = IFM_ETHER | IFM_FDX; data = (pi->port_type << 8) | pi->mod_type; switch(pi->port_type) { case FW_PORT_TYPE_BT_XFI: ifmedia_add(media, m | IFM_10G_T, data, NULL); break; case FW_PORT_TYPE_BT_XAUI: ifmedia_add(media, m | IFM_10G_T, data, NULL); /* fall through */ case FW_PORT_TYPE_BT_SGMII: ifmedia_add(media, m | IFM_1000_T, data, NULL); ifmedia_add(media, m | IFM_100_TX, data, NULL); ifmedia_add(media, IFM_ETHER | IFM_AUTO, data, NULL); ifmedia_set(media, IFM_ETHER | IFM_AUTO); break; case FW_PORT_TYPE_CX4: ifmedia_add(media, m | IFM_10G_CX4, data, NULL); ifmedia_set(media, m | IFM_10G_CX4); break; case FW_PORT_TYPE_SFP: case FW_PORT_TYPE_FIBER_XFI: case FW_PORT_TYPE_FIBER_XAUI: switch (pi->mod_type) { case FW_PORT_MOD_TYPE_LR: ifmedia_add(media, m | IFM_10G_LR, data, NULL); ifmedia_set(media, m | IFM_10G_LR); break; case FW_PORT_MOD_TYPE_SR: ifmedia_add(media, m | IFM_10G_SR, data, NULL); ifmedia_set(media, m | IFM_10G_SR); break; case FW_PORT_MOD_TYPE_LRM: ifmedia_add(media, m | IFM_10G_LRM, data, NULL); ifmedia_set(media, m | IFM_10G_LRM); break; case FW_PORT_MOD_TYPE_TWINAX_PASSIVE: case FW_PORT_MOD_TYPE_TWINAX_ACTIVE: ifmedia_add(media, m | IFM_10G_TWINAX, data, NULL); ifmedia_set(media, m | IFM_10G_TWINAX); break; case FW_PORT_MOD_TYPE_NONE: m &= ~IFM_FDX; ifmedia_add(media, m | IFM_NONE, data, NULL); ifmedia_set(media, m | IFM_NONE); break; case FW_PORT_MOD_TYPE_NA: case FW_PORT_MOD_TYPE_ER: default: ifmedia_add(media, m | IFM_UNKNOWN, data, NULL); ifmedia_set(media, m | IFM_UNKNOWN); break; } break; case FW_PORT_TYPE_KX4: case FW_PORT_TYPE_KX: case FW_PORT_TYPE_KR: default: ifmedia_add(media, m | IFM_UNKNOWN, data, NULL); ifmedia_set(media, m | IFM_UNKNOWN); break; } PORT_UNLOCK(pi); } /* * Program the port's XGMAC based on parameters in ifnet. The caller also * indicates which parameters should be programmed (the rest are left alone). */ static int update_mac_settings(struct port_info *pi, int flags) { int rc; struct ifnet *ifp = pi->ifp; struct adapter *sc = pi->adapter; int mtu = -1, promisc = -1, allmulti = -1, vlanex = -1; PORT_LOCK_ASSERT_OWNED(pi); KASSERT(flags, ("%s: not told what to update.", __func__)); if (flags & XGMAC_MTU) mtu = ifp->if_mtu; if (flags & XGMAC_PROMISC) promisc = ifp->if_flags & IFF_PROMISC ? 1 : 0; if (flags & XGMAC_ALLMULTI) allmulti = ifp->if_flags & IFF_ALLMULTI ? 1 : 0; if (flags & XGMAC_VLANEX) vlanex = ifp->if_capenable & IFCAP_VLAN_HWTAGGING ? 1 : 0; rc = -t4_set_rxmode(sc, sc->mbox, pi->viid, mtu, promisc, allmulti, 1, vlanex, false); if (rc) { if_printf(ifp, "set_rxmode (%x) failed: %d\n", flags, rc); return (rc); } if (flags & XGMAC_UCADDR) { uint8_t ucaddr[ETHER_ADDR_LEN]; bcopy(IF_LLADDR(ifp), ucaddr, sizeof(ucaddr)); rc = t4_change_mac(sc, sc->mbox, pi->viid, pi->xact_addr_filt, ucaddr, true, true); if (rc < 0) { rc = -rc; if_printf(ifp, "change_mac failed: %d\n", rc); return (rc); } else { pi->xact_addr_filt = rc; rc = 0; } } if (flags & XGMAC_MCADDRS) { const uint8_t *mcaddr; int del = 1; uint64_t hash = 0; struct ifmultiaddr *ifma; if_maddr_rlock(ifp); TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) { if (ifma->ifma_addr->sa_family != AF_LINK) continue; mcaddr = LLADDR((struct sockaddr_dl *)ifma->ifma_addr); rc = t4_alloc_mac_filt(sc, sc->mbox, pi->viid, del, 1, &mcaddr, NULL, &hash, 0); if (rc < 0) { rc = -rc; if_printf(ifp, "failed to add mc address" " %02x:%02x:%02x:%02x:%02x:%02x rc=%d\n", mcaddr[0], mcaddr[1], mcaddr[2], mcaddr[3], mcaddr[4], mcaddr[5], rc); goto mcfail; } del = 0; } rc = -t4_set_addr_hash(sc, sc->mbox, pi->viid, 0, hash, 0); if (rc != 0) if_printf(ifp, "failed to set mc address hash: %d", rc); mcfail: if_maddr_runlock(ifp); } return (rc); } static int cxgbe_init_locked(struct port_info *pi) { struct adapter *sc = pi->adapter; int rc = 0; ADAPTER_LOCK_ASSERT_OWNED(sc); while (!IS_DOOMED(pi) && IS_BUSY(sc)) { if (mtx_sleep(&sc->flags, &sc->sc_lock, PCATCH, "t4init", 0)) { rc = EINTR; goto done; } } if (IS_DOOMED(pi)) { rc = ENXIO; goto done; } KASSERT(!IS_BUSY(sc), ("%s: controller busy.", __func__)); /* Give up the adapter lock, port init code can sleep. */ SET_BUSY(sc); ADAPTER_UNLOCK(sc); rc = cxgbe_init_synchronized(pi); done: ADAPTER_LOCK(sc); KASSERT(IS_BUSY(sc), ("%s: controller not busy.", __func__)); CLR_BUSY(sc); wakeup_one(&sc->flags); ADAPTER_UNLOCK(sc); return (rc); } static int cxgbe_init_synchronized(struct port_info *pi) { struct adapter *sc = pi->adapter; struct ifnet *ifp = pi->ifp; int rc = 0, i; uint16_t *rss; struct sge_rxq *rxq; ADAPTER_LOCK_ASSERT_NOTOWNED(sc); if (isset(&sc->open_device_map, pi->port_id)) { KASSERT(ifp->if_drv_flags & IFF_DRV_RUNNING, ("mismatch between open_device_map and if_drv_flags")); return (0); /* already running */ } if (sc->open_device_map == 0 && ((rc = first_port_up(sc)) != 0)) return (rc); /* error message displayed already */ /* * Allocate tx/rx/fl queues for this port. */ rc = t4_setup_eth_queues(pi); if (rc != 0) goto done; /* error message displayed already */ /* * Setup RSS for this port. */ rss = malloc(pi->nrxq * sizeof (*rss), M_CXGBE, M_ZERO | M_WAITOK); for_each_rxq(pi, i, rxq) { rss[i] = rxq->iq.abs_id; } rc = -t4_config_rss_range(sc, sc->mbox, pi->viid, 0, pi->rss_size, rss, pi->nrxq); free(rss, M_CXGBE); if (rc != 0) { if_printf(ifp, "rss_config failed: %d\n", rc); goto done; } PORT_LOCK(pi); rc = update_mac_settings(pi, XGMAC_ALL); PORT_UNLOCK(pi); if (rc) goto done; /* error message displayed already */ rc = -t4_link_start(sc, sc->mbox, pi->tx_chan, &pi->link_cfg); if (rc != 0) { if_printf(ifp, "start_link failed: %d\n", rc); goto done; } rc = -t4_enable_vi(sc, sc->mbox, pi->viid, true, true); if (rc != 0) { if_printf(ifp, "enable_vi failed: %d\n", rc); goto done; } pi->flags |= VI_ENABLED; /* all ok */ setbit(&sc->open_device_map, pi->port_id); ifp->if_drv_flags |= IFF_DRV_RUNNING; ifp->if_drv_flags &= ~IFF_DRV_OACTIVE; callout_reset(&pi->tick, hz, cxgbe_tick, pi); done: if (rc != 0) cxgbe_uninit_synchronized(pi); return (rc); } static int cxgbe_uninit_locked(struct port_info *pi) { struct adapter *sc = pi->adapter; int rc; ADAPTER_LOCK_ASSERT_OWNED(sc); while (!IS_DOOMED(pi) && IS_BUSY(sc)) { if (mtx_sleep(&sc->flags, &sc->sc_lock, PCATCH, "t4uninit", 0)) { rc = EINTR; goto done; } } if (IS_DOOMED(pi)) { rc = ENXIO; goto done; } KASSERT(!IS_BUSY(sc), ("%s: controller busy.", __func__)); SET_BUSY(sc); ADAPTER_UNLOCK(sc); rc = cxgbe_uninit_synchronized(pi); ADAPTER_LOCK(sc); KASSERT(IS_BUSY(sc), ("%s: controller not busy.", __func__)); CLR_BUSY(sc); wakeup_one(&sc->flags); done: ADAPTER_UNLOCK(sc); return (rc); } /* * Idempotent. */ static int cxgbe_uninit_synchronized(struct port_info *pi) { struct adapter *sc = pi->adapter; struct ifnet *ifp = pi->ifp; int rc; /* * taskqueue_drain may cause a deadlock if the adapter lock is held. */ ADAPTER_LOCK_ASSERT_NOTOWNED(sc); /* * Clear this port's bit from the open device map, and then drain * tasks and callouts. */ clrbit(&sc->open_device_map, pi->port_id); PORT_LOCK(pi); ifp->if_drv_flags &= ~(IFF_DRV_RUNNING | IFF_DRV_OACTIVE); callout_stop(&pi->tick); PORT_UNLOCK(pi); callout_drain(&pi->tick); /* * Stop and then free the queues' resources, including the queues * themselves. * * XXX: we could just stop the queues here (on ifconfig down) and free * them later (on port detach), but having up/down go through the entire * allocate/activate/deactivate/free sequence is a good way to find * leaks and bugs. */ rc = t4_teardown_eth_queues(pi); if (rc != 0) if_printf(ifp, "teardown failed: %d\n", rc); if (pi->flags & VI_ENABLED) { rc = -t4_enable_vi(sc, sc->mbox, pi->viid, false, false); if (rc) if_printf(ifp, "disable_vi failed: %d\n", rc); else pi->flags &= ~VI_ENABLED; } pi->link_cfg.link_ok = 0; pi->link_cfg.speed = 0; t4_os_link_changed(sc, pi->port_id, 0); if (sc->open_device_map == 0) last_port_down(sc); return (0); } #define T4_ALLOC_IRQ(sc, irq, rid, handler, arg, name) do { \ rc = t4_alloc_irq(sc, irq, rid, handler, arg, name); \ if (rc != 0) \ goto done; \ } while (0) static int first_port_up(struct adapter *sc) { int rc, i, rid, p, q; char s[8]; struct irq *irq; struct sge_iq *intrq; ADAPTER_LOCK_ASSERT_NOTOWNED(sc); /* * queues that belong to the adapter (not any particular port). */ rc = t4_setup_adapter_queues(sc); if (rc != 0) goto done; /* * Setup interrupts. */ irq = &sc->irq[0]; rid = sc->intr_type == INTR_INTX ? 0 : 1; if (sc->intr_count == 1) { KASSERT(sc->flags & INTR_SHARED, ("%s: single interrupt but not shared?", __func__)); T4_ALLOC_IRQ(sc, irq, rid, t4_intr_all, sc, "all"); } else { /* Multiple interrupts. The first one is always error intr */ T4_ALLOC_IRQ(sc, irq, rid, t4_intr_err, sc, "err"); irq++; rid++; /* Firmware event queue normally has an interrupt of its own */ if (sc->intr_count > T4_EXTRA_INTR) { T4_ALLOC_IRQ(sc, irq, rid, t4_intr_evt, &sc->sge.fwq, "evt"); irq++; rid++; } intrq = &sc->sge.intrq[0]; if (sc->flags & INTR_SHARED) { /* All ports share these interrupt queues */ for (i = 0; i < NINTRQ(sc); i++) { snprintf(s, sizeof(s), "*.%d", i); T4_ALLOC_IRQ(sc, irq, rid, t4_intr, intrq, s); irq++; rid++; intrq++; } } else { /* Each port has its own set of interrupt queues */ for (p = 0; p < sc->params.nports; p++) { for (q = 0; q < sc->port[p]->nrxq; q++) { snprintf(s, sizeof(s), "%d.%d", p, q); T4_ALLOC_IRQ(sc, irq, rid, t4_intr, intrq, s); irq++; rid++; intrq++; } } } } t4_intr_enable(sc); sc->flags |= FULL_INIT_DONE; done: if (rc != 0) last_port_down(sc); return (rc); } #undef T4_ALLOC_IRQ /* * Idempotent. */ static int last_port_down(struct adapter *sc) { int i; ADAPTER_LOCK_ASSERT_NOTOWNED(sc); t4_intr_disable(sc); t4_teardown_adapter_queues(sc); for (i = 0; i < sc->intr_count; i++) t4_free_irq(sc, &sc->irq[i]); sc->flags &= ~FULL_INIT_DONE; return (0); } static int t4_alloc_irq(struct adapter *sc, struct irq *irq, int rid, iq_intr_handler_t *handler, void *arg, char *name) { int rc; irq->rid = rid; irq->res = bus_alloc_resource_any(sc->dev, SYS_RES_IRQ, &irq->rid, RF_SHAREABLE | RF_ACTIVE); if (irq->res == NULL) { device_printf(sc->dev, "failed to allocate IRQ for rid %d, name %s.\n", rid, name); return (ENOMEM); } rc = bus_setup_intr(sc->dev, irq->res, INTR_MPSAFE | INTR_TYPE_NET, NULL, handler, arg, &irq->tag); if (rc != 0) { device_printf(sc->dev, "failed to setup interrupt for rid %d, name %s: %d\n", rid, name, rc); } else if (name) bus_describe_intr(sc->dev, irq->res, irq->tag, name); return (rc); } static int t4_free_irq(struct adapter *sc, struct irq *irq) { if (irq->tag) bus_teardown_intr(sc->dev, irq->res, irq->tag); if (irq->res) bus_release_resource(sc->dev, SYS_RES_IRQ, irq->rid, irq->res); bzero(irq, sizeof(*irq)); return (0); } static void reg_block_dump(struct adapter *sc, uint8_t *buf, unsigned int start, unsigned int end) { uint32_t *p = (uint32_t *)(buf + start); for ( ; start <= end; start += sizeof(uint32_t)) *p++ = t4_read_reg(sc, start); } static void t4_get_regs(struct adapter *sc, struct t4_regdump *regs, uint8_t *buf) { int i; static const unsigned int reg_ranges[] = { 0x1008, 0x1108, 0x1180, 0x11b4, 0x11fc, 0x123c, 0x1300, 0x173c, 0x1800, 0x18fc, 0x3000, 0x30d8, 0x30e0, 0x5924, 0x5960, 0x59d4, 0x5a00, 0x5af8, 0x6000, 0x6098, 0x6100, 0x6150, 0x6200, 0x6208, 0x6240, 0x6248, 0x6280, 0x6338, 0x6370, 0x638c, 0x6400, 0x643c, 0x6500, 0x6524, 0x6a00, 0x6a38, 0x6a60, 0x6a78, 0x6b00, 0x6b84, 0x6bf0, 0x6c84, 0x6cf0, 0x6d84, 0x6df0, 0x6e84, 0x6ef0, 0x6f84, 0x6ff0, 0x7084, 0x70f0, 0x7184, 0x71f0, 0x7284, 0x72f0, 0x7384, 0x73f0, 0x7450, 0x7500, 0x7530, 0x7600, 0x761c, 0x7680, 0x76cc, 0x7700, 0x7798, 0x77c0, 0x77fc, 0x7900, 0x79fc, 0x7b00, 0x7c38, 0x7d00, 0x7efc, 0x8dc0, 0x8e1c, 0x8e30, 0x8e78, 0x8ea0, 0x8f6c, 0x8fc0, 0x9074, 0x90fc, 0x90fc, 0x9400, 0x9458, 0x9600, 0x96bc, 0x9800, 0x9808, 0x9820, 0x983c, 0x9850, 0x9864, 0x9c00, 0x9c6c, 0x9c80, 0x9cec, 0x9d00, 0x9d6c, 0x9d80, 0x9dec, 0x9e00, 0x9e6c, 0x9e80, 0x9eec, 0x9f00, 0x9f6c, 0x9f80, 0x9fec, 0xd004, 0xd03c, 0xdfc0, 0xdfe0, 0xe000, 0xea7c, 0xf000, 0x11190, 0x19040, 0x19124, 0x19150, 0x191b0, 0x191d0, 0x191e8, 0x19238, 0x1924c, 0x193f8, 0x19474, 0x19490, 0x194f8, 0x19800, 0x19f30, 0x1a000, 0x1a06c, 0x1a0b0, 0x1a120, 0x1a128, 0x1a138, 0x1a190, 0x1a1c4, 0x1a1fc, 0x1a1fc, 0x1e040, 0x1e04c, 0x1e240, 0x1e28c, 0x1e2c0, 0x1e2c0, 0x1e2e0, 0x1e2e0, 0x1e300, 0x1e384, 0x1e3c0, 0x1e3c8, 0x1e440, 0x1e44c, 0x1e640, 0x1e68c, 0x1e6c0, 0x1e6c0, 0x1e6e0, 0x1e6e0, 0x1e700, 0x1e784, 0x1e7c0, 0x1e7c8, 0x1e840, 0x1e84c, 0x1ea40, 0x1ea8c, 0x1eac0, 0x1eac0, 0x1eae0, 0x1eae0, 0x1eb00, 0x1eb84, 0x1ebc0, 0x1ebc8, 0x1ec40, 0x1ec4c, 0x1ee40, 0x1ee8c, 0x1eec0, 0x1eec0, 0x1eee0, 0x1eee0, 0x1ef00, 0x1ef84, 0x1efc0, 0x1efc8, 0x1f040, 0x1f04c, 0x1f240, 0x1f28c, 0x1f2c0, 0x1f2c0, 0x1f2e0, 0x1f2e0, 0x1f300, 0x1f384, 0x1f3c0, 0x1f3c8, 0x1f440, 0x1f44c, 0x1f640, 0x1f68c, 0x1f6c0, 0x1f6c0, 0x1f6e0, 0x1f6e0, 0x1f700, 0x1f784, 0x1f7c0, 0x1f7c8, 0x1f840, 0x1f84c, 0x1fa40, 0x1fa8c, 0x1fac0, 0x1fac0, 0x1fae0, 0x1fae0, 0x1fb00, 0x1fb84, 0x1fbc0, 0x1fbc8, 0x1fc40, 0x1fc4c, 0x1fe40, 0x1fe8c, 0x1fec0, 0x1fec0, 0x1fee0, 0x1fee0, 0x1ff00, 0x1ff84, 0x1ffc0, 0x1ffc8, 0x20000, 0x2002c, 0x20100, 0x2013c, 0x20190, 0x201c8, 0x20200, 0x20318, 0x20400, 0x20528, 0x20540, 0x20614, 0x21000, 0x21040, 0x2104c, 0x21060, 0x210c0, 0x210ec, 0x21200, 0x21268, 0x21270, 0x21284, 0x212fc, 0x21388, 0x21400, 0x21404, 0x21500, 0x21518, 0x2152c, 0x2153c, 0x21550, 0x21554, 0x21600, 0x21600, 0x21608, 0x21628, 0x21630, 0x2163c, 0x21700, 0x2171c, 0x21780, 0x2178c, 0x21800, 0x21c38, 0x21c80, 0x21d7c, 0x21e00, 0x21e04, 0x22000, 0x2202c, 0x22100, 0x2213c, 0x22190, 0x221c8, 0x22200, 0x22318, 0x22400, 0x22528, 0x22540, 0x22614, 0x23000, 0x23040, 0x2304c, 0x23060, 0x230c0, 0x230ec, 0x23200, 0x23268, 0x23270, 0x23284, 0x232fc, 0x23388, 0x23400, 0x23404, 0x23500, 0x23518, 0x2352c, 0x2353c, 0x23550, 0x23554, 0x23600, 0x23600, 0x23608, 0x23628, 0x23630, 0x2363c, 0x23700, 0x2371c, 0x23780, 0x2378c, 0x23800, 0x23c38, 0x23c80, 0x23d7c, 0x23e00, 0x23e04, 0x24000, 0x2402c, 0x24100, 0x2413c, 0x24190, 0x241c8, 0x24200, 0x24318, 0x24400, 0x24528, 0x24540, 0x24614, 0x25000, 0x25040, 0x2504c, 0x25060, 0x250c0, 0x250ec, 0x25200, 0x25268, 0x25270, 0x25284, 0x252fc, 0x25388, 0x25400, 0x25404, 0x25500, 0x25518, 0x2552c, 0x2553c, 0x25550, 0x25554, 0x25600, 0x25600, 0x25608, 0x25628, 0x25630, 0x2563c, 0x25700, 0x2571c, 0x25780, 0x2578c, 0x25800, 0x25c38, 0x25c80, 0x25d7c, 0x25e00, 0x25e04, 0x26000, 0x2602c, 0x26100, 0x2613c, 0x26190, 0x261c8, 0x26200, 0x26318, 0x26400, 0x26528, 0x26540, 0x26614, 0x27000, 0x27040, 0x2704c, 0x27060, 0x270c0, 0x270ec, 0x27200, 0x27268, 0x27270, 0x27284, 0x272fc, 0x27388, 0x27400, 0x27404, 0x27500, 0x27518, 0x2752c, 0x2753c, 0x27550, 0x27554, 0x27600, 0x27600, 0x27608, 0x27628, 0x27630, 0x2763c, 0x27700, 0x2771c, 0x27780, 0x2778c, 0x27800, 0x27c38, 0x27c80, 0x27d7c, 0x27e00, 0x27e04 }; regs->version = 4 | (sc->params.rev << 10); for (i = 0; i < ARRAY_SIZE(reg_ranges); i += 2) reg_block_dump(sc, buf, reg_ranges[i], reg_ranges[i + 1]); } static void cxgbe_tick(void *arg) { struct port_info *pi = arg; struct ifnet *ifp = pi->ifp; struct sge_txq *txq; int i, drops; struct port_stats *s = &pi->stats; PORT_LOCK(pi); if (!(ifp->if_drv_flags & IFF_DRV_RUNNING)) { PORT_UNLOCK(pi); return; /* without scheduling another callout */ } t4_get_port_stats(pi->adapter, pi->tx_chan, s); ifp->if_opackets = s->tx_frames; ifp->if_ipackets = s->rx_frames; ifp->if_obytes = s->tx_octets; ifp->if_ibytes = s->rx_octets; ifp->if_omcasts = s->tx_mcast_frames; ifp->if_imcasts = s->rx_mcast_frames; ifp->if_iqdrops = s->rx_ovflow0 + s->rx_ovflow1 + s->rx_ovflow2 + s->rx_ovflow3; drops = s->tx_drop; for_each_txq(pi, i, txq) drops += txq->br->br_drops; ifp->if_snd.ifq_drops = drops; ifp->if_oerrors = s->tx_error_frames; ifp->if_ierrors = s->rx_jabber + s->rx_runt + s->rx_too_long + s->rx_fcs_err + s->rx_len_err; callout_schedule(&pi->tick, hz); PORT_UNLOCK(pi); } static int t4_sysctls(struct adapter *sc) { struct sysctl_ctx_list *ctx; struct sysctl_oid *oid; struct sysctl_oid_list *children; ctx = device_get_sysctl_ctx(sc->dev); oid = device_get_sysctl_tree(sc->dev); children = SYSCTL_CHILDREN(oid); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "nports", CTLFLAG_RD, &sc->params.nports, 0, "# of ports"); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "hw_revision", CTLFLAG_RD, &sc->params.rev, 0, "chip hardware revision"); SYSCTL_ADD_STRING(ctx, children, OID_AUTO, "firmware_version", CTLFLAG_RD, &sc->fw_version, 0, "firmware version"); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "TOE", CTLFLAG_RD, &sc->params.offload, 0, "hardware is capable of TCP offload"); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "core_clock", CTLFLAG_RD, &sc->params.vpd.cclk, 0, "core clock frequency (in KHz)"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "holdoff_timers", CTLTYPE_STRING | CTLFLAG_RD, &intr_timer, sizeof(intr_timer), sysctl_int_array, "A", "interrupt holdoff timer values (us)"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "holdoff_pkt_counts", CTLTYPE_STRING | CTLFLAG_RD, &intr_pktcount, sizeof(intr_pktcount), sysctl_int_array, "A", "interrupt holdoff packet counter values"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "devlog", CTLTYPE_STRING | CTLFLAG_RD, sc, 0, sysctl_devlog, "A", "device log"); return (0); } static int cxgbe_sysctls(struct port_info *pi) { struct sysctl_ctx_list *ctx; struct sysctl_oid *oid; struct sysctl_oid_list *children; ctx = device_get_sysctl_ctx(pi->dev); /* * dev.cxgbe.X. */ oid = device_get_sysctl_tree(pi->dev); children = SYSCTL_CHILDREN(oid); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "nrxq", CTLFLAG_RD, &pi->nrxq, 0, "# of rx queues"); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "ntxq", CTLFLAG_RD, &pi->ntxq, 0, "# of tx queues"); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "first_rxq", CTLFLAG_RD, &pi->first_rxq, 0, "index of first rx queue"); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "first_txq", CTLFLAG_RD, &pi->first_txq, 0, "index of first tx queue"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "holdoff_tmr_idx", CTLTYPE_INT | CTLFLAG_RW, pi, 0, sysctl_holdoff_tmr_idx, "I", "holdoff timer index"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "holdoff_pktc_idx", CTLTYPE_INT | CTLFLAG_RW, pi, 0, sysctl_holdoff_pktc_idx, "I", "holdoff packet counter index"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "qsize_rxq", CTLTYPE_INT | CTLFLAG_RW, pi, 0, sysctl_qsize_rxq, "I", "rx queue size"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "qsize_txq", CTLTYPE_INT | CTLFLAG_RW, pi, 0, sysctl_qsize_txq, "I", "tx queue size"); /* * dev.cxgbe.X.stats. */ oid = SYSCTL_ADD_NODE(ctx, children, OID_AUTO, "stats", CTLFLAG_RD, NULL, "port statistics"); children = SYSCTL_CHILDREN(oid); #define SYSCTL_ADD_T4_REG64(pi, name, desc, reg) \ SYSCTL_ADD_OID(ctx, children, OID_AUTO, name, \ CTLTYPE_U64 | CTLFLAG_RD, pi->adapter, reg, \ sysctl_handle_t4_reg64, "QU", desc) SYSCTL_ADD_T4_REG64(pi, "tx_octets", "# of octets in good frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_BYTES_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames", "total # of good frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_FRAMES_L)); SYSCTL_ADD_T4_REG64(pi, "tx_bcast_frames", "# of broadcast frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_BCAST_L)); SYSCTL_ADD_T4_REG64(pi, "tx_mcast_frames", "# of multicast frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_MCAST_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ucast_frames", "# of unicast frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_UCAST_L)); SYSCTL_ADD_T4_REG64(pi, "tx_error_frames", "# of error frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_ERROR_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_64", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_64B_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_65_127", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_65B_127B_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_128_255", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_128B_255B_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_256_511", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_256B_511B_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_512_1023", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_512B_1023B_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_1024_1518", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_1024B_1518B_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_1519_max", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_1519B_MAX_L)); SYSCTL_ADD_T4_REG64(pi, "tx_drop", "# of dropped tx frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_DROP_L)); SYSCTL_ADD_T4_REG64(pi, "tx_pause", "# of pause frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PAUSE_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp0", "# of PPP prio 0 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP0_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp1", "# of PPP prio 1 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP1_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp2", "# of PPP prio 2 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP2_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp3", "# of PPP prio 3 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP3_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp4", "# of PPP prio 4 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP4_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp5", "# of PPP prio 5 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP5_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp6", "# of PPP prio 6 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP6_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp7", "# of PPP prio 7 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP7_L)); SYSCTL_ADD_T4_REG64(pi, "rx_octets", "# of octets in good frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_BYTES_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames", "total # of good frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_FRAMES_L)); SYSCTL_ADD_T4_REG64(pi, "rx_bcast_frames", "# of broadcast frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_BCAST_L)); SYSCTL_ADD_T4_REG64(pi, "rx_mcast_frames", "# of multicast frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_MCAST_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ucast_frames", "# of unicast frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_UCAST_L)); SYSCTL_ADD_T4_REG64(pi, "rx_too_long", "# of frames exceeding MTU", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_MTU_ERROR_L)); SYSCTL_ADD_T4_REG64(pi, "rx_jabber", "# of jabber frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_MTU_CRC_ERROR_L)); SYSCTL_ADD_T4_REG64(pi, "rx_fcs_err", "# of frames received with bad FCS", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_CRC_ERROR_L)); SYSCTL_ADD_T4_REG64(pi, "rx_len_err", "# of frames received with length error", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_LEN_ERROR_L)); SYSCTL_ADD_T4_REG64(pi, "rx_symbol_err", "symbol errors", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_SYM_ERROR_L)); SYSCTL_ADD_T4_REG64(pi, "rx_runt", "# of short frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_LESS_64B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_64", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_64B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_65_127", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_65B_127B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_128_255", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_128B_255B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_256_511", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_256B_511B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_512_1023", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_512B_1023B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_1024_1518", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_1024B_1518B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_1519_max", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_1519B_MAX_L)); SYSCTL_ADD_T4_REG64(pi, "rx_pause", "# of pause frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PAUSE_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp0", "# of PPP prio 0 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP0_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp1", "# of PPP prio 1 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP1_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp2", "# of PPP prio 2 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP2_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp3", "# of PPP prio 3 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP3_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp4", "# of PPP prio 4 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP4_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp5", "# of PPP prio 5 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP5_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp6", "# of PPP prio 6 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP6_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp7", "# of PPP prio 7 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP7_L)); #undef SYSCTL_ADD_T4_REG64 #define SYSCTL_ADD_T4_PORTSTAT(name, desc) \ SYSCTL_ADD_UQUAD(ctx, children, OID_AUTO, #name, CTLFLAG_RD, \ &pi->stats.name, desc) /* We get these from port_stats and they may be stale by upto 1s */ SYSCTL_ADD_T4_PORTSTAT(rx_ovflow0, "# drops due to buffer-group 0 overflows"); SYSCTL_ADD_T4_PORTSTAT(rx_ovflow1, "# drops due to buffer-group 1 overflows"); SYSCTL_ADD_T4_PORTSTAT(rx_ovflow2, "# drops due to buffer-group 2 overflows"); SYSCTL_ADD_T4_PORTSTAT(rx_ovflow3, "# drops due to buffer-group 3 overflows"); SYSCTL_ADD_T4_PORTSTAT(rx_trunc0, "# of buffer-group 0 truncated packets"); SYSCTL_ADD_T4_PORTSTAT(rx_trunc1, "# of buffer-group 1 truncated packets"); SYSCTL_ADD_T4_PORTSTAT(rx_trunc2, "# of buffer-group 2 truncated packets"); SYSCTL_ADD_T4_PORTSTAT(rx_trunc3, "# of buffer-group 3 truncated packets"); #undef SYSCTL_ADD_T4_PORTSTAT return (0); } static int sysctl_int_array(SYSCTL_HANDLER_ARGS) { int rc, *i; struct sbuf sb; sbuf_new(&sb, NULL, 32, SBUF_AUTOEXTEND); for (i = arg1; arg2; arg2 -= sizeof(int), i++) sbuf_printf(&sb, "%d ", *i); sbuf_trim(&sb); sbuf_finish(&sb); rc = sysctl_handle_string(oidp, sbuf_data(&sb), sbuf_len(&sb), req); sbuf_delete(&sb); return (rc); } static int sysctl_holdoff_tmr_idx(SYSCTL_HANDLER_ARGS) { struct port_info *pi = arg1; struct adapter *sc = pi->adapter; struct sge_rxq *rxq; int idx, rc, i; idx = pi->tmr_idx; rc = sysctl_handle_int(oidp, &idx, 0, req); if (rc != 0 || req->newptr == NULL) return (rc); if (idx < 0 || idx >= SGE_NTIMERS) return (EINVAL); ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc == 0) { for_each_rxq(pi, i, rxq) { rxq->iq.intr_params = V_QINTR_TIMER_IDX(idx) | V_QINTR_CNT_EN(pi->pktc_idx != -1); } pi->tmr_idx = idx; } ADAPTER_UNLOCK(sc); return (rc); } static int sysctl_holdoff_pktc_idx(SYSCTL_HANDLER_ARGS) { struct port_info *pi = arg1; struct adapter *sc = pi->adapter; int idx, rc; idx = pi->pktc_idx; rc = sysctl_handle_int(oidp, &idx, 0, req); if (rc != 0 || req->newptr == NULL) return (rc); if (idx < -1 || idx >= SGE_NCOUNTERS) return (EINVAL); ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc == 0 && pi->ifp->if_drv_flags & IFF_DRV_RUNNING) rc = EBUSY; /* can be changed only when port is down */ if (rc == 0) pi->pktc_idx = idx; ADAPTER_UNLOCK(sc); return (rc); } static int sysctl_qsize_rxq(SYSCTL_HANDLER_ARGS) { struct port_info *pi = arg1; struct adapter *sc = pi->adapter; int qsize, rc; qsize = pi->qsize_rxq; rc = sysctl_handle_int(oidp, &qsize, 0, req); if (rc != 0 || req->newptr == NULL) return (rc); if (qsize < 128 || (qsize & 7)) return (EINVAL); ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc == 0 && pi->ifp->if_drv_flags & IFF_DRV_RUNNING) rc = EBUSY; /* can be changed only when port is down */ if (rc == 0) pi->qsize_rxq = qsize; ADAPTER_UNLOCK(sc); return (rc); } static int sysctl_qsize_txq(SYSCTL_HANDLER_ARGS) { struct port_info *pi = arg1; struct adapter *sc = pi->adapter; int qsize, rc; qsize = pi->qsize_txq; rc = sysctl_handle_int(oidp, &qsize, 0, req); if (rc != 0 || req->newptr == NULL) return (rc); if (qsize < 128) return (EINVAL); ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc == 0 && pi->ifp->if_drv_flags & IFF_DRV_RUNNING) rc = EBUSY; /* can be changed only when port is down */ if (rc == 0) pi->qsize_txq = qsize; ADAPTER_UNLOCK(sc); return (rc); } static int sysctl_handle_t4_reg64(SYSCTL_HANDLER_ARGS) { struct adapter *sc = arg1; int reg = arg2; uint64_t val; val = t4_read_reg64(sc, reg); return (sysctl_handle_64(oidp, &val, 0, req)); } const char *devlog_level_strings[] = { [FW_DEVLOG_LEVEL_EMERG] = "EMERG", [FW_DEVLOG_LEVEL_CRIT] = "CRIT", [FW_DEVLOG_LEVEL_ERR] = "ERR", [FW_DEVLOG_LEVEL_NOTICE] = "NOTICE", [FW_DEVLOG_LEVEL_INFO] = "INFO", [FW_DEVLOG_LEVEL_DEBUG] = "DEBUG" }; const char *devlog_facility_strings[] = { [FW_DEVLOG_FACILITY_CORE] = "CORE", [FW_DEVLOG_FACILITY_SCHED] = "SCHED", [FW_DEVLOG_FACILITY_TIMER] = "TIMER", [FW_DEVLOG_FACILITY_RES] = "RES", [FW_DEVLOG_FACILITY_HW] = "HW", [FW_DEVLOG_FACILITY_FLR] = "FLR", [FW_DEVLOG_FACILITY_DMAQ] = "DMAQ", [FW_DEVLOG_FACILITY_PHY] = "PHY", [FW_DEVLOG_FACILITY_MAC] = "MAC", [FW_DEVLOG_FACILITY_PORT] = "PORT", [FW_DEVLOG_FACILITY_VI] = "VI", [FW_DEVLOG_FACILITY_FILTER] = "FILTER", [FW_DEVLOG_FACILITY_ACL] = "ACL", [FW_DEVLOG_FACILITY_TM] = "TM", [FW_DEVLOG_FACILITY_QFC] = "QFC", [FW_DEVLOG_FACILITY_DCB] = "DCB", [FW_DEVLOG_FACILITY_ETH] = "ETH", [FW_DEVLOG_FACILITY_OFLD] = "OFLD", [FW_DEVLOG_FACILITY_RI] = "RI", [FW_DEVLOG_FACILITY_ISCSI] = "ISCSI", [FW_DEVLOG_FACILITY_FCOE] = "FCOE", [FW_DEVLOG_FACILITY_FOISCSI] = "FOISCSI", [FW_DEVLOG_FACILITY_FOFCOE] = "FOFCOE" }; static int sysctl_devlog(SYSCTL_HANDLER_ARGS) { struct adapter *sc = arg1; struct devlog_params *dparams = &sc->params.devlog; struct fw_devlog_e *buf, *e; int i, j, rc, nentries, first = 0; struct sbuf *sb; uint64_t ftstamp = UINT64_MAX; if (dparams->start == 0) return (ENXIO); nentries = dparams->size / sizeof(struct fw_devlog_e); buf = malloc(dparams->size, M_CXGBE, M_NOWAIT); if (buf == NULL) return (ENOMEM); rc = -t4_mem_read(sc, dparams->memtype, dparams->start, dparams->size, (void *)buf); if (rc != 0) goto done; for (i = 0; i < nentries; i++) { e = &buf[i]; if (e->timestamp == 0) break; /* end */ e->timestamp = be64toh(e->timestamp); e->seqno = be32toh(e->seqno); for (j = 0; j < 8; j++) e->params[j] = be32toh(e->params[j]); if (e->timestamp < ftstamp) { ftstamp = e->timestamp; first = i; } } if (buf[first].timestamp == 0) goto done; /* nothing in the log */ rc = sysctl_wire_old_buffer(req, 0); if (rc != 0) goto done; sb = sbuf_new_for_sysctl(NULL, NULL, 4096, req); sbuf_printf(sb, "\n%10s %15s %8s %8s %s\n", "Seq#", "Tstamp", "Level", "Facility", "Message"); i = first; do { e = &buf[i]; if (e->timestamp == 0) break; /* end */ sbuf_printf(sb, "%10d %15ju %8s %8s ", e->seqno, e->timestamp, (e->level < ARRAY_SIZE(devlog_level_strings) ? devlog_level_strings[e->level] : "UNKNOWN"), (e->facility < ARRAY_SIZE(devlog_facility_strings) ? devlog_facility_strings[e->facility] : "UNKNOWN")); sbuf_printf(sb, e->fmt, e->params[0], e->params[1], e->params[2], e->params[3], e->params[4], e->params[5], e->params[6], e->params[7]); if (++i == nentries) i = 0; } while (i != first); rc = sbuf_finish(sb); sbuf_delete(sb); done: free(buf, M_CXGBE); return (rc); } static inline void txq_start(struct ifnet *ifp, struct sge_txq *txq) { struct buf_ring *br; struct mbuf *m; TXQ_LOCK_ASSERT_OWNED(txq); br = txq->br; m = txq->m ? txq->m : drbr_dequeue(ifp, br); if (m) t4_eth_tx(ifp, txq, m); } void cxgbe_txq_start(void *arg, int count) { struct sge_txq *txq = arg; TXQ_LOCK(txq); if (txq->eq.flags & EQ_CRFLUSHED) { txq->eq.flags &= ~EQ_CRFLUSHED; txq_start(txq->ifp, txq); } else wakeup_one(txq); /* txq is going away, wakeup free_txq */ TXQ_UNLOCK(txq); } static uint32_t fconf_to_mode(uint32_t fconf) { uint32_t mode; mode = T4_FILTER_IPv4 | T4_FILTER_IPv6 | T4_FILTER_IP_SADDR | T4_FILTER_IP_DADDR | T4_FILTER_IP_SPORT | T4_FILTER_IP_DPORT; if (fconf & F_FRAGMENTATION) mode |= T4_FILTER_IP_FRAGMENT; if (fconf & F_MPSHITTYPE) mode |= T4_FILTER_MPS_HIT_TYPE; if (fconf & F_MACMATCH) mode |= T4_FILTER_MAC_IDX; if (fconf & F_ETHERTYPE) mode |= T4_FILTER_ETH_TYPE; if (fconf & F_PROTOCOL) mode |= T4_FILTER_IP_PROTO; if (fconf & F_TOS) mode |= T4_FILTER_IP_TOS; if (fconf & F_VLAN) mode |= T4_FILTER_IVLAN; if (fconf & F_VNIC_ID) mode |= T4_FILTER_OVLAN; if (fconf & F_PORT) mode |= T4_FILTER_PORT; if (fconf & F_FCOE) mode |= T4_FILTER_FCoE; return (mode); } static uint32_t mode_to_fconf(uint32_t mode) { uint32_t fconf = 0; if (mode & T4_FILTER_IP_FRAGMENT) fconf |= F_FRAGMENTATION; if (mode & T4_FILTER_MPS_HIT_TYPE) fconf |= F_MPSHITTYPE; if (mode & T4_FILTER_MAC_IDX) fconf |= F_MACMATCH; if (mode & T4_FILTER_ETH_TYPE) fconf |= F_ETHERTYPE; if (mode & T4_FILTER_IP_PROTO) fconf |= F_PROTOCOL; if (mode & T4_FILTER_IP_TOS) fconf |= F_TOS; if (mode & T4_FILTER_IVLAN) fconf |= F_VLAN; if (mode & T4_FILTER_OVLAN) fconf |= F_VNIC_ID; if (mode & T4_FILTER_PORT) fconf |= F_PORT; if (mode & T4_FILTER_FCoE) fconf |= F_FCOE; return (fconf); } static uint32_t fspec_to_fconf(struct t4_filter_specification *fs) { uint32_t fconf = 0; if (fs->val.frag || fs->mask.frag) fconf |= F_FRAGMENTATION; if (fs->val.matchtype || fs->mask.matchtype) fconf |= F_MPSHITTYPE; if (fs->val.macidx || fs->mask.macidx) fconf |= F_MACMATCH; if (fs->val.ethtype || fs->mask.ethtype) fconf |= F_ETHERTYPE; if (fs->val.proto || fs->mask.proto) fconf |= F_PROTOCOL; if (fs->val.tos || fs->mask.tos) fconf |= F_TOS; if (fs->val.ivlan_vld || fs->mask.ivlan_vld) fconf |= F_VLAN; if (fs->val.ovlan_vld || fs->mask.ovlan_vld) fconf |= F_VNIC_ID; if (fs->val.iport || fs->mask.iport) fconf |= F_PORT; if (fs->val.fcoe || fs->mask.fcoe) fconf |= F_FCOE; return (fconf); } static int get_filter_mode(struct adapter *sc, uint32_t *mode) { uint32_t fconf; t4_read_indirect(sc, A_TP_PIO_ADDR, A_TP_PIO_DATA, &fconf, 1, A_TP_VLAN_PRI_MAP); *mode = fconf_to_mode(fconf); return (0); } static int set_filter_mode(struct adapter *sc, uint32_t mode) { uint32_t fconf; int rc; fconf = mode_to_fconf(mode); ADAPTER_LOCK(sc); if (IS_BUSY(sc)) { rc = EAGAIN; goto done; } if (sc->tids.ftids_in_use > 0) { rc = EBUSY; goto done; } rc = -t4_set_filter_mode(sc, fconf); done: ADAPTER_UNLOCK(sc); return (rc); } static inline uint64_t get_filter_hits(struct adapter *sc, uint32_t fid) { uint32_t tcb_base = t4_read_reg(sc, A_TP_CMM_TCB_BASE); uint64_t hits; t4_write_reg(sc, PCIE_MEM_ACCESS_REG(A_PCIE_MEM_ACCESS_OFFSET, 0), tcb_base + (fid + sc->tids.ftid_base) * TCB_SIZE); t4_read_reg(sc, PCIE_MEM_ACCESS_REG(A_PCIE_MEM_ACCESS_OFFSET, 0)); hits = t4_read_reg64(sc, MEMWIN0_BASE + 16); return (be64toh(hits)); } static int get_filter(struct adapter *sc, struct t4_filter *t) { int i, nfilters = sc->tids.nftids; struct filter_entry *f; ADAPTER_LOCK_ASSERT_OWNED(sc); if (IS_BUSY(sc)) return (EAGAIN); if (sc->tids.ftids_in_use == 0 || sc->tids.ftid_tab == NULL || t->idx >= nfilters) { t->idx = 0xffffffff; return (0); } f = &sc->tids.ftid_tab[t->idx]; for (i = t->idx; i < nfilters; i++, f++) { if (f->valid) { t->idx = i; t->l2tidx = f->l2t ? f->l2t->idx : 0; t->smtidx = f->smtidx; if (f->fs.hitcnts) t->hits = get_filter_hits(sc, t->idx); else t->hits = UINT64_MAX; t->fs = f->fs; return (0); } } t->idx = 0xffffffff; return (0); } static int set_filter(struct adapter *sc, struct t4_filter *t) { uint32_t fconf; unsigned int nfilters, nports; struct filter_entry *f; int i; ADAPTER_LOCK_ASSERT_OWNED(sc); nfilters = sc->tids.nftids; nports = sc->params.nports; if (nfilters == 0) return (ENOTSUP); if (!(sc->flags & FULL_INIT_DONE)) return (EAGAIN); if (t->idx >= nfilters) return (EINVAL); /* Validate against the global filter mode */ t4_read_indirect(sc, A_TP_PIO_ADDR, A_TP_PIO_DATA, &fconf, 1, A_TP_VLAN_PRI_MAP); if ((fconf | fspec_to_fconf(&t->fs)) != fconf) return (E2BIG); if (t->fs.action == FILTER_SWITCH && t->fs.eport >= nports) return (EINVAL); if (t->fs.val.iport >= nports) return (EINVAL); /* Can't specify an iq if not steering to it */ if (!t->fs.dirsteer && t->fs.iq) return (EINVAL); /* IPv6 filter idx must be 4 aligned */ if (t->fs.type == 1 && ((t->idx & 0x3) || t->idx + 4 >= nfilters)) return (EINVAL); if (sc->tids.ftid_tab == NULL) { KASSERT(sc->tids.ftids_in_use == 0, ("%s: no memory allocated but filters_in_use > 0", __func__)); sc->tids.ftid_tab = malloc(sizeof (struct filter_entry) * nfilters, M_CXGBE, M_NOWAIT | M_ZERO); if (sc->tids.ftid_tab == NULL) return (ENOMEM); } for (i = 0; i < 4; i++) { f = &sc->tids.ftid_tab[t->idx + i]; if (f->pending || f->valid) return (EBUSY); if (f->locked) return (EPERM); if (t->fs.type == 0) break; } f = &sc->tids.ftid_tab[t->idx]; f->fs = t->fs; return set_filter_wr(sc, t->idx); } static int del_filter(struct adapter *sc, struct t4_filter *t) { unsigned int nfilters; struct filter_entry *f; ADAPTER_LOCK_ASSERT_OWNED(sc); if (IS_BUSY(sc)) return (EAGAIN); nfilters = sc->tids.nftids; if (nfilters == 0) return (ENOTSUP); if (sc->tids.ftid_tab == NULL || sc->tids.ftids_in_use == 0 || t->idx >= nfilters) return (EINVAL); if (!(sc->flags & FULL_INIT_DONE)) return (EAGAIN); f = &sc->tids.ftid_tab[t->idx]; if (f->pending) return (EBUSY); if (f->locked) return (EPERM); if (f->valid) { t->fs = f->fs; /* extra info for the caller */ return del_filter_wr(sc, t->idx); } return (0); } static void clear_filter(struct filter_entry *f) { if (f->l2t) t4_l2t_release(f->l2t); bzero(f, sizeof (*f)); } static int set_filter_wr(struct adapter *sc, int fidx) { int rc; struct filter_entry *f = &sc->tids.ftid_tab[fidx]; struct mbuf *m; struct fw_filter_wr *fwr; unsigned int ftid; ADAPTER_LOCK_ASSERT_OWNED(sc); if (f->fs.newdmac || f->fs.newvlan) { /* This filter needs an L2T entry; allocate one. */ f->l2t = t4_l2t_alloc_switching(sc->l2t); if (f->l2t == NULL) return (EAGAIN); if (t4_l2t_set_switching(sc, f->l2t, f->fs.vlan, f->fs.eport, f->fs.dmac)) { t4_l2t_release(f->l2t); f->l2t = NULL; return (ENOMEM); } } ftid = sc->tids.ftid_base + fidx; m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) return (ENOMEM); fwr = mtod(m, struct fw_filter_wr *); m->m_len = m->m_pkthdr.len = sizeof(*fwr); bzero(fwr, sizeof (*fwr)); fwr->op_pkd = htobe32(V_FW_WR_OP(FW_FILTER_WR)); fwr->len16_pkd = htobe32(FW_LEN16(*fwr)); fwr->tid_to_iq = htobe32(V_FW_FILTER_WR_TID(ftid) | V_FW_FILTER_WR_RQTYPE(f->fs.type) | V_FW_FILTER_WR_NOREPLY(0) | V_FW_FILTER_WR_IQ(f->fs.iq)); fwr->del_filter_to_l2tix = htobe32(V_FW_FILTER_WR_RPTTID(f->fs.rpttid) | V_FW_FILTER_WR_DROP(f->fs.action == FILTER_DROP) | V_FW_FILTER_WR_DIRSTEER(f->fs.dirsteer) | V_FW_FILTER_WR_MASKHASH(f->fs.maskhash) | V_FW_FILTER_WR_DIRSTEERHASH(f->fs.dirsteerhash) | V_FW_FILTER_WR_LPBK(f->fs.action == FILTER_SWITCH) | V_FW_FILTER_WR_DMAC(f->fs.newdmac) | V_FW_FILTER_WR_SMAC(f->fs.newsmac) | V_FW_FILTER_WR_INSVLAN(f->fs.newvlan == VLAN_INSERT || f->fs.newvlan == VLAN_REWRITE) | V_FW_FILTER_WR_RMVLAN(f->fs.newvlan == VLAN_REMOVE || f->fs.newvlan == VLAN_REWRITE) | V_FW_FILTER_WR_HITCNTS(f->fs.hitcnts) | V_FW_FILTER_WR_TXCHAN(f->fs.eport) | V_FW_FILTER_WR_PRIO(f->fs.prio) | V_FW_FILTER_WR_L2TIX(f->l2t ? f->l2t->idx : 0)); fwr->ethtype = htobe16(f->fs.val.ethtype); fwr->ethtypem = htobe16(f->fs.mask.ethtype); fwr->frag_to_ovlan_vldm = (V_FW_FILTER_WR_FRAG(f->fs.val.frag) | V_FW_FILTER_WR_FRAGM(f->fs.mask.frag) | V_FW_FILTER_WR_IVLAN_VLD(f->fs.val.ivlan_vld) | V_FW_FILTER_WR_OVLAN_VLD(f->fs.val.ovlan_vld) | V_FW_FILTER_WR_IVLAN_VLDM(f->fs.mask.ivlan_vld) | V_FW_FILTER_WR_OVLAN_VLDM(f->fs.mask.ovlan_vld)); fwr->smac_sel = 0; fwr->rx_chan_rx_rpl_iq = htobe16(V_FW_FILTER_WR_RX_CHAN(0) | V_FW_FILTER_WR_RX_RPL_IQ(sc->sge.intrq[0].abs_id)); fwr->maci_to_matchtypem = htobe32(V_FW_FILTER_WR_MACI(f->fs.val.macidx) | V_FW_FILTER_WR_MACIM(f->fs.mask.macidx) | V_FW_FILTER_WR_FCOE(f->fs.val.fcoe) | V_FW_FILTER_WR_FCOEM(f->fs.mask.fcoe) | V_FW_FILTER_WR_PORT(f->fs.val.iport) | V_FW_FILTER_WR_PORTM(f->fs.mask.iport) | V_FW_FILTER_WR_MATCHTYPE(f->fs.val.matchtype) | V_FW_FILTER_WR_MATCHTYPEM(f->fs.mask.matchtype)); fwr->ptcl = f->fs.val.proto; fwr->ptclm = f->fs.mask.proto; fwr->ttyp = f->fs.val.tos; fwr->ttypm = f->fs.mask.tos; fwr->ivlan = htobe16(f->fs.val.ivlan); fwr->ivlanm = htobe16(f->fs.mask.ivlan); fwr->ovlan = htobe16(f->fs.val.ovlan); fwr->ovlanm = htobe16(f->fs.mask.ovlan); bcopy(f->fs.val.dip, fwr->lip, sizeof (fwr->lip)); bcopy(f->fs.mask.dip, fwr->lipm, sizeof (fwr->lipm)); bcopy(f->fs.val.sip, fwr->fip, sizeof (fwr->fip)); bcopy(f->fs.mask.sip, fwr->fipm, sizeof (fwr->fipm)); fwr->lp = htobe16(f->fs.val.dport); fwr->lpm = htobe16(f->fs.mask.dport); fwr->fp = htobe16(f->fs.val.sport); fwr->fpm = htobe16(f->fs.mask.sport); if (f->fs.newsmac) bcopy(f->fs.smac, fwr->sma, sizeof (fwr->sma)); f->pending = 1; sc->tids.ftids_in_use++; rc = t4_mgmt_tx(sc, m); if (rc != 0) { sc->tids.ftids_in_use--; m_freem(m); clear_filter(f); } return (rc); } static int del_filter_wr(struct adapter *sc, int fidx) { struct filter_entry *f = &sc->tids.ftid_tab[fidx]; struct mbuf *m; struct fw_filter_wr *fwr; unsigned int rc, ftid; ADAPTER_LOCK_ASSERT_OWNED(sc); ftid = sc->tids.ftid_base + fidx; m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) return (ENOMEM); fwr = mtod(m, struct fw_filter_wr *); m->m_len = m->m_pkthdr.len = sizeof(*fwr); bzero(fwr, sizeof (*fwr)); t4_mk_filtdelwr(ftid, fwr, sc->sge.intrq[0].abs_id); f->pending = 1; rc = t4_mgmt_tx(sc, m); if (rc != 0) { f->pending = 0; m_freem(m); } return (rc); } /* XXX move intr handlers to main.c and make this static */ void filter_rpl(struct adapter *sc, const struct cpl_set_tcb_rpl *rpl) { unsigned int idx = GET_TID(rpl); if (idx >= sc->tids.ftid_base && (idx -= sc->tids.ftid_base) < sc->tids.nftids) { unsigned int rc = G_COOKIE(rpl->cookie); struct filter_entry *f = &sc->tids.ftid_tab[idx]; if (rc == FW_FILTER_WR_FLT_DELETED) { /* * Clear the filter when we get confirmation from the * hardware that the filter has been deleted. */ clear_filter(f); sc->tids.ftids_in_use--; } else if (rc == FW_FILTER_WR_SMT_TBL_FULL) { device_printf(sc->dev, "filter %u setup failed due to full SMT\n", idx); clear_filter(f); sc->tids.ftids_in_use--; } else if (rc == FW_FILTER_WR_FLT_ADDED) { f->smtidx = (be64toh(rpl->oldval) >> 24) & 0xff; f->pending = 0; /* asynchronous setup completed */ f->valid = 1; } else { /* * Something went wrong. Issue a warning about the * problem and clear everything out. */ device_printf(sc->dev, "filter %u setup failed with error %u\n", idx, rc); clear_filter(f); sc->tids.ftids_in_use--; } } } int t4_os_find_pci_capability(struct adapter *sc, int cap) { int i; return (pci_find_cap(sc->dev, cap, &i) == 0 ? i : 0); } int t4_os_pci_save_state(struct adapter *sc) { device_t dev; struct pci_devinfo *dinfo; dev = sc->dev; dinfo = device_get_ivars(dev); pci_cfg_save(dev, dinfo, 0); return (0); } int t4_os_pci_restore_state(struct adapter *sc) { device_t dev; struct pci_devinfo *dinfo; dev = sc->dev; dinfo = device_get_ivars(dev); pci_cfg_restore(dev, dinfo); return (0); } void t4_os_portmod_changed(const struct adapter *sc, int idx) { struct port_info *pi = sc->port[idx]; static const char *mod_str[] = { NULL, "LR", "SR", "ER", "TWINAX", "active TWINAX", "LRM" }; if (pi->mod_type == FW_PORT_MOD_TYPE_NONE) if_printf(pi->ifp, "transceiver unplugged.\n"); else if (pi->mod_type == FW_PORT_MOD_TYPE_UNKNOWN) if_printf(pi->ifp, "unknown transceiver inserted.\n"); else if (pi->mod_type == FW_PORT_MOD_TYPE_NOTSUPPORTED) if_printf(pi->ifp, "unsupported transceiver inserted.\n"); else if (pi->mod_type > 0 && pi->mod_type < ARRAY_SIZE(mod_str)) { if_printf(pi->ifp, "%s transceiver inserted.\n", mod_str[pi->mod_type]); } else { if_printf(pi->ifp, "transceiver (type %d) inserted.\n", pi->mod_type); } } void t4_os_link_changed(struct adapter *sc, int idx, int link_stat) { struct port_info *pi = sc->port[idx]; struct ifnet *ifp = pi->ifp; if (link_stat) { ifp->if_baudrate = IF_Mbps(pi->link_cfg.speed); if_link_state_change(ifp, LINK_STATE_UP); } else if_link_state_change(ifp, LINK_STATE_DOWN); } static int t4_open(struct cdev *dev, int flags, int type, struct thread *td) { return (0); } static int t4_close(struct cdev *dev, int flags, int type, struct thread *td) { return (0); } static int t4_ioctl(struct cdev *dev, unsigned long cmd, caddr_t data, int fflag, struct thread *td) { int rc; struct adapter *sc = dev->si_drv1; rc = priv_check(td, PRIV_DRIVER); if (rc != 0) return (rc); switch (cmd) { case CHELSIO_T4_GETREG: { struct t4_reg *edata = (struct t4_reg *)data; if ((edata->addr & 0x3) != 0 || edata->addr >= sc->mmio_len) return (EFAULT); if (edata->size == 4) edata->val = t4_read_reg(sc, edata->addr); else if (edata->size == 8) edata->val = t4_read_reg64(sc, edata->addr); else return (EINVAL); break; } case CHELSIO_T4_SETREG: { struct t4_reg *edata = (struct t4_reg *)data; if ((edata->addr & 0x3) != 0 || edata->addr >= sc->mmio_len) return (EFAULT); if (edata->size == 4) { if (edata->val & 0xffffffff00000000) return (EINVAL); t4_write_reg(sc, edata->addr, (uint32_t) edata->val); } else if (edata->size == 8) t4_write_reg64(sc, edata->addr, edata->val); else return (EINVAL); break; } case CHELSIO_T4_REGDUMP: { struct t4_regdump *regs = (struct t4_regdump *)data; int reglen = T4_REGDUMP_SIZE; uint8_t *buf; if (regs->len < reglen) { regs->len = reglen; /* hint to the caller */ return (ENOBUFS); } regs->len = reglen; buf = malloc(reglen, M_CXGBE, M_WAITOK | M_ZERO); t4_get_regs(sc, regs, buf); rc = copyout(buf, regs->data, reglen); free(buf, M_CXGBE); break; } case CHELSIO_T4_GET_FILTER_MODE: rc = get_filter_mode(sc, (uint32_t *)data); break; case CHELSIO_T4_SET_FILTER_MODE: rc = set_filter_mode(sc, *(uint32_t *)data); break; case CHELSIO_T4_GET_FILTER: ADAPTER_LOCK(sc); rc = get_filter(sc, (struct t4_filter *)data); ADAPTER_UNLOCK(sc); break; case CHELSIO_T4_SET_FILTER: ADAPTER_LOCK(sc); rc = set_filter(sc, (struct t4_filter *)data); ADAPTER_UNLOCK(sc); break; case CHELSIO_T4_DEL_FILTER: ADAPTER_LOCK(sc); rc = del_filter(sc, (struct t4_filter *)data); ADAPTER_UNLOCK(sc); break; default: rc = EINVAL; } return (rc); } static int t4_mod_event(module_t mod, int cmd, void *arg) { if (cmd == MOD_LOAD) t4_sge_modload(); return (0); } static devclass_t t4_devclass; static devclass_t cxgbe_devclass; DRIVER_MODULE(t4nex, pci, t4_driver, t4_devclass, t4_mod_event, 0); MODULE_VERSION(t4nex, 1); DRIVER_MODULE(cxgbe, t4nex, cxgbe_driver, cxgbe_devclass, 0, 0); MODULE_VERSION(cxgbe, 1); Index: projects/largeSMP/sys/dev/cxgbe/t4_sge.c =================================================================== --- projects/largeSMP/sys/dev/cxgbe/t4_sge.c (revision 222782) +++ projects/largeSMP/sys/dev/cxgbe/t4_sge.c (revision 222783) @@ -1,2825 +1,2833 @@ /*- * Copyright (c) 2011 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 #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" #include "common/t4_msg.h" #include "common/t4fw_interface.h" struct fl_buf_info { int size; int type; uma_zone_t zone; }; /* Filled up by t4_sge_modload */ static struct fl_buf_info fl_buf_info[FL_BUF_SIZES]; #define FL_BUF_SIZE(x) (fl_buf_info[x].size) #define FL_BUF_TYPE(x) (fl_buf_info[x].type) #define FL_BUF_ZONE(x) (fl_buf_info[x].zone) enum { FL_PKTSHIFT = 2 }; #define FL_ALIGN min(CACHE_LINE_SIZE, 32) #if CACHE_LINE_SIZE > 64 #define SPG_LEN 128 #else #define SPG_LEN 64 #endif /* Used to track coalesced tx work request */ struct txpkts { uint64_t *flitp; /* ptr to flit where next pkt should start */ uint8_t npkt; /* # of packets in this work request */ uint8_t nflits; /* # of flits used by this work request */ uint16_t plen; /* total payload (sum of all packets) */ }; /* A packet's SGL. This + m_pkthdr has all info needed for tx */ struct sgl { int nsegs; /* # of segments in the SGL, 0 means imm. tx */ int nflits; /* # of flits needed for the SGL */ bus_dma_segment_t seg[TX_SGL_SEGS]; }; static void t4_evt_rx(void *); static void t4_eth_rx(void *); static inline void init_iq(struct sge_iq *, struct adapter *, int, int, int, int, iq_intr_handler_t *, char *); static inline void init_fl(struct sge_fl *, int, char *); static inline void init_eq(struct sge_eq *, int, char *); static int alloc_ring(struct adapter *, size_t, bus_dma_tag_t *, bus_dmamap_t *, bus_addr_t *, void **); static int free_ring(struct adapter *, bus_dma_tag_t, bus_dmamap_t, bus_addr_t, void *); static int alloc_iq_fl(struct port_info *, struct sge_iq *, struct sge_fl *, int, int); static int free_iq_fl(struct port_info *, struct sge_iq *, struct sge_fl *); static int alloc_intrq(struct adapter *, int, int, int); static int free_intrq(struct sge_iq *); static int alloc_fwq(struct adapter *, int); static int free_fwq(struct sge_iq *); static int alloc_rxq(struct port_info *, struct sge_rxq *, int, int); static int free_rxq(struct port_info *, struct sge_rxq *); static int alloc_ctrlq(struct adapter *, struct sge_ctrlq *, int); static int free_ctrlq(struct adapter *, struct sge_ctrlq *); static int alloc_txq(struct port_info *, struct sge_txq *, int); static int free_txq(struct port_info *, struct sge_txq *); static void oneseg_dma_callback(void *, bus_dma_segment_t *, int, int); static inline bool is_new_response(const struct sge_iq *, struct rsp_ctrl **); static inline void iq_next(struct sge_iq *); static inline void ring_fl_db(struct adapter *, struct sge_fl *); static void refill_fl(struct adapter *, struct sge_fl *, int, int); static int alloc_fl_sdesc(struct sge_fl *); static void free_fl_sdesc(struct sge_fl *); static int alloc_tx_maps(struct sge_txq *); static void free_tx_maps(struct sge_txq *); static void set_fl_tag_idx(struct sge_fl *, int); static int get_pkt_sgl(struct sge_txq *, struct mbuf **, struct sgl *, int); static int free_pkt_sgl(struct sge_txq *, struct sgl *); static int write_txpkt_wr(struct port_info *, struct sge_txq *, struct mbuf *, struct sgl *); static int add_to_txpkts(struct port_info *, struct sge_txq *, struct txpkts *, struct mbuf *, struct sgl *); static void write_txpkts_wr(struct sge_txq *, struct txpkts *); static inline void write_ulp_cpl_sgl(struct port_info *, struct sge_txq *, struct txpkts *, struct mbuf *, struct sgl *); static int write_sgl_to_txd(struct sge_eq *, struct sgl *, caddr_t *); static inline void copy_to_txd(struct sge_eq *, caddr_t, caddr_t *, int); static inline void ring_eq_db(struct adapter *, struct sge_eq *); static inline int reclaimable(struct sge_eq *); static int reclaim_tx_descs(struct sge_txq *, int, int); static void write_eqflush_wr(struct sge_eq *); static __be64 get_flit(bus_dma_segment_t *, int, int); static int handle_sge_egr_update(struct adapter *, const struct cpl_sge_egr_update *); static void handle_cpl(struct adapter *, struct sge_iq *); static int ctrl_tx(struct adapter *, struct sge_ctrlq *, struct mbuf *); static int sysctl_uint16(SYSCTL_HANDLER_ARGS); extern void filter_rpl(struct adapter *, const struct cpl_set_tcb_rpl *); /* * Called on MOD_LOAD and fills up fl_buf_info[]. */ void t4_sge_modload(void) { int i; int bufsize[FL_BUF_SIZES] = { MCLBYTES, #if MJUMPAGESIZE != MCLBYTES MJUMPAGESIZE, #endif MJUM9BYTES, MJUM16BYTES }; for (i = 0; i < FL_BUF_SIZES; i++) { FL_BUF_SIZE(i) = bufsize[i]; FL_BUF_TYPE(i) = m_gettype(bufsize[i]); FL_BUF_ZONE(i) = m_getzone(bufsize[i]); } } /** * t4_sge_init - initialize SGE * @sc: the adapter * * Performs SGE initialization needed every time after a chip reset. * We do not initialize any of the queues here, instead the driver * top-level must request them individually. */ void t4_sge_init(struct adapter *sc) { struct sge *s = &sc->sge; int i; t4_set_reg_field(sc, A_SGE_CONTROL, V_PKTSHIFT(M_PKTSHIFT) | V_INGPADBOUNDARY(M_INGPADBOUNDARY) | F_EGRSTATUSPAGESIZE, V_INGPADBOUNDARY(ilog2(FL_ALIGN) - 5) | V_PKTSHIFT(FL_PKTSHIFT) | F_RXPKTCPLMODE | V_EGRSTATUSPAGESIZE(SPG_LEN == 128)); t4_set_reg_field(sc, A_SGE_HOST_PAGE_SIZE, V_HOSTPAGESIZEPF0(M_HOSTPAGESIZEPF0), V_HOSTPAGESIZEPF0(PAGE_SHIFT - 10)); for (i = 0; i < FL_BUF_SIZES; i++) { t4_write_reg(sc, A_SGE_FL_BUFFER_SIZE0 + (4 * i), FL_BUF_SIZE(i)); } + i = t4_read_reg(sc, A_SGE_CONM_CTRL); + s->fl_starve_threshold = G_EGRTHRESHOLD(i) * 2 + 1; + t4_write_reg(sc, A_SGE_INGRESS_RX_THRESHOLD, V_THRESHOLD_0(s->counter_val[0]) | V_THRESHOLD_1(s->counter_val[1]) | V_THRESHOLD_2(s->counter_val[2]) | V_THRESHOLD_3(s->counter_val[3])); t4_write_reg(sc, A_SGE_TIMER_VALUE_0_AND_1, V_TIMERVALUE0(us_to_core_ticks(sc, s->timer_val[0])) | V_TIMERVALUE1(us_to_core_ticks(sc, s->timer_val[1]))); t4_write_reg(sc, A_SGE_TIMER_VALUE_2_AND_3, V_TIMERVALUE2(us_to_core_ticks(sc, s->timer_val[2])) | V_TIMERVALUE3(us_to_core_ticks(sc, s->timer_val[3]))); t4_write_reg(sc, A_SGE_TIMER_VALUE_4_AND_5, V_TIMERVALUE4(us_to_core_ticks(sc, s->timer_val[4])) | V_TIMERVALUE5(us_to_core_ticks(sc, s->timer_val[5]))); } int t4_create_dma_tag(struct adapter *sc) { int rc; rc = bus_dma_tag_create(bus_get_dma_tag(sc->dev), 1, 0, BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, BUS_SPACE_MAXSIZE, BUS_SPACE_UNRESTRICTED, BUS_SPACE_MAXSIZE, BUS_DMA_ALLOCNOW, NULL, NULL, &sc->dmat); if (rc != 0) { device_printf(sc->dev, "failed to create main DMA tag: %d\n", rc); } return (rc); } int t4_destroy_dma_tag(struct adapter *sc) { if (sc->dmat) bus_dma_tag_destroy(sc->dmat); return (0); } /* * Allocate and initialize the firmware event queue, control queues, and the * interrupt queues. The adapter owns all of these queues. * * Returns errno on failure. Resources allocated up to that point may still be * allocated. Caller is responsible for cleanup in case this function fails. */ int t4_setup_adapter_queues(struct adapter *sc) { int i, j, rc, intr_idx, qsize; struct sge_iq *iq; struct sge_ctrlq *ctrlq; iq_intr_handler_t *handler; char name[16]; ADAPTER_LOCK_ASSERT_NOTOWNED(sc); if (sysctl_ctx_init(&sc->ctx) == 0) { struct sysctl_oid *oid = device_get_sysctl_tree(sc->dev); struct sysctl_oid_list *children = SYSCTL_CHILDREN(oid); sc->oid_fwq = SYSCTL_ADD_NODE(&sc->ctx, children, OID_AUTO, "fwq", CTLFLAG_RD, NULL, "firmware event queue"); sc->oid_ctrlq = SYSCTL_ADD_NODE(&sc->ctx, children, OID_AUTO, "ctrlq", CTLFLAG_RD, NULL, "ctrl queues"); sc->oid_intrq = SYSCTL_ADD_NODE(&sc->ctx, children, OID_AUTO, "intrq", CTLFLAG_RD, NULL, "interrupt queues"); } /* * Interrupt queues */ intr_idx = sc->intr_count - NINTRQ(sc); if (sc->flags & INTR_SHARED) { qsize = max((sc->sge.nrxq + 1) * 2, INTR_IQ_QSIZE); for (i = 0; i < NINTRQ(sc); i++, intr_idx++) { snprintf(name, sizeof(name), "%s intrq%d", device_get_nameunit(sc->dev), i); iq = &sc->sge.intrq[i]; init_iq(iq, sc, 0, 0, qsize, INTR_IQ_ESIZE, NULL, name); rc = alloc_intrq(sc, i % sc->params.nports, i, intr_idx); if (rc != 0) { device_printf(sc->dev, "failed to create %s: %d\n", name, rc); return (rc); } } } else { int qidx = 0; struct port_info *pi; for (i = 0; i < sc->params.nports; i++) { pi = sc->port[i]; qsize = max((pi->nrxq + 1) * 2, INTR_IQ_QSIZE); for (j = 0; j < pi->nrxq; j++, qidx++, intr_idx++) { snprintf(name, sizeof(name), "%s intrq%d", device_get_nameunit(pi->dev), j); iq = &sc->sge.intrq[qidx]; init_iq(iq, sc, 0, 0, qsize, INTR_IQ_ESIZE, NULL, name); rc = alloc_intrq(sc, i, qidx, intr_idx); if (rc != 0) { device_printf(sc->dev, "failed to create %s: %d\n", name, rc); return (rc); } } } } /* * Firmware event queue */ snprintf(name, sizeof(name), "%s fwq", device_get_nameunit(sc->dev)); if (sc->intr_count > T4_EXTRA_INTR) { handler = NULL; intr_idx = 1; } else { handler = t4_evt_rx; intr_idx = 0; } iq = &sc->sge.fwq; init_iq(iq, sc, 0, 0, FW_IQ_QSIZE, FW_IQ_ESIZE, handler, name); rc = alloc_fwq(sc, intr_idx); if (rc != 0) { device_printf(sc->dev, "failed to create firmware event queue: %d\n", rc); return (rc); } /* * Control queues - one per port. */ ctrlq = &sc->sge.ctrlq[0]; for (i = 0; i < sc->params.nports; i++, ctrlq++) { snprintf(name, sizeof(name), "%s ctrlq%d", device_get_nameunit(sc->dev), i); init_eq(&ctrlq->eq, CTRL_EQ_QSIZE, name); rc = alloc_ctrlq(sc, ctrlq, i); if (rc != 0) { device_printf(sc->dev, "failed to create control queue %d: %d\n", i, rc); return (rc); } } return (rc); } /* * Idempotent */ int t4_teardown_adapter_queues(struct adapter *sc) { int i; struct sge_iq *iq; ADAPTER_LOCK_ASSERT_NOTOWNED(sc); /* Do this before freeing the queues */ if (sc->oid_fwq || sc->oid_ctrlq || sc->oid_intrq) { sysctl_ctx_free(&sc->ctx); sc->oid_fwq = NULL; sc->oid_ctrlq = NULL; sc->oid_intrq = NULL; } for (i = 0; i < sc->params.nports; i++) free_ctrlq(sc, &sc->sge.ctrlq[i]); iq = &sc->sge.fwq; free_fwq(iq); for (i = 0; i < NINTRQ(sc); i++) { iq = &sc->sge.intrq[i]; free_intrq(iq); } return (0); } int t4_setup_eth_queues(struct port_info *pi) { int rc = 0, i, intr_idx; struct sge_rxq *rxq; struct sge_txq *txq; char name[16]; struct adapter *sc = pi->adapter; if (sysctl_ctx_init(&pi->ctx) == 0) { struct sysctl_oid *oid = device_get_sysctl_tree(pi->dev); struct sysctl_oid_list *children = SYSCTL_CHILDREN(oid); pi->oid_rxq = SYSCTL_ADD_NODE(&pi->ctx, children, OID_AUTO, "rxq", CTLFLAG_RD, NULL, "rx queues"); pi->oid_txq = SYSCTL_ADD_NODE(&pi->ctx, children, OID_AUTO, "txq", CTLFLAG_RD, NULL, "tx queues"); } for_each_rxq(pi, i, rxq) { snprintf(name, sizeof(name), "%s rxq%d-iq", device_get_nameunit(pi->dev), i); init_iq(&rxq->iq, sc, pi->tmr_idx, pi->pktc_idx, pi->qsize_rxq, RX_IQ_ESIZE, t4_eth_rx, name); snprintf(name, sizeof(name), "%s rxq%d-fl", device_get_nameunit(pi->dev), i); init_fl(&rxq->fl, pi->qsize_rxq / 8, name); intr_idx = pi->first_rxq + i; if (sc->flags & INTR_SHARED) intr_idx %= NINTRQ(sc); rc = alloc_rxq(pi, rxq, intr_idx, i); if (rc != 0) goto done; } for_each_txq(pi, i, txq) { snprintf(name, sizeof(name), "%s txq%d", device_get_nameunit(pi->dev), i); init_eq(&txq->eq, pi->qsize_txq, name); rc = alloc_txq(pi, txq, i); if (rc != 0) goto done; } done: if (rc) t4_teardown_eth_queues(pi); return (rc); } /* * Idempotent */ int t4_teardown_eth_queues(struct port_info *pi) { int i; struct sge_rxq *rxq; struct sge_txq *txq; /* Do this before freeing the queues */ if (pi->oid_txq || pi->oid_rxq) { sysctl_ctx_free(&pi->ctx); pi->oid_txq = pi->oid_rxq = NULL; } for_each_txq(pi, i, txq) { free_txq(pi, txq); } for_each_rxq(pi, i, rxq) { free_rxq(pi, rxq); } return (0); } /* Deals with errors and the first (and only) interrupt queue */ void t4_intr_all(void *arg) { struct adapter *sc = arg; t4_intr_err(arg); t4_intr(&sc->sge.intrq[0]); } /* Deals with interrupts, and a few CPLs, on the given interrupt queue */ void t4_intr(void *arg) { struct sge_iq *iq = arg, *q; struct adapter *sc = iq->adapter; struct rsp_ctrl *ctrl; const struct rss_header *rss; int ndesc_pending = 0, ndesc_total = 0; int qid, rsp_type; if (!atomic_cmpset_32(&iq->state, IQS_IDLE, IQS_BUSY)) return; while (is_new_response(iq, &ctrl)) { rmb(); rss = (const void *)iq->cdesc; rsp_type = G_RSPD_TYPE(ctrl->u.type_gen); if (__predict_false(rsp_type == X_RSPD_TYPE_CPL)) { handle_cpl(sc, iq); goto nextdesc; } qid = ntohl(ctrl->pldbuflen_qid) - sc->sge.iq_start; q = sc->sge.iqmap[qid]; if (atomic_cmpset_32(&q->state, IQS_IDLE, IQS_BUSY)) { q->handler(q); atomic_cmpset_32(&q->state, IQS_BUSY, IQS_IDLE); } nextdesc: ndesc_total++; if (++ndesc_pending >= iq->qsize / 4) { t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(ndesc_pending) | V_INGRESSQID(iq->cntxt_id) | V_SEINTARM( V_QINTR_TIMER_IDX(X_TIMERREG_UPDATE_CIDX))); ndesc_pending = 0; } iq_next(iq); } t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(ndesc_pending) | V_INGRESSQID((u32)iq->cntxt_id) | V_SEINTARM(iq->intr_params)); atomic_cmpset_32(&iq->state, IQS_BUSY, IQS_IDLE); } /* Deals with error interrupts */ void t4_intr_err(void *arg) { struct adapter *sc = arg; t4_write_reg(sc, MYPF_REG(A_PCIE_PF_CLI), 0); t4_slow_intr_handler(sc); } /* Deals with the firmware event queue */ void t4_intr_evt(void *arg) { struct sge_iq *iq = arg; if (atomic_cmpset_32(&iq->state, IQS_IDLE, IQS_BUSY)) { t4_evt_rx(arg); atomic_cmpset_32(&iq->state, IQS_BUSY, IQS_IDLE); } } static void t4_evt_rx(void *arg) { struct sge_iq *iq = arg; struct adapter *sc = iq->adapter; struct rsp_ctrl *ctrl; int ndesc_pending = 0, ndesc_total = 0; KASSERT(iq == &sc->sge.fwq, ("%s: unexpected ingress queue", __func__)); while (is_new_response(iq, &ctrl)) { int rsp_type; rmb(); rsp_type = G_RSPD_TYPE(ctrl->u.type_gen); if (__predict_false(rsp_type != X_RSPD_TYPE_CPL)) panic("%s: unexpected rsp_type %d", __func__, rsp_type); handle_cpl(sc, iq); ndesc_total++; if (++ndesc_pending >= iq->qsize / 4) { t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(ndesc_pending) | V_INGRESSQID(iq->cntxt_id) | V_SEINTARM( V_QINTR_TIMER_IDX(X_TIMERREG_UPDATE_CIDX))); ndesc_pending = 0; } iq_next(iq); } t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(ndesc_pending) | V_INGRESSQID(iq->cntxt_id) | V_SEINTARM(iq->intr_params)); } #ifdef T4_PKT_TIMESTAMP #define RX_COPY_THRESHOLD (MINCLSIZE - 8) #else #define RX_COPY_THRESHOLD MINCLSIZE #endif static void t4_eth_rx(void *arg) { struct sge_rxq *rxq = arg; struct sge_iq *iq = arg; struct adapter *sc = iq->adapter; struct rsp_ctrl *ctrl; struct ifnet *ifp = rxq->ifp; struct sge_fl *fl = &rxq->fl; struct fl_sdesc *sd = &fl->sdesc[fl->cidx], *sd_next; const struct rss_header *rss; const struct cpl_rx_pkt *cpl; uint32_t len; int ndescs = 0, i; struct mbuf *m0, *m; #ifdef INET struct lro_ctrl *lro = &rxq->lro; struct lro_entry *l; #endif prefetch(sd->m); prefetch(sd->cl); iq->intr_next = iq->intr_params; while (is_new_response(iq, &ctrl)) { rmb(); rss = (const void *)iq->cdesc; i = G_RSPD_TYPE(ctrl->u.type_gen); KASSERT(i == X_RSPD_TYPE_FLBUF && rss->opcode == CPL_RX_PKT, ("%s: unexpected type %d CPL opcode 0x%x", __func__, i, rss->opcode)); sd_next = sd + 1; if (__predict_false(fl->cidx + 1 == fl->cap)) sd_next = fl->sdesc; prefetch(sd_next->m); prefetch(sd_next->cl); cpl = (const void *)(rss + 1); m0 = sd->m; sd->m = NULL; /* consumed */ len = be32toh(ctrl->pldbuflen_qid); if (__predict_false((len & F_RSPD_NEWBUF) == 0)) panic("%s: cannot handle packed frames", __func__); len = G_RSPD_LEN(len); bus_dmamap_sync(fl->tag[sd->tag_idx], sd->map, BUS_DMASYNC_POSTREAD); m_init(m0, NULL, 0, M_NOWAIT, MT_DATA, M_PKTHDR); #ifdef T4_PKT_TIMESTAMP *mtod(m0, uint64_t *) = be64toh(ctrl->u.last_flit & 0xfffffffffffffff); m0->m_data += 8; /* * 60 bit timestamp value is *(uint64_t *)m0->m_pktdat. Note * that it is in the leading free-space (see M_LEADINGSPACE) in * the mbuf. The kernel can clobber it during a pullup, * m_copymdata, etc. You need to make sure that the mbuf * reaches you unmolested if you care about the timestamp. */ #endif if (len < RX_COPY_THRESHOLD) { /* copy data to mbuf, buffer will be recycled */ bcopy(sd->cl, mtod(m0, caddr_t), len); m0->m_len = len; } else { bus_dmamap_unload(fl->tag[sd->tag_idx], sd->map); m_cljset(m0, sd->cl, FL_BUF_TYPE(sd->tag_idx)); sd->cl = NULL; /* consumed */ m0->m_len = min(len, FL_BUF_SIZE(sd->tag_idx)); } len -= FL_PKTSHIFT; m0->m_len -= FL_PKTSHIFT; m0->m_data += FL_PKTSHIFT; m0->m_pkthdr.len = len; m0->m_pkthdr.rcvif = ifp; m0->m_flags |= M_FLOWID; m0->m_pkthdr.flowid = rss->hash_val; if (cpl->csum_calc && !cpl->err_vec && ifp->if_capenable & IFCAP_RXCSUM) { m0->m_pkthdr.csum_flags |= (CSUM_IP_CHECKED | CSUM_IP_VALID | CSUM_DATA_VALID | CSUM_PSEUDO_HDR); if (cpl->ip_frag) m0->m_pkthdr.csum_data = be16toh(cpl->csum); else m0->m_pkthdr.csum_data = 0xffff; rxq->rxcsum++; } if (cpl->vlan_ex) { m0->m_pkthdr.ether_vtag = be16toh(cpl->vlan); m0->m_flags |= M_VLANTAG; rxq->vlan_extraction++; } i = 1; /* # of fl sdesc used */ sd = sd_next; if (__predict_false(++fl->cidx == fl->cap)) fl->cidx = 0; len -= m0->m_len; m = m0; while (len) { i++; sd_next = sd + 1; if (__predict_false(fl->cidx + 1 == fl->cap)) sd_next = fl->sdesc; prefetch(sd_next->m); prefetch(sd_next->cl); m->m_next = sd->m; sd->m = NULL; /* consumed */ m = m->m_next; bus_dmamap_sync(fl->tag[sd->tag_idx], sd->map, BUS_DMASYNC_POSTREAD); m_init(m, NULL, 0, M_NOWAIT, MT_DATA, 0); if (len <= MLEN) { bcopy(sd->cl, mtod(m, caddr_t), len); m->m_len = len; } else { bus_dmamap_unload(fl->tag[sd->tag_idx], sd->map); m_cljset(m, sd->cl, FL_BUF_TYPE(sd->tag_idx)); sd->cl = NULL; /* consumed */ m->m_len = min(len, FL_BUF_SIZE(sd->tag_idx)); } i++; sd = sd_next; if (__predict_false(++fl->cidx == fl->cap)) fl->cidx = 0; len -= m->m_len; } #ifdef INET if (cpl->l2info & htobe32(F_RXF_LRO) && rxq->flags & RXQ_LRO_ENABLED && tcp_lro_rx(lro, m0, 0) == 0) { /* queued for LRO */ } else #endif ifp->if_input(ifp, m0); FL_LOCK(fl); fl->needed += i; if (fl->needed >= 32) refill_fl(sc, fl, 64, 32); FL_UNLOCK(fl); if (++ndescs > 32) { t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(ndescs) | V_INGRESSQID((u32)iq->cntxt_id) | V_SEINTARM(V_QINTR_TIMER_IDX(X_TIMERREG_UPDATE_CIDX))); ndescs = 0; } iq_next(iq); } #ifdef INET while (!SLIST_EMPTY(&lro->lro_active)) { l = SLIST_FIRST(&lro->lro_active); SLIST_REMOVE_HEAD(&lro->lro_active, next); tcp_lro_flush(lro, l); } #endif t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(ndescs) | V_INGRESSQID((u32)iq->cntxt_id) | V_SEINTARM(iq->intr_next)); FL_LOCK(fl); if (fl->needed >= 32) refill_fl(sc, fl, 128, 8); FL_UNLOCK(fl); } int t4_mgmt_tx(struct adapter *sc, struct mbuf *m) { return ctrl_tx(sc, &sc->sge.ctrlq[0], m); } /* Per-packet header in a coalesced tx WR, before the SGL starts (in flits) */ #define TXPKTS_PKT_HDR ((\ sizeof(struct ulp_txpkt) + \ sizeof(struct ulptx_idata) + \ sizeof(struct cpl_tx_pkt_core) \ ) / 8) /* Header of a coalesced tx WR, before SGL of first packet (in flits) */ #define TXPKTS_WR_HDR (\ sizeof(struct fw_eth_tx_pkts_wr) / 8 + \ TXPKTS_PKT_HDR) /* Header of a tx WR, before SGL of first packet (in flits) */ #define TXPKT_WR_HDR ((\ sizeof(struct fw_eth_tx_pkt_wr) + \ sizeof(struct cpl_tx_pkt_core) \ ) / 8 ) /* Header of a tx LSO WR, before SGL of first packet (in flits) */ #define TXPKT_LSO_WR_HDR ((\ sizeof(struct fw_eth_tx_pkt_wr) + \ sizeof(struct cpl_tx_pkt_lso) + \ sizeof(struct cpl_tx_pkt_core) \ ) / 8 ) int t4_eth_tx(struct ifnet *ifp, struct sge_txq *txq, struct mbuf *m) { struct port_info *pi = (void *)ifp->if_softc; struct adapter *sc = pi->adapter; struct sge_eq *eq = &txq->eq; struct buf_ring *br = txq->br; struct mbuf *next; int rc, coalescing, can_reclaim; struct txpkts txpkts; struct sgl sgl; TXQ_LOCK_ASSERT_OWNED(txq); KASSERT(m, ("%s: called with nothing to do.", __func__)); prefetch(&eq->desc[eq->pidx]); prefetch(&txq->sdesc[eq->pidx]); txpkts.npkt = 0;/* indicates there's nothing in txpkts */ coalescing = 0; if (eq->avail < 8) reclaim_tx_descs(txq, 0, 8); for (; m; m = next ? next : drbr_dequeue(ifp, br)) { if (eq->avail < 8) break; next = m->m_nextpkt; m->m_nextpkt = NULL; if (next || buf_ring_peek(br)) coalescing = 1; rc = get_pkt_sgl(txq, &m, &sgl, coalescing); if (rc != 0) { if (rc == ENOMEM) { /* Short of resources, suspend tx */ m->m_nextpkt = next; break; } /* * Unrecoverable error for this packet, throw it away * and move on to the next. get_pkt_sgl may already * have freed m (it will be NULL in that case and the * m_freem here is still safe). */ m_freem(m); continue; } if (coalescing && add_to_txpkts(pi, txq, &txpkts, m, &sgl) == 0) { /* Successfully absorbed into txpkts */ write_ulp_cpl_sgl(pi, txq, &txpkts, m, &sgl); goto doorbell; } /* * We weren't coalescing to begin with, or current frame could * not be coalesced (add_to_txpkts flushes txpkts if a frame * given to it can't be coalesced). Either way there should be * nothing in txpkts. */ KASSERT(txpkts.npkt == 0, ("%s: txpkts not empty: %d", __func__, txpkts.npkt)); /* We're sending out individual packets now */ coalescing = 0; if (eq->avail < 8) reclaim_tx_descs(txq, 0, 8); rc = write_txpkt_wr(pi, txq, m, &sgl); if (rc != 0) { /* Short of hardware descriptors, suspend tx */ /* * This is an unlikely but expensive failure. We've * done all the hard work (DMA mappings etc.) and now we * can't send out the packet. What's worse, we have to * spend even more time freeing up everything in sgl. */ txq->no_desc++; free_pkt_sgl(txq, &sgl); m->m_nextpkt = next; break; } ETHER_BPF_MTAP(ifp, m); if (sgl.nsegs == 0) m_freem(m); doorbell: /* Fewer and fewer doorbells as the queue fills up */ if (eq->pending >= (1 << (fls(eq->qsize - eq->avail) / 2))) ring_eq_db(sc, eq); can_reclaim = reclaimable(eq); if (can_reclaim >= 32) reclaim_tx_descs(txq, can_reclaim, 32); } if (txpkts.npkt > 0) write_txpkts_wr(txq, &txpkts); /* * m not NULL means there was an error but we haven't thrown it away. * This can happen when we're short of tx descriptors (no_desc) or maybe * even DMA maps (no_dmamap). Either way, a credit flush and reclaim * will get things going again. * * If eq->avail is already 0 we know a credit flush was requested in the * WR that reduced it to 0 so we don't need another flush (we don't have * any descriptor for a flush WR anyway, duh). */ if (m && eq->avail > 0 && !(eq->flags & EQ_CRFLUSHED)) { struct tx_sdesc *txsd = &txq->sdesc[eq->pidx]; txsd->desc_used = 1; txsd->credits = 0; write_eqflush_wr(eq); } txq->m = m; if (eq->pending) ring_eq_db(sc, eq); can_reclaim = reclaimable(eq); if (can_reclaim >= 32) reclaim_tx_descs(txq, can_reclaim, 128); return (0); } void t4_update_fl_bufsize(struct ifnet *ifp) { struct port_info *pi = ifp->if_softc; struct sge_rxq *rxq; struct sge_fl *fl; int i; for_each_rxq(pi, i, rxq) { fl = &rxq->fl; FL_LOCK(fl); set_fl_tag_idx(fl, ifp->if_mtu); FL_UNLOCK(fl); } } /* * A non-NULL handler indicates this iq will not receive direct interrupts, the * handler will be invoked by an interrupt queue. */ static inline void init_iq(struct sge_iq *iq, struct adapter *sc, int tmr_idx, int pktc_idx, int qsize, int esize, iq_intr_handler_t *handler, char *name) { KASSERT(tmr_idx >= 0 && tmr_idx < SGE_NTIMERS, ("%s: bad tmr_idx %d", __func__, tmr_idx)); KASSERT(pktc_idx < SGE_NCOUNTERS, /* -ve is ok, means don't use */ ("%s: bad pktc_idx %d", __func__, pktc_idx)); iq->flags = 0; iq->adapter = sc; iq->intr_params = V_QINTR_TIMER_IDX(tmr_idx) | V_QINTR_CNT_EN(pktc_idx >= 0); iq->intr_pktc_idx = pktc_idx; iq->qsize = roundup(qsize, 16); /* See FW_IQ_CMD/iqsize */ iq->esize = max(esize, 16); /* See FW_IQ_CMD/iqesize */ iq->handler = handler; strlcpy(iq->lockname, name, sizeof(iq->lockname)); } static inline void init_fl(struct sge_fl *fl, int qsize, char *name) { fl->qsize = qsize; strlcpy(fl->lockname, name, sizeof(fl->lockname)); } static inline void init_eq(struct sge_eq *eq, int qsize, char *name) { eq->qsize = qsize; strlcpy(eq->lockname, name, sizeof(eq->lockname)); } static int alloc_ring(struct adapter *sc, size_t len, bus_dma_tag_t *tag, bus_dmamap_t *map, bus_addr_t *pa, void **va) { int rc; rc = bus_dma_tag_create(sc->dmat, 512, 0, BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, len, 1, len, 0, NULL, NULL, tag); if (rc != 0) { device_printf(sc->dev, "cannot allocate DMA tag: %d\n", rc); goto done; } rc = bus_dmamem_alloc(*tag, va, BUS_DMA_WAITOK | BUS_DMA_COHERENT | BUS_DMA_ZERO, map); if (rc != 0) { device_printf(sc->dev, "cannot allocate DMA memory: %d\n", rc); goto done; } rc = bus_dmamap_load(*tag, *map, *va, len, oneseg_dma_callback, pa, 0); if (rc != 0) { device_printf(sc->dev, "cannot load DMA map: %d\n", rc); goto done; } done: if (rc) free_ring(sc, *tag, *map, *pa, *va); return (rc); } static int free_ring(struct adapter *sc, bus_dma_tag_t tag, bus_dmamap_t map, bus_addr_t pa, void *va) { if (pa) bus_dmamap_unload(tag, map); if (va) bus_dmamem_free(tag, va, map); if (tag) bus_dma_tag_destroy(tag); return (0); } /* * Allocates the ring for an ingress queue and an optional freelist. If the * freelist is specified it will be allocated and then associated with the * ingress queue. * * Returns errno on failure. Resources allocated up to that point may still be * allocated. Caller is responsible for cleanup in case this function fails. * * If the ingress queue will take interrupts directly (iq->handler == NULL) then * the intr_idx specifies the vector, starting from 0. Otherwise it specifies * the index of the interrupt queue to which its interrupts will be forwarded. */ static int alloc_iq_fl(struct port_info *pi, struct sge_iq *iq, struct sge_fl *fl, int intr_idx, int cong) { int rc, i, cntxt_id; size_t len; struct fw_iq_cmd c; struct adapter *sc = iq->adapter; __be32 v = 0; len = iq->qsize * iq->esize; rc = alloc_ring(sc, len, &iq->desc_tag, &iq->desc_map, &iq->ba, (void **)&iq->desc); if (rc != 0) return (rc); 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)); /* Special handling for firmware event queue */ if (iq == &sc->sge.fwq) v |= F_FW_IQ_CMD_IQASYNCH; if (iq->handler) { KASSERT(intr_idx < NINTRQ(sc), ("%s: invalid indirect intr_idx %d", __func__, intr_idx)); v |= F_FW_IQ_CMD_IQANDST; v |= V_FW_IQ_CMD_IQANDSTINDEX(sc->sge.intrq[intr_idx].abs_id); } else { KASSERT(intr_idx < sc->intr_count, ("%s: invalid direct intr_idx %d", __func__, intr_idx)); v |= V_FW_IQ_CMD_IQANDSTINDEX(intr_idx); } c.type_to_iqandstindex = htobe32(v | V_FW_IQ_CMD_TYPE(FW_IQ_TYPE_FL_INT_CAP) | V_FW_IQ_CMD_VIID(pi->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(iq->intr_pktc_idx) | V_FW_IQ_CMD_IQESIZE(ilog2(iq->esize) - 4)); c.iqsize = htobe16(iq->qsize); c.iqaddr = htobe64(iq->ba); if (cong >= 0) c.iqns_to_fl0congen = htobe32(F_FW_IQ_CMD_IQFLINTCONGEN); if (fl) { mtx_init(&fl->fl_lock, fl->lockname, NULL, MTX_DEF); for (i = 0; i < FL_BUF_SIZES; i++) { /* * A freelist buffer must be 16 byte aligned as the SGE * uses the low 4 bits of the bus addr to figure out the * buffer size. */ rc = bus_dma_tag_create(sc->dmat, 16, 0, BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, FL_BUF_SIZE(i), 1, FL_BUF_SIZE(i), BUS_DMA_ALLOCNOW, NULL, NULL, &fl->tag[i]); if (rc != 0) { device_printf(sc->dev, "failed to create fl DMA tag[%d]: %d\n", i, rc); return (rc); } } len = fl->qsize * RX_FL_ESIZE; rc = alloc_ring(sc, len, &fl->desc_tag, &fl->desc_map, &fl->ba, (void **)&fl->desc); if (rc) return (rc); /* Allocate space for one software descriptor per buffer. */ fl->cap = (fl->qsize - SPG_LEN / RX_FL_ESIZE) * 8; FL_LOCK(fl); set_fl_tag_idx(fl, pi->ifp->if_mtu); rc = alloc_fl_sdesc(fl); FL_UNLOCK(fl); if (rc != 0) { device_printf(sc->dev, "failed to setup fl software descriptors: %d\n", rc); return (rc); } fl->needed = fl->cap; c.iqns_to_fl0congen = htobe32(V_FW_IQ_CMD_FL0HOSTFCMODE(X_HOSTFCMODE_NONE) | F_FW_IQ_CMD_FL0FETCHRO | F_FW_IQ_CMD_FL0DATARO | F_FW_IQ_CMD_FL0PADEN); if (cong >= 0) { c.iqns_to_fl0congen |= htobe32(V_FW_IQ_CMD_FL0CNGCHMAP(cong) | F_FW_IQ_CMD_FL0CONGCIF | F_FW_IQ_CMD_FL0CONGEN); } c.fl0dcaen_to_fl0cidxfthresh = htobe16(V_FW_IQ_CMD_FL0FBMIN(X_FETCHBURSTMIN_64B) | V_FW_IQ_CMD_FL0FBMAX(X_FETCHBURSTMAX_512B)); c.fl0size = htobe16(fl->qsize); c.fl0addr = htobe64(fl->ba); } rc = -t4_wr_mbox(sc, sc->mbox, &c, sizeof(c), &c); if (rc != 0) { device_printf(sc->dev, "failed to create ingress queue: %d\n", rc); return (rc); } iq->cdesc = iq->desc; iq->cidx = 0; iq->gen = 1; iq->intr_next = iq->intr_params; iq->cntxt_id = be16toh(c.iqid); iq->abs_id = be16toh(c.physiqid); iq->flags |= (IQ_ALLOCATED | IQ_STARTED); cntxt_id = iq->cntxt_id - sc->sge.iq_start; KASSERT(cntxt_id < sc->sge.niq, ("%s: iq->cntxt_id (%d) more than the max (%d)", __func__, cntxt_id, sc->sge.niq - 1)); sc->sge.iqmap[cntxt_id] = iq; if (fl) { fl->cntxt_id = be16toh(c.fl0id); fl->pidx = fl->cidx = 0; cntxt_id = fl->cntxt_id - sc->sge.eq_start; KASSERT(cntxt_id < sc->sge.neq, ("%s: fl->cntxt_id (%d) more than the max (%d)", __func__, cntxt_id, sc->sge.neq - 1)); sc->sge.eqmap[cntxt_id] = (void *)fl; FL_LOCK(fl); - refill_fl(sc, fl, -1, 8); + /* Just enough to make sure it doesn't starve right away. */ + refill_fl(sc, fl, roundup(sc->sge.fl_starve_threshold, 8), 8); FL_UNLOCK(fl); } /* Enable IQ interrupts */ atomic_store_rel_32(&iq->state, IQS_IDLE); t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_SEINTARM(iq->intr_params) | V_INGRESSQID(iq->cntxt_id)); return (0); } /* * This can be called with the iq/fl in any state - fully allocated and * functional, partially allocated, even all-zeroed out. */ static int free_iq_fl(struct port_info *pi, struct sge_iq *iq, struct sge_fl *fl) { int i, rc; struct adapter *sc = iq->adapter; device_t dev; if (sc == NULL) return (0); /* nothing to do */ dev = pi ? pi->dev : sc->dev; if (iq->flags & IQ_STARTED) { rc = -t4_iq_start_stop(sc, sc->mbox, 0, sc->pf, 0, iq->cntxt_id, fl ? fl->cntxt_id : 0xffff, 0xffff); if (rc != 0) { device_printf(dev, "failed to stop queue %p: %d\n", iq, rc); return (rc); } iq->flags &= ~IQ_STARTED; /* Synchronize with the interrupt handler */ while (!atomic_cmpset_32(&iq->state, IQS_IDLE, IQS_DISABLED)) pause("iqfree", hz / 1000); } if (iq->flags & IQ_ALLOCATED) { rc = -t4_iq_free(sc, sc->mbox, sc->pf, 0, FW_IQ_TYPE_FL_INT_CAP, iq->cntxt_id, fl ? fl->cntxt_id : 0xffff, 0xffff); if (rc != 0) { device_printf(dev, "failed to free queue %p: %d\n", iq, rc); return (rc); } iq->flags &= ~IQ_ALLOCATED; } free_ring(sc, iq->desc_tag, iq->desc_map, iq->ba, iq->desc); bzero(iq, sizeof(*iq)); if (fl) { free_ring(sc, fl->desc_tag, fl->desc_map, fl->ba, fl->desc); if (fl->sdesc) { FL_LOCK(fl); free_fl_sdesc(fl); FL_UNLOCK(fl); } if (mtx_initialized(&fl->fl_lock)) mtx_destroy(&fl->fl_lock); for (i = 0; i < FL_BUF_SIZES; i++) { if (fl->tag[i]) bus_dma_tag_destroy(fl->tag[i]); } bzero(fl, sizeof(*fl)); } return (0); } static int alloc_intrq(struct adapter *sc, int port_idx, int intrq_idx, int intr_idx) { int rc; struct sysctl_oid *oid; struct sysctl_oid_list *children; char name[16]; struct sge_iq *intrq = &sc->sge.intrq[intrq_idx]; rc = alloc_iq_fl(sc->port[port_idx], intrq, NULL, intr_idx, -1); if (rc != 0) return (rc); children = SYSCTL_CHILDREN(sc->oid_intrq); snprintf(name, sizeof(name), "%d", intrq_idx); oid = SYSCTL_ADD_NODE(&sc->ctx, children, OID_AUTO, name, CTLFLAG_RD, NULL, "interrupt queue"); children = SYSCTL_CHILDREN(oid); SYSCTL_ADD_PROC(&sc->ctx, children, OID_AUTO, "cidx", CTLTYPE_INT | CTLFLAG_RD, &intrq->cidx, 0, sysctl_uint16, "I", "consumer index"); return (rc); } static int free_intrq(struct sge_iq *iq) { return free_iq_fl(NULL, iq, NULL); } static int alloc_fwq(struct adapter *sc, int intr_idx) { int rc; struct sysctl_oid_list *children; struct sge_iq *fwq = &sc->sge.fwq; rc = alloc_iq_fl(sc->port[0], fwq, NULL, intr_idx, -1); if (rc != 0) return (rc); children = SYSCTL_CHILDREN(sc->oid_fwq); SYSCTL_ADD_PROC(&sc->ctx, children, OID_AUTO, "cidx", CTLTYPE_INT | CTLFLAG_RD, &fwq->cidx, 0, sysctl_uint16, "I", "consumer index"); return (rc); } static int free_fwq(struct sge_iq *iq) { return free_iq_fl(NULL, iq, NULL); } static int alloc_rxq(struct port_info *pi, struct sge_rxq *rxq, int intr_idx, int idx) { int rc; struct sysctl_oid *oid; struct sysctl_oid_list *children; char name[16]; rc = alloc_iq_fl(pi, &rxq->iq, &rxq->fl, intr_idx, 1 << pi->tx_chan); if (rc != 0) return (rc); + + FL_LOCK(&rxq->fl); + refill_fl(pi->adapter, &rxq->fl, rxq->fl.needed / 8, 8); + FL_UNLOCK(&rxq->fl); #ifdef INET rc = tcp_lro_init(&rxq->lro); if (rc != 0) return (rc); rxq->lro.ifp = pi->ifp; /* also indicates LRO init'ed */ if (pi->ifp->if_capenable & IFCAP_LRO) rxq->flags |= RXQ_LRO_ENABLED; #endif rxq->ifp = pi->ifp; children = SYSCTL_CHILDREN(pi->oid_rxq); snprintf(name, sizeof(name), "%d", idx); oid = SYSCTL_ADD_NODE(&pi->ctx, children, OID_AUTO, name, CTLFLAG_RD, NULL, "rx queue"); children = SYSCTL_CHILDREN(oid); SYSCTL_ADD_PROC(&pi->ctx, children, OID_AUTO, "abs_id", CTLTYPE_INT | CTLFLAG_RD, &rxq->iq.abs_id, 0, sysctl_uint16, "I", "absolute id of the queue"); #ifdef INET SYSCTL_ADD_INT(&pi->ctx, children, OID_AUTO, "lro_queued", CTLFLAG_RD, &rxq->lro.lro_queued, 0, NULL); SYSCTL_ADD_INT(&pi->ctx, children, OID_AUTO, "lro_flushed", CTLFLAG_RD, &rxq->lro.lro_flushed, 0, NULL); #endif SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "rxcsum", CTLFLAG_RD, &rxq->rxcsum, "# of times hardware assisted with checksum"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "vlan_extraction", CTLFLAG_RD, &rxq->vlan_extraction, "# of times hardware extracted 802.1Q tag"); return (rc); } static int free_rxq(struct port_info *pi, struct sge_rxq *rxq) { int rc; #ifdef INET if (rxq->lro.ifp) { tcp_lro_free(&rxq->lro); rxq->lro.ifp = NULL; } #endif rc = free_iq_fl(pi, &rxq->iq, &rxq->fl); if (rc == 0) bzero(rxq, sizeof(*rxq)); return (rc); } static int alloc_ctrlq(struct adapter *sc, struct sge_ctrlq *ctrlq, int idx) { int rc, cntxt_id; size_t len; struct fw_eq_ctrl_cmd c; struct sge_eq *eq = &ctrlq->eq; char name[16]; struct sysctl_oid *oid; struct sysctl_oid_list *children; mtx_init(&eq->eq_lock, eq->lockname, NULL, MTX_DEF); len = eq->qsize * CTRL_EQ_ESIZE; rc = alloc_ring(sc, len, &eq->desc_tag, &eq->desc_map, &eq->ba, (void **)&eq->desc); if (rc) return (rc); eq->cap = eq->qsize - SPG_LEN / CTRL_EQ_ESIZE; eq->spg = (void *)&eq->desc[eq->cap]; eq->avail = eq->cap - 1; /* one less to avoid cidx = pidx */ if (sc->flags & INTR_SHARED) eq->iqid = sc->sge.intrq[idx % NINTRQ(sc)].cntxt_id; else eq->iqid = sc->sge.intrq[sc->port[idx]->first_rxq].cntxt_id; bzero(&c, sizeof(c)); c.op_to_vfn = htobe32(V_FW_CMD_OP(FW_EQ_CTRL_CMD) | F_FW_CMD_REQUEST | F_FW_CMD_WRITE | F_FW_CMD_EXEC | V_FW_EQ_CTRL_CMD_PFN(sc->pf) | V_FW_EQ_CTRL_CMD_VFN(0)); c.alloc_to_len16 = htobe32(F_FW_EQ_CTRL_CMD_ALLOC | F_FW_EQ_CTRL_CMD_EQSTART | FW_LEN16(c)); c.cmpliqid_eqid = htonl(V_FW_EQ_CTRL_CMD_CMPLIQID(eq->iqid)); /* XXX */ c.physeqid_pkd = htobe32(0); c.fetchszm_to_iqid = htobe32(V_FW_EQ_CTRL_CMD_HOSTFCMODE(X_HOSTFCMODE_STATUS_PAGE) | V_FW_EQ_CTRL_CMD_PCIECHN(sc->port[idx]->tx_chan) | F_FW_EQ_CTRL_CMD_FETCHRO | V_FW_EQ_CTRL_CMD_IQID(eq->iqid)); c.dcaen_to_eqsize = htobe32(V_FW_EQ_CTRL_CMD_FBMIN(X_FETCHBURSTMIN_64B) | V_FW_EQ_CTRL_CMD_FBMAX(X_FETCHBURSTMAX_512B) | V_FW_EQ_CTRL_CMD_CIDXFTHRESH(X_CIDXFLUSHTHRESH_32) | V_FW_EQ_CTRL_CMD_EQSIZE(eq->qsize)); c.eqaddr = htobe64(eq->ba); rc = -t4_wr_mbox(sc, sc->mbox, &c, sizeof(c), &c); if (rc != 0) { device_printf(sc->dev, "failed to create control queue %d: %d\n", idx, rc); return (rc); } eq->pidx = eq->cidx = 0; eq->cntxt_id = G_FW_EQ_CTRL_CMD_EQID(be32toh(c.cmpliqid_eqid)); eq->flags |= (EQ_ALLOCATED | EQ_STARTED); cntxt_id = eq->cntxt_id - sc->sge.eq_start; KASSERT(cntxt_id < sc->sge.neq, ("%s: eq->cntxt_id (%d) more than the max (%d)", __func__, cntxt_id, sc->sge.neq - 1)); sc->sge.eqmap[cntxt_id] = eq; children = SYSCTL_CHILDREN(sc->oid_ctrlq); snprintf(name, sizeof(name), "%d", idx); oid = SYSCTL_ADD_NODE(&sc->ctx, children, OID_AUTO, name, CTLFLAG_RD, NULL, "ctrl queue"); children = SYSCTL_CHILDREN(oid); SYSCTL_ADD_PROC(&sc->ctx, children, OID_AUTO, "pidx", CTLTYPE_INT | CTLFLAG_RD, &ctrlq->eq.pidx, 0, sysctl_uint16, "I", "producer index"); SYSCTL_ADD_UINT(&sc->ctx, children, OID_AUTO, "no_desc", CTLFLAG_RD, &ctrlq->no_desc, 0, "# of times ctrlq ran out of hardware descriptors"); return (rc); } static int free_ctrlq(struct adapter *sc, struct sge_ctrlq *ctrlq) { int rc; struct sge_eq *eq = &ctrlq->eq; if (eq->flags & (EQ_ALLOCATED | EQ_STARTED)) { rc = -t4_ctrl_eq_free(sc, sc->mbox, sc->pf, 0, eq->cntxt_id); if (rc != 0) { device_printf(sc->dev, "failed to free ctrl queue %p: %d\n", eq, rc); return (rc); } eq->flags &= ~(EQ_ALLOCATED | EQ_STARTED); } free_ring(sc, eq->desc_tag, eq->desc_map, eq->ba, eq->desc); if (mtx_initialized(&eq->eq_lock)) mtx_destroy(&eq->eq_lock); bzero(ctrlq, sizeof(*ctrlq)); return (0); } static int alloc_txq(struct port_info *pi, struct sge_txq *txq, int idx) { int rc, cntxt_id; size_t len; struct adapter *sc = pi->adapter; struct fw_eq_eth_cmd c; struct sge_eq *eq = &txq->eq; char name[16]; struct sysctl_oid *oid; struct sysctl_oid_list *children; struct sge_iq *intrq; txq->ifp = pi->ifp; TASK_INIT(&txq->resume_tx, 0, cxgbe_txq_start, txq); mtx_init(&eq->eq_lock, eq->lockname, NULL, MTX_DEF); len = eq->qsize * TX_EQ_ESIZE; rc = alloc_ring(sc, len, &eq->desc_tag, &eq->desc_map, &eq->ba, (void **)&eq->desc); if (rc) return (rc); eq->cap = eq->qsize - SPG_LEN / TX_EQ_ESIZE; eq->spg = (void *)&eq->desc[eq->cap]; eq->avail = eq->cap - 1; /* one less to avoid cidx = pidx */ txq->sdesc = malloc(eq->cap * sizeof(struct tx_sdesc), M_CXGBE, M_ZERO | M_WAITOK); txq->br = buf_ring_alloc(eq->qsize, M_CXGBE, M_WAITOK, &eq->eq_lock); intrq = &sc->sge.intrq[0]; if (sc->flags & INTR_SHARED) eq->iqid = intrq[(pi->first_txq + idx) % NINTRQ(sc)].cntxt_id; else eq->iqid = intrq[pi->first_rxq + (idx % pi->nrxq)].cntxt_id; rc = bus_dma_tag_create(sc->dmat, 1, 0, BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, 64 * 1024, TX_SGL_SEGS, BUS_SPACE_MAXSIZE, BUS_DMA_ALLOCNOW, NULL, NULL, &txq->tx_tag); if (rc != 0) { device_printf(sc->dev, "failed to create tx DMA tag: %d\n", rc); return (rc); } rc = alloc_tx_maps(txq); if (rc != 0) { device_printf(sc->dev, "failed to setup tx DMA maps: %d\n", rc); return (rc); } 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.viid_pkd = htobe32(V_FW_EQ_ETH_CMD_VIID(pi->viid)); c.fetchszm_to_iqid = htobe32(V_FW_EQ_ETH_CMD_HOSTFCMODE(X_HOSTFCMODE_STATUS_PAGE) | V_FW_EQ_ETH_CMD_PCIECHN(pi->tx_chan) | F_FW_EQ_ETH_CMD_FETCHRO | V_FW_EQ_ETH_CMD_IQID(eq->iqid)); 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_CIDXFTHRESH(X_CIDXFLUSHTHRESH_32) | V_FW_EQ_ETH_CMD_EQSIZE(eq->qsize)); c.eqaddr = htobe64(eq->ba); rc = -t4_wr_mbox(sc, sc->mbox, &c, sizeof(c), &c); if (rc != 0) { device_printf(pi->dev, "failed to create egress queue: %d\n", rc); return (rc); } eq->pidx = eq->cidx = 0; eq->cntxt_id = G_FW_EQ_ETH_CMD_EQID(be32toh(c.eqid_pkd)); eq->flags |= (EQ_ALLOCATED | EQ_STARTED); cntxt_id = eq->cntxt_id - sc->sge.eq_start; KASSERT(cntxt_id < sc->sge.neq, ("%s: eq->cntxt_id (%d) more than the max (%d)", __func__, cntxt_id, sc->sge.neq - 1)); sc->sge.eqmap[cntxt_id] = eq; children = SYSCTL_CHILDREN(pi->oid_txq); snprintf(name, sizeof(name), "%d", idx); oid = SYSCTL_ADD_NODE(&pi->ctx, children, OID_AUTO, name, CTLFLAG_RD, NULL, "tx queue"); children = SYSCTL_CHILDREN(oid); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "txcsum", CTLFLAG_RD, &txq->txcsum, "# of times hardware assisted with checksum"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "vlan_insertion", CTLFLAG_RD, &txq->vlan_insertion, "# of times hardware inserted 802.1Q tag"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "tso_wrs", CTLFLAG_RD, &txq->tso_wrs, "# of IPv4 TSO work requests"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "imm_wrs", CTLFLAG_RD, &txq->imm_wrs, "# of work requests with immediate data"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "sgl_wrs", CTLFLAG_RD, &txq->sgl_wrs, "# of work requests with direct SGL"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "txpkt_wrs", CTLFLAG_RD, &txq->txpkt_wrs, "# of txpkt work requests (one pkt/WR)"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "txpkts_wrs", CTLFLAG_RD, &txq->txpkts_wrs, "# of txpkts work requests (multiple pkts/WR)"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "txpkts_pkts", CTLFLAG_RD, &txq->txpkts_pkts, "# of frames tx'd using txpkts work requests"); SYSCTL_ADD_UINT(&pi->ctx, children, OID_AUTO, "no_dmamap", CTLFLAG_RD, &txq->no_dmamap, 0, "# of times txq ran out of DMA maps"); SYSCTL_ADD_UINT(&pi->ctx, children, OID_AUTO, "no_desc", CTLFLAG_RD, &txq->no_desc, 0, "# of times txq ran out of hardware descriptors"); SYSCTL_ADD_UINT(&pi->ctx, children, OID_AUTO, "egr_update", CTLFLAG_RD, &txq->egr_update, 0, "egress update notifications from the SGE"); return (rc); } static int free_txq(struct port_info *pi, struct sge_txq *txq) { int rc; struct adapter *sc = pi->adapter; struct sge_eq *eq = &txq->eq; if (eq->flags & (EQ_ALLOCATED | EQ_STARTED)) { /* * Wait for the response to a credit flush if there's one * pending. Clearing the flag tells handle_sge_egr_update or * cxgbe_txq_start (depending on how far the response has made * it) that they should ignore the response and wake up free_txq * instead. * * The interface has been marked down by the time we get here * (both IFF_UP and IFF_DRV_RUNNING cleared). qflush has * emptied the tx buf_rings and we know nothing new is being * queued for tx so we don't have to worry about a new credit * flush request. */ TXQ_LOCK(txq); if (eq->flags & EQ_CRFLUSHED) { eq->flags &= ~EQ_CRFLUSHED; msleep(txq, &eq->eq_lock, 0, "crflush", 0); } TXQ_UNLOCK(txq); rc = -t4_eth_eq_free(sc, sc->mbox, sc->pf, 0, eq->cntxt_id); if (rc != 0) { device_printf(pi->dev, "failed to free egress queue %p: %d\n", eq, rc); return (rc); } eq->flags &= ~(EQ_ALLOCATED | EQ_STARTED); } free_ring(sc, eq->desc_tag, eq->desc_map, eq->ba, eq->desc); free(txq->sdesc, M_CXGBE); if (txq->maps) free_tx_maps(txq); buf_ring_free(txq->br, M_CXGBE); if (txq->tx_tag) bus_dma_tag_destroy(txq->tx_tag); if (mtx_initialized(&eq->eq_lock)) mtx_destroy(&eq->eq_lock); bzero(txq, sizeof(*txq)); return (0); } static void oneseg_dma_callback(void *arg, bus_dma_segment_t *segs, int nseg, int error) { bus_addr_t *ba = arg; KASSERT(nseg == 1, ("%s meant for single segment mappings only.", __func__)); *ba = error ? 0 : segs->ds_addr; } static inline bool is_new_response(const struct sge_iq *iq, struct rsp_ctrl **ctrl) { *ctrl = (void *)((uintptr_t)iq->cdesc + (iq->esize - sizeof(struct rsp_ctrl))); return (((*ctrl)->u.type_gen >> S_RSPD_GEN) == iq->gen); } static inline void iq_next(struct sge_iq *iq) { iq->cdesc = (void *) ((uintptr_t)iq->cdesc + iq->esize); if (__predict_false(++iq->cidx == iq->qsize - 1)) { iq->cidx = 0; iq->gen ^= 1; iq->cdesc = iq->desc; } } #define FL_HW_IDX(x) ((x) >> 3) static inline void ring_fl_db(struct adapter *sc, struct sge_fl *fl) { int ndesc = fl->pending / 8; if (FL_HW_IDX(fl->pidx) == FL_HW_IDX(fl->cidx)) ndesc--; /* hold back one credit */ if (ndesc <= 0) return; /* nothing to do */ wmb(); t4_write_reg(sc, MYPF_REG(A_SGE_PF_KDOORBELL), F_DBPRIO | V_QID(fl->cntxt_id) | V_PIDX(ndesc)); fl->pending -= ndesc * 8; } /* * Fill up the freelist by upto nbufs and ring its doorbell if the number of * buffers ready to be handed to the hardware >= dbthresh. */ static void refill_fl(struct adapter *sc, struct sge_fl *fl, int nbufs, int dbthresh) { __be64 *d = &fl->desc[fl->pidx]; struct fl_sdesc *sd = &fl->sdesc[fl->pidx]; bus_dma_tag_t tag; bus_addr_t pa; caddr_t cl; int rc; FL_LOCK_ASSERT_OWNED(fl); if (nbufs < 0 || nbufs > fl->needed) nbufs = fl->needed; while (nbufs--) { if (sd->cl != NULL) { /* * This happens when a frame small enough to fit * entirely in an mbuf was received in cl last time. * We'd held on to cl and can reuse it now. Note that * we reuse a cluster of the old size if fl->tag_idx is * no longer the same as sd->tag_idx. */ KASSERT(*d == sd->ba_tag, ("%s: recyling problem at pidx %d", __func__, fl->pidx)); d++; goto recycled; } if (fl->tag_idx != sd->tag_idx) { bus_dmamap_t map; bus_dma_tag_t newtag = fl->tag[fl->tag_idx]; bus_dma_tag_t oldtag = fl->tag[sd->tag_idx]; /* * An MTU change can get us here. Discard the old map * which was created with the old tag, but only if * we're able to get a new one. */ rc = bus_dmamap_create(newtag, 0, &map); if (rc == 0) { bus_dmamap_destroy(oldtag, sd->map); sd->map = map; sd->tag_idx = fl->tag_idx; } } tag = fl->tag[sd->tag_idx]; cl = m_cljget(NULL, M_NOWAIT, FL_BUF_SIZE(sd->tag_idx)); if (cl == NULL) break; rc = bus_dmamap_load(tag, sd->map, cl, FL_BUF_SIZE(sd->tag_idx), oneseg_dma_callback, &pa, 0); if (rc != 0 || pa == 0) { fl->dmamap_failed++; uma_zfree(FL_BUF_ZONE(sd->tag_idx), cl); break; } sd->cl = cl; *d++ = htobe64(pa | sd->tag_idx); #ifdef INVARIANTS sd->ba_tag = htobe64(pa | sd->tag_idx); #endif recycled: /* sd->m is never recycled, should always be NULL */ KASSERT(sd->m == NULL, ("%s: stray mbuf", __func__)); sd->m = m_gethdr(M_NOWAIT, MT_NOINIT); if (sd->m == NULL) break; fl->pending++; fl->needed--; sd++; if (++fl->pidx == fl->cap) { fl->pidx = 0; sd = fl->sdesc; d = fl->desc; } } if (fl->pending >= dbthresh) ring_fl_db(sc, fl); } static int alloc_fl_sdesc(struct sge_fl *fl) { struct fl_sdesc *sd; bus_dma_tag_t tag; int i, rc; FL_LOCK_ASSERT_OWNED(fl); fl->sdesc = malloc(fl->cap * sizeof(struct fl_sdesc), M_CXGBE, M_ZERO | M_WAITOK); tag = fl->tag[fl->tag_idx]; sd = fl->sdesc; for (i = 0; i < fl->cap; i++, sd++) { sd->tag_idx = fl->tag_idx; rc = bus_dmamap_create(tag, 0, &sd->map); if (rc != 0) goto failed; } return (0); failed: while (--i >= 0) { sd--; bus_dmamap_destroy(tag, sd->map); if (sd->m) { m_init(sd->m, NULL, 0, M_NOWAIT, MT_DATA, 0); m_free(sd->m); sd->m = NULL; } } KASSERT(sd == fl->sdesc, ("%s: EDOOFUS", __func__)); free(fl->sdesc, M_CXGBE); fl->sdesc = NULL; return (rc); } static void free_fl_sdesc(struct sge_fl *fl) { struct fl_sdesc *sd; int i; FL_LOCK_ASSERT_OWNED(fl); sd = fl->sdesc; for (i = 0; i < fl->cap; i++, sd++) { if (sd->m) { m_init(sd->m, NULL, 0, M_NOWAIT, MT_DATA, 0); m_free(sd->m); sd->m = NULL; } if (sd->cl) { bus_dmamap_unload(fl->tag[sd->tag_idx], sd->map); uma_zfree(FL_BUF_ZONE(sd->tag_idx), sd->cl); sd->cl = NULL; } bus_dmamap_destroy(fl->tag[sd->tag_idx], sd->map); } free(fl->sdesc, M_CXGBE); fl->sdesc = NULL; } static int alloc_tx_maps(struct sge_txq *txq) { struct tx_map *txm; int i, rc, count; /* * We can stuff ~10 frames in an 8-descriptor txpkts WR (8 is the SGE * limit for any WR). txq->no_dmamap events shouldn't occur if maps is * sized for the worst case. */ count = txq->eq.qsize * 10 / 8; txq->map_total = txq->map_avail = count; txq->map_cidx = txq->map_pidx = 0; txq->maps = malloc(count * sizeof(struct tx_map), M_CXGBE, M_ZERO | M_WAITOK); txm = txq->maps; for (i = 0; i < count; i++, txm++) { rc = bus_dmamap_create(txq->tx_tag, 0, &txm->map); if (rc != 0) goto failed; } return (0); failed: while (--i >= 0) { txm--; bus_dmamap_destroy(txq->tx_tag, txm->map); } KASSERT(txm == txq->maps, ("%s: EDOOFUS", __func__)); free(txq->maps, M_CXGBE); txq->maps = NULL; return (rc); } static void free_tx_maps(struct sge_txq *txq) { struct tx_map *txm; int i; txm = txq->maps; for (i = 0; i < txq->map_total; i++, txm++) { if (txm->m) { bus_dmamap_unload(txq->tx_tag, txm->map); m_freem(txm->m); txm->m = NULL; } bus_dmamap_destroy(txq->tx_tag, txm->map); } free(txq->maps, M_CXGBE); txq->maps = NULL; } /* * We'll do immediate data tx for non-TSO, but only when not coalescing. We're * willing to use upto 2 hardware descriptors which means a maximum of 96 bytes * of immediate data. */ #define IMM_LEN ( \ 2 * TX_EQ_ESIZE \ - sizeof(struct fw_eth_tx_pkt_wr) \ - sizeof(struct cpl_tx_pkt_core)) /* * Returns non-zero on failure, no need to cleanup anything in that case. * * Note 1: We always try to defrag the mbuf if required and return EFBIG only * if the resulting chain still won't fit in a tx descriptor. * * Note 2: We'll pullup the mbuf chain if TSO is requested and the first mbuf * does not have the TCP header in it. */ static int get_pkt_sgl(struct sge_txq *txq, struct mbuf **fp, struct sgl *sgl, int sgl_only) { struct mbuf *m = *fp; struct tx_map *txm; int rc, defragged = 0, n; TXQ_LOCK_ASSERT_OWNED(txq); if (m->m_pkthdr.tso_segsz) sgl_only = 1; /* Do not allow immediate data with LSO */ start: sgl->nsegs = 0; if (m->m_pkthdr.len <= IMM_LEN && !sgl_only) return (0); /* nsegs = 0 tells caller to use imm. tx */ if (txq->map_avail == 0) { txq->no_dmamap++; return (ENOMEM); } txm = &txq->maps[txq->map_pidx]; if (m->m_pkthdr.tso_segsz && m->m_len < 50) { *fp = m_pullup(m, 50); m = *fp; if (m == NULL) return (ENOBUFS); } rc = bus_dmamap_load_mbuf_sg(txq->tx_tag, txm->map, m, sgl->seg, &sgl->nsegs, BUS_DMA_NOWAIT); if (rc == EFBIG && defragged == 0) { m = m_defrag(m, M_DONTWAIT); if (m == NULL) return (EFBIG); defragged = 1; *fp = m; goto start; } if (rc != 0) return (rc); txm->m = m; txq->map_avail--; if (++txq->map_pidx == txq->map_total) txq->map_pidx = 0; KASSERT(sgl->nsegs > 0 && sgl->nsegs <= TX_SGL_SEGS, ("%s: bad DMA mapping (%d segments)", __func__, sgl->nsegs)); /* * Store the # of flits required to hold this frame's SGL in nflits. An * SGL has a (ULPTX header + len0, addr0) tuple optionally followed by * multiple (len0 + len1, addr0, addr1) tuples. If addr1 is not used * then len1 must be set to 0. */ n = sgl->nsegs - 1; sgl->nflits = (3 * n) / 2 + (n & 1) + 2; return (0); } /* * Releases all the txq resources used up in the specified sgl. */ static int free_pkt_sgl(struct sge_txq *txq, struct sgl *sgl) { struct tx_map *txm; TXQ_LOCK_ASSERT_OWNED(txq); if (sgl->nsegs == 0) return (0); /* didn't use any map */ /* 1 pkt uses exactly 1 map, back it out */ txq->map_avail++; if (txq->map_pidx > 0) txq->map_pidx--; else txq->map_pidx = txq->map_total - 1; txm = &txq->maps[txq->map_pidx]; bus_dmamap_unload(txq->tx_tag, txm->map); txm->m = NULL; return (0); } static int write_txpkt_wr(struct port_info *pi, struct sge_txq *txq, struct mbuf *m, struct sgl *sgl) { struct sge_eq *eq = &txq->eq; struct fw_eth_tx_pkt_wr *wr; struct cpl_tx_pkt_core *cpl; uint32_t ctrl; /* used in many unrelated places */ uint64_t ctrl1; int nflits, ndesc, pktlen; struct tx_sdesc *txsd; caddr_t dst; TXQ_LOCK_ASSERT_OWNED(txq); pktlen = m->m_pkthdr.len; /* * Do we have enough flits to send this frame out? */ ctrl = sizeof(struct cpl_tx_pkt_core); if (m->m_pkthdr.tso_segsz) { nflits = TXPKT_LSO_WR_HDR; ctrl += sizeof(struct cpl_tx_pkt_lso); } else nflits = TXPKT_WR_HDR; if (sgl->nsegs > 0) nflits += sgl->nflits; else { nflits += howmany(pktlen, 8); ctrl += pktlen; } ndesc = howmany(nflits, 8); if (ndesc > eq->avail) return (ENOMEM); /* Firmware work request header */ wr = (void *)&eq->desc[eq->pidx]; wr->op_immdlen = htobe32(V_FW_WR_OP(FW_ETH_TX_PKT_WR) | V_FW_WR_IMMDLEN(ctrl)); ctrl = V_FW_WR_LEN16(howmany(nflits, 2)); if (eq->avail == ndesc && !(eq->flags & EQ_CRFLUSHED)) { ctrl |= F_FW_WR_EQUEQ | F_FW_WR_EQUIQ; eq->flags |= EQ_CRFLUSHED; } wr->equiq_to_len16 = htobe32(ctrl); wr->r3 = 0; if (m->m_pkthdr.tso_segsz) { struct cpl_tx_pkt_lso *lso = (void *)(wr + 1); struct ether_header *eh; struct ip *ip; struct tcphdr *tcp; ctrl = V_LSO_OPCODE(CPL_TX_PKT_LSO) | F_LSO_FIRST_SLICE | F_LSO_LAST_SLICE; eh = mtod(m, struct ether_header *); if (eh->ether_type == htons(ETHERTYPE_VLAN)) { ctrl |= V_LSO_ETHHDR_LEN(1); ip = (void *)((struct ether_vlan_header *)eh + 1); } else ip = (void *)(eh + 1); tcp = (void *)((uintptr_t)ip + ip->ip_hl * 4); ctrl |= V_LSO_IPHDR_LEN(ip->ip_hl) | V_LSO_TCPHDR_LEN(tcp->th_off); lso->lso_ctrl = htobe32(ctrl); lso->ipid_ofst = htobe16(0); lso->mss = htobe16(m->m_pkthdr.tso_segsz); lso->seqno_offset = htobe32(0); lso->len = htobe32(pktlen); cpl = (void *)(lso + 1); txq->tso_wrs++; } else cpl = (void *)(wr + 1); /* Checksum offload */ ctrl1 = 0; if (!(m->m_pkthdr.csum_flags & CSUM_IP)) ctrl1 |= F_TXPKT_IPCSUM_DIS; if (!(m->m_pkthdr.csum_flags & (CSUM_TCP | CSUM_UDP))) ctrl1 |= F_TXPKT_L4CSUM_DIS; if (m->m_pkthdr.csum_flags & (CSUM_IP | CSUM_TCP | CSUM_UDP)) txq->txcsum++; /* some hardware assistance provided */ /* VLAN tag insertion */ if (m->m_flags & M_VLANTAG) { ctrl1 |= F_TXPKT_VLAN_VLD | V_TXPKT_VLAN(m->m_pkthdr.ether_vtag); txq->vlan_insertion++; } /* CPL header */ cpl->ctrl0 = htobe32(V_TXPKT_OPCODE(CPL_TX_PKT) | V_TXPKT_INTF(pi->tx_chan) | V_TXPKT_PF(pi->adapter->pf)); cpl->pack = 0; cpl->len = htobe16(pktlen); cpl->ctrl1 = htobe64(ctrl1); /* Software descriptor */ txsd = &txq->sdesc[eq->pidx]; txsd->desc_used = ndesc; eq->pending += ndesc; eq->avail -= ndesc; eq->pidx += ndesc; if (eq->pidx >= eq->cap) eq->pidx -= eq->cap; /* SGL */ dst = (void *)(cpl + 1); if (sgl->nsegs > 0) { txsd->credits = 1; txq->sgl_wrs++; write_sgl_to_txd(eq, sgl, &dst); } else { txsd->credits = 0; txq->imm_wrs++; for (; m; m = m->m_next) { copy_to_txd(eq, mtod(m, caddr_t), &dst, m->m_len); #ifdef INVARIANTS pktlen -= m->m_len; #endif } #ifdef INVARIANTS KASSERT(pktlen == 0, ("%s: %d bytes left.", __func__, pktlen)); #endif } txq->txpkt_wrs++; return (0); } /* * Returns 0 to indicate that m has been accepted into a coalesced tx work * request. It has either been folded into txpkts or txpkts was flushed and m * has started a new coalesced work request (as the first frame in a fresh * txpkts). * * Returns non-zero to indicate a failure - caller is responsible for * transmitting m, if there was anything in txpkts it has been flushed. */ static int add_to_txpkts(struct port_info *pi, struct sge_txq *txq, struct txpkts *txpkts, struct mbuf *m, struct sgl *sgl) { struct sge_eq *eq = &txq->eq; int can_coalesce; struct tx_sdesc *txsd; int flits; TXQ_LOCK_ASSERT_OWNED(txq); if (txpkts->npkt > 0) { flits = TXPKTS_PKT_HDR + sgl->nflits; can_coalesce = m->m_pkthdr.tso_segsz == 0 && txpkts->nflits + flits <= TX_WR_FLITS && txpkts->nflits + flits <= eq->avail * 8 && txpkts->plen + m->m_pkthdr.len < 65536; if (can_coalesce) { txpkts->npkt++; txpkts->nflits += flits; txpkts->plen += m->m_pkthdr.len; txsd = &txq->sdesc[eq->pidx]; txsd->credits++; return (0); } /* * Couldn't coalesce m into txpkts. The first order of business * is to send txpkts on its way. Then we'll revisit m. */ write_txpkts_wr(txq, txpkts); } /* * Check if we can start a new coalesced tx work request with m as * the first packet in it. */ KASSERT(txpkts->npkt == 0, ("%s: txpkts not empty", __func__)); flits = TXPKTS_WR_HDR + sgl->nflits; can_coalesce = m->m_pkthdr.tso_segsz == 0 && flits <= eq->avail * 8 && flits <= TX_WR_FLITS; if (can_coalesce == 0) return (EINVAL); /* * Start a fresh coalesced tx WR with m as the first frame in it. */ txpkts->npkt = 1; txpkts->nflits = flits; txpkts->flitp = &eq->desc[eq->pidx].flit[2]; txpkts->plen = m->m_pkthdr.len; txsd = &txq->sdesc[eq->pidx]; txsd->credits = 1; return (0); } /* * Note that write_txpkts_wr can never run out of hardware descriptors (but * write_txpkt_wr can). add_to_txpkts ensures that a frame is accepted for * coalescing only if sufficient hardware descriptors are available. */ static void write_txpkts_wr(struct sge_txq *txq, struct txpkts *txpkts) { struct sge_eq *eq = &txq->eq; struct fw_eth_tx_pkts_wr *wr; struct tx_sdesc *txsd; uint32_t ctrl; int ndesc; TXQ_LOCK_ASSERT_OWNED(txq); ndesc = howmany(txpkts->nflits, 8); wr = (void *)&eq->desc[eq->pidx]; wr->op_immdlen = htobe32(V_FW_WR_OP(FW_ETH_TX_PKTS_WR) | V_FW_WR_IMMDLEN(0)); /* immdlen does not matter in this WR */ ctrl = V_FW_WR_LEN16(howmany(txpkts->nflits, 2)); if (eq->avail == ndesc && !(eq->flags & EQ_CRFLUSHED)) { ctrl |= F_FW_WR_EQUEQ | F_FW_WR_EQUIQ; eq->flags |= EQ_CRFLUSHED; } wr->equiq_to_len16 = htobe32(ctrl); wr->plen = htobe16(txpkts->plen); wr->npkt = txpkts->npkt; wr->r3 = wr->type = 0; /* Everything else already written */ txsd = &txq->sdesc[eq->pidx]; txsd->desc_used = ndesc; KASSERT(eq->avail >= ndesc, ("%s: out of descriptors", __func__)); eq->pending += ndesc; eq->avail -= ndesc; eq->pidx += ndesc; if (eq->pidx >= eq->cap) eq->pidx -= eq->cap; txq->txpkts_pkts += txpkts->npkt; txq->txpkts_wrs++; txpkts->npkt = 0; /* emptied */ } static inline void write_ulp_cpl_sgl(struct port_info *pi, struct sge_txq *txq, struct txpkts *txpkts, struct mbuf *m, struct sgl *sgl) { struct ulp_txpkt *ulpmc; struct ulptx_idata *ulpsc; struct cpl_tx_pkt_core *cpl; struct sge_eq *eq = &txq->eq; uintptr_t flitp, start, end; uint64_t ctrl; caddr_t dst; KASSERT(txpkts->npkt > 0, ("%s: txpkts is empty", __func__)); start = (uintptr_t)eq->desc; end = (uintptr_t)eq->spg; /* Checksum offload */ ctrl = 0; if (!(m->m_pkthdr.csum_flags & CSUM_IP)) ctrl |= F_TXPKT_IPCSUM_DIS; if (!(m->m_pkthdr.csum_flags & (CSUM_TCP | CSUM_UDP))) ctrl |= F_TXPKT_L4CSUM_DIS; if (m->m_pkthdr.csum_flags & (CSUM_IP | CSUM_TCP | CSUM_UDP)) txq->txcsum++; /* some hardware assistance provided */ /* VLAN tag insertion */ if (m->m_flags & M_VLANTAG) { ctrl |= F_TXPKT_VLAN_VLD | V_TXPKT_VLAN(m->m_pkthdr.ether_vtag); txq->vlan_insertion++; } /* * The previous packet's SGL must have ended at a 16 byte boundary (this * is required by the firmware/hardware). It follows that flitp cannot * wrap around between the ULPTX master command and ULPTX subcommand (8 * bytes each), and that it can not wrap around in the middle of the * cpl_tx_pkt_core either. */ flitp = (uintptr_t)txpkts->flitp; KASSERT((flitp & 0xf) == 0, ("%s: last SGL did not end at 16 byte boundary: %p", __func__, txpkts->flitp)); /* ULP master command */ ulpmc = (void *)flitp; ulpmc->cmd_dest = htonl(V_ULPTX_CMD(ULP_TX_PKT) | V_ULP_TXPKT_DEST(0) | V_ULP_TXPKT_FID(eq->iqid)); ulpmc->len = htonl(howmany(sizeof(*ulpmc) + sizeof(*ulpsc) + sizeof(*cpl) + 8 * sgl->nflits, 16)); /* ULP subcommand */ ulpsc = (void *)(ulpmc + 1); ulpsc->cmd_more = htobe32(V_ULPTX_CMD((u32)ULP_TX_SC_IMM) | F_ULP_TX_SC_MORE); ulpsc->len = htobe32(sizeof(struct cpl_tx_pkt_core)); flitp += sizeof(*ulpmc) + sizeof(*ulpsc); if (flitp == end) flitp = start; /* CPL_TX_PKT */ cpl = (void *)flitp; cpl->ctrl0 = htobe32(V_TXPKT_OPCODE(CPL_TX_PKT) | V_TXPKT_INTF(pi->tx_chan) | V_TXPKT_PF(pi->adapter->pf)); cpl->pack = 0; cpl->len = htobe16(m->m_pkthdr.len); cpl->ctrl1 = htobe64(ctrl); flitp += sizeof(*cpl); if (flitp == end) flitp = start; /* SGL for this frame */ dst = (caddr_t)flitp; txpkts->nflits += write_sgl_to_txd(eq, sgl, &dst); txpkts->flitp = (void *)dst; KASSERT(((uintptr_t)dst & 0xf) == 0, ("%s: SGL ends at %p (not a 16 byte boundary)", __func__, dst)); } /* * If the SGL ends on an address that is not 16 byte aligned, this function will * add a 0 filled flit at the end. It returns 1 in that case. */ static int write_sgl_to_txd(struct sge_eq *eq, struct sgl *sgl, caddr_t *to) { __be64 *flitp, *end; struct ulptx_sgl *usgl; bus_dma_segment_t *seg; int i, padded; KASSERT(sgl->nsegs > 0 && sgl->nflits > 0, ("%s: bad SGL - nsegs=%d, nflits=%d", __func__, sgl->nsegs, sgl->nflits)); KASSERT(((uintptr_t)(*to) & 0xf) == 0, ("%s: SGL must start at a 16 byte boundary: %p", __func__, *to)); flitp = (__be64 *)(*to); end = flitp + sgl->nflits; seg = &sgl->seg[0]; usgl = (void *)flitp; /* * We start at a 16 byte boundary somewhere inside the tx descriptor * ring, so we're at least 16 bytes away from the status page. There is * no chance of a wrap around in the middle of usgl (which is 16 bytes). */ usgl->cmd_nsge = htobe32(V_ULPTX_CMD(ULP_TX_SC_DSGL) | V_ULPTX_NSGE(sgl->nsegs)); usgl->len0 = htobe32(seg->ds_len); usgl->addr0 = htobe64(seg->ds_addr); seg++; if ((uintptr_t)end <= (uintptr_t)eq->spg) { /* Won't wrap around at all */ for (i = 0; i < sgl->nsegs - 1; i++, seg++) { usgl->sge[i / 2].len[i & 1] = htobe32(seg->ds_len); usgl->sge[i / 2].addr[i & 1] = htobe64(seg->ds_addr); } if (i & 1) usgl->sge[i / 2].len[1] = htobe32(0); } else { /* Will wrap somewhere in the rest of the SGL */ /* 2 flits already written, write the rest flit by flit */ flitp = (void *)(usgl + 1); for (i = 0; i < sgl->nflits - 2; i++) { if ((uintptr_t)flitp == (uintptr_t)eq->spg) flitp = (void *)eq->desc; *flitp++ = get_flit(seg, sgl->nsegs - 1, i); } end = flitp; } if ((uintptr_t)end & 0xf) { *(uint64_t *)end = 0; end++; padded = 1; } else padded = 0; if ((uintptr_t)end == (uintptr_t)eq->spg) *to = (void *)eq->desc; else *to = (void *)end; return (padded); } static inline void copy_to_txd(struct sge_eq *eq, caddr_t from, caddr_t *to, int len) { if ((uintptr_t)(*to) + len <= (uintptr_t)eq->spg) { bcopy(from, *to, len); (*to) += len; } else { int portion = (uintptr_t)eq->spg - (uintptr_t)(*to); bcopy(from, *to, portion); from += portion; portion = len - portion; /* remaining */ bcopy(from, (void *)eq->desc, portion); (*to) = (caddr_t)eq->desc + portion; } } static inline void ring_eq_db(struct adapter *sc, struct sge_eq *eq) { wmb(); t4_write_reg(sc, MYPF_REG(A_SGE_PF_KDOORBELL), V_QID(eq->cntxt_id) | V_PIDX(eq->pending)); eq->pending = 0; } static inline int reclaimable(struct sge_eq *eq) { unsigned int cidx; cidx = eq->spg->cidx; /* stable snapshot */ cidx = be16_to_cpu(cidx); if (cidx >= eq->cidx) return (cidx - eq->cidx); else return (cidx + eq->cap - eq->cidx); } /* * There are "can_reclaim" tx descriptors ready to be reclaimed. Reclaim as * many as possible but stop when there are around "n" mbufs to free. * * The actual number reclaimed is provided as the return value. */ static int reclaim_tx_descs(struct sge_txq *txq, int can_reclaim, int n) { struct tx_sdesc *txsd; struct tx_map *txm; unsigned int reclaimed, maps; struct sge_eq *eq = &txq->eq; EQ_LOCK_ASSERT_OWNED(eq); if (can_reclaim == 0) can_reclaim = reclaimable(eq); maps = reclaimed = 0; while (can_reclaim && maps < n) { int ndesc; txsd = &txq->sdesc[eq->cidx]; ndesc = txsd->desc_used; /* Firmware doesn't return "partial" credits. */ KASSERT(can_reclaim >= ndesc, ("%s: unexpected number of credits: %d, %d", __func__, can_reclaim, ndesc)); maps += txsd->credits; reclaimed += ndesc; can_reclaim -= ndesc; eq->cidx += ndesc; if (__predict_false(eq->cidx >= eq->cap)) eq->cidx -= eq->cap; } txm = &txq->maps[txq->map_cidx]; if (maps) prefetch(txm->m); eq->avail += reclaimed; KASSERT(eq->avail < eq->cap, /* avail tops out at (cap - 1) */ ("%s: too many descriptors available", __func__)); txq->map_avail += maps; KASSERT(txq->map_avail <= txq->map_total, ("%s: too many maps available", __func__)); while (maps--) { struct tx_map *next; next = txm + 1; if (__predict_false(txq->map_cidx + 1 == txq->map_total)) next = txq->maps; prefetch(next->m); bus_dmamap_unload(txq->tx_tag, txm->map); m_freem(txm->m); txm->m = NULL; txm = next; if (__predict_false(++txq->map_cidx == txq->map_total)) txq->map_cidx = 0; } return (reclaimed); } static void write_eqflush_wr(struct sge_eq *eq) { struct fw_eq_flush_wr *wr; EQ_LOCK_ASSERT_OWNED(eq); KASSERT(eq->avail > 0, ("%s: no descriptors left.", __func__)); wr = (void *)&eq->desc[eq->pidx]; bzero(wr, sizeof(*wr)); wr->opcode = FW_EQ_FLUSH_WR; wr->equiq_to_len16 = htobe32(V_FW_WR_LEN16(sizeof(*wr) / 16) | F_FW_WR_EQUEQ | F_FW_WR_EQUIQ); eq->flags |= EQ_CRFLUSHED; eq->pending++; eq->avail--; if (++eq->pidx == eq->cap) eq->pidx = 0; } static __be64 get_flit(bus_dma_segment_t *sgl, int nsegs, int idx) { int i = (idx / 3) * 2; switch (idx % 3) { case 0: { __be64 rc; rc = htobe32(sgl[i].ds_len); if (i + 1 < nsegs) rc |= (uint64_t)htobe32(sgl[i + 1].ds_len) << 32; return (rc); } case 1: return htobe64(sgl[i].ds_addr); case 2: return htobe64(sgl[i + 1].ds_addr); } return (0); } static void set_fl_tag_idx(struct sge_fl *fl, int mtu) { int i; FL_LOCK_ASSERT_OWNED(fl); for (i = 0; i < FL_BUF_SIZES - 1; i++) { if (FL_BUF_SIZE(i) >= (mtu + FL_PKTSHIFT)) break; } fl->tag_idx = i; } static int handle_sge_egr_update(struct adapter *sc, const struct cpl_sge_egr_update *cpl) { unsigned int qid = G_EGR_QID(ntohl(cpl->opcode_qid)); struct sge *s = &sc->sge; struct sge_txq *txq; struct port_info *pi; txq = (void *)s->eqmap[qid - s->eq_start]; TXQ_LOCK(txq); if (txq->eq.flags & EQ_CRFLUSHED) { pi = txq->ifp->if_softc; taskqueue_enqueue(pi->tq, &txq->resume_tx); txq->egr_update++; } else wakeup_one(txq); /* txq is going away, wakeup free_txq */ TXQ_UNLOCK(txq); return (0); } static void handle_cpl(struct adapter *sc, struct sge_iq *iq) { const struct rss_header *rss = (const void *)iq->cdesc; const struct cpl_fw6_msg *cpl = (const void *)(rss + 1); switch (rss->opcode) { case CPL_FW4_MSG: case CPL_FW6_MSG: if (cpl->type == FW6_TYPE_CMD_RPL) t4_handle_fw_rpl(sc, cpl->data); break; case CPL_SGE_EGR_UPDATE: handle_sge_egr_update(sc, (const void *)cpl); break; case CPL_SET_TCB_RPL: filter_rpl(sc, (const void *)cpl); break; default: panic("%s: unexpected CPL opcode 0x%x", __func__, rss->opcode); } } /* * m0 is freed on successful transmission. */ static int ctrl_tx(struct adapter *sc, struct sge_ctrlq *ctrlq, struct mbuf *m0) { struct sge_eq *eq = &ctrlq->eq; int rc = 0, ndesc; int can_reclaim; caddr_t dst; struct mbuf *m; M_ASSERTPKTHDR(m0); if (m0->m_pkthdr.len > SGE_MAX_WR_LEN) { log(LOG_ERR, "%s: %s work request too long (%d)", device_get_nameunit(sc->dev), __func__, m0->m_pkthdr.len); return (EMSGSIZE); } ndesc = howmany(m0->m_pkthdr.len, CTRL_EQ_ESIZE); EQ_LOCK(eq); can_reclaim = reclaimable(eq); eq->cidx += can_reclaim; eq->avail += can_reclaim; if (__predict_false(eq->cidx >= eq->cap)) eq->cidx -= eq->cap; if (eq->avail < ndesc) { rc = EAGAIN; ctrlq->no_desc++; goto failed; } dst = (void *)&eq->desc[eq->pidx]; for (m = m0; m; m = m->m_next) copy_to_txd(eq, mtod(m, caddr_t), &dst, m->m_len); eq->pidx += ndesc; if (__predict_false(eq->pidx >= eq->cap)) eq->pidx -= eq->cap; eq->pending += ndesc; ring_eq_db(sc, eq); failed: EQ_UNLOCK(eq); if (rc == 0) m_freem(m0); return (rc); } static int sysctl_uint16(SYSCTL_HANDLER_ARGS) { uint16_t *id = arg1; int i = *id; return sysctl_handle_int(oidp, &i, 0, req); } Index: projects/largeSMP/sys/dev/pccard/pccard.c =================================================================== --- projects/largeSMP/sys/dev/pccard/pccard.c (revision 222782) +++ projects/largeSMP/sys/dev/pccard/pccard.c (revision 222783) @@ -1,1485 +1,1485 @@ /* $NetBSD: pcmcia.c,v 1.23 2000/07/28 19:17:02 drochner Exp $ */ /*- * Copyright (c) 1997 Marc Horowitz. 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 Marc Horowitz. * 4. 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. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "power_if.h" #include "card_if.h" #define PCCARDDEBUG /* sysctl vars */ SYSCTL_NODE(_hw, OID_AUTO, pccard, CTLFLAG_RD, 0, "PCCARD parameters"); int pccard_debug = 0; TUNABLE_INT("hw.pccard.debug", &pccard_debug); SYSCTL_INT(_hw_pccard, OID_AUTO, debug, CTLFLAG_RW, &pccard_debug, 0, "pccard debug"); int pccard_cis_debug = 0; TUNABLE_INT("hw.pccard.cis_debug", &pccard_cis_debug); SYSCTL_INT(_hw_pccard, OID_AUTO, cis_debug, CTLFLAG_RW, &pccard_cis_debug, 0, "pccard CIS debug"); #ifdef PCCARDDEBUG #define DPRINTF(arg) if (pccard_debug) printf arg #define DEVPRINTF(arg) if (pccard_debug) device_printf arg #define PRVERBOSE(arg) printf arg #define DEVPRVERBOSE(arg) device_printf arg #else #define DPRINTF(arg) #define DEVPRINTF(arg) #define PRVERBOSE(arg) if (bootverbose) printf arg #define DEVPRVERBOSE(arg) if (bootverbose) device_printf arg #endif static int pccard_ccr_read(struct pccard_function *pf, int ccr); static void pccard_ccr_write(struct pccard_function *pf, int ccr, int val); static int pccard_attach_card(device_t dev); static int pccard_detach_card(device_t dev); static void pccard_function_init(struct pccard_function *pf, int entry); static void pccard_function_free(struct pccard_function *pf); static int pccard_function_enable(struct pccard_function *pf); static void pccard_function_disable(struct pccard_function *pf); static int pccard_probe(device_t dev); static int pccard_attach(device_t dev); static int pccard_detach(device_t dev); static void pccard_print_resources(struct resource_list *rl, const char *name, int type, int count, const char *format); static int pccard_print_child(device_t dev, device_t child); static int pccard_set_resource(device_t dev, device_t child, int type, int rid, u_long start, u_long count); static int pccard_get_resource(device_t dev, device_t child, int type, int rid, u_long *startp, u_long *countp); static void pccard_delete_resource(device_t dev, device_t child, int type, int rid); static int pccard_set_res_flags(device_t dev, device_t child, int type, int rid, u_long flags); static int pccard_set_memory_offset(device_t dev, device_t child, int rid, uint32_t offset, uint32_t *deltap); static int pccard_probe_and_attach_child(device_t dev, device_t child, struct pccard_function *pf); static void pccard_probe_nomatch(device_t cbdev, device_t child); static int pccard_read_ivar(device_t bus, device_t child, int which, uintptr_t *result); static void pccard_driver_added(device_t dev, driver_t *driver); static struct resource *pccard_alloc_resource(device_t dev, device_t child, int type, int *rid, u_long start, u_long end, u_long count, u_int flags); static int pccard_release_resource(device_t dev, device_t child, int type, int rid, struct resource *r); static void pccard_child_detached(device_t parent, device_t dev); static int pccard_filter(void *arg); static void pccard_intr(void *arg); static int pccard_setup_intr(device_t dev, device_t child, struct resource *irq, int flags, driver_filter_t *filt, driver_intr_t *intr, void *arg, void **cookiep); static int pccard_teardown_intr(device_t dev, device_t child, struct resource *r, void *cookie); static const struct pccard_product * pccard_do_product_lookup(device_t bus, device_t dev, const struct pccard_product *tab, size_t ent_size, pccard_product_match_fn matchfn); static int pccard_ccr_read(struct pccard_function *pf, int ccr) { return (bus_space_read_1(pf->pf_ccrt, pf->pf_ccrh, pf->pf_ccr_offset + ccr)); } static void pccard_ccr_write(struct pccard_function *pf, int ccr, int val) { if ((pf->ccr_mask) & (1 << (ccr / 2))) { bus_space_write_1(pf->pf_ccrt, pf->pf_ccrh, pf->pf_ccr_offset + ccr, val); } } static int pccard_set_default_descr(device_t dev) { const char *vendorstr, *prodstr; uint32_t vendor, prod; char *str; if (pccard_get_vendor_str(dev, &vendorstr)) return (0); if (pccard_get_product_str(dev, &prodstr)) return (0); if (vendorstr != NULL && prodstr != NULL) { str = malloc(strlen(vendorstr) + strlen(prodstr) + 2, M_DEVBUF, M_WAITOK); sprintf(str, "%s %s", vendorstr, prodstr); device_set_desc_copy(dev, str); free(str, M_DEVBUF); } else { if (pccard_get_vendor(dev, &vendor)) return (0); if (pccard_get_product(dev, &prod)) return (0); str = malloc(100, M_DEVBUF, M_WAITOK); snprintf(str, 100, "vendor=%#x product=%#x", vendor, prod); device_set_desc_copy(dev, str); free(str, M_DEVBUF); } return (0); } static int pccard_attach_card(device_t dev) { struct pccard_softc *sc = PCCARD_SOFTC(dev); struct pccard_function *pf; struct pccard_ivar *ivar; device_t child; int i; if (!STAILQ_EMPTY(&sc->card.pf_head)) { if (bootverbose || pccard_debug) device_printf(dev, "Card already inserted.\n"); } DEVPRINTF((dev, "chip_socket_enable\n")); POWER_ENABLE_SOCKET(device_get_parent(dev), dev); DEVPRINTF((dev, "read_cis\n")); pccard_read_cis(sc); DEVPRINTF((dev, "check_cis_quirks\n")); pccard_check_cis_quirks(dev); /* * bail now if the card has no functions, or if there was an error in * the cis. */ if (sc->card.error) { device_printf(dev, "CARD ERROR!\n"); return (1); } if (STAILQ_EMPTY(&sc->card.pf_head)) { device_printf(dev, "Card has no functions!\n"); return (1); } if (bootverbose || pccard_debug) pccard_print_cis(dev); DEVPRINTF((dev, "functions scanning\n")); i = -1; STAILQ_FOREACH(pf, &sc->card.pf_head, pf_list) { i++; if (STAILQ_EMPTY(&pf->cfe_head)) { device_printf(dev, "Function %d has no config entries.!\n", i); continue; } pf->sc = sc; pf->cfe = NULL; pf->dev = NULL; } DEVPRINTF((dev, "Card has %d functions. pccard_mfc is %d\n", i + 1, pccard_mfc(sc))); STAILQ_FOREACH(pf, &sc->card.pf_head, pf_list) { if (STAILQ_EMPTY(&pf->cfe_head)) continue; ivar = malloc(sizeof(struct pccard_ivar), M_DEVBUF, M_WAITOK | M_ZERO); resource_list_init(&ivar->resources); child = device_add_child(dev, NULL, -1); device_set_ivars(child, ivar); ivar->pf = pf; pf->dev = child; pccard_probe_and_attach_child(dev, child, pf); } return (0); } static int pccard_probe_and_attach_child(device_t dev, device_t child, struct pccard_function *pf) { struct pccard_softc *sc = PCCARD_SOFTC(dev); int error; /* * In NetBSD, the drivers are responsible for activating each * function of a card and selecting the config to use. In * FreeBSD, all that's done automatically in the typical lazy * way we do device resoruce allocation (except we pick the * cfe up front). This is the biggest depature from the * inherited NetBSD model, apart from the FreeBSD resource code. * * This seems to work well in practice for most cards. * However, there are two cases that are problematic. If a * driver wishes to pick and chose which config entry to use, * then this method falls down. These are usually older * cards. In addition, there are some cards that have * multiple hardware units on the cards, but presents only one * CIS chain. These cards are combination cards, but only one * of these units can be on at a time. * * To overcome this limitation, while preserving the basic * model, the probe routine can select a cfe and try to * activate it. If that succeeds, then we'll keep track of * and let that information persist until we attach the card. * Probe routines that do this MUST return 0, and cannot * participate in the bidding process for a device. This * seems harsh until you realize that if a probe routine knows * enough to override the cfe we pick, then chances are very * very good that it is the only driver that could hope to * cope with the card. Bidding is for generic drivers, and * while some of them may also match, none of them will do * configuration override. */ error = device_probe(child); if (error != 0) goto out; pccard_function_init(pf, -1); if (sc->sc_enabled_count == 0) POWER_ENABLE_SOCKET(device_get_parent(dev), dev); if (pccard_function_enable(pf) == 0 && pccard_set_default_descr(child) == 0 && device_attach(child) == 0) { DEVPRINTF((sc->dev, "function %d CCR at %d offset %#x " "mask %#x: %#x %#x %#x %#x, %#x %#x %#x %#x, %#x\n", pf->number, pf->pf_ccr_window, pf->pf_ccr_offset, pf->ccr_mask, pccard_ccr_read(pf, 0x00), pccard_ccr_read(pf, 0x02), pccard_ccr_read(pf, 0x04), pccard_ccr_read(pf, 0x06), pccard_ccr_read(pf, 0x0A), pccard_ccr_read(pf, 0x0C), pccard_ccr_read(pf, 0x0E), pccard_ccr_read(pf, 0x10), pccard_ccr_read(pf, 0x12))); return (0); } error = ENXIO; out:; /* * Probe may fail AND also try to select a cfe, if so, free * it. This is how we do cfe override. Or the attach fails. * Either way, we have to clean up. */ if (pf->cfe != NULL) pccard_function_disable(pf); pf->cfe = NULL; pccard_function_free(pf); return error; } static int pccard_detach_card(device_t dev) { struct pccard_softc *sc = PCCARD_SOFTC(dev); struct pccard_function *pf; struct pccard_config_entry *cfe; struct pccard_ivar *devi; int state; /* * We are running on either the PCCARD socket's event thread * or in user context detaching a device by user request. */ STAILQ_FOREACH(pf, &sc->card.pf_head, pf_list) { if (pf->dev == NULL) continue; state = device_get_state(pf->dev); if (state == DS_ATTACHED || state == DS_BUSY) device_detach(pf->dev); if (pf->cfe != NULL) pccard_function_disable(pf); pccard_function_free(pf); devi = PCCARD_IVAR(pf->dev); device_delete_child(dev, pf->dev); free(devi, M_DEVBUF); } if (sc->sc_enabled_count == 0) POWER_DISABLE_SOCKET(device_get_parent(dev), dev); while (NULL != (pf = STAILQ_FIRST(&sc->card.pf_head))) { while (NULL != (cfe = STAILQ_FIRST(&pf->cfe_head))) { STAILQ_REMOVE_HEAD(&pf->cfe_head, cfe_list); free(cfe, M_DEVBUF); } STAILQ_REMOVE_HEAD(&sc->card.pf_head, pf_list); free(pf, M_DEVBUF); } STAILQ_INIT(&sc->card.pf_head); return (0); } static const struct pccard_product * pccard_do_product_lookup(device_t bus, device_t dev, const struct pccard_product *tab, size_t ent_size, pccard_product_match_fn matchfn) { const struct pccard_product *ent; int matches; uint32_t vendor; uint32_t prod; const char *vendorstr; const char *prodstr; const char *cis3str; const char *cis4str; #ifdef DIAGNOSTIC if (sizeof *ent > ent_size) panic("pccard_product_lookup: bogus ent_size %jd", (intmax_t) ent_size); #endif if (pccard_get_vendor(dev, &vendor)) return (NULL); if (pccard_get_product(dev, &prod)) return (NULL); if (pccard_get_vendor_str(dev, &vendorstr)) return (NULL); if (pccard_get_product_str(dev, &prodstr)) return (NULL); if (pccard_get_cis3_str(dev, &cis3str)) return (NULL); if (pccard_get_cis4_str(dev, &cis4str)) return (NULL); for (ent = tab; ent->pp_vendor != 0; ent = (const struct pccard_product *) ((const char *) ent + ent_size)) { matches = 1; if (ent->pp_vendor == PCCARD_VENDOR_ANY && ent->pp_product == PCCARD_PRODUCT_ANY && ent->pp_cis[0] == NULL && ent->pp_cis[1] == NULL) { if (ent->pp_name) device_printf(dev, "Total wildcard entry ignored for %s\n", ent->pp_name); continue; } if (matches && ent->pp_vendor != PCCARD_VENDOR_ANY && vendor != ent->pp_vendor) matches = 0; if (matches && ent->pp_product != PCCARD_PRODUCT_ANY && prod != ent->pp_product) matches = 0; if (matches && ent->pp_cis[0] && (vendorstr == NULL || strcmp(ent->pp_cis[0], vendorstr) != 0)) matches = 0; if (matches && ent->pp_cis[1] && (prodstr == NULL || strcmp(ent->pp_cis[1], prodstr) != 0)) matches = 0; if (matches && ent->pp_cis[2] && (cis3str == NULL || strcmp(ent->pp_cis[2], cis3str) != 0)) matches = 0; if (matches && ent->pp_cis[3] && (cis4str == NULL || strcmp(ent->pp_cis[3], cis4str) != 0)) matches = 0; if (matchfn != NULL) matches = (*matchfn)(dev, ent, matches); if (matches) return (ent); } return (NULL); } /** * @brief pccard_select_cfe * * Select a cfe entry to use. Should be called from the pccard's probe * routine after it knows for sure that it wants this card. * * XXX I think we need to make this symbol be static, ala the kobj stuff * we do for everything else. This is a quick hack. */ int pccard_select_cfe(device_t dev, int entry) { struct pccard_ivar *devi = PCCARD_IVAR(dev); struct pccard_function *pf = devi->pf; pccard_function_init(pf, entry); return (pf->cfe ? 0 : ENOMEM); } /* * Initialize a PCCARD function. May be called as long as the function is * disabled. * * Note: pccard_function_init should not keep resources allocated. It should * only set them up ala isa pnp, set the values in the rl lists, and return. * Any resource held after pccard_function_init is called is a bug. However, * the bus routines to get the resources also assume that pccard_function_init * does this, so they need to be fixed too. */ static void pccard_function_init(struct pccard_function *pf, int entry) { struct pccard_config_entry *cfe; struct pccard_ivar *devi = PCCARD_IVAR(pf->dev); struct resource_list *rl = &devi->resources; struct resource_list_entry *rle; struct resource *r = 0; device_t bus; u_long start, end, len; int i, rid, spaces; if (pf->pf_flags & PFF_ENABLED) { printf("pccard_function_init: function is enabled"); return; } /* * Driver probe routine requested a specific entry already * that succeeded. */ if (pf->cfe != NULL) return; /* * walk the list of configuration entries until we find one that * we can allocate all the resources to. */ bus = device_get_parent(pf->dev); STAILQ_FOREACH(cfe, &pf->cfe_head, cfe_list) { if (cfe->iftype != PCCARD_IFTYPE_IO) continue; if (entry != -1 && cfe->number != entry) continue; spaces = 0; for (i = 0; i < cfe->num_iospace; i++) { start = cfe->iospace[i].start; if (start) end = start + cfe->iospace[i].length - 1; else end = ~0UL; DEVPRINTF((bus, "I/O rid %d start %#lx end %#lx\n", i, start, end)); rid = i; len = cfe->iospace[i].length; r = bus_alloc_resource(bus, SYS_RES_IOPORT, &rid, start, end, len, rman_make_alignment_flags(len)); if (r == NULL) goto not_this_one; rle = resource_list_add(rl, SYS_RES_IOPORT, rid, rman_get_start(r), rman_get_end(r), cfe->iospace[i].length); if (rle == NULL) panic("Cannot add resource rid %d IOPORT", rid); rle->res = r; spaces++; } for (i = 0; i < cfe->num_memspace; i++) { start = cfe->memspace[i].hostaddr; if (start) end = start + cfe->memspace[i].length - 1; else end = ~0UL; DEVPRINTF((bus, "Memory rid %d start %#lx end %#lx\n", i, start, end)); rid = i; len = cfe->memspace[i].length; r = bus_alloc_resource(bus, SYS_RES_MEMORY, &rid, start, end, len, rman_make_alignment_flags(len)); if (r == NULL) goto not_this_one; rle = resource_list_add(rl, SYS_RES_MEMORY, rid, rman_get_start(r), rman_get_end(r), cfe->memspace[i].length); if (rle == NULL) panic("Cannot add resource rid %d MEM", rid); rle->res = r; spaces++; } if (spaces == 0) { DEVPRINTF((bus, "Neither memory nor I/O mapped\n")); goto not_this_one; } if (cfe->irqmask) { rid = 0; r = bus_alloc_resource_any(bus, SYS_RES_IRQ, &rid, RF_SHAREABLE); if (r == NULL) goto not_this_one; rle = resource_list_add(rl, SYS_RES_IRQ, rid, rman_get_start(r), rman_get_end(r), 1); if (rle == NULL) panic("Cannot add resource rid %d IRQ", rid); rle->res = r; } /* If we get to here, we've allocated all we need */ pf->cfe = cfe; break; not_this_one:; DEVPRVERBOSE((bus, "Allocation failed for cfe %d\n", cfe->number)); resource_list_purge(rl); } } /* * Free resources allocated by pccard_function_init(), May be called as long * as the function is disabled. * * NOTE: This function should be unnecessary. pccard_function_init should * never keep resources initialized. */ static void pccard_function_free(struct pccard_function *pf) { struct pccard_ivar *devi = PCCARD_IVAR(pf->dev); struct resource_list_entry *rle; if (pf->pf_flags & PFF_ENABLED) { printf("pccard_function_free: function is enabled"); return; } STAILQ_FOREACH(rle, &devi->resources, link) { if (rle->res) { if (rman_get_device(rle->res) != pf->sc->dev) device_printf(pf->sc->dev, "function_free: Resource still owned by " "child, oops. " "(type=%d, rid=%d, addr=%#lx)\n", rle->type, rle->rid, rman_get_start(rle->res)); BUS_RELEASE_RESOURCE(device_get_parent(pf->sc->dev), pf->sc->dev, rle->type, rle->rid, rle->res); rle->res = NULL; } } resource_list_free(&devi->resources); } static void pccard_mfc_adjust_iobase(struct pccard_function *pf, bus_addr_t addr, bus_addr_t offset, bus_size_t size) { bus_size_t iosize, tmp; if (addr != 0) { if (pf->pf_mfc_iomax == 0) { pf->pf_mfc_iobase = addr + offset; pf->pf_mfc_iomax = pf->pf_mfc_iobase + size; } else { /* this makes the assumption that nothing overlaps */ if (pf->pf_mfc_iobase > addr + offset) pf->pf_mfc_iobase = addr + offset; if (pf->pf_mfc_iomax < addr + offset + size) pf->pf_mfc_iomax = addr + offset + size; } } tmp = pf->pf_mfc_iomax - pf->pf_mfc_iobase; /* round up to nearest (2^n)-1 */ for (iosize = 1; iosize < tmp; iosize <<= 1) ; iosize--; DEVPRINTF((pf->dev, "MFC: I/O base %#jx IOSIZE %#jx\n", (uintmax_t)pf->pf_mfc_iobase, (uintmax_t)(iosize + 1))); pccard_ccr_write(pf, PCCARD_CCR_IOBASE0, pf->pf_mfc_iobase & 0xff); pccard_ccr_write(pf, PCCARD_CCR_IOBASE1, (pf->pf_mfc_iobase >> 8) & 0xff); pccard_ccr_write(pf, PCCARD_CCR_IOBASE2, 0); pccard_ccr_write(pf, PCCARD_CCR_IOBASE3, 0); pccard_ccr_write(pf, PCCARD_CCR_IOSIZE, iosize); } /* Enable a PCCARD function */ static int pccard_function_enable(struct pccard_function *pf) { struct pccard_function *tmp; int reg; device_t dev = pf->sc->dev; if (pf->cfe == NULL) { DEVPRVERBOSE((dev, "No config entry could be allocated.\n")); return (ENOMEM); } if (pf->pf_flags & PFF_ENABLED) return (0); pf->sc->sc_enabled_count++; /* * it's possible for different functions' CCRs to be in the same * underlying page. Check for that. */ STAILQ_FOREACH(tmp, &pf->sc->card.pf_head, pf_list) { if ((tmp->pf_flags & PFF_ENABLED) && (pf->ccr_base >= (tmp->ccr_base - tmp->pf_ccr_offset)) && ((pf->ccr_base + PCCARD_CCR_SIZE) <= (tmp->ccr_base - tmp->pf_ccr_offset + tmp->pf_ccr_realsize))) { pf->pf_ccrt = tmp->pf_ccrt; pf->pf_ccrh = tmp->pf_ccrh; pf->pf_ccr_realsize = tmp->pf_ccr_realsize; /* * pf->pf_ccr_offset = (tmp->pf_ccr_offset - * tmp->ccr_base) + pf->ccr_base; */ /* pf->pf_ccr_offset = (tmp->pf_ccr_offset + pf->ccr_base) - tmp->ccr_base; */ pf->pf_ccr_window = tmp->pf_ccr_window; break; } } if (tmp == NULL) { pf->ccr_rid = 0; pf->ccr_res = bus_alloc_resource(dev, SYS_RES_MEMORY, &pf->ccr_rid, 0, ~0, PCCARD_MEM_PAGE_SIZE, RF_ACTIVE); if (!pf->ccr_res) goto bad; DEVPRINTF((dev, "ccr_res == %#lx-%#lx, base=%#x\n", rman_get_start(pf->ccr_res), rman_get_end(pf->ccr_res), pf->ccr_base)); CARD_SET_RES_FLAGS(device_get_parent(dev), dev, SYS_RES_MEMORY, pf->ccr_rid, PCCARD_A_MEM_ATTR); CARD_SET_MEMORY_OFFSET(device_get_parent(dev), dev, pf->ccr_rid, pf->ccr_base, &pf->pf_ccr_offset); pf->pf_ccrt = rman_get_bustag(pf->ccr_res); pf->pf_ccrh = rman_get_bushandle(pf->ccr_res); pf->pf_ccr_realsize = 1; } reg = (pf->cfe->number & PCCARD_CCR_OPTION_CFINDEX); reg |= PCCARD_CCR_OPTION_LEVIREQ; if (pccard_mfc(pf->sc)) { reg |= (PCCARD_CCR_OPTION_FUNC_ENABLE | PCCARD_CCR_OPTION_ADDR_DECODE); /* PCCARD_CCR_OPTION_IRQ_ENABLE set elsewhere as needed */ } pccard_ccr_write(pf, PCCARD_CCR_OPTION, reg); reg = 0; if ((pf->cfe->flags & PCCARD_CFE_IO16) == 0) reg |= PCCARD_CCR_STATUS_IOIS8; if (pf->cfe->flags & PCCARD_CFE_AUDIO) reg |= PCCARD_CCR_STATUS_AUDIO; pccard_ccr_write(pf, PCCARD_CCR_STATUS, reg); pccard_ccr_write(pf, PCCARD_CCR_SOCKETCOPY, 0); if (pccard_mfc(pf->sc)) pccard_mfc_adjust_iobase(pf, 0, 0, 0); #ifdef PCCARDDEBUG if (pccard_debug) { STAILQ_FOREACH(tmp, &pf->sc->card.pf_head, pf_list) { device_printf(tmp->sc->dev, "function %d CCR at %d offset %#x: " "%#x %#x %#x %#x, %#x %#x %#x %#x, %#x\n", tmp->number, tmp->pf_ccr_window, tmp->pf_ccr_offset, pccard_ccr_read(tmp, 0x00), pccard_ccr_read(tmp, 0x02), pccard_ccr_read(tmp, 0x04), pccard_ccr_read(tmp, 0x06), pccard_ccr_read(tmp, 0x0A), pccard_ccr_read(tmp, 0x0C), pccard_ccr_read(tmp, 0x0E), pccard_ccr_read(tmp, 0x10), pccard_ccr_read(tmp, 0x12)); } } #endif pf->pf_flags |= PFF_ENABLED; return (0); bad: /* * Decrement the reference count, and power down the socket, if * necessary. */ pf->sc->sc_enabled_count--; DEVPRINTF((dev, "bad --enabled_count = %d\n", pf->sc->sc_enabled_count)); return (1); } /* Disable PCCARD function. */ static void pccard_function_disable(struct pccard_function *pf) { struct pccard_function *tmp; device_t dev = pf->sc->dev; if (pf->cfe == NULL) panic("pccard_function_disable: function not initialized"); if ((pf->pf_flags & PFF_ENABLED) == 0) return; if (pf->intr_handler != NULL) { struct pccard_ivar *devi = PCCARD_IVAR(pf->dev); struct resource_list_entry *rle = resource_list_find(&devi->resources, SYS_RES_IRQ, 0); if (rle == NULL) panic("Can't disable an interrupt with no IRQ res\n"); BUS_TEARDOWN_INTR(dev, pf->dev, rle->res, pf->intr_handler_cookie); } /* * it's possible for different functions' CCRs to be in the same * underlying page. Check for that. Note we mark us as disabled * first to avoid matching ourself. */ pf->pf_flags &= ~PFF_ENABLED; STAILQ_FOREACH(tmp, &pf->sc->card.pf_head, pf_list) { if ((tmp->pf_flags & PFF_ENABLED) && (pf->ccr_base >= (tmp->ccr_base - tmp->pf_ccr_offset)) && ((pf->ccr_base + PCCARD_CCR_SIZE) <= (tmp->ccr_base - tmp->pf_ccr_offset + tmp->pf_ccr_realsize))) break; } /* Not used by anyone else; unmap the CCR. */ if (tmp == NULL) { bus_release_resource(dev, SYS_RES_MEMORY, pf->ccr_rid, pf->ccr_res); pf->ccr_res = NULL; } /* * Decrement the reference count, and power down the socket, if * necessary. */ pf->sc->sc_enabled_count--; } #define PCCARD_NPORT 2 #define PCCARD_NMEM 5 #define PCCARD_NIRQ 1 #define PCCARD_NDRQ 0 static int pccard_probe(device_t dev) { device_set_desc(dev, "16-bit PCCard bus"); return (0); } static int pccard_attach(device_t dev) { struct pccard_softc *sc = PCCARD_SOFTC(dev); int err; sc->dev = dev; sc->sc_enabled_count = 0; if ((err = pccard_device_create(sc)) != 0) return (err); STAILQ_INIT(&sc->card.pf_head); return (bus_generic_attach(dev)); } static int pccard_detach(device_t dev) { pccard_detach_card(dev); pccard_device_destroy(device_get_softc(dev)); return (0); } static int pccard_suspend(device_t self) { pccard_detach_card(self); return (0); } static int pccard_resume(device_t self) { return (0); } static void pccard_print_resources(struct resource_list *rl, const char *name, int type, int count, const char *format) { struct resource_list_entry *rle; int printed; int i; printed = 0; for (i = 0; i < count; i++) { rle = resource_list_find(rl, type, i); if (rle != NULL) { if (printed == 0) printf(" %s ", name); else if (printed > 0) printf(","); printed++; printf(format, rle->start); if (rle->count > 1) { printf("-"); printf(format, rle->start + rle->count - 1); } } else if (i > 3) { /* check the first few regardless */ break; } } } static int pccard_print_child(device_t dev, device_t child) { struct pccard_ivar *devi = PCCARD_IVAR(child); struct resource_list *rl = &devi->resources; int retval = 0; retval += bus_print_child_header(dev, child); retval += printf(" at"); if (devi != NULL) { pccard_print_resources(rl, "port", SYS_RES_IOPORT, PCCARD_NPORT, "%#lx"); pccard_print_resources(rl, "iomem", SYS_RES_MEMORY, PCCARD_NMEM, "%#lx"); pccard_print_resources(rl, "irq", SYS_RES_IRQ, PCCARD_NIRQ, "%ld"); pccard_print_resources(rl, "drq", SYS_RES_DRQ, PCCARD_NDRQ, "%ld"); retval += printf(" function %d config %d", devi->pf->number, devi->pf->cfe->number); } retval += bus_print_child_footer(dev, child); return (retval); } static int pccard_set_resource(device_t dev, device_t child, int type, int rid, u_long start, u_long count) { struct pccard_ivar *devi = PCCARD_IVAR(child); struct resource_list *rl = &devi->resources; if (type != SYS_RES_IOPORT && type != SYS_RES_MEMORY && type != SYS_RES_IRQ && type != SYS_RES_DRQ) return (EINVAL); if (rid < 0) return (EINVAL); if (type == SYS_RES_IOPORT && rid >= PCCARD_NPORT) return (EINVAL); if (type == SYS_RES_MEMORY && rid >= PCCARD_NMEM) return (EINVAL); if (type == SYS_RES_IRQ && rid >= PCCARD_NIRQ) return (EINVAL); if (type == SYS_RES_DRQ && rid >= PCCARD_NDRQ) return (EINVAL); resource_list_add(rl, type, rid, start, start + count - 1, count); if (NULL != resource_list_alloc(rl, device_get_parent(dev), dev, type, &rid, start, start + count - 1, count, 0)) return 0; else return ENOMEM; } static int pccard_get_resource(device_t dev, device_t child, int type, int rid, u_long *startp, u_long *countp) { struct pccard_ivar *devi = PCCARD_IVAR(child); struct resource_list *rl = &devi->resources; struct resource_list_entry *rle; rle = resource_list_find(rl, type, rid); if (rle == NULL) return (ENOENT); if (startp != NULL) *startp = rle->start; if (countp != NULL) *countp = rle->count; return (0); } static void pccard_delete_resource(device_t dev, device_t child, int type, int rid) { struct pccard_ivar *devi = PCCARD_IVAR(child); struct resource_list *rl = &devi->resources; resource_list_delete(rl, type, rid); } static int pccard_set_res_flags(device_t dev, device_t child, int type, int rid, u_long flags) { return (CARD_SET_RES_FLAGS(device_get_parent(dev), child, type, rid, flags)); } static int pccard_set_memory_offset(device_t dev, device_t child, int rid, uint32_t offset, uint32_t *deltap) { return (CARD_SET_MEMORY_OFFSET(device_get_parent(dev), child, rid, offset, deltap)); } static void pccard_probe_nomatch(device_t bus, device_t child) { struct pccard_ivar *devi = PCCARD_IVAR(child); struct pccard_function *pf = devi->pf; struct pccard_softc *sc = PCCARD_SOFTC(bus); int i; device_printf(bus, ""); printf(" (manufacturer=0x%04x, product=0x%04x, function_type=%d) " "at function %d\n", sc->card.manufacturer, sc->card.product, pf->function, pf->number); device_printf(bus, " CIS info: "); for (i = 0; sc->card.cis1_info[i] != NULL && i < 4; i++) printf("%s%s", i > 0 ? ", " : "", sc->card.cis1_info[i]); printf("\n"); return; } static int pccard_child_location_str(device_t bus, device_t child, char *buf, size_t buflen) { struct pccard_ivar *devi = PCCARD_IVAR(child); struct pccard_function *pf = devi->pf; snprintf(buf, buflen, "function=%d", pf->number); return (0); } /* XXX Maybe this should be in subr_bus? */ static void pccard_safe_quote(char *dst, const char *src, size_t len) { char *walker = dst, *ep = dst + len - 1; if (len == 0) return; while (src != NULL && walker < ep) { if (*src == '"') { if (ep - walker < 2) break; *walker++ = '\\'; } *walker++ = *src++; } *walker = '\0'; } static int pccard_child_pnpinfo_str(device_t bus, device_t child, char *buf, size_t buflen) { struct pccard_ivar *devi = PCCARD_IVAR(child); struct pccard_function *pf = devi->pf; struct pccard_softc *sc = PCCARD_SOFTC(bus); char cis0[128], cis1[128]; pccard_safe_quote(cis0, sc->card.cis1_info[0], sizeof(cis0)); pccard_safe_quote(cis1, sc->card.cis1_info[1], sizeof(cis1)); snprintf(buf, buflen, "manufacturer=0x%04x product=0x%04x " "cisvendor=\"%s\" cisproduct=\"%s\" function_type=%d", sc->card.manufacturer, sc->card.product, cis0, cis1, pf->function); return (0); } static int pccard_read_ivar(device_t bus, device_t child, int which, uintptr_t *result) { struct pccard_ivar *devi = PCCARD_IVAR(child); struct pccard_function *pf = devi->pf; struct pccard_softc *sc = PCCARD_SOFTC(bus); if (!pf) panic("No pccard function pointer"); switch (which) { default: return (EINVAL); case PCCARD_IVAR_FUNCE_DISK: *(uint16_t *)result = pf->pf_funce_disk_interface | (pf->pf_funce_disk_power << 8); break; case PCCARD_IVAR_ETHADDR: bcopy(pf->pf_funce_lan_nid, result, ETHER_ADDR_LEN); break; case PCCARD_IVAR_VENDOR: *(uint32_t *)result = sc->card.manufacturer; break; case PCCARD_IVAR_PRODUCT: *(uint32_t *)result = sc->card.product; break; case PCCARD_IVAR_PRODEXT: *(uint16_t *)result = sc->card.prodext; break; case PCCARD_IVAR_FUNCTION: *(uint32_t *)result = pf->function; break; case PCCARD_IVAR_FUNCTION_NUMBER: *(uint32_t *)result = pf->number; break; case PCCARD_IVAR_VENDOR_STR: *(const char **)result = sc->card.cis1_info[0]; break; case PCCARD_IVAR_PRODUCT_STR: *(const char **)result = sc->card.cis1_info[1]; break; case PCCARD_IVAR_CIS3_STR: *(const char **)result = sc->card.cis1_info[2]; break; case PCCARD_IVAR_CIS4_STR: *(const char **)result = sc->card.cis1_info[3]; break; } return (0); } static void pccard_driver_added(device_t dev, driver_t *driver) { struct pccard_softc *sc = PCCARD_SOFTC(dev); struct pccard_function *pf; device_t child; STAILQ_FOREACH(pf, &sc->card.pf_head, pf_list) { if (STAILQ_EMPTY(&pf->cfe_head)) continue; child = pf->dev; if (device_get_state(child) != DS_NOTPRESENT) continue; pccard_probe_and_attach_child(dev, child, pf); } return; } static struct resource * pccard_alloc_resource(device_t dev, device_t child, int type, int *rid, u_long start, u_long end, u_long count, u_int flags) { struct pccard_ivar *dinfo; struct resource_list_entry *rle = 0; int passthrough = (device_get_parent(child) != dev); int isdefault = (start == 0 && end == ~0UL && count == 1); struct resource *r = NULL; /* XXX I'm no longer sure this is right */ if (passthrough) { return (BUS_ALLOC_RESOURCE(device_get_parent(dev), child, type, rid, start, end, count, flags)); } dinfo = device_get_ivars(child); rle = resource_list_find(&dinfo->resources, type, *rid); if (rle == NULL && isdefault) return (NULL); /* no resource of that type/rid */ if (rle == NULL || rle->res == NULL) { /* XXX Need to adjust flags */ r = bus_alloc_resource(dev, type, rid, start, end, count, flags); if (r == NULL) goto bad; resource_list_add(&dinfo->resources, type, *rid, rman_get_start(r), rman_get_end(r), count); rle = resource_list_find(&dinfo->resources, type, *rid); if (!rle) goto bad; rle->res = r; } /* * If dev doesn't own the device, then we can't give this device * out. */ if (rman_get_device(rle->res) != dev) return (NULL); rman_set_device(rle->res, child); if (flags & RF_ACTIVE) BUS_ACTIVATE_RESOURCE(dev, child, type, *rid, rle->res); return (rle->res); bad:; device_printf(dev, "WARNING: Resource not reserved by pccard\n"); return (NULL); } static int pccard_release_resource(device_t dev, device_t child, int type, int rid, struct resource *r) { struct pccard_ivar *dinfo; int passthrough = (device_get_parent(child) != dev); struct resource_list_entry *rle = 0; if (passthrough) return BUS_RELEASE_RESOURCE(device_get_parent(dev), child, type, rid, r); dinfo = device_get_ivars(child); rle = resource_list_find(&dinfo->resources, type, rid); if (!rle) { device_printf(dev, "Allocated resource not found, " "%d %#x %#lx %#lx\n", type, rid, rman_get_start(r), rman_get_size(r)); return ENOENT; } if (!rle->res) { device_printf(dev, "Allocated resource not recorded\n"); return ENOENT; } /* * Deactivate the resource (since it is being released), and * assign it to the bus. */ BUS_DEACTIVATE_RESOURCE(dev, child, type, rid, rle->res); rman_set_device(rle->res, dev); return (0); } static void pccard_child_detached(device_t parent, device_t dev) { struct pccard_ivar *ivar = PCCARD_IVAR(dev); struct pccard_function *pf = ivar->pf; pccard_function_disable(pf); } static int pccard_filter(void *arg) { struct pccard_function *pf = (struct pccard_function*) arg; int reg; int doisr = 1; /* * MFC cards know if they interrupted, so we have to ack the * interrupt and call the ISR. Non-MFC cards don't have these * bits, so they always get called. Many non-MFC cards have * this bit set always upon read, but some do not. * * We always ack the interrupt, even if there's no ISR * for the card. This is done on the theory that acking * the interrupt will pacify the card enough to keep an * interrupt storm from happening. Of course this won't * help in the non-MFC case. * * This has no impact for MPSAFEness of the client drivers. * We register this with whatever flags the intr_handler * was registered with. All these functions are MPSAFE. */ if (pccard_mfc(pf->sc)) { reg = pccard_ccr_read(pf, PCCARD_CCR_STATUS); if (reg & PCCARD_CCR_STATUS_INTR) pccard_ccr_write(pf, PCCARD_CCR_STATUS, reg & ~PCCARD_CCR_STATUS_INTR); else doisr = 0; } if (doisr) { if (pf->intr_filter != NULL) return (pf->intr_filter(pf->intr_handler_arg)); return (FILTER_SCHEDULE_THREAD); } return (FILTER_STRAY); } static void pccard_intr(void *arg) { struct pccard_function *pf = (struct pccard_function*) arg; pf->intr_handler(pf->intr_handler_arg); } static int pccard_setup_intr(device_t dev, device_t child, struct resource *irq, int flags, driver_filter_t *filt, driver_intr_t *intr, void *arg, void **cookiep) { struct pccard_softc *sc = PCCARD_SOFTC(dev); struct pccard_ivar *ivar = PCCARD_IVAR(child); struct pccard_function *pf = ivar->pf; int err; if (pf->intr_filter != NULL || pf->intr_handler != NULL) panic("Only one interrupt handler per function allowed"); err = bus_generic_setup_intr(dev, child, irq, flags, pccard_filter, intr ? pccard_intr : NULL, pf, cookiep); if (err != 0) return (err); pf->intr_filter = filt; pf->intr_handler = intr; pf->intr_handler_arg = arg; pf->intr_handler_cookie = *cookiep; if (pccard_mfc(sc)) { pccard_ccr_write(pf, PCCARD_CCR_OPTION, pccard_ccr_read(pf, PCCARD_CCR_OPTION) | PCCARD_CCR_OPTION_IREQ_ENABLE); } return (0); } static int pccard_teardown_intr(device_t dev, device_t child, struct resource *r, void *cookie) { struct pccard_softc *sc = PCCARD_SOFTC(dev); struct pccard_ivar *ivar = PCCARD_IVAR(child); struct pccard_function *pf = ivar->pf; int ret; if (pccard_mfc(sc)) { pccard_ccr_write(pf, PCCARD_CCR_OPTION, pccard_ccr_read(pf, PCCARD_CCR_OPTION) & ~PCCARD_CCR_OPTION_IREQ_ENABLE); } ret = bus_generic_teardown_intr(dev, child, r, cookie); if (ret == 0) { pf->intr_handler = NULL; pf->intr_handler_arg = NULL; pf->intr_handler_cookie = NULL; } return (ret); } static int pccard_activate_resource(device_t brdev, device_t child, int type, int rid, struct resource *r) { struct pccard_ivar *ivar = PCCARD_IVAR(child); struct pccard_function *pf = ivar->pf; switch(type) { case SYS_RES_IOPORT: /* * We need to adjust IOBASE[01] and IOSIZE if we're an MFC * card. */ if (pccard_mfc(pf->sc)) pccard_mfc_adjust_iobase(pf, rman_get_start(r), 0, rman_get_size(r)); break; default: break; } return (bus_generic_activate_resource(brdev, child, type, rid, r)); } static int pccard_deactivate_resource(device_t brdev, device_t child, int type, int rid, struct resource *r) { /* XXX undo pccard_activate_resource? XXX */ return (bus_generic_deactivate_resource(brdev, child, type, rid, r)); } static int pccard_attr_read_impl(device_t brdev, device_t child, uint32_t offset, uint8_t *val) { struct pccard_ivar *devi = PCCARD_IVAR(child); struct pccard_function *pf = devi->pf; /* * Optimization. Most of the time, devices want to access * the same page of the attribute memory that the CCR is in. * We take advantage of this fact here. */ if (offset / PCCARD_MEM_PAGE_SIZE == pf->ccr_base / PCCARD_MEM_PAGE_SIZE) *val = bus_space_read_1(pf->pf_ccrt, pf->pf_ccrh, offset % PCCARD_MEM_PAGE_SIZE); else { CARD_SET_MEMORY_OFFSET(brdev, child, pf->ccr_rid, offset, &offset); *val = bus_space_read_1(pf->pf_ccrt, pf->pf_ccrh, offset); CARD_SET_MEMORY_OFFSET(brdev, child, pf->ccr_rid, pf->ccr_base, &offset); } return 0; } static int pccard_attr_write_impl(device_t brdev, device_t child, uint32_t offset, uint8_t val) { struct pccard_ivar *devi = PCCARD_IVAR(child); struct pccard_function *pf = devi->pf; /* * Optimization. Most of the time, devices want to access * the same page of the attribute memory that the CCR is in. * We take advantage of this fact here. */ if (offset / PCCARD_MEM_PAGE_SIZE == pf->ccr_base / PCCARD_MEM_PAGE_SIZE) bus_space_write_1(pf->pf_ccrt, pf->pf_ccrh, offset % PCCARD_MEM_PAGE_SIZE, val); else { CARD_SET_MEMORY_OFFSET(brdev, child, pf->ccr_rid, offset, &offset); bus_space_write_1(pf->pf_ccrt, pf->pf_ccrh, offset, val); CARD_SET_MEMORY_OFFSET(brdev, child, pf->ccr_rid, pf->ccr_base, &offset); } return 0; } static int pccard_ccr_read_impl(device_t brdev, device_t child, uint32_t offset, uint8_t *val) { struct pccard_ivar *devi = PCCARD_IVAR(child); *val = pccard_ccr_read(devi->pf, offset); - device_printf(child, "ccr_read of %#x (%#x) is %#x\n", offset, - devi->pf->pf_ccr_offset, *val); + DEVPRINTF((child, "ccr_read of %#x (%#x) is %#x\n", offset, + devi->pf->pf_ccr_offset, *val)); return 0; } static int pccard_ccr_write_impl(device_t brdev, device_t child, uint32_t offset, uint8_t val) { struct pccard_ivar *devi = PCCARD_IVAR(child); struct pccard_function *pf = devi->pf; /* * Can't use pccard_ccr_write since client drivers may access * registers not contained in the 'mask' if they are non-standard. */ - device_printf(child, "ccr_write of %#x to %#x (%#x)\n", val, offset, - devi->pf->pf_ccr_offset); + DEVPRINTF((child, "ccr_write of %#x to %#x (%#x)\n", val, offset, + devi->pf->pf_ccr_offset)); bus_space_write_1(pf->pf_ccrt, pf->pf_ccrh, pf->pf_ccr_offset + offset, val); return 0; } static device_method_t pccard_methods[] = { /* Device interface */ DEVMETHOD(device_probe, pccard_probe), DEVMETHOD(device_attach, pccard_attach), DEVMETHOD(device_detach, pccard_detach), DEVMETHOD(device_shutdown, bus_generic_shutdown), DEVMETHOD(device_suspend, pccard_suspend), DEVMETHOD(device_resume, pccard_resume), /* Bus interface */ DEVMETHOD(bus_print_child, pccard_print_child), DEVMETHOD(bus_driver_added, pccard_driver_added), DEVMETHOD(bus_child_detached, pccard_child_detached), DEVMETHOD(bus_alloc_resource, pccard_alloc_resource), DEVMETHOD(bus_release_resource, pccard_release_resource), DEVMETHOD(bus_activate_resource, pccard_activate_resource), DEVMETHOD(bus_deactivate_resource, pccard_deactivate_resource), DEVMETHOD(bus_setup_intr, pccard_setup_intr), DEVMETHOD(bus_teardown_intr, pccard_teardown_intr), DEVMETHOD(bus_set_resource, pccard_set_resource), DEVMETHOD(bus_get_resource, pccard_get_resource), DEVMETHOD(bus_delete_resource, pccard_delete_resource), DEVMETHOD(bus_probe_nomatch, pccard_probe_nomatch), DEVMETHOD(bus_read_ivar, pccard_read_ivar), DEVMETHOD(bus_child_pnpinfo_str, pccard_child_pnpinfo_str), DEVMETHOD(bus_child_location_str, pccard_child_location_str), /* Card Interface */ DEVMETHOD(card_set_res_flags, pccard_set_res_flags), DEVMETHOD(card_set_memory_offset, pccard_set_memory_offset), DEVMETHOD(card_attach_card, pccard_attach_card), DEVMETHOD(card_detach_card, pccard_detach_card), DEVMETHOD(card_do_product_lookup, pccard_do_product_lookup), DEVMETHOD(card_cis_scan, pccard_scan_cis), DEVMETHOD(card_attr_read, pccard_attr_read_impl), DEVMETHOD(card_attr_write, pccard_attr_write_impl), DEVMETHOD(card_ccr_read, pccard_ccr_read_impl), DEVMETHOD(card_ccr_write, pccard_ccr_write_impl), { 0, 0 } }; static driver_t pccard_driver = { "pccard", pccard_methods, sizeof(struct pccard_softc) }; devclass_t pccard_devclass; /* Maybe we need to have a slot device? */ DRIVER_MODULE(pccard, pcic, pccard_driver, pccard_devclass, 0, 0); DRIVER_MODULE(pccard, cbb, pccard_driver, pccard_devclass, 0, 0); MODULE_VERSION(pccard, 1); Index: projects/largeSMP/sys/dev/pci/pci.c =================================================================== --- projects/largeSMP/sys/dev/pci/pci.c (revision 222782) +++ projects/largeSMP/sys/dev/pci/pci.c (revision 222783) @@ -1,4342 +1,4353 @@ /*- * Copyright (c) 1997, Stefan Esser * Copyright (c) 2000, 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 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 "opt_bus.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #if defined(__i386__) || defined(__amd64__) || defined(__powerpc__) #include #endif #include #include #include #include #include #include #include #include "pcib_if.h" #include "pci_if.h" #define PCIR_IS_BIOS(cfg, reg) \ (((cfg)->hdrtype == PCIM_HDRTYPE_NORMAL && reg == PCIR_BIOS) || \ ((cfg)->hdrtype == PCIM_HDRTYPE_BRIDGE && reg == PCIR_BIOS_1)) static pci_addr_t pci_mapbase(uint64_t mapreg); static const char *pci_maptype(uint64_t mapreg); static int pci_mapsize(uint64_t testval); static int pci_maprange(uint64_t mapreg); static pci_addr_t pci_rombase(uint64_t mapreg); static int pci_romsize(uint64_t testval); static void pci_fixancient(pcicfgregs *cfg); static int pci_printf(pcicfgregs *cfg, const char *fmt, ...); static int pci_porten(device_t dev); static int pci_memen(device_t dev); static void pci_assign_interrupt(device_t bus, device_t dev, int force_route); static int pci_add_map(device_t bus, device_t dev, int reg, struct resource_list *rl, int force, int prefetch); static int pci_probe(device_t dev); static int pci_attach(device_t dev); static void pci_load_vendor_data(void); static int pci_describe_parse_line(char **ptr, int *vendor, int *device, char **desc); static char *pci_describe_device(device_t dev); static int pci_modevent(module_t mod, int what, void *arg); static void pci_hdrtypedata(device_t pcib, int b, int s, int f, pcicfgregs *cfg); static void pci_read_cap(device_t pcib, pcicfgregs *cfg); static int pci_read_vpd_reg(device_t pcib, pcicfgregs *cfg, int reg, uint32_t *data); #if 0 static int pci_write_vpd_reg(device_t pcib, pcicfgregs *cfg, int reg, uint32_t data); #endif static void pci_read_vpd(device_t pcib, pcicfgregs *cfg); static void pci_disable_msi(device_t dev); static void pci_enable_msi(device_t dev, uint64_t address, uint16_t data); static void pci_enable_msix(device_t dev, u_int index, uint64_t address, uint32_t data); static void pci_mask_msix(device_t dev, u_int index); static void pci_unmask_msix(device_t dev, u_int index); static int pci_msi_blacklisted(void); static void pci_resume_msi(device_t dev); static void pci_resume_msix(device_t dev); static int pci_remap_intr_method(device_t bus, device_t dev, u_int irq); static device_method_t pci_methods[] = { /* Device interface */ DEVMETHOD(device_probe, pci_probe), DEVMETHOD(device_attach, pci_attach), DEVMETHOD(device_detach, bus_generic_detach), DEVMETHOD(device_shutdown, bus_generic_shutdown), DEVMETHOD(device_suspend, pci_suspend), DEVMETHOD(device_resume, pci_resume), /* Bus interface */ DEVMETHOD(bus_print_child, pci_print_child), DEVMETHOD(bus_probe_nomatch, pci_probe_nomatch), DEVMETHOD(bus_read_ivar, pci_read_ivar), DEVMETHOD(bus_write_ivar, pci_write_ivar), DEVMETHOD(bus_driver_added, pci_driver_added), DEVMETHOD(bus_setup_intr, pci_setup_intr), DEVMETHOD(bus_teardown_intr, pci_teardown_intr), DEVMETHOD(bus_get_resource_list,pci_get_resource_list), DEVMETHOD(bus_set_resource, bus_generic_rl_set_resource), DEVMETHOD(bus_get_resource, bus_generic_rl_get_resource), DEVMETHOD(bus_delete_resource, pci_delete_resource), DEVMETHOD(bus_alloc_resource, pci_alloc_resource), DEVMETHOD(bus_adjust_resource, bus_generic_adjust_resource), DEVMETHOD(bus_release_resource, bus_generic_rl_release_resource), DEVMETHOD(bus_activate_resource, pci_activate_resource), DEVMETHOD(bus_deactivate_resource, pci_deactivate_resource), DEVMETHOD(bus_child_pnpinfo_str, pci_child_pnpinfo_str_method), DEVMETHOD(bus_child_location_str, pci_child_location_str_method), DEVMETHOD(bus_remap_intr, pci_remap_intr_method), /* PCI interface */ DEVMETHOD(pci_read_config, pci_read_config_method), DEVMETHOD(pci_write_config, pci_write_config_method), DEVMETHOD(pci_enable_busmaster, pci_enable_busmaster_method), DEVMETHOD(pci_disable_busmaster, pci_disable_busmaster_method), DEVMETHOD(pci_enable_io, pci_enable_io_method), DEVMETHOD(pci_disable_io, pci_disable_io_method), DEVMETHOD(pci_get_vpd_ident, pci_get_vpd_ident_method), DEVMETHOD(pci_get_vpd_readonly, pci_get_vpd_readonly_method), DEVMETHOD(pci_get_powerstate, pci_get_powerstate_method), DEVMETHOD(pci_set_powerstate, pci_set_powerstate_method), DEVMETHOD(pci_assign_interrupt, pci_assign_interrupt_method), DEVMETHOD(pci_find_extcap, pci_find_extcap_method), DEVMETHOD(pci_alloc_msi, pci_alloc_msi_method), DEVMETHOD(pci_alloc_msix, pci_alloc_msix_method), DEVMETHOD(pci_remap_msix, pci_remap_msix_method), DEVMETHOD(pci_release_msi, pci_release_msi_method), DEVMETHOD(pci_msi_count, pci_msi_count_method), DEVMETHOD(pci_msix_count, pci_msix_count_method), { 0, 0 } }; DEFINE_CLASS_0(pci, pci_driver, pci_methods, 0); static devclass_t pci_devclass; DRIVER_MODULE(pci, pcib, pci_driver, pci_devclass, pci_modevent, 0); MODULE_VERSION(pci, 1); static char *pci_vendordata; static size_t pci_vendordata_size; struct pci_quirk { uint32_t devid; /* Vendor/device of the card */ int type; #define PCI_QUIRK_MAP_REG 1 /* PCI map register in weird place */ #define PCI_QUIRK_DISABLE_MSI 2 /* MSI/MSI-X doesn't work */ #define PCI_QUIRK_ENABLE_MSI_VM 3 /* Older chipset in VM where MSI works */ int arg1; int arg2; }; struct pci_quirk pci_quirks[] = { /* The Intel 82371AB and 82443MX has a map register at offset 0x90. */ { 0x71138086, PCI_QUIRK_MAP_REG, 0x90, 0 }, { 0x719b8086, PCI_QUIRK_MAP_REG, 0x90, 0 }, /* As does the Serverworks OSB4 (the SMBus mapping register) */ { 0x02001166, PCI_QUIRK_MAP_REG, 0x90, 0 }, /* * MSI doesn't work with the ServerWorks CNB20-HE Host Bridge * or the CMIC-SL (AKA ServerWorks GC_LE). */ { 0x00141166, PCI_QUIRK_DISABLE_MSI, 0, 0 }, { 0x00171166, PCI_QUIRK_DISABLE_MSI, 0, 0 }, /* * MSI doesn't work on earlier Intel chipsets including * E7500, E7501, E7505, 845, 865, 875/E7210, and 855. */ { 0x25408086, PCI_QUIRK_DISABLE_MSI, 0, 0 }, { 0x254c8086, PCI_QUIRK_DISABLE_MSI, 0, 0 }, { 0x25508086, PCI_QUIRK_DISABLE_MSI, 0, 0 }, { 0x25608086, PCI_QUIRK_DISABLE_MSI, 0, 0 }, { 0x25708086, PCI_QUIRK_DISABLE_MSI, 0, 0 }, { 0x25788086, PCI_QUIRK_DISABLE_MSI, 0, 0 }, { 0x35808086, PCI_QUIRK_DISABLE_MSI, 0, 0 }, /* * MSI doesn't work with devices behind the AMD 8131 HT-PCIX * bridge. */ { 0x74501022, PCI_QUIRK_DISABLE_MSI, 0, 0 }, /* * Some virtualization environments emulate an older chipset * but support MSI just fine. QEMU uses the Intel 82440. */ { 0x12378086, PCI_QUIRK_ENABLE_MSI_VM, 0, 0 }, { 0 } }; /* map register information */ #define PCI_MAPMEM 0x01 /* memory map */ #define PCI_MAPMEMP 0x02 /* prefetchable memory map */ #define PCI_MAPPORT 0x04 /* port map */ struct devlist pci_devq; uint32_t pci_generation; uint32_t pci_numdevs = 0; static int pcie_chipset, pcix_chipset; /* sysctl vars */ SYSCTL_NODE(_hw, OID_AUTO, pci, CTLFLAG_RD, 0, "PCI bus tuning parameters"); static int pci_enable_io_modes = 1; TUNABLE_INT("hw.pci.enable_io_modes", &pci_enable_io_modes); SYSCTL_INT(_hw_pci, OID_AUTO, enable_io_modes, CTLFLAG_RW, &pci_enable_io_modes, 1, "Enable I/O and memory bits in the config register. Some BIOSes do not\n\ enable these bits correctly. We'd like to do this all the time, but there\n\ are some peripherals that this causes problems with."); static int pci_do_power_nodriver = 0; TUNABLE_INT("hw.pci.do_power_nodriver", &pci_do_power_nodriver); SYSCTL_INT(_hw_pci, OID_AUTO, do_power_nodriver, CTLFLAG_RW, &pci_do_power_nodriver, 0, "Place a function into D3 state when no driver attaches to it. 0 means\n\ disable. 1 means conservatively place devices into D3 state. 2 means\n\ agressively place devices into D3 state. 3 means put absolutely everything\n\ in D3 state."); int pci_do_power_resume = 1; TUNABLE_INT("hw.pci.do_power_resume", &pci_do_power_resume); SYSCTL_INT(_hw_pci, OID_AUTO, do_power_resume, CTLFLAG_RW, &pci_do_power_resume, 1, "Transition from D3 -> D0 on resume."); int pci_do_power_suspend = 1; TUNABLE_INT("hw.pci.do_power_suspend", &pci_do_power_suspend); SYSCTL_INT(_hw_pci, OID_AUTO, do_power_suspend, CTLFLAG_RW, &pci_do_power_suspend, 1, "Transition from D0 -> D3 on suspend."); static int pci_do_msi = 1; TUNABLE_INT("hw.pci.enable_msi", &pci_do_msi); SYSCTL_INT(_hw_pci, OID_AUTO, enable_msi, CTLFLAG_RW, &pci_do_msi, 1, "Enable support for MSI interrupts"); static int pci_do_msix = 1; TUNABLE_INT("hw.pci.enable_msix", &pci_do_msix); SYSCTL_INT(_hw_pci, OID_AUTO, enable_msix, CTLFLAG_RW, &pci_do_msix, 1, "Enable support for MSI-X interrupts"); static int pci_honor_msi_blacklist = 1; TUNABLE_INT("hw.pci.honor_msi_blacklist", &pci_honor_msi_blacklist); SYSCTL_INT(_hw_pci, OID_AUTO, honor_msi_blacklist, CTLFLAG_RD, &pci_honor_msi_blacklist, 1, "Honor chipset blacklist for MSI"); #if defined(__i386__) || defined(__amd64__) static int pci_usb_takeover = 1; #else static int pci_usb_takeover = 0; #endif TUNABLE_INT("hw.pci.usb_early_takeover", &pci_usb_takeover); SYSCTL_INT(_hw_pci, OID_AUTO, usb_early_takeover, CTLFLAG_RD | CTLFLAG_TUN, &pci_usb_takeover, 1, "Enable early takeover of USB controllers.\n\ Disable this if you depend on BIOS emulation of USB devices, that is\n\ you use USB devices (like keyboard or mouse) but do not load USB drivers"); /* Find a device_t by bus/slot/function in domain 0 */ device_t pci_find_bsf(uint8_t bus, uint8_t slot, uint8_t func) { return (pci_find_dbsf(0, bus, slot, func)); } /* Find a device_t by domain/bus/slot/function */ device_t pci_find_dbsf(uint32_t domain, uint8_t bus, uint8_t slot, uint8_t func) { struct pci_devinfo *dinfo; STAILQ_FOREACH(dinfo, &pci_devq, pci_links) { if ((dinfo->cfg.domain == domain) && (dinfo->cfg.bus == bus) && (dinfo->cfg.slot == slot) && (dinfo->cfg.func == func)) { return (dinfo->cfg.dev); } } return (NULL); } /* Find a device_t by vendor/device ID */ device_t pci_find_device(uint16_t vendor, uint16_t device) { struct pci_devinfo *dinfo; STAILQ_FOREACH(dinfo, &pci_devq, pci_links) { if ((dinfo->cfg.vendor == vendor) && (dinfo->cfg.device == device)) { return (dinfo->cfg.dev); } } return (NULL); } static int pci_printf(pcicfgregs *cfg, const char *fmt, ...) { va_list ap; int retval; retval = printf("pci%d:%d:%d:%d: ", cfg->domain, cfg->bus, cfg->slot, cfg->func); va_start(ap, fmt); retval += vprintf(fmt, ap); va_end(ap); return (retval); } /* return base address of memory or port map */ static pci_addr_t pci_mapbase(uint64_t mapreg) { if (PCI_BAR_MEM(mapreg)) return (mapreg & PCIM_BAR_MEM_BASE); else return (mapreg & PCIM_BAR_IO_BASE); } /* return map type of memory or port map */ static const char * pci_maptype(uint64_t mapreg) { if (PCI_BAR_IO(mapreg)) return ("I/O Port"); if (mapreg & PCIM_BAR_MEM_PREFETCH) return ("Prefetchable Memory"); return ("Memory"); } /* return log2 of map size decoded for memory or port map */ static int pci_mapsize(uint64_t testval) { int ln2size; testval = pci_mapbase(testval); ln2size = 0; if (testval != 0) { while ((testval & 1) == 0) { ln2size++; testval >>= 1; } } return (ln2size); } /* return base address of device ROM */ static pci_addr_t pci_rombase(uint64_t mapreg) { return (mapreg & PCIM_BIOS_ADDR_MASK); } /* return log2 of map size decided for device ROM */ static int pci_romsize(uint64_t testval) { int ln2size; testval = pci_rombase(testval); ln2size = 0; if (testval != 0) { while ((testval & 1) == 0) { ln2size++; testval >>= 1; } } return (ln2size); } /* return log2 of address range supported by map register */ static int pci_maprange(uint64_t mapreg) { int ln2range = 0; if (PCI_BAR_IO(mapreg)) ln2range = 32; else switch (mapreg & PCIM_BAR_MEM_TYPE) { case PCIM_BAR_MEM_32: ln2range = 32; break; case PCIM_BAR_MEM_1MB: ln2range = 20; break; case PCIM_BAR_MEM_64: ln2range = 64; break; } return (ln2range); } /* adjust some values from PCI 1.0 devices to match 2.0 standards ... */ static void pci_fixancient(pcicfgregs *cfg) { if ((cfg->hdrtype & PCIM_HDRTYPE) != PCIM_HDRTYPE_NORMAL) return; /* PCI to PCI bridges use header type 1 */ if (cfg->baseclass == PCIC_BRIDGE && cfg->subclass == PCIS_BRIDGE_PCI) cfg->hdrtype = PCIM_HDRTYPE_BRIDGE; } /* extract header type specific config data */ static void pci_hdrtypedata(device_t pcib, int b, int s, int f, pcicfgregs *cfg) { #define REG(n, w) PCIB_READ_CONFIG(pcib, b, s, f, n, w) switch (cfg->hdrtype & PCIM_HDRTYPE) { case PCIM_HDRTYPE_NORMAL: cfg->subvendor = REG(PCIR_SUBVEND_0, 2); cfg->subdevice = REG(PCIR_SUBDEV_0, 2); cfg->nummaps = PCI_MAXMAPS_0; break; case PCIM_HDRTYPE_BRIDGE: cfg->nummaps = PCI_MAXMAPS_1; break; case PCIM_HDRTYPE_CARDBUS: cfg->subvendor = REG(PCIR_SUBVEND_2, 2); cfg->subdevice = REG(PCIR_SUBDEV_2, 2); cfg->nummaps = PCI_MAXMAPS_2; break; } #undef REG } /* read configuration header into pcicfgregs structure */ struct pci_devinfo * pci_read_device(device_t pcib, int d, int b, int s, int f, size_t size) { #define REG(n, w) PCIB_READ_CONFIG(pcib, b, s, f, n, w) pcicfgregs *cfg = NULL; struct pci_devinfo *devlist_entry; struct devlist *devlist_head; devlist_head = &pci_devq; devlist_entry = NULL; if (REG(PCIR_DEVVENDOR, 4) != 0xfffffffful) { devlist_entry = malloc(size, M_DEVBUF, M_WAITOK | M_ZERO); if (devlist_entry == NULL) return (NULL); cfg = &devlist_entry->cfg; cfg->domain = d; cfg->bus = b; cfg->slot = s; cfg->func = f; cfg->vendor = REG(PCIR_VENDOR, 2); cfg->device = REG(PCIR_DEVICE, 2); cfg->cmdreg = REG(PCIR_COMMAND, 2); cfg->statreg = REG(PCIR_STATUS, 2); cfg->baseclass = REG(PCIR_CLASS, 1); cfg->subclass = REG(PCIR_SUBCLASS, 1); cfg->progif = REG(PCIR_PROGIF, 1); cfg->revid = REG(PCIR_REVID, 1); cfg->hdrtype = REG(PCIR_HDRTYPE, 1); cfg->cachelnsz = REG(PCIR_CACHELNSZ, 1); cfg->lattimer = REG(PCIR_LATTIMER, 1); cfg->intpin = REG(PCIR_INTPIN, 1); cfg->intline = REG(PCIR_INTLINE, 1); cfg->mingnt = REG(PCIR_MINGNT, 1); cfg->maxlat = REG(PCIR_MAXLAT, 1); cfg->mfdev = (cfg->hdrtype & PCIM_MFDEV) != 0; cfg->hdrtype &= ~PCIM_MFDEV; STAILQ_INIT(&cfg->maps); pci_fixancient(cfg); pci_hdrtypedata(pcib, b, s, f, cfg); if (REG(PCIR_STATUS, 2) & PCIM_STATUS_CAPPRESENT) pci_read_cap(pcib, cfg); STAILQ_INSERT_TAIL(devlist_head, devlist_entry, pci_links); devlist_entry->conf.pc_sel.pc_domain = cfg->domain; devlist_entry->conf.pc_sel.pc_bus = cfg->bus; devlist_entry->conf.pc_sel.pc_dev = cfg->slot; devlist_entry->conf.pc_sel.pc_func = cfg->func; devlist_entry->conf.pc_hdr = cfg->hdrtype; devlist_entry->conf.pc_subvendor = cfg->subvendor; devlist_entry->conf.pc_subdevice = cfg->subdevice; devlist_entry->conf.pc_vendor = cfg->vendor; devlist_entry->conf.pc_device = cfg->device; devlist_entry->conf.pc_class = cfg->baseclass; devlist_entry->conf.pc_subclass = cfg->subclass; devlist_entry->conf.pc_progif = cfg->progif; devlist_entry->conf.pc_revid = cfg->revid; pci_numdevs++; pci_generation++; } return (devlist_entry); #undef REG } static void pci_read_cap(device_t pcib, pcicfgregs *cfg) { #define REG(n, w) PCIB_READ_CONFIG(pcib, cfg->bus, cfg->slot, cfg->func, n, w) #define WREG(n, v, w) PCIB_WRITE_CONFIG(pcib, cfg->bus, cfg->slot, cfg->func, n, v, w) #if defined(__i386__) || defined(__amd64__) || defined(__powerpc__) uint64_t addr; #endif uint32_t val; int ptr, nextptr, ptrptr; switch (cfg->hdrtype & PCIM_HDRTYPE) { case PCIM_HDRTYPE_NORMAL: case PCIM_HDRTYPE_BRIDGE: ptrptr = PCIR_CAP_PTR; break; case PCIM_HDRTYPE_CARDBUS: ptrptr = PCIR_CAP_PTR_2; /* cardbus capabilities ptr */ break; default: return; /* no extended capabilities support */ } nextptr = REG(ptrptr, 1); /* sanity check? */ /* * Read capability entries. */ while (nextptr != 0) { /* Sanity check */ if (nextptr > 255) { printf("illegal PCI extended capability offset %d\n", nextptr); return; } /* Find the next entry */ ptr = nextptr; nextptr = REG(ptr + PCICAP_NEXTPTR, 1); /* Process this entry */ switch (REG(ptr + PCICAP_ID, 1)) { case PCIY_PMG: /* PCI power management */ if (cfg->pp.pp_cap == 0) { cfg->pp.pp_cap = REG(ptr + PCIR_POWER_CAP, 2); cfg->pp.pp_status = ptr + PCIR_POWER_STATUS; cfg->pp.pp_bse = ptr + PCIR_POWER_BSE; if ((nextptr - ptr) > PCIR_POWER_DATA) cfg->pp.pp_data = ptr + PCIR_POWER_DATA; } break; case PCIY_HT: /* HyperTransport */ /* Determine HT-specific capability type. */ val = REG(ptr + PCIR_HT_COMMAND, 2); if ((val & 0xe000) == PCIM_HTCAP_SLAVE) cfg->ht.ht_slave = ptr; #if defined(__i386__) || defined(__amd64__) || defined(__powerpc__) switch (val & PCIM_HTCMD_CAP_MASK) { case PCIM_HTCAP_MSI_MAPPING: if (!(val & PCIM_HTCMD_MSI_FIXED)) { /* Sanity check the mapping window. */ addr = REG(ptr + PCIR_HTMSI_ADDRESS_HI, 4); addr <<= 32; addr |= REG(ptr + PCIR_HTMSI_ADDRESS_LO, 4); if (addr != MSI_INTEL_ADDR_BASE) device_printf(pcib, "HT device at pci%d:%d:%d:%d has non-default MSI window 0x%llx\n", cfg->domain, cfg->bus, cfg->slot, cfg->func, (long long)addr); } else addr = MSI_INTEL_ADDR_BASE; cfg->ht.ht_msimap = ptr; cfg->ht.ht_msictrl = val; cfg->ht.ht_msiaddr = addr; break; } #endif break; case PCIY_MSI: /* PCI MSI */ cfg->msi.msi_location = ptr; cfg->msi.msi_ctrl = REG(ptr + PCIR_MSI_CTRL, 2); cfg->msi.msi_msgnum = 1 << ((cfg->msi.msi_ctrl & PCIM_MSICTRL_MMC_MASK)>>1); break; case PCIY_MSIX: /* PCI MSI-X */ cfg->msix.msix_location = ptr; cfg->msix.msix_ctrl = REG(ptr + PCIR_MSIX_CTRL, 2); cfg->msix.msix_msgnum = (cfg->msix.msix_ctrl & PCIM_MSIXCTRL_TABLE_SIZE) + 1; val = REG(ptr + PCIR_MSIX_TABLE, 4); cfg->msix.msix_table_bar = PCIR_BAR(val & PCIM_MSIX_BIR_MASK); cfg->msix.msix_table_offset = val & ~PCIM_MSIX_BIR_MASK; val = REG(ptr + PCIR_MSIX_PBA, 4); cfg->msix.msix_pba_bar = PCIR_BAR(val & PCIM_MSIX_BIR_MASK); cfg->msix.msix_pba_offset = val & ~PCIM_MSIX_BIR_MASK; break; case PCIY_VPD: /* PCI Vital Product Data */ cfg->vpd.vpd_reg = ptr; break; case PCIY_SUBVENDOR: /* Should always be true. */ if ((cfg->hdrtype & PCIM_HDRTYPE) == PCIM_HDRTYPE_BRIDGE) { val = REG(ptr + PCIR_SUBVENDCAP_ID, 4); cfg->subvendor = val & 0xffff; cfg->subdevice = val >> 16; } break; case PCIY_PCIX: /* PCI-X */ /* * Assume we have a PCI-X chipset if we have * at least one PCI-PCI bridge with a PCI-X * capability. Note that some systems with * PCI-express or HT chipsets might match on * this check as well. */ if ((cfg->hdrtype & PCIM_HDRTYPE) == PCIM_HDRTYPE_BRIDGE) pcix_chipset = 1; break; case PCIY_EXPRESS: /* PCI-express */ /* * Assume we have a PCI-express chipset if we have * at least one PCI-express device. */ pcie_chipset = 1; break; default: break; } } #if defined(__i386__) || defined(__amd64__) || defined(__powerpc__) /* * Enable the MSI mapping window for all HyperTransport * slaves. PCI-PCI bridges have their windows enabled via * PCIB_MAP_MSI(). */ if (cfg->ht.ht_slave != 0 && cfg->ht.ht_msimap != 0 && !(cfg->ht.ht_msictrl & PCIM_HTCMD_MSI_ENABLE)) { device_printf(pcib, "Enabling MSI window for HyperTransport slave at pci%d:%d:%d:%d\n", cfg->domain, cfg->bus, cfg->slot, cfg->func); cfg->ht.ht_msictrl |= PCIM_HTCMD_MSI_ENABLE; WREG(cfg->ht.ht_msimap + PCIR_HT_COMMAND, cfg->ht.ht_msictrl, 2); } #endif /* REG and WREG use carry through to next functions */ } /* * PCI Vital Product Data */ #define PCI_VPD_TIMEOUT 1000000 static int pci_read_vpd_reg(device_t pcib, pcicfgregs *cfg, int reg, uint32_t *data) { int count = PCI_VPD_TIMEOUT; KASSERT((reg & 3) == 0, ("VPD register must by 4 byte aligned")); WREG(cfg->vpd.vpd_reg + PCIR_VPD_ADDR, reg, 2); while ((REG(cfg->vpd.vpd_reg + PCIR_VPD_ADDR, 2) & 0x8000) != 0x8000) { if (--count < 0) return (ENXIO); DELAY(1); /* limit looping */ } *data = (REG(cfg->vpd.vpd_reg + PCIR_VPD_DATA, 4)); return (0); } #if 0 static int pci_write_vpd_reg(device_t pcib, pcicfgregs *cfg, int reg, uint32_t data) { int count = PCI_VPD_TIMEOUT; KASSERT((reg & 3) == 0, ("VPD register must by 4 byte aligned")); WREG(cfg->vpd.vpd_reg + PCIR_VPD_DATA, data, 4); WREG(cfg->vpd.vpd_reg + PCIR_VPD_ADDR, reg | 0x8000, 2); while ((REG(cfg->vpd.vpd_reg + PCIR_VPD_ADDR, 2) & 0x8000) == 0x8000) { if (--count < 0) return (ENXIO); DELAY(1); /* limit looping */ } return (0); } #endif #undef PCI_VPD_TIMEOUT struct vpd_readstate { device_t pcib; pcicfgregs *cfg; uint32_t val; int bytesinval; int off; uint8_t cksum; }; static int vpd_nextbyte(struct vpd_readstate *vrs, uint8_t *data) { uint32_t reg; uint8_t byte; if (vrs->bytesinval == 0) { if (pci_read_vpd_reg(vrs->pcib, vrs->cfg, vrs->off, ®)) return (ENXIO); vrs->val = le32toh(reg); vrs->off += 4; byte = vrs->val & 0xff; vrs->bytesinval = 3; } else { vrs->val = vrs->val >> 8; byte = vrs->val & 0xff; vrs->bytesinval--; } vrs->cksum += byte; *data = byte; return (0); } static void pci_read_vpd(device_t pcib, pcicfgregs *cfg) { struct vpd_readstate vrs; int state; int name; int remain; int i; int alloc, off; /* alloc/off for RO/W arrays */ int cksumvalid; int dflen; uint8_t byte; uint8_t byte2; /* init vpd reader */ vrs.bytesinval = 0; vrs.off = 0; vrs.pcib = pcib; vrs.cfg = cfg; vrs.cksum = 0; state = 0; name = remain = i = 0; /* shut up stupid gcc */ alloc = off = 0; /* shut up stupid gcc */ dflen = 0; /* shut up stupid gcc */ cksumvalid = -1; while (state >= 0) { if (vpd_nextbyte(&vrs, &byte)) { state = -2; break; } #if 0 printf("vpd: val: %#x, off: %d, bytesinval: %d, byte: %#hhx, " \ "state: %d, remain: %d, name: %#x, i: %d\n", vrs.val, vrs.off, vrs.bytesinval, byte, state, remain, name, i); #endif switch (state) { case 0: /* item name */ if (byte & 0x80) { if (vpd_nextbyte(&vrs, &byte2)) { state = -2; break; } remain = byte2; if (vpd_nextbyte(&vrs, &byte2)) { state = -2; break; } remain |= byte2 << 8; if (remain > (0x7f*4 - vrs.off)) { state = -1; printf( "pci%d:%d:%d:%d: invalid VPD data, remain %#x\n", cfg->domain, cfg->bus, cfg->slot, cfg->func, remain); } name = byte & 0x7f; } else { remain = byte & 0x7; name = (byte >> 3) & 0xf; } switch (name) { case 0x2: /* String */ cfg->vpd.vpd_ident = malloc(remain + 1, M_DEVBUF, M_WAITOK); i = 0; state = 1; break; case 0xf: /* End */ state = -1; break; case 0x10: /* VPD-R */ alloc = 8; off = 0; cfg->vpd.vpd_ros = malloc(alloc * sizeof(*cfg->vpd.vpd_ros), M_DEVBUF, M_WAITOK | M_ZERO); state = 2; break; case 0x11: /* VPD-W */ alloc = 8; off = 0; cfg->vpd.vpd_w = malloc(alloc * sizeof(*cfg->vpd.vpd_w), M_DEVBUF, M_WAITOK | M_ZERO); state = 5; break; default: /* Invalid data, abort */ state = -1; break; } break; case 1: /* Identifier String */ cfg->vpd.vpd_ident[i++] = byte; remain--; if (remain == 0) { cfg->vpd.vpd_ident[i] = '\0'; state = 0; } break; case 2: /* VPD-R Keyword Header */ if (off == alloc) { cfg->vpd.vpd_ros = reallocf(cfg->vpd.vpd_ros, (alloc *= 2) * sizeof(*cfg->vpd.vpd_ros), M_DEVBUF, M_WAITOK | M_ZERO); } cfg->vpd.vpd_ros[off].keyword[0] = byte; if (vpd_nextbyte(&vrs, &byte2)) { state = -2; break; } cfg->vpd.vpd_ros[off].keyword[1] = byte2; if (vpd_nextbyte(&vrs, &byte2)) { state = -2; break; } dflen = byte2; if (dflen == 0 && strncmp(cfg->vpd.vpd_ros[off].keyword, "RV", 2) == 0) { /* * if this happens, we can't trust the rest * of the VPD. */ printf( "pci%d:%d:%d:%d: bad keyword length: %d\n", cfg->domain, cfg->bus, cfg->slot, cfg->func, dflen); cksumvalid = 0; state = -1; break; } else if (dflen == 0) { cfg->vpd.vpd_ros[off].value = malloc(1 * sizeof(*cfg->vpd.vpd_ros[off].value), M_DEVBUF, M_WAITOK); cfg->vpd.vpd_ros[off].value[0] = '\x00'; } else cfg->vpd.vpd_ros[off].value = malloc( (dflen + 1) * sizeof(*cfg->vpd.vpd_ros[off].value), M_DEVBUF, M_WAITOK); remain -= 3; i = 0; /* keep in sync w/ state 3's transistions */ if (dflen == 0 && remain == 0) state = 0; else if (dflen == 0) state = 2; else state = 3; break; case 3: /* VPD-R Keyword Value */ cfg->vpd.vpd_ros[off].value[i++] = byte; if (strncmp(cfg->vpd.vpd_ros[off].keyword, "RV", 2) == 0 && cksumvalid == -1) { if (vrs.cksum == 0) cksumvalid = 1; else { if (bootverbose) printf( "pci%d:%d:%d:%d: bad VPD cksum, remain %hhu\n", cfg->domain, cfg->bus, cfg->slot, cfg->func, vrs.cksum); cksumvalid = 0; state = -1; break; } } dflen--; remain--; /* keep in sync w/ state 2's transistions */ if (dflen == 0) cfg->vpd.vpd_ros[off++].value[i++] = '\0'; if (dflen == 0 && remain == 0) { cfg->vpd.vpd_rocnt = off; cfg->vpd.vpd_ros = reallocf(cfg->vpd.vpd_ros, off * sizeof(*cfg->vpd.vpd_ros), M_DEVBUF, M_WAITOK | M_ZERO); state = 0; } else if (dflen == 0) state = 2; break; case 4: remain--; if (remain == 0) state = 0; break; case 5: /* VPD-W Keyword Header */ if (off == alloc) { cfg->vpd.vpd_w = reallocf(cfg->vpd.vpd_w, (alloc *= 2) * sizeof(*cfg->vpd.vpd_w), M_DEVBUF, M_WAITOK | M_ZERO); } cfg->vpd.vpd_w[off].keyword[0] = byte; if (vpd_nextbyte(&vrs, &byte2)) { state = -2; break; } cfg->vpd.vpd_w[off].keyword[1] = byte2; if (vpd_nextbyte(&vrs, &byte2)) { state = -2; break; } cfg->vpd.vpd_w[off].len = dflen = byte2; cfg->vpd.vpd_w[off].start = vrs.off - vrs.bytesinval; cfg->vpd.vpd_w[off].value = malloc((dflen + 1) * sizeof(*cfg->vpd.vpd_w[off].value), M_DEVBUF, M_WAITOK); remain -= 3; i = 0; /* keep in sync w/ state 6's transistions */ if (dflen == 0 && remain == 0) state = 0; else if (dflen == 0) state = 5; else state = 6; break; case 6: /* VPD-W Keyword Value */ cfg->vpd.vpd_w[off].value[i++] = byte; dflen--; remain--; /* keep in sync w/ state 5's transistions */ if (dflen == 0) cfg->vpd.vpd_w[off++].value[i++] = '\0'; if (dflen == 0 && remain == 0) { cfg->vpd.vpd_wcnt = off; cfg->vpd.vpd_w = reallocf(cfg->vpd.vpd_w, off * sizeof(*cfg->vpd.vpd_w), M_DEVBUF, M_WAITOK | M_ZERO); state = 0; } else if (dflen == 0) state = 5; break; default: printf("pci%d:%d:%d:%d: invalid state: %d\n", cfg->domain, cfg->bus, cfg->slot, cfg->func, state); state = -1; break; } } if (cksumvalid == 0 || state < -1) { /* read-only data bad, clean up */ if (cfg->vpd.vpd_ros != NULL) { for (off = 0; cfg->vpd.vpd_ros[off].value; off++) free(cfg->vpd.vpd_ros[off].value, M_DEVBUF); free(cfg->vpd.vpd_ros, M_DEVBUF); cfg->vpd.vpd_ros = NULL; } } if (state < -1) { /* I/O error, clean up */ printf("pci%d:%d:%d:%d: failed to read VPD data.\n", cfg->domain, cfg->bus, cfg->slot, cfg->func); if (cfg->vpd.vpd_ident != NULL) { free(cfg->vpd.vpd_ident, M_DEVBUF); cfg->vpd.vpd_ident = NULL; } if (cfg->vpd.vpd_w != NULL) { for (off = 0; cfg->vpd.vpd_w[off].value; off++) free(cfg->vpd.vpd_w[off].value, M_DEVBUF); free(cfg->vpd.vpd_w, M_DEVBUF); cfg->vpd.vpd_w = NULL; } } cfg->vpd.vpd_cached = 1; #undef REG #undef WREG } int pci_get_vpd_ident_method(device_t dev, device_t child, const char **identptr) { struct pci_devinfo *dinfo = device_get_ivars(child); pcicfgregs *cfg = &dinfo->cfg; if (!cfg->vpd.vpd_cached && cfg->vpd.vpd_reg != 0) pci_read_vpd(device_get_parent(dev), cfg); *identptr = cfg->vpd.vpd_ident; if (*identptr == NULL) return (ENXIO); return (0); } int pci_get_vpd_readonly_method(device_t dev, device_t child, const char *kw, const char **vptr) { struct pci_devinfo *dinfo = device_get_ivars(child); pcicfgregs *cfg = &dinfo->cfg; int i; if (!cfg->vpd.vpd_cached && cfg->vpd.vpd_reg != 0) pci_read_vpd(device_get_parent(dev), cfg); for (i = 0; i < cfg->vpd.vpd_rocnt; i++) if (memcmp(kw, cfg->vpd.vpd_ros[i].keyword, sizeof(cfg->vpd.vpd_ros[i].keyword)) == 0) { *vptr = cfg->vpd.vpd_ros[i].value; } if (i != cfg->vpd.vpd_rocnt) return (0); *vptr = NULL; return (ENXIO); } /* * Find the requested extended capability and return the offset in * configuration space via the pointer provided. The function returns * 0 on success and error code otherwise. */ int pci_find_extcap_method(device_t dev, device_t child, int capability, int *capreg) { struct pci_devinfo *dinfo = device_get_ivars(child); pcicfgregs *cfg = &dinfo->cfg; u_int32_t status; u_int8_t ptr; /* * Check the CAP_LIST bit of the PCI status register first. */ status = pci_read_config(child, PCIR_STATUS, 2); if (!(status & PCIM_STATUS_CAPPRESENT)) return (ENXIO); /* * Determine the start pointer of the capabilities list. */ switch (cfg->hdrtype & PCIM_HDRTYPE) { case PCIM_HDRTYPE_NORMAL: case PCIM_HDRTYPE_BRIDGE: ptr = PCIR_CAP_PTR; break; case PCIM_HDRTYPE_CARDBUS: ptr = PCIR_CAP_PTR_2; break; default: /* XXX: panic? */ return (ENXIO); /* no extended capabilities support */ } ptr = pci_read_config(child, ptr, 1); /* * Traverse the capabilities list. */ while (ptr != 0) { if (pci_read_config(child, ptr + PCICAP_ID, 1) == capability) { if (capreg != NULL) *capreg = ptr; return (0); } ptr = pci_read_config(child, ptr + PCICAP_NEXTPTR, 1); } return (ENOENT); } /* * Support for MSI-X message interrupts. */ void pci_enable_msix(device_t dev, u_int index, uint64_t address, uint32_t data) { struct pci_devinfo *dinfo = device_get_ivars(dev); struct pcicfg_msix *msix = &dinfo->cfg.msix; uint32_t offset; KASSERT(msix->msix_table_len > index, ("bogus index")); offset = msix->msix_table_offset + index * 16; bus_write_4(msix->msix_table_res, offset, address & 0xffffffff); bus_write_4(msix->msix_table_res, offset + 4, address >> 32); bus_write_4(msix->msix_table_res, offset + 8, data); /* Enable MSI -> HT mapping. */ pci_ht_map_msi(dev, address); } void pci_mask_msix(device_t dev, u_int index) { struct pci_devinfo *dinfo = device_get_ivars(dev); struct pcicfg_msix *msix = &dinfo->cfg.msix; uint32_t offset, val; KASSERT(msix->msix_msgnum > index, ("bogus index")); offset = msix->msix_table_offset + index * 16 + 12; val = bus_read_4(msix->msix_table_res, offset); if (!(val & PCIM_MSIX_VCTRL_MASK)) { val |= PCIM_MSIX_VCTRL_MASK; bus_write_4(msix->msix_table_res, offset, val); } } void pci_unmask_msix(device_t dev, u_int index) { struct pci_devinfo *dinfo = device_get_ivars(dev); struct pcicfg_msix *msix = &dinfo->cfg.msix; uint32_t offset, val; KASSERT(msix->msix_table_len > index, ("bogus index")); offset = msix->msix_table_offset + index * 16 + 12; val = bus_read_4(msix->msix_table_res, offset); if (val & PCIM_MSIX_VCTRL_MASK) { val &= ~PCIM_MSIX_VCTRL_MASK; bus_write_4(msix->msix_table_res, offset, val); } } int pci_pending_msix(device_t dev, u_int index) { struct pci_devinfo *dinfo = device_get_ivars(dev); struct pcicfg_msix *msix = &dinfo->cfg.msix; uint32_t offset, bit; KASSERT(msix->msix_table_len > index, ("bogus index")); offset = msix->msix_pba_offset + (index / 32) * 4; bit = 1 << index % 32; return (bus_read_4(msix->msix_pba_res, offset) & bit); } /* * Restore MSI-X registers and table during resume. If MSI-X is * enabled then walk the virtual table to restore the actual MSI-X * table. */ static void pci_resume_msix(device_t dev) { struct pci_devinfo *dinfo = device_get_ivars(dev); struct pcicfg_msix *msix = &dinfo->cfg.msix; struct msix_table_entry *mte; struct msix_vector *mv; int i; if (msix->msix_alloc > 0) { /* First, mask all vectors. */ for (i = 0; i < msix->msix_msgnum; i++) pci_mask_msix(dev, i); /* Second, program any messages with at least one handler. */ for (i = 0; i < msix->msix_table_len; i++) { mte = &msix->msix_table[i]; if (mte->mte_vector == 0 || mte->mte_handlers == 0) continue; mv = &msix->msix_vectors[mte->mte_vector - 1]; pci_enable_msix(dev, i, mv->mv_address, mv->mv_data); pci_unmask_msix(dev, i); } } pci_write_config(dev, msix->msix_location + PCIR_MSIX_CTRL, msix->msix_ctrl, 2); } /* * Attempt to allocate *count MSI-X messages. The actual number allocated is * returned in *count. After this function returns, each message will be * available to the driver as SYS_RES_IRQ resources starting at rid 1. */ int pci_alloc_msix_method(device_t dev, device_t child, int *count) { struct pci_devinfo *dinfo = device_get_ivars(child); pcicfgregs *cfg = &dinfo->cfg; struct resource_list_entry *rle; int actual, error, i, irq, max; /* Don't let count == 0 get us into trouble. */ if (*count == 0) return (EINVAL); /* If rid 0 is allocated, then fail. */ rle = resource_list_find(&dinfo->resources, SYS_RES_IRQ, 0); if (rle != NULL && rle->res != NULL) return (ENXIO); /* Already have allocated messages? */ if (cfg->msi.msi_alloc != 0 || cfg->msix.msix_alloc != 0) return (ENXIO); /* If MSI is blacklisted for this system, fail. */ if (pci_msi_blacklisted()) return (ENXIO); /* MSI-X capability present? */ if (cfg->msix.msix_location == 0 || !pci_do_msix) return (ENODEV); /* Make sure the appropriate BARs are mapped. */ rle = resource_list_find(&dinfo->resources, SYS_RES_MEMORY, cfg->msix.msix_table_bar); if (rle == NULL || rle->res == NULL || !(rman_get_flags(rle->res) & RF_ACTIVE)) return (ENXIO); cfg->msix.msix_table_res = rle->res; if (cfg->msix.msix_pba_bar != cfg->msix.msix_table_bar) { rle = resource_list_find(&dinfo->resources, SYS_RES_MEMORY, cfg->msix.msix_pba_bar); if (rle == NULL || rle->res == NULL || !(rman_get_flags(rle->res) & RF_ACTIVE)) return (ENXIO); } cfg->msix.msix_pba_res = rle->res; if (bootverbose) device_printf(child, "attempting to allocate %d MSI-X vectors (%d supported)\n", *count, cfg->msix.msix_msgnum); max = min(*count, cfg->msix.msix_msgnum); for (i = 0; i < max; i++) { /* Allocate a message. */ error = PCIB_ALLOC_MSIX(device_get_parent(dev), child, &irq); if (error) break; resource_list_add(&dinfo->resources, SYS_RES_IRQ, i + 1, irq, irq, 1); } actual = i; if (bootverbose) { rle = resource_list_find(&dinfo->resources, SYS_RES_IRQ, 1); if (actual == 1) device_printf(child, "using IRQ %lu for MSI-X\n", rle->start); else { int run; /* * Be fancy and try to print contiguous runs of * IRQ values as ranges. 'irq' is the previous IRQ. * 'run' is true if we are in a range. */ device_printf(child, "using IRQs %lu", rle->start); irq = rle->start; run = 0; for (i = 1; i < actual; i++) { rle = resource_list_find(&dinfo->resources, SYS_RES_IRQ, i + 1); /* Still in a run? */ if (rle->start == irq + 1) { run = 1; irq++; continue; } /* Finish previous range. */ if (run) { printf("-%d", irq); run = 0; } /* Start new range. */ printf(",%lu", rle->start); irq = rle->start; } /* Unfinished range? */ if (run) printf("-%d", irq); printf(" for MSI-X\n"); } } /* Mask all vectors. */ for (i = 0; i < cfg->msix.msix_msgnum; i++) pci_mask_msix(child, i); /* Allocate and initialize vector data and virtual table. */ cfg->msix.msix_vectors = malloc(sizeof(struct msix_vector) * actual, M_DEVBUF, M_WAITOK | M_ZERO); cfg->msix.msix_table = malloc(sizeof(struct msix_table_entry) * actual, M_DEVBUF, M_WAITOK | M_ZERO); for (i = 0; i < actual; i++) { rle = resource_list_find(&dinfo->resources, SYS_RES_IRQ, i + 1); cfg->msix.msix_vectors[i].mv_irq = rle->start; cfg->msix.msix_table[i].mte_vector = i + 1; } /* Update control register to enable MSI-X. */ cfg->msix.msix_ctrl |= PCIM_MSIXCTRL_MSIX_ENABLE; pci_write_config(child, cfg->msix.msix_location + PCIR_MSIX_CTRL, cfg->msix.msix_ctrl, 2); /* Update counts of alloc'd messages. */ cfg->msix.msix_alloc = actual; cfg->msix.msix_table_len = actual; *count = actual; return (0); } /* * By default, pci_alloc_msix() will assign the allocated IRQ * resources consecutively to the first N messages in the MSI-X table. * However, device drivers may want to use different layouts if they * either receive fewer messages than they asked for, or they wish to * populate the MSI-X table sparsely. This method allows the driver * to specify what layout it wants. It must be called after a * successful pci_alloc_msix() but before any of the associated * SYS_RES_IRQ resources are allocated via bus_alloc_resource(). * * The 'vectors' array contains 'count' message vectors. The array * maps directly to the MSI-X table in that index 0 in the array * specifies the vector for the first message in the MSI-X table, etc. * The vector value in each array index can either be 0 to indicate * that no vector should be assigned to a message slot, or it can be a * number from 1 to N (where N is the count returned from a * succcessful call to pci_alloc_msix()) to indicate which message * vector (IRQ) to be used for the corresponding message. * * On successful return, each message with a non-zero vector will have * an associated SYS_RES_IRQ whose rid is equal to the array index + * 1. Additionally, if any of the IRQs allocated via the previous * call to pci_alloc_msix() are not used in the mapping, those IRQs * will be freed back to the system automatically. * * For example, suppose a driver has a MSI-X table with 6 messages and * asks for 6 messages, but pci_alloc_msix() only returns a count of * 3. Call the three vectors allocated by pci_alloc_msix() A, B, and * C. After the call to pci_alloc_msix(), the device will be setup to * have an MSI-X table of ABC--- (where - means no vector assigned). * If the driver ten passes a vector array of { 1, 0, 1, 2, 0, 2 }, * then the MSI-X table will look like A-AB-B, and the 'C' vector will * be freed back to the system. This device will also have valid * SYS_RES_IRQ rids of 1, 3, 4, and 6. * * In any case, the SYS_RES_IRQ rid X will always map to the message * at MSI-X table index X - 1 and will only be valid if a vector is * assigned to that table entry. */ int pci_remap_msix_method(device_t dev, device_t child, int count, const u_int *vectors) { struct pci_devinfo *dinfo = device_get_ivars(child); struct pcicfg_msix *msix = &dinfo->cfg.msix; struct resource_list_entry *rle; int i, irq, j, *used; /* * Have to have at least one message in the table but the * table can't be bigger than the actual MSI-X table in the * device. */ if (count == 0 || count > msix->msix_msgnum) return (EINVAL); /* Sanity check the vectors. */ for (i = 0; i < count; i++) if (vectors[i] > msix->msix_alloc) return (EINVAL); /* * Make sure there aren't any holes in the vectors to be used. * It's a big pain to support it, and it doesn't really make * sense anyway. Also, at least one vector must be used. */ used = malloc(sizeof(int) * msix->msix_alloc, M_DEVBUF, M_WAITOK | M_ZERO); for (i = 0; i < count; i++) if (vectors[i] != 0) used[vectors[i] - 1] = 1; for (i = 0; i < msix->msix_alloc - 1; i++) if (used[i] == 0 && used[i + 1] == 1) { free(used, M_DEVBUF); return (EINVAL); } if (used[0] != 1) { free(used, M_DEVBUF); return (EINVAL); } /* Make sure none of the resources are allocated. */ for (i = 0; i < msix->msix_table_len; i++) { if (msix->msix_table[i].mte_vector == 0) continue; if (msix->msix_table[i].mte_handlers > 0) return (EBUSY); rle = resource_list_find(&dinfo->resources, SYS_RES_IRQ, i + 1); KASSERT(rle != NULL, ("missing resource")); if (rle->res != NULL) return (EBUSY); } /* Free the existing resource list entries. */ for (i = 0; i < msix->msix_table_len; i++) { if (msix->msix_table[i].mte_vector == 0) continue; resource_list_delete(&dinfo->resources, SYS_RES_IRQ, i + 1); } /* * Build the new virtual table keeping track of which vectors are * used. */ free(msix->msix_table, M_DEVBUF); msix->msix_table = malloc(sizeof(struct msix_table_entry) * count, M_DEVBUF, M_WAITOK | M_ZERO); for (i = 0; i < count; i++) msix->msix_table[i].mte_vector = vectors[i]; msix->msix_table_len = count; /* Free any unused IRQs and resize the vectors array if necessary. */ j = msix->msix_alloc - 1; if (used[j] == 0) { struct msix_vector *vec; while (used[j] == 0) { PCIB_RELEASE_MSIX(device_get_parent(dev), child, msix->msix_vectors[j].mv_irq); j--; } vec = malloc(sizeof(struct msix_vector) * (j + 1), M_DEVBUF, M_WAITOK); bcopy(msix->msix_vectors, vec, sizeof(struct msix_vector) * (j + 1)); free(msix->msix_vectors, M_DEVBUF); msix->msix_vectors = vec; msix->msix_alloc = j + 1; } free(used, M_DEVBUF); /* Map the IRQs onto the rids. */ for (i = 0; i < count; i++) { if (vectors[i] == 0) continue; irq = msix->msix_vectors[vectors[i]].mv_irq; resource_list_add(&dinfo->resources, SYS_RES_IRQ, i + 1, irq, irq, 1); } if (bootverbose) { device_printf(child, "Remapped MSI-X IRQs as: "); for (i = 0; i < count; i++) { if (i != 0) printf(", "); if (vectors[i] == 0) printf("---"); else printf("%d", msix->msix_vectors[vectors[i]].mv_irq); } printf("\n"); } return (0); } static int pci_release_msix(device_t dev, device_t child) { struct pci_devinfo *dinfo = device_get_ivars(child); struct pcicfg_msix *msix = &dinfo->cfg.msix; struct resource_list_entry *rle; int i; /* Do we have any messages to release? */ if (msix->msix_alloc == 0) return (ENODEV); /* Make sure none of the resources are allocated. */ for (i = 0; i < msix->msix_table_len; i++) { if (msix->msix_table[i].mte_vector == 0) continue; if (msix->msix_table[i].mte_handlers > 0) return (EBUSY); rle = resource_list_find(&dinfo->resources, SYS_RES_IRQ, i + 1); KASSERT(rle != NULL, ("missing resource")); if (rle->res != NULL) return (EBUSY); } /* Update control register to disable MSI-X. */ msix->msix_ctrl &= ~PCIM_MSIXCTRL_MSIX_ENABLE; pci_write_config(child, msix->msix_location + PCIR_MSIX_CTRL, msix->msix_ctrl, 2); /* Free the resource list entries. */ for (i = 0; i < msix->msix_table_len; i++) { if (msix->msix_table[i].mte_vector == 0) continue; resource_list_delete(&dinfo->resources, SYS_RES_IRQ, i + 1); } free(msix->msix_table, M_DEVBUF); msix->msix_table_len = 0; /* Release the IRQs. */ for (i = 0; i < msix->msix_alloc; i++) PCIB_RELEASE_MSIX(device_get_parent(dev), child, msix->msix_vectors[i].mv_irq); free(msix->msix_vectors, M_DEVBUF); msix->msix_alloc = 0; return (0); } /* * Return the max supported MSI-X messages this device supports. * Basically, assuming the MD code can alloc messages, this function * should return the maximum value that pci_alloc_msix() can return. * Thus, it is subject to the tunables, etc. */ int pci_msix_count_method(device_t dev, device_t child) { struct pci_devinfo *dinfo = device_get_ivars(child); struct pcicfg_msix *msix = &dinfo->cfg.msix; if (pci_do_msix && msix->msix_location != 0) return (msix->msix_msgnum); return (0); } /* * HyperTransport MSI mapping control */ void pci_ht_map_msi(device_t dev, uint64_t addr) { struct pci_devinfo *dinfo = device_get_ivars(dev); struct pcicfg_ht *ht = &dinfo->cfg.ht; if (!ht->ht_msimap) return; if (addr && !(ht->ht_msictrl & PCIM_HTCMD_MSI_ENABLE) && ht->ht_msiaddr >> 20 == addr >> 20) { /* Enable MSI -> HT mapping. */ ht->ht_msictrl |= PCIM_HTCMD_MSI_ENABLE; pci_write_config(dev, ht->ht_msimap + PCIR_HT_COMMAND, ht->ht_msictrl, 2); } if (!addr && ht->ht_msictrl & PCIM_HTCMD_MSI_ENABLE) { /* Disable MSI -> HT mapping. */ ht->ht_msictrl &= ~PCIM_HTCMD_MSI_ENABLE; pci_write_config(dev, ht->ht_msimap + PCIR_HT_COMMAND, ht->ht_msictrl, 2); } } int pci_get_max_read_req(device_t dev) { int cap; uint16_t val; if (pci_find_cap(dev, PCIY_EXPRESS, &cap) != 0) return (0); val = pci_read_config(dev, cap + PCIR_EXPRESS_DEVICE_CTL, 2); val &= PCIM_EXP_CTL_MAX_READ_REQUEST; val >>= 12; return (1 << (val + 7)); } int pci_set_max_read_req(device_t dev, int size) { int cap; uint16_t val; if (pci_find_cap(dev, PCIY_EXPRESS, &cap) != 0) return (0); if (size < 128) size = 128; if (size > 4096) size = 4096; size = (1 << (fls(size) - 1)); val = pci_read_config(dev, cap + PCIR_EXPRESS_DEVICE_CTL, 2); val &= ~PCIM_EXP_CTL_MAX_READ_REQUEST; val |= (fls(size) - 8) << 12; pci_write_config(dev, cap + PCIR_EXPRESS_DEVICE_CTL, val, 2); return (size); } /* * Support for MSI message signalled interrupts. */ void pci_enable_msi(device_t dev, uint64_t address, uint16_t data) { struct pci_devinfo *dinfo = device_get_ivars(dev); struct pcicfg_msi *msi = &dinfo->cfg.msi; /* Write data and address values. */ pci_write_config(dev, msi->msi_location + PCIR_MSI_ADDR, address & 0xffffffff, 4); if (msi->msi_ctrl & PCIM_MSICTRL_64BIT) { pci_write_config(dev, msi->msi_location + PCIR_MSI_ADDR_HIGH, address >> 32, 4); pci_write_config(dev, msi->msi_location + PCIR_MSI_DATA_64BIT, data, 2); } else pci_write_config(dev, msi->msi_location + PCIR_MSI_DATA, data, 2); /* Enable MSI in the control register. */ msi->msi_ctrl |= PCIM_MSICTRL_MSI_ENABLE; pci_write_config(dev, msi->msi_location + PCIR_MSI_CTRL, msi->msi_ctrl, 2); /* Enable MSI -> HT mapping. */ pci_ht_map_msi(dev, address); } void pci_disable_msi(device_t dev) { struct pci_devinfo *dinfo = device_get_ivars(dev); struct pcicfg_msi *msi = &dinfo->cfg.msi; /* Disable MSI -> HT mapping. */ pci_ht_map_msi(dev, 0); /* Disable MSI in the control register. */ msi->msi_ctrl &= ~PCIM_MSICTRL_MSI_ENABLE; pci_write_config(dev, msi->msi_location + PCIR_MSI_CTRL, msi->msi_ctrl, 2); } /* * Restore MSI registers during resume. If MSI is enabled then * restore the data and address registers in addition to the control * register. */ static void pci_resume_msi(device_t dev) { struct pci_devinfo *dinfo = device_get_ivars(dev); struct pcicfg_msi *msi = &dinfo->cfg.msi; uint64_t address; uint16_t data; if (msi->msi_ctrl & PCIM_MSICTRL_MSI_ENABLE) { address = msi->msi_addr; data = msi->msi_data; pci_write_config(dev, msi->msi_location + PCIR_MSI_ADDR, address & 0xffffffff, 4); if (msi->msi_ctrl & PCIM_MSICTRL_64BIT) { pci_write_config(dev, msi->msi_location + PCIR_MSI_ADDR_HIGH, address >> 32, 4); pci_write_config(dev, msi->msi_location + PCIR_MSI_DATA_64BIT, data, 2); } else pci_write_config(dev, msi->msi_location + PCIR_MSI_DATA, data, 2); } pci_write_config(dev, msi->msi_location + PCIR_MSI_CTRL, msi->msi_ctrl, 2); } static int pci_remap_intr_method(device_t bus, device_t dev, u_int irq) { struct pci_devinfo *dinfo = device_get_ivars(dev); pcicfgregs *cfg = &dinfo->cfg; struct resource_list_entry *rle; struct msix_table_entry *mte; struct msix_vector *mv; uint64_t addr; uint32_t data; int error, i, j; /* * Handle MSI first. We try to find this IRQ among our list * of MSI IRQs. If we find it, we request updated address and * data registers and apply the results. */ if (cfg->msi.msi_alloc > 0) { /* If we don't have any active handlers, nothing to do. */ if (cfg->msi.msi_handlers == 0) return (0); for (i = 0; i < cfg->msi.msi_alloc; i++) { rle = resource_list_find(&dinfo->resources, SYS_RES_IRQ, i + 1); if (rle->start == irq) { error = PCIB_MAP_MSI(device_get_parent(bus), dev, irq, &addr, &data); if (error) return (error); pci_disable_msi(dev); dinfo->cfg.msi.msi_addr = addr; dinfo->cfg.msi.msi_data = data; pci_enable_msi(dev, addr, data); return (0); } } return (ENOENT); } /* * For MSI-X, we check to see if we have this IRQ. If we do, * we request the updated mapping info. If that works, we go * through all the slots that use this IRQ and update them. */ if (cfg->msix.msix_alloc > 0) { for (i = 0; i < cfg->msix.msix_alloc; i++) { mv = &cfg->msix.msix_vectors[i]; if (mv->mv_irq == irq) { error = PCIB_MAP_MSI(device_get_parent(bus), dev, irq, &addr, &data); if (error) return (error); mv->mv_address = addr; mv->mv_data = data; for (j = 0; j < cfg->msix.msix_table_len; j++) { mte = &cfg->msix.msix_table[j]; if (mte->mte_vector != i + 1) continue; if (mte->mte_handlers == 0) continue; pci_mask_msix(dev, j); pci_enable_msix(dev, j, addr, data); pci_unmask_msix(dev, j); } } } return (ENOENT); } return (ENOENT); } /* * Returns true if the specified device is blacklisted because MSI * doesn't work. */ int pci_msi_device_blacklisted(device_t dev) { struct pci_quirk *q; if (!pci_honor_msi_blacklist) return (0); for (q = &pci_quirks[0]; q->devid; q++) { if (q->devid == pci_get_devid(dev) && q->type == PCI_QUIRK_DISABLE_MSI) return (1); } return (0); } /* * Returns true if a specified chipset supports MSI when it is * emulated hardware in a virtual machine. */ static int pci_msi_vm_chipset(device_t dev) { struct pci_quirk *q; for (q = &pci_quirks[0]; q->devid; q++) { if (q->devid == pci_get_devid(dev) && q->type == PCI_QUIRK_ENABLE_MSI_VM) return (1); } return (0); } /* * Determine if MSI is blacklisted globally on this sytem. Currently, * we just check for blacklisted chipsets as represented by the * host-PCI bridge at device 0:0:0. In the future, it may become * necessary to check other system attributes, such as the kenv values * that give the motherboard manufacturer and model number. */ static int pci_msi_blacklisted(void) { device_t dev; if (!pci_honor_msi_blacklist) return (0); /* Blacklist all non-PCI-express and non-PCI-X chipsets. */ if (!(pcie_chipset || pcix_chipset)) { if (vm_guest != VM_GUEST_NO) { dev = pci_find_bsf(0, 0, 0); if (dev != NULL) return (pci_msi_vm_chipset(dev) == 0); } return (1); } dev = pci_find_bsf(0, 0, 0); if (dev != NULL) return (pci_msi_device_blacklisted(dev)); return (0); } /* * Attempt to allocate *count MSI messages. The actual number allocated is * returned in *count. After this function returns, each message will be * available to the driver as SYS_RES_IRQ resources starting at a rid 1. */ int pci_alloc_msi_method(device_t dev, device_t child, int *count) { struct pci_devinfo *dinfo = device_get_ivars(child); pcicfgregs *cfg = &dinfo->cfg; struct resource_list_entry *rle; int actual, error, i, irqs[32]; uint16_t ctrl; /* Don't let count == 0 get us into trouble. */ if (*count == 0) return (EINVAL); /* If rid 0 is allocated, then fail. */ rle = resource_list_find(&dinfo->resources, SYS_RES_IRQ, 0); if (rle != NULL && rle->res != NULL) return (ENXIO); /* Already have allocated messages? */ if (cfg->msi.msi_alloc != 0 || cfg->msix.msix_alloc != 0) return (ENXIO); /* If MSI is blacklisted for this system, fail. */ if (pci_msi_blacklisted()) return (ENXIO); /* MSI capability present? */ if (cfg->msi.msi_location == 0 || !pci_do_msi) return (ENODEV); if (bootverbose) device_printf(child, "attempting to allocate %d MSI vectors (%d supported)\n", *count, cfg->msi.msi_msgnum); /* Don't ask for more than the device supports. */ actual = min(*count, cfg->msi.msi_msgnum); /* Don't ask for more than 32 messages. */ actual = min(actual, 32); /* MSI requires power of 2 number of messages. */ if (!powerof2(actual)) return (EINVAL); for (;;) { /* Try to allocate N messages. */ error = PCIB_ALLOC_MSI(device_get_parent(dev), child, actual, actual, irqs); if (error == 0) break; if (actual == 1) return (error); /* Try N / 2. */ actual >>= 1; } /* * We now have N actual messages mapped onto SYS_RES_IRQ * resources in the irqs[] array, so add new resources * starting at rid 1. */ for (i = 0; i < actual; i++) resource_list_add(&dinfo->resources, SYS_RES_IRQ, i + 1, irqs[i], irqs[i], 1); if (bootverbose) { if (actual == 1) device_printf(child, "using IRQ %d for MSI\n", irqs[0]); else { int run; /* * Be fancy and try to print contiguous runs * of IRQ values as ranges. 'run' is true if * we are in a range. */ device_printf(child, "using IRQs %d", irqs[0]); run = 0; for (i = 1; i < actual; i++) { /* Still in a run? */ if (irqs[i] == irqs[i - 1] + 1) { run = 1; continue; } /* Finish previous range. */ if (run) { printf("-%d", irqs[i - 1]); run = 0; } /* Start new range. */ printf(",%d", irqs[i]); } /* Unfinished range? */ if (run) printf("-%d", irqs[actual - 1]); printf(" for MSI\n"); } } /* Update control register with actual count. */ ctrl = cfg->msi.msi_ctrl; ctrl &= ~PCIM_MSICTRL_MME_MASK; ctrl |= (ffs(actual) - 1) << 4; cfg->msi.msi_ctrl = ctrl; pci_write_config(child, cfg->msi.msi_location + PCIR_MSI_CTRL, ctrl, 2); /* Update counts of alloc'd messages. */ cfg->msi.msi_alloc = actual; cfg->msi.msi_handlers = 0; *count = actual; return (0); } /* Release the MSI messages associated with this device. */ int pci_release_msi_method(device_t dev, device_t child) { struct pci_devinfo *dinfo = device_get_ivars(child); struct pcicfg_msi *msi = &dinfo->cfg.msi; struct resource_list_entry *rle; int error, i, irqs[32]; /* Try MSI-X first. */ error = pci_release_msix(dev, child); if (error != ENODEV) return (error); /* Do we have any messages to release? */ if (msi->msi_alloc == 0) return (ENODEV); KASSERT(msi->msi_alloc <= 32, ("more than 32 alloc'd messages")); /* Make sure none of the resources are allocated. */ if (msi->msi_handlers > 0) return (EBUSY); for (i = 0; i < msi->msi_alloc; i++) { rle = resource_list_find(&dinfo->resources, SYS_RES_IRQ, i + 1); KASSERT(rle != NULL, ("missing MSI resource")); if (rle->res != NULL) return (EBUSY); irqs[i] = rle->start; } /* Update control register with 0 count. */ KASSERT(!(msi->msi_ctrl & PCIM_MSICTRL_MSI_ENABLE), ("%s: MSI still enabled", __func__)); msi->msi_ctrl &= ~PCIM_MSICTRL_MME_MASK; pci_write_config(child, msi->msi_location + PCIR_MSI_CTRL, msi->msi_ctrl, 2); /* Release the messages. */ PCIB_RELEASE_MSI(device_get_parent(dev), child, msi->msi_alloc, irqs); for (i = 0; i < msi->msi_alloc; i++) resource_list_delete(&dinfo->resources, SYS_RES_IRQ, i + 1); /* Update alloc count. */ msi->msi_alloc = 0; msi->msi_addr = 0; msi->msi_data = 0; return (0); } /* * Return the max supported MSI messages this device supports. * Basically, assuming the MD code can alloc messages, this function * should return the maximum value that pci_alloc_msi() can return. * Thus, it is subject to the tunables, etc. */ int pci_msi_count_method(device_t dev, device_t child) { struct pci_devinfo *dinfo = device_get_ivars(child); struct pcicfg_msi *msi = &dinfo->cfg.msi; if (pci_do_msi && msi->msi_location != 0) return (msi->msi_msgnum); return (0); } /* free pcicfgregs structure and all depending data structures */ int pci_freecfg(struct pci_devinfo *dinfo) { struct devlist *devlist_head; struct pci_map *pm, *next; int i; devlist_head = &pci_devq; if (dinfo->cfg.vpd.vpd_reg) { free(dinfo->cfg.vpd.vpd_ident, M_DEVBUF); for (i = 0; i < dinfo->cfg.vpd.vpd_rocnt; i++) free(dinfo->cfg.vpd.vpd_ros[i].value, M_DEVBUF); free(dinfo->cfg.vpd.vpd_ros, M_DEVBUF); for (i = 0; i < dinfo->cfg.vpd.vpd_wcnt; i++) free(dinfo->cfg.vpd.vpd_w[i].value, M_DEVBUF); free(dinfo->cfg.vpd.vpd_w, M_DEVBUF); } STAILQ_FOREACH_SAFE(pm, &dinfo->cfg.maps, pm_link, next) { free(pm, M_DEVBUF); } STAILQ_REMOVE(devlist_head, dinfo, pci_devinfo, pci_links); free(dinfo, M_DEVBUF); /* increment the generation count */ pci_generation++; /* we're losing one device */ pci_numdevs--; return (0); } /* * PCI power manangement */ int pci_set_powerstate_method(device_t dev, device_t child, int state) { struct pci_devinfo *dinfo = device_get_ivars(child); pcicfgregs *cfg = &dinfo->cfg; uint16_t status; int result, oldstate, highest, delay; if (cfg->pp.pp_cap == 0) return (EOPNOTSUPP); /* * Optimize a no state change request away. While it would be OK to * write to the hardware in theory, some devices have shown odd * behavior when going from D3 -> D3. */ oldstate = pci_get_powerstate(child); if (oldstate == state) return (0); /* * The PCI power management specification states that after a state * transition between PCI power states, system software must * guarantee a minimal delay before the function accesses the device. * Compute the worst case delay that we need to guarantee before we * access the device. Many devices will be responsive much more * quickly than this delay, but there are some that don't respond * instantly to state changes. Transitions to/from D3 state require * 10ms, while D2 requires 200us, and D0/1 require none. The delay * is done below with DELAY rather than a sleeper function because * this function can be called from contexts where we cannot sleep. */ highest = (oldstate > state) ? oldstate : state; if (highest == PCI_POWERSTATE_D3) delay = 10000; else if (highest == PCI_POWERSTATE_D2) delay = 200; else delay = 0; status = PCI_READ_CONFIG(dev, child, cfg->pp.pp_status, 2) & ~PCIM_PSTAT_DMASK; result = 0; switch (state) { case PCI_POWERSTATE_D0: status |= PCIM_PSTAT_D0; break; case PCI_POWERSTATE_D1: if ((cfg->pp.pp_cap & PCIM_PCAP_D1SUPP) == 0) return (EOPNOTSUPP); status |= PCIM_PSTAT_D1; break; case PCI_POWERSTATE_D2: if ((cfg->pp.pp_cap & PCIM_PCAP_D2SUPP) == 0) return (EOPNOTSUPP); status |= PCIM_PSTAT_D2; break; case PCI_POWERSTATE_D3: status |= PCIM_PSTAT_D3; break; default: return (EINVAL); } if (bootverbose) pci_printf(cfg, "Transition from D%d to D%d\n", oldstate, state); PCI_WRITE_CONFIG(dev, child, cfg->pp.pp_status, status, 2); if (delay) DELAY(delay); return (0); } int pci_get_powerstate_method(device_t dev, device_t child) { struct pci_devinfo *dinfo = device_get_ivars(child); pcicfgregs *cfg = &dinfo->cfg; uint16_t status; int result; if (cfg->pp.pp_cap != 0) { status = PCI_READ_CONFIG(dev, child, cfg->pp.pp_status, 2); switch (status & PCIM_PSTAT_DMASK) { case PCIM_PSTAT_D0: result = PCI_POWERSTATE_D0; break; case PCIM_PSTAT_D1: result = PCI_POWERSTATE_D1; break; case PCIM_PSTAT_D2: result = PCI_POWERSTATE_D2; break; case PCIM_PSTAT_D3: result = PCI_POWERSTATE_D3; break; default: result = PCI_POWERSTATE_UNKNOWN; break; } } else { /* No support, device is always at D0 */ result = PCI_POWERSTATE_D0; } return (result); } /* * Some convenience functions for PCI device drivers. */ static __inline void pci_set_command_bit(device_t dev, device_t child, uint16_t bit) { uint16_t command; command = PCI_READ_CONFIG(dev, child, PCIR_COMMAND, 2); command |= bit; PCI_WRITE_CONFIG(dev, child, PCIR_COMMAND, command, 2); } static __inline void pci_clear_command_bit(device_t dev, device_t child, uint16_t bit) { uint16_t command; command = PCI_READ_CONFIG(dev, child, PCIR_COMMAND, 2); command &= ~bit; PCI_WRITE_CONFIG(dev, child, PCIR_COMMAND, command, 2); } int pci_enable_busmaster_method(device_t dev, device_t child) { pci_set_command_bit(dev, child, PCIM_CMD_BUSMASTEREN); return (0); } int pci_disable_busmaster_method(device_t dev, device_t child) { pci_clear_command_bit(dev, child, PCIM_CMD_BUSMASTEREN); return (0); } int pci_enable_io_method(device_t dev, device_t child, int space) { uint16_t bit; switch(space) { case SYS_RES_IOPORT: bit = PCIM_CMD_PORTEN; break; case SYS_RES_MEMORY: bit = PCIM_CMD_MEMEN; break; default: return (EINVAL); } pci_set_command_bit(dev, child, bit); return (0); } int pci_disable_io_method(device_t dev, device_t child, int space) { uint16_t bit; switch(space) { case SYS_RES_IOPORT: bit = PCIM_CMD_PORTEN; break; case SYS_RES_MEMORY: bit = PCIM_CMD_MEMEN; break; default: return (EINVAL); } pci_clear_command_bit(dev, child, bit); return (0); } /* * New style pci driver. Parent device is either a pci-host-bridge or a * pci-pci-bridge. Both kinds are represented by instances of pcib. */ void pci_print_verbose(struct pci_devinfo *dinfo) { if (bootverbose) { pcicfgregs *cfg = &dinfo->cfg; printf("found->\tvendor=0x%04x, dev=0x%04x, revid=0x%02x\n", cfg->vendor, cfg->device, cfg->revid); printf("\tdomain=%d, bus=%d, slot=%d, func=%d\n", cfg->domain, cfg->bus, cfg->slot, cfg->func); printf("\tclass=%02x-%02x-%02x, hdrtype=0x%02x, mfdev=%d\n", cfg->baseclass, cfg->subclass, cfg->progif, cfg->hdrtype, cfg->mfdev); printf("\tcmdreg=0x%04x, statreg=0x%04x, cachelnsz=%d (dwords)\n", cfg->cmdreg, cfg->statreg, cfg->cachelnsz); printf("\tlattimer=0x%02x (%d ns), mingnt=0x%02x (%d ns), maxlat=0x%02x (%d ns)\n", cfg->lattimer, cfg->lattimer * 30, cfg->mingnt, cfg->mingnt * 250, cfg->maxlat, cfg->maxlat * 250); if (cfg->intpin > 0) printf("\tintpin=%c, irq=%d\n", cfg->intpin +'a' -1, cfg->intline); if (cfg->pp.pp_cap) { uint16_t status; status = pci_read_config(cfg->dev, cfg->pp.pp_status, 2); printf("\tpowerspec %d supports D0%s%s D3 current D%d\n", cfg->pp.pp_cap & PCIM_PCAP_SPEC, cfg->pp.pp_cap & PCIM_PCAP_D1SUPP ? " D1" : "", cfg->pp.pp_cap & PCIM_PCAP_D2SUPP ? " D2" : "", status & PCIM_PSTAT_DMASK); } if (cfg->msi.msi_location) { int ctrl; ctrl = cfg->msi.msi_ctrl; printf("\tMSI supports %d message%s%s%s\n", cfg->msi.msi_msgnum, (cfg->msi.msi_msgnum == 1) ? "" : "s", (ctrl & PCIM_MSICTRL_64BIT) ? ", 64 bit" : "", (ctrl & PCIM_MSICTRL_VECTOR) ? ", vector masks":""); } if (cfg->msix.msix_location) { printf("\tMSI-X supports %d message%s ", cfg->msix.msix_msgnum, (cfg->msix.msix_msgnum == 1) ? "" : "s"); if (cfg->msix.msix_table_bar == cfg->msix.msix_pba_bar) printf("in map 0x%x\n", cfg->msix.msix_table_bar); else printf("in maps 0x%x and 0x%x\n", cfg->msix.msix_table_bar, cfg->msix.msix_pba_bar); } } } static int pci_porten(device_t dev) { return (pci_read_config(dev, PCIR_COMMAND, 2) & PCIM_CMD_PORTEN) != 0; } static int pci_memen(device_t dev) { return (pci_read_config(dev, PCIR_COMMAND, 2) & PCIM_CMD_MEMEN) != 0; } static void pci_read_bar(device_t dev, int reg, pci_addr_t *mapp, pci_addr_t *testvalp) { struct pci_devinfo *dinfo; pci_addr_t map, testval; int ln2range; uint16_t cmd; /* * The device ROM BAR is special. It is always a 32-bit * memory BAR. Bit 0 is special and should not be set when * sizing the BAR. */ dinfo = device_get_ivars(dev); if (PCIR_IS_BIOS(&dinfo->cfg, reg)) { map = pci_read_config(dev, reg, 4); pci_write_config(dev, reg, 0xfffffffe, 4); testval = pci_read_config(dev, reg, 4); pci_write_config(dev, reg, map, 4); *mapp = map; *testvalp = testval; return; } map = pci_read_config(dev, reg, 4); ln2range = pci_maprange(map); if (ln2range == 64) map |= (pci_addr_t)pci_read_config(dev, reg + 4, 4) << 32; /* * Disable decoding via the command register before * determining the BAR's length since we will be placing it in * a weird state. */ cmd = pci_read_config(dev, PCIR_COMMAND, 2); pci_write_config(dev, PCIR_COMMAND, cmd & ~(PCI_BAR_MEM(map) ? PCIM_CMD_MEMEN : PCIM_CMD_PORTEN), 2); /* * Determine the BAR's length by writing all 1's. The bottom * log_2(size) bits of the BAR will stick as 0 when we read * the value back. */ pci_write_config(dev, reg, 0xffffffff, 4); testval = pci_read_config(dev, reg, 4); if (ln2range == 64) { pci_write_config(dev, reg + 4, 0xffffffff, 4); testval |= (pci_addr_t)pci_read_config(dev, reg + 4, 4) << 32; } /* * Restore the original value of the BAR. We may have reprogrammed * the BAR of the low-level console device and when booting verbose, * we need the console device addressable. */ pci_write_config(dev, reg, map, 4); if (ln2range == 64) pci_write_config(dev, reg + 4, map >> 32, 4); pci_write_config(dev, PCIR_COMMAND, cmd, 2); *mapp = map; *testvalp = testval; } static void pci_write_bar(device_t dev, struct pci_map *pm, pci_addr_t base) { struct pci_devinfo *dinfo; int ln2range; /* The device ROM BAR is always a 32-bit memory BAR. */ dinfo = device_get_ivars(dev); if (PCIR_IS_BIOS(&dinfo->cfg, pm->pm_reg)) ln2range = 32; else ln2range = pci_maprange(pm->pm_value); pci_write_config(dev, pm->pm_reg, base, 4); if (ln2range == 64) pci_write_config(dev, pm->pm_reg + 4, base >> 32, 4); pm->pm_value = pci_read_config(dev, pm->pm_reg, 4); if (ln2range == 64) pm->pm_value |= (pci_addr_t)pci_read_config(dev, pm->pm_reg + 4, 4) << 32; } struct pci_map * pci_find_bar(device_t dev, int reg) { struct pci_devinfo *dinfo; struct pci_map *pm; dinfo = device_get_ivars(dev); STAILQ_FOREACH(pm, &dinfo->cfg.maps, pm_link) { if (pm->pm_reg == reg) return (pm); } return (NULL); } int pci_bar_enabled(device_t dev, struct pci_map *pm) { struct pci_devinfo *dinfo; uint16_t cmd; dinfo = device_get_ivars(dev); if (PCIR_IS_BIOS(&dinfo->cfg, pm->pm_reg) && !(pm->pm_value & PCIM_BIOS_ENABLE)) return (0); cmd = pci_read_config(dev, PCIR_COMMAND, 2); if (PCIR_IS_BIOS(&dinfo->cfg, pm->pm_reg) || PCI_BAR_MEM(pm->pm_value)) return ((cmd & PCIM_CMD_MEMEN) != 0); else return ((cmd & PCIM_CMD_PORTEN) != 0); } static struct pci_map * pci_add_bar(device_t dev, int reg, pci_addr_t value, pci_addr_t size) { struct pci_devinfo *dinfo; struct pci_map *pm, *prev; dinfo = device_get_ivars(dev); pm = malloc(sizeof(*pm), M_DEVBUF, M_WAITOK | M_ZERO); pm->pm_reg = reg; pm->pm_value = value; pm->pm_size = size; STAILQ_FOREACH(prev, &dinfo->cfg.maps, pm_link) { KASSERT(prev->pm_reg != pm->pm_reg, ("duplicate map %02x", reg)); if (STAILQ_NEXT(prev, pm_link) == NULL || STAILQ_NEXT(prev, pm_link)->pm_reg > pm->pm_reg) break; } if (prev != NULL) STAILQ_INSERT_AFTER(&dinfo->cfg.maps, prev, pm, pm_link); else STAILQ_INSERT_TAIL(&dinfo->cfg.maps, pm, pm_link); return (pm); } static void pci_restore_bars(device_t dev) { struct pci_devinfo *dinfo; struct pci_map *pm; int ln2range; dinfo = device_get_ivars(dev); STAILQ_FOREACH(pm, &dinfo->cfg.maps, pm_link) { if (PCIR_IS_BIOS(&dinfo->cfg, pm->pm_reg)) ln2range = 32; else ln2range = pci_maprange(pm->pm_value); pci_write_config(dev, pm->pm_reg, pm->pm_value, 4); if (ln2range == 64) pci_write_config(dev, pm->pm_reg + 4, pm->pm_value >> 32, 4); } } /* * Add a resource based on a pci map register. Return 1 if the map * register is a 32bit map register or 2 if it is a 64bit register. */ static int pci_add_map(device_t bus, device_t dev, int reg, struct resource_list *rl, int force, int prefetch) { struct pci_map *pm; pci_addr_t base, map, testval; pci_addr_t start, end, count; int barlen, basezero, maprange, mapsize, type; uint16_t cmd; struct resource *res; + /* + * The BAR may already exist if the device is a CardBus card + * whose CIS is stored in this BAR. + */ + pm = pci_find_bar(dev, reg); + if (pm != NULL) { + maprange = pci_maprange(pm->pm_value); + barlen = maprange == 64 ? 2 : 1; + return (barlen); + } + pci_read_bar(dev, reg, &map, &testval); if (PCI_BAR_MEM(map)) { type = SYS_RES_MEMORY; if (map & PCIM_BAR_MEM_PREFETCH) prefetch = 1; } else type = SYS_RES_IOPORT; mapsize = pci_mapsize(testval); base = pci_mapbase(map); #ifdef __PCI_BAR_ZERO_VALID basezero = 0; #else basezero = base == 0; #endif maprange = pci_maprange(map); barlen = maprange == 64 ? 2 : 1; /* * For I/O registers, if bottom bit is set, and the next bit up * isn't clear, we know we have a BAR that doesn't conform to the * spec, so ignore it. Also, sanity check the size of the data * areas to the type of memory involved. Memory must be at least * 16 bytes in size, while I/O ranges must be at least 4. */ if (PCI_BAR_IO(testval) && (testval & PCIM_BAR_IO_RESERVED) != 0) return (barlen); if ((type == SYS_RES_MEMORY && mapsize < 4) || (type == SYS_RES_IOPORT && mapsize < 2)) return (barlen); /* Save a record of this BAR. */ pm = pci_add_bar(dev, reg, map, mapsize); if (bootverbose) { printf("\tmap[%02x]: type %s, range %2d, base %#jx, size %2d", reg, pci_maptype(map), maprange, (uintmax_t)base, mapsize); if (type == SYS_RES_IOPORT && !pci_porten(dev)) printf(", port disabled\n"); else if (type == SYS_RES_MEMORY && !pci_memen(dev)) printf(", memory disabled\n"); else printf(", enabled\n"); } /* * If base is 0, then we have problems if this architecture does * not allow that. It is best to ignore such entries for the * moment. These will be allocated later if the driver specifically * requests them. However, some removable busses look better when * all resources are allocated, so allow '0' to be overriden. * * Similarly treat maps whose values is the same as the test value * read back. These maps have had all f's written to them by the * BIOS in an attempt to disable the resources. */ if (!force && (basezero || map == testval)) return (barlen); if ((u_long)base != base) { device_printf(bus, "pci%d:%d:%d:%d bar %#x too many address bits", pci_get_domain(dev), pci_get_bus(dev), pci_get_slot(dev), pci_get_function(dev), reg); return (barlen); } /* * This code theoretically does the right thing, but has * undesirable side effects in some cases where peripherals * respond oddly to having these bits enabled. Let the user * be able to turn them off (since pci_enable_io_modes is 1 by * default). */ if (pci_enable_io_modes) { /* Turn on resources that have been left off by a lazy BIOS */ if (type == SYS_RES_IOPORT && !pci_porten(dev)) { cmd = pci_read_config(dev, PCIR_COMMAND, 2); cmd |= PCIM_CMD_PORTEN; pci_write_config(dev, PCIR_COMMAND, cmd, 2); } if (type == SYS_RES_MEMORY && !pci_memen(dev)) { cmd = pci_read_config(dev, PCIR_COMMAND, 2); cmd |= PCIM_CMD_MEMEN; pci_write_config(dev, PCIR_COMMAND, cmd, 2); } } else { if (type == SYS_RES_IOPORT && !pci_porten(dev)) return (barlen); if (type == SYS_RES_MEMORY && !pci_memen(dev)) return (barlen); } count = (pci_addr_t)1 << mapsize; if (basezero || base == pci_mapbase(testval)) { start = 0; /* Let the parent decide. */ end = ~0ULL; } else { start = base; end = base + count - 1; } resource_list_add(rl, type, reg, start, end, count); /* * Try to allocate the resource for this BAR from our parent * so that this resource range is already reserved. The * driver for this device will later inherit this resource in * pci_alloc_resource(). */ res = resource_list_reserve(rl, bus, dev, type, ®, start, end, count, prefetch ? RF_PREFETCHABLE : 0); if (res == NULL) { /* * If the allocation fails, clear the BAR and delete * the resource list entry to force * pci_alloc_resource() to allocate resources from the * parent. */ resource_list_delete(rl, type, reg); start = 0; } else start = rman_get_start(res); pci_write_bar(dev, pm, start); return (barlen); } /* * For ATA devices we need to decide early what addressing mode to use. * Legacy demands that the primary and secondary ATA ports sits on the * same addresses that old ISA hardware did. This dictates that we use * those addresses and ignore the BAR's if we cannot set PCI native * addressing mode. */ static void pci_ata_maps(device_t bus, device_t dev, struct resource_list *rl, int force, uint32_t prefetchmask) { struct resource *r; int rid, type, progif; #if 0 /* if this device supports PCI native addressing use it */ progif = pci_read_config(dev, PCIR_PROGIF, 1); if ((progif & 0x8a) == 0x8a) { if (pci_mapbase(pci_read_config(dev, PCIR_BAR(0), 4)) && pci_mapbase(pci_read_config(dev, PCIR_BAR(2), 4))) { printf("Trying ATA native PCI addressing mode\n"); pci_write_config(dev, PCIR_PROGIF, progif | 0x05, 1); } } #endif progif = pci_read_config(dev, PCIR_PROGIF, 1); type = SYS_RES_IOPORT; if (progif & PCIP_STORAGE_IDE_MODEPRIM) { pci_add_map(bus, dev, PCIR_BAR(0), rl, force, prefetchmask & (1 << 0)); pci_add_map(bus, dev, PCIR_BAR(1), rl, force, prefetchmask & (1 << 1)); } else { rid = PCIR_BAR(0); resource_list_add(rl, type, rid, 0x1f0, 0x1f7, 8); r = resource_list_reserve(rl, bus, dev, type, &rid, 0x1f0, 0x1f7, 8, 0); rid = PCIR_BAR(1); resource_list_add(rl, type, rid, 0x3f6, 0x3f6, 1); r = resource_list_reserve(rl, bus, dev, type, &rid, 0x3f6, 0x3f6, 1, 0); } if (progif & PCIP_STORAGE_IDE_MODESEC) { pci_add_map(bus, dev, PCIR_BAR(2), rl, force, prefetchmask & (1 << 2)); pci_add_map(bus, dev, PCIR_BAR(3), rl, force, prefetchmask & (1 << 3)); } else { rid = PCIR_BAR(2); resource_list_add(rl, type, rid, 0x170, 0x177, 8); r = resource_list_reserve(rl, bus, dev, type, &rid, 0x170, 0x177, 8, 0); rid = PCIR_BAR(3); resource_list_add(rl, type, rid, 0x376, 0x376, 1); r = resource_list_reserve(rl, bus, dev, type, &rid, 0x376, 0x376, 1, 0); } pci_add_map(bus, dev, PCIR_BAR(4), rl, force, prefetchmask & (1 << 4)); pci_add_map(bus, dev, PCIR_BAR(5), rl, force, prefetchmask & (1 << 5)); } static void pci_assign_interrupt(device_t bus, device_t dev, int force_route) { struct pci_devinfo *dinfo = device_get_ivars(dev); pcicfgregs *cfg = &dinfo->cfg; char tunable_name[64]; int irq; /* Has to have an intpin to have an interrupt. */ if (cfg->intpin == 0) return; /* Let the user override the IRQ with a tunable. */ irq = PCI_INVALID_IRQ; snprintf(tunable_name, sizeof(tunable_name), "hw.pci%d.%d.%d.INT%c.irq", cfg->domain, cfg->bus, cfg->slot, cfg->intpin + 'A' - 1); if (TUNABLE_INT_FETCH(tunable_name, &irq) && (irq >= 255 || irq <= 0)) irq = PCI_INVALID_IRQ; /* * If we didn't get an IRQ via the tunable, then we either use the * IRQ value in the intline register or we ask the bus to route an * interrupt for us. If force_route is true, then we only use the * value in the intline register if the bus was unable to assign an * IRQ. */ if (!PCI_INTERRUPT_VALID(irq)) { if (!PCI_INTERRUPT_VALID(cfg->intline) || force_route) irq = PCI_ASSIGN_INTERRUPT(bus, dev); if (!PCI_INTERRUPT_VALID(irq)) irq = cfg->intline; } /* If after all that we don't have an IRQ, just bail. */ if (!PCI_INTERRUPT_VALID(irq)) return; /* Update the config register if it changed. */ if (irq != cfg->intline) { cfg->intline = irq; pci_write_config(dev, PCIR_INTLINE, irq, 1); } /* Add this IRQ as rid 0 interrupt resource. */ resource_list_add(&dinfo->resources, SYS_RES_IRQ, 0, irq, irq, 1); } /* Perform early OHCI takeover from SMM. */ static void ohci_early_takeover(device_t self) { struct resource *res; uint32_t ctl; int rid; int i; rid = PCIR_BAR(0); res = bus_alloc_resource_any(self, SYS_RES_MEMORY, &rid, RF_ACTIVE); if (res == NULL) return; ctl = bus_read_4(res, OHCI_CONTROL); if (ctl & OHCI_IR) { if (bootverbose) printf("ohci early: " "SMM active, request owner change\n"); bus_write_4(res, OHCI_COMMAND_STATUS, OHCI_OCR); for (i = 0; (i < 100) && (ctl & OHCI_IR); i++) { DELAY(1000); ctl = bus_read_4(res, OHCI_CONTROL); } if (ctl & OHCI_IR) { if (bootverbose) printf("ohci early: " "SMM does not respond, resetting\n"); bus_write_4(res, OHCI_CONTROL, OHCI_HCFS_RESET); } /* Disable interrupts */ bus_write_4(res, OHCI_INTERRUPT_DISABLE, OHCI_ALL_INTRS); } bus_release_resource(self, SYS_RES_MEMORY, rid, res); } /* Perform early UHCI takeover from SMM. */ static void uhci_early_takeover(device_t self) { struct resource *res; int rid; /* * Set the PIRQD enable bit and switch off all the others. We don't * want legacy support to interfere with us XXX Does this also mean * that the BIOS won't touch the keyboard anymore if it is connected * to the ports of the root hub? */ pci_write_config(self, PCI_LEGSUP, PCI_LEGSUP_USBPIRQDEN, 2); /* Disable interrupts */ rid = PCI_UHCI_BASE_REG; res = bus_alloc_resource_any(self, SYS_RES_IOPORT, &rid, RF_ACTIVE); if (res != NULL) { bus_write_2(res, UHCI_INTR, 0); bus_release_resource(self, SYS_RES_IOPORT, rid, res); } } /* Perform early EHCI takeover from SMM. */ static void ehci_early_takeover(device_t self) { struct resource *res; uint32_t cparams; uint32_t eec; uint8_t eecp; uint8_t bios_sem; uint8_t offs; int rid; int i; rid = PCIR_BAR(0); res = bus_alloc_resource_any(self, SYS_RES_MEMORY, &rid, RF_ACTIVE); if (res == NULL) return; cparams = bus_read_4(res, EHCI_HCCPARAMS); /* Synchronise with the BIOS if it owns the controller. */ for (eecp = EHCI_HCC_EECP(cparams); eecp != 0; eecp = EHCI_EECP_NEXT(eec)) { eec = pci_read_config(self, eecp, 4); if (EHCI_EECP_ID(eec) != EHCI_EC_LEGSUP) { continue; } bios_sem = pci_read_config(self, eecp + EHCI_LEGSUP_BIOS_SEM, 1); if (bios_sem == 0) { continue; } if (bootverbose) printf("ehci early: " "SMM active, request owner change\n"); pci_write_config(self, eecp + EHCI_LEGSUP_OS_SEM, 1, 1); for (i = 0; (i < 100) && (bios_sem != 0); i++) { DELAY(1000); bios_sem = pci_read_config(self, eecp + EHCI_LEGSUP_BIOS_SEM, 1); } if (bios_sem != 0) { if (bootverbose) printf("ehci early: " "SMM does not respond\n"); } /* Disable interrupts */ offs = EHCI_CAPLENGTH(bus_read_4(res, EHCI_CAPLEN_HCIVERSION)); bus_write_4(res, offs + EHCI_USBINTR, 0); } bus_release_resource(self, SYS_RES_MEMORY, rid, res); } void pci_add_resources(device_t bus, device_t dev, int force, uint32_t prefetchmask) { struct pci_devinfo *dinfo = device_get_ivars(dev); pcicfgregs *cfg = &dinfo->cfg; struct resource_list *rl = &dinfo->resources; struct pci_quirk *q; int i; /* ATA devices needs special map treatment */ if ((pci_get_class(dev) == PCIC_STORAGE) && (pci_get_subclass(dev) == PCIS_STORAGE_IDE) && ((pci_get_progif(dev) & PCIP_STORAGE_IDE_MASTERDEV) || (!pci_read_config(dev, PCIR_BAR(0), 4) && !pci_read_config(dev, PCIR_BAR(2), 4))) ) pci_ata_maps(bus, dev, rl, force, prefetchmask); else for (i = 0; i < cfg->nummaps;) i += pci_add_map(bus, dev, PCIR_BAR(i), rl, force, prefetchmask & (1 << i)); /* * Add additional, quirked resources. */ for (q = &pci_quirks[0]; q->devid; q++) { if (q->devid == ((cfg->device << 16) | cfg->vendor) && q->type == PCI_QUIRK_MAP_REG) pci_add_map(bus, dev, q->arg1, rl, force, 0); } if (cfg->intpin > 0 && PCI_INTERRUPT_VALID(cfg->intline)) { #ifdef __PCI_REROUTE_INTERRUPT /* * Try to re-route interrupts. Sometimes the BIOS or * firmware may leave bogus values in these registers. * If the re-route fails, then just stick with what we * have. */ pci_assign_interrupt(bus, dev, 1); #else pci_assign_interrupt(bus, dev, 0); #endif } if (pci_usb_takeover && pci_get_class(dev) == PCIC_SERIALBUS && pci_get_subclass(dev) == PCIS_SERIALBUS_USB) { if (pci_get_progif(dev) == PCIP_SERIALBUS_USB_EHCI) ehci_early_takeover(dev); else if (pci_get_progif(dev) == PCIP_SERIALBUS_USB_OHCI) ohci_early_takeover(dev); else if (pci_get_progif(dev) == PCIP_SERIALBUS_USB_UHCI) uhci_early_takeover(dev); } } void pci_add_children(device_t dev, int domain, int busno, size_t dinfo_size) { #define REG(n, w) PCIB_READ_CONFIG(pcib, busno, s, f, n, w) device_t pcib = device_get_parent(dev); struct pci_devinfo *dinfo; int maxslots; int s, f, pcifunchigh; uint8_t hdrtype; KASSERT(dinfo_size >= sizeof(struct pci_devinfo), ("dinfo_size too small")); maxslots = PCIB_MAXSLOTS(pcib); for (s = 0; s <= maxslots; s++) { pcifunchigh = 0; f = 0; DELAY(1); hdrtype = REG(PCIR_HDRTYPE, 1); if ((hdrtype & PCIM_HDRTYPE) > PCI_MAXHDRTYPE) continue; if (hdrtype & PCIM_MFDEV) pcifunchigh = PCI_FUNCMAX; for (f = 0; f <= pcifunchigh; f++) { dinfo = pci_read_device(pcib, domain, busno, s, f, dinfo_size); if (dinfo != NULL) { pci_add_child(dev, dinfo); } } } #undef REG } void pci_add_child(device_t bus, struct pci_devinfo *dinfo) { dinfo->cfg.dev = device_add_child(bus, NULL, -1); device_set_ivars(dinfo->cfg.dev, dinfo); resource_list_init(&dinfo->resources); pci_cfg_save(dinfo->cfg.dev, dinfo, 0); pci_cfg_restore(dinfo->cfg.dev, dinfo); pci_print_verbose(dinfo); pci_add_resources(bus, dinfo->cfg.dev, 0, 0); } static int pci_probe(device_t dev) { device_set_desc(dev, "PCI bus"); /* Allow other subclasses to override this driver. */ return (BUS_PROBE_GENERIC); } static int pci_attach(device_t dev) { int busno, domain; /* * Since there can be multiple independantly numbered PCI * busses on systems with multiple PCI domains, we can't use * the unit number to decide which bus we are probing. We ask * the parent pcib what our domain and bus numbers are. */ domain = pcib_get_domain(dev); busno = pcib_get_bus(dev); if (bootverbose) device_printf(dev, "domain=%d, physical bus=%d\n", domain, busno); pci_add_children(dev, domain, busno, sizeof(struct pci_devinfo)); return (bus_generic_attach(dev)); } static void pci_set_power_children(device_t dev, device_t *devlist, int numdevs, int state) { device_t child, pcib; struct pci_devinfo *dinfo; int dstate, i; /* * Set the device to the given state. If the firmware suggests * a different power state, use it instead. If power management * is not present, the firmware is responsible for managing * device power. Skip children who aren't attached since they * are handled separately. */ pcib = device_get_parent(dev); for (i = 0; i < numdevs; i++) { child = devlist[i]; dinfo = device_get_ivars(child); dstate = state; if (device_is_attached(child) && PCIB_POWER_FOR_SLEEP(pcib, dev, &dstate) == 0) pci_set_powerstate(child, dstate); } } int pci_suspend(device_t dev) { device_t child, *devlist; struct pci_devinfo *dinfo; int error, i, numdevs; /* * Save the PCI configuration space for each child and set the * device in the appropriate power state for this sleep state. */ if ((error = device_get_children(dev, &devlist, &numdevs)) != 0) return (error); for (i = 0; i < numdevs; i++) { child = devlist[i]; dinfo = device_get_ivars(child); pci_cfg_save(child, dinfo, 0); } /* Suspend devices before potentially powering them down. */ error = bus_generic_suspend(dev); if (error) { free(devlist, M_TEMP); return (error); } if (pci_do_power_suspend) pci_set_power_children(dev, devlist, numdevs, PCI_POWERSTATE_D3); free(devlist, M_TEMP); return (0); } int pci_resume(device_t dev) { device_t child, *devlist; struct pci_devinfo *dinfo; int error, i, numdevs; /* * Set each child to D0 and restore its PCI configuration space. */ if ((error = device_get_children(dev, &devlist, &numdevs)) != 0) return (error); if (pci_do_power_resume) pci_set_power_children(dev, devlist, numdevs, PCI_POWERSTATE_D0); /* Now the device is powered up, restore its config space. */ for (i = 0; i < numdevs; i++) { child = devlist[i]; dinfo = device_get_ivars(child); pci_cfg_restore(child, dinfo); if (!device_is_attached(child)) pci_cfg_save(child, dinfo, 1); } /* * Resume critical devices first, then everything else later. */ for (i = 0; i < numdevs; i++) { child = devlist[i]; switch (pci_get_class(child)) { case PCIC_DISPLAY: case PCIC_MEMORY: case PCIC_BRIDGE: case PCIC_BASEPERIPH: DEVICE_RESUME(child); break; } } for (i = 0; i < numdevs; i++) { child = devlist[i]; switch (pci_get_class(child)) { case PCIC_DISPLAY: case PCIC_MEMORY: case PCIC_BRIDGE: case PCIC_BASEPERIPH: break; default: DEVICE_RESUME(child); } } free(devlist, M_TEMP); return (0); } static void pci_load_vendor_data(void) { caddr_t data; void *ptr; size_t sz; data = preload_search_by_type("pci_vendor_data"); if (data != NULL) { ptr = preload_fetch_addr(data); sz = preload_fetch_size(data); if (ptr != NULL && sz != 0) { pci_vendordata = ptr; pci_vendordata_size = sz; /* terminate the database */ pci_vendordata[pci_vendordata_size] = '\n'; } } } void pci_driver_added(device_t dev, driver_t *driver) { int numdevs; device_t *devlist; device_t child; struct pci_devinfo *dinfo; int i; if (bootverbose) device_printf(dev, "driver added\n"); DEVICE_IDENTIFY(driver, dev); if (device_get_children(dev, &devlist, &numdevs) != 0) return; for (i = 0; i < numdevs; i++) { child = devlist[i]; if (device_get_state(child) != DS_NOTPRESENT) continue; dinfo = device_get_ivars(child); pci_print_verbose(dinfo); if (bootverbose) pci_printf(&dinfo->cfg, "reprobing on driver added\n"); pci_cfg_restore(child, dinfo); if (device_probe_and_attach(child) != 0) pci_cfg_save(child, dinfo, 1); } free(devlist, M_TEMP); } int pci_setup_intr(device_t dev, device_t child, struct resource *irq, int flags, driver_filter_t *filter, driver_intr_t *intr, void *arg, void **cookiep) { struct pci_devinfo *dinfo; struct msix_table_entry *mte; struct msix_vector *mv; uint64_t addr; uint32_t data; void *cookie; int error, rid; error = bus_generic_setup_intr(dev, child, irq, flags, filter, intr, arg, &cookie); if (error) return (error); /* If this is not a direct child, just bail out. */ if (device_get_parent(child) != dev) { *cookiep = cookie; return(0); } rid = rman_get_rid(irq); if (rid == 0) { /* Make sure that INTx is enabled */ pci_clear_command_bit(dev, child, PCIM_CMD_INTxDIS); } else { /* * Check to see if the interrupt is MSI or MSI-X. * Ask our parent to map the MSI and give * us the address and data register values. * If we fail for some reason, teardown the * interrupt handler. */ dinfo = device_get_ivars(child); if (dinfo->cfg.msi.msi_alloc > 0) { if (dinfo->cfg.msi.msi_addr == 0) { KASSERT(dinfo->cfg.msi.msi_handlers == 0, ("MSI has handlers, but vectors not mapped")); error = PCIB_MAP_MSI(device_get_parent(dev), child, rman_get_start(irq), &addr, &data); if (error) goto bad; dinfo->cfg.msi.msi_addr = addr; dinfo->cfg.msi.msi_data = data; } if (dinfo->cfg.msi.msi_handlers == 0) pci_enable_msi(child, dinfo->cfg.msi.msi_addr, dinfo->cfg.msi.msi_data); dinfo->cfg.msi.msi_handlers++; } else { KASSERT(dinfo->cfg.msix.msix_alloc > 0, ("No MSI or MSI-X interrupts allocated")); KASSERT(rid <= dinfo->cfg.msix.msix_table_len, ("MSI-X index too high")); mte = &dinfo->cfg.msix.msix_table[rid - 1]; KASSERT(mte->mte_vector != 0, ("no message vector")); mv = &dinfo->cfg.msix.msix_vectors[mte->mte_vector - 1]; KASSERT(mv->mv_irq == rman_get_start(irq), ("IRQ mismatch")); if (mv->mv_address == 0) { KASSERT(mte->mte_handlers == 0, ("MSI-X table entry has handlers, but vector not mapped")); error = PCIB_MAP_MSI(device_get_parent(dev), child, rman_get_start(irq), &addr, &data); if (error) goto bad; mv->mv_address = addr; mv->mv_data = data; } if (mte->mte_handlers == 0) { pci_enable_msix(child, rid - 1, mv->mv_address, mv->mv_data); pci_unmask_msix(child, rid - 1); } mte->mte_handlers++; } /* Make sure that INTx is disabled if we are using MSI/MSIX */ pci_set_command_bit(dev, child, PCIM_CMD_INTxDIS); bad: if (error) { (void)bus_generic_teardown_intr(dev, child, irq, cookie); return (error); } } *cookiep = cookie; return (0); } int pci_teardown_intr(device_t dev, device_t child, struct resource *irq, void *cookie) { struct msix_table_entry *mte; struct resource_list_entry *rle; struct pci_devinfo *dinfo; int error, rid; if (irq == NULL || !(rman_get_flags(irq) & RF_ACTIVE)) return (EINVAL); /* If this isn't a direct child, just bail out */ if (device_get_parent(child) != dev) return(bus_generic_teardown_intr(dev, child, irq, cookie)); rid = rman_get_rid(irq); if (rid == 0) { /* Mask INTx */ pci_set_command_bit(dev, child, PCIM_CMD_INTxDIS); } else { /* * Check to see if the interrupt is MSI or MSI-X. If so, * decrement the appropriate handlers count and mask the * MSI-X message, or disable MSI messages if the count * drops to 0. */ dinfo = device_get_ivars(child); rle = resource_list_find(&dinfo->resources, SYS_RES_IRQ, rid); if (rle->res != irq) return (EINVAL); if (dinfo->cfg.msi.msi_alloc > 0) { KASSERT(rid <= dinfo->cfg.msi.msi_alloc, ("MSI-X index too high")); if (dinfo->cfg.msi.msi_handlers == 0) return (EINVAL); dinfo->cfg.msi.msi_handlers--; if (dinfo->cfg.msi.msi_handlers == 0) pci_disable_msi(child); } else { KASSERT(dinfo->cfg.msix.msix_alloc > 0, ("No MSI or MSI-X interrupts allocated")); KASSERT(rid <= dinfo->cfg.msix.msix_table_len, ("MSI-X index too high")); mte = &dinfo->cfg.msix.msix_table[rid - 1]; if (mte->mte_handlers == 0) return (EINVAL); mte->mte_handlers--; if (mte->mte_handlers == 0) pci_mask_msix(child, rid - 1); } } error = bus_generic_teardown_intr(dev, child, irq, cookie); if (rid > 0) KASSERT(error == 0, ("%s: generic teardown failed for MSI/MSI-X", __func__)); return (error); } int pci_print_child(device_t dev, device_t child) { struct pci_devinfo *dinfo; struct resource_list *rl; int retval = 0; dinfo = device_get_ivars(child); rl = &dinfo->resources; retval += bus_print_child_header(dev, child); retval += resource_list_print_type(rl, "port", SYS_RES_IOPORT, "%#lx"); retval += resource_list_print_type(rl, "mem", SYS_RES_MEMORY, "%#lx"); retval += resource_list_print_type(rl, "irq", SYS_RES_IRQ, "%ld"); if (device_get_flags(dev)) retval += printf(" flags %#x", device_get_flags(dev)); retval += printf(" at device %d.%d", pci_get_slot(child), pci_get_function(child)); retval += bus_print_child_footer(dev, child); return (retval); } static struct { int class; int subclass; char *desc; } pci_nomatch_tab[] = { {PCIC_OLD, -1, "old"}, {PCIC_OLD, PCIS_OLD_NONVGA, "non-VGA display device"}, {PCIC_OLD, PCIS_OLD_VGA, "VGA-compatible display device"}, {PCIC_STORAGE, -1, "mass storage"}, {PCIC_STORAGE, PCIS_STORAGE_SCSI, "SCSI"}, {PCIC_STORAGE, PCIS_STORAGE_IDE, "ATA"}, {PCIC_STORAGE, PCIS_STORAGE_FLOPPY, "floppy disk"}, {PCIC_STORAGE, PCIS_STORAGE_IPI, "IPI"}, {PCIC_STORAGE, PCIS_STORAGE_RAID, "RAID"}, {PCIC_STORAGE, PCIS_STORAGE_ATA_ADMA, "ATA (ADMA)"}, {PCIC_STORAGE, PCIS_STORAGE_SATA, "SATA"}, {PCIC_STORAGE, PCIS_STORAGE_SAS, "SAS"}, {PCIC_NETWORK, -1, "network"}, {PCIC_NETWORK, PCIS_NETWORK_ETHERNET, "ethernet"}, {PCIC_NETWORK, PCIS_NETWORK_TOKENRING, "token ring"}, {PCIC_NETWORK, PCIS_NETWORK_FDDI, "fddi"}, {PCIC_NETWORK, PCIS_NETWORK_ATM, "ATM"}, {PCIC_NETWORK, PCIS_NETWORK_ISDN, "ISDN"}, {PCIC_DISPLAY, -1, "display"}, {PCIC_DISPLAY, PCIS_DISPLAY_VGA, "VGA"}, {PCIC_DISPLAY, PCIS_DISPLAY_XGA, "XGA"}, {PCIC_DISPLAY, PCIS_DISPLAY_3D, "3D"}, {PCIC_MULTIMEDIA, -1, "multimedia"}, {PCIC_MULTIMEDIA, PCIS_MULTIMEDIA_VIDEO, "video"}, {PCIC_MULTIMEDIA, PCIS_MULTIMEDIA_AUDIO, "audio"}, {PCIC_MULTIMEDIA, PCIS_MULTIMEDIA_TELE, "telephony"}, {PCIC_MULTIMEDIA, PCIS_MULTIMEDIA_HDA, "HDA"}, {PCIC_MEMORY, -1, "memory"}, {PCIC_MEMORY, PCIS_MEMORY_RAM, "RAM"}, {PCIC_MEMORY, PCIS_MEMORY_FLASH, "flash"}, {PCIC_BRIDGE, -1, "bridge"}, {PCIC_BRIDGE, PCIS_BRIDGE_HOST, "HOST-PCI"}, {PCIC_BRIDGE, PCIS_BRIDGE_ISA, "PCI-ISA"}, {PCIC_BRIDGE, PCIS_BRIDGE_EISA, "PCI-EISA"}, {PCIC_BRIDGE, PCIS_BRIDGE_MCA, "PCI-MCA"}, {PCIC_BRIDGE, PCIS_BRIDGE_PCI, "PCI-PCI"}, {PCIC_BRIDGE, PCIS_BRIDGE_PCMCIA, "PCI-PCMCIA"}, {PCIC_BRIDGE, PCIS_BRIDGE_NUBUS, "PCI-NuBus"}, {PCIC_BRIDGE, PCIS_BRIDGE_CARDBUS, "PCI-CardBus"}, {PCIC_BRIDGE, PCIS_BRIDGE_RACEWAY, "PCI-RACEway"}, {PCIC_SIMPLECOMM, -1, "simple comms"}, {PCIC_SIMPLECOMM, PCIS_SIMPLECOMM_UART, "UART"}, /* could detect 16550 */ {PCIC_SIMPLECOMM, PCIS_SIMPLECOMM_PAR, "parallel port"}, {PCIC_SIMPLECOMM, PCIS_SIMPLECOMM_MULSER, "multiport serial"}, {PCIC_SIMPLECOMM, PCIS_SIMPLECOMM_MODEM, "generic modem"}, {PCIC_BASEPERIPH, -1, "base peripheral"}, {PCIC_BASEPERIPH, PCIS_BASEPERIPH_PIC, "interrupt controller"}, {PCIC_BASEPERIPH, PCIS_BASEPERIPH_DMA, "DMA controller"}, {PCIC_BASEPERIPH, PCIS_BASEPERIPH_TIMER, "timer"}, {PCIC_BASEPERIPH, PCIS_BASEPERIPH_RTC, "realtime clock"}, {PCIC_BASEPERIPH, PCIS_BASEPERIPH_PCIHOT, "PCI hot-plug controller"}, {PCIC_BASEPERIPH, PCIS_BASEPERIPH_SDHC, "SD host controller"}, {PCIC_INPUTDEV, -1, "input device"}, {PCIC_INPUTDEV, PCIS_INPUTDEV_KEYBOARD, "keyboard"}, {PCIC_INPUTDEV, PCIS_INPUTDEV_DIGITIZER,"digitizer"}, {PCIC_INPUTDEV, PCIS_INPUTDEV_MOUSE, "mouse"}, {PCIC_INPUTDEV, PCIS_INPUTDEV_SCANNER, "scanner"}, {PCIC_INPUTDEV, PCIS_INPUTDEV_GAMEPORT, "gameport"}, {PCIC_DOCKING, -1, "docking station"}, {PCIC_PROCESSOR, -1, "processor"}, {PCIC_SERIALBUS, -1, "serial bus"}, {PCIC_SERIALBUS, PCIS_SERIALBUS_FW, "FireWire"}, {PCIC_SERIALBUS, PCIS_SERIALBUS_ACCESS, "AccessBus"}, {PCIC_SERIALBUS, PCIS_SERIALBUS_SSA, "SSA"}, {PCIC_SERIALBUS, PCIS_SERIALBUS_USB, "USB"}, {PCIC_SERIALBUS, PCIS_SERIALBUS_FC, "Fibre Channel"}, {PCIC_SERIALBUS, PCIS_SERIALBUS_SMBUS, "SMBus"}, {PCIC_WIRELESS, -1, "wireless controller"}, {PCIC_WIRELESS, PCIS_WIRELESS_IRDA, "iRDA"}, {PCIC_WIRELESS, PCIS_WIRELESS_IR, "IR"}, {PCIC_WIRELESS, PCIS_WIRELESS_RF, "RF"}, {PCIC_INTELLIIO, -1, "intelligent I/O controller"}, {PCIC_INTELLIIO, PCIS_INTELLIIO_I2O, "I2O"}, {PCIC_SATCOM, -1, "satellite communication"}, {PCIC_SATCOM, PCIS_SATCOM_TV, "sat TV"}, {PCIC_SATCOM, PCIS_SATCOM_AUDIO, "sat audio"}, {PCIC_SATCOM, PCIS_SATCOM_VOICE, "sat voice"}, {PCIC_SATCOM, PCIS_SATCOM_DATA, "sat data"}, {PCIC_CRYPTO, -1, "encrypt/decrypt"}, {PCIC_CRYPTO, PCIS_CRYPTO_NETCOMP, "network/computer crypto"}, {PCIC_CRYPTO, PCIS_CRYPTO_ENTERTAIN, "entertainment crypto"}, {PCIC_DASP, -1, "dasp"}, {PCIC_DASP, PCIS_DASP_DPIO, "DPIO module"}, {0, 0, NULL} }; void pci_probe_nomatch(device_t dev, device_t child) { int i; char *cp, *scp, *device; /* * Look for a listing for this device in a loaded device database. */ if ((device = pci_describe_device(child)) != NULL) { device_printf(dev, "<%s>", device); free(device, M_DEVBUF); } else { /* * Scan the class/subclass descriptions for a general * description. */ cp = "unknown"; scp = NULL; for (i = 0; pci_nomatch_tab[i].desc != NULL; i++) { if (pci_nomatch_tab[i].class == pci_get_class(child)) { if (pci_nomatch_tab[i].subclass == -1) { cp = pci_nomatch_tab[i].desc; } else if (pci_nomatch_tab[i].subclass == pci_get_subclass(child)) { scp = pci_nomatch_tab[i].desc; } } } device_printf(dev, "<%s%s%s>", cp ? cp : "", ((cp != NULL) && (scp != NULL)) ? ", " : "", scp ? scp : ""); } printf(" at device %d.%d (no driver attached)\n", pci_get_slot(child), pci_get_function(child)); pci_cfg_save(child, device_get_ivars(child), 1); return; } /* * Parse the PCI device database, if loaded, and return a pointer to a * description of the device. * * The database is flat text formatted as follows: * * Any line not in a valid format is ignored. * Lines are terminated with newline '\n' characters. * * A VENDOR line consists of the 4 digit (hex) vendor code, a TAB, then * the vendor name. * * A DEVICE line is entered immediately below the corresponding VENDOR ID. * - devices cannot be listed without a corresponding VENDOR line. * A DEVICE line consists of a TAB, the 4 digit (hex) device code, * another TAB, then the device name. */ /* * Assuming (ptr) points to the beginning of a line in the database, * return the vendor or device and description of the next entry. * The value of (vendor) or (device) inappropriate for the entry type * is set to -1. Returns nonzero at the end of the database. * * Note that this is slightly unrobust in the face of corrupt data; * we attempt to safeguard against this by spamming the end of the * database with a newline when we initialise. */ static int pci_describe_parse_line(char **ptr, int *vendor, int *device, char **desc) { char *cp = *ptr; int left; *device = -1; *vendor = -1; **desc = '\0'; for (;;) { left = pci_vendordata_size - (cp - pci_vendordata); if (left <= 0) { *ptr = cp; return(1); } /* vendor entry? */ if (*cp != '\t' && sscanf(cp, "%x\t%80[^\n]", vendor, *desc) == 2) break; /* device entry? */ if (*cp == '\t' && sscanf(cp, "%x\t%80[^\n]", device, *desc) == 2) break; /* skip to next line */ while (*cp != '\n' && left > 0) { cp++; left--; } if (*cp == '\n') { cp++; left--; } } /* skip to next line */ while (*cp != '\n' && left > 0) { cp++; left--; } if (*cp == '\n' && left > 0) cp++; *ptr = cp; return(0); } static char * pci_describe_device(device_t dev) { int vendor, device; char *desc, *vp, *dp, *line; desc = vp = dp = NULL; /* * If we have no vendor data, we can't do anything. */ if (pci_vendordata == NULL) goto out; /* * Scan the vendor data looking for this device */ line = pci_vendordata; if ((vp = malloc(80, M_DEVBUF, M_NOWAIT)) == NULL) goto out; for (;;) { if (pci_describe_parse_line(&line, &vendor, &device, &vp)) goto out; if (vendor == pci_get_vendor(dev)) break; } if ((dp = malloc(80, M_DEVBUF, M_NOWAIT)) == NULL) goto out; for (;;) { if (pci_describe_parse_line(&line, &vendor, &device, &dp)) { *dp = 0; break; } if (vendor != -1) { *dp = 0; break; } if (device == pci_get_device(dev)) break; } if (dp[0] == '\0') snprintf(dp, 80, "0x%x", pci_get_device(dev)); if ((desc = malloc(strlen(vp) + strlen(dp) + 3, M_DEVBUF, M_NOWAIT)) != NULL) sprintf(desc, "%s, %s", vp, dp); out: if (vp != NULL) free(vp, M_DEVBUF); if (dp != NULL) free(dp, M_DEVBUF); return(desc); } int pci_read_ivar(device_t dev, device_t child, int which, uintptr_t *result) { struct pci_devinfo *dinfo; pcicfgregs *cfg; dinfo = device_get_ivars(child); cfg = &dinfo->cfg; switch (which) { case PCI_IVAR_ETHADDR: /* * The generic accessor doesn't deal with failure, so * we set the return value, then return an error. */ *((uint8_t **) result) = NULL; return (EINVAL); case PCI_IVAR_SUBVENDOR: *result = cfg->subvendor; break; case PCI_IVAR_SUBDEVICE: *result = cfg->subdevice; break; case PCI_IVAR_VENDOR: *result = cfg->vendor; break; case PCI_IVAR_DEVICE: *result = cfg->device; break; case PCI_IVAR_DEVID: *result = (cfg->device << 16) | cfg->vendor; break; case PCI_IVAR_CLASS: *result = cfg->baseclass; break; case PCI_IVAR_SUBCLASS: *result = cfg->subclass; break; case PCI_IVAR_PROGIF: *result = cfg->progif; break; case PCI_IVAR_REVID: *result = cfg->revid; break; case PCI_IVAR_INTPIN: *result = cfg->intpin; break; case PCI_IVAR_IRQ: *result = cfg->intline; break; case PCI_IVAR_DOMAIN: *result = cfg->domain; break; case PCI_IVAR_BUS: *result = cfg->bus; break; case PCI_IVAR_SLOT: *result = cfg->slot; break; case PCI_IVAR_FUNCTION: *result = cfg->func; break; case PCI_IVAR_CMDREG: *result = cfg->cmdreg; break; case PCI_IVAR_CACHELNSZ: *result = cfg->cachelnsz; break; case PCI_IVAR_MINGNT: *result = cfg->mingnt; break; case PCI_IVAR_MAXLAT: *result = cfg->maxlat; break; case PCI_IVAR_LATTIMER: *result = cfg->lattimer; break; default: return (ENOENT); } return (0); } int pci_write_ivar(device_t dev, device_t child, int which, uintptr_t value) { struct pci_devinfo *dinfo; dinfo = device_get_ivars(child); switch (which) { case PCI_IVAR_INTPIN: dinfo->cfg.intpin = value; return (0); case PCI_IVAR_ETHADDR: case PCI_IVAR_SUBVENDOR: case PCI_IVAR_SUBDEVICE: case PCI_IVAR_VENDOR: case PCI_IVAR_DEVICE: case PCI_IVAR_DEVID: case PCI_IVAR_CLASS: case PCI_IVAR_SUBCLASS: case PCI_IVAR_PROGIF: case PCI_IVAR_REVID: case PCI_IVAR_IRQ: case PCI_IVAR_DOMAIN: case PCI_IVAR_BUS: case PCI_IVAR_SLOT: case PCI_IVAR_FUNCTION: return (EINVAL); /* disallow for now */ default: return (ENOENT); } } #include "opt_ddb.h" #ifdef DDB #include #include /* * List resources based on pci map registers, used for within ddb */ DB_SHOW_COMMAND(pciregs, db_pci_dump) { struct pci_devinfo *dinfo; struct devlist *devlist_head; struct pci_conf *p; const char *name; int i, error, none_count; none_count = 0; /* get the head of the device queue */ devlist_head = &pci_devq; /* * Go through the list of devices and print out devices */ for (error = 0, i = 0, dinfo = STAILQ_FIRST(devlist_head); (dinfo != NULL) && (error == 0) && (i < pci_numdevs) && !db_pager_quit; dinfo = STAILQ_NEXT(dinfo, pci_links), i++) { /* Populate pd_name and pd_unit */ name = NULL; if (dinfo->cfg.dev) name = device_get_name(dinfo->cfg.dev); p = &dinfo->conf; db_printf("%s%d@pci%d:%d:%d:%d:\tclass=0x%06x card=0x%08x " "chip=0x%08x rev=0x%02x hdr=0x%02x\n", (name && *name) ? name : "none", (name && *name) ? (int)device_get_unit(dinfo->cfg.dev) : none_count++, p->pc_sel.pc_domain, p->pc_sel.pc_bus, p->pc_sel.pc_dev, p->pc_sel.pc_func, (p->pc_class << 16) | (p->pc_subclass << 8) | p->pc_progif, (p->pc_subdevice << 16) | p->pc_subvendor, (p->pc_device << 16) | p->pc_vendor, p->pc_revid, p->pc_hdr); } } #endif /* DDB */ static struct resource * pci_reserve_map(device_t dev, device_t child, int type, int *rid, u_long start, u_long end, u_long count, u_int flags) { struct pci_devinfo *dinfo = device_get_ivars(child); struct resource_list *rl = &dinfo->resources; struct resource_list_entry *rle; struct resource *res; struct pci_map *pm; pci_addr_t map, testval; int mapsize; res = NULL; pm = pci_find_bar(child, *rid); if (pm != NULL) { /* This is a BAR that we failed to allocate earlier. */ mapsize = pm->pm_size; map = pm->pm_value; } else { /* * Weed out the bogons, and figure out how large the * BAR/map is. BARs that read back 0 here are bogus * and unimplemented. Note: atapci in legacy mode are * special and handled elsewhere in the code. If you * have a atapci device in legacy mode and it fails * here, that other code is broken. */ pci_read_bar(child, *rid, &map, &testval); /* * Determine the size of the BAR and ignore BARs with a size * of 0. Device ROM BARs use a different mask value. */ if (PCIR_IS_BIOS(&dinfo->cfg, *rid)) mapsize = pci_romsize(testval); else mapsize = pci_mapsize(testval); if (mapsize == 0) goto out; pm = pci_add_bar(child, *rid, map, mapsize); } if (PCI_BAR_MEM(map) || PCIR_IS_BIOS(&dinfo->cfg, *rid)) { if (type != SYS_RES_MEMORY) { if (bootverbose) device_printf(dev, "child %s requested type %d for rid %#x," " but the BAR says it is an memio\n", device_get_nameunit(child), type, *rid); goto out; } } else { if (type != SYS_RES_IOPORT) { if (bootverbose) device_printf(dev, "child %s requested type %d for rid %#x," " but the BAR says it is an ioport\n", device_get_nameunit(child), type, *rid); goto out; } } /* * For real BARs, we need to override the size that * the driver requests, because that's what the BAR * actually uses and we would otherwise have a * situation where we might allocate the excess to * another driver, which won't work. */ count = (pci_addr_t)1 << mapsize; if (RF_ALIGNMENT(flags) < mapsize) flags = (flags & ~RF_ALIGNMENT_MASK) | RF_ALIGNMENT_LOG2(mapsize); if (PCI_BAR_MEM(map) && (map & PCIM_BAR_MEM_PREFETCH)) flags |= RF_PREFETCHABLE; /* * Allocate enough resource, and then write back the * appropriate BAR for that resource. */ res = BUS_ALLOC_RESOURCE(device_get_parent(dev), child, type, rid, start, end, count, flags & ~RF_ACTIVE); if (res == NULL) { device_printf(child, "%#lx bytes of rid %#x res %d failed (%#lx, %#lx).\n", count, *rid, type, start, end); goto out; } resource_list_add(rl, type, *rid, start, end, count); rle = resource_list_find(rl, type, *rid); if (rle == NULL) panic("pci_reserve_map: unexpectedly can't find resource."); rle->res = res; rle->start = rman_get_start(res); rle->end = rman_get_end(res); rle->count = count; rle->flags = RLE_RESERVED; if (bootverbose) device_printf(child, "Lazy allocation of %#lx bytes rid %#x type %d at %#lx\n", count, *rid, type, rman_get_start(res)); map = rman_get_start(res); pci_write_bar(child, pm, map); out:; return (res); } struct resource * pci_alloc_resource(device_t dev, device_t child, int type, int *rid, u_long start, u_long end, u_long count, u_int flags) { struct pci_devinfo *dinfo = device_get_ivars(child); struct resource_list *rl = &dinfo->resources; struct resource_list_entry *rle; struct resource *res; pcicfgregs *cfg = &dinfo->cfg; if (device_get_parent(child) != dev) return (BUS_ALLOC_RESOURCE(device_get_parent(dev), child, type, rid, start, end, count, flags)); /* * Perform lazy resource allocation */ switch (type) { case SYS_RES_IRQ: /* * Can't alloc legacy interrupt once MSI messages have * been allocated. */ if (*rid == 0 && (cfg->msi.msi_alloc > 0 || cfg->msix.msix_alloc > 0)) return (NULL); /* * If the child device doesn't have an interrupt * routed and is deserving of an interrupt, try to * assign it one. */ if (*rid == 0 && !PCI_INTERRUPT_VALID(cfg->intline) && (cfg->intpin != 0)) pci_assign_interrupt(dev, child, 0); break; case SYS_RES_IOPORT: case SYS_RES_MEMORY: #ifdef NEW_PCIB /* * PCI-PCI bridge I/O window resources are not BARs. * For those allocations just pass the request up the * tree. */ if (cfg->hdrtype == PCIM_HDRTYPE_BRIDGE) { switch (*rid) { case PCIR_IOBASEL_1: case PCIR_MEMBASE_1: case PCIR_PMBASEL_1: /* * XXX: Should we bother creating a resource * list entry? */ return (bus_generic_alloc_resource(dev, child, type, rid, start, end, count, flags)); } } #endif /* Reserve resources for this BAR if needed. */ rle = resource_list_find(rl, type, *rid); if (rle == NULL) { res = pci_reserve_map(dev, child, type, rid, start, end, count, flags); if (res == NULL) return (NULL); } } return (resource_list_alloc(rl, dev, child, type, rid, start, end, count, flags)); } int pci_activate_resource(device_t dev, device_t child, int type, int rid, struct resource *r) { struct pci_devinfo *dinfo; int error; error = bus_generic_activate_resource(dev, child, type, rid, r); if (error) return (error); /* Enable decoding in the command register when activating BARs. */ if (device_get_parent(child) == dev) { /* Device ROMs need their decoding explicitly enabled. */ dinfo = device_get_ivars(child); if (PCIR_IS_BIOS(&dinfo->cfg, rid)) pci_write_bar(child, pci_find_bar(child, rid), rman_get_start(r) | PCIM_BIOS_ENABLE); switch (type) { case SYS_RES_IOPORT: case SYS_RES_MEMORY: error = PCI_ENABLE_IO(dev, child, type); break; } } return (error); } int pci_deactivate_resource(device_t dev, device_t child, int type, int rid, struct resource *r) { struct pci_devinfo *dinfo; int error; error = bus_generic_deactivate_resource(dev, child, type, rid, r); if (error) return (error); /* Disable decoding for device ROMs. */ if (device_get_parent(child) == dev) { dinfo = device_get_ivars(child); if (PCIR_IS_BIOS(&dinfo->cfg, rid)) pci_write_bar(child, pci_find_bar(child, rid), rman_get_start(r)); } return (0); } void pci_delete_child(device_t dev, device_t child) { struct resource_list_entry *rle; struct resource_list *rl; struct pci_devinfo *dinfo; dinfo = device_get_ivars(child); rl = &dinfo->resources; if (device_is_attached(child)) device_detach(child); /* Turn off access to resources we're about to free */ pci_write_config(child, PCIR_COMMAND, pci_read_config(child, PCIR_COMMAND, 2) & ~(PCIM_CMD_MEMEN | PCIM_CMD_PORTEN), 2); /* Free all allocated resources */ STAILQ_FOREACH(rle, rl, link) { if (rle->res) { if (rman_get_flags(rle->res) & RF_ACTIVE || resource_list_busy(rl, rle->type, rle->rid)) { pci_printf(&dinfo->cfg, "Resource still owned, oops. " "(type=%d, rid=%d, addr=%lx)\n", rle->type, rle->rid, rman_get_start(rle->res)); bus_release_resource(child, rle->type, rle->rid, rle->res); } resource_list_unreserve(rl, dev, child, rle->type, rle->rid); } } resource_list_free(rl); device_delete_child(dev, child); pci_freecfg(dinfo); } void pci_delete_resource(device_t dev, device_t child, int type, int rid) { struct pci_devinfo *dinfo; struct resource_list *rl; struct resource_list_entry *rle; if (device_get_parent(child) != dev) return; dinfo = device_get_ivars(child); rl = &dinfo->resources; rle = resource_list_find(rl, type, rid); if (rle == NULL) return; if (rle->res) { if (rman_get_flags(rle->res) & RF_ACTIVE || resource_list_busy(rl, type, rid)) { device_printf(dev, "delete_resource: " "Resource still owned by child, oops. " "(type=%d, rid=%d, addr=%lx)\n", type, rid, rman_get_start(rle->res)); return; } #ifndef __PCI_BAR_ZERO_VALID /* * If this is a BAR, clear the BAR so it stops * decoding before releasing the resource. */ switch (type) { case SYS_RES_IOPORT: case SYS_RES_MEMORY: pci_write_bar(child, pci_find_bar(child, rid), 0); break; } #endif resource_list_unreserve(rl, dev, child, type, rid); } resource_list_delete(rl, type, rid); } struct resource_list * pci_get_resource_list (device_t dev, device_t child) { struct pci_devinfo *dinfo = device_get_ivars(child); return (&dinfo->resources); } uint32_t pci_read_config_method(device_t dev, device_t child, int reg, int width) { struct pci_devinfo *dinfo = device_get_ivars(child); pcicfgregs *cfg = &dinfo->cfg; return (PCIB_READ_CONFIG(device_get_parent(dev), cfg->bus, cfg->slot, cfg->func, reg, width)); } void pci_write_config_method(device_t dev, device_t child, int reg, uint32_t val, int width) { struct pci_devinfo *dinfo = device_get_ivars(child); pcicfgregs *cfg = &dinfo->cfg; PCIB_WRITE_CONFIG(device_get_parent(dev), cfg->bus, cfg->slot, cfg->func, reg, val, width); } int pci_child_location_str_method(device_t dev, device_t child, char *buf, size_t buflen) { snprintf(buf, buflen, "slot=%d function=%d", pci_get_slot(child), pci_get_function(child)); return (0); } int pci_child_pnpinfo_str_method(device_t dev, device_t child, char *buf, size_t buflen) { struct pci_devinfo *dinfo; pcicfgregs *cfg; dinfo = device_get_ivars(child); cfg = &dinfo->cfg; snprintf(buf, buflen, "vendor=0x%04x device=0x%04x subvendor=0x%04x " "subdevice=0x%04x class=0x%02x%02x%02x", cfg->vendor, cfg->device, cfg->subvendor, cfg->subdevice, cfg->baseclass, cfg->subclass, cfg->progif); return (0); } int pci_assign_interrupt_method(device_t dev, device_t child) { struct pci_devinfo *dinfo = device_get_ivars(child); pcicfgregs *cfg = &dinfo->cfg; return (PCIB_ROUTE_INTERRUPT(device_get_parent(dev), child, cfg->intpin)); } static int pci_modevent(module_t mod, int what, void *arg) { static struct cdev *pci_cdev; switch (what) { case MOD_LOAD: STAILQ_INIT(&pci_devq); pci_generation = 0; pci_cdev = make_dev(&pcicdev, 0, UID_ROOT, GID_WHEEL, 0644, "pci"); pci_load_vendor_data(); break; case MOD_UNLOAD: destroy_dev(pci_cdev); break; } return (0); } void pci_cfg_restore(device_t dev, struct pci_devinfo *dinfo) { /* * Only do header type 0 devices. Type 1 devices are bridges, * which we know need special treatment. Type 2 devices are * cardbus bridges which also require special treatment. * Other types are unknown, and we err on the side of safety * by ignoring them. */ if ((dinfo->cfg.hdrtype & PCIM_HDRTYPE) != PCIM_HDRTYPE_NORMAL) return; /* * Restore the device to full power mode. We must do this * before we restore the registers because moving from D3 to * D0 will cause the chip's BARs and some other registers to * be reset to some unknown power on reset values. Cut down * the noise on boot by doing nothing if we are already in * state D0. */ if (pci_get_powerstate(dev) != PCI_POWERSTATE_D0) pci_set_powerstate(dev, PCI_POWERSTATE_D0); pci_restore_bars(dev); pci_write_config(dev, PCIR_COMMAND, dinfo->cfg.cmdreg, 2); pci_write_config(dev, PCIR_INTLINE, dinfo->cfg.intline, 1); pci_write_config(dev, PCIR_INTPIN, dinfo->cfg.intpin, 1); pci_write_config(dev, PCIR_MINGNT, dinfo->cfg.mingnt, 1); pci_write_config(dev, PCIR_MAXLAT, dinfo->cfg.maxlat, 1); pci_write_config(dev, PCIR_CACHELNSZ, dinfo->cfg.cachelnsz, 1); pci_write_config(dev, PCIR_LATTIMER, dinfo->cfg.lattimer, 1); pci_write_config(dev, PCIR_PROGIF, dinfo->cfg.progif, 1); pci_write_config(dev, PCIR_REVID, dinfo->cfg.revid, 1); /* Restore MSI and MSI-X configurations if they are present. */ if (dinfo->cfg.msi.msi_location != 0) pci_resume_msi(dev); if (dinfo->cfg.msix.msix_location != 0) pci_resume_msix(dev); } void pci_cfg_save(device_t dev, struct pci_devinfo *dinfo, int setstate) { uint32_t cls; int ps; /* * Only do header type 0 devices. Type 1 devices are bridges, which * we know need special treatment. Type 2 devices are cardbus bridges * which also require special treatment. Other types are unknown, and * we err on the side of safety by ignoring them. Powering down * bridges should not be undertaken lightly. */ if ((dinfo->cfg.hdrtype & PCIM_HDRTYPE) != PCIM_HDRTYPE_NORMAL) return; /* * Some drivers apparently write to these registers w/o updating our * cached copy. No harm happens if we update the copy, so do so here * so we can restore them. The COMMAND register is modified by the * bus w/o updating the cache. This should represent the normally * writable portion of the 'defined' part of type 0 headers. In * theory we also need to save/restore the PCI capability structures * we know about, but apart from power we don't know any that are * writable. */ dinfo->cfg.subvendor = pci_read_config(dev, PCIR_SUBVEND_0, 2); dinfo->cfg.subdevice = pci_read_config(dev, PCIR_SUBDEV_0, 2); dinfo->cfg.vendor = pci_read_config(dev, PCIR_VENDOR, 2); dinfo->cfg.device = pci_read_config(dev, PCIR_DEVICE, 2); dinfo->cfg.cmdreg = pci_read_config(dev, PCIR_COMMAND, 2); dinfo->cfg.intline = pci_read_config(dev, PCIR_INTLINE, 1); dinfo->cfg.intpin = pci_read_config(dev, PCIR_INTPIN, 1); dinfo->cfg.mingnt = pci_read_config(dev, PCIR_MINGNT, 1); dinfo->cfg.maxlat = pci_read_config(dev, PCIR_MAXLAT, 1); dinfo->cfg.cachelnsz = pci_read_config(dev, PCIR_CACHELNSZ, 1); dinfo->cfg.lattimer = pci_read_config(dev, PCIR_LATTIMER, 1); dinfo->cfg.baseclass = pci_read_config(dev, PCIR_CLASS, 1); dinfo->cfg.subclass = pci_read_config(dev, PCIR_SUBCLASS, 1); dinfo->cfg.progif = pci_read_config(dev, PCIR_PROGIF, 1); dinfo->cfg.revid = pci_read_config(dev, PCIR_REVID, 1); /* * don't set the state for display devices, base peripherals and * memory devices since bad things happen when they are powered down. * We should (a) have drivers that can easily detach and (b) use * generic drivers for these devices so that some device actually * attaches. We need to make sure that when we implement (a) we don't * power the device down on a reattach. */ cls = pci_get_class(dev); if (!setstate) return; switch (pci_do_power_nodriver) { case 0: /* NO powerdown at all */ return; case 1: /* Conservative about what to power down */ if (cls == PCIC_STORAGE) return; /*FALLTHROUGH*/ case 2: /* Agressive about what to power down */ if (cls == PCIC_DISPLAY || cls == PCIC_MEMORY || cls == PCIC_BASEPERIPH) return; /*FALLTHROUGH*/ case 3: /* Power down everything */ break; } /* * PCI spec says we can only go into D3 state from D0 state. * Transition from D[12] into D0 before going to D3 state. */ ps = pci_get_powerstate(dev); if (ps != PCI_POWERSTATE_D0 && ps != PCI_POWERSTATE_D3) pci_set_powerstate(dev, PCI_POWERSTATE_D0); if (pci_get_powerstate(dev) != PCI_POWERSTATE_D3) pci_set_powerstate(dev, PCI_POWERSTATE_D3); } Index: projects/largeSMP/sys/dev/puc/pucdata.c =================================================================== --- projects/largeSMP/sys/dev/puc/pucdata.c (revision 222782) +++ projects/largeSMP/sys/dev/puc/pucdata.c (revision 222783) @@ -1,1441 +1,1439 @@ /*- * Copyright (c) 2006 Marcel Moolenaar * 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("$FreeBSD$"); /* * PCI "universal" communications card driver configuration data (used to * match/attach the cards). */ #include #include #include #include #include #include #include #include #include #include #include static puc_config_f puc_config_amc; static puc_config_f puc_config_diva; static puc_config_f puc_config_exar; static puc_config_f puc_config_icbook; +static puc_config_f puc_config_oxford_pcie; static puc_config_f puc_config_quatech; static puc_config_f puc_config_syba; static puc_config_f puc_config_siig; static puc_config_f puc_config_timedia; static puc_config_f puc_config_titan; -static puc_config_f puc_config_oxford_pcie; const struct puc_cfg puc_pci_devices[] = { { 0x0009, 0x7168, 0xffff, 0, "Sunix SUN1889", DEFAULT_RCLK * 8, PUC_PORT_2S, 0x10, 0, 8, }, { 0x103c, 0x1048, 0x103c, 0x1049, "HP Diva Serial [GSP] Multiport UART - Tosca Console", DEFAULT_RCLK, PUC_PORT_3S, 0x10, 0, -1, .config_function = puc_config_diva }, { 0x103c, 0x1048, 0x103c, 0x104a, "HP Diva Serial [GSP] Multiport UART - Tosca Secondary", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 0, -1, .config_function = puc_config_diva }, { 0x103c, 0x1048, 0x103c, 0x104b, "HP Diva Serial [GSP] Multiport UART - Maestro SP2", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 0, -1, .config_function = puc_config_diva }, { 0x103c, 0x1048, 0x103c, 0x1223, "HP Diva Serial [GSP] Multiport UART - Superdome Console", DEFAULT_RCLK, PUC_PORT_3S, 0x10, 0, -1, .config_function = puc_config_diva }, { 0x103c, 0x1048, 0x103c, 0x1226, "HP Diva Serial [GSP] Multiport UART - Keystone SP2", DEFAULT_RCLK, PUC_PORT_3S, 0x10, 0, -1, .config_function = puc_config_diva }, { 0x103c, 0x1048, 0x103c, 0x1282, "HP Diva Serial [GSP] Multiport UART - Everest SP2", DEFAULT_RCLK, PUC_PORT_3S, 0x10, 0, -1, .config_function = puc_config_diva }, { 0x10b5, 0x1076, 0x10b5, 0x1076, "VScom PCI-800", DEFAULT_RCLK * 8, PUC_PORT_8S, 0x18, 0, 8, }, { 0x10b5, 0x1077, 0x10b5, 0x1077, "VScom PCI-400", DEFAULT_RCLK * 8, PUC_PORT_4S, 0x18, 0, 8, }, { 0x10b5, 0x1103, 0x10b5, 0x1103, "VScom PCI-200", DEFAULT_RCLK * 8, PUC_PORT_2S, 0x18, 4, 0, }, /* * Boca Research Turbo Serial 658 (8 serial port) card. * Appears to be the same as Chase Research PLC PCI-FAST8 * and Perle PCI-FAST8 Multi-Port serial cards. */ { 0x10b5, 0x9050, 0x12e0, 0x0021, "Boca Research Turbo Serial 658", DEFAULT_RCLK * 4, PUC_PORT_8S, 0x18, 0, 8, }, { 0x10b5, 0x9050, 0x12e0, 0x0031, "Boca Research Turbo Serial 654", DEFAULT_RCLK * 4, PUC_PORT_4S, 0x18, 0, 8, }, /* * Dolphin Peripherals 4035 (dual serial port) card. PLX 9050, with * a seemingly-lame EEPROM setup that puts the Dolphin IDs * into the subsystem fields, and claims that it's a * network/misc (0x02/0x80) device. */ { 0x10b5, 0x9050, 0xd84d, 0x6808, "Dolphin Peripherals 4035", DEFAULT_RCLK, PUC_PORT_2S, 0x18, 4, 0, }, /* * Dolphin Peripherals 4014 (dual parallel port) card. PLX 9050, with * a seemingly-lame EEPROM setup that puts the Dolphin IDs * into the subsystem fields, and claims that it's a * network/misc (0x02/0x80) device. */ { 0x10b5, 0x9050, 0xd84d, 0x6810, "Dolphin Peripherals 4014", 0, PUC_PORT_2P, 0x20, 4, 0, }, { 0x10e8, 0x818e, 0xffff, 0, "Applied Micro Circuits 8 Port UART", DEFAULT_RCLK, PUC_PORT_8S, 0x14, -1, -1, .config_function = puc_config_amc }, { 0x11fe, 0x8010, 0xffff, 0, "Comtrol RocketPort 550/8 RJ11 part A", DEFAULT_RCLK * 4, PUC_PORT_4S, 0x10, 0, 8, }, { 0x11fe, 0x8011, 0xffff, 0, "Comtrol RocketPort 550/8 RJ11 part B", DEFAULT_RCLK * 4, PUC_PORT_4S, 0x10, 0, 8, }, { 0x11fe, 0x8012, 0xffff, 0, "Comtrol RocketPort 550/8 Octa part A", DEFAULT_RCLK * 4, PUC_PORT_4S, 0x10, 0, 8, }, { 0x11fe, 0x8013, 0xffff, 0, "Comtrol RocketPort 550/8 Octa part B", DEFAULT_RCLK * 4, PUC_PORT_4S, 0x10, 0, 8, }, { 0x11fe, 0x8014, 0xffff, 0, "Comtrol RocketPort 550/4 RJ45", DEFAULT_RCLK * 4, PUC_PORT_4S, 0x10, 0, 8, }, { 0x11fe, 0x8015, 0xffff, 0, "Comtrol RocketPort 550/Quad", DEFAULT_RCLK * 4, PUC_PORT_4S, 0x10, 0, 8, }, { 0x11fe, 0x8016, 0xffff, 0, "Comtrol RocketPort 550/16 part A", DEFAULT_RCLK * 4, PUC_PORT_4S, 0x10, 0, 8, }, { 0x11fe, 0x8017, 0xffff, 0, "Comtrol RocketPort 550/16 part B", DEFAULT_RCLK * 4, PUC_PORT_12S, 0x10, 0, 8, }, { 0x11fe, 0x8018, 0xffff, 0, "Comtrol RocketPort 550/8 part A", DEFAULT_RCLK * 4, PUC_PORT_4S, 0x10, 0, 8, }, { 0x11fe, 0x8019, 0xffff, 0, "Comtrol RocketPort 550/8 part B", DEFAULT_RCLK * 4, PUC_PORT_4S, 0x10, 0, 8, }, /* * IBM SurePOS 300 Series (481033H) serial ports * Details can be found on the IBM RSS websites */ { 0x1014, 0x0297, 0xffff, 0, "IBM SurePOS 300 Series (481033H) serial ports", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 4, 0 }, /* * SIIG Boards. * * SIIG provides documentation for their boards at: * */ { 0x131f, 0x1010, 0xffff, 0, "SIIG Cyber I/O PCI 16C550 (10x family)", DEFAULT_RCLK, PUC_PORT_1S1P, 0x18, 4, 0, }, { 0x131f, 0x1011, 0xffff, 0, "SIIG Cyber I/O PCI 16C650 (10x family)", DEFAULT_RCLK, PUC_PORT_1S1P, 0x18, 4, 0, }, { 0x131f, 0x1012, 0xffff, 0, "SIIG Cyber I/O PCI 16C850 (10x family)", DEFAULT_RCLK, PUC_PORT_1S1P, 0x18, 4, 0, }, { 0x131f, 0x1021, 0xffff, 0, "SIIG Cyber Parallel Dual PCI (10x family)", 0, PUC_PORT_2P, 0x18, 8, 0, }, { 0x131f, 0x1030, 0xffff, 0, "SIIG Cyber Serial Dual PCI 16C550 (10x family)", DEFAULT_RCLK, PUC_PORT_2S, 0x18, 4, 0, }, { 0x131f, 0x1031, 0xffff, 0, "SIIG Cyber Serial Dual PCI 16C650 (10x family)", DEFAULT_RCLK, PUC_PORT_2S, 0x18, 4, 0, }, { 0x131f, 0x1032, 0xffff, 0, "SIIG Cyber Serial Dual PCI 16C850 (10x family)", DEFAULT_RCLK, PUC_PORT_2S, 0x18, 4, 0, }, { 0x131f, 0x1034, 0xffff, 0, /* XXX really? */ "SIIG Cyber 2S1P PCI 16C550 (10x family)", DEFAULT_RCLK, PUC_PORT_2S1P, 0x18, 4, 0, }, { 0x131f, 0x1035, 0xffff, 0, /* XXX really? */ "SIIG Cyber 2S1P PCI 16C650 (10x family)", DEFAULT_RCLK, PUC_PORT_2S1P, 0x18, 4, 0, }, { 0x131f, 0x1036, 0xffff, 0, /* XXX really? */ "SIIG Cyber 2S1P PCI 16C850 (10x family)", DEFAULT_RCLK, PUC_PORT_2S1P, 0x18, 4, 0, }, { 0x131f, 0x1050, 0xffff, 0, "SIIG Cyber 4S PCI 16C550 (10x family)", DEFAULT_RCLK, PUC_PORT_4S, 0x18, 4, 0, }, { 0x131f, 0x1051, 0xffff, 0, "SIIG Cyber 4S PCI 16C650 (10x family)", DEFAULT_RCLK, PUC_PORT_4S, 0x18, 4, 0, }, { 0x131f, 0x1052, 0xffff, 0, "SIIG Cyber 4S PCI 16C850 (10x family)", DEFAULT_RCLK, PUC_PORT_4S, 0x18, 4, 0, }, { 0x131f, 0x2010, 0xffff, 0, "SIIG Cyber I/O PCI 16C550 (20x family)", DEFAULT_RCLK, PUC_PORT_1S1P, 0x10, 4, 0, }, { 0x131f, 0x2011, 0xffff, 0, "SIIG Cyber I/O PCI 16C650 (20x family)", DEFAULT_RCLK, PUC_PORT_1S1P, 0x10, 4, 0, }, { 0x131f, 0x2012, 0xffff, 0, "SIIG Cyber I/O PCI 16C850 (20x family)", DEFAULT_RCLK, PUC_PORT_1S1P, 0x10, 4, 0, }, { 0x131f, 0x2021, 0xffff, 0, "SIIG Cyber Parallel Dual PCI (20x family)", 0, PUC_PORT_2P, 0x10, 8, 0, }, { 0x131f, 0x2030, 0xffff, 0, "SIIG Cyber Serial Dual PCI 16C550 (20x family)", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 4, 0, }, { 0x131f, 0x2031, 0xffff, 0, "SIIG Cyber Serial Dual PCI 16C650 (20x family)", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 4, 0, }, { 0x131f, 0x2032, 0xffff, 0, "SIIG Cyber Serial Dual PCI 16C850 (20x family)", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 4, 0, }, { 0x131f, 0x2040, 0xffff, 0, "SIIG Cyber 2P1S PCI 16C550 (20x family)", DEFAULT_RCLK, PUC_PORT_1S2P, 0x10, -1, 0, .config_function = puc_config_siig }, { 0x131f, 0x2041, 0xffff, 0, "SIIG Cyber 2P1S PCI 16C650 (20x family)", DEFAULT_RCLK, PUC_PORT_1S2P, 0x10, -1, 0, .config_function = puc_config_siig }, { 0x131f, 0x2042, 0xffff, 0, "SIIG Cyber 2P1S PCI 16C850 (20x family)", DEFAULT_RCLK, PUC_PORT_1S2P, 0x10, -1, 0, .config_function = puc_config_siig }, { 0x131f, 0x2050, 0xffff, 0, "SIIG Cyber 4S PCI 16C550 (20x family)", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 4, 0, }, { 0x131f, 0x2051, 0xffff, 0, "SIIG Cyber 4S PCI 16C650 (20x family)", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 4, 0, }, { 0x131f, 0x2052, 0xffff, 0, "SIIG Cyber 4S PCI 16C850 (20x family)", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 4, 0, }, { 0x131f, 0x2060, 0xffff, 0, "SIIG Cyber 2S1P PCI 16C550 (20x family)", DEFAULT_RCLK, PUC_PORT_2S1P, 0x10, 4, 0, }, { 0x131f, 0x2061, 0xffff, 0, "SIIG Cyber 2S1P PCI 16C650 (20x family)", DEFAULT_RCLK, PUC_PORT_2S1P, 0x10, 4, 0, }, { 0x131f, 0x2062, 0xffff, 0, "SIIG Cyber 2S1P PCI 16C850 (20x family)", DEFAULT_RCLK, PUC_PORT_2S1P, 0x10, 4, 0, }, { 0x131f, 0x2081, 0xffff, 0, "SIIG PS8000 8S PCI 16C650 (20x family)", DEFAULT_RCLK, PUC_PORT_8S, 0x10, -1, -1, .config_function = puc_config_siig }, { 0x135c, 0x0010, 0xffff, 0, "Quatech QSC-100", -3, /* max 8x clock rate */ PUC_PORT_4S, 0x14, 0, 8, .config_function = puc_config_quatech }, { 0x135c, 0x0020, 0xffff, 0, "Quatech DSC-100", -1, /* max 2x clock rate */ PUC_PORT_2S, 0x14, 0, 8, .config_function = puc_config_quatech }, { 0x135c, 0x0030, 0xffff, 0, "Quatech DSC-200/300", -1, /* max 2x clock rate */ PUC_PORT_2S, 0x14, 0, 8, .config_function = puc_config_quatech }, { 0x135c, 0x0040, 0xffff, 0, "Quatech QSC-200/300", -3, /* max 8x clock rate */ PUC_PORT_4S, 0x14, 0, 8, .config_function = puc_config_quatech }, { 0x135c, 0x0050, 0xffff, 0, "Quatech ESC-100D", -3, /* max 8x clock rate */ PUC_PORT_8S, 0x14, 0, 8, .config_function = puc_config_quatech }, { 0x135c, 0x0060, 0xffff, 0, "Quatech ESC-100M", -3, /* max 8x clock rate */ PUC_PORT_8S, 0x14, 0, 8, .config_function = puc_config_quatech }, { 0x135c, 0x0170, 0xffff, 0, "Quatech QSCLP-100", -1, /* max 2x clock rate */ PUC_PORT_4S, 0x18, 0, 8, .config_function = puc_config_quatech }, { 0x135c, 0x0180, 0xffff, 0, "Quatech DSCLP-100", -1, /* max 3x clock rate */ PUC_PORT_2S, 0x18, 0, 8, .config_function = puc_config_quatech }, { 0x135c, 0x01b0, 0xffff, 0, "Quatech DSCLP-200/300", -1, /* max 2x clock rate */ PUC_PORT_2S, 0x18, 0, 8, .config_function = puc_config_quatech }, { 0x135c, 0x01e0, 0xffff, 0, "Quatech ESCLP-100", -3, /* max 8x clock rate */ PUC_PORT_8S, 0x10, 0, 8, .config_function = puc_config_quatech }, { 0x1393, 0x1040, 0xffff, 0, "Moxa Technologies, Smartio C104H/PCI", DEFAULT_RCLK * 8, PUC_PORT_4S, 0x18, 0, 8, }, { 0x1393, 0x1041, 0xffff, 0, "Moxa Technologies, Smartio CP-104UL/PCI", DEFAULT_RCLK * 8, PUC_PORT_4S, 0x18, 0, 8, }, { 0x1393, 0x1043, 0xffff, 0, "Moxa Technologies, Smartio CP-104EL/PCIe", DEFAULT_RCLK * 8, PUC_PORT_4S, 0x18, 0, 8, }, { 0x1393, 0x1141, 0xffff, 0, "Moxa Technologies, Industio CP-114", DEFAULT_RCLK * 8, PUC_PORT_4S, 0x18, 0, 8, }, { 0x1393, 0x1680, 0xffff, 0, "Moxa Technologies, C168H/PCI", DEFAULT_RCLK * 8, PUC_PORT_8S, 0x18, 0, 8, }, { 0x1393, 0x1681, 0xffff, 0, "Moxa Technologies, C168U/PCI", DEFAULT_RCLK * 8, PUC_PORT_8S, 0x18, 0, 8, }, { 0x1393, 0x1682, 0xffff, 0, "Moxa Technologies, CP-168EL/PCIe", DEFAULT_RCLK * 8, PUC_PORT_8S, 0x18, 0, 8, }, { 0x13a8, 0x0152, 0xffff, 0, "Exar XR17C/D152", DEFAULT_RCLK * 8, PUC_PORT_2S, 0x10, 0, -1, .config_function = puc_config_exar }, { 0x13a8, 0x0154, 0xffff, 0, "Exar XR17C154", DEFAULT_RCLK * 8, PUC_PORT_4S, 0x10, 0, -1, .config_function = puc_config_exar }, { 0x13a8, 0x0158, 0xffff, 0, "Exar XR17C158", DEFAULT_RCLK * 8, PUC_PORT_8S, 0x10, 0, -1, .config_function = puc_config_exar }, { 0x13a8, 0x0258, 0xffff, 0, "Exar XR17V258IV", DEFAULT_RCLK * 8, PUC_PORT_8S, 0x10, 0, -1, }, { 0x1407, 0x0100, 0xffff, 0, "Lava Computers Dual Serial", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 4, 0, }, { 0x1407, 0x0101, 0xffff, 0, "Lava Computers Quatro A", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 4, 0, }, { 0x1407, 0x0102, 0xffff, 0, "Lava Computers Quatro B", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 4, 0, }, { 0x1407, 0x0120, 0xffff, 0, "Lava Computers Quattro-PCI A", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 4, 0, }, { 0x1407, 0x0121, 0xffff, 0, "Lava Computers Quattro-PCI B", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 4, 0, }, { 0x1407, 0x0180, 0xffff, 0, "Lava Computers Octo A", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 4, 0, }, { 0x1407, 0x0181, 0xffff, 0, "Lava Computers Octo B", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 4, 0, }, { 0x1409, 0x7268, 0xffff, 0, "Sunix SUN1888", 0, PUC_PORT_2P, 0x10, 0, 8, }, { 0x1409, 0x7168, 0xffff, 0, NULL, DEFAULT_RCLK * 8, PUC_PORT_NONSTANDARD, 0x10, -1, -1, .config_function = puc_config_timedia }, /* * Boards with an Oxford Semiconductor chip. * * Oxford Semiconductor provides documentation for their chip at: * * * As sold by Kouwell . * I/O Flex PCI I/O Card Model-223 with 4 serial and 1 parallel ports. */ { 0x1415, 0x9501, 0x131f, 0x2050, "SIIG Cyber 4 PCI 16550", DEFAULT_RCLK * 10, PUC_PORT_4S, 0x10, 0, 8, }, { 0x1415, 0x9501, 0x131f, 0x2051, "SIIG Cyber 4S PCI 16C650 (20x family)", DEFAULT_RCLK * 10, PUC_PORT_4S, 0x10, 0, 8, }, { 0x1415, 0x9501, 0x131f, 0x2052, "SIIG Quartet Serial 850", DEFAULT_RCLK * 10, PUC_PORT_4S, 0x10, 0, 8, }, { 0x1415, 0x9501, 0x14db, 0x2150, "Kuroutoshikou SERIAL4P-LPPCI2", DEFAULT_RCLK * 10, PUC_PORT_4S, 0x10, 0, 8, }, { 0x1415, 0x9501, 0xffff, 0, "Oxford Semiconductor OX16PCI954 UARTs", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 0, 8, }, { 0x1415, 0x950a, 0x131f, 0x2030, "SIIG Cyber 2S PCIe", DEFAULT_RCLK * 10, PUC_PORT_2S, 0x10, 0, 8, }, { 0x1415, 0x950a, 0xffff, 0, "Oxford Semiconductor OX16PCI954 UARTs", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 0, 8, }, { 0x1415, 0x9511, 0xffff, 0, "Oxford Semiconductor OX9160/OX16PCI954 UARTs (function 1)", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 0, 8, }, { 0x1415, 0x9521, 0xffff, 0, "Oxford Semiconductor OX16PCI952 UARTs", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 4, 0, }, { 0x1415, 0x9538, 0xffff, 0, "Oxford Semiconductor OX16PCI958 UARTs", DEFAULT_RCLK * 10, PUC_PORT_8S, 0x18, 0, 8, }, /* * Perle boards use Oxford Semiconductor chips, but they store the * Oxford Semiconductor device ID as a subvendor device ID and use * their own device IDs. */ { 0x155f, 0x0331, 0xffff, 0, "Perle Speed4 LE", DEFAULT_RCLK * 8, PUC_PORT_4S, 0x10, 0, 8, }, /* * Oxford Semiconductor PCI Express Expresso family * * Found in many 'native' PCI Express serial boards such as: * * eMegatech MP954ER4 (4 port) and MP958ER8 (8 port) * * * Lindy 51189 (4 port) * * * StarTech.com PEX4S952 (4 port) and PEX8S952 (8 port) * */ { 0x1415, 0xc158, 0xffff, 0, "Oxford Semiconductor OXPCIe952 UARTs", DEFAULT_RCLK * 0x22, PUC_PORT_NONSTANDARD, 0x10, 0, -1, .config_function = puc_config_oxford_pcie }, { 0x1415, 0xc15d, 0xffff, 0, "Oxford Semiconductor OXPCIe952 UARTs (function 1)", DEFAULT_RCLK * 0x22, PUC_PORT_NONSTANDARD, 0x10, 0, -1, .config_function = puc_config_oxford_pcie }, { 0x1415, 0xc208, 0xffff, 0, "Oxford Semiconductor OXPCIe954 UARTs", DEFAULT_RCLK * 0x22, PUC_PORT_NONSTANDARD, 0x10, 0, -1, .config_function = puc_config_oxford_pcie }, { 0x1415, 0xc20d, 0xffff, 0, "Oxford Semiconductor OXPCIe954 UARTs (function 1)", DEFAULT_RCLK * 0x22, PUC_PORT_NONSTANDARD, 0x10, 0, -1, .config_function = puc_config_oxford_pcie }, { 0x1415, 0xc308, 0xffff, 0, "Oxford Semiconductor OXPCIe958 UARTs", DEFAULT_RCLK * 0x22, PUC_PORT_NONSTANDARD, 0x10, 0, -1, .config_function = puc_config_oxford_pcie }, { 0x1415, 0xc30d, 0xffff, 0, "Oxford Semiconductor OXPCIe958 UARTs (function 1)", DEFAULT_RCLK * 0x22, PUC_PORT_NONSTANDARD, 0x10, 0, -1, .config_function = puc_config_oxford_pcie }, { 0x14d2, 0x8010, 0xffff, 0, "VScom PCI-100L", DEFAULT_RCLK * 8, PUC_PORT_1S, 0x14, 0, 0, }, { 0x14d2, 0x8020, 0xffff, 0, "VScom PCI-200L", DEFAULT_RCLK * 8, PUC_PORT_2S, 0x14, 4, 0, }, { 0x14d2, 0x8028, 0xffff, 0, "VScom 200Li", DEFAULT_RCLK, PUC_PORT_2S, 0x20, 0, 8, }, /* * VScom (Titan?) PCI-800L. More modern variant of the * PCI-800. Uses 6 discrete 16550 UARTs, plus another * two of them obviously implemented as macro cells in * the ASIC. This causes the weird port access pattern * below, where two of the IO port ranges each access * one of the ASIC UARTs, and a block of IO addresses * access the external UARTs. */ { 0x14d2, 0x8080, 0xffff, 0, "Titan VScom PCI-800L", DEFAULT_RCLK * 8, PUC_PORT_8S, 0x14, -1, -1, .config_function = puc_config_titan }, /* * VScom PCI-800H. Uses 8 16950 UART, behind a PCI chips that offers * 4 com port on PCI device 0 and 4 on PCI device 1. PCI device 0 has * device ID 3 and PCI device 1 device ID 4. */ { 0x14d2, 0xa003, 0xffff, 0, "Titan PCI-800H", DEFAULT_RCLK * 8, PUC_PORT_4S, 0x10, 0, 8, }, { 0x14d2, 0xa004, 0xffff, 0, "Titan PCI-800H", DEFAULT_RCLK * 8, PUC_PORT_4S, 0x10, 0, 8, }, { 0x14d2, 0xa005, 0xffff, 0, "Titan PCI-200H", DEFAULT_RCLK * 8, PUC_PORT_2S, 0x10, 0, 8, }, { 0x14d2, 0xe020, 0xffff, 0, "Titan VScom PCI-200HV2", DEFAULT_RCLK * 8, PUC_PORT_2S, 0x10, 4, 0, }, { 0x14db, 0x2130, 0xffff, 0, "Avlab Technology, PCI IO 2S", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 4, 0, }, { 0x14db, 0x2150, 0xffff, 0, "Avlab Low Profile PCI 4 Serial", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 4, 0, }, { 0x14db, 0x2152, 0xffff, 0, "Avlab Low Profile PCI 4 Serial", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 4, 0, }, { 0x1592, 0x0781, 0xffff, 0, "Syba Tech Ltd. PCI-4S2P-550-ECP", DEFAULT_RCLK, PUC_PORT_4S1P, 0x10, 0, -1, .config_function = puc_config_syba }, { 0x6666, 0x0001, 0xffff, 0, "Decision Computer Inc, PCCOM 4-port serial", DEFAULT_RCLK, PUC_PORT_4S, 0x1c, 0, 8, }, { 0x6666, 0x0002, 0xffff, 0, "Decision Computer Inc, PCCOM 8-port serial", DEFAULT_RCLK, PUC_PORT_8S, 0x1c, 0, 8, }, { 0x6666, 0x0004, 0xffff, 0, "PCCOM dual port RS232/422/485", DEFAULT_RCLK, PUC_PORT_2S, 0x1c, 0, 8, }, { 0x9710, 0x9815, 0xffff, 0, "NetMos NM9815 Dual 1284 Printer port", 0, PUC_PORT_2P, 0x10, 8, 0, }, /* * This is more specific than the generic NM9835 entry that follows, and * is placed here to _prevent_ puc from claiming this single port card. * * uart(4) will claim this device. */ { 0x9710, 0x9835, 0x1000, 1, "NetMos NM9835 based 1-port serial", DEFAULT_RCLK, PUC_PORT_1S, 0x10, 4, 0, }, { 0x9710, 0x9835, 0x1000, 2, "NetMos NM9835 based 2-port serial", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 4, 0, }, { 0x9710, 0x9835, 0xffff, 0, "NetMos NM9835 Dual UART and 1284 Printer port", DEFAULT_RCLK, PUC_PORT_2S1P, 0x10, 4, 0, }, { 0x9710, 0x9845, 0x1000, 0x0006, "NetMos NM9845 6 Port UART", DEFAULT_RCLK, PUC_PORT_6S, 0x10, 4, 0, }, { 0x9710, 0x9845, 0xffff, 0, "NetMos NM9845 Quad UART and 1284 Printer port", DEFAULT_RCLK, PUC_PORT_4S1P, 0x10, 4, 0, }, { 0x9710, 0x9865, 0xa000, 0x3002, "NetMos NM9865 Dual UART", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 4, 0, }, { 0x9710, 0x9865, 0xa000, 0x3003, "NetMos NM9865 Triple UART", DEFAULT_RCLK, PUC_PORT_3S, 0x10, 4, 0, }, { 0x9710, 0x9865, 0xa000, 0x3004, "NetMos NM9865 Quad UART", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 4, 0,0 }, { 0x9710, 0x9865, 0xa000, 0x3011, "NetMos NM9865 Single UART and 1284 Printer port", DEFAULT_RCLK, PUC_PORT_1S1P, 0x10, 4, 0, }, { 0x9710, 0x9865, 0xa000, 0x3012, "NetMos NM9865 Dual UART and 1284 Printer port", DEFAULT_RCLK, PUC_PORT_2S1P, 0x10, 4, 0, }, { 0x9710, 0x9865, 0xa000, 0x3020, "NetMos NM9865 Dual 1284 Printer port", DEFAULT_RCLK, PUC_PORT_2P, 0x10, 4, 0, }, { 0xb00c, 0x021c, 0xffff, 0, "IC Book Labs Gunboat x4 Lite", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 0, 8, .config_function = puc_config_icbook }, { 0xb00c, 0x031c, 0xffff, 0, "IC Book Labs Gunboat x4 Pro", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 0, 8, .config_function = puc_config_icbook }, { 0xb00c, 0x041c, 0xffff, 0, "IC Book Labs Ironclad x8 Lite", DEFAULT_RCLK, PUC_PORT_8S, 0x10, 0, 8, .config_function = puc_config_icbook }, { 0xb00c, 0x051c, 0xffff, 0, "IC Book Labs Ironclad x8 Pro", DEFAULT_RCLK, PUC_PORT_8S, 0x10, 0, 8, .config_function = puc_config_icbook }, { 0xb00c, 0x081c, 0xffff, 0, "IC Book Labs Dreadnought x16 Pro", DEFAULT_RCLK * 8, PUC_PORT_16S, 0x10, 0, 8, .config_function = puc_config_icbook }, { 0xb00c, 0x091c, 0xffff, 0, "IC Book Labs Dreadnought x16 Lite", DEFAULT_RCLK, PUC_PORT_16S, 0x10, 0, 8, .config_function = puc_config_icbook }, { 0xb00c, 0x0a1c, 0xffff, 0, "IC Book Labs Gunboat x2 Low Profile", DEFAULT_RCLK, PUC_PORT_2S, 0x10, 0, 8, }, { 0xb00c, 0x0b1c, 0xffff, 0, "IC Book Labs Gunboat x4 Low Profile", DEFAULT_RCLK, PUC_PORT_4S, 0x10, 0, 8, .config_function = puc_config_icbook }, { 0xffff, 0, 0xffff, 0, NULL, 0 } }; static int puc_config_amc(struct puc_softc *sc, enum puc_cfg_cmd cmd, int port, intptr_t *res) { switch (cmd) { case PUC_CFG_GET_OFS: *res = 8 * (port & 1); return (0); case PUC_CFG_GET_RID: *res = 0x14 + (port >> 1) * 4; return (0); default: break; } return (ENXIO); } static int puc_config_diva(struct puc_softc *sc, enum puc_cfg_cmd cmd, int port, intptr_t *res) { const struct puc_cfg *cfg = sc->sc_cfg; if (cmd == PUC_CFG_GET_OFS) { if (cfg->subdevice == 0x1282) /* Everest SP */ port <<= 1; else if (cfg->subdevice == 0x104b) /* Maestro SP2 */ port = (port == 3) ? 4 : port; *res = port * 8 + ((port > 2) ? 0x18 : 0); return (0); } return (ENXIO); } static int puc_config_exar(struct puc_softc *sc, enum puc_cfg_cmd cmd, int port, intptr_t *res) { if (cmd == PUC_CFG_GET_OFS) { *res = port * 0x200; return (0); } return (ENXIO); } static int puc_config_icbook(struct puc_softc *sc, enum puc_cfg_cmd cmd, int port, intptr_t *res) { if (cmd == PUC_CFG_GET_ILR) { *res = PUC_ILR_DIGI; return (0); } return (ENXIO); } static int puc_config_quatech(struct puc_softc *sc, enum puc_cfg_cmd cmd, int port, intptr_t *res) { const struct puc_cfg *cfg = sc->sc_cfg; struct puc_bar *bar; uint8_t v0, v1; switch (cmd) { case PUC_CFG_SETUP: /* * Check if the scratchpad register is enabled or if the * interrupt status and options registers are active. */ bar = puc_get_bar(sc, cfg->rid); if (bar == NULL) return (ENXIO); /* Set DLAB in the LCR register of UART 0. */ bus_write_1(bar->b_res, 3, 0x80); /* Write 0 to the SPR register of UART 0. */ bus_write_1(bar->b_res, 7, 0); /* Read back the contents of the SPR register of UART 0. */ v0 = bus_read_1(bar->b_res, 7); /* Write a specific value to the SPR register of UART 0. */ bus_write_1(bar->b_res, 7, 0x80 + -cfg->clock); /* Read back the contents of the SPR register of UART 0. */ v1 = bus_read_1(bar->b_res, 7); /* Clear DLAB in the LCR register of UART 0. */ bus_write_1(bar->b_res, 3, 0); /* Save the two values read-back from the SPR register. */ sc->sc_cfg_data = (v0 << 8) | v1; if (v0 == 0 && v1 == 0x80 + -cfg->clock) { /* * The SPR register echoed the two values written * by us. This means that the SPAD jumper is set. */ device_printf(sc->sc_dev, "warning: extra features " "not usable -- SPAD compatibility enabled\n"); return (0); } if (v0 != 0) { /* * The first value doesn't match. This can only mean * that the SPAD jumper is not set and that a non- * standard fixed clock multiplier jumper is set. */ if (bootverbose) device_printf(sc->sc_dev, "fixed clock rate " "multiplier of %d\n", 1 << v0); if (v0 < -cfg->clock) device_printf(sc->sc_dev, "warning: " "suboptimal fixed clock rate multiplier " "setting\n"); return (0); } /* * The first value matched, but the second didn't. We know * that the SPAD jumper is not set. We also know that the * clock rate multiplier is software controlled *and* that * we just programmed it to the maximum allowed. */ if (bootverbose) device_printf(sc->sc_dev, "clock rate multiplier of " "%d selected\n", 1 << -cfg->clock); return (0); case PUC_CFG_GET_CLOCK: v0 = (sc->sc_cfg_data >> 8) & 0xff; v1 = sc->sc_cfg_data & 0xff; if (v0 == 0 && v1 == 0x80 + -cfg->clock) { /* * XXX With the SPAD jumper applied, there's no * easy way of knowing if there's also a clock * rate multiplier jumper installed. Let's hope * not... */ *res = DEFAULT_RCLK; } else if (v0 == 0) { /* * No clock rate multiplier jumper installed, * so we programmed the board with the maximum * multiplier allowed as given to us in the * clock field of the config record (negated). */ *res = DEFAULT_RCLK << -cfg->clock; } else *res = DEFAULT_RCLK << v0; return (0); case PUC_CFG_GET_ILR: v0 = (sc->sc_cfg_data >> 8) & 0xff; v1 = sc->sc_cfg_data & 0xff; *res = (v0 == 0 && v1 == 0x80 + -cfg->clock) ? PUC_ILR_NONE : PUC_ILR_QUATECH; return (0); default: break; } return (ENXIO); } static int puc_config_syba(struct puc_softc *sc, enum puc_cfg_cmd cmd, int port, intptr_t *res) { static int base[] = { 0x251, 0x3f0, 0 }; const struct puc_cfg *cfg = sc->sc_cfg; struct puc_bar *bar; int efir, idx, ofs; uint8_t v; switch (cmd) { case PUC_CFG_SETUP: bar = puc_get_bar(sc, cfg->rid); if (bar == NULL) return (ENXIO); /* configure both W83877TFs */ bus_write_1(bar->b_res, 0x250, 0x89); bus_write_1(bar->b_res, 0x3f0, 0x87); bus_write_1(bar->b_res, 0x3f0, 0x87); idx = 0; while (base[idx] != 0) { efir = base[idx]; bus_write_1(bar->b_res, efir, 0x09); v = bus_read_1(bar->b_res, efir + 1); if ((v & 0x0f) != 0x0c) return (ENXIO); bus_write_1(bar->b_res, efir, 0x16); v = bus_read_1(bar->b_res, efir + 1); bus_write_1(bar->b_res, efir, 0x16); bus_write_1(bar->b_res, efir + 1, v | 0x04); bus_write_1(bar->b_res, efir, 0x16); bus_write_1(bar->b_res, efir + 1, v & ~0x04); ofs = base[idx] & 0x300; bus_write_1(bar->b_res, efir, 0x23); bus_write_1(bar->b_res, efir + 1, (ofs + 0x78) >> 2); bus_write_1(bar->b_res, efir, 0x24); bus_write_1(bar->b_res, efir + 1, (ofs + 0xf8) >> 2); bus_write_1(bar->b_res, efir, 0x25); bus_write_1(bar->b_res, efir + 1, (ofs + 0xe8) >> 2); bus_write_1(bar->b_res, efir, 0x17); bus_write_1(bar->b_res, efir + 1, 0x03); bus_write_1(bar->b_res, efir, 0x28); bus_write_1(bar->b_res, efir + 1, 0x43); idx++; } bus_write_1(bar->b_res, 0x250, 0xaa); bus_write_1(bar->b_res, 0x3f0, 0xaa); return (0); case PUC_CFG_GET_OFS: switch (port) { case 0: *res = 0x2f8; return (0); case 1: *res = 0x2e8; return (0); case 2: *res = 0x3f8; return (0); case 3: *res = 0x3e8; return (0); case 4: *res = 0x278; return (0); } break; default: break; } return (ENXIO); } static int puc_config_siig(struct puc_softc *sc, enum puc_cfg_cmd cmd, int port, intptr_t *res) { const struct puc_cfg *cfg = sc->sc_cfg; switch (cmd) { case PUC_CFG_GET_OFS: if (cfg->ports == PUC_PORT_8S) { *res = (port > 4) ? 8 * (port - 4) : 0; return (0); } break; case PUC_CFG_GET_RID: if (cfg->ports == PUC_PORT_8S) { *res = 0x10 + ((port > 4) ? 0x10 : 4 * port); return (0); } if (cfg->ports == PUC_PORT_2S1P) { switch (port) { case 0: *res = 0x10; return (0); case 1: *res = 0x14; return (0); case 2: *res = 0x1c; return (0); } } break; default: break; } return (ENXIO); } static int puc_config_timedia(struct puc_softc *sc, enum puc_cfg_cmd cmd, int port, intptr_t *res) { static uint16_t dual[] = { 0x0002, 0x4036, 0x4037, 0x4038, 0x4078, 0x4079, 0x4085, 0x4088, 0x4089, 0x5037, 0x5078, 0x5079, 0x5085, 0x6079, 0x7079, 0x8079, 0x8137, 0x8138, 0x8237, 0x8238, 0x9079, 0x9137, 0x9138, 0x9237, 0x9238, 0xA079, 0xB079, 0xC079, 0xD079, 0 }; static uint16_t quad[] = { 0x4055, 0x4056, 0x4095, 0x4096, 0x5056, 0x8156, 0x8157, 0x8256, 0x8257, 0x9056, 0x9156, 0x9157, 0x9158, 0x9159, 0x9256, 0x9257, 0xA056, 0xA157, 0xA158, 0xA159, 0xB056, 0xB157, 0 }; static uint16_t octa[] = { 0x4065, 0x4066, 0x5065, 0x5066, 0x8166, 0x9066, 0x9166, 0x9167, 0x9168, 0xA066, 0xA167, 0xA168, 0 }; static struct { int ports; uint16_t *ids; } subdevs[] = { { 2, dual }, { 4, quad }, { 8, octa }, { 0, NULL } }; static char desc[64]; int dev, id; uint16_t subdev; switch (cmd) { case PUC_CFG_GET_CLOCK: if (port < 2) *res = DEFAULT_RCLK * 8; else *res = DEFAULT_RCLK; return (0); case PUC_CFG_GET_DESC: snprintf(desc, sizeof(desc), "Timedia technology %d Port Serial", (int)sc->sc_cfg_data); *res = (intptr_t)desc; return (0); case PUC_CFG_GET_NPORTS: subdev = pci_get_subdevice(sc->sc_dev); dev = 0; while (subdevs[dev].ports != 0) { id = 0; while (subdevs[dev].ids[id] != 0) { if (subdev == subdevs[dev].ids[id]) { sc->sc_cfg_data = subdevs[dev].ports; *res = sc->sc_cfg_data; return (0); } id++; } dev++; } return (ENXIO); case PUC_CFG_GET_OFS: *res = (port == 1 || port == 3) ? 8 : 0; return (0); case PUC_CFG_GET_RID: *res = 0x10 + ((port > 3) ? port - 2 : port >> 1) * 4; return (0); case PUC_CFG_GET_TYPE: *res = PUC_TYPE_SERIAL; return (0); default: break; } return (ENXIO); } static int puc_config_oxford_pcie(struct puc_softc *sc, enum puc_cfg_cmd cmd, int port, intptr_t *res) { const struct puc_cfg *cfg = sc->sc_cfg; int idx; struct puc_bar *bar; uint8_t value; switch (cmd) { case PUC_CFG_SETUP: device_printf(sc->sc_dev, "%d UARTs detected\n", sc->sc_nports); /* Set UARTs to enhanced mode */ bar = puc_get_bar(sc, cfg->rid); if (bar == NULL) return (ENXIO); - for (idx = 0; idx < sc->sc_nports; idx++) { - value = bus_read_1(bar->b_res, 0x1000 + (idx << 9) - + 0x92); + value = bus_read_1(bar->b_res, 0x1000 + (idx << 9) + + 0x92); bus_write_1(bar->b_res, 0x1000 + (idx << 9) + 0x92, - value | 0x10); + value | 0x10); } - return (0); case PUC_CFG_GET_LEN: *res = 0x200; return (0); case PUC_CFG_GET_NPORTS: /* * Check if we are being called from puc_bfe_attach() * or puc_bfe_probe(). If puc_bfe_probe(), we cannot * puc_get_bar(), so we return a value of 16. This has cosmetic * side-effects at worst; in PUC_CFG_GET_DESC, * (int)sc->sc_cfg_data will not contain the true number of * ports in PUC_CFG_GET_DESC, but we are not implementing that * call for this device family anyway. * * The check is for initialisation of sc->sc_bar[idx], which is * only done in puc_bfe_attach(). */ idx = 0; do { if (sc->sc_bar[idx++].b_rid != -1) { sc->sc_cfg_data = 16; *res = sc->sc_cfg_data; return (0); } } while (idx < PUC_PCI_BARS); bar = puc_get_bar(sc, cfg->rid); if (bar == NULL) return (ENXIO); value = bus_read_1(bar->b_res, 0x04); if (value == 0) return (ENXIO); sc->sc_cfg_data = value; *res = sc->sc_cfg_data; return (0); case PUC_CFG_GET_OFS: *res = 0x1000 + (port << 9); return (0); case PUC_CFG_GET_TYPE: *res = PUC_TYPE_SERIAL; return (0); default: break; } return (ENXIO); } static int puc_config_titan(struct puc_softc *sc, enum puc_cfg_cmd cmd, int port, intptr_t *res) { switch (cmd) { case PUC_CFG_GET_OFS: *res = (port < 3) ? 0 : (port - 2) << 3; return (0); case PUC_CFG_GET_RID: *res = 0x14 + ((port >= 2) ? 0x0c : port << 2); return (0); default: break; } return (ENXIO); } Index: projects/largeSMP/sys/fs/nfs/nfs_var.h =================================================================== --- projects/largeSMP/sys/fs/nfs/nfs_var.h (revision 222782) +++ projects/largeSMP/sys/fs/nfs/nfs_var.h (revision 222783) @@ -1,600 +1,601 @@ /*- * Copyright (c) 1989, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Rick Macklem at The University of Guelph. * * 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. * * $FreeBSD$ */ /* * XXX needs and because of typedefs */ struct uio; struct ucred; struct nfscred; NFSPROC_T; struct buf; struct sockaddr_in; struct nfs_dlmount; struct file; struct nfsmount; struct socket; struct nfsreq; struct nfssockreq; struct vattr; struct nameidata; struct nfsnode; struct nfsfh; struct sillyrename; struct componentname; struct nfsd_srvargs; struct nfsrv_descript; struct nfs_fattr; union nethostaddr; struct nfsstate; struct nfslock; struct nfsclient; struct nfslockconflict; struct nfsd_idargs; struct nfsd_clid; struct nfsusrgrp; struct nfsclowner; struct nfsclopen; struct nfsclopenhead; struct nfsclclient; struct nfscllockowner; struct nfscllock; struct nfscldeleg; struct nfsv4lock; struct nfsvattr; struct nfs_vattr; struct NFSSVCARGS; #ifdef __FreeBSD__ NFS_ACCESS_ARGS; NFS_OPEN_ARGS; NFS_GETATTR_ARGS; NFS_LOOKUP_ARGS; NFS_READDIR_ARGS; #endif /* nfs_nfsdstate.c */ int nfsrv_setclient(struct nfsrv_descript *, struct nfsclient **, nfsquad_t *, nfsquad_t *, NFSPROC_T *); int nfsrv_getclient(nfsquad_t, int, struct nfsclient **, nfsquad_t, struct nfsrv_descript *, NFSPROC_T *); int nfsrv_adminrevoke(struct nfsd_clid *, NFSPROC_T *); void nfsrv_dumpclients(struct nfsd_dumpclients *, int); void nfsrv_dumplocks(vnode_t, struct nfsd_dumplocks *, int, NFSPROC_T *); int nfsrv_lockctrl(vnode_t, struct nfsstate **, struct nfslock **, struct nfslockconflict *, nfsquad_t, nfsv4stateid_t *, struct nfsexstuff *, struct nfsrv_descript *, NFSPROC_T *); int nfsrv_openctrl(struct nfsrv_descript *, vnode_t, struct nfsstate **, nfsquad_t, nfsv4stateid_t *, nfsv4stateid_t *, u_int32_t *, struct nfsexstuff *, NFSPROC_T *, u_quad_t); int nfsrv_opencheck(nfsquad_t, nfsv4stateid_t *, struct nfsstate *, vnode_t, struct nfsrv_descript *, NFSPROC_T *, int); int nfsrv_openupdate(vnode_t, struct nfsstate *, nfsquad_t, nfsv4stateid_t *, struct nfsrv_descript *, NFSPROC_T *); int nfsrv_delegupdate(nfsquad_t, nfsv4stateid_t *, vnode_t, int, struct ucred *, NFSPROC_T *); int nfsrv_releaselckown(struct nfsstate *, nfsquad_t, NFSPROC_T *); void nfsrv_zapclient(struct nfsclient *, NFSPROC_T *); int nfssvc_idname(struct nfsd_idargs *); void nfsrv_servertimer(void); int nfsrv_getclientipaddr(struct nfsrv_descript *, struct nfsclient *); void nfsrv_setupstable(NFSPROC_T *); void nfsrv_updatestable(NFSPROC_T *); void nfsrv_writestable(u_char *, int, int, NFSPROC_T *); void nfsrv_throwawayopens(NFSPROC_T *); int nfsrv_checkremove(vnode_t, int, NFSPROC_T *); void nfsd_recalldelegation(vnode_t, NFSPROC_T *); void nfsd_disabledelegation(vnode_t, NFSPROC_T *); int nfsrv_checksetattr(vnode_t, struct nfsrv_descript *, nfsv4stateid_t *, struct nfsvattr *, nfsattrbit_t *, struct nfsexstuff *, NFSPROC_T *); int nfsrv_checkgetattr(struct nfsrv_descript *, vnode_t, struct nfsvattr *, nfsattrbit_t *, struct ucred *, NFSPROC_T *); int nfsrv_nfsuserdport(u_short, NFSPROC_T *); void nfsrv_nfsuserddelport(void); void nfsrv_throwawayallstate(NFSPROC_T *); /* nfs_nfsdserv.c */ int nfsrvd_access(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_getattr(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_setattr(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_lookup(struct nfsrv_descript *, int, vnode_t, vnode_t *, fhandle_t *, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_readlink(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_read(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_write(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_create(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_mknod(struct nfsrv_descript *, int, vnode_t, vnode_t *, fhandle_t *, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_remove(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_rename(struct nfsrv_descript *, int, vnode_t, vnode_t, NFSPROC_T *, struct nfsexstuff *, struct nfsexstuff *); int nfsrvd_link(struct nfsrv_descript *, int, vnode_t, vnode_t, NFSPROC_T *, struct nfsexstuff *, struct nfsexstuff *); int nfsrvd_symlink(struct nfsrv_descript *, int, vnode_t, vnode_t *, fhandle_t *, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_mkdir(struct nfsrv_descript *, int, vnode_t, vnode_t *, fhandle_t *, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_readdir(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_readdirplus(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_commit(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_statfs(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_fsinfo(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_close(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_delegpurge(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_delegreturn(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_getfh(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_lock(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_lockt(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_locku(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_openconfirm(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_opendowngrade(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_renew(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_secinfo(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_setclientid(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_setclientidcfrm(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_verify(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_open(struct nfsrv_descript *, int, vnode_t, vnode_t *, fhandle_t *, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_openattr(struct nfsrv_descript *, int, vnode_t, vnode_t *, fhandle_t *, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_releaselckown(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); int nfsrvd_pathconf(struct nfsrv_descript *, int, vnode_t, NFSPROC_T *, struct nfsexstuff *); /* nfs_nfsdsocket.c */ void nfsrvd_rephead(struct nfsrv_descript *); void nfsrvd_dorpc(struct nfsrv_descript *, int, NFSPROC_T *); /* nfs_nfsdcache.c */ void nfsrvd_initcache(void); int nfsrvd_getcache(struct nfsrv_descript *, struct socket *); struct nfsrvcache *nfsrvd_updatecache(struct nfsrv_descript *, struct socket *); void nfsrvd_sentcache(struct nfsrvcache *, struct socket *, int); void nfsrvd_cleancache(void); void nfsrvd_refcache(struct nfsrvcache *); void nfsrvd_derefcache(struct nfsrvcache *); void nfsrvd_delcache(struct nfsrvcache *); /* nfs_commonsubs.c */ void newnfs_init(void); int nfsaddr_match(int, union nethostaddr *, NFSSOCKADDR_T); int nfsaddr2_match(NFSSOCKADDR_T, NFSSOCKADDR_T); int nfsm_strtom(struct nfsrv_descript *, const char *, int); int nfsm_mbufuio(struct nfsrv_descript *, struct uio *, int); int nfsm_fhtom(struct nfsrv_descript *, u_int8_t *, int, int); int nfsm_advance(struct nfsrv_descript *, int, int); void *nfsm_dissct(struct nfsrv_descript *, int); void newnfs_trimleading(struct nfsrv_descript *); void newnfs_trimtrailing(struct nfsrv_descript *, mbuf_t, caddr_t); void newnfs_copycred(struct nfscred *, struct ucred *); void newnfs_copyincred(struct ucred *, struct nfscred *); int nfsrv_dissectacl(struct nfsrv_descript *, NFSACL_T *, int *, int *, NFSPROC_T *); int nfsrv_getattrbits(struct nfsrv_descript *, nfsattrbit_t *, int *, int *); int nfsv4_loadattr(struct nfsrv_descript *, vnode_t, struct nfsvattr *, struct nfsfh **, fhandle_t *, int, struct nfsv3_pathconf *, struct statfs *, struct nfsstatfs *, struct nfsfsinfo *, NFSACL_T *, int, int *, u_int32_t *, u_int32_t *, NFSPROC_T *, struct ucred *); int nfsv4_lock(struct nfsv4lock *, int, int *, void *, struct mount *); void nfsv4_unlock(struct nfsv4lock *, int); void nfsv4_relref(struct nfsv4lock *); void nfsv4_getref(struct nfsv4lock *, int *, void *, struct mount *); int nfsv4_getref_nonblock(struct nfsv4lock *); int nfsv4_testlock(struct nfsv4lock *); int nfsrv_mtostr(struct nfsrv_descript *, char *, int); int nfsrv_checkutf8(u_int8_t *, int); int newnfs_sndlock(int *); void newnfs_sndunlock(int *); /* nfs_clcomsubs.c */ void nfsm_uiombuf(struct nfsrv_descript *, struct uio *, int); void nfscl_reqstart(struct nfsrv_descript *, int, struct nfsmount *, u_int8_t *, int, u_int32_t **); nfsuint64 *nfscl_getcookie(struct nfsnode *, off_t off, int); void nfscl_fillsattr(struct nfsrv_descript *, struct vattr *, vnode_t, int, u_int32_t); u_int8_t *nfscl_getmyip(struct nfsmount *, int *); int nfsm_getfh(struct nfsrv_descript *, struct nfsfh **); int nfscl_mtofh(struct nfsrv_descript *, struct nfsfh **, struct nfsvattr *, int *); int nfscl_postop_attr(struct nfsrv_descript *, struct nfsvattr *, int *, void *); int nfscl_wcc_data(struct nfsrv_descript *, vnode_t, struct nfsvattr *, int *, int *, void *); int nfsm_loadattr(struct nfsrv_descript *, struct nfsvattr *); int nfscl_request(struct nfsrv_descript *, vnode_t, NFSPROC_T *, struct ucred *, void *); void nfsm_stateidtom(struct nfsrv_descript *, nfsv4stateid_t *, int); /* nfs_nfsdsubs.c */ void nfsd_fhtovp(struct nfsrv_descript *, struct nfsrvfh *, int, vnode_t *, struct nfsexstuff *, mount_t *, int, NFSPROC_T *); int nfsd_excred(struct nfsrv_descript *, struct nfsexstuff *, struct ucred *); int nfsrv_mtofh(struct nfsrv_descript *, struct nfsrvfh *); int nfsrv_putattrbit(struct nfsrv_descript *, nfsattrbit_t *); void nfsrv_wcc(struct nfsrv_descript *, int, struct nfsvattr *, int, struct nfsvattr *); int nfsv4_fillattr(struct nfsrv_descript *, struct mount *, vnode_t, NFSACL_T *, struct vattr *, fhandle_t *, int, nfsattrbit_t *, struct ucred *, NFSPROC_T *, int, int, int, int, uint64_t); void nfsrv_fillattr(struct nfsrv_descript *, struct nfsvattr *); void nfsrv_adj(mbuf_t, int, int); void nfsrv_postopattr(struct nfsrv_descript *, int, struct nfsvattr *); int nfsd_errmap(struct nfsrv_descript *); void nfsv4_uidtostr(uid_t, u_char **, int *, NFSPROC_T *); int nfsv4_strtouid(u_char *, int, uid_t *, NFSPROC_T *); void nfsv4_gidtostr(gid_t, u_char **, int *, NFSPROC_T *); int nfsv4_strtogid(u_char *, int, gid_t *, NFSPROC_T *); int nfsrv_checkuidgid(struct nfsrv_descript *, struct nfsvattr *); void nfsrv_fixattr(struct nfsrv_descript *, vnode_t, struct nfsvattr *, NFSACL_T *, NFSPROC_T *, nfsattrbit_t *, struct nfsexstuff *); int nfsrv_errmoved(int); int nfsrv_putreferralattr(struct nfsrv_descript *, nfsattrbit_t *, struct nfsreferral *, int, int *); int nfsrv_parsename(struct nfsrv_descript *, char *, u_long *, NFSPATHLEN_T *); void nfsd_init(void); int nfsd_checkrootexp(struct nfsrv_descript *); /* nfs_clvfsops.c */ /* nfs_commonport.c */ int nfsrv_checksockseqnum(struct socket *, tcp_seq); int nfsrv_getsockseqnum(struct socket *, tcp_seq *); int nfsrv_getsocksndseq(struct socket *, tcp_seq *, tcp_seq *); int nfsrv_lookupfilename(struct nameidata *, char *, NFSPROC_T *); void nfsrv_object_create(vnode_t, NFSPROC_T *); int nfsrv_mallocmget_limit(void); int nfsvno_v4rootexport(struct nfsrv_descript *); void newnfs_portinit(void); struct ucred *newnfs_getcred(void); void newnfs_setroot(struct ucred *); int nfs_catnap(int, int, const char *); struct nfsreferral *nfsv4root_getreferral(vnode_t, vnode_t, u_int32_t); int nfsrv_atroot(vnode_t, long *); void newnfs_timer(void *); int nfs_supportsnfsv4acls(vnode_t); /* nfs_commonacl.c */ int nfsrv_dissectace(struct nfsrv_descript *, struct acl_entry *, int *, int *, NFSPROC_T *); int nfsrv_buildacl(struct nfsrv_descript *, NFSACL_T *, enum vtype, NFSPROC_T *); int nfsrv_setacl(vnode_t, NFSACL_T *, struct ucred *, NFSPROC_T *); int nfsrv_compareacl(NFSACL_T *, NFSACL_T *); /* nfs_clrpcops.c */ int nfsrpc_null(vnode_t, struct ucred *, NFSPROC_T *); int nfsrpc_access(vnode_t, int, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *); int nfsrpc_accessrpc(vnode_t, u_int32_t, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *, u_int32_t *, void *); int nfsrpc_open(vnode_t, int, struct ucred *, NFSPROC_T *); int nfsrpc_openrpc(struct nfsmount *, vnode_t, u_int8_t *, int, u_int8_t *, int, u_int32_t, struct nfsclopen *, u_int8_t *, int, struct nfscldeleg **, int, u_int32_t, struct ucred *, NFSPROC_T *, int, int); int nfsrpc_opendowngrade(vnode_t, u_int32_t, struct nfsclopen *, struct ucred *, NFSPROC_T *); int nfsrpc_close(vnode_t, int, NFSPROC_T *); int nfsrpc_closerpc(struct nfsrv_descript *, struct nfsmount *, struct nfsclopen *, struct ucred *, NFSPROC_T *, int); int nfsrpc_openconfirm(vnode_t, u_int8_t *, int, struct nfsclopen *, struct ucred *, NFSPROC_T *); int nfsrpc_setclient(struct nfsmount *, struct nfsclclient *, struct ucred *, NFSPROC_T *); int nfsrpc_getattr(vnode_t, struct ucred *, NFSPROC_T *, struct nfsvattr *, void *); int nfsrpc_getattrnovp(struct nfsmount *, u_int8_t *, int, int, struct ucred *, NFSPROC_T *, struct nfsvattr *, u_int64_t *); int nfsrpc_setattr(vnode_t, struct vattr *, NFSACL_T *, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *, void *); int nfsrpc_lookup(vnode_t, char *, int, struct ucred *, NFSPROC_T *, struct nfsvattr *, struct nfsvattr *, struct nfsfh **, int *, int *, void *); int nfsrpc_readlink(vnode_t, struct uio *, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *, void *); int nfsrpc_read(vnode_t, struct uio *, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *, void *); int nfsrpc_write(vnode_t, struct uio *, int *, int *, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *, void *, int); int nfsrpc_mknod(vnode_t, char *, int, struct vattr *, u_int32_t, enum vtype, struct ucred *, NFSPROC_T *, struct nfsvattr *, struct nfsvattr *, struct nfsfh **, int *, int *, void *); int nfsrpc_create(vnode_t, char *, int, struct vattr *, nfsquad_t, int, struct ucred *, NFSPROC_T *, struct nfsvattr *, struct nfsvattr *, struct nfsfh **, int *, int *, void *); int nfsrpc_remove(vnode_t, char *, int, vnode_t, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *, void *); int nfsrpc_rename(vnode_t, vnode_t, char *, int, vnode_t, vnode_t, char *, int, struct ucred *, NFSPROC_T *, struct nfsvattr *, struct nfsvattr *, int *, int *, void *, void *); int nfsrpc_link(vnode_t, vnode_t, char *, int, struct ucred *, NFSPROC_T *, struct nfsvattr *, struct nfsvattr *, int *, int *, void *); int nfsrpc_symlink(vnode_t, char *, int, char *, struct vattr *, struct ucred *, NFSPROC_T *, struct nfsvattr *, struct nfsvattr *, struct nfsfh **, int *, int *, void *); int nfsrpc_mkdir(vnode_t, char *, int, struct vattr *, struct ucred *, NFSPROC_T *, struct nfsvattr *, struct nfsvattr *, struct nfsfh **, int *, int *, void *); int nfsrpc_rmdir(vnode_t, char *, int, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *, void *); int nfsrpc_readdir(vnode_t, struct uio *, nfsuint64 *, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *, int *, void *); int nfsrpc_readdirplus(vnode_t, struct uio *, nfsuint64 *, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *, int *, void *); int nfsrpc_commit(vnode_t, u_quad_t, int, struct ucred *, NFSPROC_T *, u_char *, struct nfsvattr *, int *, void *); int nfsrpc_advlock(vnode_t, off_t, int, struct flock *, int, - struct ucred *, NFSPROC_T *); + struct ucred *, NFSPROC_T *, void *, int); int nfsrpc_lockt(struct nfsrv_descript *, vnode_t, struct nfsclclient *, u_int64_t, u_int64_t, struct flock *, - struct ucred *, NFSPROC_T *); + struct ucred *, NFSPROC_T *, void *, int); int nfsrpc_lock(struct nfsrv_descript *, struct nfsmount *, vnode_t, u_int8_t *, int, struct nfscllockowner *, int, int, u_int64_t, u_int64_t, short, struct ucred *, NFSPROC_T *, int); int nfsrpc_statfs(vnode_t, struct nfsstatfs *, struct nfsfsinfo *, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *, void *); int nfsrpc_fsinfo(vnode_t, struct nfsfsinfo *, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *, void *); int nfsrpc_pathconf(vnode_t, struct nfsv3_pathconf *, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *, void *); int nfsrpc_renew(struct nfsclclient *, struct ucred *, NFSPROC_T *); int nfsrpc_rellockown(struct nfsmount *, struct nfscllockowner *, struct ucred *, NFSPROC_T *); int nfsrpc_getdirpath(struct nfsmount *, u_char *, struct ucred *, NFSPROC_T *); int nfsrpc_delegreturn(struct nfscldeleg *, struct ucred *, struct nfsmount *, NFSPROC_T *, int); int nfsrpc_getacl(vnode_t, struct ucred *, NFSPROC_T *, NFSACL_T *, void *); int nfsrpc_setacl(vnode_t, struct ucred *, NFSPROC_T *, NFSACL_T *, void *); /* nfs_clstate.c */ int nfscl_open(vnode_t, u_int8_t *, int, u_int32_t, int, struct ucred *, NFSPROC_T *, struct nfsclowner **, struct nfsclopen **, int *, int *, int); int nfscl_getstateid(vnode_t, u_int8_t *, int, u_int32_t, struct ucred *, NFSPROC_T *, nfsv4stateid_t *, void **); void nfscl_ownerrelease(struct nfsclowner *, int, int, int); void nfscl_openrelease(struct nfsclopen *, int, int); int nfscl_getcl(vnode_t, struct ucred *, NFSPROC_T *, struct nfsclclient **); struct nfsclclient *nfscl_findcl(struct nfsmount *); void nfscl_clientrelease(struct nfsclclient *); void nfscl_freelock(struct nfscllock *, int); int nfscl_getbytelock(vnode_t, u_int64_t, u_int64_t, short, - struct ucred *, NFSPROC_T *, struct nfsclclient *, int, u_int8_t *, - u_int8_t *, struct nfscllockowner **, int *, int *); + struct ucred *, NFSPROC_T *, struct nfsclclient *, int, void *, int, + u_int8_t *, u_int8_t *, struct nfscllockowner **, int *, int *); int nfscl_relbytelock(vnode_t, u_int64_t, u_int64_t, struct ucred *, NFSPROC_T *, int, struct nfsclclient *, - struct nfscllockowner **, int *); + void *, int, struct nfscllockowner **, int *); int nfscl_checkwritelocked(vnode_t, struct flock *, - struct ucred *, NFSPROC_T *); + struct ucred *, NFSPROC_T *, void *, int); void nfscl_lockrelease(struct nfscllockowner *, int, int); void nfscl_fillclid(u_int64_t, char *, u_int8_t *, u_int16_t); -void nfscl_filllockowner(NFSPROC_T *, u_int8_t *); +void nfscl_filllockowner(void *, u_int8_t *, int); void nfscl_freeopen(struct nfsclopen *, int); void nfscl_umount(struct nfsmount *, NFSPROC_T *); void nfscl_renewthread(struct nfsclclient *, NFSPROC_T *); void nfscl_initiate_recovery(struct nfsclclient *); int nfscl_hasexpired(struct nfsclclient *, u_int32_t, NFSPROC_T *); void nfscl_dumpstate(struct nfsmount *, int, int, int, int); void nfscl_dupopen(vnode_t, int); int nfscl_getclose(vnode_t, struct nfsclclient **); int nfscl_doclose(vnode_t, struct nfsclclient **, NFSPROC_T *); void nfsrpc_doclose(struct nfsmount *, struct nfsclopen *, NFSPROC_T *); int nfscl_deleg(mount_t, struct nfsclclient *, u_int8_t *, int, struct ucred *, NFSPROC_T *, struct nfscldeleg **); void nfscl_lockinit(struct nfsv4lock *); void nfscl_lockexcl(struct nfsv4lock *, void *); void nfscl_lockunlock(struct nfsv4lock *); void nfscl_lockderef(struct nfsv4lock *); void nfscl_docb(struct nfsrv_descript *, NFSPROC_T *); -void nfscl_releasealllocks(struct nfsclclient *, vnode_t, NFSPROC_T *); +void nfscl_releasealllocks(struct nfsclclient *, vnode_t, NFSPROC_T *, void *, + int); int nfscl_lockt(vnode_t, struct nfsclclient *, u_int64_t, - u_int64_t, struct flock *, NFSPROC_T *); + u_int64_t, struct flock *, NFSPROC_T *, void *, int); int nfscl_mustflush(vnode_t); int nfscl_nodeleg(vnode_t, int); int nfscl_removedeleg(vnode_t, NFSPROC_T *, nfsv4stateid_t *); int nfscl_getref(struct nfsmount *); void nfscl_relref(struct nfsmount *); int nfscl_renamedeleg(vnode_t, nfsv4stateid_t *, int *, vnode_t, nfsv4stateid_t *, int *, NFSPROC_T *); void nfscl_reclaimnode(vnode_t); void nfscl_newnode(vnode_t); void nfscl_delegmodtime(vnode_t); void nfscl_deleggetmodtime(vnode_t, struct timespec *); int nfscl_tryclose(struct nfsclopen *, struct ucred *, struct nfsmount *, NFSPROC_T *); void nfscl_cleanup(NFSPROC_T *); /* nfs_clport.c */ int nfscl_nget(mount_t, vnode_t, struct nfsfh *, struct componentname *, NFSPROC_T *, struct nfsnode **, void *, int); NFSPROC_T *nfscl_getparent(NFSPROC_T *); void nfscl_start_renewthread(struct nfsclclient *); void nfscl_loadsbinfo(struct nfsmount *, struct nfsstatfs *, void *); void nfscl_loadfsinfo (struct nfsmount *, struct nfsfsinfo *); void nfscl_delegreturn(struct nfscldeleg *, int, struct nfsmount *, struct ucred *, NFSPROC_T *); void nfsrvd_cbinit(int); int nfscl_checksattr(struct vattr *, struct nfsvattr *); int nfscl_ngetreopen(mount_t, u_int8_t *, int, NFSPROC_T *, struct nfsnode **); int nfscl_procdoesntexist(u_int8_t *); int nfscl_maperr(NFSPROC_T *, int, uid_t, gid_t); /* nfs_clsubs.c */ void nfscl_init(void); /* nfs_clbio.c */ int ncl_flush(vnode_t, int, struct ucred *, NFSPROC_T *, int, int); /* nfs_clnode.c */ void ncl_invalcaches(vnode_t); /* nfs_nfsdport.c */ int nfsvno_getattr(vnode_t, struct nfsvattr *, struct ucred *, NFSPROC_T *, int); int nfsvno_setattr(vnode_t, struct nfsvattr *, struct ucred *, NFSPROC_T *, struct nfsexstuff *); int nfsvno_getfh(vnode_t, fhandle_t *, NFSPROC_T *); int nfsvno_accchk(vnode_t, accmode_t, struct ucred *, struct nfsexstuff *, NFSPROC_T *, int, int, u_int32_t *); int nfsvno_namei(struct nfsrv_descript *, struct nameidata *, vnode_t, int, struct nfsexstuff *, NFSPROC_T *, vnode_t *); void nfsvno_setpathbuf(struct nameidata *, char **, u_long **); void nfsvno_relpathbuf(struct nameidata *); int nfsvno_readlink(vnode_t, struct ucred *, NFSPROC_T *, mbuf_t *, mbuf_t *, int *); int nfsvno_read(vnode_t, off_t, int, struct ucred *, NFSPROC_T *, mbuf_t *, mbuf_t *); int nfsvno_write(vnode_t, off_t, int, int, int, mbuf_t, char *, struct ucred *, NFSPROC_T *); int nfsvno_createsub(struct nfsrv_descript *, struct nameidata *, vnode_t *, struct nfsvattr *, int *, int32_t *, NFSDEV_T, NFSPROC_T *, struct nfsexstuff *); int nfsvno_mknod(struct nameidata *, struct nfsvattr *, struct ucred *, NFSPROC_T *); int nfsvno_mkdir(struct nameidata *, struct nfsvattr *, uid_t, struct ucred *, NFSPROC_T *, struct nfsexstuff *); int nfsvno_symlink(struct nameidata *, struct nfsvattr *, char *, int, int, uid_t, struct ucred *, NFSPROC_T *, struct nfsexstuff *); int nfsvno_getsymlink(struct nfsrv_descript *, struct nfsvattr *, NFSPROC_T *, char **, int *); int nfsvno_removesub(struct nameidata *, int, struct ucred *, NFSPROC_T *, struct nfsexstuff *); int nfsvno_rmdirsub(struct nameidata *, int, struct ucred *, NFSPROC_T *, struct nfsexstuff *); int nfsvno_rename(struct nameidata *, struct nameidata *, u_int32_t, u_int32_t, struct ucred *, NFSPROC_T *); int nfsvno_link(struct nameidata *, vnode_t, struct ucred *, NFSPROC_T *, struct nfsexstuff *); int nfsvno_fsync(vnode_t, u_int64_t, int, struct ucred *, NFSPROC_T *); int nfsvno_statfs(vnode_t, struct statfs *); void nfsvno_getfs(struct nfsfsinfo *, int); void nfsvno_open(struct nfsrv_descript *, struct nameidata *, nfsquad_t, nfsv4stateid_t *, struct nfsstate *, int *, struct nfsvattr *, int32_t *, int, NFSACL_T *, nfsattrbit_t *, struct ucred *, NFSPROC_T *, struct nfsexstuff *, vnode_t *); void nfsvno_updfilerev(vnode_t, struct nfsvattr *, struct ucred *, NFSPROC_T *); int nfsvno_fillattr(struct nfsrv_descript *, struct mount *, vnode_t, struct nfsvattr *, fhandle_t *, int, nfsattrbit_t *, struct ucred *, NFSPROC_T *, int, int, int, int, uint64_t); int nfsrv_sattr(struct nfsrv_descript *, struct nfsvattr *, nfsattrbit_t *, NFSACL_T *, NFSPROC_T *); int nfsv4_sattr(struct nfsrv_descript *, struct nfsvattr *, nfsattrbit_t *, NFSACL_T *, NFSPROC_T *); int nfsvno_checkexp(mount_t, NFSSOCKADDR_T, struct nfsexstuff *, struct ucred **); int nfsvno_fhtovp(mount_t, fhandle_t *, NFSSOCKADDR_T, int, vnode_t *, struct nfsexstuff *, struct ucred **); int nfsvno_pathconf(vnode_t, int, register_t *, struct ucred *, NFSPROC_T *); vnode_t nfsvno_getvp(fhandle_t *); int nfsvno_advlock(vnode_t, int, u_int64_t, u_int64_t, NFSPROC_T *); int nfsrv_v4rootexport(void *, struct ucred *, NFSPROC_T *); int nfsvno_testexp(struct nfsrv_descript *, struct nfsexstuff *); uint32_t nfsrv_hashfh(fhandle_t *); void nfsrv_backupstable(void); /* nfs_commonkrpc.c */ int newnfs_nmcancelreqs(struct nfsmount *); void newnfs_set_sigmask(struct thread *, sigset_t *); void newnfs_restore_sigmask(struct thread *, sigset_t *); int newnfs_msleep(struct thread *, void *, struct mtx *, int, char *, int); int newnfs_request(struct nfsrv_descript *, struct nfsmount *, struct nfsclient *, struct nfssockreq *, vnode_t, NFSPROC_T *, struct ucred *, u_int32_t, u_int32_t, u_char *, int, u_int64_t *); int newnfs_connect(struct nfsmount *, struct nfssockreq *, struct ucred *, NFSPROC_T *, int); void newnfs_disconnect(struct nfssockreq *); int newnfs_sigintr(struct nfsmount *, NFSPROC_T *); /* nfs_nfsdkrpc.c */ int nfsrvd_addsock(struct file *); int nfsrvd_nfsd(NFSPROC_T *, struct nfsd_nfsd_args *); void nfsrvd_init(int); /* nfs_clkrpc.c */ int nfscbd_addsock(struct file *); int nfscbd_nfsd(NFSPROC_T *, struct nfsd_nfscbd_args *); Index: projects/largeSMP/sys/fs/nfsclient/nfs_clport.c =================================================================== --- projects/largeSMP/sys/fs/nfsclient/nfs_clport.c (revision 222782) +++ projects/largeSMP/sys/fs/nfsclient/nfs_clport.c (revision 222783) @@ -1,1296 +1,1298 @@ /*- * Copyright (c) 1989, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Rick Macklem at The University of Guelph. * * 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. * */ #include __FBSDID("$FreeBSD$"); /* * generally, I don't like #includes inside .h files, but it seems to * be the easiest way to handle the port. */ #include #include #include #include extern u_int32_t newnfs_true, newnfs_false, newnfs_xdrneg1; extern struct vop_vector newnfs_vnodeops; extern struct vop_vector newnfs_fifoops; extern uma_zone_t newnfsnode_zone; extern struct buf_ops buf_ops_newnfs; extern int ncl_pbuf_freecnt; extern short nfsv4_cbport; extern int nfscl_enablecallb; extern int nfs_numnfscbd; extern int nfscl_inited; struct mtx nfs_clstate_mutex; struct mtx ncl_iod_mutex; NFSDLOCKMUTEX; extern void (*ncl_call_invalcaches)(struct vnode *); /* * Comparison function for vfs_hash functions. */ int newnfs_vncmpf(struct vnode *vp, void *arg) { struct nfsfh *nfhp = (struct nfsfh *)arg; struct nfsnode *np = VTONFS(vp); if (np->n_fhp->nfh_len != nfhp->nfh_len || NFSBCMP(np->n_fhp->nfh_fh, nfhp->nfh_fh, nfhp->nfh_len)) return (1); return (0); } /* * Look up a vnode/nfsnode by file handle. * Callers must check for mount points!! * In all cases, a pointer to a * nfsnode structure is returned. * This variant takes a "struct nfsfh *" as second argument and uses * that structure up, either by hanging off the nfsnode or FREEing it. */ int nfscl_nget(struct mount *mntp, struct vnode *dvp, struct nfsfh *nfhp, struct componentname *cnp, struct thread *td, struct nfsnode **npp, void *stuff, int lkflags) { struct nfsnode *np, *dnp; struct vnode *vp, *nvp; struct nfsv4node *newd, *oldd; int error; u_int hash; struct nfsmount *nmp; nmp = VFSTONFS(mntp); dnp = VTONFS(dvp); *npp = NULL; hash = fnv_32_buf(nfhp->nfh_fh, nfhp->nfh_len, FNV1_32_INIT); error = vfs_hash_get(mntp, hash, lkflags, td, &nvp, newnfs_vncmpf, nfhp); if (error == 0 && nvp != NULL) { /* * I believe there is a slight chance that vgonel() could * get called on this vnode between when vn_lock() drops * the VI_LOCK() and vget() acquires it again, so that it * hasn't yet had v_usecount incremented. If this were to * happen, the VI_DOOMED flag would be set, so check for * that here. Since we now have the v_usecount incremented, * we should be ok until we vrele() it, if the VI_DOOMED * flag isn't set now. */ VI_LOCK(nvp); if ((nvp->v_iflag & VI_DOOMED)) { VI_UNLOCK(nvp); vrele(nvp); error = ENOENT; } else { VI_UNLOCK(nvp); } } if (error) { FREE((caddr_t)nfhp, M_NFSFH); return (error); } if (nvp != NULL) { np = VTONFS(nvp); /* * For NFSv4, check to see if it is the same name and * replace the name, if it is different. */ oldd = newd = NULL; if ((nmp->nm_flag & NFSMNT_NFSV4) && np->n_v4 != NULL && nvp->v_type == VREG && (np->n_v4->n4_namelen != cnp->cn_namelen || NFSBCMP(cnp->cn_nameptr, NFS4NODENAME(np->n_v4), cnp->cn_namelen) || dnp->n_fhp->nfh_len != np->n_v4->n4_fhlen || NFSBCMP(dnp->n_fhp->nfh_fh, np->n_v4->n4_data, dnp->n_fhp->nfh_len))) { MALLOC(newd, struct nfsv4node *, sizeof (struct nfsv4node) + dnp->n_fhp->nfh_len + + cnp->cn_namelen - 1, M_NFSV4NODE, M_WAITOK); NFSLOCKNODE(np); if (newd != NULL && np->n_v4 != NULL && nvp->v_type == VREG && (np->n_v4->n4_namelen != cnp->cn_namelen || NFSBCMP(cnp->cn_nameptr, NFS4NODENAME(np->n_v4), cnp->cn_namelen) || dnp->n_fhp->nfh_len != np->n_v4->n4_fhlen || NFSBCMP(dnp->n_fhp->nfh_fh, np->n_v4->n4_data, dnp->n_fhp->nfh_len))) { oldd = np->n_v4; np->n_v4 = newd; newd = NULL; np->n_v4->n4_fhlen = dnp->n_fhp->nfh_len; np->n_v4->n4_namelen = cnp->cn_namelen; NFSBCOPY(dnp->n_fhp->nfh_fh, np->n_v4->n4_data, dnp->n_fhp->nfh_len); NFSBCOPY(cnp->cn_nameptr, NFS4NODENAME(np->n_v4), cnp->cn_namelen); } NFSUNLOCKNODE(np); } if (newd != NULL) FREE((caddr_t)newd, M_NFSV4NODE); if (oldd != NULL) FREE((caddr_t)oldd, M_NFSV4NODE); *npp = np; FREE((caddr_t)nfhp, M_NFSFH); return (0); } /* * Allocate before getnewvnode since doing so afterward * might cause a bogus v_data pointer to get dereferenced * elsewhere if zalloc should block. */ np = uma_zalloc(newnfsnode_zone, M_WAITOK | M_ZERO); error = getnewvnode("newnfs", mntp, &newnfs_vnodeops, &nvp); if (error) { uma_zfree(newnfsnode_zone, np); FREE((caddr_t)nfhp, M_NFSFH); return (error); } vp = nvp; vp->v_bufobj.bo_ops = &buf_ops_newnfs; vp->v_data = np; np->n_vnode = vp; /* * Initialize the mutex even if the vnode is going to be a loser. * This simplifies the logic in reclaim, which can then unconditionally * destroy the mutex (in the case of the loser, or if hash_insert * happened to return an error no special casing is needed). */ mtx_init(&np->n_mtx, "NEWNFSnode lock", NULL, MTX_DEF | MTX_DUPOK); /* * Are we getting the root? If so, make sure the vnode flags * are correct */ if ((nfhp->nfh_len == nmp->nm_fhsize) && !bcmp(nfhp->nfh_fh, nmp->nm_fh, nfhp->nfh_len)) { if (vp->v_type == VNON) vp->v_type = VDIR; vp->v_vflag |= VV_ROOT; } np->n_fhp = nfhp; /* * For NFSv4, we have to attach the directory file handle and * file name, so that Open Ops can be done later. */ if (nmp->nm_flag & NFSMNT_NFSV4) { MALLOC(np->n_v4, struct nfsv4node *, sizeof (struct nfsv4node) + dnp->n_fhp->nfh_len + cnp->cn_namelen - 1, M_NFSV4NODE, M_WAITOK); np->n_v4->n4_fhlen = dnp->n_fhp->nfh_len; np->n_v4->n4_namelen = cnp->cn_namelen; NFSBCOPY(dnp->n_fhp->nfh_fh, np->n_v4->n4_data, dnp->n_fhp->nfh_len); NFSBCOPY(cnp->cn_nameptr, NFS4NODENAME(np->n_v4), cnp->cn_namelen); } else { np->n_v4 = NULL; } /* * NFS supports recursive and shared locking. */ lockmgr(vp->v_vnlock, LK_EXCLUSIVE | LK_NOWITNESS, NULL); VN_LOCK_AREC(vp); VN_LOCK_ASHARE(vp); error = insmntque(vp, mntp); if (error != 0) { *npp = NULL; mtx_destroy(&np->n_mtx); FREE((caddr_t)nfhp, M_NFSFH); if (np->n_v4 != NULL) FREE((caddr_t)np->n_v4, M_NFSV4NODE); uma_zfree(newnfsnode_zone, np); return (error); } error = vfs_hash_insert(vp, hash, lkflags, td, &nvp, newnfs_vncmpf, nfhp); if (error) return (error); if (nvp != NULL) { *npp = VTONFS(nvp); /* vfs_hash_insert() vput()'s the losing vnode */ return (0); } *npp = np; return (0); } /* * Anothe variant of nfs_nget(). This one is only used by reopen. It * takes almost the same args as nfs_nget(), but only succeeds if an entry * exists in the cache. (Since files should already be "open" with a * vnode ref cnt on the node when reopen calls this, it should always * succeed.) * Also, don't get a vnode lock, since it may already be locked by some * other process that is handling it. This is ok, since all other threads * on the client are blocked by the nfsc_lock being exclusively held by the * caller of this function. */ int nfscl_ngetreopen(struct mount *mntp, u_int8_t *fhp, int fhsize, struct thread *td, struct nfsnode **npp) { struct vnode *nvp; u_int hash; struct nfsfh *nfhp; int error; *npp = NULL; /* For forced dismounts, just return error. */ if ((mntp->mnt_kern_flag & MNTK_UNMOUNTF)) return (EINTR); MALLOC(nfhp, struct nfsfh *, sizeof (struct nfsfh) + fhsize, M_NFSFH, M_WAITOK); bcopy(fhp, &nfhp->nfh_fh[0], fhsize); nfhp->nfh_len = fhsize; hash = fnv_32_buf(fhp, fhsize, FNV1_32_INIT); /* * First, try to get the vnode locked, but don't block for the lock. */ error = vfs_hash_get(mntp, hash, (LK_EXCLUSIVE | LK_NOWAIT), td, &nvp, newnfs_vncmpf, nfhp); if (error == 0 && nvp != NULL) { VOP_UNLOCK(nvp, 0); } else if (error == EBUSY) { /* * The LK_EXCLOTHER lock type tells nfs_lock1() to not try * and lock the vnode, but just get a v_usecount on it. * LK_NOWAIT is set so that when vget() returns ENOENT, * vfs_hash_get() fails instead of looping. * If this succeeds, it is safe so long as a vflush() with * FORCECLOSE has not been done. Since the Renew thread is * stopped and the MNTK_UNMOUNTF flag is set before doing * a vflush() with FORCECLOSE, we should be ok here. */ if ((mntp->mnt_kern_flag & MNTK_UNMOUNTF)) error = EINTR; else error = vfs_hash_get(mntp, hash, (LK_EXCLOTHER | LK_NOWAIT), td, &nvp, newnfs_vncmpf, nfhp); } FREE(nfhp, M_NFSFH); if (error) return (error); if (nvp != NULL) { *npp = VTONFS(nvp); return (0); } return (EINVAL); } /* * Load the attribute cache (that lives in the nfsnode entry) with * the attributes of the second argument and * Iff vaper not NULL * copy the attributes to *vaper * Similar to nfs_loadattrcache(), except the attributes are passed in * instead of being parsed out of the mbuf list. */ int nfscl_loadattrcache(struct vnode **vpp, struct nfsvattr *nap, void *nvaper, void *stuff, int writeattr, int dontshrink) { struct vnode *vp = *vpp; struct vattr *vap, *nvap = &nap->na_vattr, *vaper = nvaper; struct nfsnode *np; struct nfsmount *nmp; struct timespec mtime_save; /* * If v_type == VNON it is a new node, so fill in the v_type, * n_mtime fields. Check to see if it represents a special * device, and if so, check for a possible alias. Once the * correct vnode has been obtained, fill in the rest of the * information. */ np = VTONFS(vp); NFSLOCKNODE(np); if (vp->v_type != nvap->va_type) { vp->v_type = nvap->va_type; if (vp->v_type == VFIFO) vp->v_op = &newnfs_fifoops; np->n_mtime = nvap->va_mtime; } nmp = VFSTONFS(vp->v_mount); vap = &np->n_vattr.na_vattr; mtime_save = vap->va_mtime; if (writeattr) { np->n_vattr.na_filerev = nap->na_filerev; np->n_vattr.na_size = nap->na_size; np->n_vattr.na_mtime = nap->na_mtime; np->n_vattr.na_ctime = nap->na_ctime; np->n_vattr.na_fsid = nap->na_fsid; } else { NFSBCOPY((caddr_t)nap, (caddr_t)&np->n_vattr, sizeof (struct nfsvattr)); } /* * For NFSv4, if the node's fsid is not equal to the mount point's * fsid, return the low order 32bits of the node's fsid. This * allows getcwd(3) to work. There is a chance that the fsid might * be the same as a local fs, but since this is in an NFS mount * point, I don't think that will cause any problems? */ if (NFSHASNFSV4(nmp) && NFSHASHASSETFSID(nmp) && (nmp->nm_fsid[0] != np->n_vattr.na_filesid[0] || nmp->nm_fsid[1] != np->n_vattr.na_filesid[1])) { /* * va_fsid needs to be set to some value derived from * np->n_vattr.na_filesid that is not equal * vp->v_mount->mnt_stat.f_fsid[0], so that it changes * from the value used for the top level server volume * in the mounted subtree. */ if (vp->v_mount->mnt_stat.f_fsid.val[0] != (uint32_t)np->n_vattr.na_filesid[0]) vap->va_fsid = (uint32_t)np->n_vattr.na_filesid[0]; else vap->va_fsid = (uint32_t)hash32_buf( np->n_vattr.na_filesid, 2 * sizeof(uint64_t), 0); } else vap->va_fsid = vp->v_mount->mnt_stat.f_fsid.val[0]; np->n_attrstamp = time_second; if (vap->va_size != np->n_size) { if (vap->va_type == VREG) { if (dontshrink && vap->va_size < np->n_size) { /* * We've been told not to shrink the file; * zero np->n_attrstamp to indicate that * the attributes are stale. */ vap->va_size = np->n_size; np->n_attrstamp = 0; } else if (np->n_flag & NMODIFIED) { /* * We've modified the file: Use the larger * of our size, and the server's size. */ if (vap->va_size < np->n_size) { vap->va_size = np->n_size; } else { np->n_size = vap->va_size; np->n_flag |= NSIZECHANGED; } } else { np->n_size = vap->va_size; np->n_flag |= NSIZECHANGED; } vnode_pager_setsize(vp, np->n_size); } else { np->n_size = vap->va_size; } } /* * The following checks are added to prevent a race between (say) * a READDIR+ and a WRITE. * READDIR+, WRITE requests sent out. * READDIR+ resp, WRITE resp received on client. * However, the WRITE resp was handled before the READDIR+ resp * causing the post op attrs from the write to be loaded first * and the attrs from the READDIR+ to be loaded later. If this * happens, we have stale attrs loaded into the attrcache. * We detect this by for the mtime moving back. We invalidate the * attrcache when this happens. */ if (timespeccmp(&mtime_save, &vap->va_mtime, >)) /* Size changed or mtime went backwards */ np->n_attrstamp = 0; if (vaper != NULL) { NFSBCOPY((caddr_t)vap, (caddr_t)vaper, sizeof(*vap)); if (np->n_flag & NCHG) { if (np->n_flag & NACC) vaper->va_atime = np->n_atim; if (np->n_flag & NUPD) vaper->va_mtime = np->n_mtim; } } NFSUNLOCKNODE(np); return (0); } /* * Fill in the client id name. For these bytes: * 1 - they must be unique * 2 - they should be persistent across client reboots * 1 is more critical than 2 * Use the mount point's unique id plus either the uuid or, if that * isn't set, random junk. */ void nfscl_fillclid(u_int64_t clval, char *uuid, u_int8_t *cp, u_int16_t idlen) { int uuidlen; /* * First, put in the 64bit mount point identifier. */ if (idlen >= sizeof (u_int64_t)) { NFSBCOPY((caddr_t)&clval, cp, sizeof (u_int64_t)); cp += sizeof (u_int64_t); idlen -= sizeof (u_int64_t); } /* * If uuid is non-zero length, use it. */ uuidlen = strlen(uuid); if (uuidlen > 0 && idlen >= uuidlen) { NFSBCOPY(uuid, cp, uuidlen); cp += uuidlen; idlen -= uuidlen; } /* * This only normally happens if the uuid isn't set. */ while (idlen > 0) { *cp++ = (u_int8_t)(arc4random() % 256); idlen--; } } /* * Fill in a lock owner name. For now, pid + the process's creation time. */ void -nfscl_filllockowner(struct thread *td, u_int8_t *cp) +nfscl_filllockowner(void *id, u_int8_t *cp, int flags) { union { u_int32_t lval; u_int8_t cval[4]; } tl; struct proc *p; -if (td == NULL) { - printf("NULL td\n"); - bzero(cp, 12); - return; + if (id == NULL) { + printf("NULL id\n"); + bzero(cp, NFSV4CL_LOCKNAMELEN); + return; + } + if ((flags & F_POSIX) != 0) { + p = (struct proc *)id; + tl.lval = p->p_pid; + *cp++ = tl.cval[0]; + *cp++ = tl.cval[1]; + *cp++ = tl.cval[2]; + *cp++ = tl.cval[3]; + tl.lval = p->p_stats->p_start.tv_sec; + *cp++ = tl.cval[0]; + *cp++ = tl.cval[1]; + *cp++ = tl.cval[2]; + *cp++ = tl.cval[3]; + tl.lval = p->p_stats->p_start.tv_usec; + *cp++ = tl.cval[0]; + *cp++ = tl.cval[1]; + *cp++ = tl.cval[2]; + *cp = tl.cval[3]; + } else if ((flags & F_FLOCK) != 0) { + bcopy(&id, cp, sizeof(id)); + bzero(&cp[sizeof(id)], NFSV4CL_LOCKNAMELEN - sizeof(id)); + } else { + printf("nfscl_filllockowner: not F_POSIX or F_FLOCK\n"); + bzero(cp, NFSV4CL_LOCKNAMELEN); + } } - p = td->td_proc; -if (p == NULL) { - printf("NULL pid\n"); - bzero(cp, 12); - return; -} - tl.lval = p->p_pid; - *cp++ = tl.cval[0]; - *cp++ = tl.cval[1]; - *cp++ = tl.cval[2]; - *cp++ = tl.cval[3]; -if (p->p_stats == NULL) { - printf("pstats null\n"); - bzero(cp, 8); - return; -} - tl.lval = p->p_stats->p_start.tv_sec; - *cp++ = tl.cval[0]; - *cp++ = tl.cval[1]; - *cp++ = tl.cval[2]; - *cp++ = tl.cval[3]; - tl.lval = p->p_stats->p_start.tv_usec; - *cp++ = tl.cval[0]; - *cp++ = tl.cval[1]; - *cp++ = tl.cval[2]; - *cp = tl.cval[3]; -} /* * Find the parent process for the thread passed in as an argument. * If none exists, return NULL, otherwise return a thread for the parent. * (Can be any of the threads, since it is only used for td->td_proc.) */ NFSPROC_T * nfscl_getparent(struct thread *td) { struct proc *p; struct thread *ptd; if (td == NULL) return (NULL); p = td->td_proc; if (p->p_pid == 0) return (NULL); p = p->p_pptr; if (p == NULL) return (NULL); ptd = TAILQ_FIRST(&p->p_threads); return (ptd); } /* * Start up the renew kernel thread. */ static void start_nfscl(void *arg) { struct nfsclclient *clp; struct thread *td; clp = (struct nfsclclient *)arg; td = TAILQ_FIRST(&clp->nfsc_renewthread->p_threads); nfscl_renewthread(clp, td); kproc_exit(0); } void nfscl_start_renewthread(struct nfsclclient *clp) { kproc_create(start_nfscl, (void *)clp, &clp->nfsc_renewthread, 0, 0, "nfscl"); } /* * Handle wcc_data. * For NFSv4, it assumes that nfsv4_wccattr() was used to set up the getattr * as the first Op after PutFH. * (For NFSv4, the postop attributes are after the Op, so they can't be * parsed here. A separate call to nfscl_postop_attr() is required.) */ int nfscl_wcc_data(struct nfsrv_descript *nd, struct vnode *vp, struct nfsvattr *nap, int *flagp, int *wccflagp, void *stuff) { u_int32_t *tl; struct nfsnode *np = VTONFS(vp); struct nfsvattr nfsva; int error = 0; if (wccflagp != NULL) *wccflagp = 0; if (nd->nd_flag & ND_NFSV3) { *flagp = 0; NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (*tl == newnfs_true) { NFSM_DISSECT(tl, u_int32_t *, 6 * NFSX_UNSIGNED); if (wccflagp != NULL) { mtx_lock(&np->n_mtx); *wccflagp = (np->n_mtime.tv_sec == fxdr_unsigned(u_int32_t, *(tl + 2)) && np->n_mtime.tv_nsec == fxdr_unsigned(u_int32_t, *(tl + 3))); mtx_unlock(&np->n_mtx); } } error = nfscl_postop_attr(nd, nap, flagp, stuff); } else if ((nd->nd_flag & (ND_NOMOREDATA | ND_NFSV4 | ND_V4WCCATTR)) == (ND_NFSV4 | ND_V4WCCATTR)) { error = nfsv4_loadattr(nd, NULL, &nfsva, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL); if (error) return (error); /* * Get rid of Op# and status for next op. */ NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); if (*++tl) nd->nd_flag |= ND_NOMOREDATA; if (wccflagp != NULL && nfsva.na_vattr.va_mtime.tv_sec != 0) { mtx_lock(&np->n_mtx); *wccflagp = (np->n_mtime.tv_sec == nfsva.na_vattr.va_mtime.tv_sec && np->n_mtime.tv_nsec == nfsva.na_vattr.va_mtime.tv_sec); mtx_unlock(&np->n_mtx); } } nfsmout: return (error); } /* * Get postop attributes. */ int nfscl_postop_attr(struct nfsrv_descript *nd, struct nfsvattr *nap, int *retp, void *stuff) { u_int32_t *tl; int error = 0; *retp = 0; if (nd->nd_flag & ND_NOMOREDATA) return (error); if (nd->nd_flag & ND_NFSV3) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); *retp = fxdr_unsigned(int, *tl); } else if (nd->nd_flag & ND_NFSV4) { /* * For NFSv4, the postop attr are at the end, so no point * in looking if nd_repstat != 0. */ if (!nd->nd_repstat) { NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); if (*(tl + 1)) /* should never happen since nd_repstat != 0 */ nd->nd_flag |= ND_NOMOREDATA; else *retp = 1; } } else if (!nd->nd_repstat) { /* For NFSv2, the attributes are here iff nd_repstat == 0 */ *retp = 1; } if (*retp) { error = nfsm_loadattr(nd, nap); if (error) *retp = 0; } nfsmout: return (error); } /* * Fill in the setable attributes. The full argument indicates whether * to fill in them all or just mode and time. */ void nfscl_fillsattr(struct nfsrv_descript *nd, struct vattr *vap, struct vnode *vp, int flags, u_int32_t rdev) { u_int32_t *tl; struct nfsv2_sattr *sp; nfsattrbit_t attrbits; struct timeval curtime; switch (nd->nd_flag & (ND_NFSV2 | ND_NFSV3 | ND_NFSV4)) { case ND_NFSV2: NFSM_BUILD(sp, struct nfsv2_sattr *, NFSX_V2SATTR); if (vap->va_mode == (mode_t)VNOVAL) sp->sa_mode = newnfs_xdrneg1; else sp->sa_mode = vtonfsv2_mode(vap->va_type, vap->va_mode); if (vap->va_uid == (uid_t)VNOVAL) sp->sa_uid = newnfs_xdrneg1; else sp->sa_uid = txdr_unsigned(vap->va_uid); if (vap->va_gid == (gid_t)VNOVAL) sp->sa_gid = newnfs_xdrneg1; else sp->sa_gid = txdr_unsigned(vap->va_gid); if (flags & NFSSATTR_SIZE0) sp->sa_size = 0; else if (flags & NFSSATTR_SIZENEG1) sp->sa_size = newnfs_xdrneg1; else if (flags & NFSSATTR_SIZERDEV) sp->sa_size = txdr_unsigned(rdev); else sp->sa_size = txdr_unsigned(vap->va_size); txdr_nfsv2time(&vap->va_atime, &sp->sa_atime); txdr_nfsv2time(&vap->va_mtime, &sp->sa_mtime); break; case ND_NFSV3: getmicrotime(&curtime); if (vap->va_mode != (mode_t)VNOVAL) { NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = newnfs_true; *tl = txdr_unsigned(vap->va_mode); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_false; } if ((flags & NFSSATTR_FULL) && vap->va_uid != (uid_t)VNOVAL) { NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = newnfs_true; *tl = txdr_unsigned(vap->va_uid); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_false; } if ((flags & NFSSATTR_FULL) && vap->va_gid != (gid_t)VNOVAL) { NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = newnfs_true; *tl = txdr_unsigned(vap->va_gid); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_false; } if ((flags & NFSSATTR_FULL) && vap->va_size != VNOVAL) { NFSM_BUILD(tl, u_int32_t *, 3 * NFSX_UNSIGNED); *tl++ = newnfs_true; txdr_hyper(vap->va_size, tl); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_false; } if (vap->va_atime.tv_sec != VNOVAL) { if (vap->va_atime.tv_sec != curtime.tv_sec) { NFSM_BUILD(tl, u_int32_t *, 3 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV3SATTRTIME_TOCLIENT); txdr_nfsv3time(&vap->va_atime, tl); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV3SATTRTIME_TOSERVER); } } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV3SATTRTIME_DONTCHANGE); } if (vap->va_mtime.tv_sec != VNOVAL) { if (vap->va_mtime.tv_sec != curtime.tv_sec) { NFSM_BUILD(tl, u_int32_t *, 3 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV3SATTRTIME_TOCLIENT); txdr_nfsv3time(&vap->va_mtime, tl); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV3SATTRTIME_TOSERVER); } } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV3SATTRTIME_DONTCHANGE); } break; case ND_NFSV4: NFSZERO_ATTRBIT(&attrbits); if (vap->va_mode != (mode_t)VNOVAL) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_MODE); if ((flags & NFSSATTR_FULL) && vap->va_uid != (uid_t)VNOVAL) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_OWNER); if ((flags & NFSSATTR_FULL) && vap->va_gid != (gid_t)VNOVAL) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_OWNERGROUP); if ((flags & NFSSATTR_FULL) && vap->va_size != VNOVAL) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_SIZE); if (vap->va_atime.tv_sec != VNOVAL) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_TIMEACCESSSET); if (vap->va_mtime.tv_sec != VNOVAL) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_TIMEMODIFYSET); (void) nfsv4_fillattr(nd, vp->v_mount, vp, NULL, vap, NULL, 0, &attrbits, NULL, NULL, 0, 0, 0, 0, (uint64_t)0); break; }; } /* * nfscl_request() - mostly a wrapper for newnfs_request(). */ int nfscl_request(struct nfsrv_descript *nd, struct vnode *vp, NFSPROC_T *p, struct ucred *cred, void *stuff) { int ret, vers; struct nfsmount *nmp; nmp = VFSTONFS(vp->v_mount); if (nd->nd_flag & ND_NFSV4) vers = NFS_VER4; else if (nd->nd_flag & ND_NFSV3) vers = NFS_VER3; else vers = NFS_VER2; ret = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, vp, p, cred, NFS_PROG, vers, NULL, 1, NULL); return (ret); } /* * fill in this bsden's variant of statfs using nfsstatfs. */ void nfscl_loadsbinfo(struct nfsmount *nmp, struct nfsstatfs *sfp, void *statfs) { struct statfs *sbp = (struct statfs *)statfs; if (nmp->nm_flag & (NFSMNT_NFSV3 | NFSMNT_NFSV4)) { sbp->f_bsize = NFS_FABLKSIZE; sbp->f_blocks = sfp->sf_tbytes / NFS_FABLKSIZE; sbp->f_bfree = sfp->sf_fbytes / NFS_FABLKSIZE; /* * Although sf_abytes is uint64_t and f_bavail is int64_t, * the value after dividing by NFS_FABLKSIZE is small * enough that it will fit in 63bits, so it is ok to * assign it to f_bavail without fear that it will become * negative. */ sbp->f_bavail = sfp->sf_abytes / NFS_FABLKSIZE; sbp->f_files = sfp->sf_tfiles; /* Since f_ffree is int64_t, clip it to 63bits. */ if (sfp->sf_ffiles > INT64_MAX) sbp->f_ffree = INT64_MAX; else sbp->f_ffree = sfp->sf_ffiles; } else if ((nmp->nm_flag & NFSMNT_NFSV4) == 0) { /* * The type casts to (int32_t) ensure that this code is * compatible with the old NFS client, in that it will * propagate bit31 to the high order bits. This may or may * not be correct for NFSv2, but since it is a legacy * environment, I'd rather retain backwards compatibility. */ sbp->f_bsize = (int32_t)sfp->sf_bsize; sbp->f_blocks = (int32_t)sfp->sf_blocks; sbp->f_bfree = (int32_t)sfp->sf_bfree; sbp->f_bavail = (int32_t)sfp->sf_bavail; sbp->f_files = 0; sbp->f_ffree = 0; } } /* * Use the fsinfo stuff to update the mount point. */ void nfscl_loadfsinfo(struct nfsmount *nmp, struct nfsfsinfo *fsp) { if ((nmp->nm_wsize == 0 || fsp->fs_wtpref < nmp->nm_wsize) && fsp->fs_wtpref >= NFS_FABLKSIZE) nmp->nm_wsize = (fsp->fs_wtpref + NFS_FABLKSIZE - 1) & ~(NFS_FABLKSIZE - 1); if (fsp->fs_wtmax < nmp->nm_wsize && fsp->fs_wtmax > 0) { nmp->nm_wsize = fsp->fs_wtmax & ~(NFS_FABLKSIZE - 1); if (nmp->nm_wsize == 0) nmp->nm_wsize = fsp->fs_wtmax; } if (nmp->nm_wsize < NFS_FABLKSIZE) nmp->nm_wsize = NFS_FABLKSIZE; if ((nmp->nm_rsize == 0 || fsp->fs_rtpref < nmp->nm_rsize) && fsp->fs_rtpref >= NFS_FABLKSIZE) nmp->nm_rsize = (fsp->fs_rtpref + NFS_FABLKSIZE - 1) & ~(NFS_FABLKSIZE - 1); if (fsp->fs_rtmax < nmp->nm_rsize && fsp->fs_rtmax > 0) { nmp->nm_rsize = fsp->fs_rtmax & ~(NFS_FABLKSIZE - 1); if (nmp->nm_rsize == 0) nmp->nm_rsize = fsp->fs_rtmax; } if (nmp->nm_rsize < NFS_FABLKSIZE) nmp->nm_rsize = NFS_FABLKSIZE; if ((nmp->nm_readdirsize == 0 || fsp->fs_dtpref < nmp->nm_readdirsize) && fsp->fs_dtpref >= NFS_DIRBLKSIZ) nmp->nm_readdirsize = (fsp->fs_dtpref + NFS_DIRBLKSIZ - 1) & ~(NFS_DIRBLKSIZ - 1); if (fsp->fs_rtmax < nmp->nm_readdirsize && fsp->fs_rtmax > 0) { nmp->nm_readdirsize = fsp->fs_rtmax & ~(NFS_DIRBLKSIZ - 1); if (nmp->nm_readdirsize == 0) nmp->nm_readdirsize = fsp->fs_rtmax; } if (nmp->nm_readdirsize < NFS_DIRBLKSIZ) nmp->nm_readdirsize = NFS_DIRBLKSIZ; if (fsp->fs_maxfilesize > 0 && fsp->fs_maxfilesize < nmp->nm_maxfilesize) nmp->nm_maxfilesize = fsp->fs_maxfilesize; nmp->nm_mountp->mnt_stat.f_iosize = newnfs_iosize(nmp); nmp->nm_state |= NFSSTA_GOTFSINFO; } /* * Get a pointer to my IP addrress and return it. * Return NULL if you can't find one. */ u_int8_t * nfscl_getmyip(struct nfsmount *nmp, int *isinet6p) { struct sockaddr_in sad, *sin; struct rtentry *rt; u_int8_t *retp = NULL; static struct in_addr laddr; *isinet6p = 0; /* * Loop up a route for the destination address. */ if (nmp->nm_nam->sa_family == AF_INET) { bzero(&sad, sizeof (sad)); sin = (struct sockaddr_in *)nmp->nm_nam; sad.sin_family = AF_INET; sad.sin_len = sizeof (struct sockaddr_in); sad.sin_addr.s_addr = sin->sin_addr.s_addr; + CURVNET_SET(CRED_TO_VNET(nmp->nm_sockreq.nr_cred)); rt = rtalloc1((struct sockaddr *)&sad, 0, 0UL); if (rt != NULL) { if (rt->rt_ifp != NULL && rt->rt_ifa != NULL && ((rt->rt_ifp->if_flags & IFF_LOOPBACK) == 0) && rt->rt_ifa->ifa_addr->sa_family == AF_INET) { sin = (struct sockaddr_in *) rt->rt_ifa->ifa_addr; laddr.s_addr = sin->sin_addr.s_addr; retp = (u_int8_t *)&laddr; } RTFREE_LOCKED(rt); } + CURVNET_RESTORE(); #ifdef INET6 } else if (nmp->nm_nam->sa_family == AF_INET6) { struct sockaddr_in6 sad6, *sin6; static struct in6_addr laddr6; bzero(&sad6, sizeof (sad6)); sin6 = (struct sockaddr_in6 *)nmp->nm_nam; sad6.sin6_family = AF_INET6; sad6.sin6_len = sizeof (struct sockaddr_in6); sad6.sin6_addr = sin6->sin6_addr; + CURVNET_SET(CRED_TO_VNET(nmp->nm_sockreq.nr_cred)); rt = rtalloc1((struct sockaddr *)&sad6, 0, 0UL); if (rt != NULL) { if (rt->rt_ifp != NULL && rt->rt_ifa != NULL && ((rt->rt_ifp->if_flags & IFF_LOOPBACK) == 0) && rt->rt_ifa->ifa_addr->sa_family == AF_INET6) { sin6 = (struct sockaddr_in6 *) rt->rt_ifa->ifa_addr; laddr6 = sin6->sin6_addr; retp = (u_int8_t *)&laddr6; *isinet6p = 1; } RTFREE_LOCKED(rt); } + CURVNET_RESTORE(); #endif } return (retp); } /* * Copy NFS uid, gids from the cred structure. */ void newnfs_copyincred(struct ucred *cr, struct nfscred *nfscr) { int i; KASSERT(cr->cr_ngroups >= 0, ("newnfs_copyincred: negative cr_ngroups")); nfscr->nfsc_uid = cr->cr_uid; nfscr->nfsc_ngroups = MIN(cr->cr_ngroups, NFS_MAXGRPS + 1); for (i = 0; i < nfscr->nfsc_ngroups; i++) nfscr->nfsc_groups[i] = cr->cr_groups[i]; } /* * Do any client specific initialization. */ void nfscl_init(void) { static int inited = 0; if (inited) return; inited = 1; nfscl_inited = 1; ncl_pbuf_freecnt = nswbuf / 2 + 1; } /* * Check each of the attributes to be set, to ensure they aren't already * the correct value. Disable setting ones already correct. */ int nfscl_checksattr(struct vattr *vap, struct nfsvattr *nvap) { if (vap->va_mode != (mode_t)VNOVAL) { if (vap->va_mode == nvap->na_mode) vap->va_mode = (mode_t)VNOVAL; } if (vap->va_uid != (uid_t)VNOVAL) { if (vap->va_uid == nvap->na_uid) vap->va_uid = (uid_t)VNOVAL; } if (vap->va_gid != (gid_t)VNOVAL) { if (vap->va_gid == nvap->na_gid) vap->va_gid = (gid_t)VNOVAL; } if (vap->va_size != VNOVAL) { if (vap->va_size == nvap->na_size) vap->va_size = VNOVAL; } /* * We are normally called with only a partially initialized * VAP. Since the NFSv3 spec says that server may use the * file attributes to store the verifier, the spec requires * us to do a SETATTR RPC. FreeBSD servers store the verifier * in atime, but we can't really assume that all servers will * so we ensure that our SETATTR sets both atime and mtime. */ if (vap->va_mtime.tv_sec == VNOVAL) vfs_timestamp(&vap->va_mtime); if (vap->va_atime.tv_sec == VNOVAL) vap->va_atime = vap->va_mtime; return (1); } /* * Map nfsv4 errors to errno.h errors. * The uid and gid arguments are only used for NFSERR_BADOWNER and that * error should only be returned for the Open, Create and Setattr Ops. * As such, most calls can just pass in 0 for those arguments. */ APPLESTATIC int nfscl_maperr(struct thread *td, int error, uid_t uid, gid_t gid) { struct proc *p; if (error < 10000) return (error); if (td != NULL) p = td->td_proc; else p = NULL; switch (error) { case NFSERR_BADOWNER: tprintf(p, LOG_INFO, "No name and/or group mapping for uid,gid:(%d,%d)\n", uid, gid); return (EPERM); case NFSERR_STALECLIENTID: case NFSERR_STALESTATEID: case NFSERR_EXPIRED: case NFSERR_BADSTATEID: printf("nfsv4 recover err returned %d\n", error); return (EIO); case NFSERR_BADHANDLE: case NFSERR_SERVERFAULT: case NFSERR_BADTYPE: case NFSERR_FHEXPIRED: case NFSERR_RESOURCE: case NFSERR_MOVED: case NFSERR_NOFILEHANDLE: case NFSERR_MINORVERMISMATCH: case NFSERR_OLDSTATEID: case NFSERR_BADSEQID: case NFSERR_LEASEMOVED: case NFSERR_RECLAIMBAD: case NFSERR_BADXDR: case NFSERR_BADCHAR: case NFSERR_BADNAME: case NFSERR_OPILLEGAL: printf("nfsv4 client/server protocol prob err=%d\n", error); return (EIO); default: tprintf(p, LOG_INFO, "nfsv4 err=%d\n", error); return (EIO); }; } /* * Locate a process by number; return only "live" processes -- i.e., neither * zombies nor newly born but incompletely initialized processes. By not * returning processes in the PRS_NEW state, we allow callers to avoid * testing for that condition to avoid dereferencing p_ucred, et al. * Identical to pfind() in kern_proc.c, except it assume the list is * already locked. */ static struct proc * pfind_locked(pid_t pid) { struct proc *p; LIST_FOREACH(p, PIDHASH(pid), p_hash) if (p->p_pid == pid) { PROC_LOCK(p); if (p->p_state == PRS_NEW) { PROC_UNLOCK(p); p = NULL; } break; } return (p); } /* * Check to see if the process for this owner exists. Return 1 if it doesn't * and 0 otherwise. */ int nfscl_procdoesntexist(u_int8_t *own) { union { u_int32_t lval; u_int8_t cval[4]; } tl; struct proc *p; pid_t pid; int ret = 0; tl.cval[0] = *own++; tl.cval[1] = *own++; tl.cval[2] = *own++; tl.cval[3] = *own++; pid = tl.lval; p = pfind_locked(pid); if (p == NULL) return (1); if (p->p_stats == NULL) { PROC_UNLOCK(p); return (0); } tl.cval[0] = *own++; tl.cval[1] = *own++; tl.cval[2] = *own++; tl.cval[3] = *own++; if (tl.lval != p->p_stats->p_start.tv_sec) { ret = 1; } else { tl.cval[0] = *own++; tl.cval[1] = *own++; tl.cval[2] = *own++; tl.cval[3] = *own; if (tl.lval != p->p_stats->p_start.tv_usec) ret = 1; } PROC_UNLOCK(p); return (ret); } /* * - nfs pseudo system call for the client */ /* * MPSAFE */ static int nfssvc_nfscl(struct thread *td, struct nfssvc_args *uap) { struct file *fp; struct nfscbd_args nfscbdarg; struct nfsd_nfscbd_args nfscbdarg2; int error; if (uap->flag & NFSSVC_CBADDSOCK) { error = copyin(uap->argp, (caddr_t)&nfscbdarg, sizeof(nfscbdarg)); if (error) return (error); if ((error = fget(td, nfscbdarg.sock, &fp)) != 0) { return (error); } if (fp->f_type != DTYPE_SOCKET) { fdrop(fp, td); return (EPERM); } error = nfscbd_addsock(fp); fdrop(fp, td); if (!error && nfscl_enablecallb == 0) { nfsv4_cbport = nfscbdarg.port; nfscl_enablecallb = 1; } } else if (uap->flag & NFSSVC_NFSCBD) { if (uap->argp == NULL) return (EINVAL); error = copyin(uap->argp, (caddr_t)&nfscbdarg2, sizeof(nfscbdarg2)); if (error) return (error); error = nfscbd_nfsd(td, &nfscbdarg2); } else { error = EINVAL; } return (error); } extern int (*nfsd_call_nfscl)(struct thread *, struct nfssvc_args *); /* * Called once to initialize data structures... */ static int nfscl_modevent(module_t mod, int type, void *data) { int error = 0; static int loaded = 0; switch (type) { case MOD_LOAD: if (loaded) return (0); newnfs_portinit(); mtx_init(&nfs_clstate_mutex, "nfs_clstate_mutex", NULL, MTX_DEF); mtx_init(&ncl_iod_mutex, "ncl_iod_mutex", NULL, MTX_DEF); nfscl_init(); NFSD_LOCK(); nfsrvd_cbinit(0); NFSD_UNLOCK(); ncl_call_invalcaches = ncl_invalcaches; nfsd_call_nfscl = nfssvc_nfscl; loaded = 1; break; case MOD_UNLOAD: if (nfs_numnfscbd != 0) { error = EBUSY; break; } /* * XXX: Unloading of nfscl module is unsupported. */ #if 0 ncl_call_invalcaches = NULL; nfsd_call_nfscl = NULL; /* and get rid of the mutexes */ mtx_destroy(&nfs_clstate_mutex); mtx_destroy(&ncl_iod_mutex); loaded = 0; break; #else /* FALLTHROUGH */ #endif default: error = EOPNOTSUPP; break; } return error; } static moduledata_t nfscl_mod = { "nfscl", nfscl_modevent, NULL, }; DECLARE_MODULE(nfscl, nfscl_mod, SI_SUB_VFS, SI_ORDER_FIRST); /* So that loader and kldload(2) can find us, wherever we are.. */ MODULE_VERSION(nfscl, 1); MODULE_DEPEND(nfscl, nfscommon, 1, 1, 1); MODULE_DEPEND(nfscl, krpc, 1, 1, 1); MODULE_DEPEND(nfscl, nfssvc, 1, 1, 1); MODULE_DEPEND(nfscl, nfslock, 1, 1, 1); Index: projects/largeSMP/sys/fs/nfsclient/nfs_clrpcops.c =================================================================== --- projects/largeSMP/sys/fs/nfsclient/nfs_clrpcops.c (revision 222782) +++ projects/largeSMP/sys/fs/nfsclient/nfs_clrpcops.c (revision 222783) @@ -1,4214 +1,4214 @@ /*- * Copyright (c) 1989, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Rick Macklem at The University of Guelph. * * 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. * */ #include __FBSDID("$FreeBSD$"); /* * Rpc op calls, generally called from the vnode op calls or through the * buffer cache, for NFS v2, 3 and 4. * These do not normally make any changes to vnode arguments or use * structures that might change between the VFS variants. The returned * arguments are all at the end, after the NFSPROC_T *p one. */ #ifndef APPLEKEXT #include /* * Global variables */ extern int nfs_numnfscbd; extern struct timeval nfsboottime; extern u_int32_t newnfs_false, newnfs_true; extern nfstype nfsv34_type[9]; extern int nfsrv_useacl; extern char nfsv4_callbackaddr[INET6_ADDRSTRLEN]; NFSCLSTATEMUTEX; int nfstest_outofseq = 0; int nfscl_assumeposixlocks = 1; int nfscl_enablecallb = 0; short nfsv4_cbport = NFSV4_CBPORT; int nfstest_openallsetattr = 0; #endif /* !APPLEKEXT */ #define DIRHDSIZ (sizeof (struct dirent) - (MAXNAMLEN + 1)) static int nfsrpc_setattrrpc(vnode_t , struct vattr *, nfsv4stateid_t *, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *, void *); static int nfsrpc_readrpc(vnode_t , struct uio *, struct ucred *, nfsv4stateid_t *, NFSPROC_T *, struct nfsvattr *, int *, void *); static int nfsrpc_writerpc(vnode_t , struct uio *, int *, int *, struct ucred *, nfsv4stateid_t *, NFSPROC_T *, struct nfsvattr *, int *, void *); static int nfsrpc_createv23(vnode_t , char *, int, struct vattr *, nfsquad_t, int, struct ucred *, NFSPROC_T *, struct nfsvattr *, struct nfsvattr *, struct nfsfh **, int *, int *, void *); static int nfsrpc_createv4(vnode_t , char *, int, struct vattr *, nfsquad_t, int, struct nfsclowner *, struct nfscldeleg **, struct ucred *, NFSPROC_T *, struct nfsvattr *, struct nfsvattr *, struct nfsfh **, int *, int *, void *, int *); static int nfsrpc_locku(struct nfsrv_descript *, struct nfsmount *, struct nfscllockowner *, u_int64_t, u_int64_t, u_int32_t, struct ucred *, NFSPROC_T *, int); static int nfsrpc_setaclrpc(vnode_t, struct ucred *, NFSPROC_T *, struct acl *, nfsv4stateid_t *, void *); /* * nfs null call from vfs. */ APPLESTATIC int nfsrpc_null(vnode_t vp, struct ucred *cred, NFSPROC_T *p) { int error; struct nfsrv_descript nfsd, *nd = &nfsd; NFSCL_REQSTART(nd, NFSPROC_NULL, vp); error = nfscl_request(nd, vp, p, cred, NULL); if (nd->nd_repstat && !error) error = nd->nd_repstat; mbuf_freem(nd->nd_mrep); return (error); } /* * nfs access rpc op. * For nfs version 3 and 4, use the access rpc to check accessibility. If file * modes are changed on the server, accesses might still fail later. */ APPLESTATIC int nfsrpc_access(vnode_t vp, int acmode, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, int *attrflagp) { int error; u_int32_t mode, rmode; if (acmode & VREAD) mode = NFSACCESS_READ; else mode = 0; if (vnode_vtype(vp) == VDIR) { if (acmode & VWRITE) mode |= (NFSACCESS_MODIFY | NFSACCESS_EXTEND | NFSACCESS_DELETE); if (acmode & VEXEC) mode |= NFSACCESS_LOOKUP; } else { if (acmode & VWRITE) mode |= (NFSACCESS_MODIFY | NFSACCESS_EXTEND); if (acmode & VEXEC) mode |= NFSACCESS_EXECUTE; } /* * Now, just call nfsrpc_accessrpc() to do the actual RPC. */ error = nfsrpc_accessrpc(vp, mode, cred, p, nap, attrflagp, &rmode, NULL); /* * The NFS V3 spec does not clarify whether or not * the returned access bits can be a superset of * the ones requested, so... */ if (!error && (rmode & mode) != mode) error = EACCES; return (error); } /* * The actual rpc, separated out for Darwin. */ APPLESTATIC int nfsrpc_accessrpc(vnode_t vp, u_int32_t mode, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, int *attrflagp, u_int32_t *rmodep, void *stuff) { u_int32_t *tl; u_int32_t supported, rmode; int error; struct nfsrv_descript nfsd, *nd = &nfsd; nfsattrbit_t attrbits; *attrflagp = 0; supported = mode; NFSCL_REQSTART(nd, NFSPROC_ACCESS, vp); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(mode); if (nd->nd_flag & ND_NFSV4) { /* * And do a Getattr op. */ NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_GETATTR); NFSGETATTR_ATTRBIT(&attrbits); (void) nfsrv_putattrbit(nd, &attrbits); } error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (nd->nd_flag & ND_NFSV3) { error = nfscl_postop_attr(nd, nap, attrflagp, stuff); if (error) goto nfsmout; } if (!nd->nd_repstat) { if (nd->nd_flag & ND_NFSV4) { NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); supported = fxdr_unsigned(u_int32_t, *tl++); } else { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); } rmode = fxdr_unsigned(u_int32_t, *tl); if (nd->nd_flag & ND_NFSV4) error = nfscl_postop_attr(nd, nap, attrflagp, stuff); /* * It's not obvious what should be done about * unsupported access modes. For now, be paranoid * and clear the unsupported ones. */ rmode &= supported; *rmodep = rmode; } else error = nd->nd_repstat; nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * nfs open rpc */ APPLESTATIC int nfsrpc_open(vnode_t vp, int amode, struct ucred *cred, NFSPROC_T *p) { struct nfsclopen *op; struct nfscldeleg *dp; struct nfsfh *nfhp; struct nfsnode *np = VTONFS(vp); struct nfsmount *nmp = VFSTONFS(vnode_mount(vp)); u_int32_t mode, clidrev; int ret, newone, error, expireret = 0, retrycnt; /* * For NFSv4, Open Ops are only done on Regular Files. */ if (vnode_vtype(vp) != VREG) return (0); mode = 0; if (amode & FREAD) mode |= NFSV4OPEN_ACCESSREAD; if (amode & FWRITE) mode |= NFSV4OPEN_ACCESSWRITE; nfhp = np->n_fhp; retrycnt = 0; #ifdef notdef { char name[100]; int namel; namel = (np->n_v4->n4_namelen < 100) ? np->n_v4->n4_namelen : 99; bcopy(NFS4NODENAME(np->n_v4), name, namel); name[namel] = '\0'; printf("rpcopen p=0x%x name=%s",p->p_pid,name); if (nfhp->nfh_len > 0) printf(" fh=0x%x\n",nfhp->nfh_fh[12]); else printf(" fhl=0\n"); } #endif do { dp = NULL; error = nfscl_open(vp, nfhp->nfh_fh, nfhp->nfh_len, mode, 1, cred, p, NULL, &op, &newone, &ret, 1); if (error) { return (error); } if (nmp->nm_clp != NULL) clidrev = nmp->nm_clp->nfsc_clientidrev; else clidrev = 0; if (ret == NFSCLOPEN_DOOPEN) { if (np->n_v4 != NULL) { error = nfsrpc_openrpc(nmp, vp, np->n_v4->n4_data, np->n_v4->n4_fhlen, np->n_fhp->nfh_fh, np->n_fhp->nfh_len, mode, op, NFS4NODENAME(np->n_v4), np->n_v4->n4_namelen, &dp, 0, 0x0, cred, p, 0, 0); if (dp != NULL) { #ifdef APPLE OSBitAndAtomic((int32_t)~NDELEGMOD, (UInt32 *)&np->n_flag); #else NFSLOCKNODE(np); np->n_flag &= ~NDELEGMOD; /* * Invalidate the attribute cache, so that * attributes that pre-date the issue of a * delegation are not cached, since the * cached attributes will remain valid while * the delegation is held. */ NFSINVALATTRCACHE(np); NFSUNLOCKNODE(np); #endif (void) nfscl_deleg(nmp->nm_mountp, op->nfso_own->nfsow_clp, nfhp->nfh_fh, nfhp->nfh_len, cred, p, &dp); } } else { error = EIO; } newnfs_copyincred(cred, &op->nfso_cred); } else if (ret == NFSCLOPEN_SETCRED) /* * This is a new local open on a delegation. It needs * to have credentials so that an open can be done * against the server during recovery. */ newnfs_copyincred(cred, &op->nfso_cred); /* * nfso_opencnt is the count of how many VOP_OPEN()s have * been done on this Open successfully and a VOP_CLOSE() * is expected for each of these. * If error is non-zero, don't increment it, since the Open * hasn't succeeded yet. */ if (!error) op->nfso_opencnt++; nfscl_openrelease(op, error, newone); if (error == NFSERR_GRACE || error == NFSERR_STALECLIENTID || error == NFSERR_STALEDONTRECOVER || error == NFSERR_DELAY) { (void) nfs_catnap(PZERO, error, "nfs_open"); } else if ((error == NFSERR_EXPIRED || error == NFSERR_BADSTATEID) && clidrev != 0) { expireret = nfscl_hasexpired(nmp->nm_clp, clidrev, p); retrycnt++; } } while (error == NFSERR_GRACE || error == NFSERR_STALECLIENTID || error == NFSERR_STALEDONTRECOVER || error == NFSERR_DELAY || ((error == NFSERR_EXPIRED || error == NFSERR_BADSTATEID) && expireret == 0 && clidrev != 0 && retrycnt < 4)); if (error && retrycnt >= 4) error = EIO; return (error); } /* * the actual open rpc */ APPLESTATIC int nfsrpc_openrpc(struct nfsmount *nmp, vnode_t vp, u_int8_t *nfhp, int fhlen, u_int8_t *newfhp, int newfhlen, u_int32_t mode, struct nfsclopen *op, u_int8_t *name, int namelen, struct nfscldeleg **dpp, int reclaim, u_int32_t delegtype, struct ucred *cred, NFSPROC_T *p, int syscred, int recursed) { u_int32_t *tl; struct nfsrv_descript nfsd, *nd = &nfsd; struct nfscldeleg *dp, *ndp = NULL; struct nfsvattr nfsva; u_int32_t rflags, deleg; nfsattrbit_t attrbits; int error, ret, acesize, limitby; dp = *dpp; *dpp = NULL; nfscl_reqstart(nd, NFSPROC_OPEN, nmp, nfhp, fhlen, NULL); NFSM_BUILD(tl, u_int32_t *, 5 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(op->nfso_own->nfsow_seqid); *tl++ = txdr_unsigned(mode & NFSV4OPEN_ACCESSBOTH); *tl++ = txdr_unsigned((mode >> NFSLCK_SHIFT) & NFSV4OPEN_DENYBOTH); *tl++ = op->nfso_own->nfsow_clp->nfsc_clientid.lval[0]; *tl = op->nfso_own->nfsow_clp->nfsc_clientid.lval[1]; (void) nfsm_strtom(nd, op->nfso_own->nfsow_owner, NFSV4CL_LOCKNAMELEN); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV4OPEN_NOCREATE); if (reclaim) { *tl = txdr_unsigned(NFSV4OPEN_CLAIMPREVIOUS); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(delegtype); } else { if (dp != NULL) { *tl = txdr_unsigned(NFSV4OPEN_CLAIMDELEGATECUR); NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID); *tl++ = dp->nfsdl_stateid.seqid; *tl++ = dp->nfsdl_stateid.other[0]; *tl++ = dp->nfsdl_stateid.other[1]; *tl = dp->nfsdl_stateid.other[2]; } else { *tl = txdr_unsigned(NFSV4OPEN_CLAIMNULL); } (void) nfsm_strtom(nd, name, namelen); } NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_GETATTR); NFSZERO_ATTRBIT(&attrbits); NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_CHANGE); NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_TIMEMODIFY); (void) nfsrv_putattrbit(nd, &attrbits); if (syscred) nd->nd_flag |= ND_USEGSSNAME; error = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, vp, p, cred, NFS_PROG, NFS_VER4, NULL, 1, NULL); if (error) return (error); NFSCL_INCRSEQID(op->nfso_own->nfsow_seqid, nd); if (!nd->nd_repstat) { NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID + 6 * NFSX_UNSIGNED); op->nfso_stateid.seqid = *tl++; op->nfso_stateid.other[0] = *tl++; op->nfso_stateid.other[1] = *tl++; op->nfso_stateid.other[2] = *tl; rflags = fxdr_unsigned(u_int32_t, *(tl + 6)); error = nfsrv_getattrbits(nd, &attrbits, NULL, NULL); if (error) goto nfsmout; NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); deleg = fxdr_unsigned(u_int32_t, *tl); if (deleg == NFSV4OPEN_DELEGATEREAD || deleg == NFSV4OPEN_DELEGATEWRITE) { if (!(op->nfso_own->nfsow_clp->nfsc_flags & NFSCLFLAGS_FIRSTDELEG)) op->nfso_own->nfsow_clp->nfsc_flags |= (NFSCLFLAGS_FIRSTDELEG | NFSCLFLAGS_GOTDELEG); MALLOC(ndp, struct nfscldeleg *, sizeof (struct nfscldeleg) + newfhlen, M_NFSCLDELEG, M_WAITOK); LIST_INIT(&ndp->nfsdl_owner); LIST_INIT(&ndp->nfsdl_lock); ndp->nfsdl_clp = op->nfso_own->nfsow_clp; ndp->nfsdl_fhlen = newfhlen; NFSBCOPY(newfhp, ndp->nfsdl_fh, newfhlen); newnfs_copyincred(cred, &ndp->nfsdl_cred); nfscl_lockinit(&ndp->nfsdl_rwlock); NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID + NFSX_UNSIGNED); ndp->nfsdl_stateid.seqid = *tl++; ndp->nfsdl_stateid.other[0] = *tl++; ndp->nfsdl_stateid.other[1] = *tl++; ndp->nfsdl_stateid.other[2] = *tl++; ret = fxdr_unsigned(int, *tl); if (deleg == NFSV4OPEN_DELEGATEWRITE) { ndp->nfsdl_flags = NFSCLDL_WRITE; /* * Indicates how much the file can grow. */ NFSM_DISSECT(tl, u_int32_t *, 3 * NFSX_UNSIGNED); limitby = fxdr_unsigned(int, *tl++); switch (limitby) { case NFSV4OPEN_LIMITSIZE: ndp->nfsdl_sizelimit = fxdr_hyper(tl); break; case NFSV4OPEN_LIMITBLOCKS: ndp->nfsdl_sizelimit = fxdr_unsigned(u_int64_t, *tl++); ndp->nfsdl_sizelimit *= fxdr_unsigned(u_int64_t, *tl); break; default: error = NFSERR_BADXDR; goto nfsmout; }; } else { ndp->nfsdl_flags = NFSCLDL_READ; } if (ret) ndp->nfsdl_flags |= NFSCLDL_RECALL; error = nfsrv_dissectace(nd, &ndp->nfsdl_ace, &ret, &acesize, p); if (error) goto nfsmout; } else if (deleg != NFSV4OPEN_DELEGATENONE) { error = NFSERR_BADXDR; goto nfsmout; } NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); error = nfsv4_loadattr(nd, NULL, &nfsva, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, p, cred); if (error) goto nfsmout; if (ndp != NULL) { ndp->nfsdl_change = nfsva.na_filerev; ndp->nfsdl_modtime = nfsva.na_mtime; ndp->nfsdl_flags |= NFSCLDL_MODTIMESET; } if (!reclaim && (rflags & NFSV4OPEN_RESULTCONFIRM)) { do { ret = nfsrpc_openconfirm(vp, newfhp, newfhlen, op, cred, p); if (ret == NFSERR_DELAY) (void) nfs_catnap(PZERO, ret, "nfs_open"); } while (ret == NFSERR_DELAY); error = ret; } if ((rflags & NFSV4OPEN_LOCKTYPEPOSIX) || nfscl_assumeposixlocks) op->nfso_posixlock = 1; else op->nfso_posixlock = 0; /* * If the server is handing out delegations, but we didn't * get one because an OpenConfirm was required, try the * Open again, to get a delegation. This is a harmless no-op, * from a server's point of view. */ if (!reclaim && (rflags & NFSV4OPEN_RESULTCONFIRM) && (op->nfso_own->nfsow_clp->nfsc_flags & NFSCLFLAGS_GOTDELEG) && !error && dp == NULL && ndp == NULL && !recursed) { do { ret = nfsrpc_openrpc(nmp, vp, nfhp, fhlen, newfhp, newfhlen, mode, op, name, namelen, &ndp, 0, 0x0, cred, p, syscred, 1); if (ret == NFSERR_DELAY) (void) nfs_catnap(PZERO, ret, "nfs_open2"); } while (ret == NFSERR_DELAY); if (ret) { if (ndp != NULL) FREE((caddr_t)ndp, M_NFSCLDELEG); if (ret == NFSERR_STALECLIENTID || ret == NFSERR_STALEDONTRECOVER) error = ret; } } } if (nd->nd_repstat != 0 && error == 0) error = nd->nd_repstat; if (error == NFSERR_STALECLIENTID) nfscl_initiate_recovery(op->nfso_own->nfsow_clp); nfsmout: if (!error) *dpp = ndp; else if (ndp != NULL) FREE((caddr_t)ndp, M_NFSCLDELEG); mbuf_freem(nd->nd_mrep); return (error); } /* * open downgrade rpc */ APPLESTATIC int nfsrpc_opendowngrade(vnode_t vp, u_int32_t mode, struct nfsclopen *op, struct ucred *cred, NFSPROC_T *p) { u_int32_t *tl; struct nfsrv_descript nfsd, *nd = &nfsd; int error; NFSCL_REQSTART(nd, NFSPROC_OPENDOWNGRADE, vp); NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID + 3 * NFSX_UNSIGNED); *tl++ = op->nfso_stateid.seqid; *tl++ = op->nfso_stateid.other[0]; *tl++ = op->nfso_stateid.other[1]; *tl++ = op->nfso_stateid.other[2]; *tl++ = txdr_unsigned(op->nfso_own->nfsow_seqid); *tl++ = txdr_unsigned(mode & NFSV4OPEN_ACCESSBOTH); *tl = txdr_unsigned((mode >> NFSLCK_SHIFT) & NFSV4OPEN_DENYBOTH); error = nfscl_request(nd, vp, p, cred, NULL); if (error) return (error); NFSCL_INCRSEQID(op->nfso_own->nfsow_seqid, nd); if (!nd->nd_repstat) { NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID); op->nfso_stateid.seqid = *tl++; op->nfso_stateid.other[0] = *tl++; op->nfso_stateid.other[1] = *tl++; op->nfso_stateid.other[2] = *tl; } if (nd->nd_repstat && error == 0) error = nd->nd_repstat; if (error == NFSERR_STALESTATEID) nfscl_initiate_recovery(op->nfso_own->nfsow_clp); nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * V4 Close operation. */ APPLESTATIC int nfsrpc_close(vnode_t vp, int doclose, NFSPROC_T *p) { struct nfsclclient *clp; int error; if (vnode_vtype(vp) != VREG) return (0); if (doclose) error = nfscl_doclose(vp, &clp, p); else error = nfscl_getclose(vp, &clp); if (error) return (error); nfscl_clientrelease(clp); return (0); } /* * Close the open. */ APPLESTATIC void nfsrpc_doclose(struct nfsmount *nmp, struct nfsclopen *op, NFSPROC_T *p) { struct nfsrv_descript nfsd, *nd = &nfsd; struct nfscllockowner *lp; struct nfscllock *lop, *nlop; struct ucred *tcred; u_int64_t off = 0, len = 0; u_int32_t type = NFSV4LOCKT_READ; int error, do_unlock, trycnt; tcred = newnfs_getcred(); newnfs_copycred(&op->nfso_cred, tcred); /* * (Theoretically this could be done in the same * compound as the close, but having multiple * sequenced Ops in the same compound might be * too scary for some servers.) */ if (op->nfso_posixlock) { off = 0; len = NFS64BITSSET; type = NFSV4LOCKT_READ; } /* * Since this function is only called from VOP_INACTIVE(), no * other thread will be manipulating this Open. As such, the * lock lists are not being changed by other threads, so it should * be safe to do this without locking. */ LIST_FOREACH(lp, &op->nfso_lock, nfsl_list) { do_unlock = 1; LIST_FOREACH_SAFE(lop, &lp->nfsl_lock, nfslo_list, nlop) { if (op->nfso_posixlock == 0) { off = lop->nfslo_first; len = lop->nfslo_end - lop->nfslo_first; if (lop->nfslo_type == F_WRLCK) type = NFSV4LOCKT_WRITE; else type = NFSV4LOCKT_READ; } if (do_unlock) { trycnt = 0; do { error = nfsrpc_locku(nd, nmp, lp, off, len, type, tcred, p, 0); if ((nd->nd_repstat == NFSERR_GRACE || nd->nd_repstat == NFSERR_DELAY) && error == 0) (void) nfs_catnap(PZERO, (int)nd->nd_repstat, "nfs_close"); } while ((nd->nd_repstat == NFSERR_GRACE || nd->nd_repstat == NFSERR_DELAY) && error == 0 && trycnt++ < 5); if (op->nfso_posixlock) do_unlock = 0; } nfscl_freelock(lop, 0); } } /* * There could be other Opens for different files on the same * OpenOwner, so locking is required. */ NFSLOCKCLSTATE(); nfscl_lockexcl(&op->nfso_own->nfsow_rwlock, NFSCLSTATEMUTEXPTR); NFSUNLOCKCLSTATE(); do { error = nfscl_tryclose(op, tcred, nmp, p); if (error == NFSERR_GRACE) (void) nfs_catnap(PZERO, error, "nfs_close"); } while (error == NFSERR_GRACE); NFSLOCKCLSTATE(); nfscl_lockunlock(&op->nfso_own->nfsow_rwlock); /* * Move the lockowner to nfsc_defunctlockowner, * so the Renew thread will do the ReleaseLockOwner * Op on it later. There might still be other * opens using the same lockowner name. */ lp = LIST_FIRST(&op->nfso_lock); if (lp != NULL) { while (LIST_NEXT(lp, nfsl_list) != NULL) lp = LIST_NEXT(lp, nfsl_list); LIST_PREPEND(&nmp->nm_clp->nfsc_defunctlockowner, &op->nfso_lock, lp, nfsl_list); LIST_INIT(&op->nfso_lock); } nfscl_freeopen(op, 0); NFSUNLOCKCLSTATE(); NFSFREECRED(tcred); } /* * The actual Close RPC. */ APPLESTATIC int nfsrpc_closerpc(struct nfsrv_descript *nd, struct nfsmount *nmp, struct nfsclopen *op, struct ucred *cred, NFSPROC_T *p, int syscred) { u_int32_t *tl; int error; nfscl_reqstart(nd, NFSPROC_CLOSE, nmp, op->nfso_fh, op->nfso_fhlen, NULL); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED + NFSX_STATEID); *tl++ = txdr_unsigned(op->nfso_own->nfsow_seqid); *tl++ = op->nfso_stateid.seqid; *tl++ = op->nfso_stateid.other[0]; *tl++ = op->nfso_stateid.other[1]; *tl = op->nfso_stateid.other[2]; if (syscred) nd->nd_flag |= ND_USEGSSNAME; error = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, NULL, p, cred, NFS_PROG, NFS_VER4, NULL, 1, NULL); if (error) return (error); NFSCL_INCRSEQID(op->nfso_own->nfsow_seqid, nd); if (nd->nd_repstat == 0) NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID); error = nd->nd_repstat; if (error == NFSERR_STALESTATEID) nfscl_initiate_recovery(op->nfso_own->nfsow_clp); nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * V4 Open Confirm RPC. */ APPLESTATIC int nfsrpc_openconfirm(vnode_t vp, u_int8_t *nfhp, int fhlen, struct nfsclopen *op, struct ucred *cred, NFSPROC_T *p) { u_int32_t *tl; struct nfsrv_descript nfsd, *nd = &nfsd; int error; nfscl_reqstart(nd, NFSPROC_OPENCONFIRM, VFSTONFS(vnode_mount(vp)), nfhp, fhlen, NULL); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED + NFSX_STATEID); *tl++ = op->nfso_stateid.seqid; *tl++ = op->nfso_stateid.other[0]; *tl++ = op->nfso_stateid.other[1]; *tl++ = op->nfso_stateid.other[2]; *tl = txdr_unsigned(op->nfso_own->nfsow_seqid); error = nfscl_request(nd, vp, p, cred, NULL); if (error) return (error); NFSCL_INCRSEQID(op->nfso_own->nfsow_seqid, nd); if (!nd->nd_repstat) { NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID); op->nfso_stateid.seqid = *tl++; op->nfso_stateid.other[0] = *tl++; op->nfso_stateid.other[1] = *tl++; op->nfso_stateid.other[2] = *tl; } error = nd->nd_repstat; if (error == NFSERR_STALESTATEID) nfscl_initiate_recovery(op->nfso_own->nfsow_clp); nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * Do the setclientid and setclientid confirm RPCs. Called from nfs_statfs() * when a mount has just occurred and when the server replies NFSERR_EXPIRED. */ APPLESTATIC int nfsrpc_setclient(struct nfsmount *nmp, struct nfsclclient *clp, struct ucred *cred, NFSPROC_T *p) { u_int32_t *tl; struct nfsrv_descript nfsd; struct nfsrv_descript *nd = &nfsd; nfsattrbit_t attrbits; u_int8_t *cp = NULL, *cp2, addr[INET6_ADDRSTRLEN + 9]; u_short port; int error, isinet6 = 0, callblen; nfsquad_t confirm; u_int32_t lease; static u_int32_t rev = 0; if (nfsboottime.tv_sec == 0) NFSSETBOOTTIME(nfsboottime); nfscl_reqstart(nd, NFSPROC_SETCLIENTID, nmp, NULL, 0, NULL); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(nfsboottime.tv_sec); *tl = txdr_unsigned(rev++); (void) nfsm_strtom(nd, clp->nfsc_id, clp->nfsc_idlen); /* * set up the callback address */ NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFS_CALLBCKPROG); callblen = strlen(nfsv4_callbackaddr); if (callblen == 0) cp = nfscl_getmyip(nmp, &isinet6); if (nfscl_enablecallb && nfs_numnfscbd > 0 && (callblen > 0 || cp != NULL)) { port = htons(nfsv4_cbport); cp2 = (u_int8_t *)&port; #ifdef INET6 if ((callblen > 0 && strchr(nfsv4_callbackaddr, ':')) || isinet6) { char ip6buf[INET6_ADDRSTRLEN], *ip6add; (void) nfsm_strtom(nd, "tcp6", 4); if (callblen == 0) { ip6_sprintf(ip6buf, (struct in6_addr *)cp); ip6add = ip6buf; } else { ip6add = nfsv4_callbackaddr; } snprintf(addr, INET6_ADDRSTRLEN + 9, "%s.%d.%d", ip6add, cp2[0], cp2[1]); } else #endif { (void) nfsm_strtom(nd, "tcp", 3); if (callblen == 0) snprintf(addr, INET6_ADDRSTRLEN + 9, "%d.%d.%d.%d.%d.%d", cp[0], cp[1], cp[2], cp[3], cp2[0], cp2[1]); else snprintf(addr, INET6_ADDRSTRLEN + 9, "%s.%d.%d", nfsv4_callbackaddr, cp2[0], cp2[1]); } (void) nfsm_strtom(nd, addr, strlen(addr)); } else { (void) nfsm_strtom(nd, "tcp", 3); (void) nfsm_strtom(nd, "0.0.0.0.0.0", 11); } NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(clp->nfsc_cbident); nd->nd_flag |= ND_USEGSSNAME; error = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, NULL, p, cred, NFS_PROG, NFS_VER4, NULL, 1, NULL); if (error) return (error); if (nd->nd_repstat == 0) { NFSM_DISSECT(tl, u_int32_t *, 4 * NFSX_UNSIGNED); clp->nfsc_clientid.lval[0] = *tl++; clp->nfsc_clientid.lval[1] = *tl++; confirm.lval[0] = *tl++; confirm.lval[1] = *tl; mbuf_freem(nd->nd_mrep); nd->nd_mrep = NULL; /* * and confirm it. */ nfscl_reqstart(nd, NFSPROC_SETCLIENTIDCFRM, nmp, NULL, 0, NULL); NFSM_BUILD(tl, u_int32_t *, 4 * NFSX_UNSIGNED); *tl++ = clp->nfsc_clientid.lval[0]; *tl++ = clp->nfsc_clientid.lval[1]; *tl++ = confirm.lval[0]; *tl = confirm.lval[1]; nd->nd_flag |= ND_USEGSSNAME; error = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, NULL, p, cred, NFS_PROG, NFS_VER4, NULL, 1, NULL); if (error) return (error); mbuf_freem(nd->nd_mrep); nd->nd_mrep = NULL; if (nd->nd_repstat == 0) { nfscl_reqstart(nd, NFSPROC_GETATTR, nmp, nmp->nm_fh, nmp->nm_fhsize, NULL); NFSZERO_ATTRBIT(&attrbits); NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_LEASETIME); (void) nfsrv_putattrbit(nd, &attrbits); nd->nd_flag |= ND_USEGSSNAME; error = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, NULL, p, cred, NFS_PROG, NFS_VER4, NULL, 1, NULL); if (error) return (error); if (nd->nd_repstat == 0) { error = nfsv4_loadattr(nd, NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, NULL, &lease, NULL, p, cred); if (error) goto nfsmout; clp->nfsc_renew = NFSCL_RENEW(lease); clp->nfsc_expire = NFSD_MONOSEC + clp->nfsc_renew; clp->nfsc_clientidrev++; if (clp->nfsc_clientidrev == 0) clp->nfsc_clientidrev++; } } } error = nd->nd_repstat; nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * nfs getattr call. */ APPLESTATIC int nfsrpc_getattr(vnode_t vp, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, void *stuff) { struct nfsrv_descript nfsd, *nd = &nfsd; int error; nfsattrbit_t attrbits; NFSCL_REQSTART(nd, NFSPROC_GETATTR, vp); if (nd->nd_flag & ND_NFSV4) { NFSGETATTR_ATTRBIT(&attrbits); (void) nfsrv_putattrbit(nd, &attrbits); } error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (!nd->nd_repstat) error = nfsm_loadattr(nd, nap); else error = nd->nd_repstat; mbuf_freem(nd->nd_mrep); return (error); } /* * nfs getattr call with non-vnode arguemnts. */ APPLESTATIC int nfsrpc_getattrnovp(struct nfsmount *nmp, u_int8_t *fhp, int fhlen, int syscred, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, u_int64_t *xidp) { struct nfsrv_descript nfsd, *nd = &nfsd; int error, vers = NFS_VER2; nfsattrbit_t attrbits; nfscl_reqstart(nd, NFSPROC_GETATTR, nmp, fhp, fhlen, NULL); if (nd->nd_flag & ND_NFSV4) { vers = NFS_VER4; NFSGETATTR_ATTRBIT(&attrbits); (void) nfsrv_putattrbit(nd, &attrbits); } else if (nd->nd_flag & ND_NFSV3) { vers = NFS_VER3; } if (syscred) nd->nd_flag |= ND_USEGSSNAME; error = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, NULL, p, cred, NFS_PROG, vers, NULL, 1, xidp); if (error) return (error); if (!nd->nd_repstat) error = nfsm_loadattr(nd, nap); else error = nd->nd_repstat; mbuf_freem(nd->nd_mrep); return (error); } /* * Do an nfs setattr operation. */ APPLESTATIC int nfsrpc_setattr(vnode_t vp, struct vattr *vap, NFSACL_T *aclp, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *rnap, int *attrflagp, void *stuff) { int error, expireret = 0, openerr, retrycnt; u_int32_t clidrev = 0, mode; struct nfsmount *nmp = VFSTONFS(vnode_mount(vp)); struct nfsfh *nfhp; nfsv4stateid_t stateid; void *lckp; if (nmp->nm_clp != NULL) clidrev = nmp->nm_clp->nfsc_clientidrev; if (vap != NULL && NFSATTRISSET(u_quad_t, vap, va_size)) mode = NFSV4OPEN_ACCESSWRITE; else mode = NFSV4OPEN_ACCESSREAD; retrycnt = 0; do { lckp = NULL; openerr = 1; if (NFSHASNFSV4(nmp)) { nfhp = VTONFS(vp)->n_fhp; error = nfscl_getstateid(vp, nfhp->nfh_fh, nfhp->nfh_len, mode, cred, p, &stateid, &lckp); if (error && vnode_vtype(vp) == VREG && (mode == NFSV4OPEN_ACCESSWRITE || nfstest_openallsetattr)) { /* * No Open stateid, so try and open the file * now. */ if (mode == NFSV4OPEN_ACCESSWRITE) openerr = nfsrpc_open(vp, FWRITE, cred, p); else openerr = nfsrpc_open(vp, FREAD, cred, p); if (!openerr) (void) nfscl_getstateid(vp, nfhp->nfh_fh, nfhp->nfh_len, mode, cred, p, &stateid, &lckp); } } if (vap != NULL) error = nfsrpc_setattrrpc(vp, vap, &stateid, cred, p, rnap, attrflagp, stuff); else error = nfsrpc_setaclrpc(vp, cred, p, aclp, &stateid, stuff); if (error == NFSERR_STALESTATEID) nfscl_initiate_recovery(nmp->nm_clp); if (lckp != NULL) nfscl_lockderef(lckp); if (!openerr) (void) nfsrpc_close(vp, 0, p); if (error == NFSERR_GRACE || error == NFSERR_STALESTATEID || error == NFSERR_STALEDONTRECOVER || error == NFSERR_DELAY || error == NFSERR_OLDSTATEID) { (void) nfs_catnap(PZERO, error, "nfs_setattr"); } else if ((error == NFSERR_EXPIRED || error == NFSERR_BADSTATEID) && clidrev != 0) { expireret = nfscl_hasexpired(nmp->nm_clp, clidrev, p); } retrycnt++; } while (error == NFSERR_GRACE || error == NFSERR_STALESTATEID || error == NFSERR_STALEDONTRECOVER || error == NFSERR_DELAY || (error == NFSERR_OLDSTATEID && retrycnt < 20) || ((error == NFSERR_EXPIRED || error == NFSERR_BADSTATEID) && expireret == 0 && clidrev != 0 && retrycnt < 4)); if (error && retrycnt >= 4) error = EIO; return (error); } static int nfsrpc_setattrrpc(vnode_t vp, struct vattr *vap, nfsv4stateid_t *stateidp, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *rnap, int *attrflagp, void *stuff) { u_int32_t *tl; struct nfsrv_descript nfsd, *nd = &nfsd; int error; nfsattrbit_t attrbits; *attrflagp = 0; NFSCL_REQSTART(nd, NFSPROC_SETATTR, vp); if (nd->nd_flag & ND_NFSV4) nfsm_stateidtom(nd, stateidp, NFSSTATEID_PUTSTATEID); vap->va_type = vnode_vtype(vp); nfscl_fillsattr(nd, vap, vp, NFSSATTR_FULL, 0); if (nd->nd_flag & ND_NFSV3) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_false; } else if (nd->nd_flag & ND_NFSV4) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_GETATTR); NFSGETATTR_ATTRBIT(&attrbits); (void) nfsrv_putattrbit(nd, &attrbits); } error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (nd->nd_flag & (ND_NFSV3 | ND_NFSV4)) error = nfscl_wcc_data(nd, vp, rnap, attrflagp, NULL, stuff); if ((nd->nd_flag & ND_NFSV4) && !error) error = nfsrv_getattrbits(nd, &attrbits, NULL, NULL); if (!(nd->nd_flag & ND_NFSV3) && !nd->nd_repstat && !error) error = nfscl_postop_attr(nd, rnap, attrflagp, stuff); mbuf_freem(nd->nd_mrep); if (nd->nd_repstat && !error) error = nd->nd_repstat; return (error); } /* * nfs lookup rpc */ APPLESTATIC int nfsrpc_lookup(vnode_t dvp, char *name, int len, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *dnap, struct nfsvattr *nap, struct nfsfh **nfhpp, int *attrflagp, int *dattrflagp, void *stuff) { u_int32_t *tl; struct nfsrv_descript nfsd, *nd = &nfsd; struct nfsmount *nmp; struct nfsnode *np; struct nfsfh *nfhp; nfsattrbit_t attrbits; int error = 0, lookupp = 0; *attrflagp = 0; *dattrflagp = 0; if (vnode_vtype(dvp) != VDIR) return (ENOTDIR); nmp = VFSTONFS(vnode_mount(dvp)); if (len > NFS_MAXNAMLEN) return (ENAMETOOLONG); if (NFSHASNFSV4(nmp) && len == 1 && name[0] == '.') { /* * Just return the current dir's fh. */ np = VTONFS(dvp); MALLOC(nfhp, struct nfsfh *, sizeof (struct nfsfh) + np->n_fhp->nfh_len, M_NFSFH, M_WAITOK); nfhp->nfh_len = np->n_fhp->nfh_len; NFSBCOPY(np->n_fhp->nfh_fh, nfhp->nfh_fh, nfhp->nfh_len); *nfhpp = nfhp; return (0); } if (NFSHASNFSV4(nmp) && len == 2 && name[0] == '.' && name[1] == '.') { lookupp = 1; NFSCL_REQSTART(nd, NFSPROC_LOOKUPP, dvp); } else { NFSCL_REQSTART(nd, NFSPROC_LOOKUP, dvp); (void) nfsm_strtom(nd, name, len); } if (nd->nd_flag & ND_NFSV4) { NFSGETATTR_ATTRBIT(&attrbits); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV4OP_GETFH); *tl = txdr_unsigned(NFSV4OP_GETATTR); (void) nfsrv_putattrbit(nd, &attrbits); } error = nfscl_request(nd, dvp, p, cred, stuff); if (error) return (error); if (nd->nd_repstat) { /* * When an NFSv4 Lookupp returns ENOENT, it means that * the lookup is at the root of an fs, so return this dir. */ if (nd->nd_repstat == NFSERR_NOENT && lookupp) { np = VTONFS(dvp); MALLOC(nfhp, struct nfsfh *, sizeof (struct nfsfh) + np->n_fhp->nfh_len, M_NFSFH, M_WAITOK); nfhp->nfh_len = np->n_fhp->nfh_len; NFSBCOPY(np->n_fhp->nfh_fh, nfhp->nfh_fh, nfhp->nfh_len); *nfhpp = nfhp; mbuf_freem(nd->nd_mrep); return (0); } if (nd->nd_flag & ND_NFSV3) error = nfscl_postop_attr(nd, dnap, dattrflagp, stuff); goto nfsmout; } if ((nd->nd_flag & (ND_NFSV4 | ND_NOMOREDATA)) == ND_NFSV4) { NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); if (*(tl + 1)) { nd->nd_flag |= ND_NOMOREDATA; goto nfsmout; } } error = nfsm_getfh(nd, nfhpp); if (error) goto nfsmout; error = nfscl_postop_attr(nd, nap, attrflagp, stuff); if ((nd->nd_flag & ND_NFSV3) && !error) error = nfscl_postop_attr(nd, dnap, dattrflagp, stuff); nfsmout: mbuf_freem(nd->nd_mrep); if (!error && nd->nd_repstat) error = nd->nd_repstat; return (error); } /* * Do a readlink rpc. */ APPLESTATIC int nfsrpc_readlink(vnode_t vp, struct uio *uiop, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, int *attrflagp, void *stuff) { u_int32_t *tl; struct nfsrv_descript nfsd, *nd = &nfsd; struct nfsnode *np = VTONFS(vp); nfsattrbit_t attrbits; int error, len, cangetattr = 1; *attrflagp = 0; NFSCL_REQSTART(nd, NFSPROC_READLINK, vp); if (nd->nd_flag & ND_NFSV4) { /* * And do a Getattr op. */ NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_GETATTR); NFSGETATTR_ATTRBIT(&attrbits); (void) nfsrv_putattrbit(nd, &attrbits); } error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (nd->nd_flag & ND_NFSV3) error = nfscl_postop_attr(nd, nap, attrflagp, stuff); if (!nd->nd_repstat && !error) { NFSM_STRSIZ(len, NFS_MAXPATHLEN); /* * This seems weird to me, but must have been added to * FreeBSD for some reason. The only thing I can think of * is that there was/is some server that replies with * more link data than it should? */ if (len == NFS_MAXPATHLEN) { NFSLOCKNODE(np); if (np->n_size > 0 && np->n_size < NFS_MAXPATHLEN) { len = np->n_size; cangetattr = 0; } NFSUNLOCKNODE(np); } error = nfsm_mbufuio(nd, uiop, len); if ((nd->nd_flag & ND_NFSV4) && !error && cangetattr) error = nfscl_postop_attr(nd, nap, attrflagp, stuff); } if (nd->nd_repstat && !error) error = nd->nd_repstat; nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * Read operation. */ APPLESTATIC int nfsrpc_read(vnode_t vp, struct uio *uiop, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, int *attrflagp, void *stuff) { int error, expireret = 0, retrycnt; u_int32_t clidrev = 0; struct nfsmount *nmp = VFSTONFS(vnode_mount(vp)); struct nfsnode *np = VTONFS(vp); struct ucred *newcred; struct nfsfh *nfhp = NULL; nfsv4stateid_t stateid; void *lckp; if (nmp->nm_clp != NULL) clidrev = nmp->nm_clp->nfsc_clientidrev; newcred = cred; if (NFSHASNFSV4(nmp)) { nfhp = np->n_fhp; if (p == NULL) newcred = NFSNEWCRED(cred); } retrycnt = 0; do { lckp = NULL; if (NFSHASNFSV4(nmp)) (void)nfscl_getstateid(vp, nfhp->nfh_fh, nfhp->nfh_len, NFSV4OPEN_ACCESSREAD, newcred, p, &stateid, &lckp); error = nfsrpc_readrpc(vp, uiop, newcred, &stateid, p, nap, attrflagp, stuff); if (error == NFSERR_STALESTATEID) nfscl_initiate_recovery(nmp->nm_clp); if (lckp != NULL) nfscl_lockderef(lckp); if (error == NFSERR_GRACE || error == NFSERR_STALESTATEID || error == NFSERR_STALEDONTRECOVER || error == NFSERR_DELAY || error == NFSERR_OLDSTATEID) { (void) nfs_catnap(PZERO, error, "nfs_read"); } else if ((error == NFSERR_EXPIRED || error == NFSERR_BADSTATEID) && clidrev != 0) { expireret = nfscl_hasexpired(nmp->nm_clp, clidrev, p); } retrycnt++; } while (error == NFSERR_GRACE || error == NFSERR_STALESTATEID || error == NFSERR_STALEDONTRECOVER || error == NFSERR_DELAY || (error == NFSERR_OLDSTATEID && retrycnt < 20) || ((error == NFSERR_EXPIRED || error == NFSERR_BADSTATEID) && expireret == 0 && clidrev != 0 && retrycnt < 4)); if (error && retrycnt >= 4) error = EIO; if (NFSHASNFSV4(nmp) && p == NULL) NFSFREECRED(newcred); return (error); } /* * The actual read RPC. */ static int nfsrpc_readrpc(vnode_t vp, struct uio *uiop, struct ucred *cred, nfsv4stateid_t *stateidp, NFSPROC_T *p, struct nfsvattr *nap, int *attrflagp, void *stuff) { u_int32_t *tl; int error = 0, len, retlen, tsiz, eof = 0; struct nfsrv_descript nfsd; struct nfsmount *nmp = VFSTONFS(vnode_mount(vp)); struct nfsrv_descript *nd = &nfsd; int rsize; off_t tmp_off; *attrflagp = 0; tsiz = uio_uio_resid(uiop); tmp_off = uiop->uio_offset + tsiz; NFSLOCKMNT(nmp); if (tmp_off > nmp->nm_maxfilesize || tmp_off < uiop->uio_offset) { NFSUNLOCKMNT(nmp); return (EFBIG); } rsize = nmp->nm_rsize; NFSUNLOCKMNT(nmp); nd->nd_mrep = NULL; while (tsiz > 0) { *attrflagp = 0; len = (tsiz > rsize) ? rsize : tsiz; NFSCL_REQSTART(nd, NFSPROC_READ, vp); if (nd->nd_flag & ND_NFSV4) nfsm_stateidtom(nd, stateidp, NFSSTATEID_PUTSTATEID); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED * 3); if (nd->nd_flag & ND_NFSV2) { *tl++ = txdr_unsigned(uiop->uio_offset); *tl++ = txdr_unsigned(len); *tl = 0; } else { txdr_hyper(uiop->uio_offset, tl); *(tl + 2) = txdr_unsigned(len); } /* * Since I can't do a Getattr for NFSv4 for Write, there * doesn't seem any point in doing one here, either. * (See the comment in nfsrpc_writerpc() for more info.) */ error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (nd->nd_flag & ND_NFSV3) { error = nfscl_postop_attr(nd, nap, attrflagp, stuff); } else if (!nd->nd_repstat && (nd->nd_flag & ND_NFSV2)) { error = nfsm_loadattr(nd, nap); if (!error) *attrflagp = 1; } if (nd->nd_repstat || error) { if (!error) error = nd->nd_repstat; goto nfsmout; } if (nd->nd_flag & ND_NFSV3) { NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); eof = fxdr_unsigned(int, *(tl + 1)); } else if (nd->nd_flag & ND_NFSV4) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); eof = fxdr_unsigned(int, *tl); } NFSM_STRSIZ(retlen, rsize); error = nfsm_mbufuio(nd, uiop, retlen); if (error) goto nfsmout; mbuf_freem(nd->nd_mrep); nd->nd_mrep = NULL; tsiz -= retlen; if (!(nd->nd_flag & ND_NFSV2)) { if (eof || retlen == 0) tsiz = 0; } else if (retlen < len) tsiz = 0; } return (0); nfsmout: if (nd->nd_mrep != NULL) mbuf_freem(nd->nd_mrep); return (error); } /* * nfs write operation * When called_from_strategy != 0, it should return EIO for an error that * indicates recovery is in progress, so that the buffer will be left * dirty and be written back to the server later. If it loops around, * the recovery thread could get stuck waiting for the buffer and recovery * will then deadlock. */ APPLESTATIC int nfsrpc_write(vnode_t vp, struct uio *uiop, int *iomode, int *must_commit, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, int *attrflagp, void *stuff, int called_from_strategy) { int error, expireret = 0, retrycnt, nostateid; u_int32_t clidrev = 0; struct nfsmount *nmp = VFSTONFS(vnode_mount(vp)); struct nfsnode *np = VTONFS(vp); struct ucred *newcred; struct nfsfh *nfhp = NULL; nfsv4stateid_t stateid; void *lckp; *must_commit = 0; if (nmp->nm_clp != NULL) clidrev = nmp->nm_clp->nfsc_clientidrev; newcred = cred; if (NFSHASNFSV4(nmp)) { if (p == NULL) newcred = NFSNEWCRED(cred); nfhp = np->n_fhp; } retrycnt = 0; do { lckp = NULL; nostateid = 0; if (NFSHASNFSV4(nmp)) { (void)nfscl_getstateid(vp, nfhp->nfh_fh, nfhp->nfh_len, NFSV4OPEN_ACCESSWRITE, newcred, p, &stateid, &lckp); if (stateid.other[0] == 0 && stateid.other[1] == 0 && stateid.other[2] == 0) { nostateid = 1; printf("stateid0 in write\n"); } } /* * If there is no stateid for NFSv4, it means this is an * extraneous write after close. Basically a poorly * implemented buffer cache. Just don't do the write. */ if (nostateid) error = 0; else error = nfsrpc_writerpc(vp, uiop, iomode, must_commit, newcred, &stateid, p, nap, attrflagp, stuff); if (error == NFSERR_STALESTATEID) nfscl_initiate_recovery(nmp->nm_clp); if (lckp != NULL) nfscl_lockderef(lckp); if (error == NFSERR_GRACE || error == NFSERR_STALESTATEID || error == NFSERR_STALEDONTRECOVER || error == NFSERR_DELAY || error == NFSERR_OLDSTATEID) { (void) nfs_catnap(PZERO, error, "nfs_write"); } else if ((error == NFSERR_EXPIRED || error == NFSERR_BADSTATEID) && clidrev != 0) { expireret = nfscl_hasexpired(nmp->nm_clp, clidrev, p); } retrycnt++; } while (error == NFSERR_GRACE || error == NFSERR_DELAY || ((error == NFSERR_STALESTATEID || error == NFSERR_STALEDONTRECOVER) && called_from_strategy == 0) || (error == NFSERR_OLDSTATEID && retrycnt < 20) || ((error == NFSERR_EXPIRED || error == NFSERR_BADSTATEID) && expireret == 0 && clidrev != 0 && retrycnt < 4)); if (error != 0 && (retrycnt >= 4 || ((error == NFSERR_STALESTATEID || error == NFSERR_STALEDONTRECOVER) && called_from_strategy != 0))) error = EIO; if (NFSHASNFSV4(nmp) && p == NULL) NFSFREECRED(newcred); return (error); } /* * The actual write RPC. */ static int nfsrpc_writerpc(vnode_t vp, struct uio *uiop, int *iomode, int *must_commit, struct ucred *cred, nfsv4stateid_t *stateidp, NFSPROC_T *p, struct nfsvattr *nap, int *attrflagp, void *stuff) { u_int32_t *tl; struct nfsmount *nmp = VFSTONFS(vnode_mount(vp)); struct nfsnode *np = VTONFS(vp); int error = 0, len, tsiz, rlen, commit, committed = NFSWRITE_FILESYNC; int wccflag = 0, wsize; int32_t backup; struct nfsrv_descript nfsd; struct nfsrv_descript *nd = &nfsd; nfsattrbit_t attrbits; off_t tmp_off; KASSERT(uiop->uio_iovcnt == 1, ("nfs: writerpc iovcnt > 1")); *attrflagp = 0; tsiz = uio_uio_resid(uiop); tmp_off = uiop->uio_offset + tsiz; NFSLOCKMNT(nmp); if (tmp_off > nmp->nm_maxfilesize || tmp_off < uiop->uio_offset) { NFSUNLOCKMNT(nmp); return (EFBIG); } wsize = nmp->nm_wsize; NFSUNLOCKMNT(nmp); nd->nd_mrep = NULL; /* NFSv2 sometimes does a write with */ nd->nd_repstat = 0; /* uio_resid == 0, so the while is not done */ while (tsiz > 0) { *attrflagp = 0; len = (tsiz > wsize) ? wsize : tsiz; NFSCL_REQSTART(nd, NFSPROC_WRITE, vp); if (nd->nd_flag & ND_NFSV4) { nfsm_stateidtom(nd, stateidp, NFSSTATEID_PUTSTATEID); NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER+2*NFSX_UNSIGNED); txdr_hyper(uiop->uio_offset, tl); tl += 2; *tl++ = txdr_unsigned(*iomode); *tl = txdr_unsigned(len); } else if (nd->nd_flag & ND_NFSV3) { NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER+3*NFSX_UNSIGNED); txdr_hyper(uiop->uio_offset, tl); tl += 2; *tl++ = txdr_unsigned(len); *tl++ = txdr_unsigned(*iomode); *tl = txdr_unsigned(len); } else { u_int32_t x; NFSM_BUILD(tl, u_int32_t *, 4 * NFSX_UNSIGNED); /* * Not sure why someone changed this, since the * RFC clearly states that "beginoffset" and * "totalcount" are ignored, but it wouldn't * surprise me if there's a busted server out there. */ /* Set both "begin" and "current" to non-garbage. */ x = txdr_unsigned((u_int32_t)uiop->uio_offset); *tl++ = x; /* "begin offset" */ *tl++ = x; /* "current offset" */ x = txdr_unsigned(len); *tl++ = x; /* total to this offset */ *tl = x; /* size of this write */ } nfsm_uiombuf(nd, uiop, len); /* * Although it is tempting to do a normal Getattr Op in the * NFSv4 compound, the result can be a nearly hung client * system if the Getattr asks for Owner and/or OwnerGroup. * It occurs when the client can't map either the Owner or * Owner_group name in the Getattr reply to a uid/gid. When * there is a cache miss, the kernel does an upcall to the * nfsuserd. Then, it can try and read the local /etc/passwd * or /etc/group file. It can then block in getnewbuf(), * waiting for dirty writes to be pushed to the NFS server. * The only reason this doesn't result in a complete * deadlock, is that the upcall times out and allows * the write to complete. However, progress is so slow * that it might just as well be deadlocked. * So, we just get the attributes that change with each * write Op. * nb: nfscl_loadattrcache() needs to be told that these * partial attributes from a write rpc are being * passed in, via a argument flag. */ if (nd->nd_flag & ND_NFSV4) { NFSWRITEGETATTR_ATTRBIT(&attrbits); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_GETATTR); (void) nfsrv_putattrbit(nd, &attrbits); } error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (nd->nd_repstat) { /* * In case the rpc gets retried, roll * the uio fileds changed by nfsm_uiombuf() * back. */ uiop->uio_offset -= len; uio_uio_resid_add(uiop, len); uio_iov_base_add(uiop, -len); uio_iov_len_add(uiop, len); } if (nd->nd_flag & (ND_NFSV3 | ND_NFSV4)) { error = nfscl_wcc_data(nd, vp, nap, attrflagp, &wccflag, stuff); if (error) goto nfsmout; } if (!nd->nd_repstat) { if (nd->nd_flag & (ND_NFSV3 | ND_NFSV4)) { NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED + NFSX_VERF); rlen = fxdr_unsigned(int, *tl++); if (rlen == 0) { error = NFSERR_IO; goto nfsmout; } else if (rlen < len) { backup = len - rlen; uio_iov_base_add(uiop, -(backup)); uio_iov_len_add(uiop, backup); uiop->uio_offset -= backup; uio_uio_resid_add(uiop, backup); len = rlen; } commit = fxdr_unsigned(int, *tl++); /* * Return the lowest committment level * obtained by any of the RPCs. */ if (committed == NFSWRITE_FILESYNC) committed = commit; else if (committed == NFSWRITE_DATASYNC && commit == NFSWRITE_UNSTABLE) committed = commit; NFSLOCKMNT(nmp); if (!NFSHASWRITEVERF(nmp)) { NFSBCOPY((caddr_t)tl, (caddr_t)&nmp->nm_verf[0], NFSX_VERF); NFSSETWRITEVERF(nmp); } else if (NFSBCMP(tl, nmp->nm_verf, NFSX_VERF)) { *must_commit = 1; NFSBCOPY(tl, nmp->nm_verf, NFSX_VERF); } NFSUNLOCKMNT(nmp); } if (nd->nd_flag & ND_NFSV4) NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); if (nd->nd_flag & (ND_NFSV2 | ND_NFSV4)) { error = nfsm_loadattr(nd, nap); if (!error) *attrflagp = NFS_LATTR_NOSHRINK; } } else { error = nd->nd_repstat; } if (error) goto nfsmout; NFSWRITERPC_SETTIME(wccflag, np, (nd->nd_flag & ND_NFSV4)); mbuf_freem(nd->nd_mrep); nd->nd_mrep = NULL; tsiz -= len; } nfsmout: if (nd->nd_mrep != NULL) mbuf_freem(nd->nd_mrep); *iomode = committed; if (nd->nd_repstat && !error) error = nd->nd_repstat; return (error); } /* * nfs mknod rpc * For NFS v2 this is a kludge. Use a create rpc but with the IFMT bits of the * mode set to specify the file type and the size field for rdev. */ APPLESTATIC int nfsrpc_mknod(vnode_t dvp, char *name, int namelen, struct vattr *vap, u_int32_t rdev, enum vtype vtyp, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *dnap, struct nfsvattr *nnap, struct nfsfh **nfhpp, int *attrflagp, int *dattrflagp, void *dstuff) { u_int32_t *tl; int error = 0; struct nfsrv_descript nfsd, *nd = &nfsd; nfsattrbit_t attrbits; *nfhpp = NULL; *attrflagp = 0; *dattrflagp = 0; if (namelen > NFS_MAXNAMLEN) return (ENAMETOOLONG); NFSCL_REQSTART(nd, NFSPROC_MKNOD, dvp); if (nd->nd_flag & ND_NFSV4) { if (vtyp == VBLK || vtyp == VCHR) { NFSM_BUILD(tl, u_int32_t *, 3 * NFSX_UNSIGNED); *tl++ = vtonfsv34_type(vtyp); *tl++ = txdr_unsigned(NFSMAJOR(rdev)); *tl = txdr_unsigned(NFSMINOR(rdev)); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = vtonfsv34_type(vtyp); } } (void) nfsm_strtom(nd, name, namelen); if (nd->nd_flag & ND_NFSV3) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = vtonfsv34_type(vtyp); } if (nd->nd_flag & (ND_NFSV3 | ND_NFSV4)) nfscl_fillsattr(nd, vap, dvp, 0, 0); if ((nd->nd_flag & ND_NFSV3) && (vtyp == VCHR || vtyp == VBLK)) { NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSMAJOR(rdev)); *tl = txdr_unsigned(NFSMINOR(rdev)); } if (nd->nd_flag & ND_NFSV4) { NFSGETATTR_ATTRBIT(&attrbits); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV4OP_GETFH); *tl = txdr_unsigned(NFSV4OP_GETATTR); (void) nfsrv_putattrbit(nd, &attrbits); } if (nd->nd_flag & ND_NFSV2) nfscl_fillsattr(nd, vap, dvp, NFSSATTR_SIZERDEV, rdev); error = nfscl_request(nd, dvp, p, cred, dstuff); if (error) return (error); if (nd->nd_flag & ND_NFSV4) error = nfscl_wcc_data(nd, dvp, dnap, dattrflagp, NULL, dstuff); if (!nd->nd_repstat) { if (nd->nd_flag & ND_NFSV4) { NFSM_DISSECT(tl, u_int32_t *, 5 * NFSX_UNSIGNED); error = nfsrv_getattrbits(nd, &attrbits, NULL, NULL); if (error) goto nfsmout; } error = nfscl_mtofh(nd, nfhpp, nnap, attrflagp); if (error) goto nfsmout; } if (nd->nd_flag & ND_NFSV3) error = nfscl_wcc_data(nd, dvp, dnap, dattrflagp, NULL, dstuff); if (!error && nd->nd_repstat) error = nd->nd_repstat; nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * nfs file create call * Mostly just call the approriate routine. (I separated out v4, so that * error recovery wouldn't be as difficult.) */ APPLESTATIC int nfsrpc_create(vnode_t dvp, char *name, int namelen, struct vattr *vap, nfsquad_t cverf, int fmode, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *dnap, struct nfsvattr *nnap, struct nfsfh **nfhpp, int *attrflagp, int *dattrflagp, void *dstuff) { int error = 0, newone, expireret = 0, retrycnt, unlocked; struct nfsclowner *owp; struct nfscldeleg *dp; struct nfsmount *nmp = VFSTONFS(vnode_mount(dvp)); u_int32_t clidrev; if (NFSHASNFSV4(nmp)) { retrycnt = 0; do { dp = NULL; error = nfscl_open(dvp, NULL, 0, (NFSV4OPEN_ACCESSWRITE | NFSV4OPEN_ACCESSREAD), 0, cred, p, &owp, NULL, &newone, NULL, 1); if (error) return (error); if (nmp->nm_clp != NULL) clidrev = nmp->nm_clp->nfsc_clientidrev; else clidrev = 0; error = nfsrpc_createv4(dvp, name, namelen, vap, cverf, fmode, owp, &dp, cred, p, dnap, nnap, nfhpp, attrflagp, dattrflagp, dstuff, &unlocked); /* * There is no need to invalidate cached attributes here, * since new post-delegation issue attributes are always * returned by nfsrpc_createv4() and these will update the * attribute cache. */ if (dp != NULL) (void) nfscl_deleg(nmp->nm_mountp, owp->nfsow_clp, (*nfhpp)->nfh_fh, (*nfhpp)->nfh_len, cred, p, &dp); nfscl_ownerrelease(owp, error, newone, unlocked); if (error == NFSERR_GRACE || error == NFSERR_STALECLIENTID || error == NFSERR_STALEDONTRECOVER || error == NFSERR_DELAY) { (void) nfs_catnap(PZERO, error, "nfs_open"); } else if ((error == NFSERR_EXPIRED || error == NFSERR_BADSTATEID) && clidrev != 0) { expireret = nfscl_hasexpired(nmp->nm_clp, clidrev, p); retrycnt++; } } while (error == NFSERR_GRACE || error == NFSERR_STALECLIENTID || error == NFSERR_STALEDONTRECOVER || error == NFSERR_DELAY || ((error == NFSERR_EXPIRED || error == NFSERR_BADSTATEID) && expireret == 0 && clidrev != 0 && retrycnt < 4)); if (error && retrycnt >= 4) error = EIO; } else { error = nfsrpc_createv23(dvp, name, namelen, vap, cverf, fmode, cred, p, dnap, nnap, nfhpp, attrflagp, dattrflagp, dstuff); } return (error); } /* * The create rpc for v2 and 3. */ static int nfsrpc_createv23(vnode_t dvp, char *name, int namelen, struct vattr *vap, nfsquad_t cverf, int fmode, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *dnap, struct nfsvattr *nnap, struct nfsfh **nfhpp, int *attrflagp, int *dattrflagp, void *dstuff) { u_int32_t *tl; int error = 0; struct nfsrv_descript nfsd, *nd = &nfsd; *nfhpp = NULL; *attrflagp = 0; *dattrflagp = 0; if (namelen > NFS_MAXNAMLEN) return (ENAMETOOLONG); NFSCL_REQSTART(nd, NFSPROC_CREATE, dvp); (void) nfsm_strtom(nd, name, namelen); if (nd->nd_flag & ND_NFSV3) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); if (fmode & O_EXCL) { *tl = txdr_unsigned(NFSCREATE_EXCLUSIVE); NFSM_BUILD(tl, u_int32_t *, NFSX_VERF); *tl++ = cverf.lval[0]; *tl = cverf.lval[1]; } else { *tl = txdr_unsigned(NFSCREATE_UNCHECKED); nfscl_fillsattr(nd, vap, dvp, 0, 0); } } else { nfscl_fillsattr(nd, vap, dvp, NFSSATTR_SIZE0, 0); } error = nfscl_request(nd, dvp, p, cred, dstuff); if (error) return (error); if (nd->nd_repstat == 0) { error = nfscl_mtofh(nd, nfhpp, nnap, attrflagp); if (error) goto nfsmout; } if (nd->nd_flag & ND_NFSV3) error = nfscl_wcc_data(nd, dvp, dnap, dattrflagp, NULL, dstuff); if (nd->nd_repstat != 0 && error == 0) error = nd->nd_repstat; nfsmout: mbuf_freem(nd->nd_mrep); return (error); } static int nfsrpc_createv4(vnode_t dvp, char *name, int namelen, struct vattr *vap, nfsquad_t cverf, int fmode, struct nfsclowner *owp, struct nfscldeleg **dpp, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *dnap, struct nfsvattr *nnap, struct nfsfh **nfhpp, int *attrflagp, int *dattrflagp, void *dstuff, int *unlockedp) { u_int32_t *tl; int error = 0, deleg, newone, ret, acesize, limitby; struct nfsrv_descript nfsd, *nd = &nfsd; struct nfsclopen *op; struct nfscldeleg *dp = NULL; struct nfsnode *np; struct nfsfh *nfhp; nfsattrbit_t attrbits; nfsv4stateid_t stateid; u_int32_t rflags; *unlockedp = 0; *nfhpp = NULL; *dpp = NULL; *attrflagp = 0; *dattrflagp = 0; if (namelen > NFS_MAXNAMLEN) return (ENAMETOOLONG); NFSCL_REQSTART(nd, NFSPROC_CREATE, dvp); /* * For V4, this is actually an Open op. */ NFSM_BUILD(tl, u_int32_t *, 5 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(owp->nfsow_seqid); *tl++ = txdr_unsigned(NFSV4OPEN_ACCESSWRITE | NFSV4OPEN_ACCESSREAD); *tl++ = txdr_unsigned(NFSV4OPEN_DENYNONE); *tl++ = owp->nfsow_clp->nfsc_clientid.lval[0]; *tl = owp->nfsow_clp->nfsc_clientid.lval[1]; (void) nfsm_strtom(nd, owp->nfsow_owner, NFSV4CL_LOCKNAMELEN); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV4OPEN_CREATE); if (fmode & O_EXCL) { *tl = txdr_unsigned(NFSCREATE_EXCLUSIVE); NFSM_BUILD(tl, u_int32_t *, NFSX_VERF); *tl++ = cverf.lval[0]; *tl = cverf.lval[1]; } else { *tl = txdr_unsigned(NFSCREATE_UNCHECKED); nfscl_fillsattr(nd, vap, dvp, 0, 0); } NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OPEN_CLAIMNULL); (void) nfsm_strtom(nd, name, namelen); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV4OP_GETFH); *tl = txdr_unsigned(NFSV4OP_GETATTR); NFSGETATTR_ATTRBIT(&attrbits); (void) nfsrv_putattrbit(nd, &attrbits); error = nfscl_request(nd, dvp, p, cred, dstuff); if (error) return (error); error = nfscl_wcc_data(nd, dvp, dnap, dattrflagp, NULL, dstuff); if (error) goto nfsmout; NFSCL_INCRSEQID(owp->nfsow_seqid, nd); if (nd->nd_repstat == 0) { NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID + 6 * NFSX_UNSIGNED); stateid.seqid = *tl++; stateid.other[0] = *tl++; stateid.other[1] = *tl++; stateid.other[2] = *tl; rflags = fxdr_unsigned(u_int32_t, *(tl + 6)); (void) nfsrv_getattrbits(nd, &attrbits, NULL, NULL); NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); deleg = fxdr_unsigned(int, *tl); if (deleg == NFSV4OPEN_DELEGATEREAD || deleg == NFSV4OPEN_DELEGATEWRITE) { if (!(owp->nfsow_clp->nfsc_flags & NFSCLFLAGS_FIRSTDELEG)) owp->nfsow_clp->nfsc_flags |= (NFSCLFLAGS_FIRSTDELEG | NFSCLFLAGS_GOTDELEG); MALLOC(dp, struct nfscldeleg *, sizeof (struct nfscldeleg) + NFSX_V4FHMAX, M_NFSCLDELEG, M_WAITOK); LIST_INIT(&dp->nfsdl_owner); LIST_INIT(&dp->nfsdl_lock); dp->nfsdl_clp = owp->nfsow_clp; newnfs_copyincred(cred, &dp->nfsdl_cred); nfscl_lockinit(&dp->nfsdl_rwlock); NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID + NFSX_UNSIGNED); dp->nfsdl_stateid.seqid = *tl++; dp->nfsdl_stateid.other[0] = *tl++; dp->nfsdl_stateid.other[1] = *tl++; dp->nfsdl_stateid.other[2] = *tl++; ret = fxdr_unsigned(int, *tl); if (deleg == NFSV4OPEN_DELEGATEWRITE) { dp->nfsdl_flags = NFSCLDL_WRITE; /* * Indicates how much the file can grow. */ NFSM_DISSECT(tl, u_int32_t *, 3 * NFSX_UNSIGNED); limitby = fxdr_unsigned(int, *tl++); switch (limitby) { case NFSV4OPEN_LIMITSIZE: dp->nfsdl_sizelimit = fxdr_hyper(tl); break; case NFSV4OPEN_LIMITBLOCKS: dp->nfsdl_sizelimit = fxdr_unsigned(u_int64_t, *tl++); dp->nfsdl_sizelimit *= fxdr_unsigned(u_int64_t, *tl); break; default: error = NFSERR_BADXDR; goto nfsmout; }; } else { dp->nfsdl_flags = NFSCLDL_READ; } if (ret) dp->nfsdl_flags |= NFSCLDL_RECALL; error = nfsrv_dissectace(nd, &dp->nfsdl_ace, &ret, &acesize, p); if (error) goto nfsmout; } else if (deleg != NFSV4OPEN_DELEGATENONE) { error = NFSERR_BADXDR; goto nfsmout; } error = nfscl_mtofh(nd, nfhpp, nnap, attrflagp); if (error) goto nfsmout; if (dp != NULL && *attrflagp) { dp->nfsdl_change = nnap->na_filerev; dp->nfsdl_modtime = nnap->na_mtime; dp->nfsdl_flags |= NFSCLDL_MODTIMESET; } /* * We can now complete the Open state. */ nfhp = *nfhpp; if (dp != NULL) { dp->nfsdl_fhlen = nfhp->nfh_len; NFSBCOPY(nfhp->nfh_fh, dp->nfsdl_fh, nfhp->nfh_len); } /* * Get an Open structure that will be * attached to the OpenOwner, acquired already. */ error = nfscl_open(dvp, nfhp->nfh_fh, nfhp->nfh_len, (NFSV4OPEN_ACCESSWRITE | NFSV4OPEN_ACCESSREAD), 0, cred, p, NULL, &op, &newone, NULL, 0); if (error) goto nfsmout; op->nfso_stateid = stateid; newnfs_copyincred(cred, &op->nfso_cred); if ((rflags & NFSV4OPEN_RESULTCONFIRM)) { do { ret = nfsrpc_openconfirm(dvp, nfhp->nfh_fh, nfhp->nfh_len, op, cred, p); if (ret == NFSERR_DELAY) (void) nfs_catnap(PZERO, ret, "nfs_create"); } while (ret == NFSERR_DELAY); error = ret; } /* * If the server is handing out delegations, but we didn't * get one because an OpenConfirm was required, try the * Open again, to get a delegation. This is a harmless no-op, * from a server's point of view. */ if ((rflags & NFSV4OPEN_RESULTCONFIRM) && (owp->nfsow_clp->nfsc_flags & NFSCLFLAGS_GOTDELEG) && !error && dp == NULL) { np = VTONFS(dvp); do { ret = nfsrpc_openrpc(VFSTONFS(vnode_mount(dvp)), dvp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len, nfhp->nfh_fh, nfhp->nfh_len, (NFSV4OPEN_ACCESSWRITE | NFSV4OPEN_ACCESSREAD), op, name, namelen, &dp, 0, 0x0, cred, p, 0, 1); if (ret == NFSERR_DELAY) (void) nfs_catnap(PZERO, ret, "nfs_crt2"); } while (ret == NFSERR_DELAY); if (ret) { if (dp != NULL) FREE((caddr_t)dp, M_NFSCLDELEG); if (ret == NFSERR_STALECLIENTID || ret == NFSERR_STALEDONTRECOVER) error = ret; } } nfscl_openrelease(op, error, newone); *unlockedp = 1; } if (nd->nd_repstat != 0 && error == 0) error = nd->nd_repstat; if (error == NFSERR_STALECLIENTID) nfscl_initiate_recovery(owp->nfsow_clp); nfsmout: if (!error) *dpp = dp; else if (dp != NULL) FREE((caddr_t)dp, M_NFSCLDELEG); mbuf_freem(nd->nd_mrep); return (error); } /* * Nfs remove rpc */ APPLESTATIC int nfsrpc_remove(vnode_t dvp, char *name, int namelen, vnode_t vp, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *dnap, int *dattrflagp, void *dstuff) { u_int32_t *tl; struct nfsrv_descript nfsd, *nd = &nfsd; struct nfsnode *np; struct nfsmount *nmp; nfsv4stateid_t dstateid; int error, ret = 0, i; *dattrflagp = 0; if (namelen > NFS_MAXNAMLEN) return (ENAMETOOLONG); nmp = VFSTONFS(vnode_mount(dvp)); tryagain: if (NFSHASNFSV4(nmp) && ret == 0) { ret = nfscl_removedeleg(vp, p, &dstateid); if (ret == 1) { NFSCL_REQSTART(nd, NFSPROC_RETDELEGREMOVE, vp); NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID + NFSX_UNSIGNED); *tl++ = dstateid.seqid; *tl++ = dstateid.other[0]; *tl++ = dstateid.other[1]; *tl++ = dstateid.other[2]; *tl = txdr_unsigned(NFSV4OP_PUTFH); np = VTONFS(dvp); (void) nfsm_fhtom(nd, np->n_fhp->nfh_fh, np->n_fhp->nfh_len, 0); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_REMOVE); } } else { ret = 0; } if (ret == 0) NFSCL_REQSTART(nd, NFSPROC_REMOVE, dvp); (void) nfsm_strtom(nd, name, namelen); error = nfscl_request(nd, dvp, p, cred, dstuff); if (error) return (error); if (nd->nd_flag & (ND_NFSV3 | ND_NFSV4)) { /* For NFSv4, parse out any Delereturn replies. */ if (ret > 0 && nd->nd_repstat != 0 && (nd->nd_flag & ND_NOMOREDATA)) { /* * If the Delegreturn failed, try again without * it. The server will Recall, as required. */ mbuf_freem(nd->nd_mrep); goto tryagain; } for (i = 0; i < (ret * 2); i++) { if ((nd->nd_flag & (ND_NFSV4 | ND_NOMOREDATA)) == ND_NFSV4) { NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); if (*(tl + 1)) nd->nd_flag |= ND_NOMOREDATA; } } error = nfscl_wcc_data(nd, dvp, dnap, dattrflagp, NULL, dstuff); } if (nd->nd_repstat && !error) error = nd->nd_repstat; nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * Do an nfs rename rpc. */ APPLESTATIC int nfsrpc_rename(vnode_t fdvp, vnode_t fvp, char *fnameptr, int fnamelen, vnode_t tdvp, vnode_t tvp, char *tnameptr, int tnamelen, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *fnap, struct nfsvattr *tnap, int *fattrflagp, int *tattrflagp, void *fstuff, void *tstuff) { u_int32_t *tl; struct nfsrv_descript nfsd, *nd = &nfsd; struct nfsmount *nmp; struct nfsnode *np; nfsattrbit_t attrbits; nfsv4stateid_t fdstateid, tdstateid; int error = 0, ret = 0, gottd = 0, gotfd = 0, i; *fattrflagp = 0; *tattrflagp = 0; nmp = VFSTONFS(vnode_mount(fdvp)); if (fnamelen > NFS_MAXNAMLEN || tnamelen > NFS_MAXNAMLEN) return (ENAMETOOLONG); tryagain: if (NFSHASNFSV4(nmp) && ret == 0) { ret = nfscl_renamedeleg(fvp, &fdstateid, &gotfd, tvp, &tdstateid, &gottd, p); if (gotfd && gottd) { NFSCL_REQSTART(nd, NFSPROC_RETDELEGRENAME2, fvp); } else if (gotfd) { NFSCL_REQSTART(nd, NFSPROC_RETDELEGRENAME1, fvp); } else if (gottd) { NFSCL_REQSTART(nd, NFSPROC_RETDELEGRENAME1, tvp); } if (gotfd) { NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID); *tl++ = fdstateid.seqid; *tl++ = fdstateid.other[0]; *tl++ = fdstateid.other[1]; *tl = fdstateid.other[2]; if (gottd) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_PUTFH); np = VTONFS(tvp); (void) nfsm_fhtom(nd, np->n_fhp->nfh_fh, np->n_fhp->nfh_len, 0); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_DELEGRETURN); } } if (gottd) { NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID); *tl++ = tdstateid.seqid; *tl++ = tdstateid.other[0]; *tl++ = tdstateid.other[1]; *tl = tdstateid.other[2]; } if (ret > 0) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_PUTFH); np = VTONFS(fdvp); (void) nfsm_fhtom(nd, np->n_fhp->nfh_fh, np->n_fhp->nfh_len, 0); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_SAVEFH); } } else { ret = 0; } if (ret == 0) NFSCL_REQSTART(nd, NFSPROC_RENAME, fdvp); if (nd->nd_flag & ND_NFSV4) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_GETATTR); NFSWCCATTR_ATTRBIT(&attrbits); (void) nfsrv_putattrbit(nd, &attrbits); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_PUTFH); (void) nfsm_fhtom(nd, VTONFS(tdvp)->n_fhp->nfh_fh, VTONFS(tdvp)->n_fhp->nfh_len, 0); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_GETATTR); (void) nfsrv_putattrbit(nd, &attrbits); nd->nd_flag |= ND_V4WCCATTR; NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_RENAME); } (void) nfsm_strtom(nd, fnameptr, fnamelen); if (!(nd->nd_flag & ND_NFSV4)) (void) nfsm_fhtom(nd, VTONFS(tdvp)->n_fhp->nfh_fh, VTONFS(tdvp)->n_fhp->nfh_len, 0); (void) nfsm_strtom(nd, tnameptr, tnamelen); error = nfscl_request(nd, fdvp, p, cred, fstuff); if (error) return (error); if (nd->nd_flag & (ND_NFSV3 | ND_NFSV4)) { /* For NFSv4, parse out any Delereturn replies. */ if (ret > 0 && nd->nd_repstat != 0 && (nd->nd_flag & ND_NOMOREDATA)) { /* * If the Delegreturn failed, try again without * it. The server will Recall, as required. */ mbuf_freem(nd->nd_mrep); goto tryagain; } for (i = 0; i < (ret * 2); i++) { if ((nd->nd_flag & (ND_NFSV4 | ND_NOMOREDATA)) == ND_NFSV4) { NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); if (*(tl + 1)) { if (i == 0 && ret > 1) { /* * If the Delegreturn failed, try again * without it. The server will Recall, as * required. * If ret > 1, the first iteration of this * loop is the second DelegReturn result. */ mbuf_freem(nd->nd_mrep); goto tryagain; } else { nd->nd_flag |= ND_NOMOREDATA; } } } } /* Now, the first wcc attribute reply. */ if ((nd->nd_flag & (ND_NFSV4 | ND_NOMOREDATA)) == ND_NFSV4) { NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); if (*(tl + 1)) nd->nd_flag |= ND_NOMOREDATA; } error = nfscl_wcc_data(nd, fdvp, fnap, fattrflagp, NULL, fstuff); /* and the second wcc attribute reply. */ if ((nd->nd_flag & (ND_NFSV4 | ND_NOMOREDATA)) == ND_NFSV4 && !error) { NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); if (*(tl + 1)) nd->nd_flag |= ND_NOMOREDATA; } if (!error) error = nfscl_wcc_data(nd, tdvp, tnap, tattrflagp, NULL, tstuff); } if (nd->nd_repstat && !error) error = nd->nd_repstat; nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * nfs hard link create rpc */ APPLESTATIC int nfsrpc_link(vnode_t dvp, vnode_t vp, char *name, int namelen, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *dnap, struct nfsvattr *nap, int *attrflagp, int *dattrflagp, void *dstuff) { u_int32_t *tl; struct nfsrv_descript nfsd, *nd = &nfsd; nfsattrbit_t attrbits; int error = 0; *attrflagp = 0; *dattrflagp = 0; if (namelen > NFS_MAXNAMLEN) return (ENAMETOOLONG); NFSCL_REQSTART(nd, NFSPROC_LINK, vp); if (nd->nd_flag & ND_NFSV4) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_PUTFH); } (void) nfsm_fhtom(nd, VTONFS(dvp)->n_fhp->nfh_fh, VTONFS(dvp)->n_fhp->nfh_len, 0); if (nd->nd_flag & ND_NFSV4) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_GETATTR); NFSWCCATTR_ATTRBIT(&attrbits); (void) nfsrv_putattrbit(nd, &attrbits); nd->nd_flag |= ND_V4WCCATTR; NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_LINK); } (void) nfsm_strtom(nd, name, namelen); error = nfscl_request(nd, vp, p, cred, dstuff); if (error) return (error); if (nd->nd_flag & ND_NFSV3) { error = nfscl_postop_attr(nd, nap, attrflagp, dstuff); if (!error) error = nfscl_wcc_data(nd, dvp, dnap, dattrflagp, NULL, dstuff); } else if ((nd->nd_flag & (ND_NFSV4 | ND_NOMOREDATA)) == ND_NFSV4) { /* * First, parse out the PutFH and Getattr result. */ NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); if (!(*(tl + 1))) NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); if (*(tl + 1)) nd->nd_flag |= ND_NOMOREDATA; /* * Get the pre-op attributes. */ error = nfscl_wcc_data(nd, dvp, dnap, dattrflagp, NULL, dstuff); } if (nd->nd_repstat && !error) error = nd->nd_repstat; nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * nfs symbolic link create rpc */ APPLESTATIC int nfsrpc_symlink(vnode_t dvp, char *name, int namelen, char *target, struct vattr *vap, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *dnap, struct nfsvattr *nnap, struct nfsfh **nfhpp, int *attrflagp, int *dattrflagp, void *dstuff) { u_int32_t *tl; struct nfsrv_descript nfsd, *nd = &nfsd; struct nfsmount *nmp; int slen, error = 0; *nfhpp = NULL; *attrflagp = 0; *dattrflagp = 0; nmp = VFSTONFS(vnode_mount(dvp)); slen = strlen(target); if (slen > NFS_MAXPATHLEN || namelen > NFS_MAXNAMLEN) return (ENAMETOOLONG); NFSCL_REQSTART(nd, NFSPROC_SYMLINK, dvp); if (nd->nd_flag & ND_NFSV4) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFLNK); (void) nfsm_strtom(nd, target, slen); } (void) nfsm_strtom(nd, name, namelen); if (nd->nd_flag & (ND_NFSV3 | ND_NFSV4)) nfscl_fillsattr(nd, vap, dvp, 0, 0); if (!(nd->nd_flag & ND_NFSV4)) (void) nfsm_strtom(nd, target, slen); if (nd->nd_flag & ND_NFSV2) nfscl_fillsattr(nd, vap, dvp, NFSSATTR_SIZENEG1, 0); error = nfscl_request(nd, dvp, p, cred, dstuff); if (error) return (error); if (nd->nd_flag & ND_NFSV4) error = nfscl_wcc_data(nd, dvp, dnap, dattrflagp, NULL, dstuff); if ((nd->nd_flag & ND_NFSV3) && !error) { if (!nd->nd_repstat) error = nfscl_mtofh(nd, nfhpp, nnap, attrflagp); if (!error) error = nfscl_wcc_data(nd, dvp, dnap, dattrflagp, NULL, dstuff); } if (nd->nd_repstat && !error) error = nd->nd_repstat; mbuf_freem(nd->nd_mrep); /* * Kludge: Map EEXIST => 0 assuming that it is a reply to a retry. */ if (error == EEXIST) error = 0; return (error); } /* * nfs make dir rpc */ APPLESTATIC int nfsrpc_mkdir(vnode_t dvp, char *name, int namelen, struct vattr *vap, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *dnap, struct nfsvattr *nnap, struct nfsfh **nfhpp, int *attrflagp, int *dattrflagp, void *dstuff) { u_int32_t *tl; struct nfsrv_descript nfsd, *nd = &nfsd; nfsattrbit_t attrbits; int error = 0; *nfhpp = NULL; *attrflagp = 0; *dattrflagp = 0; if (namelen > NFS_MAXNAMLEN) return (ENAMETOOLONG); NFSCL_REQSTART(nd, NFSPROC_MKDIR, dvp); if (nd->nd_flag & ND_NFSV4) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFDIR); } (void) nfsm_strtom(nd, name, namelen); nfscl_fillsattr(nd, vap, dvp, NFSSATTR_SIZENEG1, 0); if (nd->nd_flag & ND_NFSV4) { NFSGETATTR_ATTRBIT(&attrbits); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV4OP_GETFH); *tl = txdr_unsigned(NFSV4OP_GETATTR); (void) nfsrv_putattrbit(nd, &attrbits); } error = nfscl_request(nd, dvp, p, cred, dstuff); if (error) return (error); if (nd->nd_flag & ND_NFSV4) error = nfscl_wcc_data(nd, dvp, dnap, dattrflagp, NULL, dstuff); if (!nd->nd_repstat && !error) { if (nd->nd_flag & ND_NFSV4) { NFSM_DISSECT(tl, u_int32_t *, 5 * NFSX_UNSIGNED); error = nfsrv_getattrbits(nd, &attrbits, NULL, NULL); } if (!error) error = nfscl_mtofh(nd, nfhpp, nnap, attrflagp); } if ((nd->nd_flag & ND_NFSV3) && !error) error = nfscl_wcc_data(nd, dvp, dnap, dattrflagp, NULL, dstuff); if (nd->nd_repstat && !error) error = nd->nd_repstat; nfsmout: mbuf_freem(nd->nd_mrep); /* * Kludge: Map EEXIST => 0 assuming that you have a reply to a retry. */ if (error == EEXIST) error = 0; return (error); } /* * nfs remove directory call */ APPLESTATIC int nfsrpc_rmdir(vnode_t dvp, char *name, int namelen, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *dnap, int *dattrflagp, void *dstuff) { struct nfsrv_descript nfsd, *nd = &nfsd; int error = 0; *dattrflagp = 0; if (namelen > NFS_MAXNAMLEN) return (ENAMETOOLONG); NFSCL_REQSTART(nd, NFSPROC_RMDIR, dvp); (void) nfsm_strtom(nd, name, namelen); error = nfscl_request(nd, dvp, p, cred, dstuff); if (error) return (error); if (nd->nd_flag & (ND_NFSV3 | ND_NFSV4)) error = nfscl_wcc_data(nd, dvp, dnap, dattrflagp, NULL, dstuff); if (nd->nd_repstat && !error) error = nd->nd_repstat; mbuf_freem(nd->nd_mrep); /* * Kludge: Map ENOENT => 0 assuming that you have a reply to a retry. */ if (error == ENOENT) error = 0; return (error); } /* * Readdir rpc. * Always returns with either uio_resid unchanged, if you are at the * end of the directory, or uio_resid == 0, with all DIRBLKSIZ chunks * filled in. * I felt this would allow caching of directory blocks more easily * than returning a pertially filled block. * Directory offset cookies: * Oh my, what to do with them... * I can think of three ways to deal with them: * 1 - have the layer above these RPCs maintain a map between logical * directory byte offsets and the NFS directory offset cookies * 2 - pass the opaque directory offset cookies up into userland * and let the libc functions deal with them, via the system call * 3 - return them to userland in the "struct dirent", so future versions * of libc can use them and do whatever is necessary to amke things work * above these rpc calls, in the meantime * For now, I do #3 by "hiding" the directory offset cookies after the * d_name field in struct dirent. This is space inside d_reclen that * will be ignored by anything that doesn't know about them. * The directory offset cookies are filled in as the last 8 bytes of * each directory entry, after d_name. Someday, the userland libc * functions may be able to use these. In the meantime, it satisfies * OpenBSD's requirements for cookies being returned. * If expects the directory offset cookie for the read to be in uio_offset * and returns the one for the next entry after this directory block in * there, as well. */ APPLESTATIC int nfsrpc_readdir(vnode_t vp, struct uio *uiop, nfsuint64 *cookiep, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, int *attrflagp, int *eofp, void *stuff) { int len, left; struct dirent *dp = NULL; u_int32_t *tl; nfsquad_t cookie, ncookie; struct nfsmount *nmp = VFSTONFS(vnode_mount(vp)); struct nfsnode *dnp = VTONFS(vp); struct nfsvattr nfsva; struct nfsrv_descript nfsd, *nd = &nfsd; int error = 0, tlen, more_dirs = 1, blksiz = 0, bigenough = 1; int reqsize, tryformoredirs = 1, readsize, eof = 0, gotmnton = 0; long dotfileid, dotdotfileid = 0; u_int32_t fakefileno = 0xffffffff, rderr; char *cp; nfsattrbit_t attrbits, dattrbits; u_int32_t *tl2 = NULL; size_t tresid; KASSERT(uiop->uio_iovcnt == 1 && (uio_uio_resid(uiop) & (DIRBLKSIZ - 1)) == 0, ("nfs readdirrpc bad uio")); /* * There is no point in reading a lot more than uio_resid, however * adding one additional DIRBLKSIZ makes sense. Since uio_resid * and nm_readdirsize are both exact multiples of DIRBLKSIZ, this * will never make readsize > nm_readdirsize. */ readsize = nmp->nm_readdirsize; if (readsize > uio_uio_resid(uiop)) readsize = uio_uio_resid(uiop) + DIRBLKSIZ; *attrflagp = 0; if (eofp) *eofp = 0; tresid = uio_uio_resid(uiop); cookie.lval[0] = cookiep->nfsuquad[0]; cookie.lval[1] = cookiep->nfsuquad[1]; nd->nd_mrep = NULL; /* * For NFSv4, first create the "." and ".." entries. */ if (NFSHASNFSV4(nmp)) { reqsize = 6 * NFSX_UNSIGNED; NFSGETATTR_ATTRBIT(&dattrbits); NFSZERO_ATTRBIT(&attrbits); NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_FILEID); NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_TYPE); if (NFSISSET_ATTRBIT(&dnp->n_vattr.na_suppattr, NFSATTRBIT_MOUNTEDONFILEID)) { NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_MOUNTEDONFILEID); gotmnton = 1; } else { /* * Must fake it. Use the fileno, except when the * fsid is != to that of the directory. For that * case, generate a fake fileno that is not the same. */ NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_FSID); gotmnton = 0; } /* * Joy, oh joy. For V4 we get to hand craft '.' and '..'. */ if (uiop->uio_offset == 0) { #if defined(__FreeBSD_version) && __FreeBSD_version >= 800000 error = VOP_GETATTR(vp, &nfsva.na_vattr, cred); #else error = VOP_GETATTR(vp, &nfsva.na_vattr, cred, p); #endif if (error) return (error); dotfileid = nfsva.na_fileid; NFSCL_REQSTART(nd, NFSPROC_LOOKUPP, vp); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV4OP_GETFH); *tl = txdr_unsigned(NFSV4OP_GETATTR); (void) nfsrv_putattrbit(nd, &attrbits); error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (nd->nd_repstat == 0) { NFSM_DISSECT(tl, u_int32_t *, 3*NFSX_UNSIGNED); len = fxdr_unsigned(int, *(tl + 2)); if (len > 0 && len <= NFSX_V4FHMAX) error = nfsm_advance(nd, NFSM_RNDUP(len), -1); else error = EPERM; if (!error) { NFSM_DISSECT(tl, u_int32_t *, 2*NFSX_UNSIGNED); nfsva.na_mntonfileno = 0xffffffff; error = nfsv4_loadattr(nd, NULL, &nfsva, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, p, cred); if (error) { dotdotfileid = dotfileid; } else if (gotmnton) { if (nfsva.na_mntonfileno != 0xffffffff) dotdotfileid = nfsva.na_mntonfileno; else dotdotfileid = nfsva.na_fileid; } else if (nfsva.na_filesid[0] == dnp->n_vattr.na_filesid[0] && nfsva.na_filesid[1] == dnp->n_vattr.na_filesid[1]) { dotdotfileid = nfsva.na_fileid; } else { do { fakefileno--; } while (fakefileno == nfsva.na_fileid); dotdotfileid = fakefileno; } } } else if (nd->nd_repstat == NFSERR_NOENT) { /* * Lookupp returns NFSERR_NOENT when we are * at the root, so just use the current dir. */ nd->nd_repstat = 0; dotdotfileid = dotfileid; } else { error = nd->nd_repstat; } mbuf_freem(nd->nd_mrep); if (error) return (error); nd->nd_mrep = NULL; dp = (struct dirent *) CAST_DOWN(caddr_t, uio_iov_base(uiop)); dp->d_type = DT_DIR; dp->d_fileno = dotfileid; dp->d_namlen = 1; dp->d_name[0] = '.'; dp->d_name[1] = '\0'; dp->d_reclen = DIRENT_SIZE(dp) + NFSX_HYPER; /* * Just make these offset cookie 0. */ tl = (u_int32_t *)&dp->d_name[4]; *tl++ = 0; *tl = 0; blksiz += dp->d_reclen; uio_uio_resid_add(uiop, -(dp->d_reclen)); uiop->uio_offset += dp->d_reclen; uio_iov_base_add(uiop, dp->d_reclen); uio_iov_len_add(uiop, -(dp->d_reclen)); dp = (struct dirent *) CAST_DOWN(caddr_t, uio_iov_base(uiop)); dp->d_type = DT_DIR; dp->d_fileno = dotdotfileid; dp->d_namlen = 2; dp->d_name[0] = '.'; dp->d_name[1] = '.'; dp->d_name[2] = '\0'; dp->d_reclen = DIRENT_SIZE(dp) + NFSX_HYPER; /* * Just make these offset cookie 0. */ tl = (u_int32_t *)&dp->d_name[4]; *tl++ = 0; *tl = 0; blksiz += dp->d_reclen; uio_uio_resid_add(uiop, -(dp->d_reclen)); uiop->uio_offset += dp->d_reclen; uio_iov_base_add(uiop, dp->d_reclen); uio_iov_len_add(uiop, -(dp->d_reclen)); } NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_RDATTRERROR); } else { reqsize = 5 * NFSX_UNSIGNED; } /* * Loop around doing readdir rpc's of size readsize. * The stopping criteria is EOF or buffer full. */ while (more_dirs && bigenough) { *attrflagp = 0; NFSCL_REQSTART(nd, NFSPROC_READDIR, vp); if (nd->nd_flag & ND_NFSV2) { NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = cookie.lval[1]; *tl = txdr_unsigned(readsize); } else { NFSM_BUILD(tl, u_int32_t *, reqsize); *tl++ = cookie.lval[0]; *tl++ = cookie.lval[1]; if (cookie.qval == 0) { *tl++ = 0; *tl++ = 0; } else { NFSLOCKNODE(dnp); *tl++ = dnp->n_cookieverf.nfsuquad[0]; *tl++ = dnp->n_cookieverf.nfsuquad[1]; NFSUNLOCKNODE(dnp); } if (nd->nd_flag & ND_NFSV4) { *tl++ = txdr_unsigned(readsize); *tl = txdr_unsigned(readsize); (void) nfsrv_putattrbit(nd, &attrbits); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_GETATTR); (void) nfsrv_putattrbit(nd, &dattrbits); } else { *tl = txdr_unsigned(readsize); } } error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (!(nd->nd_flag & ND_NFSV2)) { if (nd->nd_flag & ND_NFSV3) error = nfscl_postop_attr(nd, nap, attrflagp, stuff); if (!nd->nd_repstat && !error) { NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); NFSLOCKNODE(dnp); dnp->n_cookieverf.nfsuquad[0] = *tl++; dnp->n_cookieverf.nfsuquad[1] = *tl; NFSUNLOCKNODE(dnp); } } if (nd->nd_repstat || error) { if (!error) error = nd->nd_repstat; goto nfsmout; } NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); more_dirs = fxdr_unsigned(int, *tl); if (!more_dirs) tryformoredirs = 0; /* loop thru the dir entries, doctoring them to 4bsd form */ while (more_dirs && bigenough) { if (nd->nd_flag & ND_NFSV4) { NFSM_DISSECT(tl, u_int32_t *, 3*NFSX_UNSIGNED); ncookie.lval[0] = *tl++; ncookie.lval[1] = *tl++; len = fxdr_unsigned(int, *tl); } else if (nd->nd_flag & ND_NFSV3) { NFSM_DISSECT(tl, u_int32_t *, 3*NFSX_UNSIGNED); nfsva.na_fileid = fxdr_hyper(tl); tl += 2; len = fxdr_unsigned(int, *tl); } else { NFSM_DISSECT(tl, u_int32_t *, 2*NFSX_UNSIGNED); nfsva.na_fileid = fxdr_unsigned(long, *tl++); len = fxdr_unsigned(int, *tl); } if (len <= 0 || len > NFS_MAXNAMLEN) { error = EBADRPC; goto nfsmout; } tlen = NFSM_RNDUP(len); if (tlen == len) tlen += 4; /* To ensure null termination */ left = DIRBLKSIZ - blksiz; if ((int)(tlen + DIRHDSIZ + NFSX_HYPER) > left) { dp->d_reclen += left; uio_iov_base_add(uiop, left); uio_iov_len_add(uiop, -(left)); uio_uio_resid_add(uiop, -(left)); uiop->uio_offset += left; blksiz = 0; } if ((int)(tlen + DIRHDSIZ + NFSX_HYPER) > uio_uio_resid(uiop)) bigenough = 0; if (bigenough) { dp = (struct dirent *) CAST_DOWN(caddr_t, uio_iov_base(uiop)); dp->d_namlen = len; dp->d_reclen = tlen + DIRHDSIZ + NFSX_HYPER; dp->d_type = DT_UNKNOWN; blksiz += dp->d_reclen; if (blksiz == DIRBLKSIZ) blksiz = 0; uio_uio_resid_add(uiop, -(DIRHDSIZ)); uiop->uio_offset += DIRHDSIZ; uio_iov_base_add(uiop, DIRHDSIZ); uio_iov_len_add(uiop, -(DIRHDSIZ)); error = nfsm_mbufuio(nd, uiop, len); if (error) goto nfsmout; cp = CAST_DOWN(caddr_t, uio_iov_base(uiop)); tlen -= len; *cp = '\0'; /* null terminate */ cp += tlen; /* points to cookie storage */ tl2 = (u_int32_t *)cp; uio_iov_base_add(uiop, (tlen + NFSX_HYPER)); uio_iov_len_add(uiop, -(tlen + NFSX_HYPER)); uio_uio_resid_add(uiop, -(tlen + NFSX_HYPER)); uiop->uio_offset += (tlen + NFSX_HYPER); } else { error = nfsm_advance(nd, NFSM_RNDUP(len), -1); if (error) goto nfsmout; } if (nd->nd_flag & ND_NFSV4) { rderr = 0; nfsva.na_mntonfileno = 0xffffffff; error = nfsv4_loadattr(nd, NULL, &nfsva, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, NULL, NULL, &rderr, p, cred); if (error) goto nfsmout; NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); } else if (nd->nd_flag & ND_NFSV3) { NFSM_DISSECT(tl, u_int32_t *, 3*NFSX_UNSIGNED); ncookie.lval[0] = *tl++; ncookie.lval[1] = *tl++; } else { NFSM_DISSECT(tl, u_int32_t *, 2*NFSX_UNSIGNED); ncookie.lval[0] = 0; ncookie.lval[1] = *tl++; } if (bigenough) { if (nd->nd_flag & ND_NFSV4) { if (rderr) { dp->d_fileno = 0; } else { if (gotmnton) { if (nfsva.na_mntonfileno != 0xffffffff) dp->d_fileno = nfsva.na_mntonfileno; else dp->d_fileno = nfsva.na_fileid; } else if (nfsva.na_filesid[0] == dnp->n_vattr.na_filesid[0] && nfsva.na_filesid[1] == dnp->n_vattr.na_filesid[1]) { dp->d_fileno = nfsva.na_fileid; } else { do { fakefileno--; } while (fakefileno == nfsva.na_fileid); dp->d_fileno = fakefileno; } dp->d_type = vtonfs_dtype(nfsva.na_type); } } else { dp->d_fileno = nfsva.na_fileid; } *tl2++ = cookiep->nfsuquad[0] = cookie.lval[0] = ncookie.lval[0]; *tl2 = cookiep->nfsuquad[1] = cookie.lval[1] = ncookie.lval[1]; } more_dirs = fxdr_unsigned(int, *tl); } /* * If at end of rpc data, get the eof boolean */ if (!more_dirs) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); eof = fxdr_unsigned(int, *tl); if (tryformoredirs) more_dirs = !eof; if (nd->nd_flag & ND_NFSV4) { error = nfscl_postop_attr(nd, nap, attrflagp, stuff); if (error) goto nfsmout; } } mbuf_freem(nd->nd_mrep); nd->nd_mrep = NULL; } /* * Fill last record, iff any, out to a multiple of DIRBLKSIZ * by increasing d_reclen for the last record. */ if (blksiz > 0) { left = DIRBLKSIZ - blksiz; dp->d_reclen += left; uio_iov_base_add(uiop, left); uio_iov_len_add(uiop, -(left)); uio_uio_resid_add(uiop, -(left)); uiop->uio_offset += left; } /* * If returning no data, assume end of file. * If not bigenough, return not end of file, since you aren't * returning all the data * Otherwise, return the eof flag from the server. */ if (eofp) { if (tresid == ((size_t)(uio_uio_resid(uiop)))) *eofp = 1; else if (!bigenough) *eofp = 0; else *eofp = eof; } /* * Add extra empty records to any remaining DIRBLKSIZ chunks. */ while (uio_uio_resid(uiop) > 0 && ((size_t)(uio_uio_resid(uiop))) != tresid) { dp = (struct dirent *) CAST_DOWN(caddr_t, uio_iov_base(uiop)); dp->d_type = DT_UNKNOWN; dp->d_fileno = 0; dp->d_namlen = 0; dp->d_name[0] = '\0'; tl = (u_int32_t *)&dp->d_name[4]; *tl++ = cookie.lval[0]; *tl = cookie.lval[1]; dp->d_reclen = DIRBLKSIZ; uio_iov_base_add(uiop, DIRBLKSIZ); uio_iov_len_add(uiop, -(DIRBLKSIZ)); uio_uio_resid_add(uiop, -(DIRBLKSIZ)); uiop->uio_offset += DIRBLKSIZ; } nfsmout: if (nd->nd_mrep != NULL) mbuf_freem(nd->nd_mrep); return (error); } #ifndef APPLE /* * NFS V3 readdir plus RPC. Used in place of nfsrpc_readdir(). * (Also used for NFS V4 when mount flag set.) * (ditto above w.r.t. multiple of DIRBLKSIZ, etc.) */ APPLESTATIC int nfsrpc_readdirplus(vnode_t vp, struct uio *uiop, nfsuint64 *cookiep, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, int *attrflagp, int *eofp, void *stuff) { int len, left; struct dirent *dp = NULL; u_int32_t *tl; vnode_t newvp = NULLVP; struct nfsrv_descript nfsd, *nd = &nfsd; struct nameidata nami, *ndp = &nami; struct componentname *cnp = &ndp->ni_cnd; struct nfsmount *nmp = VFSTONFS(vnode_mount(vp)); struct nfsnode *dnp = VTONFS(vp), *np; struct nfsvattr nfsva; struct nfsfh *nfhp; nfsquad_t cookie, ncookie; int error = 0, tlen, more_dirs = 1, blksiz = 0, bigenough = 1; int attrflag, tryformoredirs = 1, eof = 0, gotmnton = 0; int isdotdot = 0, unlocknewvp = 0; long dotfileid, dotdotfileid = 0, fileno = 0; char *cp; nfsattrbit_t attrbits, dattrbits; size_t tresid; u_int32_t *tl2 = NULL, fakefileno = 0xffffffff, rderr; KASSERT(uiop->uio_iovcnt == 1 && (uio_uio_resid(uiop) & (DIRBLKSIZ - 1)) == 0, ("nfs readdirplusrpc bad uio")); *attrflagp = 0; if (eofp != NULL) *eofp = 0; ndp->ni_dvp = vp; nd->nd_mrep = NULL; cookie.lval[0] = cookiep->nfsuquad[0]; cookie.lval[1] = cookiep->nfsuquad[1]; tresid = uio_uio_resid(uiop); /* * For NFSv4, first create the "." and ".." entries. */ if (NFSHASNFSV4(nmp)) { NFSGETATTR_ATTRBIT(&dattrbits); NFSZERO_ATTRBIT(&attrbits); NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_FILEID); if (NFSISSET_ATTRBIT(&dnp->n_vattr.na_suppattr, NFSATTRBIT_MOUNTEDONFILEID)) { NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_MOUNTEDONFILEID); gotmnton = 1; } else { /* * Must fake it. Use the fileno, except when the * fsid is != to that of the directory. For that * case, generate a fake fileno that is not the same. */ NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_FSID); gotmnton = 0; } /* * Joy, oh joy. For V4 we get to hand craft '.' and '..'. */ if (uiop->uio_offset == 0) { #if defined(__FreeBSD_version) && __FreeBSD_version >= 800000 error = VOP_GETATTR(vp, &nfsva.na_vattr, cred); #else error = VOP_GETATTR(vp, &nfsva.na_vattr, cred, p); #endif if (error) return (error); dotfileid = nfsva.na_fileid; NFSCL_REQSTART(nd, NFSPROC_LOOKUPP, vp); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV4OP_GETFH); *tl = txdr_unsigned(NFSV4OP_GETATTR); (void) nfsrv_putattrbit(nd, &attrbits); error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (nd->nd_repstat == 0) { NFSM_DISSECT(tl, u_int32_t *, 3*NFSX_UNSIGNED); len = fxdr_unsigned(int, *(tl + 2)); if (len > 0 && len <= NFSX_V4FHMAX) error = nfsm_advance(nd, NFSM_RNDUP(len), -1); else error = EPERM; if (!error) { NFSM_DISSECT(tl, u_int32_t *, 2*NFSX_UNSIGNED); nfsva.na_mntonfileno = 0xffffffff; error = nfsv4_loadattr(nd, NULL, &nfsva, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, p, cred); if (error) { dotdotfileid = dotfileid; } else if (gotmnton) { if (nfsva.na_mntonfileno != 0xffffffff) dotdotfileid = nfsva.na_mntonfileno; else dotdotfileid = nfsva.na_fileid; } else if (nfsva.na_filesid[0] == dnp->n_vattr.na_filesid[0] && nfsva.na_filesid[1] == dnp->n_vattr.na_filesid[1]) { dotdotfileid = nfsva.na_fileid; } else { do { fakefileno--; } while (fakefileno == nfsva.na_fileid); dotdotfileid = fakefileno; } } } else if (nd->nd_repstat == NFSERR_NOENT) { /* * Lookupp returns NFSERR_NOENT when we are * at the root, so just use the current dir. */ nd->nd_repstat = 0; dotdotfileid = dotfileid; } else { error = nd->nd_repstat; } mbuf_freem(nd->nd_mrep); if (error) return (error); nd->nd_mrep = NULL; dp = (struct dirent *)uio_iov_base(uiop); dp->d_type = DT_DIR; dp->d_fileno = dotfileid; dp->d_namlen = 1; dp->d_name[0] = '.'; dp->d_name[1] = '\0'; dp->d_reclen = DIRENT_SIZE(dp) + NFSX_HYPER; /* * Just make these offset cookie 0. */ tl = (u_int32_t *)&dp->d_name[4]; *tl++ = 0; *tl = 0; blksiz += dp->d_reclen; uio_uio_resid_add(uiop, -(dp->d_reclen)); uiop->uio_offset += dp->d_reclen; uio_iov_base_add(uiop, dp->d_reclen); uio_iov_len_add(uiop, -(dp->d_reclen)); dp = (struct dirent *)uio_iov_base(uiop); dp->d_type = DT_DIR; dp->d_fileno = dotdotfileid; dp->d_namlen = 2; dp->d_name[0] = '.'; dp->d_name[1] = '.'; dp->d_name[2] = '\0'; dp->d_reclen = DIRENT_SIZE(dp) + NFSX_HYPER; /* * Just make these offset cookie 0. */ tl = (u_int32_t *)&dp->d_name[4]; *tl++ = 0; *tl = 0; blksiz += dp->d_reclen; uio_uio_resid_add(uiop, -(dp->d_reclen)); uiop->uio_offset += dp->d_reclen; uio_iov_base_add(uiop, dp->d_reclen); uio_iov_len_add(uiop, -(dp->d_reclen)); } NFSREADDIRPLUS_ATTRBIT(&attrbits); if (gotmnton) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_MOUNTEDONFILEID); } /* * Loop around doing readdir rpc's of size nm_readdirsize. * The stopping criteria is EOF or buffer full. */ while (more_dirs && bigenough) { *attrflagp = 0; NFSCL_REQSTART(nd, NFSPROC_READDIRPLUS, vp); NFSM_BUILD(tl, u_int32_t *, 6 * NFSX_UNSIGNED); *tl++ = cookie.lval[0]; *tl++ = cookie.lval[1]; if (cookie.qval == 0) { *tl++ = 0; *tl++ = 0; } else { NFSLOCKNODE(dnp); *tl++ = dnp->n_cookieverf.nfsuquad[0]; *tl++ = dnp->n_cookieverf.nfsuquad[1]; NFSUNLOCKNODE(dnp); } *tl++ = txdr_unsigned(nmp->nm_readdirsize); *tl = txdr_unsigned(nmp->nm_readdirsize); if (nd->nd_flag & ND_NFSV4) { (void) nfsrv_putattrbit(nd, &attrbits); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_GETATTR); (void) nfsrv_putattrbit(nd, &dattrbits); } error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (nd->nd_flag & ND_NFSV3) error = nfscl_postop_attr(nd, nap, attrflagp, stuff); if (nd->nd_repstat || error) { if (!error) error = nd->nd_repstat; goto nfsmout; } NFSM_DISSECT(tl, u_int32_t *, 3 * NFSX_UNSIGNED); NFSLOCKNODE(dnp); dnp->n_cookieverf.nfsuquad[0] = *tl++; dnp->n_cookieverf.nfsuquad[1] = *tl++; NFSUNLOCKNODE(dnp); more_dirs = fxdr_unsigned(int, *tl); if (!more_dirs) tryformoredirs = 0; /* loop thru the dir entries, doctoring them to 4bsd form */ while (more_dirs && bigenough) { NFSM_DISSECT(tl, u_int32_t *, 3 * NFSX_UNSIGNED); if (nd->nd_flag & ND_NFSV4) { ncookie.lval[0] = *tl++; ncookie.lval[1] = *tl++; } else { fileno = fxdr_unsigned(long, *++tl); tl++; } len = fxdr_unsigned(int, *tl); if (len <= 0 || len > NFS_MAXNAMLEN) { error = EBADRPC; goto nfsmout; } tlen = NFSM_RNDUP(len); if (tlen == len) tlen += 4; /* To ensure null termination */ left = DIRBLKSIZ - blksiz; if ((tlen + DIRHDSIZ + NFSX_HYPER) > left) { dp->d_reclen += left; uio_iov_base_add(uiop, left); uio_iov_len_add(uiop, -(left)); uio_uio_resid_add(uiop, -(left)); uiop->uio_offset += left; blksiz = 0; } if ((tlen + DIRHDSIZ + NFSX_HYPER) > uio_uio_resid(uiop)) bigenough = 0; if (bigenough) { dp = (struct dirent *)uio_iov_base(uiop); dp->d_namlen = len; dp->d_reclen = tlen + DIRHDSIZ + NFSX_HYPER; dp->d_type = DT_UNKNOWN; blksiz += dp->d_reclen; if (blksiz == DIRBLKSIZ) blksiz = 0; uio_uio_resid_add(uiop, -(DIRHDSIZ)); uiop->uio_offset += DIRHDSIZ; uio_iov_base_add(uiop, DIRHDSIZ); uio_iov_len_add(uiop, -(DIRHDSIZ)); cnp->cn_nameptr = uio_iov_base(uiop); cnp->cn_namelen = len; NFSCNHASHZERO(cnp); error = nfsm_mbufuio(nd, uiop, len); if (error) goto nfsmout; cp = uio_iov_base(uiop); tlen -= len; *cp = '\0'; cp += tlen; /* points to cookie storage */ tl2 = (u_int32_t *)cp; if (len == 2 && cnp->cn_nameptr[0] == '.' && cnp->cn_nameptr[1] == '.') isdotdot = 1; else isdotdot = 0; uio_iov_base_add(uiop, (tlen + NFSX_HYPER)); uio_iov_len_add(uiop, -(tlen + NFSX_HYPER)); uio_uio_resid_add(uiop, -(tlen + NFSX_HYPER)); uiop->uio_offset += (tlen + NFSX_HYPER); } else { error = nfsm_advance(nd, NFSM_RNDUP(len), -1); if (error) goto nfsmout; } nfhp = NULL; if (nd->nd_flag & ND_NFSV3) { NFSM_DISSECT(tl, u_int32_t *, 3*NFSX_UNSIGNED); ncookie.lval[0] = *tl++; ncookie.lval[1] = *tl++; attrflag = fxdr_unsigned(int, *tl); if (attrflag) { error = nfsm_loadattr(nd, &nfsva); if (error) goto nfsmout; } NFSM_DISSECT(tl,u_int32_t *,NFSX_UNSIGNED); if (*tl) { error = nfsm_getfh(nd, &nfhp); if (error) goto nfsmout; } if (!attrflag && nfhp != NULL) { FREE((caddr_t)nfhp, M_NFSFH); nfhp = NULL; } } else { rderr = 0; nfsva.na_mntonfileno = 0xffffffff; error = nfsv4_loadattr(nd, NULL, &nfsva, &nfhp, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, NULL, NULL, &rderr, p, cred); if (error) goto nfsmout; } if (bigenough) { if (nd->nd_flag & ND_NFSV4) { if (rderr) { dp->d_fileno = 0; } else if (gotmnton) { if (nfsva.na_mntonfileno != 0xffffffff) dp->d_fileno = nfsva.na_mntonfileno; else dp->d_fileno = nfsva.na_fileid; } else if (nfsva.na_filesid[0] == dnp->n_vattr.na_filesid[0] && nfsva.na_filesid[1] == dnp->n_vattr.na_filesid[1]) { dp->d_fileno = nfsva.na_fileid; } else { do { fakefileno--; } while (fakefileno == nfsva.na_fileid); dp->d_fileno = fakefileno; } } else { dp->d_fileno = fileno; } *tl2++ = cookiep->nfsuquad[0] = cookie.lval[0] = ncookie.lval[0]; *tl2 = cookiep->nfsuquad[1] = cookie.lval[1] = ncookie.lval[1]; if (nfhp != NULL) { if (NFSRV_CMPFH(nfhp->nfh_fh, nfhp->nfh_len, dnp->n_fhp->nfh_fh, dnp->n_fhp->nfh_len)) { VREF(vp); newvp = vp; unlocknewvp = 0; FREE((caddr_t)nfhp, M_NFSFH); np = dnp; } else if (isdotdot != 0) { /* * Skip doing a nfscl_nget() call for "..". * There's a race between acquiring the nfs * node here and lookups that look for the * directory being read (in the parent). * It would try to get a lock on ".." here, * owning the lock on the directory being * read. Lookup will hold the lock on ".." * and try to acquire the lock on the * directory being read. * If the directory is unlocked/relocked, * then there is a LOR with the buflock * vp is relocked. */ free(nfhp, M_NFSFH); } else { error = nfscl_nget(vnode_mount(vp), vp, nfhp, cnp, p, &np, NULL, LK_EXCLUSIVE); if (!error) { newvp = NFSTOV(np); unlocknewvp = 1; } } nfhp = NULL; if (newvp != NULLVP) { error = nfscl_loadattrcache(&newvp, &nfsva, NULL, NULL, 0, 0); if (error) { if (unlocknewvp) vput(newvp); else vrele(newvp); goto nfsmout; } dp->d_type = vtonfs_dtype(np->n_vattr.na_type); ndp->ni_vp = newvp; NFSCNHASH(cnp, HASHINIT); if (cnp->cn_namelen <= NCHNAMLEN) { np->n_ctime = np->n_vattr.na_ctime; cache_enter(ndp->ni_dvp,ndp->ni_vp,cnp); } if (unlocknewvp) vput(newvp); else vrele(newvp); newvp = NULLVP; } } } else if (nfhp != NULL) { FREE((caddr_t)nfhp, M_NFSFH); } NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); more_dirs = fxdr_unsigned(int, *tl); } /* * If at end of rpc data, get the eof boolean */ if (!more_dirs) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); eof = fxdr_unsigned(int, *tl); if (tryformoredirs) more_dirs = !eof; if (nd->nd_flag & ND_NFSV4) { error = nfscl_postop_attr(nd, nap, attrflagp, stuff); if (error) goto nfsmout; } } mbuf_freem(nd->nd_mrep); nd->nd_mrep = NULL; } /* * Fill last record, iff any, out to a multiple of DIRBLKSIZ * by increasing d_reclen for the last record. */ if (blksiz > 0) { left = DIRBLKSIZ - blksiz; dp->d_reclen += left; uio_iov_base_add(uiop, left); uio_iov_len_add(uiop, -(left)); uio_uio_resid_add(uiop, -(left)); uiop->uio_offset += left; } /* * If returning no data, assume end of file. * If not bigenough, return not end of file, since you aren't * returning all the data * Otherwise, return the eof flag from the server. */ if (eofp != NULL) { if (tresid == uio_uio_resid(uiop)) *eofp = 1; else if (!bigenough) *eofp = 0; else *eofp = eof; } /* * Add extra empty records to any remaining DIRBLKSIZ chunks. */ while (uio_uio_resid(uiop) > 0 && uio_uio_resid(uiop) != tresid) { dp = (struct dirent *)uio_iov_base(uiop); dp->d_type = DT_UNKNOWN; dp->d_fileno = 0; dp->d_namlen = 0; dp->d_name[0] = '\0'; tl = (u_int32_t *)&dp->d_name[4]; *tl++ = cookie.lval[0]; *tl = cookie.lval[1]; dp->d_reclen = DIRBLKSIZ; uio_iov_base_add(uiop, DIRBLKSIZ); uio_iov_len_add(uiop, -(DIRBLKSIZ)); uio_uio_resid_add(uiop, -(DIRBLKSIZ)); uiop->uio_offset += DIRBLKSIZ; } nfsmout: if (nd->nd_mrep != NULL) mbuf_freem(nd->nd_mrep); return (error); } #endif /* !APPLE */ /* * Nfs commit rpc */ APPLESTATIC int nfsrpc_commit(vnode_t vp, u_quad_t offset, int cnt, struct ucred *cred, NFSPROC_T *p, u_char *verfp, struct nfsvattr *nap, int *attrflagp, void *stuff) { u_int32_t *tl; struct nfsrv_descript nfsd, *nd = &nfsd; nfsattrbit_t attrbits; int error; *attrflagp = 0; NFSCL_REQSTART(nd, NFSPROC_COMMIT, vp); NFSM_BUILD(tl, u_int32_t *, 3 * NFSX_UNSIGNED); txdr_hyper(offset, tl); tl += 2; *tl = txdr_unsigned(cnt); if (nd->nd_flag & ND_NFSV4) { /* * And do a Getattr op. */ NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_GETATTR); NFSGETATTR_ATTRBIT(&attrbits); (void) nfsrv_putattrbit(nd, &attrbits); } error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); error = nfscl_wcc_data(nd, vp, nap, attrflagp, NULL, stuff); if (!error && !nd->nd_repstat) { NFSM_DISSECT(tl, u_int32_t *, NFSX_VERF); NFSBCOPY((caddr_t)tl, verfp, NFSX_VERF); if (nd->nd_flag & ND_NFSV4) error = nfscl_postop_attr(nd, nap, attrflagp, stuff); } nfsmout: if (!error && nd->nd_repstat) error = nd->nd_repstat; mbuf_freem(nd->nd_mrep); return (error); } /* * NFS byte range lock rpc. * (Mostly just calls one of the three lower level RPC routines.) */ APPLESTATIC int nfsrpc_advlock(vnode_t vp, off_t size, int op, struct flock *fl, - int reclaim, struct ucred *cred, NFSPROC_T *p) + int reclaim, struct ucred *cred, NFSPROC_T *p, void *id, int flags) { struct nfscllockowner *lp; struct nfsclclient *clp; struct nfsfh *nfhp; struct nfsrv_descript nfsd, *nd = &nfsd; struct nfsmount *nmp = VFSTONFS(vnode_mount(vp)); u_int64_t off, len; off_t start, end; u_int32_t clidrev = 0; int error = 0, newone = 0, expireret = 0, retrycnt, donelocally; int callcnt, dorpc; /* * Convert the flock structure into a start and end and do POSIX * bounds checking. */ switch (fl->l_whence) { case SEEK_SET: case SEEK_CUR: /* * Caller is responsible for adding any necessary offset * when SEEK_CUR is used. */ start = fl->l_start; off = fl->l_start; break; case SEEK_END: start = size + fl->l_start; off = size + fl->l_start; break; default: return (EINVAL); }; if (start < 0) return (EINVAL); if (fl->l_len != 0) { end = start + fl->l_len - 1; if (end < start) return (EINVAL); } len = fl->l_len; if (len == 0) len = NFS64BITSSET; retrycnt = 0; do { nd->nd_repstat = 0; if (op == F_GETLK) { error = nfscl_getcl(vp, cred, p, &clp); if (error) return (error); - error = nfscl_lockt(vp, clp, off, len, fl, p); + error = nfscl_lockt(vp, clp, off, len, fl, p, id, flags); if (!error) { clidrev = clp->nfsc_clientidrev; error = nfsrpc_lockt(nd, vp, clp, off, len, fl, cred, - p); + p, id, flags); } else if (error == -1) { error = 0; } nfscl_clientrelease(clp); } else if (op == F_UNLCK && fl->l_type == F_UNLCK) { /* * We must loop around for all lockowner cases. */ callcnt = 0; error = nfscl_getcl(vp, cred, p, &clp); if (error) return (error); do { error = nfscl_relbytelock(vp, off, len, cred, p, callcnt, - clp, &lp, &dorpc); + clp, id, flags, &lp, &dorpc); /* * If it returns a NULL lp, we're done. */ if (lp == NULL) { if (callcnt == 0) nfscl_clientrelease(clp); else - nfscl_releasealllocks(clp, vp, p); + nfscl_releasealllocks(clp, vp, p, id, flags); return (error); } if (nmp->nm_clp != NULL) clidrev = nmp->nm_clp->nfsc_clientidrev; else clidrev = 0; /* * If the server doesn't support Posix lock semantics, * only allow locks on the entire file, since it won't * handle overlapping byte ranges. * There might still be a problem when a lock * upgrade/downgrade (read<->write) occurs, since the * server "might" expect an unlock first? */ if (dorpc && (lp->nfsl_open->nfso_posixlock || (off == 0 && len == NFS64BITSSET))) { /* * Since the lock records will go away, we must * wait for grace and delay here. */ do { error = nfsrpc_locku(nd, nmp, lp, off, len, NFSV4LOCKT_READ, cred, p, 0); if ((nd->nd_repstat == NFSERR_GRACE || nd->nd_repstat == NFSERR_DELAY) && error == 0) (void) nfs_catnap(PZERO, (int)nd->nd_repstat, "nfs_advlock"); } while ((nd->nd_repstat == NFSERR_GRACE || nd->nd_repstat == NFSERR_DELAY) && error == 0); } callcnt++; } while (error == 0 && nd->nd_repstat == 0); - nfscl_releasealllocks(clp, vp, p); + nfscl_releasealllocks(clp, vp, p, id, flags); } else if (op == F_SETLK) { error = nfscl_getbytelock(vp, off, len, fl->l_type, cred, p, - NULL, 0, NULL, NULL, &lp, &newone, &donelocally); + NULL, 0, id, flags, NULL, NULL, &lp, &newone, &donelocally); if (error || donelocally) { return (error); } if (nmp->nm_clp != NULL) clidrev = nmp->nm_clp->nfsc_clientidrev; else clidrev = 0; nfhp = VTONFS(vp)->n_fhp; if (!lp->nfsl_open->nfso_posixlock && (off != 0 || len != NFS64BITSSET)) { error = EINVAL; } else { error = nfsrpc_lock(nd, nmp, vp, nfhp->nfh_fh, nfhp->nfh_len, lp, newone, reclaim, off, len, fl->l_type, cred, p, 0); } if (!error) error = nd->nd_repstat; nfscl_lockrelease(lp, error, newone); } else { error = EINVAL; } if (!error) error = nd->nd_repstat; if (error == NFSERR_GRACE || error == NFSERR_STALESTATEID || error == NFSERR_STALEDONTRECOVER || error == NFSERR_STALECLIENTID || error == NFSERR_DELAY) { (void) nfs_catnap(PZERO, error, "nfs_advlock"); } else if ((error == NFSERR_EXPIRED || error == NFSERR_BADSTATEID) && clidrev != 0) { expireret = nfscl_hasexpired(nmp->nm_clp, clidrev, p); retrycnt++; } } while (error == NFSERR_GRACE || error == NFSERR_STALECLIENTID || error == NFSERR_DELAY || error == NFSERR_STALEDONTRECOVER || error == NFSERR_STALESTATEID || ((error == NFSERR_EXPIRED || error == NFSERR_BADSTATEID) && expireret == 0 && clidrev != 0 && retrycnt < 4)); if (error && retrycnt >= 4) error = EIO; return (error); } /* * The lower level routine for the LockT case. */ APPLESTATIC int nfsrpc_lockt(struct nfsrv_descript *nd, vnode_t vp, struct nfsclclient *clp, u_int64_t off, u_int64_t len, struct flock *fl, - struct ucred *cred, NFSPROC_T *p) + struct ucred *cred, NFSPROC_T *p, void *id, int flags) { u_int32_t *tl; int error, type, size; u_int8_t own[NFSV4CL_LOCKNAMELEN]; NFSCL_REQSTART(nd, NFSPROC_LOCKT, vp); NFSM_BUILD(tl, u_int32_t *, 7 * NFSX_UNSIGNED); if (fl->l_type == F_RDLCK) *tl++ = txdr_unsigned(NFSV4LOCKT_READ); else *tl++ = txdr_unsigned(NFSV4LOCKT_WRITE); txdr_hyper(off, tl); tl += 2; txdr_hyper(len, tl); tl += 2; *tl++ = clp->nfsc_clientid.lval[0]; *tl = clp->nfsc_clientid.lval[1]; - nfscl_filllockowner(p, own); + nfscl_filllockowner(id, own, flags); (void) nfsm_strtom(nd, own, NFSV4CL_LOCKNAMELEN); error = nfscl_request(nd, vp, p, cred, NULL); if (error) return (error); if (nd->nd_repstat == 0) { fl->l_type = F_UNLCK; } else if (nd->nd_repstat == NFSERR_DENIED) { nd->nd_repstat = 0; fl->l_whence = SEEK_SET; NFSM_DISSECT(tl, u_int32_t *, 8 * NFSX_UNSIGNED); fl->l_start = fxdr_hyper(tl); tl += 2; len = fxdr_hyper(tl); tl += 2; if (len == NFS64BITSSET) fl->l_len = 0; else fl->l_len = len; type = fxdr_unsigned(int, *tl++); if (type == NFSV4LOCKT_WRITE) fl->l_type = F_WRLCK; else fl->l_type = F_RDLCK; /* * XXX For now, I have no idea what to do with the * conflicting lock_owner, so I'll just set the pid == 0 * and skip over the lock_owner. */ fl->l_pid = (pid_t)0; tl += 2; size = fxdr_unsigned(int, *tl); if (size < 0 || size > NFSV4_OPAQUELIMIT) error = EBADRPC; if (!error) error = nfsm_advance(nd, NFSM_RNDUP(size), -1); } else if (nd->nd_repstat == NFSERR_STALECLIENTID) nfscl_initiate_recovery(clp); nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * Lower level function that performs the LockU RPC. */ static int nfsrpc_locku(struct nfsrv_descript *nd, struct nfsmount *nmp, struct nfscllockowner *lp, u_int64_t off, u_int64_t len, u_int32_t type, struct ucred *cred, NFSPROC_T *p, int syscred) { u_int32_t *tl; int error; nfscl_reqstart(nd, NFSPROC_LOCKU, nmp, lp->nfsl_open->nfso_fh, lp->nfsl_open->nfso_fhlen, NULL); NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID + 6 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(type); *tl = txdr_unsigned(lp->nfsl_seqid); if (nfstest_outofseq && (arc4random() % nfstest_outofseq) == 0) *tl = txdr_unsigned(lp->nfsl_seqid + 1); tl++; *tl++ = lp->nfsl_stateid.seqid; *tl++ = lp->nfsl_stateid.other[0]; *tl++ = lp->nfsl_stateid.other[1]; *tl++ = lp->nfsl_stateid.other[2]; txdr_hyper(off, tl); tl += 2; txdr_hyper(len, tl); if (syscred) nd->nd_flag |= ND_USEGSSNAME; error = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, NULL, p, cred, NFS_PROG, NFS_VER4, NULL, 1, NULL); NFSCL_INCRSEQID(lp->nfsl_seqid, nd); if (error) return (error); if (nd->nd_repstat == 0) { NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID); lp->nfsl_stateid.seqid = *tl++; lp->nfsl_stateid.other[0] = *tl++; lp->nfsl_stateid.other[1] = *tl++; lp->nfsl_stateid.other[2] = *tl; } else if (nd->nd_repstat == NFSERR_STALESTATEID) nfscl_initiate_recovery(lp->nfsl_open->nfso_own->nfsow_clp); nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * The actual Lock RPC. */ APPLESTATIC int nfsrpc_lock(struct nfsrv_descript *nd, struct nfsmount *nmp, vnode_t vp, u_int8_t *nfhp, int fhlen, struct nfscllockowner *lp, int newone, int reclaim, u_int64_t off, u_int64_t len, short type, struct ucred *cred, NFSPROC_T *p, int syscred) { u_int32_t *tl; int error, size; nfscl_reqstart(nd, NFSPROC_LOCK, nmp, nfhp, fhlen, NULL); NFSM_BUILD(tl, u_int32_t *, 7 * NFSX_UNSIGNED); if (type == F_RDLCK) *tl++ = txdr_unsigned(NFSV4LOCKT_READ); else *tl++ = txdr_unsigned(NFSV4LOCKT_WRITE); *tl++ = txdr_unsigned(reclaim); txdr_hyper(off, tl); tl += 2; txdr_hyper(len, tl); tl += 2; if (newone) { *tl = newnfs_true; NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID + 2 * NFSX_UNSIGNED + NFSX_HYPER); *tl++ = txdr_unsigned(lp->nfsl_open->nfso_own->nfsow_seqid); *tl++ = lp->nfsl_open->nfso_stateid.seqid; *tl++ = lp->nfsl_open->nfso_stateid.other[0]; *tl++ = lp->nfsl_open->nfso_stateid.other[1]; *tl++ = lp->nfsl_open->nfso_stateid.other[2]; *tl++ = txdr_unsigned(lp->nfsl_seqid); *tl++ = lp->nfsl_open->nfso_own->nfsow_clp->nfsc_clientid.lval[0]; *tl = lp->nfsl_open->nfso_own->nfsow_clp->nfsc_clientid.lval[1]; (void) nfsm_strtom(nd, lp->nfsl_owner, NFSV4CL_LOCKNAMELEN); } else { *tl = newnfs_false; NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID + NFSX_UNSIGNED); *tl++ = lp->nfsl_stateid.seqid; *tl++ = lp->nfsl_stateid.other[0]; *tl++ = lp->nfsl_stateid.other[1]; *tl++ = lp->nfsl_stateid.other[2]; *tl = txdr_unsigned(lp->nfsl_seqid); if (nfstest_outofseq && (arc4random() % nfstest_outofseq) == 0) *tl = txdr_unsigned(lp->nfsl_seqid + 1); } if (syscred) nd->nd_flag |= ND_USEGSSNAME; error = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, vp, p, cred, NFS_PROG, NFS_VER4, NULL, 1, NULL); if (error) return (error); if (newone) NFSCL_INCRSEQID(lp->nfsl_open->nfso_own->nfsow_seqid, nd); NFSCL_INCRSEQID(lp->nfsl_seqid, nd); if (nd->nd_repstat == 0) { NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID); lp->nfsl_stateid.seqid = *tl++; lp->nfsl_stateid.other[0] = *tl++; lp->nfsl_stateid.other[1] = *tl++; lp->nfsl_stateid.other[2] = *tl; } else if (nd->nd_repstat == NFSERR_DENIED) { NFSM_DISSECT(tl, u_int32_t *, 8 * NFSX_UNSIGNED); size = fxdr_unsigned(int, *(tl + 7)); if (size < 0 || size > NFSV4_OPAQUELIMIT) error = EBADRPC; if (!error) error = nfsm_advance(nd, NFSM_RNDUP(size), -1); } else if (nd->nd_repstat == NFSERR_STALESTATEID) nfscl_initiate_recovery(lp->nfsl_open->nfso_own->nfsow_clp); nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * nfs statfs rpc * (always called with the vp for the mount point) */ APPLESTATIC int nfsrpc_statfs(vnode_t vp, struct nfsstatfs *sbp, struct nfsfsinfo *fsp, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, int *attrflagp, void *stuff) { u_int32_t *tl = NULL; struct nfsrv_descript nfsd, *nd = &nfsd; struct nfsmount *nmp; nfsattrbit_t attrbits; int error; *attrflagp = 0; nmp = VFSTONFS(vnode_mount(vp)); if (NFSHASNFSV4(nmp)) { /* * For V4, you actually do a getattr. */ NFSCL_REQSTART(nd, NFSPROC_GETATTR, vp); NFSSTATFS_GETATTRBIT(&attrbits); (void) nfsrv_putattrbit(nd, &attrbits); nd->nd_flag |= ND_USEGSSNAME; error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (nd->nd_repstat == 0) { error = nfsv4_loadattr(nd, NULL, nap, NULL, NULL, 0, NULL, NULL, sbp, fsp, NULL, 0, NULL, NULL, NULL, p, cred); if (!error) { nmp->nm_fsid[0] = nap->na_filesid[0]; nmp->nm_fsid[1] = nap->na_filesid[1]; NFSSETHASSETFSID(nmp); *attrflagp = 1; } } else { error = nd->nd_repstat; } if (error) goto nfsmout; } else { NFSCL_REQSTART(nd, NFSPROC_FSSTAT, vp); error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (nd->nd_flag & ND_NFSV3) { error = nfscl_postop_attr(nd, nap, attrflagp, stuff); if (error) goto nfsmout; } if (nd->nd_repstat) { error = nd->nd_repstat; goto nfsmout; } NFSM_DISSECT(tl, u_int32_t *, NFSX_STATFS(nd->nd_flag & ND_NFSV3)); } if (NFSHASNFSV3(nmp)) { sbp->sf_tbytes = fxdr_hyper(tl); tl += 2; sbp->sf_fbytes = fxdr_hyper(tl); tl += 2; sbp->sf_abytes = fxdr_hyper(tl); tl += 2; sbp->sf_tfiles = fxdr_hyper(tl); tl += 2; sbp->sf_ffiles = fxdr_hyper(tl); tl += 2; sbp->sf_afiles = fxdr_hyper(tl); tl += 2; sbp->sf_invarsec = fxdr_unsigned(u_int32_t, *tl); } else if (NFSHASNFSV4(nmp) == 0) { sbp->sf_tsize = fxdr_unsigned(u_int32_t, *tl++); sbp->sf_bsize = fxdr_unsigned(u_int32_t, *tl++); sbp->sf_blocks = fxdr_unsigned(u_int32_t, *tl++); sbp->sf_bfree = fxdr_unsigned(u_int32_t, *tl++); sbp->sf_bavail = fxdr_unsigned(u_int32_t, *tl); } nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * nfs pathconf rpc */ APPLESTATIC int nfsrpc_pathconf(vnode_t vp, struct nfsv3_pathconf *pc, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, int *attrflagp, void *stuff) { struct nfsrv_descript nfsd, *nd = &nfsd; struct nfsmount *nmp; u_int32_t *tl; nfsattrbit_t attrbits; int error; *attrflagp = 0; nmp = VFSTONFS(vnode_mount(vp)); if (NFSHASNFSV4(nmp)) { /* * For V4, you actually do a getattr. */ NFSCL_REQSTART(nd, NFSPROC_GETATTR, vp); NFSPATHCONF_GETATTRBIT(&attrbits); (void) nfsrv_putattrbit(nd, &attrbits); nd->nd_flag |= ND_USEGSSNAME; error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (nd->nd_repstat == 0) { error = nfsv4_loadattr(nd, NULL, nap, NULL, NULL, 0, pc, NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, p, cred); if (!error) *attrflagp = 1; } else { error = nd->nd_repstat; } } else { NFSCL_REQSTART(nd, NFSPROC_PATHCONF, vp); error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); error = nfscl_postop_attr(nd, nap, attrflagp, stuff); if (nd->nd_repstat && !error) error = nd->nd_repstat; if (!error) { NFSM_DISSECT(tl, u_int32_t *, NFSX_V3PATHCONF); pc->pc_linkmax = fxdr_unsigned(u_int32_t, *tl++); pc->pc_namemax = fxdr_unsigned(u_int32_t, *tl++); pc->pc_notrunc = fxdr_unsigned(u_int32_t, *tl++); pc->pc_chownrestricted = fxdr_unsigned(u_int32_t, *tl++); pc->pc_caseinsensitive = fxdr_unsigned(u_int32_t, *tl++); pc->pc_casepreserving = fxdr_unsigned(u_int32_t, *tl); } } nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * nfs version 3 fsinfo rpc call */ APPLESTATIC int nfsrpc_fsinfo(vnode_t vp, struct nfsfsinfo *fsp, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, int *attrflagp, void *stuff) { u_int32_t *tl; struct nfsrv_descript nfsd, *nd = &nfsd; int error; *attrflagp = 0; NFSCL_REQSTART(nd, NFSPROC_FSINFO, vp); error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); error = nfscl_postop_attr(nd, nap, attrflagp, stuff); if (nd->nd_repstat && !error) error = nd->nd_repstat; if (!error) { NFSM_DISSECT(tl, u_int32_t *, NFSX_V3FSINFO); fsp->fs_rtmax = fxdr_unsigned(u_int32_t, *tl++); fsp->fs_rtpref = fxdr_unsigned(u_int32_t, *tl++); fsp->fs_rtmult = fxdr_unsigned(u_int32_t, *tl++); fsp->fs_wtmax = fxdr_unsigned(u_int32_t, *tl++); fsp->fs_wtpref = fxdr_unsigned(u_int32_t, *tl++); fsp->fs_wtmult = fxdr_unsigned(u_int32_t, *tl++); fsp->fs_dtpref = fxdr_unsigned(u_int32_t, *tl++); fsp->fs_maxfilesize = fxdr_hyper(tl); tl += 2; fxdr_nfsv3time(tl, &fsp->fs_timedelta); tl += 2; fsp->fs_properties = fxdr_unsigned(u_int32_t, *tl); } nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * This function performs the Renew RPC. */ APPLESTATIC int nfsrpc_renew(struct nfsclclient *clp, struct ucred *cred, NFSPROC_T *p) { u_int32_t *tl; struct nfsrv_descript nfsd; struct nfsrv_descript *nd = &nfsd; struct nfsmount *nmp; int error; nmp = clp->nfsc_nmp; if (nmp == NULL) return (0); nfscl_reqstart(nd, NFSPROC_RENEW, nmp, NULL, 0, NULL); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = clp->nfsc_clientid.lval[0]; *tl = clp->nfsc_clientid.lval[1]; nd->nd_flag |= ND_USEGSSNAME; error = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, NULL, p, cred, NFS_PROG, NFS_VER4, NULL, 1, NULL); if (error) return (error); error = nd->nd_repstat; mbuf_freem(nd->nd_mrep); return (error); } /* * This function performs the Releaselockowner RPC. */ APPLESTATIC int nfsrpc_rellockown(struct nfsmount *nmp, struct nfscllockowner *lp, struct ucred *cred, NFSPROC_T *p) { struct nfsrv_descript nfsd, *nd = &nfsd; u_int32_t *tl; int error; nfscl_reqstart(nd, NFSPROC_RELEASELCKOWN, nmp, NULL, 0, NULL); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = nmp->nm_clp->nfsc_clientid.lval[0]; *tl = nmp->nm_clp->nfsc_clientid.lval[1]; (void) nfsm_strtom(nd, lp->nfsl_owner, NFSV4CL_LOCKNAMELEN); nd->nd_flag |= ND_USEGSSNAME; error = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, NULL, p, cred, NFS_PROG, NFS_VER4, NULL, 1, NULL); if (error) return (error); error = nd->nd_repstat; mbuf_freem(nd->nd_mrep); return (error); } /* * This function performs the Compound to get the mount pt FH. */ APPLESTATIC int nfsrpc_getdirpath(struct nfsmount *nmp, u_char *dirpath, struct ucred *cred, NFSPROC_T *p) { u_int32_t *tl; struct nfsrv_descript nfsd; struct nfsrv_descript *nd = &nfsd; u_char *cp, *cp2; int error, cnt, len, setnil; u_int32_t *opcntp; nfscl_reqstart(nd, NFSPROC_PUTROOTFH, nmp, NULL, 0, &opcntp); cp = dirpath; cnt = 0; do { setnil = 0; while (*cp == '/') cp++; cp2 = cp; while (*cp2 != '\0' && *cp2 != '/') cp2++; if (*cp2 == '/') { setnil = 1; *cp2 = '\0'; } if (cp2 != cp) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_LOOKUP); nfsm_strtom(nd, cp, strlen(cp)); cnt++; } if (setnil) *cp2++ = '/'; cp = cp2; } while (*cp != '\0'); *opcntp = txdr_unsigned(2 + cnt); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OP_GETFH); nd->nd_flag |= ND_USEGSSNAME; error = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, NULL, p, cred, NFS_PROG, NFS_VER4, NULL, 1, NULL); if (error) return (error); if (nd->nd_repstat == 0) { NFSM_DISSECT(tl, u_int32_t *, (3 + 2 * cnt) * NFSX_UNSIGNED); tl += (2 + 2 * cnt); if ((len = fxdr_unsigned(int, *tl)) <= 0 || len > NFSX_FHMAX) { nd->nd_repstat = NFSERR_BADXDR; } else { nd->nd_repstat = nfsrv_mtostr(nd, nmp->nm_fh, len); if (nd->nd_repstat == 0) nmp->nm_fhsize = len; } } error = nd->nd_repstat; nfsmout: mbuf_freem(nd->nd_mrep); return (error); } /* * This function performs the Delegreturn RPC. */ APPLESTATIC int nfsrpc_delegreturn(struct nfscldeleg *dp, struct ucred *cred, struct nfsmount *nmp, NFSPROC_T *p, int syscred) { u_int32_t *tl; struct nfsrv_descript nfsd; struct nfsrv_descript *nd = &nfsd; int error; nfscl_reqstart(nd, NFSPROC_DELEGRETURN, nmp, dp->nfsdl_fh, dp->nfsdl_fhlen, NULL); NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID); *tl++ = dp->nfsdl_stateid.seqid; *tl++ = dp->nfsdl_stateid.other[0]; *tl++ = dp->nfsdl_stateid.other[1]; *tl = dp->nfsdl_stateid.other[2]; if (syscred) nd->nd_flag |= ND_USEGSSNAME; error = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, NULL, p, cred, NFS_PROG, NFS_VER4, NULL, 1, NULL); if (error) return (error); error = nd->nd_repstat; mbuf_freem(nd->nd_mrep); return (error); } /* * nfs getacl call. */ APPLESTATIC int nfsrpc_getacl(vnode_t vp, struct ucred *cred, NFSPROC_T *p, struct acl *aclp, void *stuff) { struct nfsrv_descript nfsd, *nd = &nfsd; int error; nfsattrbit_t attrbits; struct nfsmount *nmp = VFSTONFS(vnode_mount(vp)); if (nfsrv_useacl == 0 || !NFSHASNFSV4(nmp)) return (EOPNOTSUPP); NFSCL_REQSTART(nd, NFSPROC_GETACL, vp); NFSZERO_ATTRBIT(&attrbits); NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_ACL); (void) nfsrv_putattrbit(nd, &attrbits); error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); if (!nd->nd_repstat) error = nfsv4_loadattr(nd, vp, NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, aclp, 0, NULL, NULL, NULL, p, cred); else error = nd->nd_repstat; mbuf_freem(nd->nd_mrep); return (error); } /* * nfs setacl call. */ APPLESTATIC int nfsrpc_setacl(vnode_t vp, struct ucred *cred, NFSPROC_T *p, struct acl *aclp, void *stuff) { int error; struct nfsmount *nmp = VFSTONFS(vnode_mount(vp)); if (nfsrv_useacl == 0 || !NFSHASNFSV4(nmp)) return (EOPNOTSUPP); error = nfsrpc_setattr(vp, NULL, aclp, cred, p, NULL, NULL, stuff); return (error); } /* * nfs setacl call. */ static int nfsrpc_setaclrpc(vnode_t vp, struct ucred *cred, NFSPROC_T *p, struct acl *aclp, nfsv4stateid_t *stateidp, void *stuff) { struct nfsrv_descript nfsd, *nd = &nfsd; int error; nfsattrbit_t attrbits; struct nfsmount *nmp = VFSTONFS(vnode_mount(vp)); if (!NFSHASNFSV4(nmp)) return (EOPNOTSUPP); NFSCL_REQSTART(nd, NFSPROC_SETACL, vp); nfsm_stateidtom(nd, stateidp, NFSSTATEID_PUTSTATEID); NFSZERO_ATTRBIT(&attrbits); NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_ACL); (void) nfsv4_fillattr(nd, vnode_mount(vp), vp, aclp, NULL, NULL, 0, &attrbits, NULL, NULL, 0, 0, 0, 0, (uint64_t)0); error = nfscl_request(nd, vp, p, cred, stuff); if (error) return (error); /* Don't care about the pre/postop attributes */ mbuf_freem(nd->nd_mrep); return (nd->nd_repstat); } Index: projects/largeSMP/sys/fs/nfsclient/nfs_clstate.c =================================================================== --- projects/largeSMP/sys/fs/nfsclient/nfs_clstate.c (revision 222782) +++ projects/largeSMP/sys/fs/nfsclient/nfs_clstate.c (revision 222783) @@ -1,4263 +1,4265 @@ /*- * Copyright (c) 2009 Rick Macklem, University of Guelph * 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("$FreeBSD$"); /* * These functions implement the client side state handling for NFSv4. * NFSv4 state handling: * - A lockowner is used to determine lock contention, so it * corresponds directly to a Posix pid. (1 to 1 mapping) * - The correct granularity of an OpenOwner is not nearly so * obvious. An OpenOwner does the following: * - provides a serial sequencing of Open/Close/Lock-with-new-lockowner * - is used to check for Open/Share contention (not applicable to * this client, since all Opens are Deny_None) * As such, I considered both extreme. * 1 OpenOwner per ClientID - Simple to manage, but fully serializes * all Open, Close and Lock (with a new lockowner) Ops. * 1 OpenOwner for each Open - This one results in an OpenConfirm for * every Open, for most servers. * So, I chose to use the same mapping as I did for LockOwnwers. * The main concern here is that you can end up with multiple Opens * for the same File Handle, but on different OpenOwners (opens * inherited from parents, grandparents...) and you do not know * which of these the vnodeop close applies to. This is handled by * delaying the Close Op(s) until all of the Opens have been closed. * (It is not yet obvious if this is the correct granularity.) * - How the code handles serialization: * - For the ClientId, it uses an exclusive lock while getting its * SetClientId and during recovery. Otherwise, it uses a shared * lock via a reference count. * - For the rest of the data structures, it uses an SMP mutex * (once the nfs client is SMP safe) and doesn't sleep while * manipulating the linked lists. * - The serialization of Open/Close/Lock/LockU falls out in the * "wash", since OpenOwners and LockOwners are both mapped from * Posix pid. In other words, there is only one Posix pid using * any given owner, so that owner is serialized. (If you change * the granularity of the OpenOwner, then code must be added to * serialize Ops on the OpenOwner.) * - When to get rid of OpenOwners and LockOwners. * - When a process exits, it calls nfscl_cleanup(), which goes * through the client list looking for all Open and Lock Owners. * When one is found, it is marked "defunct" or in the case of * an OpenOwner without any Opens, freed. * The renew thread scans for defunct Owners and gets rid of them, * if it can. The LockOwners will also be deleted when the * associated Open is closed. * - If the LockU or Close Op(s) fail during close in a way * that could be recovered upon retry, they are relinked to the * ClientId's defunct open list and retried by the renew thread * until they succeed or an unmount/recovery occurs. * (Since we are done with them, they do not need to be recovered.) */ #ifndef APPLEKEXT #include /* * Global variables */ extern struct nfsstats newnfsstats; extern struct nfsreqhead nfsd_reqq; NFSREQSPINLOCK; NFSCLSTATEMUTEX; int nfscl_inited = 0; struct nfsclhead nfsclhead; /* Head of clientid list */ int nfscl_deleghighwater = NFSCLDELEGHIGHWATER; #endif /* !APPLEKEXT */ static int nfscl_delegcnt = 0; static int nfscl_getopen(struct nfsclownerhead *, u_int8_t *, int, u_int8_t *, NFSPROC_T *, u_int32_t, struct nfsclowner **, struct nfsclopen **); static void nfscl_clrelease(struct nfsclclient *); static void nfscl_cleanclient(struct nfsclclient *); static void nfscl_expireclient(struct nfsclclient *, struct nfsmount *, struct ucred *, NFSPROC_T *); static int nfscl_expireopen(struct nfsclclient *, struct nfsclopen *, struct nfsmount *, struct ucred *, NFSPROC_T *); static void nfscl_recover(struct nfsclclient *, struct ucred *, NFSPROC_T *); static void nfscl_insertlock(struct nfscllockowner *, struct nfscllock *, struct nfscllock *, int); static int nfscl_updatelock(struct nfscllockowner *, struct nfscllock **, struct nfscllock **, int); static void nfscl_delegreturnall(struct nfsclclient *, NFSPROC_T *); static u_int32_t nfscl_nextcbident(void); static mount_t nfscl_getmnt(u_int32_t); static struct nfscldeleg *nfscl_finddeleg(struct nfsclclient *, u_int8_t *, int); static int nfscl_checkconflict(struct nfscllockownerhead *, struct nfscllock *, u_int8_t *, struct nfscllock **); static void nfscl_freelockowner(struct nfscllockowner *, int); static void nfscl_freealllocks(struct nfscllockownerhead *, int); static int nfscl_localconflict(struct nfsclclient *, u_int8_t *, int, struct nfscllock *, u_int8_t *, struct nfscldeleg *, struct nfscllock **); static void nfscl_newopen(struct nfsclclient *, struct nfscldeleg *, struct nfsclowner **, struct nfsclowner **, struct nfsclopen **, struct nfsclopen **, u_int8_t *, u_int8_t *, int, int *); static int nfscl_moveopen(vnode_t , struct nfsclclient *, struct nfsmount *, struct nfsclopen *, struct nfsclowner *, struct nfscldeleg *, struct ucred *, NFSPROC_T *); static void nfscl_totalrecall(struct nfsclclient *); static int nfscl_relock(vnode_t , struct nfsclclient *, struct nfsmount *, struct nfscllockowner *, struct nfscllock *, struct ucred *, NFSPROC_T *); static int nfscl_tryopen(struct nfsmount *, vnode_t , u_int8_t *, int, u_int8_t *, int, u_int32_t, struct nfsclopen *, u_int8_t *, int, struct nfscldeleg **, int, u_int32_t, struct ucred *, NFSPROC_T *); static int nfscl_trylock(struct nfsmount *, vnode_t , u_int8_t *, int, struct nfscllockowner *, int, int, u_int64_t, u_int64_t, short, struct ucred *, NFSPROC_T *); static int nfsrpc_reopen(struct nfsmount *, u_int8_t *, int, u_int32_t, struct nfsclopen *, struct nfscldeleg **, struct ucred *, NFSPROC_T *); static void nfscl_freedeleg(struct nfscldeleghead *, struct nfscldeleg *); static int nfscl_errmap(struct nfsrv_descript *); static void nfscl_cleanup_common(struct nfsclclient *, u_int8_t *); static int nfscl_recalldeleg(struct nfsclclient *, struct nfsmount *, struct nfscldeleg *, vnode_t, struct ucred *, NFSPROC_T *, int); static void nfscl_freeopenowner(struct nfsclowner *, int); static void nfscl_cleandeleg(struct nfscldeleg *); static int nfscl_trydelegreturn(struct nfscldeleg *, struct ucred *, struct nfsmount *, NFSPROC_T *); static short nfscberr_null[] = { 0, 0, }; static short nfscberr_getattr[] = { NFSERR_RESOURCE, NFSERR_BADHANDLE, NFSERR_BADXDR, NFSERR_RESOURCE, NFSERR_SERVERFAULT, 0, }; static short nfscberr_recall[] = { NFSERR_RESOURCE, NFSERR_BADHANDLE, NFSERR_BADSTATEID, NFSERR_BADXDR, NFSERR_RESOURCE, NFSERR_SERVERFAULT, 0, }; static short *nfscl_cberrmap[] = { nfscberr_null, nfscberr_null, nfscberr_null, nfscberr_getattr, nfscberr_recall }; #define NETFAMILY(clp) \ (((clp)->nfsc_flags & NFSCLFLAGS_AFINET6) ? AF_INET6 : AF_INET) /* * Called for an open operation. * If the nfhp argument is NULL, just get an openowner. */ APPLESTATIC int nfscl_open(vnode_t vp, u_int8_t *nfhp, int fhlen, u_int32_t amode, int usedeleg, struct ucred *cred, NFSPROC_T *p, struct nfsclowner **owpp, struct nfsclopen **opp, int *newonep, int *retp, int lockit) { struct nfsclclient *clp; struct nfsclowner *owp, *nowp; struct nfsclopen *op = NULL, *nop = NULL; struct nfscldeleg *dp; struct nfsclownerhead *ohp; u_int8_t own[NFSV4CL_LOCKNAMELEN]; int ret; if (newonep != NULL) *newonep = 0; if (opp != NULL) *opp = NULL; if (owpp != NULL) *owpp = NULL; /* * Might need one or both of these, so MALLOC them now, to * avoid a tsleep() in MALLOC later. */ MALLOC(nowp, struct nfsclowner *, sizeof (struct nfsclowner), M_NFSCLOWNER, M_WAITOK); if (nfhp != NULL) MALLOC(nop, struct nfsclopen *, sizeof (struct nfsclopen) + fhlen - 1, M_NFSCLOPEN, M_WAITOK); ret = nfscl_getcl(vp, cred, p, &clp); if (ret != 0) { FREE((caddr_t)nowp, M_NFSCLOWNER); if (nop != NULL) FREE((caddr_t)nop, M_NFSCLOPEN); return (ret); } /* * Get the Open iff it already exists. * If none found, add the new one or return error, depending upon * "create". */ - nfscl_filllockowner(p, own); + nfscl_filllockowner(p->td_proc, own, F_POSIX); NFSLOCKCLSTATE(); dp = NULL; /* First check the delegation list */ if (nfhp != NULL && usedeleg) { LIST_FOREACH(dp, NFSCLDELEGHASH(clp, nfhp, fhlen), nfsdl_hash) { if (dp->nfsdl_fhlen == fhlen && !NFSBCMP(nfhp, dp->nfsdl_fh, fhlen)) { if (!(amode & NFSV4OPEN_ACCESSWRITE) || (dp->nfsdl_flags & NFSCLDL_WRITE)) break; dp = NULL; break; } } } if (dp != NULL) ohp = &dp->nfsdl_owner; else ohp = &clp->nfsc_owner; /* Now, search for an openowner */ LIST_FOREACH(owp, ohp, nfsow_list) { if (!NFSBCMP(owp->nfsow_owner, own, NFSV4CL_LOCKNAMELEN)) break; } /* * Create a new open, as required. */ nfscl_newopen(clp, dp, &owp, &nowp, &op, &nop, own, nfhp, fhlen, newonep); /* * Serialize modifications to the open owner for multiple threads * within the same process using a read/write sleep lock. */ if (lockit) nfscl_lockexcl(&owp->nfsow_rwlock, NFSCLSTATEMUTEXPTR); NFSUNLOCKCLSTATE(); if (nowp != NULL) FREE((caddr_t)nowp, M_NFSCLOWNER); if (nop != NULL) FREE((caddr_t)nop, M_NFSCLOPEN); if (owpp != NULL) *owpp = owp; if (opp != NULL) *opp = op; if (retp != NULL) { if (nfhp != NULL && dp != NULL && nop == NULL) /* new local open on delegation */ *retp = NFSCLOPEN_SETCRED; else *retp = NFSCLOPEN_OK; } /* * Now, check the mode on the open and return the appropriate * value. */ if (op != NULL && (amode & ~(op->nfso_mode))) { op->nfso_mode |= amode; if (retp != NULL && dp == NULL) *retp = NFSCLOPEN_DOOPEN; } return (0); } /* * Create a new open, as required. */ static void nfscl_newopen(struct nfsclclient *clp, struct nfscldeleg *dp, struct nfsclowner **owpp, struct nfsclowner **nowpp, struct nfsclopen **opp, struct nfsclopen **nopp, u_int8_t *own, u_int8_t *fhp, int fhlen, int *newonep) { struct nfsclowner *owp = *owpp, *nowp; struct nfsclopen *op, *nop; if (nowpp != NULL) nowp = *nowpp; else nowp = NULL; if (nopp != NULL) nop = *nopp; else nop = NULL; if (owp == NULL && nowp != NULL) { NFSBCOPY(own, nowp->nfsow_owner, NFSV4CL_LOCKNAMELEN); LIST_INIT(&nowp->nfsow_open); nowp->nfsow_clp = clp; nowp->nfsow_seqid = 0; nowp->nfsow_defunct = 0; nfscl_lockinit(&nowp->nfsow_rwlock); if (dp != NULL) { newnfsstats.cllocalopenowners++; LIST_INSERT_HEAD(&dp->nfsdl_owner, nowp, nfsow_list); } else { newnfsstats.clopenowners++; LIST_INSERT_HEAD(&clp->nfsc_owner, nowp, nfsow_list); } owp = *owpp = nowp; *nowpp = NULL; if (newonep != NULL) *newonep = 1; } /* If an fhp has been specified, create an Open as well. */ if (fhp != NULL) { /* and look for the correct open, based upon FH */ LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { if (op->nfso_fhlen == fhlen && !NFSBCMP(op->nfso_fh, fhp, fhlen)) break; } if (op == NULL && nop != NULL) { nop->nfso_own = owp; nop->nfso_mode = 0; nop->nfso_opencnt = 0; nop->nfso_posixlock = 1; nop->nfso_fhlen = fhlen; NFSBCOPY(fhp, nop->nfso_fh, fhlen); LIST_INIT(&nop->nfso_lock); nop->nfso_stateid.seqid = 0; nop->nfso_stateid.other[0] = 0; nop->nfso_stateid.other[1] = 0; nop->nfso_stateid.other[2] = 0; if (dp != NULL) { TAILQ_REMOVE(&clp->nfsc_deleg, dp, nfsdl_list); TAILQ_INSERT_HEAD(&clp->nfsc_deleg, dp, nfsdl_list); dp->nfsdl_timestamp = NFSD_MONOSEC + 120; newnfsstats.cllocalopens++; } else { newnfsstats.clopens++; } LIST_INSERT_HEAD(&owp->nfsow_open, nop, nfso_list); *opp = nop; *nopp = NULL; if (newonep != NULL) *newonep = 1; } else { *opp = op; } } } /* * Called to find/add a delegation to a client. */ APPLESTATIC int nfscl_deleg(mount_t mp, struct nfsclclient *clp, u_int8_t *nfhp, int fhlen, struct ucred *cred, NFSPROC_T *p, struct nfscldeleg **dpp) { struct nfscldeleg *dp = *dpp, *tdp; /* * First, if we have received a Read delegation for a file on a * read/write file system, just return it, because they aren't * useful, imho. */ if (mp != NULL && dp != NULL && !NFSMNT_RDONLY(mp) && (dp->nfsdl_flags & NFSCLDL_READ)) { (void) nfscl_trydelegreturn(dp, cred, VFSTONFS(mp), p); FREE((caddr_t)dp, M_NFSCLDELEG); *dpp = NULL; return (0); } /* Look for the correct deleg, based upon FH */ NFSLOCKCLSTATE(); tdp = nfscl_finddeleg(clp, nfhp, fhlen); if (tdp == NULL) { if (dp == NULL) { NFSUNLOCKCLSTATE(); return (NFSERR_BADSTATEID); } *dpp = NULL; TAILQ_INSERT_HEAD(&clp->nfsc_deleg, dp, nfsdl_list); LIST_INSERT_HEAD(NFSCLDELEGHASH(clp, nfhp, fhlen), dp, nfsdl_hash); dp->nfsdl_timestamp = NFSD_MONOSEC + 120; newnfsstats.cldelegates++; nfscl_delegcnt++; } else { /* * Delegation already exists, what do we do if a new one?? */ if (dp != NULL) { printf("Deleg already exists!\n"); FREE((caddr_t)dp, M_NFSCLDELEG); *dpp = NULL; } else { *dpp = tdp; } } NFSUNLOCKCLSTATE(); return (0); } /* * Find a delegation for this file handle. Return NULL upon failure. */ static struct nfscldeleg * nfscl_finddeleg(struct nfsclclient *clp, u_int8_t *fhp, int fhlen) { struct nfscldeleg *dp; LIST_FOREACH(dp, NFSCLDELEGHASH(clp, fhp, fhlen), nfsdl_hash) { if (dp->nfsdl_fhlen == fhlen && !NFSBCMP(dp->nfsdl_fh, fhp, fhlen)) break; } return (dp); } /* * Get a stateid for an I/O operation. First, look for an open and iff * found, return either a lockowner stateid or the open stateid. * If no Open is found, just return error and the special stateid of all zeros. */ APPLESTATIC int nfscl_getstateid(vnode_t vp, u_int8_t *nfhp, int fhlen, u_int32_t mode, struct ucred *cred, NFSPROC_T *p, nfsv4stateid_t *stateidp, void **lckpp) { struct nfsclclient *clp; struct nfsclowner *owp; struct nfsclopen *op = NULL; struct nfscllockowner *lp; struct nfscldeleg *dp; struct nfsnode *np; u_int8_t own[NFSV4CL_LOCKNAMELEN]; int error, done; *lckpp = NULL; /* * Initially, just set the special stateid of all zeros. */ stateidp->seqid = 0; stateidp->other[0] = 0; stateidp->other[1] = 0; stateidp->other[2] = 0; if (vnode_vtype(vp) != VREG) return (EISDIR); np = VTONFS(vp); NFSLOCKCLSTATE(); clp = nfscl_findcl(VFSTONFS(vnode_mount(vp))); if (clp == NULL) { NFSUNLOCKCLSTATE(); return (EACCES); } /* * Wait for recovery to complete. */ while ((clp->nfsc_flags & NFSCLFLAGS_RECVRINPROG)) (void) nfsmsleep(&clp->nfsc_flags, NFSCLSTATEMUTEXPTR, PZERO, "nfsrecvr", NULL); /* * First, look for a delegation. */ LIST_FOREACH(dp, NFSCLDELEGHASH(clp, nfhp, fhlen), nfsdl_hash) { if (dp->nfsdl_fhlen == fhlen && !NFSBCMP(nfhp, dp->nfsdl_fh, fhlen)) { if (!(mode & NFSV4OPEN_ACCESSWRITE) || (dp->nfsdl_flags & NFSCLDL_WRITE)) { stateidp->seqid = dp->nfsdl_stateid.seqid; stateidp->other[0] = dp->nfsdl_stateid.other[0]; stateidp->other[1] = dp->nfsdl_stateid.other[1]; stateidp->other[2] = dp->nfsdl_stateid.other[2]; if (!(np->n_flag & NDELEGRECALL)) { TAILQ_REMOVE(&clp->nfsc_deleg, dp, nfsdl_list); TAILQ_INSERT_HEAD(&clp->nfsc_deleg, dp, nfsdl_list); dp->nfsdl_timestamp = NFSD_MONOSEC + 120; dp->nfsdl_rwlock.nfslock_usecnt++; *lckpp = (void *)&dp->nfsdl_rwlock; } NFSUNLOCKCLSTATE(); return (0); } break; } } if (p != NULL) { /* * If p != NULL, we want to search the parentage tree * for a matching OpenOwner and use that. */ - nfscl_filllockowner(p, own); + nfscl_filllockowner(p->td_proc, own, F_POSIX); error = nfscl_getopen(&clp->nfsc_owner, nfhp, fhlen, NULL, p, mode, NULL, &op); if (error == 0) { /* now look for a lockowner */ LIST_FOREACH(lp, &op->nfso_lock, nfsl_list) { if (!NFSBCMP(lp->nfsl_owner, own, NFSV4CL_LOCKNAMELEN)) { stateidp->seqid = lp->nfsl_stateid.seqid; stateidp->other[0] = lp->nfsl_stateid.other[0]; stateidp->other[1] = lp->nfsl_stateid.other[1]; stateidp->other[2] = lp->nfsl_stateid.other[2]; NFSUNLOCKCLSTATE(); return (0); } } } } if (op == NULL) { /* If not found, just look for any OpenOwner that will work. */ done = 0; owp = LIST_FIRST(&clp->nfsc_owner); while (!done && owp != NULL) { LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { if (op->nfso_fhlen == fhlen && !NFSBCMP(op->nfso_fh, nfhp, fhlen) && (mode & op->nfso_mode) == mode) { done = 1; break; } } if (!done) owp = LIST_NEXT(owp, nfsow_list); } if (!done) { NFSUNLOCKCLSTATE(); return (ENOENT); } /* for read aheads or write behinds, use the open cred */ newnfs_copycred(&op->nfso_cred, cred); } /* * No lock stateid, so return the open stateid. */ stateidp->seqid = op->nfso_stateid.seqid; stateidp->other[0] = op->nfso_stateid.other[0]; stateidp->other[1] = op->nfso_stateid.other[1]; stateidp->other[2] = op->nfso_stateid.other[2]; NFSUNLOCKCLSTATE(); return (0); } /* * Get an existing open. Search up the parentage tree for a match and * return with the first one found. */ static int nfscl_getopen(struct nfsclownerhead *ohp, u_int8_t *nfhp, int fhlen, u_int8_t *rown, NFSPROC_T *p, u_int32_t mode, struct nfsclowner **owpp, struct nfsclopen **opp) { struct nfsclowner *owp = NULL; struct nfsclopen *op; NFSPROC_T *nproc; u_int8_t own[NFSV4CL_LOCKNAMELEN], *ownp; nproc = p; op = NULL; while (op == NULL && (nproc != NULL || rown != NULL)) { if (nproc != NULL) { - nfscl_filllockowner(nproc, own); + nfscl_filllockowner(nproc->td_proc, own, F_POSIX); ownp = own; } else { ownp = rown; } /* Search the client list */ LIST_FOREACH(owp, ohp, nfsow_list) { if (!NFSBCMP(owp->nfsow_owner, ownp, NFSV4CL_LOCKNAMELEN)) break; } if (owp != NULL) { /* and look for the correct open */ LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { if (op->nfso_fhlen == fhlen && !NFSBCMP(op->nfso_fh, nfhp, fhlen) && (op->nfso_mode & mode) == mode) { break; } } } if (rown != NULL) break; if (op == NULL) nproc = nfscl_getparent(nproc); } if (op == NULL) { return (EBADF); } if (owpp) *owpp = owp; *opp = op; return (0); } /* * Release use of an open owner. Called when open operations are done * with the open owner. */ APPLESTATIC void nfscl_ownerrelease(struct nfsclowner *owp, __unused int error, __unused int candelete, int unlocked) { if (owp == NULL) return; NFSLOCKCLSTATE(); if (!unlocked) nfscl_lockunlock(&owp->nfsow_rwlock); nfscl_clrelease(owp->nfsow_clp); NFSUNLOCKCLSTATE(); } /* * Release use of an open structure under an open owner. */ APPLESTATIC void nfscl_openrelease(struct nfsclopen *op, int error, int candelete) { struct nfsclclient *clp; struct nfsclowner *owp; if (op == NULL) return; NFSLOCKCLSTATE(); owp = op->nfso_own; nfscl_lockunlock(&owp->nfsow_rwlock); clp = owp->nfsow_clp; if (error && candelete && op->nfso_opencnt == 0) nfscl_freeopen(op, 0); nfscl_clrelease(clp); NFSUNLOCKCLSTATE(); } /* * Called to get a clientid structure. It will optionally lock the * client data structures to do the SetClientId/SetClientId_confirm, * but will release that lock and return the clientid with a refernce * count on it. * If the "cred" argument is NULL, a new clientid should not be created. * If the "p" argument is NULL, a SetClientID/SetClientIDConfirm cannot * be done. * It always clpp with a reference count on it, unless returning an error. */ APPLESTATIC int nfscl_getcl(vnode_t vp, struct ucred *cred, NFSPROC_T *p, struct nfsclclient **clpp) { struct nfsclclient *clp; struct nfsclclient *newclp = NULL; struct nfscllockowner *lp, *nlp; struct mount *mp; struct nfsmount *nmp; char uuid[HOSTUUIDLEN]; int igotlock = 0, error, trystalecnt, clidinusedelay, i; u_int16_t idlen = 0; mp = vnode_mount(vp); nmp = VFSTONFS(mp); if (cred != NULL) { getcredhostuuid(cred, uuid, sizeof uuid); idlen = strlen(uuid); if (idlen > 0) idlen += sizeof (u_int64_t); else idlen += sizeof (u_int64_t) + 16; /* 16 random bytes */ MALLOC(newclp, struct nfsclclient *, sizeof (struct nfsclclient) + idlen - 1, M_NFSCLCLIENT, M_WAITOK); } NFSLOCKCLSTATE(); /* * If a forced dismount is already in progress, don't * allocate a new clientid and get out now. For the case where * clp != NULL, this is a harmless optimization. */ if ((mp->mnt_kern_flag & MNTK_UNMOUNTF) != 0) { NFSUNLOCKCLSTATE(); if (newclp != NULL) free(newclp, M_NFSCLCLIENT); return (EBADF); } clp = nmp->nm_clp; if (clp == NULL) { if (newclp == NULL) { NFSUNLOCKCLSTATE(); return (EACCES); } clp = newclp; NFSBZERO((caddr_t)clp, sizeof(struct nfsclclient) + idlen - 1); clp->nfsc_idlen = idlen; LIST_INIT(&clp->nfsc_owner); TAILQ_INIT(&clp->nfsc_deleg); for (i = 0; i < NFSCLDELEGHASHSIZE; i++) LIST_INIT(&clp->nfsc_deleghash[i]); LIST_INIT(&clp->nfsc_defunctlockowner); clp->nfsc_flags = NFSCLFLAGS_INITED; clp->nfsc_clientidrev = 1; clp->nfsc_cbident = nfscl_nextcbident(); nfscl_fillclid(nmp->nm_clval, uuid, clp->nfsc_id, clp->nfsc_idlen); LIST_INSERT_HEAD(&nfsclhead, clp, nfsc_list); nmp->nm_clp = clp; clp->nfsc_nmp = nmp; NFSUNLOCKCLSTATE(); nfscl_start_renewthread(clp); } else { NFSUNLOCKCLSTATE(); if (newclp != NULL) FREE((caddr_t)newclp, M_NFSCLCLIENT); } NFSLOCKCLSTATE(); while ((clp->nfsc_flags & NFSCLFLAGS_HASCLIENTID) == 0 && !igotlock) igotlock = nfsv4_lock(&clp->nfsc_lock, 1, NULL, NFSCLSTATEMUTEXPTR, mp); if (!igotlock) nfsv4_getref(&clp->nfsc_lock, NULL, NFSCLSTATEMUTEXPTR, mp); if (igotlock == 0 && (mp->mnt_kern_flag & MNTK_UNMOUNTF) != 0) { /* * Both nfsv4_lock() and nfsv4_getref() know to check * for MNTK_UNMOUNTF and return without sleeping to * wait for the exclusive lock to be released, since it * might be held by nfscl_umount() and we need to get out * now for that case and not wait until nfscl_umount() * releases it. */ NFSUNLOCKCLSTATE(); return (EBADF); } NFSUNLOCKCLSTATE(); /* * If it needs a clientid, do the setclientid now. */ if ((clp->nfsc_flags & NFSCLFLAGS_HASCLIENTID) == 0) { if (!igotlock) panic("nfscl_clget"); if (p == NULL || cred == NULL) { NFSLOCKCLSTATE(); nfsv4_unlock(&clp->nfsc_lock, 0); NFSUNLOCKCLSTATE(); return (EACCES); } /* get rid of defunct lockowners */ LIST_FOREACH_SAFE(lp, &clp->nfsc_defunctlockowner, nfsl_list, nlp) { nfscl_freelockowner(lp, 0); } /* * If RFC3530 Sec. 14.2.33 is taken literally, * NFSERR_CLIDINUSE will be returned persistently for the * case where a new mount of the same file system is using * a different principal. In practice, NFSERR_CLIDINUSE is * only returned when there is outstanding unexpired state * on the clientid. As such, try for twice the lease * interval, if we know what that is. Otherwise, make a * wild ass guess. * The case of returning NFSERR_STALECLIENTID is far less * likely, but might occur if there is a significant delay * between doing the SetClientID and SetClientIDConfirm Ops, * such that the server throws away the clientid before * receiving the SetClientIDConfirm. */ if (clp->nfsc_renew > 0) clidinusedelay = NFSCL_LEASE(clp->nfsc_renew) * 2; else clidinusedelay = 120; trystalecnt = 3; do { error = nfsrpc_setclient(VFSTONFS(vnode_mount(vp)), clp, cred, p); if (error == NFSERR_STALECLIENTID || error == NFSERR_STALEDONTRECOVER || error == NFSERR_CLIDINUSE) { (void) nfs_catnap(PZERO, error, "nfs_setcl"); } } while (((error == NFSERR_STALECLIENTID || error == NFSERR_STALEDONTRECOVER) && --trystalecnt > 0) || (error == NFSERR_CLIDINUSE && --clidinusedelay > 0)); if (error) { NFSLOCKCLSTATE(); nfsv4_unlock(&clp->nfsc_lock, 0); NFSUNLOCKCLSTATE(); return (error); } clp->nfsc_flags |= NFSCLFLAGS_HASCLIENTID; } if (igotlock) { NFSLOCKCLSTATE(); nfsv4_unlock(&clp->nfsc_lock, 1); NFSUNLOCKCLSTATE(); } *clpp = clp; return (0); } /* * Get a reference to a clientid and return it, if valid. */ APPLESTATIC struct nfsclclient * nfscl_findcl(struct nfsmount *nmp) { struct nfsclclient *clp; clp = nmp->nm_clp; if (clp == NULL || !(clp->nfsc_flags & NFSCLFLAGS_HASCLIENTID)) return (NULL); return (clp); } /* * Release the clientid structure. It may be locked or reference counted. */ static void nfscl_clrelease(struct nfsclclient *clp) { if (clp->nfsc_lock.nfslock_lock & NFSV4LOCK_LOCK) nfsv4_unlock(&clp->nfsc_lock, 0); else nfsv4_relref(&clp->nfsc_lock); } /* * External call for nfscl_clrelease. */ APPLESTATIC void nfscl_clientrelease(struct nfsclclient *clp) { NFSLOCKCLSTATE(); if (clp->nfsc_lock.nfslock_lock & NFSV4LOCK_LOCK) nfsv4_unlock(&clp->nfsc_lock, 0); else nfsv4_relref(&clp->nfsc_lock); NFSUNLOCKCLSTATE(); } /* * Called when wanting to lock a byte region. */ APPLESTATIC int nfscl_getbytelock(vnode_t vp, u_int64_t off, u_int64_t len, short type, struct ucred *cred, NFSPROC_T *p, struct nfsclclient *rclp, - int recovery, u_int8_t *rownp, u_int8_t *ropenownp, + int recovery, void *id, int flags, u_int8_t *rownp, u_int8_t *ropenownp, struct nfscllockowner **lpp, int *newonep, int *donelocallyp) { struct nfscllockowner *lp; struct nfsclopen *op; struct nfsclclient *clp; struct nfscllockowner *nlp; struct nfscllock *nlop, *otherlop; struct nfscldeleg *dp = NULL, *ldp = NULL; struct nfscllockownerhead *lhp = NULL; struct nfsnode *np; u_int8_t own[NFSV4CL_LOCKNAMELEN], *ownp; int error = 0, ret, donelocally = 0; u_int32_t mode; if (type == F_WRLCK) mode = NFSV4OPEN_ACCESSWRITE; else mode = NFSV4OPEN_ACCESSREAD; np = VTONFS(vp); *lpp = NULL; *newonep = 0; *donelocallyp = 0; /* * Might need these, so MALLOC them now, to * avoid a tsleep() in MALLOC later. */ MALLOC(nlp, struct nfscllockowner *, sizeof (struct nfscllockowner), M_NFSCLLOCKOWNER, M_WAITOK); MALLOC(otherlop, struct nfscllock *, sizeof (struct nfscllock), M_NFSCLLOCK, M_WAITOK); MALLOC(nlop, struct nfscllock *, sizeof (struct nfscllock), M_NFSCLLOCK, M_WAITOK); nlop->nfslo_type = type; nlop->nfslo_first = off; if (len == NFS64BITSSET) { nlop->nfslo_end = NFS64BITSSET; } else { nlop->nfslo_end = off + len; if (nlop->nfslo_end <= nlop->nfslo_first) error = NFSERR_INVAL; } if (!error) { if (recovery) clp = rclp; else error = nfscl_getcl(vp, cred, p, &clp); } if (error) { FREE((caddr_t)nlp, M_NFSCLLOCKOWNER); FREE((caddr_t)otherlop, M_NFSCLLOCK); FREE((caddr_t)nlop, M_NFSCLLOCK); return (error); } op = NULL; if (recovery) { ownp = rownp; } else { - nfscl_filllockowner(p, own); + nfscl_filllockowner(id, own, flags); ownp = own; } if (!recovery) { NFSLOCKCLSTATE(); /* * First, search for a delegation. If one exists for this file, * the lock can be done locally against it, so long as there * isn't a local lock conflict. */ ldp = dp = nfscl_finddeleg(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len); /* Just sanity check for correct type of delegation */ if (dp != NULL && ((dp->nfsdl_flags & (NFSCLDL_RECALL | NFSCLDL_DELEGRET)) != 0 || (type == F_WRLCK && (dp->nfsdl_flags & NFSCLDL_WRITE) == 0))) dp = NULL; } if (dp != NULL) { /* Now, find the associated open to get the correct openowner */ ret = nfscl_getopen(&dp->nfsdl_owner, np->n_fhp->nfh_fh, np->n_fhp->nfh_len, NULL, p, mode, NULL, &op); if (ret) ret = nfscl_getopen(&clp->nfsc_owner, np->n_fhp->nfh_fh, np->n_fhp->nfh_len, NULL, p, mode, NULL, &op); if (!ret) { lhp = &dp->nfsdl_lock; TAILQ_REMOVE(&clp->nfsc_deleg, dp, nfsdl_list); TAILQ_INSERT_HEAD(&clp->nfsc_deleg, dp, nfsdl_list); dp->nfsdl_timestamp = NFSD_MONOSEC + 120; donelocally = 1; } else { dp = NULL; } } if (!donelocally) { /* * Get the related Open. */ if (recovery) error = nfscl_getopen(&clp->nfsc_owner, np->n_fhp->nfh_fh, np->n_fhp->nfh_len, ropenownp, NULL, mode, NULL, &op); else error = nfscl_getopen(&clp->nfsc_owner, np->n_fhp->nfh_fh, np->n_fhp->nfh_len, NULL, p, mode, NULL, &op); if (!error) lhp = &op->nfso_lock; } if (!error && !recovery) error = nfscl_localconflict(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len, nlop, ownp, ldp, NULL); if (error) { if (!recovery) { nfscl_clrelease(clp); NFSUNLOCKCLSTATE(); } FREE((caddr_t)nlp, M_NFSCLLOCKOWNER); FREE((caddr_t)otherlop, M_NFSCLLOCK); FREE((caddr_t)nlop, M_NFSCLLOCK); return (error); } /* * Ok, see if a lockowner exists and create one, as required. */ LIST_FOREACH(lp, lhp, nfsl_list) { if (!NFSBCMP(lp->nfsl_owner, ownp, NFSV4CL_LOCKNAMELEN)) break; } if (lp == NULL) { NFSBCOPY(ownp, nlp->nfsl_owner, NFSV4CL_LOCKNAMELEN); if (recovery) NFSBCOPY(ropenownp, nlp->nfsl_openowner, NFSV4CL_LOCKNAMELEN); else NFSBCOPY(op->nfso_own->nfsow_owner, nlp->nfsl_openowner, NFSV4CL_LOCKNAMELEN); nlp->nfsl_seqid = 0; nlp->nfsl_defunct = 0; nlp->nfsl_inprog = NULL; nfscl_lockinit(&nlp->nfsl_rwlock); LIST_INIT(&nlp->nfsl_lock); if (donelocally) { nlp->nfsl_open = NULL; newnfsstats.cllocallockowners++; } else { nlp->nfsl_open = op; newnfsstats.cllockowners++; } LIST_INSERT_HEAD(lhp, nlp, nfsl_list); lp = nlp; nlp = NULL; *newonep = 1; } /* * Now, update the byte ranges for locks. */ ret = nfscl_updatelock(lp, &nlop, &otherlop, donelocally); if (!ret) donelocally = 1; if (donelocally) { *donelocallyp = 1; if (!recovery) nfscl_clrelease(clp); } else { /* * Serial modifications on the lock owner for multiple threads * for the same process using a read/write lock. */ if (!recovery) nfscl_lockexcl(&lp->nfsl_rwlock, NFSCLSTATEMUTEXPTR); } if (!recovery) NFSUNLOCKCLSTATE(); if (nlp) FREE((caddr_t)nlp, M_NFSCLLOCKOWNER); if (nlop) FREE((caddr_t)nlop, M_NFSCLLOCK); if (otherlop) FREE((caddr_t)otherlop, M_NFSCLLOCK); *lpp = lp; return (0); } /* * Called to unlock a byte range, for LockU. */ APPLESTATIC int nfscl_relbytelock(vnode_t vp, u_int64_t off, u_int64_t len, __unused struct ucred *cred, NFSPROC_T *p, int callcnt, - struct nfsclclient *clp, struct nfscllockowner **lpp, int *dorpcp) + struct nfsclclient *clp, void *id, int flags, + struct nfscllockowner **lpp, int *dorpcp) { struct nfscllockowner *lp; struct nfsclowner *owp; struct nfsclopen *op; struct nfscllock *nlop, *other_lop = NULL; struct nfscldeleg *dp; struct nfsnode *np; u_int8_t own[NFSV4CL_LOCKNAMELEN]; int ret = 0, fnd; np = VTONFS(vp); *lpp = NULL; *dorpcp = 0; /* * Might need these, so MALLOC them now, to * avoid a tsleep() in MALLOC later. */ MALLOC(nlop, struct nfscllock *, sizeof (struct nfscllock), M_NFSCLLOCK, M_WAITOK); nlop->nfslo_type = F_UNLCK; nlop->nfslo_first = off; if (len == NFS64BITSSET) { nlop->nfslo_end = NFS64BITSSET; } else { nlop->nfslo_end = off + len; if (nlop->nfslo_end <= nlop->nfslo_first) { FREE((caddr_t)nlop, M_NFSCLLOCK); return (NFSERR_INVAL); } } if (callcnt == 0) { MALLOC(other_lop, struct nfscllock *, sizeof (struct nfscllock), M_NFSCLLOCK, M_WAITOK); *other_lop = *nlop; } - nfscl_filllockowner(p, own); + nfscl_filllockowner(id, own, flags); dp = NULL; NFSLOCKCLSTATE(); if (callcnt == 0) dp = nfscl_finddeleg(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len); /* * First, unlock any local regions on a delegation. */ if (dp != NULL) { /* Look for this lockowner. */ LIST_FOREACH(lp, &dp->nfsdl_lock, nfsl_list) { if (!NFSBCMP(lp->nfsl_owner, own, NFSV4CL_LOCKNAMELEN)) break; } if (lp != NULL) /* Use other_lop, so nlop is still available */ (void)nfscl_updatelock(lp, &other_lop, NULL, 1); } /* * Now, find a matching open/lockowner that hasn't already been done, * as marked by nfsl_inprog. */ lp = NULL; fnd = 0; LIST_FOREACH(owp, &clp->nfsc_owner, nfsow_list) { LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { if (op->nfso_fhlen == np->n_fhp->nfh_len && !NFSBCMP(op->nfso_fh, np->n_fhp->nfh_fh, op->nfso_fhlen)) { LIST_FOREACH(lp, &op->nfso_lock, nfsl_list) { if (lp->nfsl_inprog == NULL && !NFSBCMP(lp->nfsl_owner, own, NFSV4CL_LOCKNAMELEN)) { fnd = 1; break; } } if (fnd) break; } } if (fnd) break; } if (lp != NULL) { ret = nfscl_updatelock(lp, &nlop, NULL, 0); if (ret) *dorpcp = 1; /* * Serial modifications on the lock owner for multiple * threads for the same process using a read/write lock. */ lp->nfsl_inprog = p; nfscl_lockexcl(&lp->nfsl_rwlock, NFSCLSTATEMUTEXPTR); *lpp = lp; } NFSUNLOCKCLSTATE(); if (nlop) FREE((caddr_t)nlop, M_NFSCLLOCK); if (other_lop) FREE((caddr_t)other_lop, M_NFSCLLOCK); return (0); } /* * Release all lockowners marked in progess for this process and file. */ APPLESTATIC void -nfscl_releasealllocks(struct nfsclclient *clp, vnode_t vp, NFSPROC_T *p) +nfscl_releasealllocks(struct nfsclclient *clp, vnode_t vp, NFSPROC_T *p, + void *id, int flags) { struct nfsclowner *owp; struct nfsclopen *op; struct nfscllockowner *lp; struct nfsnode *np; u_int8_t own[NFSV4CL_LOCKNAMELEN]; np = VTONFS(vp); - nfscl_filllockowner(p, own); + nfscl_filllockowner(id, own, flags); NFSLOCKCLSTATE(); LIST_FOREACH(owp, &clp->nfsc_owner, nfsow_list) { LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { if (op->nfso_fhlen == np->n_fhp->nfh_len && !NFSBCMP(op->nfso_fh, np->n_fhp->nfh_fh, op->nfso_fhlen)) { LIST_FOREACH(lp, &op->nfso_lock, nfsl_list) { if (lp->nfsl_inprog == p && !NFSBCMP(lp->nfsl_owner, own, NFSV4CL_LOCKNAMELEN)) { lp->nfsl_inprog = NULL; nfscl_lockunlock(&lp->nfsl_rwlock); } } } } } nfscl_clrelease(clp); NFSUNLOCKCLSTATE(); } /* * Called to find out if any bytes within the byte range specified are * write locked by the calling process. Used to determine if flushing * is required before a LockU. * If in doubt, return 1, so the flush will occur. */ APPLESTATIC int nfscl_checkwritelocked(vnode_t vp, struct flock *fl, - struct ucred *cred, NFSPROC_T *p) + struct ucred *cred, NFSPROC_T *p, void *id, int flags) { struct nfsclowner *owp; struct nfscllockowner *lp; struct nfsclopen *op; struct nfsclclient *clp; struct nfscllock *lop; struct nfscldeleg *dp; struct nfsnode *np; u_int64_t off, end; u_int8_t own[NFSV4CL_LOCKNAMELEN]; int error = 0; np = VTONFS(vp); switch (fl->l_whence) { case SEEK_SET: case SEEK_CUR: /* * Caller is responsible for adding any necessary offset * when SEEK_CUR is used. */ off = fl->l_start; break; case SEEK_END: off = np->n_size + fl->l_start; break; default: return (1); }; if (fl->l_len != 0) { end = off + fl->l_len; if (end < off) return (1); } else { end = NFS64BITSSET; } error = nfscl_getcl(vp, cred, p, &clp); if (error) return (1); - nfscl_filllockowner(p, own); + nfscl_filllockowner(id, own, flags); NFSLOCKCLSTATE(); /* * First check the delegation locks. */ dp = nfscl_finddeleg(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len); if (dp != NULL) { LIST_FOREACH(lp, &dp->nfsdl_lock, nfsl_list) { if (!NFSBCMP(lp->nfsl_owner, own, NFSV4CL_LOCKNAMELEN)) break; } if (lp != NULL) { LIST_FOREACH(lop, &lp->nfsl_lock, nfslo_list) { if (lop->nfslo_first >= end) break; if (lop->nfslo_end <= off) continue; if (lop->nfslo_type == F_WRLCK) { nfscl_clrelease(clp); NFSUNLOCKCLSTATE(); return (1); } } } } /* * Now, check state against the server. */ LIST_FOREACH(owp, &clp->nfsc_owner, nfsow_list) { LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { if (op->nfso_fhlen == np->n_fhp->nfh_len && !NFSBCMP(op->nfso_fh, np->n_fhp->nfh_fh, op->nfso_fhlen)) { LIST_FOREACH(lp, &op->nfso_lock, nfsl_list) { if (!NFSBCMP(lp->nfsl_owner, own, NFSV4CL_LOCKNAMELEN)) break; } if (lp != NULL) { LIST_FOREACH(lop, &lp->nfsl_lock, nfslo_list) { if (lop->nfslo_first >= end) break; if (lop->nfslo_end <= off) continue; if (lop->nfslo_type == F_WRLCK) { nfscl_clrelease(clp); NFSUNLOCKCLSTATE(); return (1); } } } } } } nfscl_clrelease(clp); NFSUNLOCKCLSTATE(); return (0); } /* * Release a byte range lock owner structure. */ APPLESTATIC void nfscl_lockrelease(struct nfscllockowner *lp, int error, int candelete) { struct nfsclclient *clp; if (lp == NULL) return; NFSLOCKCLSTATE(); clp = lp->nfsl_open->nfso_own->nfsow_clp; if (error != 0 && candelete && (lp->nfsl_rwlock.nfslock_lock & NFSV4LOCK_WANTED) == 0) nfscl_freelockowner(lp, 0); else nfscl_lockunlock(&lp->nfsl_rwlock); nfscl_clrelease(clp); NFSUNLOCKCLSTATE(); } /* * Free up an open structure and any associated byte range lock structures. */ APPLESTATIC void nfscl_freeopen(struct nfsclopen *op, int local) { LIST_REMOVE(op, nfso_list); nfscl_freealllocks(&op->nfso_lock, local); FREE((caddr_t)op, M_NFSCLOPEN); if (local) newnfsstats.cllocalopens--; else newnfsstats.clopens--; } /* * Free up all lock owners and associated locks. */ static void nfscl_freealllocks(struct nfscllockownerhead *lhp, int local) { struct nfscllockowner *lp, *nlp; LIST_FOREACH_SAFE(lp, lhp, nfsl_list, nlp) { if ((lp->nfsl_rwlock.nfslock_lock & NFSV4LOCK_WANTED)) panic("nfscllckw"); nfscl_freelockowner(lp, local); } } /* * Called for an Open when NFSERR_EXPIRED is received from the server. * If there are no byte range locks nor a Share Deny lost, try to do a * fresh Open. Otherwise, free the open. */ static int nfscl_expireopen(struct nfsclclient *clp, struct nfsclopen *op, struct nfsmount *nmp, struct ucred *cred, NFSPROC_T *p) { struct nfscllockowner *lp; struct nfscldeleg *dp; int mustdelete = 0, error; /* * Look for any byte range lock(s). */ LIST_FOREACH(lp, &op->nfso_lock, nfsl_list) { if (!LIST_EMPTY(&lp->nfsl_lock)) { mustdelete = 1; break; } } /* * If no byte range lock(s) nor a Share deny, try to re-open. */ if (!mustdelete && (op->nfso_mode & NFSLCK_DENYBITS) == 0) { newnfs_copycred(&op->nfso_cred, cred); dp = NULL; error = nfsrpc_reopen(nmp, op->nfso_fh, op->nfso_fhlen, op->nfso_mode, op, &dp, cred, p); if (error) { mustdelete = 1; if (dp != NULL) { FREE((caddr_t)dp, M_NFSCLDELEG); dp = NULL; } } if (dp != NULL) nfscl_deleg(nmp->nm_mountp, clp, op->nfso_fh, op->nfso_fhlen, cred, p, &dp); } /* * If a byte range lock or Share deny or couldn't re-open, free it. */ if (mustdelete) nfscl_freeopen(op, 0); return (mustdelete); } /* * Free up an open owner structure. */ static void nfscl_freeopenowner(struct nfsclowner *owp, int local) { LIST_REMOVE(owp, nfsow_list); FREE((caddr_t)owp, M_NFSCLOWNER); if (local) newnfsstats.cllocalopenowners--; else newnfsstats.clopenowners--; } /* * Free up a byte range lock owner structure. */ static void nfscl_freelockowner(struct nfscllockowner *lp, int local) { struct nfscllock *lop, *nlop; LIST_REMOVE(lp, nfsl_list); LIST_FOREACH_SAFE(lop, &lp->nfsl_lock, nfslo_list, nlop) { nfscl_freelock(lop, local); } FREE((caddr_t)lp, M_NFSCLLOCKOWNER); if (local) newnfsstats.cllocallockowners--; else newnfsstats.cllockowners--; } /* * Free up a byte range lock structure. */ APPLESTATIC void nfscl_freelock(struct nfscllock *lop, int local) { LIST_REMOVE(lop, nfslo_list); FREE((caddr_t)lop, M_NFSCLLOCK); if (local) newnfsstats.cllocallocks--; else newnfsstats.cllocks--; } /* * Clean out the state related to a delegation. */ static void nfscl_cleandeleg(struct nfscldeleg *dp) { struct nfsclowner *owp, *nowp; struct nfsclopen *op; LIST_FOREACH_SAFE(owp, &dp->nfsdl_owner, nfsow_list, nowp) { op = LIST_FIRST(&owp->nfsow_open); if (op != NULL) { if (LIST_NEXT(op, nfso_list) != NULL) panic("nfscleandel"); nfscl_freeopen(op, 1); } nfscl_freeopenowner(owp, 1); } nfscl_freealllocks(&dp->nfsdl_lock, 1); } /* * Free a delegation. */ static void nfscl_freedeleg(struct nfscldeleghead *hdp, struct nfscldeleg *dp) { TAILQ_REMOVE(hdp, dp, nfsdl_list); LIST_REMOVE(dp, nfsdl_hash); FREE((caddr_t)dp, M_NFSCLDELEG); newnfsstats.cldelegates--; nfscl_delegcnt--; } /* * Free up all state related to this client structure. */ static void nfscl_cleanclient(struct nfsclclient *clp) { struct nfsclowner *owp, *nowp; struct nfsclopen *op, *nop; struct nfscllockowner *lp, *nlp; /* get rid of defunct lockowners */ LIST_FOREACH_SAFE(lp, &clp->nfsc_defunctlockowner, nfsl_list, nlp) { nfscl_freelockowner(lp, 0); } /* Now, all the OpenOwners, etc. */ LIST_FOREACH_SAFE(owp, &clp->nfsc_owner, nfsow_list, nowp) { LIST_FOREACH_SAFE(op, &owp->nfsow_open, nfso_list, nop) { nfscl_freeopen(op, 0); } nfscl_freeopenowner(owp, 0); } } /* * Called when an NFSERR_EXPIRED is received from the server. */ static void nfscl_expireclient(struct nfsclclient *clp, struct nfsmount *nmp, struct ucred *cred, NFSPROC_T *p) { struct nfsclowner *owp, *nowp, *towp; struct nfsclopen *op, *nop, *top; struct nfscldeleg *dp, *ndp; int ret, printed = 0; /* * First, merge locally issued Opens into the list for the server. */ dp = TAILQ_FIRST(&clp->nfsc_deleg); while (dp != NULL) { ndp = TAILQ_NEXT(dp, nfsdl_list); owp = LIST_FIRST(&dp->nfsdl_owner); while (owp != NULL) { nowp = LIST_NEXT(owp, nfsow_list); op = LIST_FIRST(&owp->nfsow_open); if (op != NULL) { if (LIST_NEXT(op, nfso_list) != NULL) panic("nfsclexp"); LIST_FOREACH(towp, &clp->nfsc_owner, nfsow_list) { if (!NFSBCMP(towp->nfsow_owner, owp->nfsow_owner, NFSV4CL_LOCKNAMELEN)) break; } if (towp != NULL) { /* Merge opens in */ LIST_FOREACH(top, &towp->nfsow_open, nfso_list) { if (top->nfso_fhlen == op->nfso_fhlen && !NFSBCMP(top->nfso_fh, op->nfso_fh, op->nfso_fhlen)) { top->nfso_mode |= op->nfso_mode; top->nfso_opencnt += op->nfso_opencnt; break; } } if (top == NULL) { /* Just add the open to the owner list */ LIST_REMOVE(op, nfso_list); op->nfso_own = towp; LIST_INSERT_HEAD(&towp->nfsow_open, op, nfso_list); newnfsstats.cllocalopens--; newnfsstats.clopens++; } } else { /* Just add the openowner to the client list */ LIST_REMOVE(owp, nfsow_list); owp->nfsow_clp = clp; LIST_INSERT_HEAD(&clp->nfsc_owner, owp, nfsow_list); newnfsstats.cllocalopenowners--; newnfsstats.clopenowners++; newnfsstats.cllocalopens--; newnfsstats.clopens++; } } owp = nowp; } if (!printed && !LIST_EMPTY(&dp->nfsdl_lock)) { printed = 1; printf("nfsv4 expired locks lost\n"); } nfscl_cleandeleg(dp); nfscl_freedeleg(&clp->nfsc_deleg, dp); dp = ndp; } if (!TAILQ_EMPTY(&clp->nfsc_deleg)) panic("nfsclexp"); /* * Now, try and reopen against the server. */ LIST_FOREACH_SAFE(owp, &clp->nfsc_owner, nfsow_list, nowp) { owp->nfsow_seqid = 0; LIST_FOREACH_SAFE(op, &owp->nfsow_open, nfso_list, nop) { ret = nfscl_expireopen(clp, op, nmp, cred, p); if (ret && !printed) { printed = 1; printf("nfsv4 expired locks lost\n"); } } if (LIST_EMPTY(&owp->nfsow_open)) nfscl_freeopenowner(owp, 0); } } #ifndef __FreeBSD__ /* * Called from exit() upon process termination. */ APPLESTATIC void nfscl_cleanup(NFSPROC_T *p) { struct nfsclclient *clp; u_int8_t own[NFSV4CL_LOCKNAMELEN]; if (!nfscl_inited) return; - nfscl_filllockowner(p, own); + nfscl_filllockowner(p->td_proc, own, F_POSIX); NFSLOCKCLSTATE(); /* * Loop through all the clientids, looking for the OpenOwners. */ LIST_FOREACH(clp, &nfsclhead, nfsc_list) nfscl_cleanup_common(clp, own); NFSUNLOCKCLSTATE(); } #endif /* !__FreeBSD__ */ /* * Common code used by nfscl_cleanup() and nfscl_cleanupkext(). * Must be called with CLSTATE lock held. */ static void nfscl_cleanup_common(struct nfsclclient *clp, u_int8_t *own) { struct nfsclowner *owp, *nowp; struct nfsclopen *op; struct nfscllockowner *lp, *nlp; struct nfscldeleg *dp; /* First, get rid of local locks on delegations. */ TAILQ_FOREACH(dp, &clp->nfsc_deleg, nfsdl_list) { LIST_FOREACH_SAFE(lp, &dp->nfsdl_lock, nfsl_list, nlp) { if (!NFSBCMP(lp->nfsl_owner, own, NFSV4CL_LOCKNAMELEN)) { if ((lp->nfsl_rwlock.nfslock_lock & NFSV4LOCK_WANTED)) panic("nfscllckw"); nfscl_freelockowner(lp, 1); } } } owp = LIST_FIRST(&clp->nfsc_owner); while (owp != NULL) { nowp = LIST_NEXT(owp, nfsow_list); if (!NFSBCMP(owp->nfsow_owner, own, NFSV4CL_LOCKNAMELEN)) { /* * If there are children that haven't closed the * file descriptors yet, the opens will still be * here. For that case, let the renew thread clear * out the OpenOwner later. */ if (LIST_EMPTY(&owp->nfsow_open)) nfscl_freeopenowner(owp, 0); else owp->nfsow_defunct = 1; } else { /* look for lockowners on other opens */ LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { LIST_FOREACH(lp, &op->nfso_lock, nfsl_list) { if (!NFSBCMP(lp->nfsl_owner, own, NFSV4CL_LOCKNAMELEN)) lp->nfsl_defunct = 1; } } } owp = nowp; } /* and check the defunct list */ LIST_FOREACH(lp, &clp->nfsc_defunctlockowner, nfsl_list) { if (!NFSBCMP(lp->nfsl_owner, own, NFSV4CL_LOCKNAMELEN)) lp->nfsl_defunct = 1; } } #if defined(APPLEKEXT) || defined(__FreeBSD__) /* * Simulate the call nfscl_cleanup() by looking for open owners associated * with processes that no longer exist, since a call to nfscl_cleanup() * can't be patched into exit(). */ static void nfscl_cleanupkext(struct nfsclclient *clp) { struct nfsclowner *owp, *nowp; struct nfscllockowner *lp; NFSPROCLISTLOCK(); NFSLOCKCLSTATE(); LIST_FOREACH_SAFE(owp, &clp->nfsc_owner, nfsow_list, nowp) { if (nfscl_procdoesntexist(owp->nfsow_owner)) nfscl_cleanup_common(clp, owp->nfsow_owner); } /* and check the defunct list */ LIST_FOREACH(lp, &clp->nfsc_defunctlockowner, nfsl_list) { if (nfscl_procdoesntexist(lp->nfsl_owner)) lp->nfsl_defunct = 1; } NFSUNLOCKCLSTATE(); NFSPROCLISTUNLOCK(); } #endif /* APPLEKEXT || __FreeBSD__ */ static int fake_global; /* Used to force visibility of MNTK_UNMOUNTF */ /* * Called from nfs umount to free up the clientid. */ APPLESTATIC void nfscl_umount(struct nfsmount *nmp, NFSPROC_T *p) { struct nfsclclient *clp; struct ucred *cred; int igotlock; /* * For the case that matters, this is the thread that set * MNTK_UNMOUNTF, so it will see it set. The code that follows is * done to ensure that any thread executing nfscl_getcl() after * this time, will see MNTK_UNMOUNTF set. nfscl_getcl() uses the * mutex for NFSLOCKCLSTATE(), so it is "m" for the following * explanation, courtesy of Alan Cox. * What follows is a snippet from Alan Cox's email at: * http://docs.FreeBSD.org/cgi/ * mid.cgi?BANLkTikR3d65zPHo9==08ZfJ2vmqZucEvw * * 1. Set MNTK_UNMOUNTF * 2. Acquire a standard FreeBSD mutex "m". * 3. Update some data structures. * 4. Release mutex "m". * * Then, other threads that acquire "m" after step 4 has occurred will * see MNTK_UNMOUNTF as set. But, other threads that beat thread X to * step 2 may or may not see MNTK_UNMOUNTF as set. */ NFSLOCKCLSTATE(); if ((nmp->nm_mountp->mnt_kern_flag & MNTK_UNMOUNTF) != 0) { fake_global++; NFSUNLOCKCLSTATE(); NFSLOCKCLSTATE(); } clp = nmp->nm_clp; if (clp != NULL) { if ((clp->nfsc_flags & NFSCLFLAGS_INITED) == 0) panic("nfscl umount"); /* * First, handshake with the nfscl renew thread, to terminate * it. */ clp->nfsc_flags |= NFSCLFLAGS_UMOUNT; while (clp->nfsc_flags & NFSCLFLAGS_HASTHREAD) (void)mtx_sleep(clp, NFSCLSTATEMUTEXPTR, PWAIT, "nfsclumnt", hz); /* * Now, get the exclusive lock on the client state, so * that no uses of the state are still in progress. */ do { igotlock = nfsv4_lock(&clp->nfsc_lock, 1, NULL, NFSCLSTATEMUTEXPTR, NULL); } while (!igotlock); NFSUNLOCKCLSTATE(); /* * Free up all the state. It will expire on the server, but * maybe we should do a SetClientId/SetClientIdConfirm so * the server throws it away? */ LIST_REMOVE(clp, nfsc_list); nfscl_delegreturnall(clp, p); cred = newnfs_getcred(); (void) nfsrpc_setclient(nmp, clp, cred, p); nfscl_cleanclient(clp); nmp->nm_clp = NULL; NFSFREECRED(cred); FREE((caddr_t)clp, M_NFSCLCLIENT); } else NFSUNLOCKCLSTATE(); } /* * This function is called when a server replies with NFSERR_STALECLIENTID * or NFSERR_STALESTATEID. It traverses the clientid lists, doing Opens * and Locks with reclaim. If these fail, it deletes the corresponding state. */ static void nfscl_recover(struct nfsclclient *clp, struct ucred *cred, NFSPROC_T *p) { struct nfsclowner *owp, *nowp; struct nfsclopen *op, *nop; struct nfscllockowner *lp, *nlp; struct nfscllock *lop, *nlop; struct nfscldeleg *dp, *ndp, *tdp; struct nfsmount *nmp; struct ucred *tcred; struct nfsclopenhead extra_open; struct nfscldeleghead extra_deleg; struct nfsreq *rep; u_int64_t len; u_int32_t delegtype = NFSV4OPEN_DELEGATEWRITE, mode; int igotlock = 0, error, trycnt, firstlock, s; /* * First, lock the client structure, so everyone else will * block when trying to use state. */ NFSLOCKCLSTATE(); clp->nfsc_flags |= NFSCLFLAGS_RECVRINPROG; do { igotlock = nfsv4_lock(&clp->nfsc_lock, 1, NULL, NFSCLSTATEMUTEXPTR, NULL); } while (!igotlock); NFSUNLOCKCLSTATE(); nmp = clp->nfsc_nmp; if (nmp == NULL) panic("nfscl recover"); trycnt = 5; do { error = nfsrpc_setclient(nmp, clp, cred, p); } while ((error == NFSERR_STALECLIENTID || error == NFSERR_STALEDONTRECOVER) && --trycnt > 0); if (error) { nfscl_cleanclient(clp); NFSLOCKCLSTATE(); clp->nfsc_flags &= ~(NFSCLFLAGS_HASCLIENTID | NFSCLFLAGS_RECOVER | NFSCLFLAGS_RECVRINPROG); wakeup(&clp->nfsc_flags); nfsv4_unlock(&clp->nfsc_lock, 0); NFSUNLOCKCLSTATE(); return; } clp->nfsc_flags |= NFSCLFLAGS_HASCLIENTID; clp->nfsc_flags &= ~NFSCLFLAGS_RECOVER; /* * Mark requests already queued on the server, so that they don't * initiate another recovery cycle. Any requests already in the * queue that handle state information will have the old stale * clientid/stateid and will get a NFSERR_STALESTATEID or * NFSERR_STALECLIENTID reply from the server. This will be * translated to NFSERR_STALEDONTRECOVER when R_DONTRECOVER is set. */ s = splsoftclock(); NFSLOCKREQ(); TAILQ_FOREACH(rep, &nfsd_reqq, r_chain) { if (rep->r_nmp == nmp) rep->r_flags |= R_DONTRECOVER; } NFSUNLOCKREQ(); splx(s); /* get rid of defunct lockowners */ LIST_FOREACH_SAFE(lp, &clp->nfsc_defunctlockowner, nfsl_list, nlp) { nfscl_freelockowner(lp, 0); } /* * Now, mark all delegations "need reclaim". */ TAILQ_FOREACH(dp, &clp->nfsc_deleg, nfsdl_list) dp->nfsdl_flags |= NFSCLDL_NEEDRECLAIM; TAILQ_INIT(&extra_deleg); LIST_INIT(&extra_open); /* * Now traverse the state lists, doing Open and Lock Reclaims. */ tcred = newnfs_getcred(); owp = LIST_FIRST(&clp->nfsc_owner); while (owp != NULL) { nowp = LIST_NEXT(owp, nfsow_list); owp->nfsow_seqid = 0; op = LIST_FIRST(&owp->nfsow_open); while (op != NULL) { nop = LIST_NEXT(op, nfso_list); if (error != NFSERR_NOGRACE) { /* Search for a delegation to reclaim with the open */ TAILQ_FOREACH(dp, &clp->nfsc_deleg, nfsdl_list) { if (!(dp->nfsdl_flags & NFSCLDL_NEEDRECLAIM)) continue; if ((dp->nfsdl_flags & NFSCLDL_WRITE)) { mode = NFSV4OPEN_ACCESSWRITE; delegtype = NFSV4OPEN_DELEGATEWRITE; } else { mode = NFSV4OPEN_ACCESSREAD; delegtype = NFSV4OPEN_DELEGATEREAD; } if ((op->nfso_mode & mode) == mode && op->nfso_fhlen == dp->nfsdl_fhlen && !NFSBCMP(op->nfso_fh, dp->nfsdl_fh, op->nfso_fhlen)) break; } ndp = dp; if (dp == NULL) delegtype = NFSV4OPEN_DELEGATENONE; newnfs_copycred(&op->nfso_cred, tcred); error = nfscl_tryopen(nmp, NULL, op->nfso_fh, op->nfso_fhlen, op->nfso_fh, op->nfso_fhlen, op->nfso_mode, op, NULL, 0, &ndp, 1, delegtype, tcred, p); if (!error) { /* Handle any replied delegation */ if (ndp != NULL && ((ndp->nfsdl_flags & NFSCLDL_WRITE) || NFSMNT_RDONLY(nmp->nm_mountp))) { if ((ndp->nfsdl_flags & NFSCLDL_WRITE)) mode = NFSV4OPEN_ACCESSWRITE; else mode = NFSV4OPEN_ACCESSREAD; TAILQ_FOREACH(dp, &clp->nfsc_deleg, nfsdl_list) { if (!(dp->nfsdl_flags & NFSCLDL_NEEDRECLAIM)) continue; if ((op->nfso_mode & mode) == mode && op->nfso_fhlen == dp->nfsdl_fhlen && !NFSBCMP(op->nfso_fh, dp->nfsdl_fh, op->nfso_fhlen)) { dp->nfsdl_stateid = ndp->nfsdl_stateid; dp->nfsdl_sizelimit = ndp->nfsdl_sizelimit; dp->nfsdl_ace = ndp->nfsdl_ace; dp->nfsdl_change = ndp->nfsdl_change; dp->nfsdl_flags &= ~NFSCLDL_NEEDRECLAIM; if ((ndp->nfsdl_flags & NFSCLDL_RECALL)) dp->nfsdl_flags |= NFSCLDL_RECALL; FREE((caddr_t)ndp, M_NFSCLDELEG); ndp = NULL; break; } } } if (ndp != NULL) TAILQ_INSERT_HEAD(&extra_deleg, ndp, nfsdl_list); /* and reclaim all byte range locks */ lp = LIST_FIRST(&op->nfso_lock); while (lp != NULL) { nlp = LIST_NEXT(lp, nfsl_list); lp->nfsl_seqid = 0; firstlock = 1; lop = LIST_FIRST(&lp->nfsl_lock); while (lop != NULL) { nlop = LIST_NEXT(lop, nfslo_list); if (lop->nfslo_end == NFS64BITSSET) len = NFS64BITSSET; else len = lop->nfslo_end - lop->nfslo_first; if (error != NFSERR_NOGRACE) error = nfscl_trylock(nmp, NULL, op->nfso_fh, op->nfso_fhlen, lp, firstlock, 1, lop->nfslo_first, len, lop->nfslo_type, tcred, p); if (error != 0) nfscl_freelock(lop, 0); else firstlock = 0; lop = nlop; } /* If no locks, but a lockowner, just delete it. */ if (LIST_EMPTY(&lp->nfsl_lock)) nfscl_freelockowner(lp, 0); lp = nlp; } } else { nfscl_freeopen(op, 0); } } op = nop; } owp = nowp; } /* * Now, try and get any delegations not yet reclaimed by cobbling * to-gether an appropriate open. */ nowp = NULL; dp = TAILQ_FIRST(&clp->nfsc_deleg); while (dp != NULL) { ndp = TAILQ_NEXT(dp, nfsdl_list); if ((dp->nfsdl_flags & NFSCLDL_NEEDRECLAIM)) { if (nowp == NULL) { MALLOC(nowp, struct nfsclowner *, sizeof (struct nfsclowner), M_NFSCLOWNER, M_WAITOK); /* * Name must be as long an largest possible * NFSV4CL_LOCKNAMELEN. 12 for now. */ NFSBCOPY("RECLAIMDELEG", nowp->nfsow_owner, NFSV4CL_LOCKNAMELEN); LIST_INIT(&nowp->nfsow_open); nowp->nfsow_clp = clp; nowp->nfsow_seqid = 0; nowp->nfsow_defunct = 0; nfscl_lockinit(&nowp->nfsow_rwlock); } nop = NULL; if (error != NFSERR_NOGRACE) { MALLOC(nop, struct nfsclopen *, sizeof (struct nfsclopen) + dp->nfsdl_fhlen - 1, M_NFSCLOPEN, M_WAITOK); nop->nfso_own = nowp; if ((dp->nfsdl_flags & NFSCLDL_WRITE)) { nop->nfso_mode = NFSV4OPEN_ACCESSWRITE; delegtype = NFSV4OPEN_DELEGATEWRITE; } else { nop->nfso_mode = NFSV4OPEN_ACCESSREAD; delegtype = NFSV4OPEN_DELEGATEREAD; } nop->nfso_opencnt = 0; nop->nfso_posixlock = 1; nop->nfso_fhlen = dp->nfsdl_fhlen; NFSBCOPY(dp->nfsdl_fh, nop->nfso_fh, dp->nfsdl_fhlen); LIST_INIT(&nop->nfso_lock); nop->nfso_stateid.seqid = 0; nop->nfso_stateid.other[0] = 0; nop->nfso_stateid.other[1] = 0; nop->nfso_stateid.other[2] = 0; newnfs_copycred(&dp->nfsdl_cred, tcred); newnfs_copyincred(tcred, &nop->nfso_cred); tdp = NULL; error = nfscl_tryopen(nmp, NULL, nop->nfso_fh, nop->nfso_fhlen, nop->nfso_fh, nop->nfso_fhlen, nop->nfso_mode, nop, NULL, 0, &tdp, 1, delegtype, tcred, p); if (tdp != NULL) { if ((tdp->nfsdl_flags & NFSCLDL_WRITE)) mode = NFSV4OPEN_ACCESSWRITE; else mode = NFSV4OPEN_ACCESSREAD; if ((nop->nfso_mode & mode) == mode && nop->nfso_fhlen == tdp->nfsdl_fhlen && !NFSBCMP(nop->nfso_fh, tdp->nfsdl_fh, nop->nfso_fhlen)) { dp->nfsdl_stateid = tdp->nfsdl_stateid; dp->nfsdl_sizelimit = tdp->nfsdl_sizelimit; dp->nfsdl_ace = tdp->nfsdl_ace; dp->nfsdl_change = tdp->nfsdl_change; dp->nfsdl_flags &= ~NFSCLDL_NEEDRECLAIM; if ((tdp->nfsdl_flags & NFSCLDL_RECALL)) dp->nfsdl_flags |= NFSCLDL_RECALL; FREE((caddr_t)tdp, M_NFSCLDELEG); } else { TAILQ_INSERT_HEAD(&extra_deleg, tdp, nfsdl_list); } } } if (error) { if (nop != NULL) FREE((caddr_t)nop, M_NFSCLOPEN); /* * Couldn't reclaim it, so throw the state * away. Ouch!! */ nfscl_cleandeleg(dp); nfscl_freedeleg(&clp->nfsc_deleg, dp); } else { LIST_INSERT_HEAD(&extra_open, nop, nfso_list); } } dp = ndp; } /* * Now, get rid of extra Opens and Delegations. */ LIST_FOREACH_SAFE(op, &extra_open, nfso_list, nop) { do { newnfs_copycred(&op->nfso_cred, tcred); error = nfscl_tryclose(op, tcred, nmp, p); if (error == NFSERR_GRACE) (void) nfs_catnap(PZERO, error, "nfsexcls"); } while (error == NFSERR_GRACE); LIST_REMOVE(op, nfso_list); FREE((caddr_t)op, M_NFSCLOPEN); } if (nowp != NULL) FREE((caddr_t)nowp, M_NFSCLOWNER); TAILQ_FOREACH_SAFE(dp, &extra_deleg, nfsdl_list, ndp) { do { newnfs_copycred(&dp->nfsdl_cred, tcred); error = nfscl_trydelegreturn(dp, tcred, nmp, p); if (error == NFSERR_GRACE) (void) nfs_catnap(PZERO, error, "nfsexdlg"); } while (error == NFSERR_GRACE); TAILQ_REMOVE(&extra_deleg, dp, nfsdl_list); FREE((caddr_t)dp, M_NFSCLDELEG); } NFSLOCKCLSTATE(); clp->nfsc_flags &= ~NFSCLFLAGS_RECVRINPROG; wakeup(&clp->nfsc_flags); nfsv4_unlock(&clp->nfsc_lock, 0); NFSUNLOCKCLSTATE(); NFSFREECRED(tcred); } /* * This function is called when a server replies with NFSERR_EXPIRED. * It deletes all state for the client and does a fresh SetClientId/confirm. * XXX Someday it should post a signal to the process(es) that hold the * state, so they know that lock state has been lost. */ APPLESTATIC int nfscl_hasexpired(struct nfsclclient *clp, u_int32_t clidrev, NFSPROC_T *p) { struct nfscllockowner *lp, *nlp; struct nfsmount *nmp; struct ucred *cred; int igotlock = 0, error, trycnt; /* * If the clientid has gone away or a new SetClientid has already * been done, just return ok. */ if (clp == NULL || clidrev != clp->nfsc_clientidrev) return (0); /* * First, lock the client structure, so everyone else will * block when trying to use state. Also, use NFSCLFLAGS_EXPIREIT so * that only one thread does the work. */ NFSLOCKCLSTATE(); clp->nfsc_flags |= NFSCLFLAGS_EXPIREIT; do { igotlock = nfsv4_lock(&clp->nfsc_lock, 1, NULL, NFSCLSTATEMUTEXPTR, NULL); } while (!igotlock && (clp->nfsc_flags & NFSCLFLAGS_EXPIREIT)); if ((clp->nfsc_flags & NFSCLFLAGS_EXPIREIT) == 0) { if (igotlock) nfsv4_unlock(&clp->nfsc_lock, 0); NFSUNLOCKCLSTATE(); return (0); } clp->nfsc_flags |= NFSCLFLAGS_RECVRINPROG; NFSUNLOCKCLSTATE(); nmp = clp->nfsc_nmp; if (nmp == NULL) panic("nfscl expired"); cred = newnfs_getcred(); trycnt = 5; do { error = nfsrpc_setclient(nmp, clp, cred, p); } while ((error == NFSERR_STALECLIENTID || error == NFSERR_STALEDONTRECOVER) && --trycnt > 0); if (error) { /* * Clear out any state. */ nfscl_cleanclient(clp); NFSLOCKCLSTATE(); clp->nfsc_flags &= ~(NFSCLFLAGS_HASCLIENTID | NFSCLFLAGS_RECOVER); } else { /* get rid of defunct lockowners */ LIST_FOREACH_SAFE(lp, &clp->nfsc_defunctlockowner, nfsl_list, nlp) { nfscl_freelockowner(lp, 0); } /* * Expire the state for the client. */ nfscl_expireclient(clp, nmp, cred, p); NFSLOCKCLSTATE(); clp->nfsc_flags |= NFSCLFLAGS_HASCLIENTID; clp->nfsc_flags &= ~NFSCLFLAGS_RECOVER; } clp->nfsc_flags &= ~(NFSCLFLAGS_EXPIREIT | NFSCLFLAGS_RECVRINPROG); wakeup(&clp->nfsc_flags); nfsv4_unlock(&clp->nfsc_lock, 0); NFSUNLOCKCLSTATE(); NFSFREECRED(cred); return (error); } /* * This function inserts a lock in the list after insert_lop. */ static void nfscl_insertlock(struct nfscllockowner *lp, struct nfscllock *new_lop, struct nfscllock *insert_lop, int local) { if ((struct nfscllockowner *)insert_lop == lp) LIST_INSERT_HEAD(&lp->nfsl_lock, new_lop, nfslo_list); else LIST_INSERT_AFTER(insert_lop, new_lop, nfslo_list); if (local) newnfsstats.cllocallocks++; else newnfsstats.cllocks++; } /* * This function updates the locking for a lock owner and given file. It * maintains a list of lock ranges ordered on increasing file offset that * are NFSCLLOCK_READ or NFSCLLOCK_WRITE and non-overlapping (aka POSIX style). * It always adds new_lop to the list and sometimes uses the one pointed * at by other_lopp. * Returns 1 if the locks were modified, 0 otherwise. */ static int nfscl_updatelock(struct nfscllockowner *lp, struct nfscllock **new_lopp, struct nfscllock **other_lopp, int local) { struct nfscllock *new_lop = *new_lopp; struct nfscllock *lop, *tlop, *ilop; struct nfscllock *other_lop; int unlock = 0, modified = 0; u_int64_t tmp; /* * Work down the list until the lock is merged. */ if (new_lop->nfslo_type == F_UNLCK) unlock = 1; ilop = (struct nfscllock *)lp; lop = LIST_FIRST(&lp->nfsl_lock); while (lop != NULL) { /* * Only check locks for this file that aren't before the start of * new lock's range. */ if (lop->nfslo_end >= new_lop->nfslo_first) { if (new_lop->nfslo_end < lop->nfslo_first) { /* * If the new lock ends before the start of the * current lock's range, no merge, just insert * the new lock. */ break; } if (new_lop->nfslo_type == lop->nfslo_type || (new_lop->nfslo_first <= lop->nfslo_first && new_lop->nfslo_end >= lop->nfslo_end)) { /* * This lock can be absorbed by the new lock/unlock. * This happens when it covers the entire range * of the old lock or is contiguous * with the old lock and is of the same type or an * unlock. */ if (new_lop->nfslo_type != lop->nfslo_type || new_lop->nfslo_first != lop->nfslo_first || new_lop->nfslo_end != lop->nfslo_end) modified = 1; if (lop->nfslo_first < new_lop->nfslo_first) new_lop->nfslo_first = lop->nfslo_first; if (lop->nfslo_end > new_lop->nfslo_end) new_lop->nfslo_end = lop->nfslo_end; tlop = lop; lop = LIST_NEXT(lop, nfslo_list); nfscl_freelock(tlop, local); continue; } /* * All these cases are for contiguous locks that are not the * same type, so they can't be merged. */ if (new_lop->nfslo_first <= lop->nfslo_first) { /* * This case is where the new lock overlaps with the * first part of the old lock. Move the start of the * old lock to just past the end of the new lock. The * new lock will be inserted in front of the old, since * ilop hasn't been updated. (We are done now.) */ if (lop->nfslo_first != new_lop->nfslo_end) { lop->nfslo_first = new_lop->nfslo_end; modified = 1; } break; } if (new_lop->nfslo_end >= lop->nfslo_end) { /* * This case is where the new lock overlaps with the * end of the old lock's range. Move the old lock's * end to just before the new lock's first and insert * the new lock after the old lock. * Might not be done yet, since the new lock could * overlap further locks with higher ranges. */ if (lop->nfslo_end != new_lop->nfslo_first) { lop->nfslo_end = new_lop->nfslo_first; modified = 1; } ilop = lop; lop = LIST_NEXT(lop, nfslo_list); continue; } /* * The final case is where the new lock's range is in the * middle of the current lock's and splits the current lock * up. Use *other_lopp to handle the second part of the * split old lock range. (We are done now.) * For unlock, we use new_lop as other_lop and tmp, since * other_lop and new_lop are the same for this case. * We noted the unlock case above, so we don't need * new_lop->nfslo_type any longer. */ tmp = new_lop->nfslo_first; if (unlock) { other_lop = new_lop; *new_lopp = NULL; } else { other_lop = *other_lopp; *other_lopp = NULL; } other_lop->nfslo_first = new_lop->nfslo_end; other_lop->nfslo_end = lop->nfslo_end; other_lop->nfslo_type = lop->nfslo_type; lop->nfslo_end = tmp; nfscl_insertlock(lp, other_lop, lop, local); ilop = lop; modified = 1; break; } ilop = lop; lop = LIST_NEXT(lop, nfslo_list); if (lop == NULL) break; } /* * Insert the new lock in the list at the appropriate place. */ if (!unlock) { nfscl_insertlock(lp, new_lop, ilop, local); *new_lopp = NULL; modified = 1; } return (modified); } /* * This function must be run as a kernel thread. * It does Renew Ops and recovery, when required. */ APPLESTATIC void nfscl_renewthread(struct nfsclclient *clp, NFSPROC_T *p) { struct nfsclowner *owp, *nowp; struct nfsclopen *op; struct nfscllockowner *lp, *nlp, *olp; struct nfscldeleghead dh; struct nfscllockownerhead lh; struct nfscldeleg *dp, *ndp; struct ucred *cred; u_int32_t clidrev; int error, cbpathdown, islept, igotlock, ret, clearok; uint32_t recover_done_time = 0; cred = newnfs_getcred(); NFSLOCKCLSTATE(); clp->nfsc_flags |= NFSCLFLAGS_HASTHREAD; NFSUNLOCKCLSTATE(); for(;;) { newnfs_setroot(cred); cbpathdown = 0; if (clp->nfsc_flags & NFSCLFLAGS_RECOVER) { /* * Only allow one recover within 1/2 of the lease * duration (nfsc_renew). */ if (recover_done_time < NFSD_MONOSEC) { recover_done_time = NFSD_MONOSEC + clp->nfsc_renew; nfscl_recover(clp, cred, p); } else { NFSLOCKCLSTATE(); clp->nfsc_flags &= ~NFSCLFLAGS_RECOVER; NFSUNLOCKCLSTATE(); } } if (clp->nfsc_expire <= NFSD_MONOSEC && (clp->nfsc_flags & NFSCLFLAGS_HASCLIENTID)) { clp->nfsc_expire = NFSD_MONOSEC + clp->nfsc_renew; clidrev = clp->nfsc_clientidrev; error = nfsrpc_renew(clp, cred, p); if (error == NFSERR_CBPATHDOWN) cbpathdown = 1; else if (error == NFSERR_STALECLIENTID) { NFSLOCKCLSTATE(); clp->nfsc_flags |= NFSCLFLAGS_RECOVER; NFSUNLOCKCLSTATE(); } else if (error == NFSERR_EXPIRED) (void) nfscl_hasexpired(clp, clidrev, p); } LIST_INIT(&lh); TAILQ_INIT(&dh); NFSLOCKCLSTATE(); if (cbpathdown) /* It's a Total Recall! */ nfscl_totalrecall(clp); /* * Now, handle defunct owners. */ owp = LIST_FIRST(&clp->nfsc_owner); while (owp != NULL) { nowp = LIST_NEXT(owp, nfsow_list); if (LIST_EMPTY(&owp->nfsow_open)) { if (owp->nfsow_defunct) nfscl_freeopenowner(owp, 0); } else { LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { lp = LIST_FIRST(&op->nfso_lock); while (lp != NULL) { nlp = LIST_NEXT(lp, nfsl_list); if (lp->nfsl_defunct && LIST_EMPTY(&lp->nfsl_lock)) { LIST_FOREACH(olp, &lh, nfsl_list) { if (!NFSBCMP(olp->nfsl_owner, lp->nfsl_owner,NFSV4CL_LOCKNAMELEN)) break; } if (olp == NULL) { LIST_REMOVE(lp, nfsl_list); LIST_INSERT_HEAD(&lh, lp, nfsl_list); } else { nfscl_freelockowner(lp, 0); } } lp = nlp; } } } owp = nowp; } /* also search the defunct list */ lp = LIST_FIRST(&clp->nfsc_defunctlockowner); while (lp != NULL) { nlp = LIST_NEXT(lp, nfsl_list); if (lp->nfsl_defunct) { LIST_FOREACH(olp, &lh, nfsl_list) { if (!NFSBCMP(olp->nfsl_owner, lp->nfsl_owner, NFSV4CL_LOCKNAMELEN)) break; } if (olp == NULL) { LIST_REMOVE(lp, nfsl_list); LIST_INSERT_HEAD(&lh, lp, nfsl_list); } else { nfscl_freelockowner(lp, 0); } } lp = nlp; } /* and release defunct lock owners */ LIST_FOREACH_SAFE(lp, &lh, nfsl_list, nlp) { nfscl_freelockowner(lp, 0); } /* * Do the recall on any delegations. To avoid trouble, always * come back up here after having slept. */ igotlock = 0; tryagain: dp = TAILQ_FIRST(&clp->nfsc_deleg); while (dp != NULL) { ndp = TAILQ_NEXT(dp, nfsdl_list); if ((dp->nfsdl_flags & NFSCLDL_RECALL)) { /* * Wait for outstanding I/O ops to be done. */ if (dp->nfsdl_rwlock.nfslock_usecnt > 0) { if (igotlock) { nfsv4_unlock(&clp->nfsc_lock, 0); igotlock = 0; } dp->nfsdl_rwlock.nfslock_lock |= NFSV4LOCK_WANTED; (void) nfsmsleep(&dp->nfsdl_rwlock, NFSCLSTATEMUTEXPTR, PZERO, "nfscld", NULL); goto tryagain; } while (!igotlock) { igotlock = nfsv4_lock(&clp->nfsc_lock, 1, &islept, NFSCLSTATEMUTEXPTR, NULL); if (islept) goto tryagain; } NFSUNLOCKCLSTATE(); newnfs_copycred(&dp->nfsdl_cred, cred); ret = nfscl_recalldeleg(clp, clp->nfsc_nmp, dp, NULL, cred, p, 1); if (!ret) { nfscl_cleandeleg(dp); TAILQ_REMOVE(&clp->nfsc_deleg, dp, nfsdl_list); LIST_REMOVE(dp, nfsdl_hash); TAILQ_INSERT_HEAD(&dh, dp, nfsdl_list); nfscl_delegcnt--; newnfsstats.cldelegates--; } NFSLOCKCLSTATE(); } dp = ndp; } /* * Clear out old delegations, if we are above the high water * mark. Only clear out ones with no state related to them. * The tailq list is in LRU order. */ dp = TAILQ_LAST(&clp->nfsc_deleg, nfscldeleghead); while (nfscl_delegcnt > nfscl_deleghighwater && dp != NULL) { ndp = TAILQ_PREV(dp, nfscldeleghead, nfsdl_list); if (dp->nfsdl_rwlock.nfslock_usecnt == 0 && dp->nfsdl_rwlock.nfslock_lock == 0 && dp->nfsdl_timestamp < NFSD_MONOSEC && (dp->nfsdl_flags & (NFSCLDL_RECALL | NFSCLDL_ZAPPED | NFSCLDL_NEEDRECLAIM | NFSCLDL_DELEGRET)) == 0) { clearok = 1; LIST_FOREACH(owp, &dp->nfsdl_owner, nfsow_list) { op = LIST_FIRST(&owp->nfsow_open); if (op != NULL) { clearok = 0; break; } } if (clearok) { LIST_FOREACH(lp, &dp->nfsdl_lock, nfsl_list) { if (!LIST_EMPTY(&lp->nfsl_lock)) { clearok = 0; break; } } } if (clearok) { TAILQ_REMOVE(&clp->nfsc_deleg, dp, nfsdl_list); LIST_REMOVE(dp, nfsdl_hash); TAILQ_INSERT_HEAD(&dh, dp, nfsdl_list); nfscl_delegcnt--; newnfsstats.cldelegates--; } } dp = ndp; } if (igotlock) nfsv4_unlock(&clp->nfsc_lock, 0); NFSUNLOCKCLSTATE(); /* * Delegreturn any delegations cleaned out or recalled. */ TAILQ_FOREACH_SAFE(dp, &dh, nfsdl_list, ndp) { newnfs_copycred(&dp->nfsdl_cred, cred); (void) nfscl_trydelegreturn(dp, cred, clp->nfsc_nmp, p); TAILQ_REMOVE(&dh, dp, nfsdl_list); FREE((caddr_t)dp, M_NFSCLDELEG); } #if defined(APPLEKEXT) || defined(__FreeBSD__) /* * Simulate the calls to nfscl_cleanup() when a process * exits, since the call can't be patched into exit(). */ { struct timespec mytime; static time_t prevsec = 0; NFSGETNANOTIME(&mytime); if (prevsec != mytime.tv_sec) { prevsec = mytime.tv_sec; nfscl_cleanupkext(clp); } } #endif /* APPLEKEXT || __FreeBSD__ */ NFSLOCKCLSTATE(); if ((clp->nfsc_flags & NFSCLFLAGS_RECOVER) == 0) (void)mtx_sleep(clp, NFSCLSTATEMUTEXPTR, PWAIT, "nfscl", hz); if (clp->nfsc_flags & NFSCLFLAGS_UMOUNT) { clp->nfsc_flags &= ~NFSCLFLAGS_HASTHREAD; NFSUNLOCKCLSTATE(); NFSFREECRED(cred); wakeup((caddr_t)clp); return; } NFSUNLOCKCLSTATE(); } } /* * Initiate state recovery. Called when NFSERR_STALECLIENTID or * NFSERR_STALESTATEID is received. */ APPLESTATIC void nfscl_initiate_recovery(struct nfsclclient *clp) { if (clp == NULL) return; NFSLOCKCLSTATE(); clp->nfsc_flags |= NFSCLFLAGS_RECOVER; NFSUNLOCKCLSTATE(); wakeup((caddr_t)clp); } /* * Dump out the state stuff for debugging. */ APPLESTATIC void nfscl_dumpstate(struct nfsmount *nmp, int openowner, int opens, int lockowner, int locks) { struct nfsclclient *clp; struct nfsclowner *owp; struct nfsclopen *op; struct nfscllockowner *lp; struct nfscllock *lop; struct nfscldeleg *dp; clp = nmp->nm_clp; if (clp == NULL) { printf("nfscl dumpstate NULL clp\n"); return; } NFSLOCKCLSTATE(); TAILQ_FOREACH(dp, &clp->nfsc_deleg, nfsdl_list) { LIST_FOREACH(owp, &dp->nfsdl_owner, nfsow_list) { if (openowner && !LIST_EMPTY(&owp->nfsow_open)) printf("owner=0x%x 0x%x 0x%x 0x%x seqid=%d\n", owp->nfsow_owner[0], owp->nfsow_owner[1], owp->nfsow_owner[2], owp->nfsow_owner[3], owp->nfsow_seqid); LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { if (opens) printf("open st=0x%x 0x%x 0x%x cnt=%d fh12=0x%x\n", op->nfso_stateid.other[0], op->nfso_stateid.other[1], op->nfso_stateid.other[2], op->nfso_opencnt, op->nfso_fh[12]); LIST_FOREACH(lp, &op->nfso_lock, nfsl_list) { if (lockowner) printf("lckown=0x%x 0x%x 0x%x 0x%x seqid=%d st=0x%x 0x%x 0x%x\n", lp->nfsl_owner[0], lp->nfsl_owner[1], lp->nfsl_owner[2], lp->nfsl_owner[3], lp->nfsl_seqid, lp->nfsl_stateid.other[0], lp->nfsl_stateid.other[1], lp->nfsl_stateid.other[2]); LIST_FOREACH(lop, &lp->nfsl_lock, nfslo_list) { if (locks) #ifdef __FreeBSD__ printf("lck typ=%d fst=%ju end=%ju\n", lop->nfslo_type, (intmax_t)lop->nfslo_first, (intmax_t)lop->nfslo_end); #else printf("lck typ=%d fst=%qd end=%qd\n", lop->nfslo_type, lop->nfslo_first, lop->nfslo_end); #endif } } } } } LIST_FOREACH(owp, &clp->nfsc_owner, nfsow_list) { if (openowner && !LIST_EMPTY(&owp->nfsow_open)) printf("owner=0x%x 0x%x 0x%x 0x%x seqid=%d\n", owp->nfsow_owner[0], owp->nfsow_owner[1], owp->nfsow_owner[2], owp->nfsow_owner[3], owp->nfsow_seqid); LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { if (opens) printf("open st=0x%x 0x%x 0x%x cnt=%d fh12=0x%x\n", op->nfso_stateid.other[0], op->nfso_stateid.other[1], op->nfso_stateid.other[2], op->nfso_opencnt, op->nfso_fh[12]); LIST_FOREACH(lp, &op->nfso_lock, nfsl_list) { if (lockowner) printf("lckown=0x%x 0x%x 0x%x 0x%x seqid=%d st=0x%x 0x%x 0x%x\n", lp->nfsl_owner[0], lp->nfsl_owner[1], lp->nfsl_owner[2], lp->nfsl_owner[3], lp->nfsl_seqid, lp->nfsl_stateid.other[0], lp->nfsl_stateid.other[1], lp->nfsl_stateid.other[2]); LIST_FOREACH(lop, &lp->nfsl_lock, nfslo_list) { if (locks) #ifdef __FreeBSD__ printf("lck typ=%d fst=%ju end=%ju\n", lop->nfslo_type, (intmax_t)lop->nfslo_first, (intmax_t)lop->nfslo_end); #else printf("lck typ=%d fst=%qd end=%qd\n", lop->nfslo_type, lop->nfslo_first, lop->nfslo_end); #endif } } } } NFSUNLOCKCLSTATE(); } /* * Check for duplicate open owners and opens. * (Only used as a diagnostic aid.) */ APPLESTATIC void nfscl_dupopen(vnode_t vp, int dupopens) { struct nfsclclient *clp; struct nfsclowner *owp, *owp2; struct nfsclopen *op, *op2; struct nfsfh *nfhp; clp = VFSTONFS(vnode_mount(vp))->nm_clp; if (clp == NULL) { printf("nfscl dupopen NULL clp\n"); return; } nfhp = VTONFS(vp)->n_fhp; NFSLOCKCLSTATE(); /* * First, search for duplicate owners. * These should never happen! */ LIST_FOREACH(owp2, &clp->nfsc_owner, nfsow_list) { LIST_FOREACH(owp, &clp->nfsc_owner, nfsow_list) { if (owp != owp2 && !NFSBCMP(owp->nfsow_owner, owp2->nfsow_owner, NFSV4CL_LOCKNAMELEN)) { NFSUNLOCKCLSTATE(); printf("DUP OWNER\n"); nfscl_dumpstate(VFSTONFS(vnode_mount(vp)), 1, 1, 0, 0); return; } } } /* * Now, search for duplicate stateids. * These shouldn't happen, either. */ LIST_FOREACH(owp2, &clp->nfsc_owner, nfsow_list) { LIST_FOREACH(op2, &owp2->nfsow_open, nfso_list) { LIST_FOREACH(owp, &clp->nfsc_owner, nfsow_list) { LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { if (op != op2 && (op->nfso_stateid.other[0] != 0 || op->nfso_stateid.other[1] != 0 || op->nfso_stateid.other[2] != 0) && op->nfso_stateid.other[0] == op2->nfso_stateid.other[0] && op->nfso_stateid.other[1] == op2->nfso_stateid.other[1] && op->nfso_stateid.other[2] == op2->nfso_stateid.other[2]) { NFSUNLOCKCLSTATE(); printf("DUP STATEID\n"); nfscl_dumpstate(VFSTONFS(vnode_mount(vp)), 1, 1, 0, 0); return; } } } } } /* * Now search for duplicate opens. * Duplicate opens for the same owner * should never occur. Other duplicates are * possible and are checked for if "dupopens" * is true. */ LIST_FOREACH(owp2, &clp->nfsc_owner, nfsow_list) { LIST_FOREACH(op2, &owp2->nfsow_open, nfso_list) { if (nfhp->nfh_len == op2->nfso_fhlen && !NFSBCMP(nfhp->nfh_fh, op2->nfso_fh, nfhp->nfh_len)) { LIST_FOREACH(owp, &clp->nfsc_owner, nfsow_list) { LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { if (op != op2 && nfhp->nfh_len == op->nfso_fhlen && !NFSBCMP(nfhp->nfh_fh, op->nfso_fh, nfhp->nfh_len) && (!NFSBCMP(op->nfso_own->nfsow_owner, op2->nfso_own->nfsow_owner, NFSV4CL_LOCKNAMELEN) || dupopens)) { if (!NFSBCMP(op->nfso_own->nfsow_owner, op2->nfso_own->nfsow_owner, NFSV4CL_LOCKNAMELEN)) { NFSUNLOCKCLSTATE(); printf("BADDUP OPEN\n"); } else { NFSUNLOCKCLSTATE(); printf("DUP OPEN\n"); } nfscl_dumpstate(VFSTONFS(vnode_mount(vp)), 1, 1, 0, 0); return; } } } } } } NFSUNLOCKCLSTATE(); } /* * During close, find an open that needs to be dereferenced and * dereference it. If there are no more opens for this file, * log a message to that effect. * Opens aren't actually Close'd until VOP_INACTIVE() is performed * on the file's vnode. * This is the safe way, since it is difficult to identify * which open the close is for and I/O can be performed after the * close(2) system call when a file is mmap'd. * If it returns 0 for success, there will be a referenced * clp returned via clpp. */ APPLESTATIC int nfscl_getclose(vnode_t vp, struct nfsclclient **clpp) { struct nfsclclient *clp; struct nfsclowner *owp; struct nfsclopen *op; struct nfscldeleg *dp; struct nfsfh *nfhp; int error, notdecr; error = nfscl_getcl(vp, NULL, NULL, &clp); if (error) return (error); *clpp = clp; nfhp = VTONFS(vp)->n_fhp; notdecr = 1; NFSLOCKCLSTATE(); /* * First, look for one under a delegation that was locally issued * and just decrement the opencnt for it. Since all my Opens against * the server are DENY_NONE, I don't see a problem with hanging * onto them. (It is much easier to use one of the extant Opens * that I already have on the server when a Delegation is recalled * than to do fresh Opens.) Someday, I might need to rethink this, but. */ dp = nfscl_finddeleg(clp, nfhp->nfh_fh, nfhp->nfh_len); if (dp != NULL) { LIST_FOREACH(owp, &dp->nfsdl_owner, nfsow_list) { op = LIST_FIRST(&owp->nfsow_open); if (op != NULL) { /* * Since a delegation is for a file, there * should never be more than one open for * each openowner. */ if (LIST_NEXT(op, nfso_list) != NULL) panic("nfscdeleg opens"); if (notdecr && op->nfso_opencnt > 0) { notdecr = 0; op->nfso_opencnt--; break; } } } } /* Now process the opens against the server. */ LIST_FOREACH(owp, &clp->nfsc_owner, nfsow_list) { LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { if (op->nfso_fhlen == nfhp->nfh_len && !NFSBCMP(op->nfso_fh, nfhp->nfh_fh, nfhp->nfh_len)) { /* Found an open, decrement cnt if possible */ if (notdecr && op->nfso_opencnt > 0) { notdecr = 0; op->nfso_opencnt--; } /* * There are more opens, so just return. */ if (op->nfso_opencnt > 0) { NFSUNLOCKCLSTATE(); return (0); } } } } NFSUNLOCKCLSTATE(); if (notdecr) printf("nfscl: never fnd open\n"); return (0); } APPLESTATIC int nfscl_doclose(vnode_t vp, struct nfsclclient **clpp, NFSPROC_T *p) { struct nfsclclient *clp; struct nfsclowner *owp, *nowp; struct nfsclopen *op; struct nfscldeleg *dp; struct nfsfh *nfhp; int error; error = nfscl_getcl(vp, NULL, NULL, &clp); if (error) return (error); *clpp = clp; nfhp = VTONFS(vp)->n_fhp; NFSLOCKCLSTATE(); /* * First get rid of the local Open structures, which should be no * longer in use. */ dp = nfscl_finddeleg(clp, nfhp->nfh_fh, nfhp->nfh_len); if (dp != NULL) { LIST_FOREACH_SAFE(owp, &dp->nfsdl_owner, nfsow_list, nowp) { op = LIST_FIRST(&owp->nfsow_open); if (op != NULL) { KASSERT((op->nfso_opencnt == 0), ("nfscl: bad open cnt on deleg")); nfscl_freeopen(op, 1); } nfscl_freeopenowner(owp, 1); } } /* Now process the opens against the server. */ lookformore: LIST_FOREACH(owp, &clp->nfsc_owner, nfsow_list) { op = LIST_FIRST(&owp->nfsow_open); while (op != NULL) { if (op->nfso_fhlen == nfhp->nfh_len && !NFSBCMP(op->nfso_fh, nfhp->nfh_fh, nfhp->nfh_len)) { /* Found an open, close it. */ KASSERT((op->nfso_opencnt == 0), ("nfscl: bad open cnt on server")); NFSUNLOCKCLSTATE(); nfsrpc_doclose(VFSTONFS(vnode_mount(vp)), op, p); NFSLOCKCLSTATE(); goto lookformore; } op = LIST_NEXT(op, nfso_list); } } NFSUNLOCKCLSTATE(); return (0); } /* * Return all delegations on this client. * (Must be called with client sleep lock.) */ static void nfscl_delegreturnall(struct nfsclclient *clp, NFSPROC_T *p) { struct nfscldeleg *dp, *ndp; struct ucred *cred; cred = newnfs_getcred(); TAILQ_FOREACH_SAFE(dp, &clp->nfsc_deleg, nfsdl_list, ndp) { nfscl_cleandeleg(dp); (void) nfscl_trydelegreturn(dp, cred, clp->nfsc_nmp, p); nfscl_freedeleg(&clp->nfsc_deleg, dp); } NFSFREECRED(cred); } /* * Do a callback RPC. */ APPLESTATIC void nfscl_docb(struct nfsrv_descript *nd, NFSPROC_T *p) { int i, op; u_int32_t *tl; struct nfsclclient *clp; struct nfscldeleg *dp = NULL; int numops, taglen = -1, error = 0, trunc, ret = 0; u_int32_t minorvers, retops = 0, *retopsp = NULL, *repp, cbident; u_char tag[NFSV4_SMALLSTR + 1], *tagstr; vnode_t vp = NULL; struct nfsnode *np; struct vattr va; struct nfsfh *nfhp; mount_t mp; nfsattrbit_t attrbits, rattrbits; nfsv4stateid_t stateid; nfsrvd_rephead(nd); NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); taglen = fxdr_unsigned(int, *tl); if (taglen < 0) { error = EBADRPC; goto nfsmout; } if (taglen <= NFSV4_SMALLSTR) tagstr = tag; else tagstr = malloc(taglen + 1, M_TEMP, M_WAITOK); error = nfsrv_mtostr(nd, tagstr, taglen); if (error) { if (taglen > NFSV4_SMALLSTR) free(tagstr, M_TEMP); taglen = -1; goto nfsmout; } (void) nfsm_strtom(nd, tag, taglen); if (taglen > NFSV4_SMALLSTR) { free(tagstr, M_TEMP); } NFSM_BUILD(retopsp, u_int32_t *, NFSX_UNSIGNED); NFSM_DISSECT(tl, u_int32_t *, 3 * NFSX_UNSIGNED); minorvers = fxdr_unsigned(u_int32_t, *tl++); if (minorvers != NFSV4_MINORVERSION) nd->nd_repstat = NFSERR_MINORVERMISMATCH; cbident = fxdr_unsigned(u_int32_t, *tl++); if (nd->nd_repstat) numops = 0; else numops = fxdr_unsigned(int, *tl); /* * Loop around doing the sub ops. */ for (i = 0; i < numops; i++) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); NFSM_BUILD(repp, u_int32_t *, 2 * NFSX_UNSIGNED); *repp++ = *tl; op = fxdr_unsigned(int, *tl); if (op < NFSV4OP_CBGETATTR || op > NFSV4OP_CBRECALL) { nd->nd_repstat = NFSERR_OPILLEGAL; *repp = nfscl_errmap(nd); retops++; break; } nd->nd_procnum = op; newnfsstats.cbrpccnt[nd->nd_procnum]++; switch (op) { case NFSV4OP_CBGETATTR: clp = NULL; error = nfsm_getfh(nd, &nfhp); if (!error) error = nfsrv_getattrbits(nd, &attrbits, NULL, NULL); if (!error) { mp = nfscl_getmnt(cbident); if (mp == NULL) error = NFSERR_SERVERFAULT; } if (!error) { dp = NULL; NFSLOCKCLSTATE(); clp = nfscl_findcl(VFSTONFS(mp)); if (clp != NULL) dp = nfscl_finddeleg(clp, nfhp->nfh_fh, nfhp->nfh_len); NFSUNLOCKCLSTATE(); if (dp == NULL) error = NFSERR_SERVERFAULT; } if (!error) { ret = nfscl_ngetreopen(mp, nfhp->nfh_fh, nfhp->nfh_len, p, &np); if (!ret) vp = NFSTOV(np); } if (nfhp != NULL) FREE((caddr_t)nfhp, M_NFSFH); if (!error) { NFSZERO_ATTRBIT(&rattrbits); if (NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_SIZE)) { if (!ret) va.va_size = np->n_size; else va.va_size = dp->nfsdl_size; NFSSETBIT_ATTRBIT(&rattrbits, NFSATTRBIT_SIZE); } if (NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_CHANGE)) { va.va_filerev = dp->nfsdl_change; if (ret || (np->n_flag & NDELEGMOD)) va.va_filerev++; NFSSETBIT_ATTRBIT(&rattrbits, NFSATTRBIT_CHANGE); } (void) nfsv4_fillattr(nd, NULL, NULL, NULL, &va, NULL, 0, &rattrbits, NULL, NULL, 0, 0, 0, 0, (uint64_t)0); if (!ret) vrele(vp); } break; case NFSV4OP_CBRECALL: clp = NULL; NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID + NFSX_UNSIGNED); stateid.seqid = *tl++; NFSBCOPY((caddr_t)tl, (caddr_t)stateid.other, NFSX_STATEIDOTHER); tl += (NFSX_STATEIDOTHER / NFSX_UNSIGNED); trunc = fxdr_unsigned(int, *tl); error = nfsm_getfh(nd, &nfhp); if (!error) { mp = nfscl_getmnt(cbident); if (mp == NULL) error = NFSERR_SERVERFAULT; } if (!error) { NFSLOCKCLSTATE(); clp = nfscl_findcl(VFSTONFS(mp)); if (clp != NULL) { dp = nfscl_finddeleg(clp, nfhp->nfh_fh, nfhp->nfh_len); if (dp != NULL && (dp->nfsdl_flags & NFSCLDL_DELEGRET) == 0) { dp->nfsdl_flags |= NFSCLDL_RECALL; wakeup((caddr_t)clp); } } else { error = NFSERR_SERVERFAULT; } NFSUNLOCKCLSTATE(); } if (nfhp != NULL) FREE((caddr_t)nfhp, M_NFSFH); break; }; if (error) { if (error == EBADRPC || error == NFSERR_BADXDR) { nd->nd_repstat = NFSERR_BADXDR; } else { nd->nd_repstat = error; } error = 0; } retops++; if (nd->nd_repstat) { *repp = nfscl_errmap(nd); break; } else *repp = 0; /* NFS4_OK */ } nfsmout: if (error) { if (error == EBADRPC || error == NFSERR_BADXDR) nd->nd_repstat = NFSERR_BADXDR; else printf("nfsv4 comperr1=%d\n", error); } if (taglen == -1) { NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = 0; *tl = 0; } else { *retopsp = txdr_unsigned(retops); } *nd->nd_errp = nfscl_errmap(nd); } /* * Generate the next cbident value. Basically just increment a static value * and then check that it isn't already in the list, if it has wrapped around. */ static u_int32_t nfscl_nextcbident(void) { struct nfsclclient *clp; int matched; static u_int32_t nextcbident = 0; static int haswrapped = 0; nextcbident++; if (nextcbident == 0) haswrapped = 1; if (haswrapped) { /* * Search the clientid list for one already using this cbident. */ do { matched = 0; NFSLOCKCLSTATE(); LIST_FOREACH(clp, &nfsclhead, nfsc_list) { if (clp->nfsc_cbident == nextcbident) { matched = 1; break; } } NFSUNLOCKCLSTATE(); if (matched == 1) nextcbident++; } while (matched); } return (nextcbident); } /* * Get the mount point related to a given cbident. */ static mount_t nfscl_getmnt(u_int32_t cbident) { struct nfsclclient *clp; struct nfsmount *nmp; NFSLOCKCLSTATE(); LIST_FOREACH(clp, &nfsclhead, nfsc_list) { if (clp->nfsc_cbident == cbident) break; } if (clp == NULL) { NFSUNLOCKCLSTATE(); return (NULL); } nmp = clp->nfsc_nmp; NFSUNLOCKCLSTATE(); return (nmp->nm_mountp); } /* * Search for a lock conflict locally on the client. A conflict occurs if * - not same owner and overlapping byte range and at least one of them is * a write lock or this is an unlock. */ static int nfscl_localconflict(struct nfsclclient *clp, u_int8_t *fhp, int fhlen, struct nfscllock *nlop, u_int8_t *own, struct nfscldeleg *dp, struct nfscllock **lopp) { struct nfsclowner *owp; struct nfsclopen *op; int ret; if (dp != NULL) { ret = nfscl_checkconflict(&dp->nfsdl_lock, nlop, own, lopp); if (ret) return (ret); } LIST_FOREACH(owp, &clp->nfsc_owner, nfsow_list) { LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { if (op->nfso_fhlen == fhlen && !NFSBCMP(op->nfso_fh, fhp, fhlen)) { ret = nfscl_checkconflict(&op->nfso_lock, nlop, own, lopp); if (ret) return (ret); } } } return (0); } static int nfscl_checkconflict(struct nfscllockownerhead *lhp, struct nfscllock *nlop, u_int8_t *own, struct nfscllock **lopp) { struct nfscllockowner *lp; struct nfscllock *lop; LIST_FOREACH(lp, lhp, nfsl_list) { if (NFSBCMP(lp->nfsl_owner, own, NFSV4CL_LOCKNAMELEN)) { LIST_FOREACH(lop, &lp->nfsl_lock, nfslo_list) { if (lop->nfslo_first >= nlop->nfslo_end) break; if (lop->nfslo_end <= nlop->nfslo_first) continue; if (lop->nfslo_type == F_WRLCK || nlop->nfslo_type == F_WRLCK || nlop->nfslo_type == F_UNLCK) { if (lopp != NULL) *lopp = lop; return (NFSERR_DENIED); } } } } return (0); } /* * Check for a local conflicting lock. */ APPLESTATIC int nfscl_lockt(vnode_t vp, struct nfsclclient *clp, u_int64_t off, - u_int64_t len, struct flock *fl, NFSPROC_T *p) + u_int64_t len, struct flock *fl, NFSPROC_T *p, void *id, int flags) { struct nfscllock *lop, nlck; struct nfscldeleg *dp; struct nfsnode *np; u_int8_t own[NFSV4CL_LOCKNAMELEN]; int error; nlck.nfslo_type = fl->l_type; nlck.nfslo_first = off; if (len == NFS64BITSSET) { nlck.nfslo_end = NFS64BITSSET; } else { nlck.nfslo_end = off + len; if (nlck.nfslo_end <= nlck.nfslo_first) return (NFSERR_INVAL); } np = VTONFS(vp); - nfscl_filllockowner(p, own); + nfscl_filllockowner(id, own, flags); NFSLOCKCLSTATE(); dp = nfscl_finddeleg(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len); error = nfscl_localconflict(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len, &nlck, own, dp, &lop); if (error != 0) { fl->l_whence = SEEK_SET; fl->l_start = lop->nfslo_first; if (lop->nfslo_end == NFS64BITSSET) fl->l_len = 0; else fl->l_len = lop->nfslo_end - lop->nfslo_first; fl->l_pid = (pid_t)0; fl->l_type = lop->nfslo_type; error = -1; /* no RPC required */ } else if (dp != NULL && ((dp->nfsdl_flags & NFSCLDL_WRITE) || fl->l_type == F_RDLCK)) { /* * The delegation ensures that there isn't a conflicting * lock on the server, so return -1 to indicate an RPC * isn't required. */ fl->l_type = F_UNLCK; error = -1; } NFSUNLOCKCLSTATE(); return (error); } /* * Handle Recall of a delegation. * The clp must be exclusive locked when this is called. */ static int nfscl_recalldeleg(struct nfsclclient *clp, struct nfsmount *nmp, struct nfscldeleg *dp, vnode_t vp, struct ucred *cred, NFSPROC_T *p, int called_from_renewthread) { struct nfsclowner *owp, *lowp, *nowp; struct nfsclopen *op, *lop; struct nfscllockowner *lp; struct nfscllock *lckp; struct nfsnode *np; int error = 0, ret, gotvp = 0; if (vp == NULL) { /* * First, get a vnode for the file. This is needed to do RPCs. */ ret = nfscl_ngetreopen(nmp->nm_mountp, dp->nfsdl_fh, dp->nfsdl_fhlen, p, &np); if (ret) { /* * File isn't open, so nothing to move over to the * server. */ return (0); } vp = NFSTOV(np); gotvp = 1; } else { np = VTONFS(vp); } dp->nfsdl_flags &= ~NFSCLDL_MODTIMESET; /* * Ok, if it's a write delegation, flush data to the server, so * that close/open consistency is retained. */ ret = 0; NFSLOCKNODE(np); if ((dp->nfsdl_flags & NFSCLDL_WRITE) && (np->n_flag & NMODIFIED)) { np->n_flag |= NDELEGRECALL; NFSUNLOCKNODE(np); ret = ncl_flush(vp, MNT_WAIT, cred, p, 1, called_from_renewthread); NFSLOCKNODE(np); np->n_flag &= ~NDELEGRECALL; } NFSINVALATTRCACHE(np); NFSUNLOCKNODE(np); if (ret == EIO && called_from_renewthread != 0) { /* * If the flush failed with EIO for the renew thread, * return now, so that the dirty buffer will be flushed * later. */ if (gotvp != 0) vrele(vp); return (ret); } /* * Now, for each openowner with opens issued locally, move them * over to state against the server. */ LIST_FOREACH(lowp, &dp->nfsdl_owner, nfsow_list) { lop = LIST_FIRST(&lowp->nfsow_open); if (lop != NULL) { if (LIST_NEXT(lop, nfso_list) != NULL) panic("nfsdlg mult opens"); /* * Look for the same openowner against the server. */ LIST_FOREACH(owp, &clp->nfsc_owner, nfsow_list) { if (!NFSBCMP(lowp->nfsow_owner, owp->nfsow_owner, NFSV4CL_LOCKNAMELEN)) { newnfs_copycred(&dp->nfsdl_cred, cred); ret = nfscl_moveopen(vp, clp, nmp, lop, owp, dp, cred, p); if (ret == NFSERR_STALECLIENTID || ret == NFSERR_STALEDONTRECOVER) { if (gotvp) vrele(vp); return (ret); } if (ret) { nfscl_freeopen(lop, 1); if (!error) error = ret; } break; } } /* * If no openowner found, create one and get an open * for it. */ if (owp == NULL) { MALLOC(nowp, struct nfsclowner *, sizeof (struct nfsclowner), M_NFSCLOWNER, M_WAITOK); nfscl_newopen(clp, NULL, &owp, &nowp, &op, NULL, lowp->nfsow_owner, dp->nfsdl_fh, dp->nfsdl_fhlen, NULL); newnfs_copycred(&dp->nfsdl_cred, cred); ret = nfscl_moveopen(vp, clp, nmp, lop, owp, dp, cred, p); if (ret) { nfscl_freeopenowner(owp, 0); if (ret == NFSERR_STALECLIENTID || ret == NFSERR_STALEDONTRECOVER) { if (gotvp) vrele(vp); return (ret); } if (ret) { nfscl_freeopen(lop, 1); if (!error) error = ret; } } } } } /* * Now, get byte range locks for any locks done locally. */ LIST_FOREACH(lp, &dp->nfsdl_lock, nfsl_list) { LIST_FOREACH(lckp, &lp->nfsl_lock, nfslo_list) { newnfs_copycred(&dp->nfsdl_cred, cred); ret = nfscl_relock(vp, clp, nmp, lp, lckp, cred, p); if (ret == NFSERR_STALESTATEID || ret == NFSERR_STALEDONTRECOVER || ret == NFSERR_STALECLIENTID) { if (gotvp) vrele(vp); return (ret); } if (ret && !error) error = ret; } } if (gotvp) vrele(vp); return (error); } /* * Move a locally issued open over to an owner on the state list. * SIDE EFFECT: If it needs to sleep (do an rpc), it unlocks clstate and * returns with it unlocked. */ static int nfscl_moveopen(vnode_t vp, struct nfsclclient *clp, struct nfsmount *nmp, struct nfsclopen *lop, struct nfsclowner *owp, struct nfscldeleg *dp, struct ucred *cred, NFSPROC_T *p) { struct nfsclopen *op, *nop; struct nfscldeleg *ndp; struct nfsnode *np; int error = 0, newone; /* * First, look for an appropriate open, If found, just increment the * opencnt in it. */ LIST_FOREACH(op, &owp->nfsow_open, nfso_list) { if ((op->nfso_mode & lop->nfso_mode) == lop->nfso_mode && op->nfso_fhlen == lop->nfso_fhlen && !NFSBCMP(op->nfso_fh, lop->nfso_fh, op->nfso_fhlen)) { op->nfso_opencnt += lop->nfso_opencnt; nfscl_freeopen(lop, 1); return (0); } } /* No appropriate open, so we have to do one against the server. */ np = VTONFS(vp); MALLOC(nop, struct nfsclopen *, sizeof (struct nfsclopen) + lop->nfso_fhlen - 1, M_NFSCLOPEN, M_WAITOK); newone = 0; nfscl_newopen(clp, NULL, &owp, NULL, &op, &nop, owp->nfsow_owner, lop->nfso_fh, lop->nfso_fhlen, &newone); ndp = dp; error = nfscl_tryopen(nmp, vp, np->n_v4->n4_data, np->n_v4->n4_fhlen, lop->nfso_fh, lop->nfso_fhlen, lop->nfso_mode, op, NFS4NODENAME(np->n_v4), np->n_v4->n4_namelen, &ndp, 0, 0, cred, p); if (error) { if (newone) nfscl_freeopen(op, 0); } else { if (newone) newnfs_copyincred(cred, &op->nfso_cred); op->nfso_mode |= lop->nfso_mode; op->nfso_opencnt += lop->nfso_opencnt; nfscl_freeopen(lop, 1); } if (nop != NULL) FREE((caddr_t)nop, M_NFSCLOPEN); if (ndp != NULL) { /* * What should I do with the returned delegation, since the * delegation is being recalled? For now, just printf and * through it away. */ printf("Moveopen returned deleg\n"); FREE((caddr_t)ndp, M_NFSCLDELEG); } return (error); } /* * Recall all delegations on this client. */ static void nfscl_totalrecall(struct nfsclclient *clp) { struct nfscldeleg *dp; TAILQ_FOREACH(dp, &clp->nfsc_deleg, nfsdl_list) { if ((dp->nfsdl_flags & NFSCLDL_DELEGRET) == 0) dp->nfsdl_flags |= NFSCLDL_RECALL; } } /* * Relock byte ranges. Called for delegation recall and state expiry. */ static int nfscl_relock(vnode_t vp, struct nfsclclient *clp, struct nfsmount *nmp, struct nfscllockowner *lp, struct nfscllock *lop, struct ucred *cred, NFSPROC_T *p) { struct nfscllockowner *nlp; struct nfsfh *nfhp; u_int64_t off, len; u_int32_t clidrev = 0; int error, newone, donelocally; off = lop->nfslo_first; len = lop->nfslo_end - lop->nfslo_first; error = nfscl_getbytelock(vp, off, len, lop->nfslo_type, cred, p, - clp, 1, lp->nfsl_owner, lp->nfsl_openowner, &nlp, &newone, + clp, 1, NULL, 0, lp->nfsl_owner, lp->nfsl_openowner, &nlp, &newone, &donelocally); if (error || donelocally) return (error); if (nmp->nm_clp != NULL) clidrev = nmp->nm_clp->nfsc_clientidrev; else clidrev = 0; nfhp = VTONFS(vp)->n_fhp; error = nfscl_trylock(nmp, vp, nfhp->nfh_fh, nfhp->nfh_len, nlp, newone, 0, off, len, lop->nfslo_type, cred, p); if (error) nfscl_freelockowner(nlp, 0); return (error); } /* * Called to re-open a file. Basically get a vnode for the file handle * and then call nfsrpc_openrpc() to do the rest. */ static int nfsrpc_reopen(struct nfsmount *nmp, u_int8_t *fhp, int fhlen, u_int32_t mode, struct nfsclopen *op, struct nfscldeleg **dpp, struct ucred *cred, NFSPROC_T *p) { struct nfsnode *np; vnode_t vp; int error; error = nfscl_ngetreopen(nmp->nm_mountp, fhp, fhlen, p, &np); if (error) return (error); vp = NFSTOV(np); if (np->n_v4 != NULL) { error = nfscl_tryopen(nmp, vp, np->n_v4->n4_data, np->n_v4->n4_fhlen, fhp, fhlen, mode, op, NFS4NODENAME(np->n_v4), np->n_v4->n4_namelen, dpp, 0, 0, cred, p); } else { error = EINVAL; } vrele(vp); return (error); } /* * Try an open against the server. Just call nfsrpc_openrpc(), retrying while * NFSERR_DELAY. Also, try system credentials, if the passed in credentials * fail. */ static int nfscl_tryopen(struct nfsmount *nmp, vnode_t vp, u_int8_t *fhp, int fhlen, u_int8_t *newfhp, int newfhlen, u_int32_t mode, struct nfsclopen *op, u_int8_t *name, int namelen, struct nfscldeleg **ndpp, int reclaim, u_int32_t delegtype, struct ucred *cred, NFSPROC_T *p) { int error; do { error = nfsrpc_openrpc(nmp, vp, fhp, fhlen, newfhp, newfhlen, mode, op, name, namelen, ndpp, reclaim, delegtype, cred, p, 0, 0); if (error == NFSERR_DELAY) (void) nfs_catnap(PZERO, error, "nfstryop"); } while (error == NFSERR_DELAY); if (error == EAUTH || error == EACCES) { /* Try again using system credentials */ newnfs_setroot(cred); do { error = nfsrpc_openrpc(nmp, vp, fhp, fhlen, newfhp, newfhlen, mode, op, name, namelen, ndpp, reclaim, delegtype, cred, p, 1, 0); if (error == NFSERR_DELAY) (void) nfs_catnap(PZERO, error, "nfstryop"); } while (error == NFSERR_DELAY); } return (error); } /* * Try a byte range lock. Just loop on nfsrpc_lock() while it returns * NFSERR_DELAY. Also, retry with system credentials, if the provided * cred don't work. */ static int nfscl_trylock(struct nfsmount *nmp, vnode_t vp, u_int8_t *fhp, int fhlen, struct nfscllockowner *nlp, int newone, int reclaim, u_int64_t off, u_int64_t len, short type, struct ucred *cred, NFSPROC_T *p) { struct nfsrv_descript nfsd, *nd = &nfsd; int error; do { error = nfsrpc_lock(nd, nmp, vp, fhp, fhlen, nlp, newone, reclaim, off, len, type, cred, p, 0); if (!error && nd->nd_repstat == NFSERR_DELAY) (void) nfs_catnap(PZERO, (int)nd->nd_repstat, "nfstrylck"); } while (!error && nd->nd_repstat == NFSERR_DELAY); if (!error) error = nd->nd_repstat; if (error == EAUTH || error == EACCES) { /* Try again using root credentials */ newnfs_setroot(cred); do { error = nfsrpc_lock(nd, nmp, vp, fhp, fhlen, nlp, newone, reclaim, off, len, type, cred, p, 1); if (!error && nd->nd_repstat == NFSERR_DELAY) (void) nfs_catnap(PZERO, (int)nd->nd_repstat, "nfstrylck"); } while (!error && nd->nd_repstat == NFSERR_DELAY); if (!error) error = nd->nd_repstat; } return (error); } /* * Try a delegreturn against the server. Just call nfsrpc_delegreturn(), * retrying while NFSERR_DELAY. Also, try system credentials, if the passed in * credentials fail. */ static int nfscl_trydelegreturn(struct nfscldeleg *dp, struct ucred *cred, struct nfsmount *nmp, NFSPROC_T *p) { int error; do { error = nfsrpc_delegreturn(dp, cred, nmp, p, 0); if (error == NFSERR_DELAY) (void) nfs_catnap(PZERO, error, "nfstrydp"); } while (error == NFSERR_DELAY); if (error == EAUTH || error == EACCES) { /* Try again using system credentials */ newnfs_setroot(cred); do { error = nfsrpc_delegreturn(dp, cred, nmp, p, 1); if (error == NFSERR_DELAY) (void) nfs_catnap(PZERO, error, "nfstrydp"); } while (error == NFSERR_DELAY); } return (error); } /* * Try a close against the server. Just call nfsrpc_closerpc(), * retrying while NFSERR_DELAY. Also, try system credentials, if the passed in * credentials fail. */ APPLESTATIC int nfscl_tryclose(struct nfsclopen *op, struct ucred *cred, struct nfsmount *nmp, NFSPROC_T *p) { struct nfsrv_descript nfsd, *nd = &nfsd; int error; do { error = nfsrpc_closerpc(nd, nmp, op, cred, p, 0); if (error == NFSERR_DELAY) (void) nfs_catnap(PZERO, error, "nfstrycl"); } while (error == NFSERR_DELAY); if (error == EAUTH || error == EACCES) { /* Try again using system credentials */ newnfs_setroot(cred); do { error = nfsrpc_closerpc(nd, nmp, op, cred, p, 1); if (error == NFSERR_DELAY) (void) nfs_catnap(PZERO, error, "nfstrycl"); } while (error == NFSERR_DELAY); } return (error); } /* * Decide if a delegation on a file permits close without flushing writes * to the server. This might be a big performance win in some environments. * (Not useful until the client does caching on local stable storage.) */ APPLESTATIC int nfscl_mustflush(vnode_t vp) { struct nfsclclient *clp; struct nfscldeleg *dp; struct nfsnode *np; struct nfsmount *nmp; np = VTONFS(vp); nmp = VFSTONFS(vnode_mount(vp)); if (!NFSHASNFSV4(nmp)) return (1); NFSLOCKCLSTATE(); clp = nfscl_findcl(nmp); if (clp == NULL) { NFSUNLOCKCLSTATE(); return (1); } dp = nfscl_finddeleg(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len); if (dp != NULL && (dp->nfsdl_flags & (NFSCLDL_WRITE | NFSCLDL_RECALL | NFSCLDL_DELEGRET)) == NFSCLDL_WRITE && (dp->nfsdl_sizelimit >= np->n_size || !NFSHASSTRICT3530(nmp))) { NFSUNLOCKCLSTATE(); return (0); } NFSUNLOCKCLSTATE(); return (1); } /* * See if a (write) delegation exists for this file. */ APPLESTATIC int nfscl_nodeleg(vnode_t vp, int writedeleg) { struct nfsclclient *clp; struct nfscldeleg *dp; struct nfsnode *np; struct nfsmount *nmp; np = VTONFS(vp); nmp = VFSTONFS(vnode_mount(vp)); if (!NFSHASNFSV4(nmp)) return (1); NFSLOCKCLSTATE(); clp = nfscl_findcl(nmp); if (clp == NULL) { NFSUNLOCKCLSTATE(); return (1); } dp = nfscl_finddeleg(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len); if (dp != NULL && (dp->nfsdl_flags & (NFSCLDL_RECALL | NFSCLDL_DELEGRET)) == 0 && (writedeleg == 0 || (dp->nfsdl_flags & NFSCLDL_WRITE) == NFSCLDL_WRITE)) { NFSUNLOCKCLSTATE(); return (0); } NFSUNLOCKCLSTATE(); return (1); } /* * Look for an associated delegation that should be DelegReturned. */ APPLESTATIC int nfscl_removedeleg(vnode_t vp, NFSPROC_T *p, nfsv4stateid_t *stp) { struct nfsclclient *clp; struct nfscldeleg *dp; struct nfsclowner *owp; struct nfscllockowner *lp; struct nfsmount *nmp; struct ucred *cred; struct nfsnode *np; int igotlock = 0, triedrecall = 0, needsrecall, retcnt = 0, islept; nmp = VFSTONFS(vnode_mount(vp)); np = VTONFS(vp); NFSLOCKCLSTATE(); /* * Loop around waiting for: * - outstanding I/O operations on delegations to complete * - for a delegation on vp that has state, lock the client and * do a recall * - return delegation with no state */ while (1) { clp = nfscl_findcl(nmp); if (clp == NULL) { NFSUNLOCKCLSTATE(); return (retcnt); } dp = nfscl_finddeleg(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len); if (dp != NULL) { /* * Wait for outstanding I/O ops to be done. */ if (dp->nfsdl_rwlock.nfslock_usecnt > 0) { if (igotlock) { nfsv4_unlock(&clp->nfsc_lock, 0); igotlock = 0; } dp->nfsdl_rwlock.nfslock_lock |= NFSV4LOCK_WANTED; (void) nfsmsleep(&dp->nfsdl_rwlock, NFSCLSTATEMUTEXPTR, PZERO, "nfscld", NULL); continue; } needsrecall = 0; LIST_FOREACH(owp, &dp->nfsdl_owner, nfsow_list) { if (!LIST_EMPTY(&owp->nfsow_open)) { needsrecall = 1; break; } } if (!needsrecall) { LIST_FOREACH(lp, &dp->nfsdl_lock, nfsl_list) { if (!LIST_EMPTY(&lp->nfsl_lock)) { needsrecall = 1; break; } } } if (needsrecall && !triedrecall) { dp->nfsdl_flags |= NFSCLDL_DELEGRET; islept = 0; while (!igotlock) { igotlock = nfsv4_lock(&clp->nfsc_lock, 1, &islept, NFSCLSTATEMUTEXPTR, NULL); if (islept) break; } if (islept) continue; NFSUNLOCKCLSTATE(); cred = newnfs_getcred(); newnfs_copycred(&dp->nfsdl_cred, cred); (void) nfscl_recalldeleg(clp, nmp, dp, vp, cred, p, 0); NFSFREECRED(cred); triedrecall = 1; NFSLOCKCLSTATE(); nfsv4_unlock(&clp->nfsc_lock, 0); igotlock = 0; continue; } *stp = dp->nfsdl_stateid; retcnt = 1; nfscl_cleandeleg(dp); nfscl_freedeleg(&clp->nfsc_deleg, dp); } if (igotlock) nfsv4_unlock(&clp->nfsc_lock, 0); NFSUNLOCKCLSTATE(); return (retcnt); } } /* * Look for associated delegation(s) that should be DelegReturned. */ APPLESTATIC int nfscl_renamedeleg(vnode_t fvp, nfsv4stateid_t *fstp, int *gotfdp, vnode_t tvp, nfsv4stateid_t *tstp, int *gottdp, NFSPROC_T *p) { struct nfsclclient *clp; struct nfscldeleg *dp; struct nfsclowner *owp; struct nfscllockowner *lp; struct nfsmount *nmp; struct ucred *cred; struct nfsnode *np; int igotlock = 0, triedrecall = 0, needsrecall, retcnt = 0, islept; nmp = VFSTONFS(vnode_mount(fvp)); *gotfdp = 0; *gottdp = 0; NFSLOCKCLSTATE(); /* * Loop around waiting for: * - outstanding I/O operations on delegations to complete * - for a delegation on fvp that has state, lock the client and * do a recall * - return delegation(s) with no state. */ while (1) { clp = nfscl_findcl(nmp); if (clp == NULL) { NFSUNLOCKCLSTATE(); return (retcnt); } np = VTONFS(fvp); dp = nfscl_finddeleg(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len); if (dp != NULL && *gotfdp == 0) { /* * Wait for outstanding I/O ops to be done. */ if (dp->nfsdl_rwlock.nfslock_usecnt > 0) { if (igotlock) { nfsv4_unlock(&clp->nfsc_lock, 0); igotlock = 0; } dp->nfsdl_rwlock.nfslock_lock |= NFSV4LOCK_WANTED; (void) nfsmsleep(&dp->nfsdl_rwlock, NFSCLSTATEMUTEXPTR, PZERO, "nfscld", NULL); continue; } needsrecall = 0; LIST_FOREACH(owp, &dp->nfsdl_owner, nfsow_list) { if (!LIST_EMPTY(&owp->nfsow_open)) { needsrecall = 1; break; } } if (!needsrecall) { LIST_FOREACH(lp, &dp->nfsdl_lock, nfsl_list) { if (!LIST_EMPTY(&lp->nfsl_lock)) { needsrecall = 1; break; } } } if (needsrecall && !triedrecall) { dp->nfsdl_flags |= NFSCLDL_DELEGRET; islept = 0; while (!igotlock) { igotlock = nfsv4_lock(&clp->nfsc_lock, 1, &islept, NFSCLSTATEMUTEXPTR, NULL); if (islept) break; } if (islept) continue; NFSUNLOCKCLSTATE(); cred = newnfs_getcred(); newnfs_copycred(&dp->nfsdl_cred, cred); (void) nfscl_recalldeleg(clp, nmp, dp, fvp, cred, p, 0); NFSFREECRED(cred); triedrecall = 1; NFSLOCKCLSTATE(); nfsv4_unlock(&clp->nfsc_lock, 0); igotlock = 0; continue; } *fstp = dp->nfsdl_stateid; retcnt++; *gotfdp = 1; nfscl_cleandeleg(dp); nfscl_freedeleg(&clp->nfsc_deleg, dp); } if (igotlock) { nfsv4_unlock(&clp->nfsc_lock, 0); igotlock = 0; } if (tvp != NULL) { np = VTONFS(tvp); dp = nfscl_finddeleg(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len); if (dp != NULL && *gottdp == 0) { /* * Wait for outstanding I/O ops to be done. */ if (dp->nfsdl_rwlock.nfslock_usecnt > 0) { dp->nfsdl_rwlock.nfslock_lock |= NFSV4LOCK_WANTED; (void) nfsmsleep(&dp->nfsdl_rwlock, NFSCLSTATEMUTEXPTR, PZERO, "nfscld", NULL); continue; } LIST_FOREACH(owp, &dp->nfsdl_owner, nfsow_list) { if (!LIST_EMPTY(&owp->nfsow_open)) { NFSUNLOCKCLSTATE(); return (retcnt); } } LIST_FOREACH(lp, &dp->nfsdl_lock, nfsl_list) { if (!LIST_EMPTY(&lp->nfsl_lock)) { NFSUNLOCKCLSTATE(); return (retcnt); } } *tstp = dp->nfsdl_stateid; retcnt++; *gottdp = 1; nfscl_cleandeleg(dp); nfscl_freedeleg(&clp->nfsc_deleg, dp); } } NFSUNLOCKCLSTATE(); return (retcnt); } } /* * Get a reference on the clientid associated with the mount point. * Return 1 if success, 0 otherwise. */ APPLESTATIC int nfscl_getref(struct nfsmount *nmp) { struct nfsclclient *clp; NFSLOCKCLSTATE(); clp = nfscl_findcl(nmp); if (clp == NULL) { NFSUNLOCKCLSTATE(); return (0); } nfsv4_getref(&clp->nfsc_lock, NULL, NFSCLSTATEMUTEXPTR, NULL); NFSUNLOCKCLSTATE(); return (1); } /* * Release a reference on a clientid acquired with the above call. */ APPLESTATIC void nfscl_relref(struct nfsmount *nmp) { struct nfsclclient *clp; NFSLOCKCLSTATE(); clp = nfscl_findcl(nmp); if (clp == NULL) { NFSUNLOCKCLSTATE(); return; } nfsv4_relref(&clp->nfsc_lock); NFSUNLOCKCLSTATE(); } /* * Save the size attribute in the delegation, since the nfsnode * is going away. */ APPLESTATIC void nfscl_reclaimnode(vnode_t vp) { struct nfsclclient *clp; struct nfscldeleg *dp; struct nfsnode *np = VTONFS(vp); struct nfsmount *nmp; nmp = VFSTONFS(vnode_mount(vp)); if (!NFSHASNFSV4(nmp)) return; NFSLOCKCLSTATE(); clp = nfscl_findcl(nmp); if (clp == NULL) { NFSUNLOCKCLSTATE(); return; } dp = nfscl_finddeleg(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len); if (dp != NULL && (dp->nfsdl_flags & NFSCLDL_WRITE)) dp->nfsdl_size = np->n_size; NFSUNLOCKCLSTATE(); } /* * Get the saved size attribute in the delegation, since it is a * newly allocated nfsnode. */ APPLESTATIC void nfscl_newnode(vnode_t vp) { struct nfsclclient *clp; struct nfscldeleg *dp; struct nfsnode *np = VTONFS(vp); struct nfsmount *nmp; nmp = VFSTONFS(vnode_mount(vp)); if (!NFSHASNFSV4(nmp)) return; NFSLOCKCLSTATE(); clp = nfscl_findcl(nmp); if (clp == NULL) { NFSUNLOCKCLSTATE(); return; } dp = nfscl_finddeleg(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len); if (dp != NULL && (dp->nfsdl_flags & NFSCLDL_WRITE)) np->n_size = dp->nfsdl_size; NFSUNLOCKCLSTATE(); } /* * If there is a valid write delegation for this file, set the modtime * to the local clock time. */ APPLESTATIC void nfscl_delegmodtime(vnode_t vp) { struct nfsclclient *clp; struct nfscldeleg *dp; struct nfsnode *np = VTONFS(vp); struct nfsmount *nmp; nmp = VFSTONFS(vnode_mount(vp)); if (!NFSHASNFSV4(nmp)) return; NFSLOCKCLSTATE(); clp = nfscl_findcl(nmp); if (clp == NULL) { NFSUNLOCKCLSTATE(); return; } dp = nfscl_finddeleg(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len); if (dp != NULL && (dp->nfsdl_flags & NFSCLDL_WRITE)) { NFSGETNANOTIME(&dp->nfsdl_modtime); dp->nfsdl_flags |= NFSCLDL_MODTIMESET; } NFSUNLOCKCLSTATE(); } /* * If there is a valid write delegation for this file with a modtime set, * put that modtime in mtime. */ APPLESTATIC void nfscl_deleggetmodtime(vnode_t vp, struct timespec *mtime) { struct nfsclclient *clp; struct nfscldeleg *dp; struct nfsnode *np = VTONFS(vp); struct nfsmount *nmp; nmp = VFSTONFS(vnode_mount(vp)); if (!NFSHASNFSV4(nmp)) return; NFSLOCKCLSTATE(); clp = nfscl_findcl(nmp); if (clp == NULL) { NFSUNLOCKCLSTATE(); return; } dp = nfscl_finddeleg(clp, np->n_fhp->nfh_fh, np->n_fhp->nfh_len); if (dp != NULL && (dp->nfsdl_flags & (NFSCLDL_WRITE | NFSCLDL_MODTIMESET)) == (NFSCLDL_WRITE | NFSCLDL_MODTIMESET)) *mtime = dp->nfsdl_modtime; NFSUNLOCKCLSTATE(); } static int nfscl_errmap(struct nfsrv_descript *nd) { short *defaulterrp, *errp; if (!nd->nd_repstat) return (0); if (nd->nd_procnum == NFSPROC_NOOP) return (txdr_unsigned(nd->nd_repstat & 0xffff)); if (nd->nd_repstat == EBADRPC) return (txdr_unsigned(NFSERR_BADXDR)); if (nd->nd_repstat == NFSERR_MINORVERMISMATCH || nd->nd_repstat == NFSERR_OPILLEGAL) return (txdr_unsigned(nd->nd_repstat)); errp = defaulterrp = nfscl_cberrmap[nd->nd_procnum]; while (*++errp) if (*errp == (short)nd->nd_repstat) return (txdr_unsigned(nd->nd_repstat)); return (txdr_unsigned(*defaulterrp)); } Index: projects/largeSMP/sys/fs/nfsclient/nfs_clvnops.c =================================================================== --- projects/largeSMP/sys/fs/nfsclient/nfs_clvnops.c (revision 222782) +++ projects/largeSMP/sys/fs/nfsclient/nfs_clvnops.c (revision 222783) @@ -1,3403 +1,3407 @@ /*- * Copyright (c) 1989, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Rick Macklem at The University of Guelph. * * 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 nfs_vnops.c 8.16 (Berkeley) 5/27/95 */ #include __FBSDID("$FreeBSD$"); /* * vnode op calls for Sun NFS version 2, 3 and 4 */ #include "opt_inet.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 /* Defs */ #define TRUE 1 #define FALSE 0 extern struct nfsstats newnfsstats; extern int nfsrv_useacl; MALLOC_DECLARE(M_NEWNFSREQ); /* * Ifdef for FreeBSD-current merged buffer cache. It is unfortunate that these * calls are not in getblk() and brelse() so that they would not be necessary * here. */ #ifndef B_VMIO #define vfs_busy_pages(bp, f) #endif static vop_read_t nfsfifo_read; static vop_write_t nfsfifo_write; static vop_close_t nfsfifo_close; static int nfs_setattrrpc(struct vnode *, struct vattr *, struct ucred *, struct thread *); static vop_lookup_t nfs_lookup; static vop_create_t nfs_create; static vop_mknod_t nfs_mknod; static vop_open_t nfs_open; static vop_pathconf_t nfs_pathconf; static vop_close_t nfs_close; static vop_access_t nfs_access; static vop_getattr_t nfs_getattr; static vop_setattr_t nfs_setattr; static vop_read_t nfs_read; static vop_fsync_t nfs_fsync; static vop_remove_t nfs_remove; static vop_link_t nfs_link; static vop_rename_t nfs_rename; static vop_mkdir_t nfs_mkdir; static vop_rmdir_t nfs_rmdir; static vop_symlink_t nfs_symlink; static vop_readdir_t nfs_readdir; static vop_strategy_t nfs_strategy; static vop_lock1_t nfs_lock1; static int nfs_lookitup(struct vnode *, char *, int, struct ucred *, struct thread *, struct nfsnode **); static int nfs_sillyrename(struct vnode *, struct vnode *, struct componentname *); static vop_access_t nfsspec_access; static vop_readlink_t nfs_readlink; static vop_print_t nfs_print; static vop_advlock_t nfs_advlock; static vop_advlockasync_t nfs_advlockasync; static vop_getacl_t nfs_getacl; static vop_setacl_t nfs_setacl; /* * Global vfs data structures for nfs */ struct vop_vector newnfs_vnodeops = { .vop_default = &default_vnodeops, .vop_access = nfs_access, .vop_advlock = nfs_advlock, .vop_advlockasync = nfs_advlockasync, .vop_close = nfs_close, .vop_create = nfs_create, .vop_fsync = nfs_fsync, .vop_getattr = nfs_getattr, .vop_getpages = ncl_getpages, .vop_putpages = ncl_putpages, .vop_inactive = ncl_inactive, .vop_link = nfs_link, .vop_lock1 = nfs_lock1, .vop_lookup = nfs_lookup, .vop_mkdir = nfs_mkdir, .vop_mknod = nfs_mknod, .vop_open = nfs_open, .vop_pathconf = nfs_pathconf, .vop_print = nfs_print, .vop_read = nfs_read, .vop_readdir = nfs_readdir, .vop_readlink = nfs_readlink, .vop_reclaim = ncl_reclaim, .vop_remove = nfs_remove, .vop_rename = nfs_rename, .vop_rmdir = nfs_rmdir, .vop_setattr = nfs_setattr, .vop_strategy = nfs_strategy, .vop_symlink = nfs_symlink, .vop_write = ncl_write, .vop_getacl = nfs_getacl, .vop_setacl = nfs_setacl, }; struct vop_vector newnfs_fifoops = { .vop_default = &fifo_specops, .vop_access = nfsspec_access, .vop_close = nfsfifo_close, .vop_fsync = nfs_fsync, .vop_getattr = nfs_getattr, .vop_inactive = ncl_inactive, .vop_print = nfs_print, .vop_read = nfsfifo_read, .vop_reclaim = ncl_reclaim, .vop_setattr = nfs_setattr, .vop_write = nfsfifo_write, }; static int nfs_mknodrpc(struct vnode *dvp, struct vnode **vpp, struct componentname *cnp, struct vattr *vap); static int nfs_removerpc(struct vnode *dvp, struct vnode *vp, char *name, int namelen, struct ucred *cred, struct thread *td); static int nfs_renamerpc(struct vnode *fdvp, struct vnode *fvp, char *fnameptr, int fnamelen, struct vnode *tdvp, struct vnode *tvp, char *tnameptr, int tnamelen, struct ucred *cred, struct thread *td); static int nfs_renameit(struct vnode *sdvp, struct vnode *svp, struct componentname *scnp, struct sillyrename *sp); /* * Global variables */ #define DIRHDSIZ (sizeof (struct dirent) - (MAXNAMLEN + 1)) SYSCTL_DECL(_vfs_nfs); static int nfsaccess_cache_timeout = NFS_MAXATTRTIMO; SYSCTL_INT(_vfs_nfs, OID_AUTO, access_cache_timeout, CTLFLAG_RW, &nfsaccess_cache_timeout, 0, "NFS ACCESS cache timeout"); static int nfs_prime_access_cache = 0; SYSCTL_INT(_vfs_nfs, OID_AUTO, prime_access_cache, CTLFLAG_RW, &nfs_prime_access_cache, 0, "Prime NFS ACCESS cache when fetching attributes"); static int newnfs_commit_on_close = 0; SYSCTL_INT(_vfs_nfs, OID_AUTO, commit_on_close, CTLFLAG_RW, &newnfs_commit_on_close, 0, "write+commit on close, else only write"); static int nfs_clean_pages_on_close = 1; SYSCTL_INT(_vfs_nfs, OID_AUTO, clean_pages_on_close, CTLFLAG_RW, &nfs_clean_pages_on_close, 0, "NFS clean dirty pages on close"); int newnfs_directio_enable = 0; SYSCTL_INT(_vfs_nfs, OID_AUTO, nfs_directio_enable, CTLFLAG_RW, &newnfs_directio_enable, 0, "Enable NFS directio"); /* * This sysctl allows other processes to mmap a file that has been opened * O_DIRECT by a process. In general, having processes mmap the file while * Direct IO is in progress can lead to Data Inconsistencies. But, we allow * this by default to prevent DoS attacks - to prevent a malicious user from * opening up files O_DIRECT preventing other users from mmap'ing these * files. "Protected" environments where stricter consistency guarantees are * required can disable this knob. The process that opened the file O_DIRECT * cannot mmap() the file, because mmap'ed IO on an O_DIRECT open() is not * meaningful. */ int newnfs_directio_allow_mmap = 1; SYSCTL_INT(_vfs_nfs, OID_AUTO, nfs_directio_allow_mmap, CTLFLAG_RW, &newnfs_directio_allow_mmap, 0, "Enable mmaped IO on file with O_DIRECT opens"); #if 0 SYSCTL_INT(_vfs_nfs, OID_AUTO, access_cache_hits, CTLFLAG_RD, &newnfsstats.accesscache_hits, 0, "NFS ACCESS cache hit count"); SYSCTL_INT(_vfs_nfs, OID_AUTO, access_cache_misses, CTLFLAG_RD, &newnfsstats.accesscache_misses, 0, "NFS ACCESS cache miss count"); #endif #define NFSACCESS_ALL (NFSACCESS_READ | NFSACCESS_MODIFY \ | NFSACCESS_EXTEND | NFSACCESS_EXECUTE \ | NFSACCESS_DELETE | NFSACCESS_LOOKUP) /* * SMP Locking Note : * The list of locks after the description of the lock is the ordering * of other locks acquired with the lock held. * np->n_mtx : Protects the fields in the nfsnode. VM Object Lock VI_MTX (acquired indirectly) * nmp->nm_mtx : Protects the fields in the nfsmount. rep->r_mtx * ncl_iod_mutex : Global lock, protects shared nfsiod state. * nfs_reqq_mtx : Global lock, protects the nfs_reqq list. nmp->nm_mtx rep->r_mtx * rep->r_mtx : Protects the fields in an nfsreq. */ static int nfs34_access_otw(struct vnode *vp, int wmode, struct thread *td, struct ucred *cred, u_int32_t *retmode) { int error = 0, attrflag, i, lrupos; u_int32_t rmode; struct nfsnode *np = VTONFS(vp); struct nfsvattr nfsva; error = nfsrpc_accessrpc(vp, wmode, cred, td, &nfsva, &attrflag, &rmode, NULL); if (attrflag) (void) nfscl_loadattrcache(&vp, &nfsva, NULL, NULL, 0, 1); if (!error) { lrupos = 0; mtx_lock(&np->n_mtx); for (i = 0; i < NFS_ACCESSCACHESIZE; i++) { if (np->n_accesscache[i].uid == cred->cr_uid) { np->n_accesscache[i].mode = rmode; np->n_accesscache[i].stamp = time_second; break; } if (i > 0 && np->n_accesscache[i].stamp < np->n_accesscache[lrupos].stamp) lrupos = i; } if (i == NFS_ACCESSCACHESIZE) { np->n_accesscache[lrupos].uid = cred->cr_uid; np->n_accesscache[lrupos].mode = rmode; np->n_accesscache[lrupos].stamp = time_second; } mtx_unlock(&np->n_mtx); if (retmode != NULL) *retmode = rmode; } else if (NFS_ISV4(vp)) { error = nfscl_maperr(td, error, (uid_t)0, (gid_t)0); } return (error); } /* * nfs access vnode op. * For nfs version 2, just return ok. File accesses may fail later. * For nfs version 3, use the access rpc to check accessibility. If file modes * are changed on the server, accesses might still fail later. */ static int nfs_access(struct vop_access_args *ap) { struct vnode *vp = ap->a_vp; int error = 0, i, gotahit; u_int32_t mode, wmode, rmode; int v34 = NFS_ISV34(vp); struct nfsnode *np = VTONFS(vp); /* * Disallow write attempts on filesystems mounted read-only; * unless the file is a socket, fifo, or a block or character * device resident on the filesystem. */ if ((ap->a_accmode & (VWRITE | VAPPEND | VWRITE_NAMED_ATTRS | VDELETE_CHILD | VWRITE_ATTRIBUTES | VDELETE | VWRITE_ACL | VWRITE_OWNER)) != 0 && (vp->v_mount->mnt_flag & MNT_RDONLY) != 0) { switch (vp->v_type) { case VREG: case VDIR: case VLNK: return (EROFS); default: break; } } /* * For nfs v3 or v4, check to see if we have done this recently, and if * so return our cached result instead of making an ACCESS call. * If not, do an access rpc, otherwise you are stuck emulating * ufs_access() locally using the vattr. This may not be correct, * since the server may apply other access criteria such as * client uid-->server uid mapping that we do not know about. */ if (v34) { if (ap->a_accmode & VREAD) mode = NFSACCESS_READ; else mode = 0; if (vp->v_type != VDIR) { if (ap->a_accmode & VWRITE) mode |= (NFSACCESS_MODIFY | NFSACCESS_EXTEND); if (ap->a_accmode & VAPPEND) mode |= NFSACCESS_EXTEND; if (ap->a_accmode & VEXEC) mode |= NFSACCESS_EXECUTE; if (ap->a_accmode & VDELETE) mode |= NFSACCESS_DELETE; } else { if (ap->a_accmode & VWRITE) mode |= (NFSACCESS_MODIFY | NFSACCESS_EXTEND); if (ap->a_accmode & VAPPEND) mode |= NFSACCESS_EXTEND; if (ap->a_accmode & VEXEC) mode |= NFSACCESS_LOOKUP; if (ap->a_accmode & VDELETE) mode |= NFSACCESS_DELETE; if (ap->a_accmode & VDELETE_CHILD) mode |= NFSACCESS_MODIFY; } /* XXX safety belt, only make blanket request if caching */ if (nfsaccess_cache_timeout > 0) { wmode = NFSACCESS_READ | NFSACCESS_MODIFY | NFSACCESS_EXTEND | NFSACCESS_EXECUTE | NFSACCESS_DELETE | NFSACCESS_LOOKUP; } else { wmode = mode; } /* * Does our cached result allow us to give a definite yes to * this request? */ gotahit = 0; mtx_lock(&np->n_mtx); for (i = 0; i < NFS_ACCESSCACHESIZE; i++) { if (ap->a_cred->cr_uid == np->n_accesscache[i].uid) { if (time_second < (np->n_accesscache[i].stamp + nfsaccess_cache_timeout) && (np->n_accesscache[i].mode & mode) == mode) { NFSINCRGLOBAL(newnfsstats.accesscache_hits); gotahit = 1; } break; } } mtx_unlock(&np->n_mtx); if (gotahit == 0) { /* * Either a no, or a don't know. Go to the wire. */ NFSINCRGLOBAL(newnfsstats.accesscache_misses); error = nfs34_access_otw(vp, wmode, ap->a_td, ap->a_cred, &rmode); if (!error && (rmode & mode) != mode) error = EACCES; } return (error); } else { if ((error = nfsspec_access(ap)) != 0) { return (error); } /* * Attempt to prevent a mapped root from accessing a file * which it shouldn't. We try to read a byte from the file * if the user is root and the file is not zero length. * After calling nfsspec_access, we should have the correct * file size cached. */ mtx_lock(&np->n_mtx); if (ap->a_cred->cr_uid == 0 && (ap->a_accmode & VREAD) && VTONFS(vp)->n_size > 0) { struct iovec aiov; struct uio auio; char buf[1]; mtx_unlock(&np->n_mtx); aiov.iov_base = buf; aiov.iov_len = 1; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_offset = 0; auio.uio_resid = 1; auio.uio_segflg = UIO_SYSSPACE; auio.uio_rw = UIO_READ; auio.uio_td = ap->a_td; if (vp->v_type == VREG) error = ncl_readrpc(vp, &auio, ap->a_cred); else if (vp->v_type == VDIR) { char* bp; bp = malloc(NFS_DIRBLKSIZ, M_TEMP, M_WAITOK); aiov.iov_base = bp; aiov.iov_len = auio.uio_resid = NFS_DIRBLKSIZ; error = ncl_readdirrpc(vp, &auio, ap->a_cred, ap->a_td); free(bp, M_TEMP); } else if (vp->v_type == VLNK) error = ncl_readlinkrpc(vp, &auio, ap->a_cred); else error = EACCES; } else mtx_unlock(&np->n_mtx); return (error); } } /* * nfs open vnode op * Check to see if the type is ok * and that deletion is not in progress. * For paged in text files, you will need to flush the page cache * if consistency is lost. */ /* ARGSUSED */ static int nfs_open(struct vop_open_args *ap) { struct vnode *vp = ap->a_vp; struct nfsnode *np = VTONFS(vp); struct vattr vattr; int error; int fmode = ap->a_mode; if (vp->v_type != VREG && vp->v_type != VDIR && vp->v_type != VLNK) return (EOPNOTSUPP); /* * For NFSv4, we need to do the Open Op before cache validation, * so that we conform to RFC3530 Sec. 9.3.1. */ if (NFS_ISV4(vp)) { error = nfsrpc_open(vp, fmode, ap->a_cred, ap->a_td); if (error) { error = nfscl_maperr(ap->a_td, error, (uid_t)0, (gid_t)0); return (error); } } /* * Now, if this Open will be doing reading, re-validate/flush the * cache, so that Close/Open coherency is maintained. */ mtx_lock(&np->n_mtx); if (np->n_flag & NMODIFIED) { mtx_unlock(&np->n_mtx); error = ncl_vinvalbuf(vp, V_SAVE, ap->a_td, 1); if (error == EINTR || error == EIO) { if (NFS_ISV4(vp)) (void) nfsrpc_close(vp, 0, ap->a_td); return (error); } mtx_lock(&np->n_mtx); np->n_attrstamp = 0; if (vp->v_type == VDIR) np->n_direofoffset = 0; mtx_unlock(&np->n_mtx); error = VOP_GETATTR(vp, &vattr, ap->a_cred); if (error) { if (NFS_ISV4(vp)) (void) nfsrpc_close(vp, 0, ap->a_td); return (error); } mtx_lock(&np->n_mtx); np->n_mtime = vattr.va_mtime; if (NFS_ISV4(vp)) np->n_change = vattr.va_filerev; } else { mtx_unlock(&np->n_mtx); error = VOP_GETATTR(vp, &vattr, ap->a_cred); if (error) { if (NFS_ISV4(vp)) (void) nfsrpc_close(vp, 0, ap->a_td); return (error); } mtx_lock(&np->n_mtx); if ((NFS_ISV4(vp) && np->n_change != vattr.va_filerev) || NFS_TIMESPEC_COMPARE(&np->n_mtime, &vattr.va_mtime)) { if (vp->v_type == VDIR) np->n_direofoffset = 0; mtx_unlock(&np->n_mtx); error = ncl_vinvalbuf(vp, V_SAVE, ap->a_td, 1); if (error == EINTR || error == EIO) { if (NFS_ISV4(vp)) (void) nfsrpc_close(vp, 0, ap->a_td); return (error); } mtx_lock(&np->n_mtx); np->n_mtime = vattr.va_mtime; if (NFS_ISV4(vp)) np->n_change = vattr.va_filerev; } } /* * If the object has >= 1 O_DIRECT active opens, we disable caching. */ if (newnfs_directio_enable && (fmode & O_DIRECT) && (vp->v_type == VREG)) { if (np->n_directio_opens == 0) { mtx_unlock(&np->n_mtx); error = ncl_vinvalbuf(vp, V_SAVE, ap->a_td, 1); if (error) { if (NFS_ISV4(vp)) (void) nfsrpc_close(vp, 0, ap->a_td); return (error); } mtx_lock(&np->n_mtx); np->n_flag |= NNONCACHE; } np->n_directio_opens++; } mtx_unlock(&np->n_mtx); vnode_create_vobject(vp, vattr.va_size, ap->a_td); return (0); } /* * nfs close vnode op * What an NFS client should do upon close after writing is a debatable issue. * Most NFS clients push delayed writes to the server upon close, basically for * two reasons: * 1 - So that any write errors may be reported back to the client process * doing the close system call. By far the two most likely errors are * NFSERR_NOSPC and NFSERR_DQUOT to indicate space allocation failure. * 2 - To put a worst case upper bound on cache inconsistency between * multiple clients for the file. * There is also a consistency problem for Version 2 of the protocol w.r.t. * not being able to tell if other clients are writing a file concurrently, * since there is no way of knowing if the changed modify time in the reply * is only due to the write for this client. * (NFS Version 3 provides weak cache consistency data in the reply that * should be sufficient to detect and handle this case.) * * The current code does the following: * for NFS Version 2 - play it safe and flush/invalidate all dirty buffers * for NFS Version 3 - flush dirty buffers to the server but don't invalidate * or commit them (this satisfies 1 and 2 except for the * case where the server crashes after this close but * before the commit RPC, which is felt to be "good * enough". Changing the last argument to ncl_flush() to * a 1 would force a commit operation, if it is felt a * commit is necessary now. * for NFS Version 4 - flush the dirty buffers and commit them, if * nfscl_mustflush() says this is necessary. * It is necessary if there is no write delegation held, * in order to satisfy open/close coherency. * If the file isn't cached on local stable storage, * it may be necessary in order to detect "out of space" * errors from the server, if the write delegation * issued by the server doesn't allow the file to grow. */ /* ARGSUSED */ static int nfs_close(struct vop_close_args *ap) { struct vnode *vp = ap->a_vp; struct nfsnode *np = VTONFS(vp); struct nfsvattr nfsva; struct ucred *cred; int error = 0, ret, localcred = 0; int fmode = ap->a_fflag; if ((vp->v_mount->mnt_kern_flag & MNTK_UNMOUNTF)) return (0); /* * During shutdown, a_cred isn't valid, so just use root. */ if (ap->a_cred == NOCRED) { cred = newnfs_getcred(); localcred = 1; } else { cred = ap->a_cred; } if (vp->v_type == VREG) { /* * Examine and clean dirty pages, regardless of NMODIFIED. * This closes a major hole in close-to-open consistency. * We want to push out all dirty pages (and buffers) on * close, regardless of whether they were dirtied by * mmap'ed writes or via write(). */ if (nfs_clean_pages_on_close && vp->v_object) { VM_OBJECT_LOCK(vp->v_object); vm_object_page_clean(vp->v_object, 0, 0, 0); VM_OBJECT_UNLOCK(vp->v_object); } mtx_lock(&np->n_mtx); if (np->n_flag & NMODIFIED) { mtx_unlock(&np->n_mtx); if (NFS_ISV3(vp)) { /* * Under NFSv3 we have dirty buffers to dispose of. We * must flush them to the NFS server. We have the option * of waiting all the way through the commit rpc or just * waiting for the initial write. The default is to only * wait through the initial write so the data is in the * server's cache, which is roughly similar to the state * a standard disk subsystem leaves the file in on close(). * * We cannot clear the NMODIFIED bit in np->n_flag due to * potential races with other processes, and certainly * cannot clear it if we don't commit. * These races occur when there is no longer the old * traditional vnode locking implemented for Vnode Ops. */ int cm = newnfs_commit_on_close ? 1 : 0; error = ncl_flush(vp, MNT_WAIT, cred, ap->a_td, cm, 0); /* np->n_flag &= ~NMODIFIED; */ } else if (NFS_ISV4(vp)) { if (nfscl_mustflush(vp) != 0) { int cm = newnfs_commit_on_close ? 1 : 0; error = ncl_flush(vp, MNT_WAIT, cred, ap->a_td, cm, 0); /* * as above w.r.t races when clearing * NMODIFIED. * np->n_flag &= ~NMODIFIED; */ } } else error = ncl_vinvalbuf(vp, V_SAVE, ap->a_td, 1); mtx_lock(&np->n_mtx); } /* * Invalidate the attribute cache in all cases. * An open is going to fetch fresh attrs any way, other procs * on this node that have file open will be forced to do an * otw attr fetch, but this is safe. * --> A user found that their RPC count dropped by 20% when * this was commented out and I can't see any requirement * for it, so I've disabled it when negative lookups are * enabled. (What does this have to do with negative lookup * caching? Well nothing, except it was reported by the * same user that needed negative lookup caching and I wanted * there to be a way to disable it to see if it * is the cause of some caching/coherency issue that might * crop up.) */ if (VFSTONFS(vp->v_mount)->nm_negnametimeo == 0) np->n_attrstamp = 0; if (np->n_flag & NWRITEERR) { np->n_flag &= ~NWRITEERR; error = np->n_error; } mtx_unlock(&np->n_mtx); } if (NFS_ISV4(vp)) { /* * Get attributes so "change" is up to date. */ if (error == 0 && nfscl_mustflush(vp) != 0) { ret = nfsrpc_getattr(vp, cred, ap->a_td, &nfsva, NULL); if (!ret) { np->n_change = nfsva.na_filerev; (void) nfscl_loadattrcache(&vp, &nfsva, NULL, NULL, 0, 0); } } /* * and do the close. */ ret = nfsrpc_close(vp, 0, ap->a_td); if (!error && ret) error = ret; if (error) error = nfscl_maperr(ap->a_td, error, (uid_t)0, (gid_t)0); } if (newnfs_directio_enable) KASSERT((np->n_directio_asyncwr == 0), ("nfs_close: dirty unflushed (%d) directio buffers\n", np->n_directio_asyncwr)); if (newnfs_directio_enable && (fmode & O_DIRECT) && (vp->v_type == VREG)) { mtx_lock(&np->n_mtx); KASSERT((np->n_directio_opens > 0), ("nfs_close: unexpectedly value (0) of n_directio_opens\n")); np->n_directio_opens--; if (np->n_directio_opens == 0) np->n_flag &= ~NNONCACHE; mtx_unlock(&np->n_mtx); } if (localcred) NFSFREECRED(cred); return (error); } /* * nfs getattr call from vfs. */ static int nfs_getattr(struct vop_getattr_args *ap) { struct vnode *vp = ap->a_vp; struct thread *td = curthread; /* XXX */ struct nfsnode *np = VTONFS(vp); int error = 0; struct nfsvattr nfsva; struct vattr *vap = ap->a_vap; struct vattr vattr; /* * Update local times for special files. */ mtx_lock(&np->n_mtx); if (np->n_flag & (NACC | NUPD)) np->n_flag |= NCHG; mtx_unlock(&np->n_mtx); /* * First look in the cache. */ if (ncl_getattrcache(vp, &vattr) == 0) { vap->va_type = vattr.va_type; vap->va_mode = vattr.va_mode; vap->va_nlink = vattr.va_nlink; vap->va_uid = vattr.va_uid; vap->va_gid = vattr.va_gid; vap->va_fsid = vattr.va_fsid; vap->va_fileid = vattr.va_fileid; vap->va_size = vattr.va_size; vap->va_blocksize = vattr.va_blocksize; vap->va_atime = vattr.va_atime; vap->va_mtime = vattr.va_mtime; vap->va_ctime = vattr.va_ctime; vap->va_gen = vattr.va_gen; vap->va_flags = vattr.va_flags; vap->va_rdev = vattr.va_rdev; vap->va_bytes = vattr.va_bytes; vap->va_filerev = vattr.va_filerev; /* * Get the local modify time for the case of a write * delegation. */ nfscl_deleggetmodtime(vp, &vap->va_mtime); return (0); } if (NFS_ISV34(vp) && nfs_prime_access_cache && nfsaccess_cache_timeout > 0) { NFSINCRGLOBAL(newnfsstats.accesscache_misses); nfs34_access_otw(vp, NFSACCESS_ALL, td, ap->a_cred, NULL); if (ncl_getattrcache(vp, ap->a_vap) == 0) { nfscl_deleggetmodtime(vp, &ap->a_vap->va_mtime); return (0); } } error = nfsrpc_getattr(vp, ap->a_cred, td, &nfsva, NULL); if (!error) error = nfscl_loadattrcache(&vp, &nfsva, vap, NULL, 0, 0); if (!error) { /* * Get the local modify time for the case of a write * delegation. */ nfscl_deleggetmodtime(vp, &vap->va_mtime); } else if (NFS_ISV4(vp)) { error = nfscl_maperr(td, error, (uid_t)0, (gid_t)0); } return (error); } /* * nfs setattr call. */ static int nfs_setattr(struct vop_setattr_args *ap) { struct vnode *vp = ap->a_vp; struct nfsnode *np = VTONFS(vp); struct thread *td = curthread; /* XXX */ struct vattr *vap = ap->a_vap; int error = 0; u_quad_t tsize; #ifndef nolint tsize = (u_quad_t)0; #endif /* * Setting of flags and marking of atimes are not supported. */ if (vap->va_flags != VNOVAL) return (EOPNOTSUPP); /* * Disallow write attempts if the filesystem is mounted read-only. */ if ((vap->va_flags != VNOVAL || vap->va_uid != (uid_t)VNOVAL || vap->va_gid != (gid_t)VNOVAL || vap->va_atime.tv_sec != VNOVAL || vap->va_mtime.tv_sec != VNOVAL || vap->va_mode != (mode_t)VNOVAL) && (vp->v_mount->mnt_flag & MNT_RDONLY)) return (EROFS); if (vap->va_size != VNOVAL) { switch (vp->v_type) { case VDIR: return (EISDIR); case VCHR: case VBLK: case VSOCK: case VFIFO: if (vap->va_mtime.tv_sec == VNOVAL && vap->va_atime.tv_sec == VNOVAL && vap->va_mode == (mode_t)VNOVAL && vap->va_uid == (uid_t)VNOVAL && vap->va_gid == (gid_t)VNOVAL) return (0); vap->va_size = VNOVAL; break; default: /* * Disallow write attempts if the filesystem is * mounted read-only. */ if (vp->v_mount->mnt_flag & MNT_RDONLY) return (EROFS); /* * We run vnode_pager_setsize() early (why?), * we must set np->n_size now to avoid vinvalbuf * V_SAVE races that might setsize a lower * value. */ mtx_lock(&np->n_mtx); tsize = np->n_size; mtx_unlock(&np->n_mtx); error = ncl_meta_setsize(vp, ap->a_cred, td, vap->va_size); mtx_lock(&np->n_mtx); if (np->n_flag & NMODIFIED) { tsize = np->n_size; mtx_unlock(&np->n_mtx); if (vap->va_size == 0) error = ncl_vinvalbuf(vp, 0, td, 1); else error = ncl_vinvalbuf(vp, V_SAVE, td, 1); if (error) { vnode_pager_setsize(vp, tsize); return (error); } /* * Call nfscl_delegmodtime() to set the modify time * locally, as required. */ nfscl_delegmodtime(vp); } else mtx_unlock(&np->n_mtx); /* * np->n_size has already been set to vap->va_size * in ncl_meta_setsize(). We must set it again since * nfs_loadattrcache() could be called through * ncl_meta_setsize() and could modify np->n_size. */ mtx_lock(&np->n_mtx); np->n_vattr.na_size = np->n_size = vap->va_size; mtx_unlock(&np->n_mtx); }; } else { mtx_lock(&np->n_mtx); if ((vap->va_mtime.tv_sec != VNOVAL || vap->va_atime.tv_sec != VNOVAL) && (np->n_flag & NMODIFIED) && vp->v_type == VREG) { mtx_unlock(&np->n_mtx); if ((error = ncl_vinvalbuf(vp, V_SAVE, td, 1)) != 0 && (error == EINTR || error == EIO)) return (error); } else mtx_unlock(&np->n_mtx); } error = nfs_setattrrpc(vp, vap, ap->a_cred, td); if (error && vap->va_size != VNOVAL) { mtx_lock(&np->n_mtx); np->n_size = np->n_vattr.na_size = tsize; vnode_pager_setsize(vp, tsize); mtx_unlock(&np->n_mtx); } return (error); } /* * Do an nfs setattr rpc. */ static int nfs_setattrrpc(struct vnode *vp, struct vattr *vap, struct ucred *cred, struct thread *td) { struct nfsnode *np = VTONFS(vp); int error, ret, attrflag, i; struct nfsvattr nfsva; if (NFS_ISV34(vp)) { mtx_lock(&np->n_mtx); for (i = 0; i < NFS_ACCESSCACHESIZE; i++) np->n_accesscache[i].stamp = 0; np->n_flag |= NDELEGMOD; mtx_unlock(&np->n_mtx); } error = nfsrpc_setattr(vp, vap, NULL, cred, td, &nfsva, &attrflag, NULL); if (attrflag) { ret = nfscl_loadattrcache(&vp, &nfsva, NULL, NULL, 0, 1); if (ret && !error) error = ret; } if (error && NFS_ISV4(vp)) error = nfscl_maperr(td, error, vap->va_uid, vap->va_gid); return (error); } /* * nfs lookup call, one step at a time... * First look in cache * If not found, unlock the directory nfsnode and do the rpc */ static int nfs_lookup(struct vop_lookup_args *ap) { struct componentname *cnp = ap->a_cnp; struct vnode *dvp = ap->a_dvp; struct vnode **vpp = ap->a_vpp; struct mount *mp = dvp->v_mount; int flags = cnp->cn_flags; struct vnode *newvp; struct nfsmount *nmp; struct nfsnode *np, *newnp; int error = 0, attrflag, dattrflag, ltype; struct thread *td = cnp->cn_thread; struct nfsfh *nfhp; struct nfsvattr dnfsva, nfsva; struct vattr vattr; struct timespec dmtime; *vpp = NULLVP; if ((flags & ISLASTCN) && (mp->mnt_flag & MNT_RDONLY) && (cnp->cn_nameiop == DELETE || cnp->cn_nameiop == RENAME)) return (EROFS); if (dvp->v_type != VDIR) return (ENOTDIR); nmp = VFSTONFS(mp); np = VTONFS(dvp); /* For NFSv4, wait until any remove is done. */ mtx_lock(&np->n_mtx); while (NFSHASNFSV4(nmp) && (np->n_flag & NREMOVEINPROG)) { np->n_flag |= NREMOVEWANT; (void) msleep((caddr_t)np, &np->n_mtx, PZERO, "nfslkup", 0); } mtx_unlock(&np->n_mtx); if ((error = VOP_ACCESS(dvp, VEXEC, cnp->cn_cred, td)) != 0) return (error); error = cache_lookup(dvp, vpp, cnp); if (error > 0 && error != ENOENT) return (error); if (error == -1) { /* * We only accept a positive hit in the cache if the * change time of the file matches our cached copy. * Otherwise, we discard the cache entry and fallback * to doing a lookup RPC. * * To better handle stale file handles and attributes, * clear the attribute cache of this node if it is a * leaf component, part of an open() call, and not * locally modified before fetching the attributes. * This should allow stale file handles to be detected * here where we can fall back to a LOOKUP RPC to * recover rather than having nfs_open() detect the * stale file handle and failing open(2) with ESTALE. */ newvp = *vpp; newnp = VTONFS(newvp); if (!(nmp->nm_flag & NFSMNT_NOCTO) && (flags & (ISLASTCN | ISOPEN)) == (ISLASTCN | ISOPEN) && !(newnp->n_flag & NMODIFIED)) { mtx_lock(&newnp->n_mtx); newnp->n_attrstamp = 0; mtx_unlock(&newnp->n_mtx); } if (nfscl_nodeleg(newvp, 0) == 0 || (VOP_GETATTR(newvp, &vattr, cnp->cn_cred) == 0 && timespeccmp(&vattr.va_ctime, &newnp->n_ctime, ==))) { NFSINCRGLOBAL(newnfsstats.lookupcache_hits); if (cnp->cn_nameiop != LOOKUP && (flags & ISLASTCN)) cnp->cn_flags |= SAVENAME; return (0); } cache_purge(newvp); if (dvp != newvp) vput(newvp); else vrele(newvp); *vpp = NULLVP; } else if (error == ENOENT) { if (dvp->v_iflag & VI_DOOMED) return (ENOENT); /* * We only accept a negative hit in the cache if the * modification time of the parent directory matches * our cached copy. Otherwise, we discard all of the * negative cache entries for this directory. We also * only trust -ve cache entries for less than * nm_negative_namecache_timeout seconds. */ if ((u_int)(ticks - np->n_dmtime_ticks) < (nmp->nm_negnametimeo * hz) && VOP_GETATTR(dvp, &vattr, cnp->cn_cred) == 0 && timespeccmp(&vattr.va_mtime, &np->n_dmtime, ==)) { NFSINCRGLOBAL(newnfsstats.lookupcache_hits); return (ENOENT); } cache_purge_negative(dvp); mtx_lock(&np->n_mtx); timespecclear(&np->n_dmtime); mtx_unlock(&np->n_mtx); } /* * Cache the modification time of the parent directory in case * the lookup fails and results in adding the first negative * name cache entry for the directory. Since this is reading * a single time_t, don't bother with locking. The * modification time may be a bit stale, but it must be read * before performing the lookup RPC to prevent a race where * another lookup updates the timestamp on the directory after * the lookup RPC has been performed on the server but before * n_dmtime is set at the end of this function. */ dmtime = np->n_vattr.na_mtime; error = 0; newvp = NULLVP; NFSINCRGLOBAL(newnfsstats.lookupcache_misses); error = nfsrpc_lookup(dvp, cnp->cn_nameptr, cnp->cn_namelen, cnp->cn_cred, td, &dnfsva, &nfsva, &nfhp, &attrflag, &dattrflag, NULL); if (dattrflag) (void) nfscl_loadattrcache(&dvp, &dnfsva, NULL, NULL, 0, 1); if (error) { if (newvp != NULLVP) { vput(newvp); *vpp = NULLVP; } if (error != ENOENT) { if (NFS_ISV4(dvp)) error = nfscl_maperr(td, error, (uid_t)0, (gid_t)0); return (error); } /* The requested file was not found. */ if ((cnp->cn_nameiop == CREATE || cnp->cn_nameiop == RENAME) && (flags & ISLASTCN)) { /* * XXX: UFS does a full VOP_ACCESS(dvp, * VWRITE) here instead of just checking * MNT_RDONLY. */ if (mp->mnt_flag & MNT_RDONLY) return (EROFS); cnp->cn_flags |= SAVENAME; return (EJUSTRETURN); } if ((cnp->cn_flags & MAKEENTRY) && cnp->cn_nameiop != CREATE) { /* * Maintain n_dmtime as the modification time * of the parent directory when the oldest -ve * name cache entry for this directory was * added. If a -ve cache entry has already * been added with a newer modification time * by a concurrent lookup, then don't bother * adding a cache entry. The modification * time of the directory might have changed * due to the file this lookup failed to find * being created. In that case a subsequent * lookup would incorrectly use the entry * added here instead of doing an extra * lookup. */ mtx_lock(&np->n_mtx); if (timespeccmp(&np->n_dmtime, &dmtime, <=)) { if (!timespecisset(&np->n_dmtime)) { np->n_dmtime = dmtime; np->n_dmtime_ticks = ticks; } mtx_unlock(&np->n_mtx); cache_enter(dvp, NULL, cnp); } else mtx_unlock(&np->n_mtx); } return (ENOENT); } /* * Handle RENAME case... */ if (cnp->cn_nameiop == RENAME && (flags & ISLASTCN)) { if (NFS_CMPFH(np, nfhp->nfh_fh, nfhp->nfh_len)) { FREE((caddr_t)nfhp, M_NFSFH); return (EISDIR); } error = nfscl_nget(mp, dvp, nfhp, cnp, td, &np, NULL, LK_EXCLUSIVE); if (error) return (error); newvp = NFSTOV(np); if (attrflag) (void) nfscl_loadattrcache(&newvp, &nfsva, NULL, NULL, 0, 1); *vpp = newvp; cnp->cn_flags |= SAVENAME; return (0); } if (flags & ISDOTDOT) { ltype = VOP_ISLOCKED(dvp); error = vfs_busy(mp, MBF_NOWAIT); if (error != 0) { vfs_ref(mp); VOP_UNLOCK(dvp, 0); error = vfs_busy(mp, 0); vn_lock(dvp, ltype | LK_RETRY); vfs_rel(mp); if (error == 0 && (dvp->v_iflag & VI_DOOMED)) { vfs_unbusy(mp); error = ENOENT; } if (error != 0) return (error); } VOP_UNLOCK(dvp, 0); error = nfscl_nget(mp, dvp, nfhp, cnp, td, &np, NULL, cnp->cn_lkflags); if (error == 0) newvp = NFSTOV(np); vfs_unbusy(mp); if (newvp != dvp) vn_lock(dvp, ltype | LK_RETRY); if (dvp->v_iflag & VI_DOOMED) { if (error == 0) { if (newvp == dvp) vrele(newvp); else vput(newvp); } error = ENOENT; } if (error != 0) return (error); if (attrflag) (void) nfscl_loadattrcache(&newvp, &nfsva, NULL, NULL, 0, 1); } else if (NFS_CMPFH(np, nfhp->nfh_fh, nfhp->nfh_len)) { FREE((caddr_t)nfhp, M_NFSFH); VREF(dvp); newvp = dvp; if (attrflag) (void) nfscl_loadattrcache(&newvp, &nfsva, NULL, NULL, 0, 1); } else { error = nfscl_nget(mp, dvp, nfhp, cnp, td, &np, NULL, cnp->cn_lkflags); if (error) return (error); newvp = NFSTOV(np); if (attrflag) (void) nfscl_loadattrcache(&newvp, &nfsva, NULL, NULL, 0, 1); else if ((flags & (ISLASTCN | ISOPEN)) == (ISLASTCN | ISOPEN) && !(np->n_flag & NMODIFIED)) { /* * Flush the attribute cache when opening a * leaf node to ensure that fresh attributes * are fetched in nfs_open() since we did not * fetch attributes from the LOOKUP reply. */ mtx_lock(&np->n_mtx); np->n_attrstamp = 0; mtx_unlock(&np->n_mtx); } } if (cnp->cn_nameiop != LOOKUP && (flags & ISLASTCN)) cnp->cn_flags |= SAVENAME; if ((cnp->cn_flags & MAKEENTRY) && (cnp->cn_nameiop != DELETE || !(flags & ISLASTCN))) { np->n_ctime = np->n_vattr.na_vattr.va_ctime; cache_enter(dvp, newvp, cnp); } *vpp = newvp; return (0); } /* * nfs read call. * Just call ncl_bioread() to do the work. */ static int nfs_read(struct vop_read_args *ap) { struct vnode *vp = ap->a_vp; switch (vp->v_type) { case VREG: return (ncl_bioread(vp, ap->a_uio, ap->a_ioflag, ap->a_cred)); case VDIR: return (EISDIR); default: return (EOPNOTSUPP); } } /* * nfs readlink call */ static int nfs_readlink(struct vop_readlink_args *ap) { struct vnode *vp = ap->a_vp; if (vp->v_type != VLNK) return (EINVAL); return (ncl_bioread(vp, ap->a_uio, 0, ap->a_cred)); } /* * Do a readlink rpc. * Called by ncl_doio() from below the buffer cache. */ int ncl_readlinkrpc(struct vnode *vp, struct uio *uiop, struct ucred *cred) { int error, ret, attrflag; struct nfsvattr nfsva; error = nfsrpc_readlink(vp, uiop, cred, uiop->uio_td, &nfsva, &attrflag, NULL); if (attrflag) { ret = nfscl_loadattrcache(&vp, &nfsva, NULL, NULL, 0, 1); if (ret && !error) error = ret; } if (error && NFS_ISV4(vp)) error = nfscl_maperr(uiop->uio_td, error, (uid_t)0, (gid_t)0); return (error); } /* * nfs read rpc call * Ditto above */ int ncl_readrpc(struct vnode *vp, struct uio *uiop, struct ucred *cred) { int error, ret, attrflag; struct nfsvattr nfsva; error = nfsrpc_read(vp, uiop, cred, uiop->uio_td, &nfsva, &attrflag, NULL); if (attrflag) { ret = nfscl_loadattrcache(&vp, &nfsva, NULL, NULL, 0, 1); if (ret && !error) error = ret; } if (error && NFS_ISV4(vp)) error = nfscl_maperr(uiop->uio_td, error, (uid_t)0, (gid_t)0); return (error); } /* * nfs write call */ int ncl_writerpc(struct vnode *vp, struct uio *uiop, struct ucred *cred, int *iomode, int *must_commit, int called_from_strategy) { struct nfsvattr nfsva; int error = 0, attrflag, ret; error = nfsrpc_write(vp, uiop, iomode, must_commit, cred, uiop->uio_td, &nfsva, &attrflag, NULL, called_from_strategy); if (attrflag) { if (VTONFS(vp)->n_flag & ND_NFSV4) ret = nfscl_loadattrcache(&vp, &nfsva, NULL, NULL, 1, 1); else ret = nfscl_loadattrcache(&vp, &nfsva, NULL, NULL, 0, 1); if (ret && !error) error = ret; } if (vp->v_mount->mnt_kern_flag & MNTK_ASYNC) *iomode = NFSWRITE_FILESYNC; if (error && NFS_ISV4(vp)) error = nfscl_maperr(uiop->uio_td, error, (uid_t)0, (gid_t)0); return (error); } /* * nfs mknod rpc * For NFS v2 this is a kludge. Use a create rpc but with the IFMT bits of the * mode set to specify the file type and the size field for rdev. */ static int nfs_mknodrpc(struct vnode *dvp, struct vnode **vpp, struct componentname *cnp, struct vattr *vap) { struct nfsvattr nfsva, dnfsva; struct vnode *newvp = NULL; struct nfsnode *np = NULL, *dnp; struct nfsfh *nfhp; struct vattr vattr; int error = 0, attrflag, dattrflag; u_int32_t rdev; if (vap->va_type == VCHR || vap->va_type == VBLK) rdev = vap->va_rdev; else if (vap->va_type == VFIFO || vap->va_type == VSOCK) rdev = 0xffffffff; else return (EOPNOTSUPP); if ((error = VOP_GETATTR(dvp, &vattr, cnp->cn_cred))) return (error); error = nfsrpc_mknod(dvp, cnp->cn_nameptr, cnp->cn_namelen, vap, rdev, vap->va_type, cnp->cn_cred, cnp->cn_thread, &dnfsva, &nfsva, &nfhp, &attrflag, &dattrflag, NULL); if (!error) { if (!nfhp) (void) nfsrpc_lookup(dvp, cnp->cn_nameptr, cnp->cn_namelen, cnp->cn_cred, cnp->cn_thread, &dnfsva, &nfsva, &nfhp, &attrflag, &dattrflag, NULL); if (nfhp) error = nfscl_nget(dvp->v_mount, dvp, nfhp, cnp, cnp->cn_thread, &np, NULL, LK_EXCLUSIVE); } if (dattrflag) (void) nfscl_loadattrcache(&dvp, &dnfsva, NULL, NULL, 0, 1); if (!error) { newvp = NFSTOV(np); if (attrflag != 0) { error = nfscl_loadattrcache(&newvp, &nfsva, NULL, NULL, 0, 1); if (error != 0) vput(newvp); } } if (!error) { if ((cnp->cn_flags & MAKEENTRY)) cache_enter(dvp, newvp, cnp); *vpp = newvp; } else if (NFS_ISV4(dvp)) { error = nfscl_maperr(cnp->cn_thread, error, vap->va_uid, vap->va_gid); } dnp = VTONFS(dvp); mtx_lock(&dnp->n_mtx); dnp->n_flag |= NMODIFIED; if (!dattrflag) dnp->n_attrstamp = 0; mtx_unlock(&dnp->n_mtx); return (error); } /* * nfs mknod vop * just call nfs_mknodrpc() to do the work. */ /* ARGSUSED */ static int nfs_mknod(struct vop_mknod_args *ap) { return (nfs_mknodrpc(ap->a_dvp, ap->a_vpp, ap->a_cnp, ap->a_vap)); } static struct mtx nfs_cverf_mtx; MTX_SYSINIT(nfs_cverf_mtx, &nfs_cverf_mtx, "NFS create verifier mutex", MTX_DEF); static nfsquad_t nfs_get_cverf(void) { static nfsquad_t cverf; nfsquad_t ret; static int cverf_initialized = 0; mtx_lock(&nfs_cverf_mtx); if (cverf_initialized == 0) { cverf.lval[0] = arc4random(); cverf.lval[1] = arc4random(); cverf_initialized = 1; } else cverf.qval++; ret = cverf; mtx_unlock(&nfs_cverf_mtx); return (ret); } /* * nfs file create call */ static int nfs_create(struct vop_create_args *ap) { struct vnode *dvp = ap->a_dvp; struct vattr *vap = ap->a_vap; struct componentname *cnp = ap->a_cnp; struct nfsnode *np = NULL, *dnp; struct vnode *newvp = NULL; struct nfsmount *nmp; struct nfsvattr dnfsva, nfsva; struct nfsfh *nfhp; nfsquad_t cverf; int error = 0, attrflag, dattrflag, fmode = 0; struct vattr vattr; /* * Oops, not for me.. */ if (vap->va_type == VSOCK) return (nfs_mknodrpc(dvp, ap->a_vpp, cnp, vap)); if ((error = VOP_GETATTR(dvp, &vattr, cnp->cn_cred))) return (error); if (vap->va_vaflags & VA_EXCLUSIVE) fmode |= O_EXCL; dnp = VTONFS(dvp); nmp = VFSTONFS(vnode_mount(dvp)); again: /* For NFSv4, wait until any remove is done. */ mtx_lock(&dnp->n_mtx); while (NFSHASNFSV4(nmp) && (dnp->n_flag & NREMOVEINPROG)) { dnp->n_flag |= NREMOVEWANT; (void) msleep((caddr_t)dnp, &dnp->n_mtx, PZERO, "nfscrt", 0); } mtx_unlock(&dnp->n_mtx); cverf = nfs_get_cverf(); error = nfsrpc_create(dvp, cnp->cn_nameptr, cnp->cn_namelen, vap, cverf, fmode, cnp->cn_cred, cnp->cn_thread, &dnfsva, &nfsva, &nfhp, &attrflag, &dattrflag, NULL); if (!error) { if (nfhp == NULL) (void) nfsrpc_lookup(dvp, cnp->cn_nameptr, cnp->cn_namelen, cnp->cn_cred, cnp->cn_thread, &dnfsva, &nfsva, &nfhp, &attrflag, &dattrflag, NULL); if (nfhp != NULL) error = nfscl_nget(dvp->v_mount, dvp, nfhp, cnp, cnp->cn_thread, &np, NULL, LK_EXCLUSIVE); } if (dattrflag) (void) nfscl_loadattrcache(&dvp, &dnfsva, NULL, NULL, 0, 1); if (!error) { newvp = NFSTOV(np); if (attrflag) error = nfscl_loadattrcache(&newvp, &nfsva, NULL, NULL, 0, 1); } if (error) { if (newvp != NULL) { vput(newvp); newvp = NULL; } if (NFS_ISV34(dvp) && (fmode & O_EXCL) && error == NFSERR_NOTSUPP) { fmode &= ~O_EXCL; goto again; } } else if (NFS_ISV34(dvp) && (fmode & O_EXCL)) { if (nfscl_checksattr(vap, &nfsva)) { /* * We are normally called with only a partially * initialized VAP. Since the NFSv3 spec says that * the server may use the file attributes to * store the verifier, the spec requires us to do a * SETATTR RPC. FreeBSD servers store the verifier in * atime, but we can't really assume that all servers * will so we ensure that our SETATTR sets both atime * and mtime. */ if (vap->va_mtime.tv_sec == VNOVAL) vfs_timestamp(&vap->va_mtime); if (vap->va_atime.tv_sec == VNOVAL) vap->va_atime = vap->va_mtime; error = nfsrpc_setattr(newvp, vap, NULL, cnp->cn_cred, cnp->cn_thread, &nfsva, &attrflag, NULL); if (error && (vap->va_uid != (uid_t)VNOVAL || vap->va_gid != (gid_t)VNOVAL)) { /* try again without setting uid/gid */ vap->va_uid = (uid_t)VNOVAL; vap->va_gid = (uid_t)VNOVAL; error = nfsrpc_setattr(newvp, vap, NULL, cnp->cn_cred, cnp->cn_thread, &nfsva, &attrflag, NULL); } if (attrflag) (void) nfscl_loadattrcache(&newvp, &nfsva, NULL, NULL, 0, 1); } } if (!error) { if (cnp->cn_flags & MAKEENTRY) cache_enter(dvp, newvp, cnp); *ap->a_vpp = newvp; } else if (NFS_ISV4(dvp)) { error = nfscl_maperr(cnp->cn_thread, error, vap->va_uid, vap->va_gid); } mtx_lock(&dnp->n_mtx); dnp->n_flag |= NMODIFIED; if (!dattrflag) dnp->n_attrstamp = 0; mtx_unlock(&dnp->n_mtx); return (error); } /* * nfs file remove call * To try and make nfs semantics closer to ufs semantics, a file that has * other processes using the vnode is renamed instead of removed and then * removed later on the last close. * - If v_usecount > 1 * If a rename is not already in the works * call nfs_sillyrename() to set it up * else * do the remove rpc */ static int nfs_remove(struct vop_remove_args *ap) { struct vnode *vp = ap->a_vp; struct vnode *dvp = ap->a_dvp; struct componentname *cnp = ap->a_cnp; struct nfsnode *np = VTONFS(vp); int error = 0; struct vattr vattr; KASSERT((cnp->cn_flags & HASBUF) != 0, ("nfs_remove: no name")); KASSERT(vrefcnt(vp) > 0, ("nfs_remove: bad v_usecount")); if (vp->v_type == VDIR) error = EPERM; else if (vrefcnt(vp) == 1 || (np->n_sillyrename && VOP_GETATTR(vp, &vattr, cnp->cn_cred) == 0 && vattr.va_nlink > 1)) { /* * Purge the name cache so that the chance of a lookup for * the name succeeding while the remove is in progress is * minimized. Without node locking it can still happen, such * that an I/O op returns ESTALE, but since you get this if * another host removes the file.. */ cache_purge(vp); /* * throw away biocache buffers, mainly to avoid * unnecessary delayed writes later. */ error = ncl_vinvalbuf(vp, 0, cnp->cn_thread, 1); /* Do the rpc */ if (error != EINTR && error != EIO) error = nfs_removerpc(dvp, vp, cnp->cn_nameptr, cnp->cn_namelen, cnp->cn_cred, cnp->cn_thread); /* * Kludge City: If the first reply to the remove rpc is lost.. * the reply to the retransmitted request will be ENOENT * since the file was in fact removed * Therefore, we cheat and return success. */ if (error == ENOENT) error = 0; } else if (!np->n_sillyrename) error = nfs_sillyrename(dvp, vp, cnp); mtx_lock(&np->n_mtx); np->n_attrstamp = 0; mtx_unlock(&np->n_mtx); return (error); } /* * nfs file remove rpc called from nfs_inactive */ int ncl_removeit(struct sillyrename *sp, struct vnode *vp) { /* * Make sure that the directory vnode is still valid. * XXX we should lock sp->s_dvp here. */ if (sp->s_dvp->v_type == VBAD) return (0); return (nfs_removerpc(sp->s_dvp, vp, sp->s_name, sp->s_namlen, sp->s_cred, NULL)); } /* * Nfs remove rpc, called from nfs_remove() and ncl_removeit(). */ static int nfs_removerpc(struct vnode *dvp, struct vnode *vp, char *name, int namelen, struct ucred *cred, struct thread *td) { struct nfsvattr dnfsva; struct nfsnode *dnp = VTONFS(dvp); int error = 0, dattrflag; mtx_lock(&dnp->n_mtx); dnp->n_flag |= NREMOVEINPROG; mtx_unlock(&dnp->n_mtx); error = nfsrpc_remove(dvp, name, namelen, vp, cred, td, &dnfsva, &dattrflag, NULL); mtx_lock(&dnp->n_mtx); if ((dnp->n_flag & NREMOVEWANT)) { dnp->n_flag &= ~(NREMOVEWANT | NREMOVEINPROG); mtx_unlock(&dnp->n_mtx); wakeup((caddr_t)dnp); } else { dnp->n_flag &= ~NREMOVEINPROG; mtx_unlock(&dnp->n_mtx); } if (dattrflag) (void) nfscl_loadattrcache(&dvp, &dnfsva, NULL, NULL, 0, 1); mtx_lock(&dnp->n_mtx); dnp->n_flag |= NMODIFIED; if (!dattrflag) dnp->n_attrstamp = 0; mtx_unlock(&dnp->n_mtx); if (error && NFS_ISV4(dvp)) error = nfscl_maperr(td, error, (uid_t)0, (gid_t)0); return (error); } /* * nfs file rename call */ static int nfs_rename(struct vop_rename_args *ap) { struct vnode *fvp = ap->a_fvp; struct vnode *tvp = ap->a_tvp; struct vnode *fdvp = ap->a_fdvp; struct vnode *tdvp = ap->a_tdvp; struct componentname *tcnp = ap->a_tcnp; struct componentname *fcnp = ap->a_fcnp; struct nfsnode *fnp = VTONFS(ap->a_fvp); struct nfsnode *tdnp = VTONFS(ap->a_tdvp); struct nfsv4node *newv4 = NULL; int error; KASSERT((tcnp->cn_flags & HASBUF) != 0 && (fcnp->cn_flags & HASBUF) != 0, ("nfs_rename: no name")); /* Check for cross-device rename */ if ((fvp->v_mount != tdvp->v_mount) || (tvp && (fvp->v_mount != tvp->v_mount))) { error = EXDEV; goto out; } if (fvp == tvp) { ncl_printf("nfs_rename: fvp == tvp (can't happen)\n"); error = 0; goto out; } if ((error = vn_lock(fvp, LK_EXCLUSIVE)) != 0) goto out; /* * We have to flush B_DELWRI data prior to renaming * the file. If we don't, the delayed-write buffers * can be flushed out later after the file has gone stale * under NFSV3. NFSV2 does not have this problem because * ( as far as I can tell ) it flushes dirty buffers more * often. * * Skip the rename operation if the fsync fails, this can happen * due to the server's volume being full, when we pushed out data * that was written back to our cache earlier. Not checking for * this condition can result in potential (silent) data loss. */ error = VOP_FSYNC(fvp, MNT_WAIT, fcnp->cn_thread); VOP_UNLOCK(fvp, 0); if (!error && tvp) error = VOP_FSYNC(tvp, MNT_WAIT, tcnp->cn_thread); if (error) goto out; /* * If the tvp exists and is in use, sillyrename it before doing the * rename of the new file over it. * XXX Can't sillyrename a directory. */ if (tvp && vrefcnt(tvp) > 1 && !VTONFS(tvp)->n_sillyrename && tvp->v_type != VDIR && !nfs_sillyrename(tdvp, tvp, tcnp)) { vput(tvp); tvp = NULL; } error = nfs_renamerpc(fdvp, fvp, fcnp->cn_nameptr, fcnp->cn_namelen, tdvp, tvp, tcnp->cn_nameptr, tcnp->cn_namelen, tcnp->cn_cred, tcnp->cn_thread); if (error == 0 && NFS_ISV4(tdvp)) { /* * For NFSv4, check to see if it is the same name and * replace the name, if it is different. */ MALLOC(newv4, struct nfsv4node *, sizeof (struct nfsv4node) + tdnp->n_fhp->nfh_len + tcnp->cn_namelen - 1, M_NFSV4NODE, M_WAITOK); mtx_lock(&tdnp->n_mtx); mtx_lock(&fnp->n_mtx); if (fnp->n_v4 != NULL && fvp->v_type == VREG && (fnp->n_v4->n4_namelen != tcnp->cn_namelen || NFSBCMP(tcnp->cn_nameptr, NFS4NODENAME(fnp->n_v4), tcnp->cn_namelen) || tdnp->n_fhp->nfh_len != fnp->n_v4->n4_fhlen || NFSBCMP(tdnp->n_fhp->nfh_fh, fnp->n_v4->n4_data, tdnp->n_fhp->nfh_len))) { #ifdef notdef { char nnn[100]; int nnnl; nnnl = (tcnp->cn_namelen < 100) ? tcnp->cn_namelen : 99; bcopy(tcnp->cn_nameptr, nnn, nnnl); nnn[nnnl] = '\0'; printf("ren replace=%s\n",nnn); } #endif FREE((caddr_t)fnp->n_v4, M_NFSV4NODE); fnp->n_v4 = newv4; newv4 = NULL; fnp->n_v4->n4_fhlen = tdnp->n_fhp->nfh_len; fnp->n_v4->n4_namelen = tcnp->cn_namelen; NFSBCOPY(tdnp->n_fhp->nfh_fh, fnp->n_v4->n4_data, tdnp->n_fhp->nfh_len); NFSBCOPY(tcnp->cn_nameptr, NFS4NODENAME(fnp->n_v4), tcnp->cn_namelen); } mtx_unlock(&tdnp->n_mtx); mtx_unlock(&fnp->n_mtx); if (newv4 != NULL) FREE((caddr_t)newv4, M_NFSV4NODE); } if (fvp->v_type == VDIR) { if (tvp != NULL && tvp->v_type == VDIR) cache_purge(tdvp); cache_purge(fdvp); } out: if (tdvp == tvp) vrele(tdvp); else vput(tdvp); if (tvp) vput(tvp); vrele(fdvp); vrele(fvp); /* * Kludge: Map ENOENT => 0 assuming that it is a reply to a retry. */ if (error == ENOENT) error = 0; return (error); } /* * nfs file rename rpc called from nfs_remove() above */ static int nfs_renameit(struct vnode *sdvp, struct vnode *svp, struct componentname *scnp, struct sillyrename *sp) { return (nfs_renamerpc(sdvp, svp, scnp->cn_nameptr, scnp->cn_namelen, sdvp, NULL, sp->s_name, sp->s_namlen, scnp->cn_cred, scnp->cn_thread)); } /* * Do an nfs rename rpc. Called from nfs_rename() and nfs_renameit(). */ static int nfs_renamerpc(struct vnode *fdvp, struct vnode *fvp, char *fnameptr, int fnamelen, struct vnode *tdvp, struct vnode *tvp, char *tnameptr, int tnamelen, struct ucred *cred, struct thread *td) { struct nfsvattr fnfsva, tnfsva; struct nfsnode *fdnp = VTONFS(fdvp); struct nfsnode *tdnp = VTONFS(tdvp); int error = 0, fattrflag, tattrflag; error = nfsrpc_rename(fdvp, fvp, fnameptr, fnamelen, tdvp, tvp, tnameptr, tnamelen, cred, td, &fnfsva, &tnfsva, &fattrflag, &tattrflag, NULL, NULL); mtx_lock(&fdnp->n_mtx); fdnp->n_flag |= NMODIFIED; if (fattrflag != 0) { mtx_unlock(&fdnp->n_mtx); (void) nfscl_loadattrcache(&fdvp, &fnfsva, NULL, NULL, 0, 1); } else { fdnp->n_attrstamp = 0; mtx_unlock(&fdnp->n_mtx); } mtx_lock(&tdnp->n_mtx); tdnp->n_flag |= NMODIFIED; if (tattrflag != 0) { mtx_unlock(&tdnp->n_mtx); (void) nfscl_loadattrcache(&tdvp, &tnfsva, NULL, NULL, 0, 1); } else { tdnp->n_attrstamp = 0; mtx_unlock(&tdnp->n_mtx); } if (error && NFS_ISV4(fdvp)) error = nfscl_maperr(td, error, (uid_t)0, (gid_t)0); return (error); } /* * nfs hard link create call */ static int nfs_link(struct vop_link_args *ap) { struct vnode *vp = ap->a_vp; struct vnode *tdvp = ap->a_tdvp; struct componentname *cnp = ap->a_cnp; struct nfsnode *np, *tdnp; struct nfsvattr nfsva, dnfsva; int error = 0, attrflag, dattrflag; if (vp->v_mount != tdvp->v_mount) { return (EXDEV); } /* * Push all writes to the server, so that the attribute cache * doesn't get "out of sync" with the server. * XXX There should be a better way! */ VOP_FSYNC(vp, MNT_WAIT, cnp->cn_thread); error = nfsrpc_link(tdvp, vp, cnp->cn_nameptr, cnp->cn_namelen, cnp->cn_cred, cnp->cn_thread, &dnfsva, &nfsva, &attrflag, &dattrflag, NULL); tdnp = VTONFS(tdvp); mtx_lock(&tdnp->n_mtx); tdnp->n_flag |= NMODIFIED; if (dattrflag != 0) { mtx_unlock(&tdnp->n_mtx); (void) nfscl_loadattrcache(&tdvp, &dnfsva, NULL, NULL, 0, 1); } else { tdnp->n_attrstamp = 0; mtx_unlock(&tdnp->n_mtx); } if (attrflag) (void) nfscl_loadattrcache(&vp, &nfsva, NULL, NULL, 0, 1); else { np = VTONFS(vp); mtx_lock(&np->n_mtx); np->n_attrstamp = 0; mtx_unlock(&np->n_mtx); } /* * If negative lookup caching is enabled, I might as well * add an entry for this node. Not necessary for correctness, * but if negative caching is enabled, then the system * must care about lookup caching hit rate, so... */ if (VFSTONFS(vp->v_mount)->nm_negnametimeo != 0 && (cnp->cn_flags & MAKEENTRY)) cache_enter(tdvp, vp, cnp); if (error && NFS_ISV4(vp)) error = nfscl_maperr(cnp->cn_thread, error, (uid_t)0, (gid_t)0); return (error); } /* * nfs symbolic link create call */ static int nfs_symlink(struct vop_symlink_args *ap) { struct vnode *dvp = ap->a_dvp; struct vattr *vap = ap->a_vap; struct componentname *cnp = ap->a_cnp; struct nfsvattr nfsva, dnfsva; struct nfsfh *nfhp; struct nfsnode *np = NULL, *dnp; struct vnode *newvp = NULL; int error = 0, attrflag, dattrflag, ret; vap->va_type = VLNK; error = nfsrpc_symlink(dvp, cnp->cn_nameptr, cnp->cn_namelen, ap->a_target, vap, cnp->cn_cred, cnp->cn_thread, &dnfsva, &nfsva, &nfhp, &attrflag, &dattrflag, NULL); if (nfhp) { ret = nfscl_nget(dvp->v_mount, dvp, nfhp, cnp, cnp->cn_thread, &np, NULL, LK_EXCLUSIVE); if (!ret) newvp = NFSTOV(np); else if (!error) error = ret; } if (newvp != NULL) { if (attrflag) (void) nfscl_loadattrcache(&newvp, &nfsva, NULL, NULL, 0, 1); } else if (!error) { /* * If we do not have an error and we could not extract the * newvp from the response due to the request being NFSv2, we * have to do a lookup in order to obtain a newvp to return. */ error = nfs_lookitup(dvp, cnp->cn_nameptr, cnp->cn_namelen, cnp->cn_cred, cnp->cn_thread, &np); if (!error) newvp = NFSTOV(np); } if (error) { if (newvp) vput(newvp); if (NFS_ISV4(dvp)) error = nfscl_maperr(cnp->cn_thread, error, vap->va_uid, vap->va_gid); } else { /* * If negative lookup caching is enabled, I might as well * add an entry for this node. Not necessary for correctness, * but if negative caching is enabled, then the system * must care about lookup caching hit rate, so... */ if (VFSTONFS(dvp->v_mount)->nm_negnametimeo != 0 && (cnp->cn_flags & MAKEENTRY)) cache_enter(dvp, newvp, cnp); *ap->a_vpp = newvp; } dnp = VTONFS(dvp); mtx_lock(&dnp->n_mtx); dnp->n_flag |= NMODIFIED; if (dattrflag != 0) { mtx_unlock(&dnp->n_mtx); (void) nfscl_loadattrcache(&dvp, &dnfsva, NULL, NULL, 0, 1); } else { dnp->n_attrstamp = 0; mtx_unlock(&dnp->n_mtx); } return (error); } /* * nfs make dir call */ static int nfs_mkdir(struct vop_mkdir_args *ap) { struct vnode *dvp = ap->a_dvp; struct vattr *vap = ap->a_vap; struct componentname *cnp = ap->a_cnp; struct nfsnode *np = NULL, *dnp; struct vnode *newvp = NULL; struct vattr vattr; struct nfsfh *nfhp; struct nfsvattr nfsva, dnfsva; int error = 0, attrflag, dattrflag, ret; if ((error = VOP_GETATTR(dvp, &vattr, cnp->cn_cred)) != 0) return (error); vap->va_type = VDIR; error = nfsrpc_mkdir(dvp, cnp->cn_nameptr, cnp->cn_namelen, vap, cnp->cn_cred, cnp->cn_thread, &dnfsva, &nfsva, &nfhp, &attrflag, &dattrflag, NULL); dnp = VTONFS(dvp); mtx_lock(&dnp->n_mtx); dnp->n_flag |= NMODIFIED; if (dattrflag != 0) { mtx_unlock(&dnp->n_mtx); (void) nfscl_loadattrcache(&dvp, &dnfsva, NULL, NULL, 0, 1); } else { dnp->n_attrstamp = 0; mtx_unlock(&dnp->n_mtx); } if (nfhp) { ret = nfscl_nget(dvp->v_mount, dvp, nfhp, cnp, cnp->cn_thread, &np, NULL, LK_EXCLUSIVE); if (!ret) { newvp = NFSTOV(np); if (attrflag) (void) nfscl_loadattrcache(&newvp, &nfsva, NULL, NULL, 0, 1); } else if (!error) error = ret; } if (!error && newvp == NULL) { error = nfs_lookitup(dvp, cnp->cn_nameptr, cnp->cn_namelen, cnp->cn_cred, cnp->cn_thread, &np); if (!error) { newvp = NFSTOV(np); if (newvp->v_type != VDIR) error = EEXIST; } } if (error) { if (newvp) vput(newvp); if (NFS_ISV4(dvp)) error = nfscl_maperr(cnp->cn_thread, error, vap->va_uid, vap->va_gid); } else { /* * If negative lookup caching is enabled, I might as well * add an entry for this node. Not necessary for correctness, * but if negative caching is enabled, then the system * must care about lookup caching hit rate, so... */ if (VFSTONFS(dvp->v_mount)->nm_negnametimeo != 0 && (cnp->cn_flags & MAKEENTRY)) cache_enter(dvp, newvp, cnp); *ap->a_vpp = newvp; } return (error); } /* * nfs remove directory call */ static int nfs_rmdir(struct vop_rmdir_args *ap) { struct vnode *vp = ap->a_vp; struct vnode *dvp = ap->a_dvp; struct componentname *cnp = ap->a_cnp; struct nfsnode *dnp; struct nfsvattr dnfsva; int error, dattrflag; if (dvp == vp) return (EINVAL); error = nfsrpc_rmdir(dvp, cnp->cn_nameptr, cnp->cn_namelen, cnp->cn_cred, cnp->cn_thread, &dnfsva, &dattrflag, NULL); dnp = VTONFS(dvp); mtx_lock(&dnp->n_mtx); dnp->n_flag |= NMODIFIED; if (dattrflag != 0) { mtx_unlock(&dnp->n_mtx); (void) nfscl_loadattrcache(&dvp, &dnfsva, NULL, NULL, 0, 1); } else { dnp->n_attrstamp = 0; mtx_unlock(&dnp->n_mtx); } cache_purge(dvp); cache_purge(vp); if (error && NFS_ISV4(dvp)) error = nfscl_maperr(cnp->cn_thread, error, (uid_t)0, (gid_t)0); /* * Kludge: Map ENOENT => 0 assuming that you have a reply to a retry. */ if (error == ENOENT) error = 0; return (error); } /* * nfs readdir call */ static int nfs_readdir(struct vop_readdir_args *ap) { struct vnode *vp = ap->a_vp; struct nfsnode *np = VTONFS(vp); struct uio *uio = ap->a_uio; int tresid, error = 0; struct vattr vattr; if (vp->v_type != VDIR) return(EPERM); /* * First, check for hit on the EOF offset cache */ if (np->n_direofoffset > 0 && uio->uio_offset >= np->n_direofoffset && (np->n_flag & NMODIFIED) == 0) { if (VOP_GETATTR(vp, &vattr, ap->a_cred) == 0) { mtx_lock(&np->n_mtx); if ((NFS_ISV4(vp) && np->n_change == vattr.va_filerev) || !NFS_TIMESPEC_COMPARE(&np->n_mtime, &vattr.va_mtime)) { mtx_unlock(&np->n_mtx); NFSINCRGLOBAL(newnfsstats.direofcache_hits); return (0); } else mtx_unlock(&np->n_mtx); } } /* * Call ncl_bioread() to do the real work. */ tresid = uio->uio_resid; error = ncl_bioread(vp, uio, 0, ap->a_cred); if (!error && uio->uio_resid == tresid) NFSINCRGLOBAL(newnfsstats.direofcache_misses); return (error); } /* * Readdir rpc call. * Called from below the buffer cache by ncl_doio(). */ int ncl_readdirrpc(struct vnode *vp, struct uio *uiop, struct ucred *cred, struct thread *td) { struct nfsvattr nfsva; nfsuint64 *cookiep, cookie; struct nfsnode *dnp = VTONFS(vp); struct nfsmount *nmp = VFSTONFS(vp->v_mount); int error = 0, eof, attrflag; KASSERT(uiop->uio_iovcnt == 1 && (uiop->uio_offset & (DIRBLKSIZ - 1)) == 0 && (uiop->uio_resid & (DIRBLKSIZ - 1)) == 0, ("nfs readdirrpc bad uio")); /* * If there is no cookie, assume directory was stale. */ ncl_dircookie_lock(dnp); cookiep = ncl_getcookie(dnp, uiop->uio_offset, 0); if (cookiep) { cookie = *cookiep; ncl_dircookie_unlock(dnp); } else { ncl_dircookie_unlock(dnp); return (NFSERR_BAD_COOKIE); } if (NFSHASNFSV3(nmp) && !NFSHASGOTFSINFO(nmp)) (void)ncl_fsinfo(nmp, vp, cred, td); error = nfsrpc_readdir(vp, uiop, &cookie, cred, td, &nfsva, &attrflag, &eof, NULL); if (attrflag) (void) nfscl_loadattrcache(&vp, &nfsva, NULL, NULL, 0, 1); if (!error) { /* * We are now either at the end of the directory or have filled * the block. */ if (eof) dnp->n_direofoffset = uiop->uio_offset; else { if (uiop->uio_resid > 0) ncl_printf("EEK! readdirrpc resid > 0\n"); ncl_dircookie_lock(dnp); cookiep = ncl_getcookie(dnp, uiop->uio_offset, 1); *cookiep = cookie; ncl_dircookie_unlock(dnp); } } else if (NFS_ISV4(vp)) { error = nfscl_maperr(td, error, (uid_t)0, (gid_t)0); } return (error); } /* * NFS V3 readdir plus RPC. Used in place of ncl_readdirrpc(). */ int ncl_readdirplusrpc(struct vnode *vp, struct uio *uiop, struct ucred *cred, struct thread *td) { struct nfsvattr nfsva; nfsuint64 *cookiep, cookie; struct nfsnode *dnp = VTONFS(vp); struct nfsmount *nmp = VFSTONFS(vp->v_mount); int error = 0, attrflag, eof; KASSERT(uiop->uio_iovcnt == 1 && (uiop->uio_offset & (DIRBLKSIZ - 1)) == 0 && (uiop->uio_resid & (DIRBLKSIZ - 1)) == 0, ("nfs readdirplusrpc bad uio")); /* * If there is no cookie, assume directory was stale. */ ncl_dircookie_lock(dnp); cookiep = ncl_getcookie(dnp, uiop->uio_offset, 0); if (cookiep) { cookie = *cookiep; ncl_dircookie_unlock(dnp); } else { ncl_dircookie_unlock(dnp); return (NFSERR_BAD_COOKIE); } if (NFSHASNFSV3(nmp) && !NFSHASGOTFSINFO(nmp)) (void)ncl_fsinfo(nmp, vp, cred, td); error = nfsrpc_readdirplus(vp, uiop, &cookie, cred, td, &nfsva, &attrflag, &eof, NULL); if (attrflag) (void) nfscl_loadattrcache(&vp, &nfsva, NULL, NULL, 0, 1); if (!error) { /* * We are now either at end of the directory or have filled the * the block. */ if (eof) dnp->n_direofoffset = uiop->uio_offset; else { if (uiop->uio_resid > 0) ncl_printf("EEK! readdirplusrpc resid > 0\n"); ncl_dircookie_lock(dnp); cookiep = ncl_getcookie(dnp, uiop->uio_offset, 1); *cookiep = cookie; ncl_dircookie_unlock(dnp); } } else if (NFS_ISV4(vp)) { error = nfscl_maperr(td, error, (uid_t)0, (gid_t)0); } return (error); } /* * Silly rename. To make the NFS filesystem that is stateless look a little * more like the "ufs" a remove of an active vnode is translated to a rename * to a funny looking filename that is removed by nfs_inactive on the * nfsnode. There is the potential for another process on a different client * to create the same funny name between the nfs_lookitup() fails and the * nfs_rename() completes, but... */ static int nfs_sillyrename(struct vnode *dvp, struct vnode *vp, struct componentname *cnp) { struct sillyrename *sp; struct nfsnode *np; int error; short pid; unsigned int lticks; cache_purge(dvp); np = VTONFS(vp); KASSERT(vp->v_type != VDIR, ("nfs: sillyrename dir")); MALLOC(sp, struct sillyrename *, sizeof (struct sillyrename), M_NEWNFSREQ, M_WAITOK); sp->s_cred = crhold(cnp->cn_cred); sp->s_dvp = dvp; VREF(dvp); /* * Fudge together a funny name. * Changing the format of the funny name to accomodate more * sillynames per directory. * The name is now changed to .nfs...4, where ticks is * CPU ticks since boot. */ pid = cnp->cn_thread->td_proc->p_pid; lticks = (unsigned int)ticks; for ( ; ; ) { sp->s_namlen = sprintf(sp->s_name, ".nfs.%08x.%04x4.4", lticks, pid); if (nfs_lookitup(dvp, sp->s_name, sp->s_namlen, sp->s_cred, cnp->cn_thread, NULL)) break; lticks++; } error = nfs_renameit(dvp, vp, cnp, sp); if (error) goto bad; error = nfs_lookitup(dvp, sp->s_name, sp->s_namlen, sp->s_cred, cnp->cn_thread, &np); np->n_sillyrename = sp; return (0); bad: vrele(sp->s_dvp); crfree(sp->s_cred); free((caddr_t)sp, M_NEWNFSREQ); return (error); } /* * Look up a file name and optionally either update the file handle or * allocate an nfsnode, depending on the value of npp. * npp == NULL --> just do the lookup * *npp == NULL --> allocate a new nfsnode and make sure attributes are * handled too * *npp != NULL --> update the file handle in the vnode */ static int nfs_lookitup(struct vnode *dvp, char *name, int len, struct ucred *cred, struct thread *td, struct nfsnode **npp) { struct vnode *newvp = NULL, *vp; struct nfsnode *np, *dnp = VTONFS(dvp); struct nfsfh *nfhp, *onfhp; struct nfsvattr nfsva, dnfsva; struct componentname cn; int error = 0, attrflag, dattrflag; u_int hash; error = nfsrpc_lookup(dvp, name, len, cred, td, &dnfsva, &nfsva, &nfhp, &attrflag, &dattrflag, NULL); if (dattrflag) (void) nfscl_loadattrcache(&dvp, &dnfsva, NULL, NULL, 0, 1); if (npp && !error) { if (*npp != NULL) { np = *npp; vp = NFSTOV(np); /* * For NFSv4, check to see if it is the same name and * replace the name, if it is different. */ if (np->n_v4 != NULL && nfsva.na_type == VREG && (np->n_v4->n4_namelen != len || NFSBCMP(name, NFS4NODENAME(np->n_v4), len) || dnp->n_fhp->nfh_len != np->n_v4->n4_fhlen || NFSBCMP(dnp->n_fhp->nfh_fh, np->n_v4->n4_data, dnp->n_fhp->nfh_len))) { #ifdef notdef { char nnn[100]; int nnnl; nnnl = (len < 100) ? len : 99; bcopy(name, nnn, nnnl); nnn[nnnl] = '\0'; printf("replace=%s\n",nnn); } #endif FREE((caddr_t)np->n_v4, M_NFSV4NODE); MALLOC(np->n_v4, struct nfsv4node *, sizeof (struct nfsv4node) + dnp->n_fhp->nfh_len + len - 1, M_NFSV4NODE, M_WAITOK); np->n_v4->n4_fhlen = dnp->n_fhp->nfh_len; np->n_v4->n4_namelen = len; NFSBCOPY(dnp->n_fhp->nfh_fh, np->n_v4->n4_data, dnp->n_fhp->nfh_len); NFSBCOPY(name, NFS4NODENAME(np->n_v4), len); } hash = fnv_32_buf(nfhp->nfh_fh, nfhp->nfh_len, FNV1_32_INIT); onfhp = np->n_fhp; /* * Rehash node for new file handle. */ vfs_hash_rehash(vp, hash); np->n_fhp = nfhp; if (onfhp != NULL) FREE((caddr_t)onfhp, M_NFSFH); newvp = NFSTOV(np); } else if (NFS_CMPFH(dnp, nfhp->nfh_fh, nfhp->nfh_len)) { FREE((caddr_t)nfhp, M_NFSFH); VREF(dvp); newvp = dvp; } else { cn.cn_nameptr = name; cn.cn_namelen = len; error = nfscl_nget(dvp->v_mount, dvp, nfhp, &cn, td, &np, NULL, LK_EXCLUSIVE); if (error) return (error); newvp = NFSTOV(np); } if (!attrflag && *npp == NULL) { if (newvp == dvp) vrele(newvp); else vput(newvp); return (ENOENT); } if (attrflag) (void) nfscl_loadattrcache(&newvp, &nfsva, NULL, NULL, 0, 1); } if (npp && *npp == NULL) { if (error) { if (newvp) { if (newvp == dvp) vrele(newvp); else vput(newvp); } } else *npp = np; } if (error && NFS_ISV4(dvp)) error = nfscl_maperr(td, error, (uid_t)0, (gid_t)0); return (error); } /* * Nfs Version 3 and 4 commit rpc */ int ncl_commit(struct vnode *vp, u_quad_t offset, int cnt, struct ucred *cred, struct thread *td) { struct nfsvattr nfsva; struct nfsmount *nmp = VFSTONFS(vp->v_mount); int error, attrflag; u_char verf[NFSX_VERF]; mtx_lock(&nmp->nm_mtx); if ((nmp->nm_state & NFSSTA_HASWRITEVERF) == 0) { mtx_unlock(&nmp->nm_mtx); return (0); } mtx_unlock(&nmp->nm_mtx); error = nfsrpc_commit(vp, offset, cnt, cred, td, verf, &nfsva, &attrflag, NULL); if (!error) { mtx_lock(&nmp->nm_mtx); if (NFSBCMP((caddr_t)nmp->nm_verf, verf, NFSX_VERF)) { NFSBCOPY(verf, (caddr_t)nmp->nm_verf, NFSX_VERF); error = NFSERR_STALEWRITEVERF; } mtx_unlock(&nmp->nm_mtx); if (!error && attrflag) (void) nfscl_loadattrcache(&vp, &nfsva, NULL, NULL, 0, 1); } else if (NFS_ISV4(vp)) { error = nfscl_maperr(td, error, (uid_t)0, (gid_t)0); } return (error); } /* * Strategy routine. * For async requests when nfsiod(s) are running, queue the request by * calling ncl_asyncio(), otherwise just all ncl_doio() to do the * request. */ static int nfs_strategy(struct vop_strategy_args *ap) { struct buf *bp = ap->a_bp; struct ucred *cr; KASSERT(!(bp->b_flags & B_DONE), ("nfs_strategy: buffer %p unexpectedly marked B_DONE", bp)); BUF_ASSERT_HELD(bp); if (bp->b_iocmd == BIO_READ) cr = bp->b_rcred; else cr = bp->b_wcred; /* * If the op is asynchronous and an i/o daemon is waiting * queue the request, wake it up and wait for completion * otherwise just do it ourselves. */ if ((bp->b_flags & B_ASYNC) == 0 || ncl_asyncio(VFSTONFS(ap->a_vp->v_mount), bp, NOCRED, curthread)) (void) ncl_doio(ap->a_vp, bp, cr, curthread, 1); return (0); } /* * fsync vnode op. Just call ncl_flush() with commit == 1. */ /* ARGSUSED */ static int nfs_fsync(struct vop_fsync_args *ap) { return (ncl_flush(ap->a_vp, ap->a_waitfor, NULL, ap->a_td, 1, 0)); } /* * Flush all the blocks associated with a vnode. * Walk through the buffer pool and push any dirty pages * associated with the vnode. * If the called_from_renewthread argument is TRUE, it has been called * from the NFSv4 renew thread and, as such, cannot block indefinitely * waiting for a buffer write to complete. */ int ncl_flush(struct vnode *vp, int waitfor, struct ucred *cred, struct thread *td, int commit, int called_from_renewthread) { struct nfsnode *np = VTONFS(vp); struct buf *bp; int i; struct buf *nbp; struct nfsmount *nmp = VFSTONFS(vp->v_mount); int error = 0, slptimeo = 0, slpflag = 0, retv, bvecpos; int passone = 1, trycnt = 0; u_quad_t off, endoff, toff; struct ucred* wcred = NULL; struct buf **bvec = NULL; struct bufobj *bo; #ifndef NFS_COMMITBVECSIZ #define NFS_COMMITBVECSIZ 20 #endif struct buf *bvec_on_stack[NFS_COMMITBVECSIZ]; int bvecsize = 0, bveccount; if (called_from_renewthread != 0) slptimeo = hz; if (nmp->nm_flag & NFSMNT_INT) slpflag = NFS_PCATCH; if (!commit) passone = 0; bo = &vp->v_bufobj; /* * A b_flags == (B_DELWRI | B_NEEDCOMMIT) block has been written to the * server, but has not been committed to stable storage on the server * yet. On the first pass, the byte range is worked out and the commit * rpc is done. On the second pass, ncl_writebp() is called to do the * job. */ again: off = (u_quad_t)-1; endoff = 0; bvecpos = 0; if (NFS_ISV34(vp) && commit) { if (bvec != NULL && bvec != bvec_on_stack) free(bvec, M_TEMP); /* * Count up how many buffers waiting for a commit. */ bveccount = 0; BO_LOCK(bo); TAILQ_FOREACH_SAFE(bp, &bo->bo_dirty.bv_hd, b_bobufs, nbp) { if (!BUF_ISLOCKED(bp) && (bp->b_flags & (B_DELWRI | B_NEEDCOMMIT)) == (B_DELWRI | B_NEEDCOMMIT)) bveccount++; } /* * Allocate space to remember the list of bufs to commit. It is * important to use M_NOWAIT here to avoid a race with nfs_write. * If we can't get memory (for whatever reason), we will end up * committing the buffers one-by-one in the loop below. */ if (bveccount > NFS_COMMITBVECSIZ) { /* * Release the vnode interlock to avoid a lock * order reversal. */ BO_UNLOCK(bo); bvec = (struct buf **) malloc(bveccount * sizeof(struct buf *), M_TEMP, M_NOWAIT); BO_LOCK(bo); if (bvec == NULL) { bvec = bvec_on_stack; bvecsize = NFS_COMMITBVECSIZ; } else bvecsize = bveccount; } else { bvec = bvec_on_stack; bvecsize = NFS_COMMITBVECSIZ; } TAILQ_FOREACH_SAFE(bp, &bo->bo_dirty.bv_hd, b_bobufs, nbp) { if (bvecpos >= bvecsize) break; if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_NOWAIT, NULL)) { nbp = TAILQ_NEXT(bp, b_bobufs); continue; } if ((bp->b_flags & (B_DELWRI | B_NEEDCOMMIT)) != (B_DELWRI | B_NEEDCOMMIT)) { BUF_UNLOCK(bp); nbp = TAILQ_NEXT(bp, b_bobufs); continue; } BO_UNLOCK(bo); bremfree(bp); /* * Work out if all buffers are using the same cred * so we can deal with them all with one commit. * * NOTE: we are not clearing B_DONE here, so we have * to do it later on in this routine if we intend to * initiate I/O on the bp. * * Note: to avoid loopback deadlocks, we do not * assign b_runningbufspace. */ if (wcred == NULL) wcred = bp->b_wcred; else if (wcred != bp->b_wcred) wcred = NOCRED; vfs_busy_pages(bp, 1); BO_LOCK(bo); /* * bp is protected by being locked, but nbp is not * and vfs_busy_pages() may sleep. We have to * recalculate nbp. */ nbp = TAILQ_NEXT(bp, b_bobufs); /* * A list of these buffers is kept so that the * second loop knows which buffers have actually * been committed. This is necessary, since there * may be a race between the commit rpc and new * uncommitted writes on the file. */ bvec[bvecpos++] = bp; toff = ((u_quad_t)bp->b_blkno) * DEV_BSIZE + bp->b_dirtyoff; if (toff < off) off = toff; toff += (u_quad_t)(bp->b_dirtyend - bp->b_dirtyoff); if (toff > endoff) endoff = toff; } BO_UNLOCK(bo); } if (bvecpos > 0) { /* * Commit data on the server, as required. * If all bufs are using the same wcred, then use that with * one call for all of them, otherwise commit each one * separately. */ if (wcred != NOCRED) retv = ncl_commit(vp, off, (int)(endoff - off), wcred, td); else { retv = 0; for (i = 0; i < bvecpos; i++) { off_t off, size; bp = bvec[i]; off = ((u_quad_t)bp->b_blkno) * DEV_BSIZE + bp->b_dirtyoff; size = (u_quad_t)(bp->b_dirtyend - bp->b_dirtyoff); retv = ncl_commit(vp, off, (int)size, bp->b_wcred, td); if (retv) break; } } if (retv == NFSERR_STALEWRITEVERF) ncl_clearcommit(vp->v_mount); /* * Now, either mark the blocks I/O done or mark the * blocks dirty, depending on whether the commit * succeeded. */ for (i = 0; i < bvecpos; i++) { bp = bvec[i]; bp->b_flags &= ~(B_NEEDCOMMIT | B_CLUSTEROK); if (retv) { /* * Error, leave B_DELWRI intact */ vfs_unbusy_pages(bp); brelse(bp); } else { /* * Success, remove B_DELWRI ( bundirty() ). * * b_dirtyoff/b_dirtyend seem to be NFS * specific. We should probably move that * into bundirty(). XXX */ bufobj_wref(bo); bp->b_flags |= B_ASYNC; bundirty(bp); bp->b_flags &= ~B_DONE; bp->b_ioflags &= ~BIO_ERROR; bp->b_dirtyoff = bp->b_dirtyend = 0; bufdone(bp); } } } /* * Start/do any write(s) that are required. */ loop: BO_LOCK(bo); TAILQ_FOREACH_SAFE(bp, &bo->bo_dirty.bv_hd, b_bobufs, nbp) { if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_NOWAIT, NULL)) { if (waitfor != MNT_WAIT || passone) continue; error = BUF_TIMELOCK(bp, LK_EXCLUSIVE | LK_SLEEPFAIL | LK_INTERLOCK, BO_MTX(bo), "nfsfsync", slpflag, slptimeo); if (error == 0) { BUF_UNLOCK(bp); goto loop; } if (error == ENOLCK) { error = 0; goto loop; } if (called_from_renewthread != 0) { /* * Return EIO so the flush will be retried * later. */ error = EIO; goto done; } if (newnfs_sigintr(nmp, td)) { error = EINTR; goto done; } if (slpflag & PCATCH) { slpflag = 0; slptimeo = 2 * hz; } goto loop; } if ((bp->b_flags & B_DELWRI) == 0) panic("nfs_fsync: not dirty"); if ((passone || !commit) && (bp->b_flags & B_NEEDCOMMIT)) { BUF_UNLOCK(bp); continue; } BO_UNLOCK(bo); bremfree(bp); if (passone || !commit) bp->b_flags |= B_ASYNC; else bp->b_flags |= B_ASYNC; bwrite(bp); if (newnfs_sigintr(nmp, td)) { error = EINTR; goto done; } goto loop; } if (passone) { passone = 0; BO_UNLOCK(bo); goto again; } if (waitfor == MNT_WAIT) { while (bo->bo_numoutput) { error = bufobj_wwait(bo, slpflag, slptimeo); if (error) { BO_UNLOCK(bo); if (called_from_renewthread != 0) { /* * Return EIO so that the flush will be * retried later. */ error = EIO; goto done; } error = newnfs_sigintr(nmp, td); if (error) goto done; if (slpflag & PCATCH) { slpflag = 0; slptimeo = 2 * hz; } BO_LOCK(bo); } } if (bo->bo_dirty.bv_cnt != 0 && commit) { BO_UNLOCK(bo); goto loop; } /* * Wait for all the async IO requests to drain */ BO_UNLOCK(bo); mtx_lock(&np->n_mtx); while (np->n_directio_asyncwr > 0) { np->n_flag |= NFSYNCWAIT; error = newnfs_msleep(td, &np->n_directio_asyncwr, &np->n_mtx, slpflag | (PRIBIO + 1), "nfsfsync", 0); if (error) { if (newnfs_sigintr(nmp, td)) { mtx_unlock(&np->n_mtx); error = EINTR; goto done; } } } mtx_unlock(&np->n_mtx); } else BO_UNLOCK(bo); mtx_lock(&np->n_mtx); if (np->n_flag & NWRITEERR) { error = np->n_error; np->n_flag &= ~NWRITEERR; } if (commit && bo->bo_dirty.bv_cnt == 0 && bo->bo_numoutput == 0 && np->n_directio_asyncwr == 0) np->n_flag &= ~NMODIFIED; mtx_unlock(&np->n_mtx); done: if (bvec != NULL && bvec != bvec_on_stack) free(bvec, M_TEMP); if (error == 0 && commit != 0 && waitfor == MNT_WAIT && (bo->bo_dirty.bv_cnt != 0 || bo->bo_numoutput != 0 || np->n_directio_asyncwr != 0) && trycnt++ < 5) { /* try, try again... */ passone = 1; wcred = NULL; bvec = NULL; bvecsize = 0; printf("try%d\n", trycnt); goto again; } return (error); } /* * NFS advisory byte-level locks. */ static int nfs_advlock(struct vop_advlock_args *ap) { struct vnode *vp = ap->a_vp; struct ucred *cred; struct nfsnode *np = VTONFS(ap->a_vp); struct proc *p = (struct proc *)ap->a_id; struct thread *td = curthread; /* XXX */ struct vattr va; int ret, error = EOPNOTSUPP; u_quad_t size; - if (NFS_ISV4(vp) && (ap->a_flags & F_POSIX)) { - cred = p->p_ucred; + if (NFS_ISV4(vp) && (ap->a_flags & (F_POSIX | F_FLOCK)) != 0) { + if ((ap->a_flags & F_POSIX) != 0) + cred = p->p_ucred; + else + cred = td->td_ucred; vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); if (vp->v_iflag & VI_DOOMED) { VOP_UNLOCK(vp, 0); return (EBADF); } /* * If this is unlocking a write locked region, flush and * commit them before unlocking. This is required by * RFC3530 Sec. 9.3.2. */ if (ap->a_op == F_UNLCK && - nfscl_checkwritelocked(vp, ap->a_fl, cred, td)) + nfscl_checkwritelocked(vp, ap->a_fl, cred, td, ap->a_id, + ap->a_flags)) (void) ncl_flush(vp, MNT_WAIT, cred, td, 1, 0); /* * Loop around doing the lock op, while a blocking lock * must wait for the lock op to succeed. */ do { ret = nfsrpc_advlock(vp, np->n_size, ap->a_op, - ap->a_fl, 0, cred, td); + ap->a_fl, 0, cred, td, ap->a_id, ap->a_flags); if (ret == NFSERR_DENIED && (ap->a_flags & F_WAIT) && ap->a_op == F_SETLK) { VOP_UNLOCK(vp, 0); error = nfs_catnap(PZERO | PCATCH, ret, "ncladvl"); if (error) return (EINTR); vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); if (vp->v_iflag & VI_DOOMED) { VOP_UNLOCK(vp, 0); return (EBADF); } } } while (ret == NFSERR_DENIED && (ap->a_flags & F_WAIT) && ap->a_op == F_SETLK); if (ret == NFSERR_DENIED) { VOP_UNLOCK(vp, 0); return (EAGAIN); } else if (ret == EINVAL || ret == EBADF || ret == EINTR) { VOP_UNLOCK(vp, 0); return (ret); } else if (ret != 0) { VOP_UNLOCK(vp, 0); return (EACCES); } /* * Now, if we just got a lock, invalidate data in the buffer * cache, as required, so that the coherency conforms with * RFC3530 Sec. 9.3.2. */ if (ap->a_op == F_SETLK) { if ((np->n_flag & NMODIFIED) == 0) { np->n_attrstamp = 0; ret = VOP_GETATTR(vp, &va, cred); } if ((np->n_flag & NMODIFIED) || ret || np->n_change != va.va_filerev) { (void) ncl_vinvalbuf(vp, V_SAVE, td, 1); np->n_attrstamp = 0; ret = VOP_GETATTR(vp, &va, cred); if (!ret) { np->n_mtime = va.va_mtime; np->n_change = va.va_filerev; } } } VOP_UNLOCK(vp, 0); return (0); } else if (!NFS_ISV4(vp)) { error = vn_lock(vp, LK_SHARED); if (error) return (error); if ((VFSTONFS(vp->v_mount)->nm_flag & NFSMNT_NOLOCKD) != 0) { size = VTONFS(vp)->n_size; VOP_UNLOCK(vp, 0); error = lf_advlock(ap, &(vp->v_lockf), size); } else { if (nfs_advlock_p != NULL) error = nfs_advlock_p(ap); else { VOP_UNLOCK(vp, 0); error = ENOLCK; } } } return (error); } /* * NFS advisory byte-level locks. */ static int nfs_advlockasync(struct vop_advlockasync_args *ap) { struct vnode *vp = ap->a_vp; u_quad_t size; int error; if (NFS_ISV4(vp)) return (EOPNOTSUPP); error = vn_lock(vp, LK_SHARED); if (error) return (error); if ((VFSTONFS(vp->v_mount)->nm_flag & NFSMNT_NOLOCKD) != 0) { size = VTONFS(vp)->n_size; VOP_UNLOCK(vp, 0); error = lf_advlockasync(ap, &(vp->v_lockf), size); } else { VOP_UNLOCK(vp, 0); error = EOPNOTSUPP; } return (error); } /* * Print out the contents of an nfsnode. */ static int nfs_print(struct vop_print_args *ap) { struct vnode *vp = ap->a_vp; struct nfsnode *np = VTONFS(vp); ncl_printf("\tfileid %ld fsid 0x%x", np->n_vattr.na_fileid, np->n_vattr.na_fsid); if (vp->v_type == VFIFO) fifo_printinfo(vp); printf("\n"); return (0); } /* * This is the "real" nfs::bwrite(struct buf*). * We set B_CACHE if this is a VMIO buffer. */ int ncl_writebp(struct buf *bp, int force __unused, struct thread *td) { int s; int oldflags = bp->b_flags; #if 0 int retv = 1; off_t off; #endif BUF_ASSERT_HELD(bp); if (bp->b_flags & B_INVAL) { brelse(bp); return(0); } bp->b_flags |= B_CACHE; /* * Undirty the bp. We will redirty it later if the I/O fails. */ s = splbio(); bundirty(bp); bp->b_flags &= ~B_DONE; bp->b_ioflags &= ~BIO_ERROR; bp->b_iocmd = BIO_WRITE; bufobj_wref(bp->b_bufobj); curthread->td_ru.ru_oublock++; splx(s); /* * Note: to avoid loopback deadlocks, we do not * assign b_runningbufspace. */ vfs_busy_pages(bp, 1); BUF_KERNPROC(bp); bp->b_iooffset = dbtob(bp->b_blkno); bstrategy(bp); if( (oldflags & B_ASYNC) == 0) { int rtval = bufwait(bp); if (oldflags & B_DELWRI) { s = splbio(); reassignbuf(bp); splx(s); } brelse(bp); return (rtval); } return (0); } /* * nfs special file access vnode op. * Essentially just get vattr and then imitate iaccess() since the device is * local to the client. */ static int nfsspec_access(struct vop_access_args *ap) { struct vattr *vap; struct ucred *cred = ap->a_cred; struct vnode *vp = ap->a_vp; accmode_t accmode = ap->a_accmode; struct vattr vattr; int error; /* * Disallow write attempts on filesystems mounted read-only; * unless the file is a socket, fifo, or a block or character * device resident on the filesystem. */ if ((accmode & VWRITE) && (vp->v_mount->mnt_flag & MNT_RDONLY)) { switch (vp->v_type) { case VREG: case VDIR: case VLNK: return (EROFS); default: break; } } vap = &vattr; error = VOP_GETATTR(vp, vap, cred); if (error) goto out; error = vaccess(vp->v_type, vap->va_mode, vap->va_uid, vap->va_gid, accmode, cred, NULL); out: return error; } /* * Read wrapper for fifos. */ static int nfsfifo_read(struct vop_read_args *ap) { struct nfsnode *np = VTONFS(ap->a_vp); int error; /* * Set access flag. */ mtx_lock(&np->n_mtx); np->n_flag |= NACC; getnanotime(&np->n_atim); mtx_unlock(&np->n_mtx); error = fifo_specops.vop_read(ap); return error; } /* * Write wrapper for fifos. */ static int nfsfifo_write(struct vop_write_args *ap) { struct nfsnode *np = VTONFS(ap->a_vp); /* * Set update flag. */ mtx_lock(&np->n_mtx); np->n_flag |= NUPD; getnanotime(&np->n_mtim); mtx_unlock(&np->n_mtx); return(fifo_specops.vop_write(ap)); } /* * Close wrapper for fifos. * * Update the times on the nfsnode then do fifo close. */ static int nfsfifo_close(struct vop_close_args *ap) { struct vnode *vp = ap->a_vp; struct nfsnode *np = VTONFS(vp); struct vattr vattr; struct timespec ts; mtx_lock(&np->n_mtx); if (np->n_flag & (NACC | NUPD)) { getnanotime(&ts); if (np->n_flag & NACC) np->n_atim = ts; if (np->n_flag & NUPD) np->n_mtim = ts; np->n_flag |= NCHG; if (vrefcnt(vp) == 1 && (vp->v_mount->mnt_flag & MNT_RDONLY) == 0) { VATTR_NULL(&vattr); if (np->n_flag & NACC) vattr.va_atime = np->n_atim; if (np->n_flag & NUPD) vattr.va_mtime = np->n_mtim; mtx_unlock(&np->n_mtx); (void)VOP_SETATTR(vp, &vattr, ap->a_cred); goto out; } } mtx_unlock(&np->n_mtx); out: return (fifo_specops.vop_close(ap)); } /* * Just call ncl_writebp() with the force argument set to 1. * * NOTE: B_DONE may or may not be set in a_bp on call. */ static int nfs_bwrite(struct buf *bp) { return (ncl_writebp(bp, 1, curthread)); } struct buf_ops buf_ops_newnfs = { .bop_name = "buf_ops_nfs", .bop_write = nfs_bwrite, .bop_strategy = bufstrategy, .bop_sync = bufsync, .bop_bdflush = bufbdflush, }; /* * Cloned from vop_stdlock(), and then the ugly hack added. */ static int nfs_lock1(struct vop_lock1_args *ap) { struct vnode *vp = ap->a_vp; int error = 0; /* * Since vfs_hash_get() calls vget() and it will no longer work * for FreeBSD8 with flags == 0, I can only think of this horrible * hack to work around it. I call vfs_hash_get() with LK_EXCLOTHER * and then handle it here. All I want for this case is a v_usecount * on the vnode to use for recovery, while another thread might * hold a lock on the vnode. I have the other threads blocked, so * there isn't any race problem. */ if ((ap->a_flags & LK_TYPE_MASK) == LK_EXCLOTHER) { if ((ap->a_flags & LK_INTERLOCK) == 0) panic("ncllock1"); if ((vp->v_iflag & VI_DOOMED)) error = ENOENT; VI_UNLOCK(vp); return (error); } return (_lockmgr_args(vp->v_vnlock, ap->a_flags, VI_MTX(vp), LK_WMESG_DEFAULT, LK_PRIO_DEFAULT, LK_TIMO_DEFAULT, ap->a_file, ap->a_line)); } static int nfs_getacl(struct vop_getacl_args *ap) { int error; if (ap->a_type != ACL_TYPE_NFS4) return (EOPNOTSUPP); error = nfsrpc_getacl(ap->a_vp, ap->a_cred, ap->a_td, ap->a_aclp, NULL); if (error > NFSERR_STALE) { (void) nfscl_maperr(ap->a_td, error, (uid_t)0, (gid_t)0); error = EPERM; } return (error); } static int nfs_setacl(struct vop_setacl_args *ap) { int error; if (ap->a_type != ACL_TYPE_NFS4) return (EOPNOTSUPP); error = nfsrpc_setacl(ap->a_vp, ap->a_cred, ap->a_td, ap->a_aclp, NULL); if (error > NFSERR_STALE) { (void) nfscl_maperr(ap->a_td, error, (uid_t)0, (gid_t)0); error = EPERM; } return (error); } /* * Return POSIX pathconf information applicable to nfs filesystems. */ static int nfs_pathconf(struct vop_pathconf_args *ap) { struct nfsv3_pathconf pc; struct nfsvattr nfsva; struct vnode *vp = ap->a_vp; struct thread *td = curthread; int attrflag, error; if (NFS_ISV4(vp) || (NFS_ISV3(vp) && (ap->a_name == _PC_LINK_MAX || ap->a_name == _PC_NAME_MAX || ap->a_name == _PC_CHOWN_RESTRICTED || ap->a_name == _PC_NO_TRUNC))) { /* * Since only the above 4 a_names are returned by the NFSv3 * Pathconf RPC, there is no point in doing it for others. */ error = nfsrpc_pathconf(vp, &pc, td->td_ucred, td, &nfsva, &attrflag, NULL); if (attrflag != 0) (void) nfscl_loadattrcache(&vp, &nfsva, NULL, NULL, 0, 1); if (error != 0) return (error); } else { /* * For NFSv2 (or NFSv3 when not one of the above 4 a_names), * just fake them. */ pc.pc_linkmax = LINK_MAX; pc.pc_namemax = NFS_MAXNAMLEN; pc.pc_notrunc = 1; pc.pc_chownrestricted = 1; pc.pc_caseinsensitive = 0; pc.pc_casepreserving = 1; error = 0; } switch (ap->a_name) { case _PC_LINK_MAX: *ap->a_retval = pc.pc_linkmax; break; case _PC_NAME_MAX: *ap->a_retval = pc.pc_namemax; break; case _PC_PATH_MAX: *ap->a_retval = PATH_MAX; break; case _PC_PIPE_BUF: *ap->a_retval = PIPE_BUF; break; case _PC_CHOWN_RESTRICTED: *ap->a_retval = pc.pc_chownrestricted; break; case _PC_NO_TRUNC: *ap->a_retval = pc.pc_notrunc; break; case _PC_ACL_EXTENDED: *ap->a_retval = 0; break; case _PC_ACL_NFS4: if (NFS_ISV4(vp) && nfsrv_useacl != 0 && attrflag != 0 && NFSISSET_ATTRBIT(&nfsva.na_suppattr, NFSATTRBIT_ACL)) *ap->a_retval = 1; else *ap->a_retval = 0; break; case _PC_ACL_PATH_MAX: if (NFS_ISV4(vp)) *ap->a_retval = ACL_MAX_ENTRIES; else *ap->a_retval = 3; break; case _PC_MAC_PRESENT: *ap->a_retval = 0; break; case _PC_ASYNC_IO: /* _PC_ASYNC_IO should have been handled by upper layers. */ KASSERT(0, ("_PC_ASYNC_IO should not get here")); error = EINVAL; break; case _PC_PRIO_IO: *ap->a_retval = 0; break; case _PC_SYNC_IO: *ap->a_retval = 0; break; case _PC_ALLOC_SIZE_MIN: *ap->a_retval = vp->v_mount->mnt_stat.f_bsize; break; case _PC_FILESIZEBITS: if (NFS_ISV34(vp)) *ap->a_retval = 64; else *ap->a_retval = 32; break; case _PC_REC_INCR_XFER_SIZE: *ap->a_retval = vp->v_mount->mnt_stat.f_iosize; break; case _PC_REC_MAX_XFER_SIZE: *ap->a_retval = -1; /* means ``unlimited'' */ break; case _PC_REC_MIN_XFER_SIZE: *ap->a_retval = vp->v_mount->mnt_stat.f_iosize; break; case _PC_REC_XFER_ALIGN: *ap->a_retval = PAGE_SIZE; break; case _PC_SYMLINK_MAX: *ap->a_retval = NFS_MAXPATHLEN; break; default: error = EINVAL; break; } return (error); } Index: projects/largeSMP/sys/i386/i386/mp_machdep.c =================================================================== --- projects/largeSMP/sys/i386/i386/mp_machdep.c (revision 222782) +++ projects/largeSMP/sys/i386/i386/mp_machdep.c (revision 222783) @@ -1,1733 +1,1736 @@ /*- * 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_mp_watchdog.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; /* Hotwire a 0->4MB V==P mapping */ extern pt_entry_t *KPTphys; struct pcb stoppcbs[MAXCPU]; /* 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 /* * 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 */ static 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 int hlt_logical_cpus; static u_int hyperthreading_cpus; /* logical cpus sharing L1 cache */ static cpuset_t hyperthreading_cpus_mask; static int hyperthreading_allowed = 1; static struct sysctl_ctx_list logical_cpu_clist; 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. */ - cpuid_count(0x04, 0, p); - max_cores = ((p[0] >> 26) & 0x3f) + 1; + 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)); /* 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) { cpuset_t tcpuset, tallcpus; struct pcpu *pc; vm_offset_t addr; int gsel_tss; int x, myid; u_int 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); /* Disable local APIC just to be sure. */ lapic_disable(); /* 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 /* Initialize the PAT MSR if present. */ pmap_init_pat(); /* set up CPU registers and state */ cpu_setregs(); /* set up FPU state on the AP */ npxinit(); /* set up SSE registers */ enable_sse(); #ifdef PAE /* Enable the PTE no-execute bit. */ if ((amd_feature & AMDID_NX) != 0) { uint64_t msr; msr = rdmsr(MSR_EFER) | EFER_NXE; wrmsr(MSR_EFER, msr); } #endif /* A quick check from sanity claus */ if (PCPU_GET(apic_id) != lapic_id()) { printf("SMP: cpuid = %d\n", PCPU_GET(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", PCPU_GET(cpuid)); printf("SMP: AP CPU #%d Launched!\n", PCPU_GET(cpuid)); tcpuset = PCPU_GET(cpumask); /* 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_OR(&logical_cpus_mask, &tcpuset); /* Determine if we are a hyperthread. */ if (hyperthreading_cpus > 1 && PCPU_GET(apic_id) % hyperthreading_cpus != 0) CPU_OR(&hyperthreading_cpus_mask, &tcpuset); /* Build our map of 'other' CPUs. */ tallcpus = all_cpus; CPU_NAND(&tallcpus, &tcpuset); PCPU_SET(other_cpus, tallcpus); if (bootverbose) lapic_dump("AP"); if (smp_cpus == mp_ncpus) { /* enable IPI's, tlb shootdown, freezes etc */ atomic_store_rel_int(&smp_started, 1); smp_active = 1; /* historic */ } 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) 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; #if defined(SCHED_ULE) /* * Don't use HT CPU if it has been disabled by a * tunable. */ if (hyperthreading_allowed == 0) { cpu_info[i].cpu_disabled = 1; continue; } #endif } /* 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; } } /* * 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) { cpuset_t tallcpus; #ifndef PC98 u_char mpbiosreason; #endif uintptr_t kptbase; 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 */ kptbase = (uintptr_t)(void *)KPTphys; for (i = TMPMAP_START; i < NKPT; i++) PTD[i] = (pd_entry_t)(PG_V | PG_RW | ((kptbase + i * PAGE_SIZE) & PG_FRAME)); 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_alloc(kernel_map, KSTACK_PAGES * PAGE_SIZE); dpcpu = (void *)kmem_alloc(kernel_map, DPCPU_SIZE); /* 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 */ } /* build our map of 'other' CPUs */ tallcpus = all_cpus; CPU_NAND(&tallcpus, PCPU_PTR(cpumask)); PCPU_SET(other_cpus, tallcpus); /* 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; /* * first we do an INIT/RESET IPI this INIT IPI might be run, reseting * 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. */ /* do an INIT IPI: assert RESET */ lapic_ipi_raw(APIC_DEST_DESTFLD | APIC_TRIGMOD_EDGE | APIC_LEVEL_ASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_INIT, apic_id); /* wait for pending status end */ lapic_ipi_wait(-1); /* do an INIT IPI: deassert RESET */ lapic_ipi_raw(APIC_DEST_ALLESELF | APIC_TRIGMOD_LEVEL | APIC_LEVEL_DEASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_INIT, 0); /* wait for pending status end */ DELAY(10000); /* wait ~10mS */ lapic_ipi_wait(-1); /* * 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. */ /* do a STARTUP IPI */ lapic_ipi_raw(APIC_DEST_DESTFLD | APIC_TRIGMOD_EDGE | APIC_LEVEL_DEASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_STARTUP | vector, apic_id); lapic_ipi_wait(-1); 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_DEASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_STARTUP | vector, apic_id); lapic_ipi_wait(-1); DELAY(200); /* wait ~200uS */ /* 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]; 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 */ /* * 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 { sched_pin(); CPU_NAND(&mask, PCPU_PTR(cpumask)); sched_unpin(); 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 = cpusetobj_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 = cpusetobj_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) { sched_pin(); if (IPI_IS_BITMAPED(ipi)) { ipi_selected(PCPU_GET(other_cpus), ipi); sched_unpin(); 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, PCPU_PTR(other_cpus)); sched_unpin(); CTR2(KTR_SMP, "%s: ipi: %x", __func__, ipi); lapic_ipi_vectored(ipi, APIC_IPI_DEST_OTHERS); } int ipi_nmi_handler() { cpuset_t cpumask; /* * 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. */ sched_pin(); cpumask = PCPU_GET(cpumask); sched_unpin(); if (!CPU_OVERLAP(&ipi_nmi_pending, &cpumask)) return (1); CPU_NAND_ATOMIC(&ipi_nmi_pending, &cpumask); cpustop_handler(); return (0); } /* * Handle an IPI_STOP by saving our current context and spinning until we * are resumed. */ void cpustop_handler(void) { cpuset_t cpumask; u_int cpu; sched_pin(); cpu = PCPU_GET(cpuid); cpumask = PCPU_GET(cpumask); sched_unpin(); savectx(&stoppcbs[cpu]); /* Indicate that we are stopped */ CPU_OR_ATOMIC(&stopped_cpus, &cpumask); /* Wait for restart */ while (!CPU_OVERLAP(&started_cpus, &cpumask)) ia32_pause(); CPU_NAND_ATOMIC(&started_cpus, &cpumask); CPU_NAND_ATOMIC(&stopped_cpus, &cpumask); if (cpu == 0 && cpustop_restartfunc != NULL) { cpustop_restartfunc(); cpustop_restartfunc = NULL; } } /* * 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); static int sysctl_hlt_cpus(SYSCTL_HANDLER_ARGS) { cpuset_t mask; int error; mask = hlt_cpus_mask; error = sysctl_handle_opaque(oidp, &mask, sizeof(mask), req); if (error || !req->newptr) return (error); if (!CPU_EMPTY(&logical_cpus_mask) && CPU_SUBSET(&mask, &logical_cpus_mask)) hlt_logical_cpus = 1; else hlt_logical_cpus = 0; if (! hyperthreading_allowed) CPU_OR(&mask, &hyperthreading_cpus_mask); if (CPU_SUBSET(&mask, &all_cpus)) CPU_CLR(0, &mask); hlt_cpus_mask = mask; return (error); } SYSCTL_PROC(_machdep, OID_AUTO, hlt_cpus, CTLTYPE_STRUCT | CTLFLAG_RW | CTLFLAG_MPSAFE, 0, 0, sysctl_hlt_cpus, "S", "Bitmap of CPUs to halt. 101 (binary) will halt CPUs 0 and 2."); static int sysctl_hlt_logical_cpus(SYSCTL_HANDLER_ARGS) { int disable, error; disable = hlt_logical_cpus; error = sysctl_handle_int(oidp, &disable, 0, req); if (error || !req->newptr) return (error); if (disable) CPU_OR(&hlt_cpus_mask, &logical_cpus_mask); else CPU_NAND(&hlt_cpus_mask, &logical_cpus_mask); if (! hyperthreading_allowed) CPU_OR(&hlt_cpus_mask, &hyperthreading_cpus_mask); if (CPU_SUBSET(&hlt_cpus_mask, &all_cpus)) CPU_CLR(0, &hlt_cpus_mask); hlt_logical_cpus = disable; return (error); } static int sysctl_hyperthreading_allowed(SYSCTL_HANDLER_ARGS) { int allowed, error; allowed = hyperthreading_allowed; error = sysctl_handle_int(oidp, &allowed, 0, req); if (error || !req->newptr) return (error); #ifdef SCHED_ULE /* * SCHED_ULE doesn't allow enabling/disabling HT cores at * run-time. */ if (allowed != hyperthreading_allowed) return (ENOTSUP); return (error); #endif if (allowed) CPU_NAND(&hlt_cpus_mask, &hyperthreading_cpus_mask); else CPU_OR(&hlt_cpus_mask, &hyperthreading_cpus_mask); if (!CPU_EMPTY(&logical_cpus_mask) && CPU_SUBSET(&hlt_cpus_mask, &logical_cpus_mask)) hlt_logical_cpus = 1; else hlt_logical_cpus = 0; if (CPU_SUBSET(&hlt_cpus_mask, &all_cpus)) CPU_CLR(0, &hlt_cpus_mask); hyperthreading_allowed = allowed; return (error); } static void cpu_hlt_setup(void *dummy __unused) { if (!CPU_EMPTY(&logical_cpus_mask)) { TUNABLE_INT_FETCH("machdep.hlt_logical_cpus", &hlt_logical_cpus); sysctl_ctx_init(&logical_cpu_clist); SYSCTL_ADD_PROC(&logical_cpu_clist, SYSCTL_STATIC_CHILDREN(_machdep), OID_AUTO, "hlt_logical_cpus", CTLTYPE_INT|CTLFLAG_RW, 0, 0, sysctl_hlt_logical_cpus, "IU", ""); SYSCTL_ADD_UINT(&logical_cpu_clist, SYSCTL_STATIC_CHILDREN(_machdep), OID_AUTO, "logical_cpus_mask", CTLTYPE_INT|CTLFLAG_RD, &logical_cpus_mask, 0, ""); if (hlt_logical_cpus) CPU_OR(&hlt_cpus_mask, &logical_cpus_mask); /* * If necessary for security purposes, force * hyperthreading off, regardless of the value * of hlt_logical_cpus. */ if (!CPU_EMPTY(&hyperthreading_cpus_mask)) { SYSCTL_ADD_PROC(&logical_cpu_clist, SYSCTL_STATIC_CHILDREN(_machdep), OID_AUTO, "hyperthreading_allowed", CTLTYPE_INT|CTLFLAG_RW, 0, 0, sysctl_hyperthreading_allowed, "IU", ""); if (! hyperthreading_allowed) CPU_OR(&hlt_cpus_mask, &hyperthreading_cpus_mask); } } } SYSINIT(cpu_hlt, SI_SUB_SMP, SI_ORDER_ANY, cpu_hlt_setup, NULL); int mp_grab_cpu_hlt(void) { cpuset_t mask; #ifdef MP_WATCHDOG u_int cpuid; #endif int retval; mask = PCPU_GET(cpumask); #ifdef MP_WATCHDOG cpuid = PCPU_GET(cpuid); ap_watchdog(cpuid); #endif retval = 0; while (CPU_OVERLAP(&mask, &hlt_cpus_mask)) { retval = 1; __asm __volatile("sti; hlt" : : : "memory"); } return (retval); } #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: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/largeSMP/sys/ia64/acpica/acpi_machdep.c =================================================================== --- projects/largeSMP/sys/ia64/acpica/acpi_machdep.c (revision 222782) +++ projects/largeSMP/sys/ia64/acpica/acpi_machdep.c (revision 222783) @@ -1,94 +1,101 @@ /*- * Copyright (c) 2001 Doug Rabson * 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. * * $FreeBSD$ */ #include #include #include #include #include #include #include int acpi_machdep_init(device_t dev) { struct acpi_softc *sc; sc = device_get_softc(dev); acpi_install_wakeup_handler(sc); return (0); } int acpi_machdep_quirks(int *quirks) { return (0); } void acpi_cpu_c1() { +#ifdef INVARIANTS + register_t ie; + + ie = intr_disable(); + KASSERT(ie == 0, ("%s called with interrupts enabled\n", __func__)); +#endif ia64_call_pal_static(PAL_HALT_LIGHT, 0, 0, 0); + ia64_enable_intr(); } void * acpi_find_table(const char *sig) { ACPI_PHYSICAL_ADDRESS rsdp_ptr; ACPI_TABLE_RSDP *rsdp; ACPI_TABLE_XSDT *xsdt; ACPI_TABLE_HEADER *table; UINT64 addr; u_int i, count; if ((rsdp_ptr = AcpiOsGetRootPointer()) == 0) return (NULL); rsdp = (ACPI_TABLE_RSDP *)IA64_PHYS_TO_RR7(rsdp_ptr); xsdt = (ACPI_TABLE_XSDT *)IA64_PHYS_TO_RR7(rsdp->XsdtPhysicalAddress); count = (UINT64 *)((char *)xsdt + xsdt->Header.Length) - xsdt->TableOffsetEntry; for (i = 0; i < count; i++) { addr = xsdt->TableOffsetEntry[i]; table = (ACPI_TABLE_HEADER *)IA64_PHYS_TO_RR7(addr); if (strncmp(table->Signature, sig, ACPI_NAME_SIZE) != 0) continue; if (ACPI_FAILURE(AcpiTbChecksum((void *)table, table->Length))) continue; return (table); } return (NULL); } Index: projects/largeSMP/sys/ia64/ia64/machdep.c =================================================================== --- projects/largeSMP/sys/ia64/ia64/machdep.c (revision 222782) +++ projects/largeSMP/sys/ia64/ia64/machdep.c (revision 222783) @@ -1,1555 +1,1580 @@ /*- * Copyright (c) 2003,2004 Marcel Moolenaar * Copyright (c) 2000,2001 Doug Rabson * 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_compat.h" #include "opt_ddb.h" #include "opt_kstack_pages.h" #include "opt_sched.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 #include #include #include #include #include #ifdef SMP #include #endif #include #include SYSCTL_NODE(_hw, OID_AUTO, freq, CTLFLAG_RD, 0, ""); SYSCTL_NODE(_machdep, OID_AUTO, cpu, CTLFLAG_RD, 0, ""); static u_int bus_freq; SYSCTL_UINT(_hw_freq, OID_AUTO, bus, CTLFLAG_RD, &bus_freq, 0, "Bus clock frequency"); static u_int cpu_freq; SYSCTL_UINT(_hw_freq, OID_AUTO, cpu, CTLFLAG_RD, &cpu_freq, 0, "CPU clock frequency"); static u_int itc_freq; SYSCTL_UINT(_hw_freq, OID_AUTO, itc, CTLFLAG_RD, &itc_freq, 0, "ITC frequency"); int cold = 1; struct bootinfo *bootinfo; struct pcpu pcpu0; extern u_int64_t kernel_text[], _end[]; extern u_int64_t ia64_gateway_page[]; extern u_int64_t break_sigtramp[]; extern u_int64_t epc_sigtramp[]; struct fpswa_iface *fpswa_iface; vm_size_t ia64_pal_size; vm_paddr_t ia64_pal_base; vm_offset_t ia64_port_base; u_int64_t ia64_lapic_addr = PAL_PIB_DEFAULT_ADDR; struct ia64_pib *ia64_pib; static int ia64_sync_icache_needed; char machine[] = MACHINE; SYSCTL_STRING(_hw, HW_MACHINE, machine, CTLFLAG_RD, machine, 0, ""); static char cpu_model[64]; SYSCTL_STRING(_hw, HW_MODEL, model, CTLFLAG_RD, cpu_model, 0, "The CPU model name"); static char cpu_family[64]; SYSCTL_STRING(_hw, OID_AUTO, family, CTLFLAG_RD, cpu_family, 0, "The CPU family name"); #ifdef DDB extern vm_offset_t ksym_start, ksym_end; #endif struct msgbuf *msgbufp = NULL; /* Other subsystems (e.g., ACPI) can hook this later. */ void (*cpu_idle_hook)(void) = NULL; long Maxmem = 0; long realmem = 0; #define PHYSMAP_SIZE (2 * VM_PHYSSEG_MAX) vm_paddr_t phys_avail[PHYSMAP_SIZE + 2]; /* must be 2 less so 0 0 can signal end of chunks */ #define PHYS_AVAIL_ARRAY_END ((sizeof(phys_avail) / sizeof(vm_offset_t)) - 2) struct kva_md_info kmi; #define Mhz 1000000L #define Ghz (1000L*Mhz) static void identifycpu(void) { char vendor[17]; char *family_name, *model_name; u_int64_t features, tmp; int number, revision, model, family, archrev; /* * Assumes little-endian. */ *(u_int64_t *) &vendor[0] = ia64_get_cpuid(0); *(u_int64_t *) &vendor[8] = ia64_get_cpuid(1); vendor[16] = '\0'; tmp = ia64_get_cpuid(3); number = (tmp >> 0) & 0xff; revision = (tmp >> 8) & 0xff; model = (tmp >> 16) & 0xff; family = (tmp >> 24) & 0xff; archrev = (tmp >> 32) & 0xff; family_name = model_name = "unknown"; switch (family) { case 0x07: family_name = "Itanium"; model_name = "Merced"; break; case 0x1f: family_name = "Itanium 2"; switch (model) { case 0x00: model_name = "McKinley"; break; case 0x01: /* * Deerfield is a low-voltage variant based on the * Madison core. We need circumstantial evidence * (i.e. the clock frequency) to identify those. * Allow for roughly 1% error margin. */ if (cpu_freq > 990 && cpu_freq < 1010) model_name = "Deerfield"; else model_name = "Madison"; break; case 0x02: model_name = "Madison II"; break; } break; case 0x20: ia64_sync_icache_needed = 1; family_name = "Itanium 2"; switch (model) { case 0x00: model_name = "Montecito"; break; } break; } snprintf(cpu_family, sizeof(cpu_family), "%s", family_name); snprintf(cpu_model, sizeof(cpu_model), "%s", model_name); features = ia64_get_cpuid(4); printf("CPU: %s (", model_name); if (cpu_freq) printf("%u Mhz ", cpu_freq); printf("%s)\n", family_name); printf(" Origin = \"%s\" Revision = %d\n", vendor, revision); printf(" Features = 0x%b\n", (u_int32_t) features, "\020" "\001LB" /* long branch (brl) instruction. */ "\002SD" /* Spontaneous deferral. */ "\003AO" /* 16-byte atomic operations (ld, st, cmpxchg). */ ); } static void cpu_startup(void *dummy) { char nodename[16]; struct pcpu *pc; struct pcpu_stats *pcs; /* * Good {morning,afternoon,evening,night}. */ identifycpu(); #ifdef PERFMON perfmon_init(); #endif printf("real memory = %ld (%ld MB)\n", ia64_ptob(Maxmem), ia64_ptob(Maxmem) / 1048576); realmem = Maxmem; /* * Display any holes after the first chunk of extended memory. */ if (bootverbose) { int indx; printf("Physical memory chunk(s):\n"); for (indx = 0; phys_avail[indx + 1] != 0; indx += 2) { long size1 = phys_avail[indx + 1] - phys_avail[indx]; printf("0x%08lx - 0x%08lx, %ld bytes (%ld pages)\n", phys_avail[indx], phys_avail[indx + 1] - 1, size1, size1 >> PAGE_SHIFT); } } vm_ksubmap_init(&kmi); printf("avail memory = %ld (%ld MB)\n", ptoa(cnt.v_free_count), ptoa(cnt.v_free_count) / 1048576); if (fpswa_iface == NULL) printf("Warning: no FPSWA package supplied\n"); else printf("FPSWA Revision = 0x%lx, Entry = %p\n", (long)fpswa_iface->if_rev, (void *)fpswa_iface->if_fpswa); /* * Set up buffers, so they can be used to read disk labels. */ bufinit(); vm_pager_bufferinit(); /* * Traverse the MADT to discover IOSAPIC and Local SAPIC * information. */ ia64_probe_sapics(); ia64_pib = pmap_mapdev(ia64_lapic_addr, sizeof(*ia64_pib)); ia64_mca_init(); /* * Create sysctl tree for per-CPU information. */ STAILQ_FOREACH(pc, &cpuhead, pc_allcpu) { snprintf(nodename, sizeof(nodename), "%u", pc->pc_cpuid); sysctl_ctx_init(&pc->pc_md.sysctl_ctx); pc->pc_md.sysctl_tree = SYSCTL_ADD_NODE(&pc->pc_md.sysctl_ctx, SYSCTL_STATIC_CHILDREN(_machdep_cpu), OID_AUTO, nodename, CTLFLAG_RD, NULL, ""); if (pc->pc_md.sysctl_tree == NULL) continue; pcs = &pc->pc_md.stats; SYSCTL_ADD_ULONG(&pc->pc_md.sysctl_ctx, SYSCTL_CHILDREN(pc->pc_md.sysctl_tree), OID_AUTO, "nasts", CTLFLAG_RD, &pcs->pcs_nasts, "Number of IPI_AST interrupts"); SYSCTL_ADD_ULONG(&pc->pc_md.sysctl_ctx, SYSCTL_CHILDREN(pc->pc_md.sysctl_tree), OID_AUTO, "nclks", CTLFLAG_RD, &pcs->pcs_nclks, "Number of clock interrupts"); SYSCTL_ADD_ULONG(&pc->pc_md.sysctl_ctx, SYSCTL_CHILDREN(pc->pc_md.sysctl_tree), OID_AUTO, "nextints", CTLFLAG_RD, &pcs->pcs_nextints, "Number of ExtINT interrupts"); SYSCTL_ADD_ULONG(&pc->pc_md.sysctl_ctx, SYSCTL_CHILDREN(pc->pc_md.sysctl_tree), OID_AUTO, "nhighfps", CTLFLAG_RD, &pcs->pcs_nhighfps, "Number of IPI_HIGH_FP interrupts"); SYSCTL_ADD_ULONG(&pc->pc_md.sysctl_ctx, SYSCTL_CHILDREN(pc->pc_md.sysctl_tree), OID_AUTO, "nhwints", CTLFLAG_RD, &pcs->pcs_nhwints, "Number of hardware (device) interrupts"); SYSCTL_ADD_ULONG(&pc->pc_md.sysctl_ctx, SYSCTL_CHILDREN(pc->pc_md.sysctl_tree), OID_AUTO, "npreempts", CTLFLAG_RD, &pcs->pcs_npreempts, "Number of IPI_PREEMPT interrupts"); SYSCTL_ADD_ULONG(&pc->pc_md.sysctl_ctx, SYSCTL_CHILDREN(pc->pc_md.sysctl_tree), OID_AUTO, "nrdvs", CTLFLAG_RD, &pcs->pcs_nrdvs, "Number of IPI_RENDEZVOUS interrupts"); SYSCTL_ADD_ULONG(&pc->pc_md.sysctl_ctx, SYSCTL_CHILDREN(pc->pc_md.sysctl_tree), OID_AUTO, "nstops", CTLFLAG_RD, &pcs->pcs_nstops, "Number of IPI_STOP interrupts"); SYSCTL_ADD_ULONG(&pc->pc_md.sysctl_ctx, SYSCTL_CHILDREN(pc->pc_md.sysctl_tree), OID_AUTO, "nstrays", CTLFLAG_RD, &pcs->pcs_nstrays, "Number of stray interrupts"); } } SYSINIT(cpu_startup, SI_SUB_CPU, SI_ORDER_FIRST, cpu_startup, NULL); void cpu_flush_dcache(void *ptr, size_t len) { vm_offset_t lim, va; va = (uintptr_t)ptr & ~31; lim = (uintptr_t)ptr + len; while (va < lim) { ia64_fc(va); va += 32; } ia64_srlz_d(); } /* Get current clock frequency for the given cpu id. */ int cpu_est_clockrate(int cpu_id, uint64_t *rate) { if (pcpu_find(cpu_id) == NULL || rate == NULL) return (EINVAL); *rate = (u_long)cpu_freq * 1000000ul; return (0); } void cpu_halt() { efi_reset_system(); } void cpu_idle(int busy) { - struct ia64_pal_result res; + register_t ie; - if (cpu_idle_hook != NULL) +#if 0 + if (!busy) { + critical_enter(); + cpu_idleclock(); + } +#endif + + ie = intr_disable(); + KASSERT(ie != 0, ("%s called with interrupts disabled\n", __func__)); + + if (sched_runnable()) + ia64_enable_intr(); + else if (cpu_idle_hook != NULL) { (*cpu_idle_hook)(); - else - res = ia64_call_pal_static(PAL_HALT_LIGHT, 0, 0, 0); + /* The hook must enable interrupts! */ + } else { + ia64_call_pal_static(PAL_HALT_LIGHT, 0, 0, 0); + ia64_enable_intr(); + } + +#if 0 + if (!busy) { + cpu_activeclock(); + critical_exit(); + } +#endif } int cpu_idle_wakeup(int cpu) { return (0); } void cpu_reset() { efi_reset_system(); } void cpu_switch(struct thread *old, struct thread *new, struct mtx *mtx) { struct pcb *oldpcb, *newpcb; oldpcb = old->td_pcb; #ifdef COMPAT_FREEBSD32 ia32_savectx(oldpcb); #endif if (PCPU_GET(fpcurthread) == old) old->td_frame->tf_special.psr |= IA64_PSR_DFH; if (!savectx(oldpcb)) { atomic_store_rel_ptr(&old->td_lock, mtx); newpcb = new->td_pcb; oldpcb->pcb_current_pmap = pmap_switch(newpcb->pcb_current_pmap); #if defined(SCHED_ULE) && defined(SMP) while (atomic_load_acq_ptr(&new->td_lock) == &blocked_lock) cpu_spinwait(); #endif PCPU_SET(curthread, new); #ifdef COMPAT_FREEBSD32 ia32_restorectx(newpcb); #endif if (PCPU_GET(fpcurthread) == new) new->td_frame->tf_special.psr &= ~IA64_PSR_DFH; restorectx(newpcb); /* We should not get here. */ panic("cpu_switch: restorectx() returned"); /* NOTREACHED */ } } void cpu_throw(struct thread *old __unused, struct thread *new) { struct pcb *newpcb; newpcb = new->td_pcb; (void)pmap_switch(newpcb->pcb_current_pmap); #if defined(SCHED_ULE) && defined(SMP) while (atomic_load_acq_ptr(&new->td_lock) == &blocked_lock) cpu_spinwait(); #endif PCPU_SET(curthread, new); #ifdef COMPAT_FREEBSD32 ia32_restorectx(newpcb); #endif restorectx(newpcb); /* We should not get here. */ panic("cpu_throw: restorectx() returned"); /* NOTREACHED */ } void cpu_pcpu_init(struct pcpu *pcpu, int cpuid, size_t size) { /* * Set pc_acpi_id to "uninitialized". * See sys/dev/acpica/acpi_cpu.c */ pcpu->pc_acpi_id = 0xffffffff; } void spinlock_enter(void) { struct thread *td; int intr; td = curthread; if (td->td_md.md_spinlock_count == 0) { intr = intr_disable(); td->td_md.md_spinlock_count = 1; td->td_md.md_saved_intr = intr; } else td->td_md.md_spinlock_count++; critical_enter(); } void spinlock_exit(void) { struct thread *td; int intr; td = curthread; critical_exit(); intr = td->td_md.md_saved_intr; td->td_md.md_spinlock_count--; if (td->td_md.md_spinlock_count == 0) intr_restore(intr); } void map_vhpt(uintptr_t vhpt) { pt_entry_t pte; uint64_t psr; pte = PTE_PRESENT | PTE_MA_WB | PTE_ACCESSED | PTE_DIRTY | PTE_PL_KERN | PTE_AR_RW; pte |= vhpt & PTE_PPN_MASK; __asm __volatile("ptr.d %0,%1" :: "r"(vhpt), "r"(pmap_vhpt_log2size << 2)); __asm __volatile("mov %0=psr" : "=r"(psr)); __asm __volatile("rsm psr.ic|psr.i"); ia64_srlz_i(); ia64_set_ifa(vhpt); ia64_set_itir(pmap_vhpt_log2size << 2); ia64_srlz_d(); __asm __volatile("itr.d dtr[%0]=%1" :: "r"(3), "r"(pte)); __asm __volatile("mov psr.l=%0" :: "r" (psr)); ia64_srlz_i(); } void map_pal_code(void) { pt_entry_t pte; vm_offset_t va; vm_size_t sz; uint64_t psr; u_int shft; if (ia64_pal_size == 0) return; va = IA64_PHYS_TO_RR7(ia64_pal_base); sz = ia64_pal_size; shft = 0; while (sz > 1) { shft++; sz >>= 1; } pte = PTE_PRESENT | PTE_MA_WB | PTE_ACCESSED | PTE_DIRTY | PTE_PL_KERN | PTE_AR_RWX; pte |= ia64_pal_base & PTE_PPN_MASK; __asm __volatile("ptr.d %0,%1; ptr.i %0,%1" :: "r"(va), "r"(shft<<2)); __asm __volatile("mov %0=psr" : "=r"(psr)); __asm __volatile("rsm psr.ic|psr.i"); ia64_srlz_i(); ia64_set_ifa(va); ia64_set_itir(shft << 2); ia64_srlz_d(); __asm __volatile("itr.d dtr[%0]=%1" :: "r"(4), "r"(pte)); ia64_srlz_d(); __asm __volatile("itr.i itr[%0]=%1" :: "r"(1), "r"(pte)); __asm __volatile("mov psr.l=%0" :: "r" (psr)); ia64_srlz_i(); } void map_gateway_page(void) { pt_entry_t pte; uint64_t psr; pte = PTE_PRESENT | PTE_MA_WB | PTE_ACCESSED | PTE_DIRTY | PTE_PL_KERN | PTE_AR_X_RX; pte |= ia64_tpa((uint64_t)ia64_gateway_page) & PTE_PPN_MASK; __asm __volatile("ptr.d %0,%1; ptr.i %0,%1" :: "r"(VM_MAXUSER_ADDRESS), "r"(PAGE_SHIFT << 2)); __asm __volatile("mov %0=psr" : "=r"(psr)); __asm __volatile("rsm psr.ic|psr.i"); ia64_srlz_i(); ia64_set_ifa(VM_MAXUSER_ADDRESS); ia64_set_itir(PAGE_SHIFT << 2); ia64_srlz_d(); __asm __volatile("itr.d dtr[%0]=%1" :: "r"(5), "r"(pte)); ia64_srlz_d(); __asm __volatile("itr.i itr[%0]=%1" :: "r"(2), "r"(pte)); __asm __volatile("mov psr.l=%0" :: "r" (psr)); ia64_srlz_i(); /* Expose the mapping to userland in ar.k5 */ ia64_set_k5(VM_MAXUSER_ADDRESS); } static u_int freq_ratio(u_long base, u_long ratio) { u_long f; f = (base * (ratio >> 32)) / (ratio & 0xfffffffful); return ((f + 500000) / 1000000); } static void calculate_frequencies(void) { struct ia64_sal_result sal; struct ia64_pal_result pal; + register_t ie; + ie = intr_disable(); sal = ia64_sal_entry(SAL_FREQ_BASE, 0, 0, 0, 0, 0, 0, 0); pal = ia64_call_pal_static(PAL_FREQ_RATIOS, 0, 0, 0); + intr_restore(ie); if (sal.sal_status == 0 && pal.pal_status == 0) { if (bootverbose) { printf("Platform clock frequency %ld Hz\n", sal.sal_result[0]); printf("Processor ratio %ld/%ld, Bus ratio %ld/%ld, " "ITC ratio %ld/%ld\n", pal.pal_result[0] >> 32, pal.pal_result[0] & ((1L << 32) - 1), pal.pal_result[1] >> 32, pal.pal_result[1] & ((1L << 32) - 1), pal.pal_result[2] >> 32, pal.pal_result[2] & ((1L << 32) - 1)); } cpu_freq = freq_ratio(sal.sal_result[0], pal.pal_result[0]); bus_freq = freq_ratio(sal.sal_result[0], pal.pal_result[1]); itc_freq = freq_ratio(sal.sal_result[0], pal.pal_result[2]); } } struct ia64_init_return ia64_init(void) { struct ia64_init_return ret; int phys_avail_cnt; vm_offset_t kernstart, kernend; vm_offset_t kernstartpfn, kernendpfn, pfn0, pfn1; char *p; struct efi_md *md; int metadata_missing; /* NO OUTPUT ALLOWED UNTIL FURTHER NOTICE */ /* * TODO: Disable interrupts, floating point etc. * Maybe flush cache and tlb */ ia64_set_fpsr(IA64_FPSR_DEFAULT); /* * TODO: Get critical system information (if possible, from the * information provided by the boot program). */ /* * Look for the I/O ports first - we need them for console * probing. */ for (md = efi_md_first(); md != NULL; md = efi_md_next(md)) { switch (md->md_type) { case EFI_MD_TYPE_IOPORT: ia64_port_base = (uintptr_t)pmap_mapdev(md->md_phys, md->md_pages * EFI_PAGE_SIZE); break; case EFI_MD_TYPE_PALCODE: ia64_pal_size = md->md_pages * EFI_PAGE_SIZE; ia64_pal_base = md->md_phys; break; } } metadata_missing = 0; if (bootinfo->bi_modulep) preload_metadata = (caddr_t)bootinfo->bi_modulep; else metadata_missing = 1; if (envmode == 0 && bootinfo->bi_envp) kern_envp = (caddr_t)bootinfo->bi_envp; else kern_envp = static_env; /* * Look at arguments passed to us and compute boothowto. */ boothowto = bootinfo->bi_boothowto; if (boothowto & RB_VERBOSE) bootverbose = 1; /* * Find the beginning and end of the kernel. */ kernstart = trunc_page(kernel_text); #ifdef DDB ksym_start = bootinfo->bi_symtab; ksym_end = bootinfo->bi_esymtab; kernend = (vm_offset_t)round_page(ksym_end); #else kernend = (vm_offset_t)round_page(_end); #endif /* But if the bootstrap tells us otherwise, believe it! */ if (bootinfo->bi_kernend) kernend = round_page(bootinfo->bi_kernend); /* * Region 6 is direct mapped UC and region 7 is direct mapped * WC. The details of this is controlled by the Alt {I,D}TLB * handlers. Here we just make sure that they have the largest * possible page size to minimise TLB usage. */ ia64_set_rr(IA64_RR_BASE(6), (6 << 8) | (PAGE_SHIFT << 2)); ia64_set_rr(IA64_RR_BASE(7), (7 << 8) | (PAGE_SHIFT << 2)); ia64_srlz_d(); /* * Wire things up so we can call the firmware. */ map_pal_code(); efi_boot_minimal(bootinfo->bi_systab); ia64_xiv_init(); ia64_sal_init(); calculate_frequencies(); /* * Setup the PCPU data for the bootstrap processor. It is needed * by printf(). Also, since printf() has critical sections, we * need to initialize at least pc_curthread. */ pcpup = &pcpu0; ia64_set_k4((u_int64_t)pcpup); pcpu_init(pcpup, 0, sizeof(pcpu0)); dpcpu_init((void *)kernend, 0); PCPU_SET(md.lid, ia64_get_lid()); kernend += DPCPU_SIZE; PCPU_SET(curthread, &thread0); /* * Initialize the console before we print anything out. */ cninit(); /* OUTPUT NOW ALLOWED */ if (metadata_missing) printf("WARNING: loader(8) metadata is missing!\n"); /* Get FPSWA interface */ fpswa_iface = (bootinfo->bi_fpswa == 0) ? NULL : (struct fpswa_iface *)IA64_PHYS_TO_RR7(bootinfo->bi_fpswa); /* Init basic tunables, including hz */ init_param1(); p = getenv("kernelname"); if (p != NULL) { strlcpy(kernelname, p, sizeof(kernelname)); freeenv(p); } kernstartpfn = atop(IA64_RR_MASK(kernstart)); kernendpfn = atop(IA64_RR_MASK(kernend)); /* * Size the memory regions and load phys_avail[] with the results. */ /* * Find out how much memory is available, by looking at * the memory descriptors. */ #ifdef DEBUG_MD printf("Memory descriptor count: %d\n", mdcount); #endif phys_avail_cnt = 0; for (md = efi_md_first(); md != NULL; md = efi_md_next(md)) { #ifdef DEBUG_MD printf("MD %p: type %d pa 0x%lx cnt 0x%lx\n", md, md->md_type, md->md_phys, md->md_pages); #endif pfn0 = ia64_btop(round_page(md->md_phys)); pfn1 = ia64_btop(trunc_page(md->md_phys + md->md_pages * 4096)); if (pfn1 <= pfn0) continue; if (md->md_type != EFI_MD_TYPE_FREE) continue; /* * We have a memory descriptor that describes conventional * memory that is for general use. We must determine if the * loader has put the kernel in this region. */ physmem += (pfn1 - pfn0); if (pfn0 <= kernendpfn && kernstartpfn <= pfn1) { /* * Must compute the location of the kernel * within the segment. */ #ifdef DEBUG_MD printf("Descriptor %p contains kernel\n", mp); #endif if (pfn0 < kernstartpfn) { /* * There is a chunk before the kernel. */ #ifdef DEBUG_MD printf("Loading chunk before kernel: " "0x%lx / 0x%lx\n", pfn0, kernstartpfn); #endif phys_avail[phys_avail_cnt] = ia64_ptob(pfn0); phys_avail[phys_avail_cnt+1] = ia64_ptob(kernstartpfn); phys_avail_cnt += 2; } if (kernendpfn < pfn1) { /* * There is a chunk after the kernel. */ #ifdef DEBUG_MD printf("Loading chunk after kernel: " "0x%lx / 0x%lx\n", kernendpfn, pfn1); #endif phys_avail[phys_avail_cnt] = ia64_ptob(kernendpfn); phys_avail[phys_avail_cnt+1] = ia64_ptob(pfn1); phys_avail_cnt += 2; } } else { /* * Just load this cluster as one chunk. */ #ifdef DEBUG_MD printf("Loading descriptor %d: 0x%lx / 0x%lx\n", i, pfn0, pfn1); #endif phys_avail[phys_avail_cnt] = ia64_ptob(pfn0); phys_avail[phys_avail_cnt+1] = ia64_ptob(pfn1); phys_avail_cnt += 2; } } phys_avail[phys_avail_cnt] = 0; Maxmem = physmem; init_param2(physmem); /* * Initialize error message buffer (at end of core). */ msgbufp = (struct msgbuf *)pmap_steal_memory(msgbufsize); msgbufinit(msgbufp, msgbufsize); proc_linkup0(&proc0, &thread0); /* * Init mapping for kernel stack for proc 0 */ thread0.td_kstack = pmap_steal_memory(KSTACK_PAGES * PAGE_SIZE); thread0.td_kstack_pages = KSTACK_PAGES; mutex_init(); /* * Initialize the rest of proc 0's PCB. * * Set the kernel sp, reserving space for an (empty) trapframe, * and make proc0's trapframe pointer point to it for sanity. * Initialise proc0's backing store to start after u area. */ cpu_thread_alloc(&thread0); thread0.td_frame->tf_flags = FRAME_SYSCALL; thread0.td_pcb->pcb_special.sp = (u_int64_t)thread0.td_frame - 16; thread0.td_pcb->pcb_special.bspstore = thread0.td_kstack; /* * Initialize the virtual memory system. */ pmap_bootstrap(); /* * Initialize debuggers, and break into them if appropriate. */ kdb_init(); #ifdef KDB if (boothowto & RB_KDB) kdb_enter(KDB_WHY_BOOTFLAGS, "Boot flags requested debugger\n"); #endif ia64_set_tpr(0); ia64_srlz_d(); ret.bspstore = thread0.td_pcb->pcb_special.bspstore; ret.sp = thread0.td_pcb->pcb_special.sp; return (ret); } uint64_t ia64_get_hcdp(void) { return (bootinfo->bi_hcdp); } void bzero(void *buf, size_t len) { caddr_t 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--; } } u_int ia64_itc_freq(void) { return (itc_freq); } void DELAY(int n) { u_int64_t start, end, now; sched_pin(); start = ia64_get_itc(); end = start + itc_freq * n; /* printf("DELAY from 0x%lx to 0x%lx\n", start, end); */ do { now = ia64_get_itc(); } while (now < end || (now > start && end < start)); sched_unpin(); } /* * Send an interrupt (signal) to a process. */ void sendsig(sig_t catcher, ksiginfo_t *ksi, sigset_t *mask) { struct proc *p; struct thread *td; struct trapframe *tf; struct sigacts *psp; struct sigframe sf, *sfp; u_int64_t sbs, sp; int oonstack; int sig; u_long code; td = curthread; p = td->td_proc; PROC_LOCK_ASSERT(p, MA_OWNED); sig = ksi->ksi_signo; code = ksi->ksi_code; psp = p->p_sigacts; mtx_assert(&psp->ps_mtx, MA_OWNED); tf = td->td_frame; sp = tf->tf_special.sp; oonstack = sigonstack(sp); sbs = 0; /* save user context */ bzero(&sf, sizeof(struct sigframe)); 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; /* * Allocate and validate space for the signal handler * context. Note that if the stack is in P0 space, the * call to grow() is a nop, and the useracc() check * will fail if the process has not already allocated * the space with a `brk'. */ if ((td->td_pflags & TDP_ALTSTACK) != 0 && !oonstack && SIGISMEMBER(psp->ps_sigonstack, sig)) { sbs = (u_int64_t)td->td_sigstk.ss_sp; sbs = (sbs + 15) & ~15; sfp = (struct sigframe *)(sbs + td->td_sigstk.ss_size); #if defined(COMPAT_43) td->td_sigstk.ss_flags |= SS_ONSTACK; #endif } else sfp = (struct sigframe *)sp; sfp = (struct sigframe *)((u_int64_t)(sfp - 1) & ~15); /* Fill in the siginfo structure for POSIX handlers. */ if (SIGISMEMBER(psp->ps_siginfo, sig)) { sf.sf_si = ksi->ksi_info; sf.sf_si.si_signo = sig; /* * XXX this shouldn't be here after code in trap.c * is fixed */ sf.sf_si.si_addr = (void*)tf->tf_special.ifa; code = (u_int64_t)&sfp->sf_si; } mtx_unlock(&psp->ps_mtx); PROC_UNLOCK(p); get_mcontext(td, &sf.sf_uc.uc_mcontext, 0); /* Copy the frame out to userland. */ if (copyout(&sf, sfp, sizeof(sf)) != 0) { /* * Process has trashed its stack; give it an illegal * instruction to halt it in its tracks. */ PROC_LOCK(p); sigexit(td, SIGILL); return; } if ((tf->tf_flags & FRAME_SYSCALL) == 0) { tf->tf_special.psr &= ~IA64_PSR_RI; tf->tf_special.iip = ia64_get_k5() + ((uint64_t)break_sigtramp - (uint64_t)ia64_gateway_page); } else tf->tf_special.iip = ia64_get_k5() + ((uint64_t)epc_sigtramp - (uint64_t)ia64_gateway_page); /* * Setup the trapframe to return to the signal trampoline. We pass * information to the trampoline in the following registers: * * gp new backing store or NULL * r8 signal number * r9 signal code or siginfo pointer * r10 signal handler (function descriptor) */ tf->tf_special.sp = (u_int64_t)sfp - 16; tf->tf_special.gp = sbs; tf->tf_special.bspstore = sf.sf_uc.uc_mcontext.mc_special.bspstore; tf->tf_special.ndirty = 0; tf->tf_special.rnat = sf.sf_uc.uc_mcontext.mc_special.rnat; tf->tf_scratch.gr8 = sig; tf->tf_scratch.gr9 = code; tf->tf_scratch.gr10 = (u_int64_t)catcher; PROC_LOCK(p); mtx_lock(&psp->ps_mtx); } /* * System call to cleanup state after a signal * has been taken. Reset signal mask and * stack state from context left by sendsig (above). * Return to previous pc and psl as specified by * context left by sendsig. Check carefully to * make sure that the user has not modified the * state to gain improper privileges. * * MPSAFE */ int sigreturn(struct thread *td, struct sigreturn_args /* { ucontext_t *sigcntxp; } */ *uap) { ucontext_t uc; struct trapframe *tf; struct pcb *pcb; tf = td->td_frame; pcb = td->td_pcb; /* * Fetch the entire context structure at once for speed. * We don't use a normal argument to simplify RSE handling. */ if (copyin(uap->sigcntxp, (caddr_t)&uc, sizeof(uc))) return (EFAULT); set_mcontext(td, &uc.uc_mcontext); #if defined(COMPAT_43) if (sigonstack(tf->tf_special.sp)) td->td_sigstk.ss_flags |= SS_ONSTACK; else td->td_sigstk.ss_flags &= ~SS_ONSTACK; #endif kern_sigprocmask(td, SIG_SETMASK, &uc.uc_sigmask, NULL, 0); return (EJUSTRETURN); } #ifdef COMPAT_FREEBSD4 int freebsd4_sigreturn(struct thread *td, struct freebsd4_sigreturn_args *uap) { return 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_special = tf->tf_special; pcb->pcb_special.__spare = ~0UL; /* XXX see unwind.c */ save_callee_saved(&pcb->pcb_preserved); save_callee_saved_fp(&pcb->pcb_preserved_fp); } int ia64_flush_dirty(struct thread *td, struct _special *r) { struct iovec iov; struct uio uio; uint64_t bspst, kstk, rnat; int error, locked; if (r->ndirty == 0) return (0); kstk = td->td_kstack + (r->bspstore & 0x1ffUL); if (td == curthread) { __asm __volatile("mov ar.rsc=0;;"); __asm __volatile("mov %0=ar.bspstore" : "=r"(bspst)); /* Make sure we have all the user registers written out. */ if (bspst - kstk < r->ndirty) { __asm __volatile("flushrs;;"); __asm __volatile("mov %0=ar.bspstore" : "=r"(bspst)); } __asm __volatile("mov %0=ar.rnat;;" : "=r"(rnat)); __asm __volatile("mov ar.rsc=3"); error = copyout((void*)kstk, (void*)r->bspstore, r->ndirty); kstk += r->ndirty; r->rnat = (bspst > kstk && (bspst & 0x1ffL) < (kstk & 0x1ffL)) ? *(uint64_t*)(kstk | 0x1f8L) : rnat; } else { locked = PROC_LOCKED(td->td_proc); if (!locked) PHOLD(td->td_proc); iov.iov_base = (void*)(uintptr_t)kstk; iov.iov_len = r->ndirty; uio.uio_iov = &iov; uio.uio_iovcnt = 1; uio.uio_offset = r->bspstore; uio.uio_resid = r->ndirty; uio.uio_segflg = UIO_SYSSPACE; uio.uio_rw = UIO_WRITE; uio.uio_td = td; error = proc_rwmem(td->td_proc, &uio); /* * XXX proc_rwmem() doesn't currently return ENOSPC, * so I think it can bogusly return 0. Neither do * we allow short writes. */ if (uio.uio_resid != 0 && error == 0) error = ENOSPC; if (!locked) PRELE(td->td_proc); } r->bspstore += r->ndirty; r->ndirty = 0; return (error); } int get_mcontext(struct thread *td, mcontext_t *mc, int flags) { struct trapframe *tf; int error; tf = td->td_frame; bzero(mc, sizeof(*mc)); mc->mc_special = tf->tf_special; error = ia64_flush_dirty(td, &mc->mc_special); if (tf->tf_flags & FRAME_SYSCALL) { mc->mc_flags |= _MC_FLAGS_SYSCALL_CONTEXT; mc->mc_scratch = tf->tf_scratch; if (flags & GET_MC_CLEAR_RET) { mc->mc_scratch.gr8 = 0; mc->mc_scratch.gr9 = 0; mc->mc_scratch.gr10 = 0; mc->mc_scratch.gr11 = 0; } } else { mc->mc_flags |= _MC_FLAGS_ASYNC_CONTEXT; mc->mc_scratch = tf->tf_scratch; mc->mc_scratch_fp = tf->tf_scratch_fp; /* * XXX If the thread never used the high FP registers, we * probably shouldn't waste time saving them. */ ia64_highfp_save(td); mc->mc_flags |= _MC_FLAGS_HIGHFP_VALID; mc->mc_high_fp = td->td_pcb->pcb_high_fp; } save_callee_saved(&mc->mc_preserved); save_callee_saved_fp(&mc->mc_preserved_fp); return (error); } int set_mcontext(struct thread *td, const mcontext_t *mc) { struct _special s; struct trapframe *tf; uint64_t psrmask; tf = td->td_frame; KASSERT((tf->tf_special.ndirty & ~PAGE_MASK) == 0, ("Whoa there! We have more than 8KB of dirty registers!")); s = mc->mc_special; /* * Only copy the user mask and the restart instruction bit from * the new context. */ psrmask = IA64_PSR_BE | IA64_PSR_UP | IA64_PSR_AC | IA64_PSR_MFL | IA64_PSR_MFH | IA64_PSR_RI; s.psr = (tf->tf_special.psr & ~psrmask) | (s.psr & psrmask); /* We don't have any dirty registers of the new context. */ s.ndirty = 0; if (mc->mc_flags & _MC_FLAGS_ASYNC_CONTEXT) { /* * We can get an async context passed to us while we * entered the kernel through a syscall: sigreturn(2) * takes contexts that could previously be the result of * a trap or interrupt. * Hence, we cannot assert that the trapframe is not * a syscall frame, but we can assert that it's at * least an expected syscall. */ if (tf->tf_flags & FRAME_SYSCALL) { KASSERT(tf->tf_scratch.gr15 == SYS_sigreturn, ("foo")); tf->tf_flags &= ~FRAME_SYSCALL; } tf->tf_scratch = mc->mc_scratch; tf->tf_scratch_fp = mc->mc_scratch_fp; if (mc->mc_flags & _MC_FLAGS_HIGHFP_VALID) td->td_pcb->pcb_high_fp = mc->mc_high_fp; } else { KASSERT((tf->tf_flags & FRAME_SYSCALL) != 0, ("foo")); if ((mc->mc_flags & _MC_FLAGS_SYSCALL_CONTEXT) == 0) { s.cfm = tf->tf_special.cfm; s.iip = tf->tf_special.iip; tf->tf_scratch.gr15 = 0; /* Clear syscall nr. */ } else tf->tf_scratch = mc->mc_scratch; } tf->tf_special = s; restore_callee_saved(&mc->mc_preserved); restore_callee_saved_fp(&mc->mc_preserved_fp); return (0); } /* * Clear registers on exec. */ void exec_setregs(struct thread *td, struct image_params *imgp, u_long stack) { struct trapframe *tf; uint64_t *ksttop, *kst; tf = td->td_frame; ksttop = (uint64_t*)(td->td_kstack + tf->tf_special.ndirty + (tf->tf_special.bspstore & 0x1ffUL)); /* * We can ignore up to 8KB of dirty registers by masking off the * lower 13 bits in exception_restore() or epc_syscall(). This * should be enough for a couple of years, but if there are more * than 8KB of dirty registers, we lose track of the bottom of * the kernel stack. The solution is to copy the active part of * the kernel stack down 1 page (or 2, but not more than that) * so that we always have less than 8KB of dirty registers. */ KASSERT((tf->tf_special.ndirty & ~PAGE_MASK) == 0, ("Whoa there! We have more than 8KB of dirty registers!")); bzero(&tf->tf_special, sizeof(tf->tf_special)); if ((tf->tf_flags & FRAME_SYSCALL) == 0) { /* break syscalls. */ bzero(&tf->tf_scratch, sizeof(tf->tf_scratch)); bzero(&tf->tf_scratch_fp, sizeof(tf->tf_scratch_fp)); tf->tf_special.cfm = (1UL<<63) | (3UL<<7) | 3UL; tf->tf_special.bspstore = IA64_BACKINGSTORE; /* * Copy the arguments onto the kernel register stack so that * they get loaded by the loadrs instruction. Skip over the * NaT collection points. */ kst = ksttop - 1; if (((uintptr_t)kst & 0x1ff) == 0x1f8) *kst-- = 0; *kst-- = 0; if (((uintptr_t)kst & 0x1ff) == 0x1f8) *kst-- = 0; *kst-- = imgp->ps_strings; if (((uintptr_t)kst & 0x1ff) == 0x1f8) *kst-- = 0; *kst = stack; tf->tf_special.ndirty = (ksttop - kst) << 3; } else { /* epc syscalls (default). */ tf->tf_special.cfm = (3UL<<62) | (3UL<<7) | 3UL; tf->tf_special.bspstore = IA64_BACKINGSTORE + 24; /* * Write values for out0, out1 and out2 to the user's backing * store and arrange for them to be restored into the user's * initial register frame. * Assumes that (bspstore & 0x1f8) < 0x1e0. */ suword((caddr_t)tf->tf_special.bspstore - 24, stack); suword((caddr_t)tf->tf_special.bspstore - 16, imgp->ps_strings); suword((caddr_t)tf->tf_special.bspstore - 8, 0); } tf->tf_special.iip = imgp->entry_addr; tf->tf_special.sp = (stack & ~15) - 16; tf->tf_special.rsc = 0xf; tf->tf_special.fpsr = IA64_FPSR_DEFAULT; tf->tf_special.psr = IA64_PSR_IC | IA64_PSR_I | IA64_PSR_IT | IA64_PSR_DT | IA64_PSR_RT | IA64_PSR_DFH | IA64_PSR_BN | IA64_PSR_CPL_USER; } int ptrace_set_pc(struct thread *td, unsigned long addr) { uint64_t slot; switch (addr & 0xFUL) { case 0: slot = IA64_PSR_RI_0; break; case 1: /* XXX we need to deal with MLX bundles here */ slot = IA64_PSR_RI_1; break; case 2: slot = IA64_PSR_RI_2; break; default: return (EINVAL); } td->td_frame->tf_special.iip = addr & ~0x0FULL; td->td_frame->tf_special.psr = (td->td_frame->tf_special.psr & ~IA64_PSR_RI) | slot; return (0); } int ptrace_single_step(struct thread *td) { struct trapframe *tf; /* * There's no way to set single stepping when we're leaving the * kernel through the EPC syscall path. The way we solve this is * by enabling the lower-privilege trap so that we re-enter the * kernel as soon as the privilege level changes. See trap.c for * how we proceed from there. */ tf = td->td_frame; if (tf->tf_flags & FRAME_SYSCALL) tf->tf_special.psr |= IA64_PSR_LP; else tf->tf_special.psr |= IA64_PSR_SS; return (0); } int ptrace_clear_single_step(struct thread *td) { struct trapframe *tf; /* * Clear any and all status bits we may use to implement single * stepping. */ tf = td->td_frame; tf->tf_special.psr &= ~IA64_PSR_SS; tf->tf_special.psr &= ~IA64_PSR_LP; tf->tf_special.psr &= ~IA64_PSR_TB; return (0); } int fill_regs(struct thread *td, struct reg *regs) { struct trapframe *tf; tf = td->td_frame; regs->r_special = tf->tf_special; regs->r_scratch = tf->tf_scratch; save_callee_saved(®s->r_preserved); return (0); } int set_regs(struct thread *td, struct reg *regs) { struct trapframe *tf; int error; tf = td->td_frame; error = ia64_flush_dirty(td, &tf->tf_special); if (!error) { tf->tf_special = regs->r_special; tf->tf_special.bspstore += tf->tf_special.ndirty; tf->tf_special.ndirty = 0; tf->tf_scratch = regs->r_scratch; restore_callee_saved(®s->r_preserved); } return (error); } int fill_dbregs(struct thread *td, struct dbreg *dbregs) { return (ENOSYS); } int set_dbregs(struct thread *td, struct dbreg *dbregs) { return (ENOSYS); } int fill_fpregs(struct thread *td, struct fpreg *fpregs) { struct trapframe *frame = td->td_frame; struct pcb *pcb = td->td_pcb; /* Save the high FP registers. */ ia64_highfp_save(td); fpregs->fpr_scratch = frame->tf_scratch_fp; save_callee_saved_fp(&fpregs->fpr_preserved); fpregs->fpr_high = pcb->pcb_high_fp; return (0); } int set_fpregs(struct thread *td, struct fpreg *fpregs) { struct trapframe *frame = td->td_frame; struct pcb *pcb = td->td_pcb; /* Throw away the high FP registers (should be redundant). */ ia64_highfp_drop(td); frame->tf_scratch_fp = fpregs->fpr_scratch; restore_callee_saved_fp(&fpregs->fpr_preserved); pcb->pcb_high_fp = fpregs->fpr_high; return (0); } void ia64_sync_icache(vm_offset_t va, vm_offset_t sz) { vm_offset_t lim; if (!ia64_sync_icache_needed) return; lim = va + sz; while (va < lim) { ia64_fc_i(va); va += 32; /* XXX */ } ia64_sync_i(); ia64_srlz_i(); } Index: projects/largeSMP/sys/ia64/ia64/pal.S =================================================================== --- projects/largeSMP/sys/ia64/ia64/pal.S (revision 222782) +++ projects/largeSMP/sys/ia64/ia64/pal.S (revision 222783) @@ -1,119 +1,116 @@ /*- * Copyright (c) 2000-2001 Doug Rabson * 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. * * $FreeBSD$ */ #include .data .global ia64_pal_entry ia64_pal_entry: .quad 0 .text /* * struct ia64_pal_result ia64_call_pal_static(u_int64_t proc, * u_int64_t arg1, u_int64_t arg2, u_int64_t arg3) */ ENTRY(ia64_call_pal_static, 4) - - .regstk 4,5,0,0 + + .regstk 4,4,0,0 palret = loc0 entry = loc1 rpsave = loc2 pfssave = loc3 -psrsave = loc4 - alloc pfssave=ar.pfs,4,5,0,0 + alloc pfssave=ar.pfs,4,4,0,0 ;; mov rpsave=rp - movl entry=@gprel(ia64_pal_entry) + 1: mov palret=ip // for return address ;; add entry=entry,gp - mov psrsave=psr + add palret=2f-1b,palret // calculate return address mov r28=in0 // procedure number - ;; - ld8 entry=[entry] // read entry point mov r29=in1 // copy arguments mov r30=in2 mov r31=in3 ;; - mov b6=entry - add palret=2f-1b,palret // calculate return address - ;; + ld8 entry=[entry] // read entry point mov b0=palret - rsm psr.i // disable interrupts ;; + mov b6=entry + ;; br.cond.sptk b6 // call into firmware -2: mov psr.l=psrsave + ;; +2: mov rp=rpsave mov ar.pfs=pfssave ;; - srlz.d br.ret.sptk rp + ;; END(ia64_call_pal_static) /* * struct ia64_pal_result ia64_call_pal_stacked(u_int64_t proc, * u_int64_t arg1, u_int64_t arg2, u_int64_t arg3) */ ENTRY(ia64_call_pal_stacked, 4) .regstk 4,4,4,0 entry = loc0 rpsave = loc1 pfssave = loc2 psrsave = loc3 alloc pfssave=ar.pfs,4,4,4,0 ;; mov rpsave=rp movl entry=@gprel(ia64_pal_entry) ;; add entry=entry,gp mov psrsave=psr mov r28=in0 // procedure number mov out0=in0 ;; ld8 entry=[entry] // read entry point mov out1=in1 // copy arguments mov out2=in2 mov out3=in3 ;; mov b6=entry ;; rsm psr.i // disable interrupts ;; br.call.sptk.many rp=b6 // call into firmware mov psr.l=psrsave mov rp=rpsave mov ar.pfs=pfssave ;; srlz.d br.ret.sptk rp END(ia64_call_pal_stacked) Index: projects/largeSMP/sys/kern/subr_rman.c =================================================================== --- projects/largeSMP/sys/kern/subr_rman.c (revision 222782) +++ projects/largeSMP/sys/kern/subr_rman.c (revision 222783) @@ -1,1154 +1,1155 @@ /*- * Copyright 1998 Massachusetts Institute of Technology * * Permission to use, copy, modify, and distribute this software and * its documentation for any purpose and without fee is hereby * granted, provided that both the above copyright notice and this * permission notice appear in all copies, that both the above * copyright notice and this permission notice appear in all * supporting documentation, and that the name of M.I.T. not be used * in advertising or publicity pertaining to distribution of the * software without specific, written prior permission. M.I.T. makes * no representations about the suitability of this software for any * purpose. It is provided "as is" without express or implied * warranty. * * THIS SOFTWARE IS PROVIDED BY M.I.T. ``AS IS''. M.I.T. DISCLAIMS * ALL EXPRESS OR IMPLIED WARRANTIES WITH REGARD TO THIS SOFTWARE, * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT * SHALL M.I.T. 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. */ /* * The kernel resource manager. This code is responsible for keeping track * of hardware resources which are apportioned out to various drivers. * It does not actually assign those resources, and it is not expected * that end-device drivers will call into this code directly. Rather, * the code which implements the buses that those devices are attached to, * and the code which manages CPU resources, will call this code, and the * end-device drivers will make upcalls to that code to actually perform * the allocation. * * There are two sorts of resources managed by this code. The first is * the more familiar array (RMAN_ARRAY) type; resources in this class * consist of a sequence of individually-allocatable objects which have * been numbered in some well-defined order. Most of the resources * are of this type, as it is the most familiar. The second type is * called a gauge (RMAN_GAUGE), and models fungible resources (i.e., * resources in which each instance is indistinguishable from every * other instance). The principal anticipated application of gauges * is in the context of power consumption, where a bus may have a specific * power budget which all attached devices share. RMAN_GAUGE is not * implemented yet. * * For array resources, we make one simplifying assumption: two clients * sharing the same resource must use the same range of indices. That * is to say, sharing of overlapping-but-not-identical regions is not * permitted. */ #include "opt_ddb.h" #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include /* XXX debugging */ #include #include #include #ifdef DDB #include #endif /* * We use a linked list rather than a bitmap because we need to be able to * represent potentially huge objects (like all of a processor's physical * address space). That is also why the indices are defined to have type * `unsigned long' -- that being the largest integral type in ISO C (1990). * The 1999 version of C allows `long long'; we may need to switch to that * at some point in the future, particularly if we want to support 36-bit * addresses on IA32 hardware. */ struct resource_i { struct resource r_r; TAILQ_ENTRY(resource_i) r_link; LIST_ENTRY(resource_i) r_sharelink; LIST_HEAD(, resource_i) *r_sharehead; u_long r_start; /* index of the first entry in this resource */ u_long r_end; /* index of the last entry (inclusive) */ u_int r_flags; void *r_virtual; /* virtual address of this resource */ struct device *r_dev; /* device which has allocated this resource */ struct rman *r_rm; /* resource manager from whence this came */ int r_rid; /* optional rid for this resource. */ }; static int rman_debug = 0; TUNABLE_INT("debug.rman_debug", &rman_debug); SYSCTL_INT(_debug, OID_AUTO, rman_debug, CTLFLAG_RW, &rman_debug, 0, "rman debug"); #define DPRINTF(params) if (rman_debug) printf params static MALLOC_DEFINE(M_RMAN, "rman", "Resource manager"); struct rman_head rman_head; static struct mtx rman_mtx; /* mutex to protect rman_head */ static int int_rman_activate_resource(struct rman *rm, struct resource_i *r, struct resource_i **whohas); static int int_rman_deactivate_resource(struct resource_i *r); static int int_rman_release_resource(struct rman *rm, struct resource_i *r); static __inline struct resource_i * int_alloc_resource(int malloc_flag) { struct resource_i *r; r = malloc(sizeof *r, M_RMAN, malloc_flag | M_ZERO); if (r != NULL) { r->r_r.__r_i = r; } return (r); } int rman_init(struct rman *rm) { static int once = 0; if (once == 0) { once = 1; TAILQ_INIT(&rman_head); mtx_init(&rman_mtx, "rman head", NULL, MTX_DEF); } if (rm->rm_start == 0 && rm->rm_end == 0) rm->rm_end = ~0ul; if (rm->rm_type == RMAN_UNINIT) panic("rman_init"); if (rm->rm_type == RMAN_GAUGE) panic("implement RMAN_GAUGE"); TAILQ_INIT(&rm->rm_list); rm->rm_mtx = malloc(sizeof *rm->rm_mtx, M_RMAN, M_NOWAIT | M_ZERO); if (rm->rm_mtx == NULL) return ENOMEM; mtx_init(rm->rm_mtx, "rman", NULL, MTX_DEF); mtx_lock(&rman_mtx); TAILQ_INSERT_TAIL(&rman_head, rm, rm_link); mtx_unlock(&rman_mtx); return 0; } int rman_manage_region(struct rman *rm, u_long start, u_long end) { struct resource_i *r, *s, *t; DPRINTF(("rman_manage_region: <%s> request: start %#lx, end %#lx\n", rm->rm_descr, start, end)); if (start < rm->rm_start || end > rm->rm_end) return EINVAL; r = int_alloc_resource(M_NOWAIT); if (r == NULL) return ENOMEM; r->r_start = start; r->r_end = end; r->r_rm = rm; mtx_lock(rm->rm_mtx); /* Skip entries before us. */ TAILQ_FOREACH(s, &rm->rm_list, r_link) { if (s->r_end == ULONG_MAX) break; if (s->r_end + 1 >= r->r_start) break; } /* If we ran off the end of the list, insert at the tail. */ if (s == NULL) { TAILQ_INSERT_TAIL(&rm->rm_list, r, r_link); } else { /* Check for any overlap with the current region. */ if (r->r_start <= s->r_end && r->r_end >= s->r_start) return EBUSY; /* Check for any overlap with the next region. */ t = TAILQ_NEXT(s, r_link); if (t && r->r_start <= t->r_end && r->r_end >= t->r_start) return EBUSY; /* * See if this region can be merged with the next region. If * not, clear the pointer. */ if (t && (r->r_end + 1 != t->r_start || t->r_flags != 0)) t = NULL; /* See if we can merge with the current region. */ if (s->r_end + 1 == r->r_start && s->r_flags == 0) { /* Can we merge all 3 regions? */ if (t != NULL) { s->r_end = t->r_end; TAILQ_REMOVE(&rm->rm_list, t, r_link); free(r, M_RMAN); free(t, M_RMAN); } else { s->r_end = r->r_end; free(r, M_RMAN); } } else if (t != NULL) { /* Can we merge with just the next region? */ t->r_start = r->r_start; free(r, M_RMAN); } else if (s->r_end < r->r_start) { TAILQ_INSERT_AFTER(&rm->rm_list, s, r, r_link); } else { TAILQ_INSERT_BEFORE(s, r, r_link); } } mtx_unlock(rm->rm_mtx); return 0; } int rman_init_from_resource(struct rman *rm, struct resource *r) { int rv; if ((rv = rman_init(rm)) != 0) return (rv); return (rman_manage_region(rm, r->__r_i->r_start, r->__r_i->r_end)); } int rman_fini(struct rman *rm) { struct resource_i *r; mtx_lock(rm->rm_mtx); TAILQ_FOREACH(r, &rm->rm_list, r_link) { if (r->r_flags & RF_ALLOCATED) { mtx_unlock(rm->rm_mtx); return EBUSY; } } /* * There really should only be one of these if we are in this * state and the code is working properly, but it can't hurt. */ while (!TAILQ_EMPTY(&rm->rm_list)) { r = TAILQ_FIRST(&rm->rm_list); TAILQ_REMOVE(&rm->rm_list, r, r_link); free(r, M_RMAN); } mtx_unlock(rm->rm_mtx); mtx_lock(&rman_mtx); TAILQ_REMOVE(&rman_head, rm, rm_link); mtx_unlock(&rman_mtx); mtx_destroy(rm->rm_mtx); free(rm->rm_mtx, M_RMAN); return 0; } int rman_first_free_region(struct rman *rm, u_long *start, u_long *end) { struct resource_i *r; mtx_lock(rm->rm_mtx); TAILQ_FOREACH(r, &rm->rm_list, r_link) { if (!(r->r_flags & RF_ALLOCATED)) { *start = r->r_start; *end = r->r_end; mtx_unlock(rm->rm_mtx); return (0); } } mtx_unlock(rm->rm_mtx); return (ENOENT); } int rman_last_free_region(struct rman *rm, u_long *start, u_long *end) { struct resource_i *r; mtx_lock(rm->rm_mtx); TAILQ_FOREACH_REVERSE(r, &rm->rm_list, resource_head, r_link) { if (!(r->r_flags & RF_ALLOCATED)) { *start = r->r_start; *end = r->r_end; mtx_unlock(rm->rm_mtx); return (0); } } mtx_unlock(rm->rm_mtx); return (ENOENT); } /* Shrink or extend one or both ends of an allocated resource. */ int rman_adjust_resource(struct resource *rr, u_long start, u_long end) { struct resource_i *r, *s, *t, *new; struct rman *rm; /* Not supported for shared resources. */ r = rr->__r_i; if (r->r_flags & (RF_TIMESHARE | RF_SHAREABLE)) return (EINVAL); /* * This does not support wholesale moving of a resource. At * least part of the desired new range must overlap with the * existing resource. */ if (end < r->r_start || r->r_end < start) return (EINVAL); /* * Find the two resource regions immediately adjacent to the * allocated resource. */ rm = r->r_rm; mtx_lock(rm->rm_mtx); #ifdef INVARIANTS TAILQ_FOREACH(s, &rm->rm_list, r_link) { if (s == r) break; } if (s == NULL) panic("resource not in list"); #endif s = TAILQ_PREV(r, resource_head, r_link); t = TAILQ_NEXT(r, r_link); KASSERT(s == NULL || s->r_end + 1 == r->r_start, ("prev resource mismatch")); KASSERT(t == NULL || r->r_end + 1 == t->r_start, ("next resource mismatch")); /* * See if the changes are permitted. Shrinking is always allowed, * but growing requires sufficient room in the adjacent region. */ if (start < r->r_start && (s == NULL || (s->r_flags & RF_ALLOCATED) || s->r_start > start)) { mtx_unlock(rm->rm_mtx); return (EBUSY); } if (end > r->r_end && (t == NULL || (t->r_flags & RF_ALLOCATED) || t->r_end < end)) { mtx_unlock(rm->rm_mtx); return (EBUSY); } /* * While holding the lock, grow either end of the resource as * needed and shrink either end if the shrinking does not require * allocating a new resource. We can safely drop the lock and then * insert a new range to handle the shrinking case afterwards. */ if (start < r->r_start || (start > r->r_start && s != NULL && !(s->r_flags & RF_ALLOCATED))) { KASSERT(s->r_flags == 0, ("prev is busy")); r->r_start = start; if (s->r_start == start) { TAILQ_REMOVE(&rm->rm_list, s, r_link); free(s, M_RMAN); } else s->r_end = start - 1; } if (end > r->r_end || (end < r->r_end && t != NULL && !(t->r_flags & RF_ALLOCATED))) { KASSERT(t->r_flags == 0, ("next is busy")); r->r_end = end; if (t->r_end == end) { TAILQ_REMOVE(&rm->rm_list, t, r_link); free(t, M_RMAN); } else t->r_start = end + 1; } mtx_unlock(rm->rm_mtx); /* * Handle the shrinking cases that require allocating a new * resource to hold the newly-free region. We have to recheck * if we still need this new region after acquiring the lock. */ if (start > r->r_start) { new = int_alloc_resource(M_WAITOK); new->r_start = r->r_start; new->r_end = start - 1; new->r_rm = rm; mtx_lock(rm->rm_mtx); r->r_start = start; s = TAILQ_PREV(r, resource_head, r_link); if (s != NULL && !(s->r_flags & RF_ALLOCATED)) { s->r_end = start - 1; free(new, M_RMAN); } else TAILQ_INSERT_BEFORE(r, new, r_link); mtx_unlock(rm->rm_mtx); } if (end < r->r_end) { new = int_alloc_resource(M_WAITOK); new->r_start = end + 1; new->r_end = r->r_end; new->r_rm = rm; mtx_lock(rm->rm_mtx); r->r_end = end; t = TAILQ_NEXT(r, r_link); if (t != NULL && !(t->r_flags & RF_ALLOCATED)) { t->r_start = end + 1; free(new, M_RMAN); } else TAILQ_INSERT_AFTER(&rm->rm_list, r, new, r_link); mtx_unlock(rm->rm_mtx); } return (0); } struct resource * rman_reserve_resource_bound(struct rman *rm, u_long start, u_long end, u_long count, u_long bound, u_int flags, struct device *dev) { u_int want_activate; struct resource_i *r, *s, *rv; u_long rstart, rend, amask, bmask; rv = NULL; DPRINTF(("rman_reserve_resource_bound: <%s> request: [%#lx, %#lx], " "length %#lx, flags %u, device %s\n", rm->rm_descr, start, end, count, flags, dev == NULL ? "" : device_get_nameunit(dev))); want_activate = (flags & RF_ACTIVE); flags &= ~RF_ACTIVE; mtx_lock(rm->rm_mtx); for (r = TAILQ_FIRST(&rm->rm_list); r && r->r_end < start; r = TAILQ_NEXT(r, r_link)) ; if (r == NULL) { DPRINTF(("could not find a region\n")); goto out; } amask = (1ul << RF_ALIGNMENT(flags)) - 1; /* If bound is 0, bmask will also be 0 */ bmask = ~(bound - 1); /* * First try to find an acceptable totally-unshared region. */ for (s = r; s; s = TAILQ_NEXT(s, r_link)) { DPRINTF(("considering [%#lx, %#lx]\n", s->r_start, s->r_end)); if (s->r_start + count - 1 > end) { DPRINTF(("s->r_start (%#lx) + count - 1> end (%#lx)\n", s->r_start, end)); break; } if (s->r_flags & RF_ALLOCATED) { DPRINTF(("region is allocated\n")); continue; } rstart = ulmax(s->r_start, start); /* * Try to find a region by adjusting to boundary and alignment * until both conditions are satisfied. This is not an optimal * algorithm, but in most cases it isn't really bad, either. */ do { rstart = (rstart + amask) & ~amask; if (((rstart ^ (rstart + count - 1)) & bmask) != 0) rstart += bound - (rstart & ~bmask); } while ((rstart & amask) != 0 && rstart < end && rstart < s->r_end); rend = ulmin(s->r_end, ulmax(rstart + count - 1, end)); if (rstart > rend) { DPRINTF(("adjusted start exceeds end\n")); continue; } DPRINTF(("truncated region: [%#lx, %#lx]; size %#lx (requested %#lx)\n", rstart, rend, (rend - rstart + 1), count)); if ((rend - rstart + 1) >= count) { DPRINTF(("candidate region: [%#lx, %#lx], size %#lx\n", rstart, rend, (rend - rstart + 1))); if ((s->r_end - s->r_start + 1) == count) { DPRINTF(("candidate region is entire chunk\n")); rv = s; rv->r_flags |= RF_ALLOCATED | flags; rv->r_dev = dev; goto out; } /* * If s->r_start < rstart and * s->r_end > rstart + count - 1, then * we need to split the region into three pieces * (the middle one will get returned to the user). * Otherwise, we are allocating at either the * beginning or the end of s, so we only need to * split it in two. The first case requires * two new allocations; the second requires but one. */ rv = int_alloc_resource(M_NOWAIT); if (rv == NULL) goto out; rv->r_start = rstart; rv->r_end = rstart + count - 1; rv->r_flags = flags | RF_ALLOCATED; rv->r_dev = dev; rv->r_rm = rm; if (s->r_start < rv->r_start && s->r_end > rv->r_end) { DPRINTF(("splitting region in three parts: " "[%#lx, %#lx]; [%#lx, %#lx]; [%#lx, %#lx]\n", s->r_start, rv->r_start - 1, rv->r_start, rv->r_end, rv->r_end + 1, s->r_end)); /* * We are allocating in the middle. */ r = int_alloc_resource(M_NOWAIT); if (r == NULL) { free(rv, M_RMAN); rv = NULL; goto out; } r->r_start = rv->r_end + 1; r->r_end = s->r_end; r->r_flags = s->r_flags; r->r_rm = rm; s->r_end = rv->r_start - 1; TAILQ_INSERT_AFTER(&rm->rm_list, s, rv, r_link); TAILQ_INSERT_AFTER(&rm->rm_list, rv, r, r_link); } else if (s->r_start == rv->r_start) { DPRINTF(("allocating from the beginning\n")); /* * We are allocating at the beginning. */ s->r_start = rv->r_end + 1; TAILQ_INSERT_BEFORE(s, rv, r_link); } else { DPRINTF(("allocating at the end\n")); /* * We are allocating at the end. */ s->r_end = rv->r_start - 1; TAILQ_INSERT_AFTER(&rm->rm_list, s, rv, r_link); } goto out; } } /* * Now find an acceptable shared region, if the client's requirements * allow sharing. By our implementation restriction, a candidate * region must match exactly by both size and sharing type in order * to be considered compatible with the client's request. (The * former restriction could probably be lifted without too much * additional work, but this does not seem warranted.) */ DPRINTF(("no unshared regions found\n")); if ((flags & (RF_SHAREABLE | RF_TIMESHARE)) == 0) goto out; for (s = r; s; s = TAILQ_NEXT(s, r_link)) { if (s->r_start > end) break; if ((s->r_flags & flags) != flags) continue; rstart = ulmax(s->r_start, start); rend = ulmin(s->r_end, ulmax(start + count - 1, end)); if (s->r_start >= start && s->r_end <= end && (s->r_end - s->r_start + 1) == count && (s->r_start & amask) == 0 && ((s->r_start ^ s->r_end) & bmask) == 0) { rv = int_alloc_resource(M_NOWAIT); if (rv == NULL) goto out; rv->r_start = s->r_start; rv->r_end = s->r_end; rv->r_flags = s->r_flags & (RF_ALLOCATED | RF_SHAREABLE | RF_TIMESHARE); rv->r_dev = dev; rv->r_rm = rm; if (s->r_sharehead == NULL) { s->r_sharehead = malloc(sizeof *s->r_sharehead, M_RMAN, M_NOWAIT | M_ZERO); if (s->r_sharehead == NULL) { free(rv, M_RMAN); rv = NULL; goto out; } LIST_INIT(s->r_sharehead); LIST_INSERT_HEAD(s->r_sharehead, s, r_sharelink); s->r_flags |= RF_FIRSTSHARE; } rv->r_sharehead = s->r_sharehead; LIST_INSERT_HEAD(s->r_sharehead, rv, r_sharelink); goto out; } } /* * We couldn't find anything. */ out: /* * If the user specified RF_ACTIVE in the initial flags, * which is reflected in `want_activate', we attempt to atomically * activate the resource. If this fails, we release the resource * and indicate overall failure. (This behavior probably doesn't * make sense for RF_TIMESHARE-type resources.) */ if (rv && want_activate) { struct resource_i *whohas; if (int_rman_activate_resource(rm, rv, &whohas)) { int_rman_release_resource(rm, rv); rv = NULL; } } mtx_unlock(rm->rm_mtx); return (rv == NULL ? NULL : &rv->r_r); } struct resource * rman_reserve_resource(struct rman *rm, u_long start, u_long end, u_long count, u_int flags, struct device *dev) { return (rman_reserve_resource_bound(rm, start, end, count, 0, flags, dev)); } static int int_rman_activate_resource(struct rman *rm, struct resource_i *r, struct resource_i **whohas) { struct resource_i *s; int ok; /* * If we are not timesharing, then there is nothing much to do. * If we already have the resource, then there is nothing at all to do. * If we are not on a sharing list with anybody else, then there is * little to do. */ if ((r->r_flags & RF_TIMESHARE) == 0 || (r->r_flags & RF_ACTIVE) != 0 || r->r_sharehead == NULL) { r->r_flags |= RF_ACTIVE; return 0; } ok = 1; for (s = LIST_FIRST(r->r_sharehead); s && ok; s = LIST_NEXT(s, r_sharelink)) { if ((s->r_flags & RF_ACTIVE) != 0) { ok = 0; *whohas = s; } } if (ok) { r->r_flags |= RF_ACTIVE; return 0; } return EBUSY; } int rman_activate_resource(struct resource *re) { int rv; struct resource_i *r, *whohas; struct rman *rm; r = re->__r_i; rm = r->r_rm; mtx_lock(rm->rm_mtx); rv = int_rman_activate_resource(rm, r, &whohas); mtx_unlock(rm->rm_mtx); return rv; } int rman_await_resource(struct resource *re, int pri, int timo) { int rv; struct resource_i *r, *whohas; struct rman *rm; r = re->__r_i; rm = r->r_rm; mtx_lock(rm->rm_mtx); for (;;) { rv = int_rman_activate_resource(rm, r, &whohas); if (rv != EBUSY) return (rv); /* returns with mutex held */ if (r->r_sharehead == NULL) panic("rman_await_resource"); whohas->r_flags |= RF_WANTED; rv = msleep(r->r_sharehead, rm->rm_mtx, pri, "rmwait", timo); if (rv) { mtx_unlock(rm->rm_mtx); return (rv); } } } static int int_rman_deactivate_resource(struct resource_i *r) { r->r_flags &= ~RF_ACTIVE; if (r->r_flags & RF_WANTED) { r->r_flags &= ~RF_WANTED; wakeup(r->r_sharehead); } return 0; } int rman_deactivate_resource(struct resource *r) { struct rman *rm; rm = r->__r_i->r_rm; mtx_lock(rm->rm_mtx); int_rman_deactivate_resource(r->__r_i); mtx_unlock(rm->rm_mtx); return 0; } static int int_rman_release_resource(struct rman *rm, struct resource_i *r) { struct resource_i *s, *t; if (r->r_flags & RF_ACTIVE) int_rman_deactivate_resource(r); /* * Check for a sharing list first. If there is one, then we don't * have to think as hard. */ if (r->r_sharehead) { /* * If a sharing list exists, then we know there are at * least two sharers. * * If we are in the main circleq, appoint someone else. */ LIST_REMOVE(r, r_sharelink); s = LIST_FIRST(r->r_sharehead); if (r->r_flags & RF_FIRSTSHARE) { s->r_flags |= RF_FIRSTSHARE; TAILQ_INSERT_BEFORE(r, s, r_link); TAILQ_REMOVE(&rm->rm_list, r, r_link); } /* * Make sure that the sharing list goes away completely * if the resource is no longer being shared at all. */ if (LIST_NEXT(s, r_sharelink) == NULL) { free(s->r_sharehead, M_RMAN); s->r_sharehead = NULL; s->r_flags &= ~RF_FIRSTSHARE; } goto out; } /* * Look at the adjacent resources in the list and see if our * segment can be merged with any of them. If either of the * resources is allocated or is not exactly adjacent then they * cannot be merged with our segment. */ s = TAILQ_PREV(r, resource_head, r_link); if (s != NULL && ((s->r_flags & RF_ALLOCATED) != 0 || s->r_end + 1 != r->r_start)) s = NULL; t = TAILQ_NEXT(r, r_link); if (t != NULL && ((t->r_flags & RF_ALLOCATED) != 0 || r->r_end + 1 != t->r_start)) t = NULL; if (s != NULL && t != NULL) { /* * Merge all three segments. */ s->r_end = t->r_end; TAILQ_REMOVE(&rm->rm_list, r, r_link); TAILQ_REMOVE(&rm->rm_list, t, r_link); free(t, M_RMAN); } else if (s != NULL) { /* * Merge previous segment with ours. */ s->r_end = r->r_end; TAILQ_REMOVE(&rm->rm_list, r, r_link); } else if (t != NULL) { /* * Merge next segment with ours. */ t->r_start = r->r_start; TAILQ_REMOVE(&rm->rm_list, r, r_link); } else { /* * At this point, we know there is nothing we * can potentially merge with, because on each * side, there is either nothing there or what is * there is still allocated. In that case, we don't * want to remove r from the list; we simply want to * change it to an unallocated region and return * without freeing anything. */ r->r_flags &= ~RF_ALLOCATED; + r->r_dev = NULL; return 0; } out: free(r, M_RMAN); return 0; } int rman_release_resource(struct resource *re) { int rv; struct resource_i *r; struct rman *rm; r = re->__r_i; rm = r->r_rm; mtx_lock(rm->rm_mtx); rv = int_rman_release_resource(rm, r); mtx_unlock(rm->rm_mtx); return (rv); } uint32_t rman_make_alignment_flags(uint32_t size) { int i; /* * Find the hightest bit set, and add one if more than one bit * set. We're effectively computing the ceil(log2(size)) here. */ for (i = 31; i > 0; i--) if ((1 << i) & size) break; if (~(1 << i) & size) i++; return(RF_ALIGNMENT_LOG2(i)); } void rman_set_start(struct resource *r, u_long start) { r->__r_i->r_start = start; } u_long rman_get_start(struct resource *r) { return (r->__r_i->r_start); } void rman_set_end(struct resource *r, u_long end) { r->__r_i->r_end = end; } u_long rman_get_end(struct resource *r) { return (r->__r_i->r_end); } u_long rman_get_size(struct resource *r) { return (r->__r_i->r_end - r->__r_i->r_start + 1); } u_int rman_get_flags(struct resource *r) { return (r->__r_i->r_flags); } void rman_set_virtual(struct resource *r, void *v) { r->__r_i->r_virtual = v; } void * rman_get_virtual(struct resource *r) { return (r->__r_i->r_virtual); } void rman_set_bustag(struct resource *r, bus_space_tag_t t) { r->r_bustag = t; } bus_space_tag_t rman_get_bustag(struct resource *r) { return (r->r_bustag); } void rman_set_bushandle(struct resource *r, bus_space_handle_t h) { r->r_bushandle = h; } bus_space_handle_t rman_get_bushandle(struct resource *r) { return (r->r_bushandle); } void rman_set_rid(struct resource *r, int rid) { r->__r_i->r_rid = rid; } int rman_get_rid(struct resource *r) { return (r->__r_i->r_rid); } void rman_set_device(struct resource *r, struct device *dev) { r->__r_i->r_dev = dev; } struct device * rman_get_device(struct resource *r) { return (r->__r_i->r_dev); } int rman_is_region_manager(struct resource *r, struct rman *rm) { return (r->__r_i->r_rm == rm); } /* * Sysctl interface for scanning the resource lists. * * We take two input parameters; the index into the list of resource * managers, and the resource offset into the list. */ static int sysctl_rman(SYSCTL_HANDLER_ARGS) { int *name = (int *)arg1; u_int namelen = arg2; int rman_idx, res_idx; struct rman *rm; struct resource_i *res; struct resource_i *sres; struct u_rman urm; struct u_resource ures; int error; if (namelen != 3) return (EINVAL); if (bus_data_generation_check(name[0])) return (EINVAL); rman_idx = name[1]; res_idx = name[2]; /* * Find the indexed resource manager */ mtx_lock(&rman_mtx); TAILQ_FOREACH(rm, &rman_head, rm_link) { if (rman_idx-- == 0) break; } mtx_unlock(&rman_mtx); if (rm == NULL) return (ENOENT); /* * If the resource index is -1, we want details on the * resource manager. */ if (res_idx == -1) { bzero(&urm, sizeof(urm)); urm.rm_handle = (uintptr_t)rm; if (rm->rm_descr != NULL) strlcpy(urm.rm_descr, rm->rm_descr, RM_TEXTLEN); urm.rm_start = rm->rm_start; urm.rm_size = rm->rm_end - rm->rm_start + 1; urm.rm_type = rm->rm_type; error = SYSCTL_OUT(req, &urm, sizeof(urm)); return (error); } /* * Find the indexed resource and return it. */ mtx_lock(rm->rm_mtx); TAILQ_FOREACH(res, &rm->rm_list, r_link) { if (res->r_sharehead != NULL) { LIST_FOREACH(sres, res->r_sharehead, r_sharelink) if (res_idx-- == 0) { res = sres; goto found; } } else if (res_idx-- == 0) goto found; } mtx_unlock(rm->rm_mtx); return (ENOENT); found: bzero(&ures, sizeof(ures)); ures.r_handle = (uintptr_t)res; ures.r_parent = (uintptr_t)res->r_rm; ures.r_device = (uintptr_t)res->r_dev; if (res->r_dev != NULL) { if (device_get_name(res->r_dev) != NULL) { snprintf(ures.r_devname, RM_TEXTLEN, "%s%d", device_get_name(res->r_dev), device_get_unit(res->r_dev)); } else { strlcpy(ures.r_devname, "nomatch", RM_TEXTLEN); } } else { ures.r_devname[0] = '\0'; } ures.r_start = res->r_start; ures.r_size = res->r_end - res->r_start + 1; ures.r_flags = res->r_flags; mtx_unlock(rm->rm_mtx); error = SYSCTL_OUT(req, &ures, sizeof(ures)); return (error); } SYSCTL_NODE(_hw_bus, OID_AUTO, rman, CTLFLAG_RD, sysctl_rman, "kernel resource manager"); #ifdef DDB static void dump_rman_header(struct rman *rm) { if (db_pager_quit) return; db_printf("rman %p: %s (0x%lx-0x%lx full range)\n", rm, rm->rm_descr, rm->rm_start, rm->rm_end); } static void dump_rman(struct rman *rm) { struct resource_i *r; const char *devname; if (db_pager_quit) return; TAILQ_FOREACH(r, &rm->rm_list, r_link) { if (r->r_dev != NULL) { devname = device_get_nameunit(r->r_dev); if (devname == NULL) devname = "nomatch"; } else devname = NULL; db_printf(" 0x%lx-0x%lx ", r->r_start, r->r_end); if (devname != NULL) db_printf("(%s)\n", devname); else db_printf("----\n"); if (db_pager_quit) return; } } DB_SHOW_COMMAND(rman, db_show_rman) { if (have_addr) { dump_rman_header((struct rman *)addr); dump_rman((struct rman *)addr); } } DB_SHOW_COMMAND(rmans, db_show_rmans) { struct rman *rm; TAILQ_FOREACH(rm, &rman_head, rm_link) { dump_rman_header(rm); } } DB_SHOW_ALL_COMMAND(rman, db_show_all_rman) { struct rman *rm; TAILQ_FOREACH(rm, &rman_head, rm_link) { dump_rman_header(rm); dump_rman(rm); } } DB_SHOW_ALIAS(allrman, db_show_all_rman); #endif Index: projects/largeSMP/sys/netinet/icmp6.h =================================================================== --- projects/largeSMP/sys/netinet/icmp6.h (revision 222782) +++ projects/largeSMP/sys/netinet/icmp6.h (revision 222783) @@ -1,741 +1,758 @@ /* $FreeBSD$ */ /* $KAME: icmp6.h,v 1.46 2001/04/27 15:09:48 itojun Exp $ */ /*- * 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. */ /*- * 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_icmp.h 8.1 (Berkeley) 6/10/93 */ #ifndef _NETINET_ICMP6_H_ #define _NETINET_ICMP6_H_ #define ICMPV6_PLD_MAXLEN 1232 /* IPV6_MMTU - sizeof(struct ip6_hdr) - sizeof(struct icmp6_hdr) */ struct icmp6_hdr { u_int8_t icmp6_type; /* type field */ u_int8_t icmp6_code; /* code field */ u_int16_t icmp6_cksum; /* checksum field */ union { u_int32_t icmp6_un_data32[1]; /* type-specific field */ u_int16_t icmp6_un_data16[2]; /* type-specific field */ u_int8_t icmp6_un_data8[4]; /* type-specific field */ } icmp6_dataun; } __packed; #define icmp6_data32 icmp6_dataun.icmp6_un_data32 #define icmp6_data16 icmp6_dataun.icmp6_un_data16 #define icmp6_data8 icmp6_dataun.icmp6_un_data8 #define icmp6_pptr icmp6_data32[0] /* parameter prob */ #define icmp6_mtu icmp6_data32[0] /* packet too big */ #define icmp6_id icmp6_data16[0] /* echo request/reply */ #define icmp6_seq icmp6_data16[1] /* echo request/reply */ #define icmp6_maxdelay icmp6_data16[0] /* mcast group membership */ #define ICMP6_DST_UNREACH 1 /* dest unreachable, codes: */ #define ICMP6_PACKET_TOO_BIG 2 /* packet too big */ #define ICMP6_TIME_EXCEEDED 3 /* time exceeded, code: */ #define ICMP6_PARAM_PROB 4 /* ip6 header bad */ #define ICMP6_ECHO_REQUEST 128 /* echo service */ #define ICMP6_ECHO_REPLY 129 /* echo reply */ #define MLD_LISTENER_QUERY 130 /* multicast listener query */ #define MLD_LISTENER_REPORT 131 /* multicast listener report */ #define MLD_LISTENER_DONE 132 /* multicast listener done */ #define MLD_LISTENER_REDUCTION MLD_LISTENER_DONE /* RFC3542 definition */ /* RFC2292 decls */ #define ICMP6_MEMBERSHIP_QUERY 130 /* group membership query */ #define ICMP6_MEMBERSHIP_REPORT 131 /* group membership report */ #define ICMP6_MEMBERSHIP_REDUCTION 132 /* group membership termination */ #ifndef _KERNEL /* the followings are for backward compatibility to old KAME apps. */ #define MLD6_LISTENER_QUERY MLD_LISTENER_QUERY #define MLD6_LISTENER_REPORT MLD_LISTENER_REPORT #define MLD6_LISTENER_DONE MLD_LISTENER_DONE #endif #define ND_ROUTER_SOLICIT 133 /* router solicitation */ #define ND_ROUTER_ADVERT 134 /* router advertisement */ #define ND_NEIGHBOR_SOLICIT 135 /* neighbor solicitation */ #define ND_NEIGHBOR_ADVERT 136 /* neighbor advertisement */ #define ND_REDIRECT 137 /* redirect */ #define ICMP6_ROUTER_RENUMBERING 138 /* router renumbering */ #define ICMP6_WRUREQUEST 139 /* who are you request */ #define ICMP6_WRUREPLY 140 /* who are you reply */ #define ICMP6_FQDN_QUERY 139 /* FQDN query */ #define ICMP6_FQDN_REPLY 140 /* FQDN reply */ #define ICMP6_NI_QUERY 139 /* node information request */ #define ICMP6_NI_REPLY 140 /* node information reply */ #define MLDV2_LISTENER_REPORT 143 /* RFC3810 listener report */ /* The definitions below are experimental. TBA */ #define MLD_MTRACE_RESP 200 /* mtrace resp (to sender) */ #define MLD_MTRACE 201 /* mtrace messages */ #ifndef _KERNEL #define MLD6_MTRACE_RESP MLD_MTRACE_RESP #define MLD6_MTRACE MLD_MTRACE #endif #define ICMP6_MAXTYPE 201 #define ICMP6_DST_UNREACH_NOROUTE 0 /* no route to destination */ #define ICMP6_DST_UNREACH_ADMIN 1 /* administratively prohibited */ #define ICMP6_DST_UNREACH_NOTNEIGHBOR 2 /* not a neighbor(obsolete) */ #define ICMP6_DST_UNREACH_BEYONDSCOPE 2 /* beyond scope of source address */ #define ICMP6_DST_UNREACH_ADDR 3 /* address unreachable */ #define ICMP6_DST_UNREACH_NOPORT 4 /* port unreachable */ #define ICMP6_TIME_EXCEED_TRANSIT 0 /* ttl==0 in transit */ #define ICMP6_TIME_EXCEED_REASSEMBLY 1 /* ttl==0 in reass */ #define ICMP6_PARAMPROB_HEADER 0 /* erroneous header field */ #define ICMP6_PARAMPROB_NEXTHEADER 1 /* unrecognized next header */ #define ICMP6_PARAMPROB_OPTION 2 /* unrecognized option */ #define ICMP6_INFOMSG_MASK 0x80 /* all informational messages */ #define ICMP6_NI_SUBJ_IPV6 0 /* Query Subject is an IPv6 address */ #define ICMP6_NI_SUBJ_FQDN 1 /* Query Subject is a Domain name */ #define ICMP6_NI_SUBJ_IPV4 2 /* Query Subject is an IPv4 address */ #define ICMP6_NI_SUCCESS 0 /* node information successful reply */ #define ICMP6_NI_REFUSED 1 /* node information request is refused */ #define ICMP6_NI_UNKNOWN 2 /* unknown Qtype */ #define ICMP6_ROUTER_RENUMBERING_COMMAND 0 /* rr command */ #define ICMP6_ROUTER_RENUMBERING_RESULT 1 /* rr result */ #define ICMP6_ROUTER_RENUMBERING_SEQNUM_RESET 255 /* rr seq num reset */ /* Used in kernel only */ #define ND_REDIRECT_ONLINK 0 /* redirect to an on-link node */ #define ND_REDIRECT_ROUTER 1 /* redirect to a better router */ /* * Multicast Listener Discovery */ struct mld_hdr { struct icmp6_hdr mld_icmp6_hdr; struct in6_addr mld_addr; /* multicast address */ } __packed; /* definitions to provide backward compatibility to old KAME applications */ #ifndef _KERNEL #define mld6_hdr mld_hdr #define mld6_type mld_type #define mld6_code mld_code #define mld6_cksum mld_cksum #define mld6_maxdelay mld_maxdelay #define mld6_reserved mld_reserved #define mld6_addr mld_addr #endif /* shortcut macro definitions */ #define mld_type mld_icmp6_hdr.icmp6_type #define mld_code mld_icmp6_hdr.icmp6_code #define mld_cksum mld_icmp6_hdr.icmp6_cksum #define mld_maxdelay mld_icmp6_hdr.icmp6_data16[0] #define mld_reserved mld_icmp6_hdr.icmp6_data16[1] #define mld_v2_reserved mld_icmp6_hdr.icmp6_data16[0] #define mld_v2_numrecs mld_icmp6_hdr.icmp6_data16[1] /* * Neighbor Discovery */ struct nd_router_solicit { /* router solicitation */ struct icmp6_hdr nd_rs_hdr; /* could be followed by options */ } __packed; #define nd_rs_type nd_rs_hdr.icmp6_type #define nd_rs_code nd_rs_hdr.icmp6_code #define nd_rs_cksum nd_rs_hdr.icmp6_cksum #define nd_rs_reserved nd_rs_hdr.icmp6_data32[0] struct nd_router_advert { /* router advertisement */ struct icmp6_hdr nd_ra_hdr; u_int32_t nd_ra_reachable; /* reachable time */ u_int32_t nd_ra_retransmit; /* retransmit timer */ /* could be followed by options */ } __packed; #define nd_ra_type nd_ra_hdr.icmp6_type #define nd_ra_code nd_ra_hdr.icmp6_code #define nd_ra_cksum nd_ra_hdr.icmp6_cksum #define nd_ra_curhoplimit nd_ra_hdr.icmp6_data8[0] #define nd_ra_flags_reserved nd_ra_hdr.icmp6_data8[1] #define ND_RA_FLAG_MANAGED 0x80 #define ND_RA_FLAG_OTHER 0x40 #define ND_RA_FLAG_HA 0x20 /* * Router preference values based on draft-draves-ipngwg-router-selection-01. * These are non-standard definitions. */ #define ND_RA_FLAG_RTPREF_MASK 0x18 /* 00011000 */ #define ND_RA_FLAG_RTPREF_HIGH 0x08 /* 00001000 */ #define ND_RA_FLAG_RTPREF_MEDIUM 0x00 /* 00000000 */ #define ND_RA_FLAG_RTPREF_LOW 0x18 /* 00011000 */ #define ND_RA_FLAG_RTPREF_RSV 0x10 /* 00010000 */ #define nd_ra_router_lifetime nd_ra_hdr.icmp6_data16[1] struct nd_neighbor_solicit { /* neighbor solicitation */ struct icmp6_hdr nd_ns_hdr; struct in6_addr nd_ns_target; /*target address */ /* could be followed by options */ } __packed; #define nd_ns_type nd_ns_hdr.icmp6_type #define nd_ns_code nd_ns_hdr.icmp6_code #define nd_ns_cksum nd_ns_hdr.icmp6_cksum #define nd_ns_reserved nd_ns_hdr.icmp6_data32[0] struct nd_neighbor_advert { /* neighbor advertisement */ struct icmp6_hdr nd_na_hdr; struct in6_addr nd_na_target; /* target address */ /* could be followed by options */ } __packed; #define nd_na_type nd_na_hdr.icmp6_type #define nd_na_code nd_na_hdr.icmp6_code #define nd_na_cksum nd_na_hdr.icmp6_cksum #define nd_na_flags_reserved nd_na_hdr.icmp6_data32[0] #if BYTE_ORDER == BIG_ENDIAN #define ND_NA_FLAG_ROUTER 0x80000000 #define ND_NA_FLAG_SOLICITED 0x40000000 #define ND_NA_FLAG_OVERRIDE 0x20000000 #else #if BYTE_ORDER == LITTLE_ENDIAN #define ND_NA_FLAG_ROUTER 0x80 #define ND_NA_FLAG_SOLICITED 0x40 #define ND_NA_FLAG_OVERRIDE 0x20 #endif #endif struct nd_redirect { /* redirect */ struct icmp6_hdr nd_rd_hdr; struct in6_addr nd_rd_target; /* target address */ struct in6_addr nd_rd_dst; /* destination address */ /* could be followed by options */ } __packed; #define nd_rd_type nd_rd_hdr.icmp6_type #define nd_rd_code nd_rd_hdr.icmp6_code #define nd_rd_cksum nd_rd_hdr.icmp6_cksum #define nd_rd_reserved nd_rd_hdr.icmp6_data32[0] struct nd_opt_hdr { /* Neighbor discovery option header */ u_int8_t nd_opt_type; u_int8_t nd_opt_len; /* followed by option specific data*/ } __packed; #define ND_OPT_SOURCE_LINKADDR 1 #define ND_OPT_TARGET_LINKADDR 2 #define ND_OPT_PREFIX_INFORMATION 3 #define ND_OPT_REDIRECTED_HEADER 4 #define ND_OPT_MTU 5 +#define ND_OPT_ROUTE_INFO 24 /* RFC 4191 */ +#define ND_OPT_RDNSS 25 /* RFC 6016 */ +#define ND_OPT_DNSSL 31 /* RFC 6016 */ -#define ND_OPT_ROUTE_INFO 200 /* draft-ietf-ipngwg-router-preference, not officially assigned yet */ - struct nd_opt_prefix_info { /* prefix information */ u_int8_t nd_opt_pi_type; u_int8_t nd_opt_pi_len; u_int8_t nd_opt_pi_prefix_len; u_int8_t nd_opt_pi_flags_reserved; u_int32_t nd_opt_pi_valid_time; u_int32_t nd_opt_pi_preferred_time; u_int32_t nd_opt_pi_reserved2; struct in6_addr nd_opt_pi_prefix; } __packed; #define ND_OPT_PI_FLAG_ONLINK 0x80 #define ND_OPT_PI_FLAG_AUTO 0x40 struct nd_opt_rd_hdr { /* redirected header */ u_int8_t nd_opt_rh_type; u_int8_t nd_opt_rh_len; u_int16_t nd_opt_rh_reserved1; u_int32_t nd_opt_rh_reserved2; /* followed by IP header and data */ } __packed; struct nd_opt_mtu { /* MTU option */ u_int8_t nd_opt_mtu_type; u_int8_t nd_opt_mtu_len; u_int16_t nd_opt_mtu_reserved; u_int32_t nd_opt_mtu_mtu; } __packed; struct nd_opt_route_info { /* route info */ u_int8_t nd_opt_rti_type; u_int8_t nd_opt_rti_len; u_int8_t nd_opt_rti_prefixlen; u_int8_t nd_opt_rti_flags; u_int32_t nd_opt_rti_lifetime; /* prefix follows */ +} __packed; + +struct nd_opt_rdnss { /* RDNSS option (RFC 6106) */ + u_int8_t nd_opt_rdnss_type; + u_int8_t nd_opt_rdnss_len; + u_int16_t nd_opt_rdnss_reserved; + u_int32_t nd_opt_rdnss_lifetime; + /* followed by list of recursive DNS servers */ +} __packed; + +struct nd_opt_dnssl { /* DNSSL option (RFC 6106) */ + u_int8_t nd_opt_dnssl_type; + u_int8_t nd_opt_dnssl_len; + u_int16_t nd_opt_dnssl_reserved; + u_int32_t nd_opt_dnssl_lifetime; + /* followed by list of DNS search domains */ } __packed; /* * icmp6 namelookup */ struct icmp6_namelookup { struct icmp6_hdr icmp6_nl_hdr; u_int8_t icmp6_nl_nonce[8]; int32_t icmp6_nl_ttl; #if 0 u_int8_t icmp6_nl_len; u_int8_t icmp6_nl_name[3]; #endif /* could be followed by options */ } __packed; /* * icmp6 node information */ struct icmp6_nodeinfo { struct icmp6_hdr icmp6_ni_hdr; u_int8_t icmp6_ni_nonce[8]; /* could be followed by reply data */ } __packed; #define ni_type icmp6_ni_hdr.icmp6_type #define ni_code icmp6_ni_hdr.icmp6_code #define ni_cksum icmp6_ni_hdr.icmp6_cksum #define ni_qtype icmp6_ni_hdr.icmp6_data16[0] #define ni_flags icmp6_ni_hdr.icmp6_data16[1] #define NI_QTYPE_NOOP 0 /* NOOP */ #define NI_QTYPE_SUPTYPES 1 /* Supported Qtypes */ #define NI_QTYPE_FQDN 2 /* FQDN (draft 04) */ #define NI_QTYPE_DNSNAME 2 /* DNS Name */ #define NI_QTYPE_NODEADDR 3 /* Node Addresses */ #define NI_QTYPE_IPV4ADDR 4 /* IPv4 Addresses */ #if BYTE_ORDER == BIG_ENDIAN #define NI_SUPTYPE_FLAG_COMPRESS 0x1 #define NI_FQDN_FLAG_VALIDTTL 0x1 #elif BYTE_ORDER == LITTLE_ENDIAN #define NI_SUPTYPE_FLAG_COMPRESS 0x0100 #define NI_FQDN_FLAG_VALIDTTL 0x0100 #endif #ifdef NAME_LOOKUPS_04 #if BYTE_ORDER == BIG_ENDIAN #define NI_NODEADDR_FLAG_LINKLOCAL 0x1 #define NI_NODEADDR_FLAG_SITELOCAL 0x2 #define NI_NODEADDR_FLAG_GLOBAL 0x4 #define NI_NODEADDR_FLAG_ALL 0x8 #define NI_NODEADDR_FLAG_TRUNCATE 0x10 #define NI_NODEADDR_FLAG_ANYCAST 0x20 /* just experimental. not in spec */ #elif BYTE_ORDER == LITTLE_ENDIAN #define NI_NODEADDR_FLAG_LINKLOCAL 0x0100 #define NI_NODEADDR_FLAG_SITELOCAL 0x0200 #define NI_NODEADDR_FLAG_GLOBAL 0x0400 #define NI_NODEADDR_FLAG_ALL 0x0800 #define NI_NODEADDR_FLAG_TRUNCATE 0x1000 #define NI_NODEADDR_FLAG_ANYCAST 0x2000 /* just experimental. not in spec */ #endif #else /* draft-ietf-ipngwg-icmp-name-lookups-05 (and later?) */ #if BYTE_ORDER == BIG_ENDIAN #define NI_NODEADDR_FLAG_TRUNCATE 0x1 #define NI_NODEADDR_FLAG_ALL 0x2 #define NI_NODEADDR_FLAG_COMPAT 0x4 #define NI_NODEADDR_FLAG_LINKLOCAL 0x8 #define NI_NODEADDR_FLAG_SITELOCAL 0x10 #define NI_NODEADDR_FLAG_GLOBAL 0x20 #define NI_NODEADDR_FLAG_ANYCAST 0x40 /* just experimental. not in spec */ #elif BYTE_ORDER == LITTLE_ENDIAN #define NI_NODEADDR_FLAG_TRUNCATE 0x0100 #define NI_NODEADDR_FLAG_ALL 0x0200 #define NI_NODEADDR_FLAG_COMPAT 0x0400 #define NI_NODEADDR_FLAG_LINKLOCAL 0x0800 #define NI_NODEADDR_FLAG_SITELOCAL 0x1000 #define NI_NODEADDR_FLAG_GLOBAL 0x2000 #define NI_NODEADDR_FLAG_ANYCAST 0x4000 /* just experimental. not in spec */ #endif #endif struct ni_reply_fqdn { u_int32_t ni_fqdn_ttl; /* TTL */ u_int8_t ni_fqdn_namelen; /* length in octets of the FQDN */ u_int8_t ni_fqdn_name[3]; /* XXX: alignment */ } __packed; /* * Router Renumbering. as router-renum-08.txt */ struct icmp6_router_renum { /* router renumbering header */ struct icmp6_hdr rr_hdr; u_int8_t rr_segnum; u_int8_t rr_flags; u_int16_t rr_maxdelay; u_int32_t rr_reserved; } __packed; #define ICMP6_RR_FLAGS_TEST 0x80 #define ICMP6_RR_FLAGS_REQRESULT 0x40 #define ICMP6_RR_FLAGS_FORCEAPPLY 0x20 #define ICMP6_RR_FLAGS_SPECSITE 0x10 #define ICMP6_RR_FLAGS_PREVDONE 0x08 #define rr_type rr_hdr.icmp6_type #define rr_code rr_hdr.icmp6_code #define rr_cksum rr_hdr.icmp6_cksum #define rr_seqnum rr_hdr.icmp6_data32[0] struct rr_pco_match { /* match prefix part */ u_int8_t rpm_code; u_int8_t rpm_len; u_int8_t rpm_ordinal; u_int8_t rpm_matchlen; u_int8_t rpm_minlen; u_int8_t rpm_maxlen; u_int16_t rpm_reserved; struct in6_addr rpm_prefix; } __packed; #define RPM_PCO_ADD 1 #define RPM_PCO_CHANGE 2 #define RPM_PCO_SETGLOBAL 3 #define RPM_PCO_MAX 4 struct rr_pco_use { /* use prefix part */ u_int8_t rpu_uselen; u_int8_t rpu_keeplen; u_int8_t rpu_ramask; u_int8_t rpu_raflags; u_int32_t rpu_vltime; u_int32_t rpu_pltime; u_int32_t rpu_flags; struct in6_addr rpu_prefix; } __packed; #define ICMP6_RR_PCOUSE_RAFLAGS_ONLINK 0x80 #define ICMP6_RR_PCOUSE_RAFLAGS_AUTO 0x40 #if BYTE_ORDER == BIG_ENDIAN #define ICMP6_RR_PCOUSE_FLAGS_DECRVLTIME 0x80000000 #define ICMP6_RR_PCOUSE_FLAGS_DECRPLTIME 0x40000000 #elif BYTE_ORDER == LITTLE_ENDIAN #define ICMP6_RR_PCOUSE_FLAGS_DECRVLTIME 0x80 #define ICMP6_RR_PCOUSE_FLAGS_DECRPLTIME 0x40 #endif struct rr_result { /* router renumbering result message */ u_int16_t rrr_flags; u_int8_t rrr_ordinal; u_int8_t rrr_matchedlen; u_int32_t rrr_ifid; struct in6_addr rrr_prefix; } __packed; #if BYTE_ORDER == BIG_ENDIAN #define ICMP6_RR_RESULT_FLAGS_OOB 0x0002 #define ICMP6_RR_RESULT_FLAGS_FORBIDDEN 0x0001 #elif BYTE_ORDER == LITTLE_ENDIAN #define ICMP6_RR_RESULT_FLAGS_OOB 0x0200 #define ICMP6_RR_RESULT_FLAGS_FORBIDDEN 0x0100 #endif /* * icmp6 filter structures. */ struct icmp6_filter { u_int32_t icmp6_filt[8]; }; #ifdef _KERNEL #define ICMP6_FILTER_SETPASSALL(filterp) \ do { \ int i; u_char *p; \ p = (u_char *)filterp; \ for (i = 0; i < sizeof(struct icmp6_filter); i++) \ p[i] = 0xff; \ } while (/*CONSTCOND*/ 0) #define ICMP6_FILTER_SETBLOCKALL(filterp) \ bzero(filterp, sizeof(struct icmp6_filter)) #else /* _KERNEL */ #define ICMP6_FILTER_SETPASSALL(filterp) \ memset(filterp, 0xff, sizeof(struct icmp6_filter)) #define ICMP6_FILTER_SETBLOCKALL(filterp) \ memset(filterp, 0x00, sizeof(struct icmp6_filter)) #endif /* _KERNEL */ #define ICMP6_FILTER_SETPASS(type, filterp) \ (((filterp)->icmp6_filt[(type) >> 5]) |= (1 << ((type) & 31))) #define ICMP6_FILTER_SETBLOCK(type, filterp) \ (((filterp)->icmp6_filt[(type) >> 5]) &= ~(1 << ((type) & 31))) #define ICMP6_FILTER_WILLPASS(type, filterp) \ ((((filterp)->icmp6_filt[(type) >> 5]) & (1 << ((type) & 31))) != 0) #define ICMP6_FILTER_WILLBLOCK(type, filterp) \ ((((filterp)->icmp6_filt[(type) >> 5]) & (1 << ((type) & 31))) == 0) /* * Variables related to this implementation * of the internet control message protocol version 6. */ struct icmp6errstat { u_quad_t icp6errs_dst_unreach_noroute; u_quad_t icp6errs_dst_unreach_admin; u_quad_t icp6errs_dst_unreach_beyondscope; u_quad_t icp6errs_dst_unreach_addr; u_quad_t icp6errs_dst_unreach_noport; u_quad_t icp6errs_packet_too_big; u_quad_t icp6errs_time_exceed_transit; u_quad_t icp6errs_time_exceed_reassembly; u_quad_t icp6errs_paramprob_header; u_quad_t icp6errs_paramprob_nextheader; u_quad_t icp6errs_paramprob_option; u_quad_t icp6errs_redirect; /* we regard redirect as an error here */ u_quad_t icp6errs_unknown; }; struct icmp6stat { /* statistics related to icmp6 packets generated */ u_quad_t icp6s_error; /* # of calls to icmp6_error */ u_quad_t icp6s_canterror; /* no error 'cuz old was icmp */ u_quad_t icp6s_toofreq; /* no error 'cuz rate limitation */ u_quad_t icp6s_outhist[256]; /* statistics related to input message processed */ u_quad_t icp6s_badcode; /* icmp6_code out of range */ u_quad_t icp6s_tooshort; /* packet < sizeof(struct icmp6_hdr) */ u_quad_t icp6s_checksum; /* bad checksum */ u_quad_t icp6s_badlen; /* calculated bound mismatch */ /* * number of responses: this member is inherited from netinet code, but * for netinet6 code, it is already available in icp6s_outhist[]. */ u_quad_t icp6s_reflect; u_quad_t icp6s_inhist[256]; u_quad_t icp6s_nd_toomanyopt; /* too many ND options */ struct icmp6errstat icp6s_outerrhist; #define icp6s_odst_unreach_noroute \ icp6s_outerrhist.icp6errs_dst_unreach_noroute #define icp6s_odst_unreach_admin icp6s_outerrhist.icp6errs_dst_unreach_admin #define icp6s_odst_unreach_beyondscope \ icp6s_outerrhist.icp6errs_dst_unreach_beyondscope #define icp6s_odst_unreach_addr icp6s_outerrhist.icp6errs_dst_unreach_addr #define icp6s_odst_unreach_noport icp6s_outerrhist.icp6errs_dst_unreach_noport #define icp6s_opacket_too_big icp6s_outerrhist.icp6errs_packet_too_big #define icp6s_otime_exceed_transit \ icp6s_outerrhist.icp6errs_time_exceed_transit #define icp6s_otime_exceed_reassembly \ icp6s_outerrhist.icp6errs_time_exceed_reassembly #define icp6s_oparamprob_header icp6s_outerrhist.icp6errs_paramprob_header #define icp6s_oparamprob_nextheader \ icp6s_outerrhist.icp6errs_paramprob_nextheader #define icp6s_oparamprob_option icp6s_outerrhist.icp6errs_paramprob_option #define icp6s_oredirect icp6s_outerrhist.icp6errs_redirect #define icp6s_ounknown icp6s_outerrhist.icp6errs_unknown u_quad_t icp6s_pmtuchg; /* path MTU changes */ u_quad_t icp6s_nd_badopt; /* bad ND options */ u_quad_t icp6s_badns; /* bad neighbor solicitation */ u_quad_t icp6s_badna; /* bad neighbor advertisement */ u_quad_t icp6s_badrs; /* bad router advertisement */ u_quad_t icp6s_badra; /* bad router advertisement */ u_quad_t icp6s_badredirect; /* bad redirect message */ }; #ifdef _KERNEL /* * In-kernel consumers can use these accessor macros directly to update * stats. */ #define ICMP6STAT_ADD(name, val) V_icmp6stat.name += (val) #define ICMP6STAT_INC(name) ICMP6STAT_ADD(name, 1) /* * Kernel module consumers must use this accessor macro. */ void kmod_icmp6stat_inc(int statnum); #define KMOD_ICMP6STAT_INC(name) \ kmod_icmp6stat_inc(offsetof(struct icmp6stat, name) / sizeof(u_quad_t)) #endif /* * Names for ICMP sysctl objects */ #define ICMPV6CTL_STATS 1 #define ICMPV6CTL_REDIRACCEPT 2 /* accept/process redirects */ #define ICMPV6CTL_REDIRTIMEOUT 3 /* redirect cache time */ #if 0 /*obsoleted*/ #define ICMPV6CTL_ERRRATELIMIT 5 /* ICMPv6 error rate limitation */ #endif #define ICMPV6CTL_ND6_PRUNE 6 #define ICMPV6CTL_ND6_DELAY 8 #define ICMPV6CTL_ND6_UMAXTRIES 9 #define ICMPV6CTL_ND6_MMAXTRIES 10 #define ICMPV6CTL_ND6_USELOOPBACK 11 /*#define ICMPV6CTL_ND6_PROXYALL 12 obsoleted, do not reuse here */ #define ICMPV6CTL_NODEINFO 13 #define ICMPV6CTL_ERRPPSLIMIT 14 /* ICMPv6 error pps limitation */ #define ICMPV6CTL_ND6_MAXNUDHINT 15 #define ICMPV6CTL_MTUDISC_HIWAT 16 #define ICMPV6CTL_MTUDISC_LOWAT 17 #define ICMPV6CTL_ND6_DEBUG 18 #define ICMPV6CTL_ND6_DRLIST 19 #define ICMPV6CTL_ND6_PRLIST 20 #define ICMPV6CTL_MLD_MAXSRCFILTER 21 #define ICMPV6CTL_MLD_SOMAXSRC 22 #define ICMPV6CTL_MLD_VERSION 23 #define ICMPV6CTL_ND6_MAXQLEN 24 #define ICMPV6CTL_MAXID 25 #define RTF_PROBEMTU RTF_PROTO1 #ifdef _KERNEL # ifdef __STDC__ struct rtentry; struct rttimer; struct in6_multi; # endif void icmp6_paramerror(struct mbuf *, int); void icmp6_error(struct mbuf *, int, int, int); void icmp6_error2(struct mbuf *, int, int, int, struct ifnet *); int icmp6_input(struct mbuf **, int *, int); void icmp6_fasttimo(void); void icmp6_slowtimo(void); void icmp6_reflect(struct mbuf *, size_t); void icmp6_prepare(struct mbuf *); void icmp6_redirect_input(struct mbuf *, int); void icmp6_redirect_output(struct mbuf *, struct rtentry *); struct ip6ctlparam; void icmp6_mtudisc_update(struct ip6ctlparam *, int); /* XXX: is this the right place for these macros? */ #define icmp6_ifstat_inc(ifp, tag) \ do { \ if (ifp) \ ((struct in6_ifextra *)((ifp)->if_afdata[AF_INET6]))->icmp6_ifstat->tag++; \ } while (/*CONSTCOND*/ 0) #define icmp6_ifoutstat_inc(ifp, type, code) \ do { \ icmp6_ifstat_inc(ifp, ifs6_out_msg); \ if (type < ICMP6_INFOMSG_MASK) \ icmp6_ifstat_inc(ifp, ifs6_out_error); \ switch (type) { \ case ICMP6_DST_UNREACH: \ icmp6_ifstat_inc(ifp, ifs6_out_dstunreach); \ if (code == ICMP6_DST_UNREACH_ADMIN) \ icmp6_ifstat_inc(ifp, ifs6_out_adminprohib); \ break; \ case ICMP6_PACKET_TOO_BIG: \ icmp6_ifstat_inc(ifp, ifs6_out_pkttoobig); \ break; \ case ICMP6_TIME_EXCEEDED: \ icmp6_ifstat_inc(ifp, ifs6_out_timeexceed); \ break; \ case ICMP6_PARAM_PROB: \ icmp6_ifstat_inc(ifp, ifs6_out_paramprob); \ break; \ case ICMP6_ECHO_REQUEST: \ icmp6_ifstat_inc(ifp, ifs6_out_echo); \ break; \ case ICMP6_ECHO_REPLY: \ icmp6_ifstat_inc(ifp, ifs6_out_echoreply); \ break; \ case MLD_LISTENER_QUERY: \ icmp6_ifstat_inc(ifp, ifs6_out_mldquery); \ break; \ case MLD_LISTENER_REPORT: \ icmp6_ifstat_inc(ifp, ifs6_out_mldreport); \ break; \ case MLD_LISTENER_DONE: \ icmp6_ifstat_inc(ifp, ifs6_out_mlddone); \ break; \ case ND_ROUTER_SOLICIT: \ icmp6_ifstat_inc(ifp, ifs6_out_routersolicit); \ break; \ case ND_ROUTER_ADVERT: \ icmp6_ifstat_inc(ifp, ifs6_out_routeradvert); \ break; \ case ND_NEIGHBOR_SOLICIT: \ icmp6_ifstat_inc(ifp, ifs6_out_neighborsolicit); \ break; \ case ND_NEIGHBOR_ADVERT: \ icmp6_ifstat_inc(ifp, ifs6_out_neighboradvert); \ break; \ case ND_REDIRECT: \ icmp6_ifstat_inc(ifp, ifs6_out_redirect); \ break; \ } \ } while (/*CONSTCOND*/ 0) VNET_DECLARE(int, icmp6_rediraccept); /* accept/process redirects */ VNET_DECLARE(int, icmp6_redirtimeout); /* cache time for redirect routes */ #define V_icmp6_rediraccept VNET(icmp6_rediraccept) #define V_icmp6_redirtimeout VNET(icmp6_redirtimeout) #define ICMP6_NODEINFO_FQDNOK 0x1 #define ICMP6_NODEINFO_NODEADDROK 0x2 #define ICMP6_NODEINFO_TMPADDROK 0x4 #define ICMP6_NODEINFO_GLOBALOK 0x8 #endif /* _KERNEL */ #endif /* not _NETINET_ICMP6_H_ */ Index: projects/largeSMP/sys/netinet/in_pcb.c =================================================================== --- projects/largeSMP/sys/netinet/in_pcb.c (revision 222782) +++ projects/largeSMP/sys/netinet/in_pcb.c (revision 222783) @@ -1,2228 +1,2449 @@ /*- * Copyright (c) 1982, 1986, 1991, 1993, 1995 * The Regents of the University of California. * Copyright (c) 2007-2009 Robert N. M. Watson * 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. * 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 8.4 (Berkeley) 5/24/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_ddb.h" #include "opt_ipsec.h" #include "opt_inet.h" #include "opt_inet6.h" +#include "opt_pcbgroup.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef DDB #include #endif #include #include #include #include #include #if defined(INET) || defined(INET6) #include #include #include #include #include #include #endif #ifdef INET #include #endif #ifdef INET6 #include #include #include #include #endif /* INET6 */ #ifdef IPSEC #include #include #endif /* IPSEC */ #include static struct callout ipport_tick_callout; /* * These configure the range of local port addresses assigned to * "unspecified" outgoing connections/packets/whatever. */ VNET_DEFINE(int, ipport_lowfirstauto) = IPPORT_RESERVED - 1; /* 1023 */ VNET_DEFINE(int, ipport_lowlastauto) = IPPORT_RESERVEDSTART; /* 600 */ VNET_DEFINE(int, ipport_firstauto) = IPPORT_EPHEMERALFIRST; /* 10000 */ VNET_DEFINE(int, ipport_lastauto) = IPPORT_EPHEMERALLAST; /* 65535 */ VNET_DEFINE(int, ipport_hifirstauto) = IPPORT_HIFIRSTAUTO; /* 49152 */ VNET_DEFINE(int, ipport_hilastauto) = IPPORT_HILASTAUTO; /* 65535 */ /* * Reserved ports accessible only to root. There are significant * security considerations that must be accounted for when changing these, * but the security benefits can be great. Please be careful. */ VNET_DEFINE(int, ipport_reservedhigh) = IPPORT_RESERVED - 1; /* 1023 */ VNET_DEFINE(int, ipport_reservedlow); /* Variables dealing with random ephemeral port allocation. */ VNET_DEFINE(int, ipport_randomized) = 1; /* user controlled via sysctl */ VNET_DEFINE(int, ipport_randomcps) = 10; /* user controlled via sysctl */ VNET_DEFINE(int, ipport_randomtime) = 45; /* user controlled via sysctl */ VNET_DEFINE(int, ipport_stoprandom); /* toggled by ipport_tick */ VNET_DEFINE(int, ipport_tcpallocs); static VNET_DEFINE(int, ipport_tcplastcount); #define V_ipport_tcplastcount VNET(ipport_tcplastcount) static void in_pcbremlists(struct inpcb *inp); #ifdef INET static struct inpcb *in_pcblookup_hash_locked(struct inpcbinfo *pcbinfo, struct in_addr faddr, u_int fport_arg, struct in_addr laddr, u_int lport_arg, int lookupflags, struct ifnet *ifp); #define RANGECHK(var, min, max) \ if ((var) < (min)) { (var) = (min); } \ else if ((var) > (max)) { (var) = (max); } static int sysctl_net_ipport_check(SYSCTL_HANDLER_ARGS) { int error; #ifdef VIMAGE error = vnet_sysctl_handle_int(oidp, arg1, arg2, req); #else error = sysctl_handle_int(oidp, arg1, arg2, req); #endif if (error == 0) { RANGECHK(V_ipport_lowfirstauto, 1, IPPORT_RESERVED - 1); RANGECHK(V_ipport_lowlastauto, 1, IPPORT_RESERVED - 1); RANGECHK(V_ipport_firstauto, IPPORT_RESERVED, IPPORT_MAX); RANGECHK(V_ipport_lastauto, IPPORT_RESERVED, IPPORT_MAX); RANGECHK(V_ipport_hifirstauto, IPPORT_RESERVED, IPPORT_MAX); RANGECHK(V_ipport_hilastauto, IPPORT_RESERVED, IPPORT_MAX); } return (error); } #undef RANGECHK SYSCTL_NODE(_net_inet_ip, IPPROTO_IP, portrange, CTLFLAG_RW, 0, "IP Ports"); SYSCTL_VNET_PROC(_net_inet_ip_portrange, OID_AUTO, lowfirst, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(ipport_lowfirstauto), 0, &sysctl_net_ipport_check, "I", ""); SYSCTL_VNET_PROC(_net_inet_ip_portrange, OID_AUTO, lowlast, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(ipport_lowlastauto), 0, &sysctl_net_ipport_check, "I", ""); SYSCTL_VNET_PROC(_net_inet_ip_portrange, OID_AUTO, first, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(ipport_firstauto), 0, &sysctl_net_ipport_check, "I", ""); SYSCTL_VNET_PROC(_net_inet_ip_portrange, OID_AUTO, last, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(ipport_lastauto), 0, &sysctl_net_ipport_check, "I", ""); SYSCTL_VNET_PROC(_net_inet_ip_portrange, OID_AUTO, hifirst, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(ipport_hifirstauto), 0, &sysctl_net_ipport_check, "I", ""); SYSCTL_VNET_PROC(_net_inet_ip_portrange, OID_AUTO, hilast, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(ipport_hilastauto), 0, &sysctl_net_ipport_check, "I", ""); SYSCTL_VNET_INT(_net_inet_ip_portrange, OID_AUTO, reservedhigh, CTLFLAG_RW|CTLFLAG_SECURE, &VNET_NAME(ipport_reservedhigh), 0, ""); SYSCTL_VNET_INT(_net_inet_ip_portrange, OID_AUTO, reservedlow, CTLFLAG_RW|CTLFLAG_SECURE, &VNET_NAME(ipport_reservedlow), 0, ""); SYSCTL_VNET_INT(_net_inet_ip_portrange, OID_AUTO, randomized, CTLFLAG_RW, &VNET_NAME(ipport_randomized), 0, "Enable random port allocation"); SYSCTL_VNET_INT(_net_inet_ip_portrange, OID_AUTO, randomcps, CTLFLAG_RW, &VNET_NAME(ipport_randomcps), 0, "Maximum number of random port " "allocations before switching to a sequental one"); SYSCTL_VNET_INT(_net_inet_ip_portrange, OID_AUTO, randomtime, CTLFLAG_RW, &VNET_NAME(ipport_randomtime), 0, "Minimum time to keep sequental port " "allocation before switching to a random one"); #endif /* * in_pcb.c: manage the Protocol Control Blocks. * * NOTE: It is assumed that most of these functions will be called with * the pcbinfo lock held, and often, the inpcb lock held, as these utility * functions often modify hash chains or addresses in pcbs. */ /* * Initialize an inpcbinfo -- we should be able to reduce the number of * arguments in time. */ void in_pcbinfo_init(struct inpcbinfo *pcbinfo, const char *name, struct inpcbhead *listhead, int hash_nelements, int porthash_nelements, char *inpcbzone_name, uma_init inpcbzone_init, uma_fini inpcbzone_fini, - uint32_t inpcbzone_flags) + uint32_t inpcbzone_flags, u_int hashfields) { INP_INFO_LOCK_INIT(pcbinfo, name); INP_HASH_LOCK_INIT(pcbinfo, "pcbinfohash"); /* XXXRW: argument? */ #ifdef VIMAGE pcbinfo->ipi_vnet = curvnet; #endif pcbinfo->ipi_listhead = listhead; LIST_INIT(pcbinfo->ipi_listhead); pcbinfo->ipi_count = 0; pcbinfo->ipi_hashbase = hashinit(hash_nelements, M_PCB, &pcbinfo->ipi_hashmask); pcbinfo->ipi_porthashbase = hashinit(porthash_nelements, M_PCB, &pcbinfo->ipi_porthashmask); +#ifdef PCBGROUP + in_pcbgroup_init(pcbinfo, hashfields, hash_nelements); +#endif pcbinfo->ipi_zone = uma_zcreate(inpcbzone_name, sizeof(struct inpcb), NULL, NULL, inpcbzone_init, inpcbzone_fini, UMA_ALIGN_PTR, inpcbzone_flags); uma_zone_set_max(pcbinfo->ipi_zone, maxsockets); } /* * Destroy an inpcbinfo. */ void in_pcbinfo_destroy(struct inpcbinfo *pcbinfo) { KASSERT(pcbinfo->ipi_count == 0, ("%s: ipi_count = %u", __func__, pcbinfo->ipi_count)); hashdestroy(pcbinfo->ipi_hashbase, M_PCB, pcbinfo->ipi_hashmask); hashdestroy(pcbinfo->ipi_porthashbase, M_PCB, pcbinfo->ipi_porthashmask); +#ifdef PCBGROUP + in_pcbgroup_destroy(pcbinfo); +#endif uma_zdestroy(pcbinfo->ipi_zone); INP_HASH_LOCK_DESTROY(pcbinfo); INP_INFO_LOCK_DESTROY(pcbinfo); } /* * Allocate a PCB and associate it with the socket. * On success return with the PCB locked. */ int in_pcballoc(struct socket *so, struct inpcbinfo *pcbinfo) { struct inpcb *inp; int error; INP_INFO_WLOCK_ASSERT(pcbinfo); error = 0; inp = uma_zalloc(pcbinfo->ipi_zone, M_NOWAIT); if (inp == NULL) return (ENOBUFS); bzero(inp, inp_zero_size); inp->inp_pcbinfo = pcbinfo; inp->inp_socket = so; inp->inp_cred = crhold(so->so_cred); inp->inp_inc.inc_fibnum = so->so_fibnum; #ifdef MAC error = mac_inpcb_init(inp, M_NOWAIT); if (error != 0) goto out; mac_inpcb_create(so, inp); #endif #ifdef IPSEC error = ipsec_init_policy(so, &inp->inp_sp); if (error != 0) { #ifdef MAC mac_inpcb_destroy(inp); #endif goto out; } #endif /*IPSEC*/ #ifdef INET6 if (INP_SOCKAF(so) == AF_INET6) { inp->inp_vflag |= INP_IPV6PROTO; if (V_ip6_v6only) inp->inp_flags |= IN6P_IPV6_V6ONLY; } #endif LIST_INSERT_HEAD(pcbinfo->ipi_listhead, inp, inp_list); pcbinfo->ipi_count++; so->so_pcb = (caddr_t)inp; #ifdef INET6 if (V_ip6_auto_flowlabel) inp->inp_flags |= IN6P_AUTOFLOWLABEL; #endif INP_WLOCK(inp); inp->inp_gencnt = ++pcbinfo->ipi_gencnt; refcount_init(&inp->inp_refcount, 1); /* Reference from inpcbinfo */ #if defined(IPSEC) || defined(MAC) out: if (error != 0) { crfree(inp->inp_cred); uma_zfree(pcbinfo->ipi_zone, inp); } #endif return (error); } #ifdef INET int in_pcbbind(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred) { int anonport, error; INP_WLOCK_ASSERT(inp); INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); if (inp->inp_lport != 0 || inp->inp_laddr.s_addr != INADDR_ANY) return (EINVAL); anonport = inp->inp_lport == 0 && (nam == NULL || ((struct sockaddr_in *)nam)->sin_port == 0); error = in_pcbbind_setup(inp, nam, &inp->inp_laddr.s_addr, &inp->inp_lport, cred); if (error) return (error); if (in_pcbinshash(inp) != 0) { inp->inp_laddr.s_addr = INADDR_ANY; inp->inp_lport = 0; return (EAGAIN); } if (anonport) inp->inp_flags |= INP_ANONPORT; return (0); } #endif #if defined(INET) || defined(INET6) int in_pcb_lport(struct inpcb *inp, struct in_addr *laddrp, u_short *lportp, struct ucred *cred, int lookupflags) { struct inpcbinfo *pcbinfo; struct inpcb *tmpinp; unsigned short *lastport; int count, dorandom, error; u_short aux, first, last, lport; #ifdef INET struct in_addr laddr; #endif pcbinfo = inp->inp_pcbinfo; /* * Because no actual state changes occur here, a global write lock on * the pcbinfo isn't required. */ INP_LOCK_ASSERT(inp); INP_HASH_LOCK_ASSERT(pcbinfo); if (inp->inp_flags & INP_HIGHPORT) { first = V_ipport_hifirstauto; /* sysctl */ last = V_ipport_hilastauto; lastport = &pcbinfo->ipi_lasthi; } else if (inp->inp_flags & INP_LOWPORT) { error = priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT, 0); if (error) return (error); first = V_ipport_lowfirstauto; /* 1023 */ last = V_ipport_lowlastauto; /* 600 */ lastport = &pcbinfo->ipi_lastlow; } else { first = V_ipport_firstauto; /* sysctl */ last = V_ipport_lastauto; lastport = &pcbinfo->ipi_lastport; } /* * For UDP, use random port allocation as long as the user * allows it. For TCP (and as of yet unknown) connections, * use random port allocation only if the user allows it AND * ipport_tick() allows it. */ if (V_ipport_randomized && (!V_ipport_stoprandom || pcbinfo == &V_udbinfo)) dorandom = 1; else dorandom = 0; /* * It makes no sense to do random port allocation if * we have the only port available. */ if (first == last) dorandom = 0; /* Make sure to not include UDP packets in the count. */ if (pcbinfo != &V_udbinfo) V_ipport_tcpallocs++; /* * Instead of having two loops further down counting up or down * make sure that first is always <= last and go with only one * code path implementing all logic. */ if (first > last) { aux = first; first = last; last = aux; } #ifdef INET /* Make the compiler happy. */ laddr.s_addr = 0; if ((inp->inp_vflag & (INP_IPV4|INP_IPV6)) == INP_IPV4) { KASSERT(laddrp != NULL, ("%s: laddrp NULL for v4 inp %p", __func__, inp)); laddr = *laddrp; } #endif tmpinp = NULL; /* Make compiler happy. */ lport = *lportp; if (dorandom) *lastport = first + (arc4random() % (last - first)); count = last - first; do { if (count-- < 0) /* completely used? */ return (EADDRNOTAVAIL); ++*lastport; if (*lastport < first || *lastport > last) *lastport = first; lport = htons(*lastport); #ifdef INET6 if ((inp->inp_vflag & INP_IPV6) != 0) tmpinp = in6_pcblookup_local(pcbinfo, &inp->in6p_laddr, lport, lookupflags, cred); #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET tmpinp = in_pcblookup_local(pcbinfo, laddr, lport, lookupflags, cred); #endif } while (tmpinp != NULL); #ifdef INET if ((inp->inp_vflag & (INP_IPV4|INP_IPV6)) == INP_IPV4) laddrp->s_addr = laddr.s_addr; #endif *lportp = lport; return (0); } #endif /* INET || INET6 */ #ifdef INET /* * Set up a bind operation on a PCB, performing port allocation * as required, but do not actually modify the PCB. Callers can * either complete the bind by setting inp_laddr/inp_lport and * calling in_pcbinshash(), or they can just use the resulting * port and address to authorise the sending of a once-off packet. * * On error, the values of *laddrp and *lportp are not changed. */ int in_pcbbind_setup(struct inpcb *inp, struct sockaddr *nam, in_addr_t *laddrp, u_short *lportp, struct ucred *cred) { struct socket *so = inp->inp_socket; struct sockaddr_in *sin; struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; struct in_addr laddr; u_short lport = 0; int lookupflags = 0, reuseport = (so->so_options & SO_REUSEPORT); int error; /* * No state changes, so read locks are sufficient here. */ INP_LOCK_ASSERT(inp); INP_HASH_LOCK_ASSERT(pcbinfo); if (TAILQ_EMPTY(&V_in_ifaddrhead)) /* XXX broken! */ return (EADDRNOTAVAIL); laddr.s_addr = *laddrp; if (nam != NULL && laddr.s_addr != INADDR_ANY) return (EINVAL); if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) lookupflags = INPLOOKUP_WILDCARD; if (nam == NULL) { if ((error = prison_local_ip4(cred, &laddr)) != 0) return (error); } else { sin = (struct sockaddr_in *)nam; if (nam->sa_len != sizeof (*sin)) return (EINVAL); #ifdef notdef /* * We should check the family, but old programs * incorrectly fail to initialize it. */ if (sin->sin_family != AF_INET) return (EAFNOSUPPORT); #endif error = prison_local_ip4(cred, &sin->sin_addr); if (error) return (error); if (sin->sin_port != *lportp) { /* Don't allow the port to change. */ if (*lportp != 0) return (EINVAL); lport = sin->sin_port; } /* NB: lport is left as 0 if the port isn't being changed. */ if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) { /* * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; * allow complete 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) reuseport = SO_REUSEADDR|SO_REUSEPORT; } else if (sin->sin_addr.s_addr != INADDR_ANY) { sin->sin_port = 0; /* yech... */ bzero(&sin->sin_zero, sizeof(sin->sin_zero)); /* * Is the address a local IP address? * If INP_BINDANY is set, then the socket may be bound * to any endpoint address, local or not. */ if ((inp->inp_flags & INP_BINDANY) == 0 && ifa_ifwithaddr_check((struct sockaddr *)sin) == 0) return (EADDRNOTAVAIL); } laddr = sin->sin_addr; 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 (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)) && priv_check_cred(inp->inp_cred, PRIV_NETINET_REUSEPORT, 0) != 0) { t = in_pcblookup_local(pcbinfo, sin->sin_addr, lport, INPLOOKUP_WILDCARD, cred); /* * XXX * This entire block sorely needs a rewrite. */ if (t && ((t->inp_flags & INP_TIMEWAIT) == 0) && (so->so_type != SOCK_STREAM || ntohl(t->inp_faddr.s_addr) == INADDR_ANY) && (ntohl(sin->sin_addr.s_addr) != INADDR_ANY || ntohl(t->inp_laddr.s_addr) != INADDR_ANY || (t->inp_socket->so_options & SO_REUSEPORT) == 0) && (inp->inp_cred->cr_uid != t->inp_cred->cr_uid)) return (EADDRINUSE); } t = in_pcblookup_local(pcbinfo, sin->sin_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(inp); if (tw == NULL || (reuseport & tw->tw_so_options) == 0) return (EADDRINUSE); } else if (t && (reuseport & t->inp_socket->so_options) == 0) { #ifdef INET6 if (ntohl(sin->sin_addr.s_addr) != INADDR_ANY || ntohl(t->inp_laddr.s_addr) != INADDR_ANY || INP_SOCKAF(so) == INP_SOCKAF(t->inp_socket)) #endif return (EADDRINUSE); } } } if (*lportp != 0) lport = *lportp; if (lport == 0) { error = in_pcb_lport(inp, &laddr, &lport, cred, lookupflags); if (error != 0) return (error); } *laddrp = laddr.s_addr; *lportp = lport; return (0); } /* * 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 in_pcbconnect_mbuf(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred, struct mbuf *m) { u_short lport, fport; in_addr_t laddr, faddr; int anonport, error; INP_WLOCK_ASSERT(inp); INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); lport = inp->inp_lport; laddr = inp->inp_laddr.s_addr; anonport = (lport == 0); error = in_pcbconnect_setup(inp, nam, &laddr, &lport, &faddr, &fport, NULL, cred); if (error) return (error); /* Do the initial binding of the local address if required. */ if (inp->inp_laddr.s_addr == INADDR_ANY && inp->inp_lport == 0) { inp->inp_lport = lport; inp->inp_laddr.s_addr = laddr; if (in_pcbinshash(inp) != 0) { inp->inp_laddr.s_addr = INADDR_ANY; inp->inp_lport = 0; return (EAGAIN); } } /* Commit the remaining changes. */ inp->inp_lport = lport; inp->inp_laddr.s_addr = laddr; inp->inp_faddr.s_addr = faddr; inp->inp_fport = fport; in_pcbrehash_mbuf(inp, m); if (anonport) inp->inp_flags |= INP_ANONPORT; return (0); } int in_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred) { return (in_pcbconnect_mbuf(inp, nam, cred, NULL)); } /* * Do proper source address selection on an unbound socket in case * of connect. Take jails into account as well. */ static int in_pcbladdr(struct inpcb *inp, struct in_addr *faddr, struct in_addr *laddr, struct ucred *cred) { struct ifaddr *ifa; struct sockaddr *sa; struct sockaddr_in *sin; struct route sro; int error; KASSERT(laddr != NULL, ("%s: laddr NULL", __func__)); /* * Bypass source address selection and use the primary jail IP * if requested. */ if (cred != NULL && !prison_saddrsel_ip4(cred, laddr)) return (0); error = 0; bzero(&sro, sizeof(sro)); sin = (struct sockaddr_in *)&sro.ro_dst; sin->sin_family = AF_INET; sin->sin_len = sizeof(struct sockaddr_in); sin->sin_addr.s_addr = faddr->s_addr; /* * If route is known our src addr is taken from the i/f, * else punt. * * Find out route to destination. */ if ((inp->inp_socket->so_options & SO_DONTROUTE) == 0) in_rtalloc_ign(&sro, 0, inp->inp_inc.inc_fibnum); /* * If we found a route, use the address corresponding to * the outgoing interface. * * Otherwise assume faddr is reachable on a directly connected * network and try to find a corresponding interface to take * the source address from. */ if (sro.ro_rt == NULL || sro.ro_rt->rt_ifp == NULL) { struct in_ifaddr *ia; struct ifnet *ifp; ia = ifatoia(ifa_ifwithdstaddr((struct sockaddr *)sin)); if (ia == NULL) ia = ifatoia(ifa_ifwithnet((struct sockaddr *)sin, 0)); if (ia == NULL) { error = ENETUNREACH; goto done; } if (cred == NULL || !prison_flag(cred, PR_IP4)) { laddr->s_addr = ia->ia_addr.sin_addr.s_addr; ifa_free(&ia->ia_ifa); goto done; } ifp = ia->ia_ifp; ifa_free(&ia->ia_ifa); ia = NULL; IF_ADDR_LOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { sa = ifa->ifa_addr; if (sa->sa_family != AF_INET) continue; sin = (struct sockaddr_in *)sa; if (prison_check_ip4(cred, &sin->sin_addr) == 0) { ia = (struct in_ifaddr *)ifa; break; } } if (ia != NULL) { laddr->s_addr = ia->ia_addr.sin_addr.s_addr; IF_ADDR_UNLOCK(ifp); goto done; } IF_ADDR_UNLOCK(ifp); /* 3. As a last resort return the 'default' jail address. */ error = prison_get_ip4(cred, laddr); goto done; } /* * If the outgoing interface on the route found is not * a loopback interface, use the address from that interface. * In case of jails do those three steps: * 1. check if the interface address belongs to the jail. If so use it. * 2. check if we have any address on the outgoing interface * belonging to this jail. If so use it. * 3. as a last resort return the 'default' jail address. */ if ((sro.ro_rt->rt_ifp->if_flags & IFF_LOOPBACK) == 0) { struct in_ifaddr *ia; struct ifnet *ifp; /* If not jailed, use the default returned. */ if (cred == NULL || !prison_flag(cred, PR_IP4)) { ia = (struct in_ifaddr *)sro.ro_rt->rt_ifa; laddr->s_addr = ia->ia_addr.sin_addr.s_addr; goto done; } /* Jailed. */ /* 1. Check if the iface address belongs to the jail. */ sin = (struct sockaddr_in *)sro.ro_rt->rt_ifa->ifa_addr; if (prison_check_ip4(cred, &sin->sin_addr) == 0) { ia = (struct in_ifaddr *)sro.ro_rt->rt_ifa; laddr->s_addr = ia->ia_addr.sin_addr.s_addr; goto done; } /* * 2. Check if we have any address on the outgoing interface * belonging to this jail. */ ia = NULL; ifp = sro.ro_rt->rt_ifp; IF_ADDR_LOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { sa = ifa->ifa_addr; if (sa->sa_family != AF_INET) continue; sin = (struct sockaddr_in *)sa; if (prison_check_ip4(cred, &sin->sin_addr) == 0) { ia = (struct in_ifaddr *)ifa; break; } } if (ia != NULL) { laddr->s_addr = ia->ia_addr.sin_addr.s_addr; IF_ADDR_UNLOCK(ifp); goto done; } IF_ADDR_UNLOCK(ifp); /* 3. As a last resort return the 'default' jail address. */ error = prison_get_ip4(cred, laddr); goto done; } /* * The outgoing interface is marked with 'loopback net', so a route * to ourselves is here. * Try to find the interface of the destination address and then * take the address from there. That interface is not necessarily * a loopback interface. * In case of jails, check that it is an address of the jail * and if we cannot find, fall back to the 'default' jail address. */ if ((sro.ro_rt->rt_ifp->if_flags & IFF_LOOPBACK) != 0) { struct sockaddr_in sain; struct in_ifaddr *ia; bzero(&sain, sizeof(struct sockaddr_in)); sain.sin_family = AF_INET; sain.sin_len = sizeof(struct sockaddr_in); sain.sin_addr.s_addr = faddr->s_addr; ia = ifatoia(ifa_ifwithdstaddr(sintosa(&sain))); if (ia == NULL) ia = ifatoia(ifa_ifwithnet(sintosa(&sain), 0)); if (ia == NULL) ia = ifatoia(ifa_ifwithaddr(sintosa(&sain))); if (cred == NULL || !prison_flag(cred, PR_IP4)) { if (ia == NULL) { error = ENETUNREACH; goto done; } laddr->s_addr = ia->ia_addr.sin_addr.s_addr; ifa_free(&ia->ia_ifa); goto done; } /* Jailed. */ if (ia != NULL) { struct ifnet *ifp; ifp = ia->ia_ifp; ifa_free(&ia->ia_ifa); ia = NULL; IF_ADDR_LOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { sa = ifa->ifa_addr; if (sa->sa_family != AF_INET) continue; sin = (struct sockaddr_in *)sa; if (prison_check_ip4(cred, &sin->sin_addr) == 0) { ia = (struct in_ifaddr *)ifa; break; } } if (ia != NULL) { laddr->s_addr = ia->ia_addr.sin_addr.s_addr; IF_ADDR_UNLOCK(ifp); goto done; } IF_ADDR_UNLOCK(ifp); } /* 3. As a last resort return the 'default' jail address. */ error = prison_get_ip4(cred, laddr); goto done; } done: if (sro.ro_rt != NULL) RTFREE(sro.ro_rt); return (error); } /* * Set up for a connect from a socket to the specified address. * On entry, *laddrp and *lportp should contain the current local * address and port for the PCB; these are updated to the values * that should be placed in inp_laddr and inp_lport to complete * the connect. * * On success, *faddrp and *fportp will be set to the remote address * and port. These are not updated in the error case. * * If the operation fails because the connection already exists, * *oinpp will be set to the PCB of that connection so that the * caller can decide to override it. In all other cases, *oinpp * is set to NULL. */ int in_pcbconnect_setup(struct inpcb *inp, struct sockaddr *nam, in_addr_t *laddrp, u_short *lportp, in_addr_t *faddrp, u_short *fportp, struct inpcb **oinpp, struct ucred *cred) { struct sockaddr_in *sin = (struct sockaddr_in *)nam; struct in_ifaddr *ia; struct inpcb *oinp; struct in_addr laddr, faddr; u_short lport, fport; int error; /* * Because a global state change doesn't actually occur here, a read * lock is sufficient. */ INP_LOCK_ASSERT(inp); INP_HASH_LOCK_ASSERT(inp->inp_pcbinfo); if (oinpp != NULL) *oinpp = NULL; if (nam->sa_len != sizeof (*sin)) return (EINVAL); if (sin->sin_family != AF_INET) return (EAFNOSUPPORT); if (sin->sin_port == 0) return (EADDRNOTAVAIL); laddr.s_addr = *laddrp; lport = *lportp; faddr = sin->sin_addr; fport = sin->sin_port; if (!TAILQ_EMPTY(&V_in_ifaddrhead)) { /* * If the destination address is INADDR_ANY, * use the primary local address. * If the supplied address is INADDR_BROADCAST, * and the primary interface supports broadcast, * choose the broadcast address for that interface. */ if (faddr.s_addr == INADDR_ANY) { IN_IFADDR_RLOCK(); faddr = IA_SIN(TAILQ_FIRST(&V_in_ifaddrhead))->sin_addr; IN_IFADDR_RUNLOCK(); if (cred != NULL && (error = prison_get_ip4(cred, &faddr)) != 0) return (error); } else if (faddr.s_addr == (u_long)INADDR_BROADCAST) { IN_IFADDR_RLOCK(); if (TAILQ_FIRST(&V_in_ifaddrhead)->ia_ifp->if_flags & IFF_BROADCAST) faddr = satosin(&TAILQ_FIRST( &V_in_ifaddrhead)->ia_broadaddr)->sin_addr; IN_IFADDR_RUNLOCK(); } } if (laddr.s_addr == INADDR_ANY) { error = in_pcbladdr(inp, &faddr, &laddr, cred); /* * If the destination address is multicast and an outgoing * interface has been set as a multicast option, prefer the * address of that interface as our source address. */ if (IN_MULTICAST(ntohl(faddr.s_addr)) && inp->inp_moptions != NULL) { struct ip_moptions *imo; struct ifnet *ifp; imo = inp->inp_moptions; if (imo->imo_multicast_ifp != NULL) { ifp = imo->imo_multicast_ifp; IN_IFADDR_RLOCK(); TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) { if ((ia->ia_ifp == ifp) && (cred == NULL || prison_check_ip4(cred, &ia->ia_addr.sin_addr) == 0)) break; } if (ia == NULL) error = EADDRNOTAVAIL; else { laddr = ia->ia_addr.sin_addr; error = 0; } IN_IFADDR_RUNLOCK(); } } if (error) return (error); } oinp = in_pcblookup_hash_locked(inp->inp_pcbinfo, faddr, fport, laddr, lport, 0, NULL); if (oinp != NULL) { if (oinpp != NULL) *oinpp = oinp; return (EADDRINUSE); } if (lport == 0) { error = in_pcbbind_setup(inp, NULL, &laddr.s_addr, &lport, cred); if (error) return (error); } *laddrp = laddr.s_addr; *lportp = lport; *faddrp = faddr.s_addr; *fportp = fport; return (0); } void in_pcbdisconnect(struct inpcb *inp) { INP_WLOCK_ASSERT(inp); INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); inp->inp_faddr.s_addr = INADDR_ANY; inp->inp_fport = 0; in_pcbrehash(inp); } #endif /* * in_pcbdetach() is responsibe for disassociating a socket from an inpcb. * For most protocols, this will be invoked immediately prior to calling * in_pcbfree(). However, with TCP the inpcb may significantly outlive the * socket, in which case in_pcbfree() is deferred. */ void in_pcbdetach(struct inpcb *inp) { KASSERT(inp->inp_socket != NULL, ("%s: inp_socket == NULL", __func__)); inp->inp_socket->so_pcb = NULL; inp->inp_socket = NULL; } /* * in_pcbref() bumps the reference count on an inpcb in order to maintain * stability of an inpcb pointer despite the inpcb lock being released. This * is used in TCP when the inpcbinfo lock needs to be acquired or upgraded, - * but where the inpcb lock is already held. + * but where the inpcb lock may already held, or when acquiring a reference + * via a pcbgroup. * * in_pcbref() should be used only to provide brief memory stability, and * must always be followed by a call to INP_WLOCK() and in_pcbrele() to * garbage collect the inpcb if it has been in_pcbfree()'d from another * context. Until in_pcbrele() has returned that the inpcb is still valid, * lock and rele are the *only* safe operations that may be performed on the * inpcb. * * While the inpcb will not be freed, releasing the inpcb lock means that the * connection's state may change, so the caller should be careful to * revalidate any cached state on reacquiring the lock. Drop the reference * using in_pcbrele(). */ void in_pcbref(struct inpcb *inp) { KASSERT(inp->inp_refcount > 0, ("%s: refcount 0", __func__)); refcount_acquire(&inp->inp_refcount); } /* * Drop a refcount on an inpcb elevated using in_pcbref(); because a call to * in_pcbfree() may have been made between in_pcbref() and in_pcbrele(), we * return a flag indicating whether or not the inpcb remains valid. If it is * valid, we return with the inpcb lock held. * * Notice that, unlike in_pcbref(), the inpcb lock must be held to drop a * reference on an inpcb. Historically more work was done here (actually, in * in_pcbfree_internal()) but has been moved to in_pcbfree() to avoid the * need for the pcbinfo lock in in_pcbrele(). Deferring the free is entirely * about memory stability (and continued use of the write lock). */ int in_pcbrele_rlocked(struct inpcb *inp) { struct inpcbinfo *pcbinfo; KASSERT(inp->inp_refcount > 0, ("%s: refcount 0", __func__)); INP_RLOCK_ASSERT(inp); if (refcount_release(&inp->inp_refcount) == 0) return (0); KASSERT(inp->inp_socket == NULL, ("%s: inp_socket != NULL", __func__)); INP_RUNLOCK(inp); pcbinfo = inp->inp_pcbinfo; uma_zfree(pcbinfo->ipi_zone, inp); return (1); } int in_pcbrele_wlocked(struct inpcb *inp) { struct inpcbinfo *pcbinfo; KASSERT(inp->inp_refcount > 0, ("%s: refcount 0", __func__)); INP_WLOCK_ASSERT(inp); if (refcount_release(&inp->inp_refcount) == 0) return (0); KASSERT(inp->inp_socket == NULL, ("%s: inp_socket != NULL", __func__)); INP_WUNLOCK(inp); pcbinfo = inp->inp_pcbinfo; uma_zfree(pcbinfo->ipi_zone, inp); return (1); } /* * Temporary wrapper. */ int in_pcbrele(struct inpcb *inp) { return (in_pcbrele_wlocked(inp)); } /* * Unconditionally schedule an inpcb to be freed by decrementing its * reference count, which should occur only after the inpcb has been detached * from its socket. If another thread holds a temporary reference (acquired * using in_pcbref()) then the free is deferred until that reference is * released using in_pcbrele(), but the inpcb is still unlocked. Almost all * work, including removal from global lists, is done in this context, where * the pcbinfo lock is held. */ void in_pcbfree(struct inpcb *inp) { struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; KASSERT(inp->inp_socket == NULL, ("%s: inp_socket != NULL", __func__)); INP_INFO_WLOCK_ASSERT(pcbinfo); INP_WLOCK_ASSERT(inp); /* XXXRW: Do as much as possible here. */ #ifdef IPSEC if (inp->inp_sp != NULL) ipsec_delete_pcbpolicy(inp); #endif /* IPSEC */ inp->inp_gencnt = ++pcbinfo->ipi_gencnt; in_pcbremlists(inp); #ifdef INET6 if (inp->inp_vflag & INP_IPV6PROTO) { ip6_freepcbopts(inp->in6p_outputopts); if (inp->in6p_moptions != NULL) ip6_freemoptions(inp->in6p_moptions); } #endif if (inp->inp_options) (void)m_free(inp->inp_options); #ifdef INET if (inp->inp_moptions != NULL) inp_freemoptions(inp->inp_moptions); #endif inp->inp_vflag = 0; crfree(inp->inp_cred); #ifdef MAC mac_inpcb_destroy(inp); #endif if (!in_pcbrele_wlocked(inp)) INP_WUNLOCK(inp); } /* * in_pcbdrop() removes an inpcb from hashed lists, releasing its address and * port reservation, and preventing it from being returned by inpcb lookups. * * It is used by TCP to mark an inpcb as unused and avoid future packet * delivery or event notification when a socket remains open but TCP has * closed. This might occur as a result of a shutdown()-initiated TCP close * or a RST on the wire, and allows the port binding to be reused while still * maintaining the invariant that so_pcb always points to a valid inpcb until * in_pcbdetach(). * * XXXRW: Possibly in_pcbdrop() should also prevent future notifications by * in_pcbnotifyall() and in_pcbpurgeif0()? */ void in_pcbdrop(struct inpcb *inp) { INP_WLOCK_ASSERT(inp); /* * XXXRW: Possibly we should protect the setting of INP_DROPPED with * the hash lock...? */ inp->inp_flags |= INP_DROPPED; if (inp->inp_flags & INP_INHASHLIST) { struct inpcbport *phd = inp->inp_phd; INP_HASH_WLOCK(inp->inp_pcbinfo); LIST_REMOVE(inp, inp_hash); LIST_REMOVE(inp, inp_portlist); if (LIST_FIRST(&phd->phd_pcblist) == NULL) { LIST_REMOVE(phd, phd_hash); free(phd, M_PCB); } INP_HASH_WUNLOCK(inp->inp_pcbinfo); inp->inp_flags &= ~INP_INHASHLIST; +#ifdef PCBGROUP + in_pcbgroup_remove(inp); +#endif } } #ifdef INET /* * Common routines to return the socket addresses associated with inpcbs. */ struct sockaddr * in_sockaddr(in_port_t port, struct in_addr *addr_p) { struct sockaddr_in *sin; sin = malloc(sizeof *sin, M_SONAME, M_WAITOK | M_ZERO); sin->sin_family = AF_INET; sin->sin_len = sizeof(*sin); sin->sin_addr = *addr_p; sin->sin_port = port; return (struct sockaddr *)sin; } int in_getsockaddr(struct socket *so, struct sockaddr **nam) { struct inpcb *inp; struct in_addr addr; in_port_t port; inp = sotoinpcb(so); KASSERT(inp != NULL, ("in_getsockaddr: inp == NULL")); INP_RLOCK(inp); port = inp->inp_lport; addr = inp->inp_laddr; INP_RUNLOCK(inp); *nam = in_sockaddr(port, &addr); return 0; } int in_getpeeraddr(struct socket *so, struct sockaddr **nam) { struct inpcb *inp; struct in_addr addr; in_port_t port; inp = sotoinpcb(so); KASSERT(inp != NULL, ("in_getpeeraddr: inp == NULL")); INP_RLOCK(inp); port = inp->inp_fport; addr = inp->inp_faddr; INP_RUNLOCK(inp); *nam = in_sockaddr(port, &addr); return 0; } void in_pcbnotifyall(struct inpcbinfo *pcbinfo, struct in_addr faddr, int errno, struct inpcb *(*notify)(struct inpcb *, int)) { struct inpcb *inp, *inp_temp; INP_INFO_WLOCK(pcbinfo); LIST_FOREACH_SAFE(inp, pcbinfo->ipi_listhead, inp_list, inp_temp) { INP_WLOCK(inp); #ifdef INET6 if ((inp->inp_vflag & INP_IPV4) == 0) { INP_WUNLOCK(inp); continue; } #endif if (inp->inp_faddr.s_addr != faddr.s_addr || inp->inp_socket == NULL) { INP_WUNLOCK(inp); continue; } if ((*notify)(inp, errno)) INP_WUNLOCK(inp); } INP_INFO_WUNLOCK(pcbinfo); } void in_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp) { struct inpcb *inp; struct ip_moptions *imo; int i, gap; INP_INFO_RLOCK(pcbinfo); LIST_FOREACH(inp, pcbinfo->ipi_listhead, inp_list) { INP_WLOCK(inp); imo = inp->inp_moptions; if ((inp->inp_vflag & INP_IPV4) && imo != NULL) { /* * Unselect the outgoing interface if it is being * detached. */ if (imo->imo_multicast_ifp == ifp) imo->imo_multicast_ifp = NULL; /* * Drop multicast group membership if we joined * through the interface being detached. */ for (i = 0, gap = 0; i < imo->imo_num_memberships; i++) { if (imo->imo_membership[i]->inm_ifp == ifp) { in_delmulti(imo->imo_membership[i]); gap++; } else if (gap != 0) imo->imo_membership[i - gap] = imo->imo_membership[i]; } imo->imo_num_memberships -= gap; } INP_WUNLOCK(inp); } INP_INFO_RUNLOCK(pcbinfo); } /* * Lookup a PCB based on the local address and port. Caller must hold the * hash lock. No inpcb locks or references are acquired. */ #define INP_LOOKUP_MAPPED_PCB_COST 3 struct inpcb * in_pcblookup_local(struct inpcbinfo *pcbinfo, struct in_addr laddr, u_short lport, int lookupflags, struct ucred *cred) { struct inpcb *inp; #ifdef INET6 int matchwild = 3 + INP_LOOKUP_MAPPED_PCB_COST; #else int matchwild = 3; #endif int wildcard; KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0, ("%s: invalid lookup flags %d", __func__, lookupflags)); INP_HASH_LOCK_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(INADDR_ANY, lport, 0, pcbinfo->ipi_hashmask)]; LIST_FOREACH(inp, head, inp_hash) { #ifdef INET6 /* XXX inp locking */ if ((inp->inp_vflag & INP_IPV4) == 0) continue; #endif if (inp->inp_faddr.s_addr == INADDR_ANY && inp->inp_laddr.s_addr == laddr.s_addr && inp->inp_lport == lport) { /* * Found? */ if (cred == NULL || prison_equal_ip4(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_ip4(inp->inp_cred->cr_prison, cred->cr_prison)) continue; #ifdef INET6 /* XXX inp locking */ if ((inp->inp_vflag & INP_IPV4) == 0) continue; /* * We never select the PCB that has * INP_IPV6 flag and is bound to :: if * we have another PCB which is bound * to 0.0.0.0. If a PCB has the * INP_IPV6 flag, then we set its cost * higher than IPv4 only PCBs. * * Note that the case only happens * when a socket is bound to ::, under * the condition that the use of the * mapped address is allowed. */ if ((inp->inp_vflag & INP_IPV6) != 0) wildcard += INP_LOOKUP_MAPPED_PCB_COST; #endif if (inp->inp_faddr.s_addr != INADDR_ANY) wildcard++; if (inp->inp_laddr.s_addr != INADDR_ANY) { if (laddr.s_addr == INADDR_ANY) wildcard++; else if (inp->inp_laddr.s_addr != laddr.s_addr) continue; } else { if (laddr.s_addr != INADDR_ANY) wildcard++; } if (wildcard < matchwild) { match = inp; matchwild = wildcard; if (matchwild == 0) break; } } } return (match); } } #undef INP_LOOKUP_MAPPED_PCB_COST +#ifdef PCBGROUP /* + * Lookup PCB in hash list, using pcbgroup tables. + */ +static struct inpcb * +in_pcblookup_group(struct inpcbinfo *pcbinfo, struct inpcbgroup *pcbgroup, + struct in_addr faddr, u_int fport_arg, struct in_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(faddr.s_addr, lport, fport, + pcbgroup->ipg_hashmask)]; + LIST_FOREACH(inp, head, inp_pcbgrouphash) { +#ifdef INET6 + /* XXX inp locking */ + if ((inp->inp_vflag & INP_IPV4) == 0) + continue; +#endif + if (inp->inp_faddr.s_addr == faddr.s_addr && + inp->inp_laddr.s_addr == laddr.s_addr && + 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_IP4)) + goto found; + if (tmpinp == NULL) + tmpinp = inp; + } + } + if (tmpinp != NULL) { + inp = tmpinp; + goto found; + } + + /* + * Then look for a wildcard match, if requested. + */ + if ((lookupflags & INPLOOKUP_WILDCARD) != 0) { + struct inpcb *local_wild = NULL, *local_exact = NULL; +#ifdef INET6 + struct inpcb *local_wild_mapped = NULL; +#endif + struct inpcb *jail_wild = NULL; + struct inpcbhead *head; + 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(INADDR_ANY, lport, + 0, pcbinfo->ipi_wildmask)]; + LIST_FOREACH(inp, head, inp_pcbgroup_wild) { +#ifdef INET6 + /* XXX inp locking */ + if ((inp->inp_vflag & INP_IPV4) == 0) + continue; +#endif + if (inp->inp_faddr.s_addr != INADDR_ANY || + inp->inp_lport != lport) + continue; + + /* XXX inp locking */ + if (ifp && ifp->if_type == IFT_FAITH && + (inp->inp_flags & INP_FAITH) == 0) + continue; + + injail = prison_flag(inp->inp_cred, PR_IP4); + if (injail) { + if (prison_check_ip4(inp->inp_cred, + &laddr) != 0) + continue; + } else { + if (local_exact != NULL) + continue; + } + + if (inp->inp_laddr.s_addr == laddr.s_addr) { + if (injail) + goto found; + else + local_exact = inp; + } else if (inp->inp_laddr.s_addr == INADDR_ANY) { +#ifdef INET6 + /* XXX inp locking, NULL check */ + if (inp->inp_vflag & INP_IPV6PROTO) + local_wild_mapped = inp; + else +#endif /* INET6 */ + if (injail) + jail_wild = inp; + else + local_wild = inp; + } + } /* LIST_FOREACH */ + inp = jail_wild; + if (inp == NULL) + inp = local_exact; + if (inp == NULL) + inp = local_wild; +#ifdef INET6 + if (inp == NULL) + inp = local_wild_mapped; +#endif /* defined(INET6) */ + if (inp != NULL) + goto found; + } /* if (lookupflags & INPLOOKUP_WILDCARD) */ + 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 bug", __func__); + return (inp); +} +#endif /* PCBGROUP */ + +/* * Lookup PCB in hash list, using pcbinfo tables. This variation assumes * that the caller has locked the hash list, and will not perform any further * locking or reference operations on either the hash list or the connection. */ static struct inpcb * in_pcblookup_hash_locked(struct inpcbinfo *pcbinfo, struct in_addr faddr, u_int fport_arg, struct in_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(faddr.s_addr, lport, fport, pcbinfo->ipi_hashmask)]; LIST_FOREACH(inp, head, inp_hash) { #ifdef INET6 /* XXX inp locking */ if ((inp->inp_vflag & INP_IPV4) == 0) continue; #endif if (inp->inp_faddr.s_addr == faddr.s_addr && inp->inp_laddr.s_addr == laddr.s_addr && 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_IP4)) 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; #ifdef INET6 struct inpcb *local_wild_mapped = NULL; #endif 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(INADDR_ANY, lport, 0, pcbinfo->ipi_hashmask)]; LIST_FOREACH(inp, head, inp_hash) { #ifdef INET6 /* XXX inp locking */ if ((inp->inp_vflag & INP_IPV4) == 0) continue; #endif if (inp->inp_faddr.s_addr != INADDR_ANY || inp->inp_lport != lport) continue; /* XXX inp locking */ if (ifp && ifp->if_type == IFT_FAITH && (inp->inp_flags & INP_FAITH) == 0) continue; injail = prison_flag(inp->inp_cred, PR_IP4); if (injail) { if (prison_check_ip4(inp->inp_cred, &laddr) != 0) continue; } else { if (local_exact != NULL) continue; } if (inp->inp_laddr.s_addr == laddr.s_addr) { if (injail) return (inp); else local_exact = inp; } else if (inp->inp_laddr.s_addr == INADDR_ANY) { #ifdef INET6 /* XXX inp locking, NULL check */ if (inp->inp_vflag & INP_IPV6PROTO) local_wild_mapped = inp; else #endif /* INET6 */ 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); #ifdef INET6 if (local_wild_mapped != NULL) return (local_wild_mapped); #endif /* defined(INET6) */ } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */ 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 * in_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in_addr faddr, u_int fport, struct in_addr laddr, u_int lport, int lookupflags, struct ifnet *ifp) { struct inpcb *inp; INP_HASH_RLOCK(pcbinfo); inp = in_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 in_pcbgroup.c. */ struct inpcb * in_pcblookup(struct inpcbinfo *pcbinfo, struct in_addr faddr, u_int fport, struct in_addr laddr, u_int lport, int lookupflags, struct ifnet *ifp) { +#if defined(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__)); +#if defined(PCBGROUP) + if (in_pcbgroup_enabled(pcbinfo)) { + pcbgroup = in_pcbgroup_bytuple(pcbinfo, laddr, lport, faddr, + fport); + return (in_pcblookup_group(pcbinfo, pcbgroup, faddr, fport, + laddr, lport, lookupflags, ifp)); + } +#endif return (in_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport, lookupflags, ifp)); } struct inpcb * in_pcblookup_mbuf(struct inpcbinfo *pcbinfo, struct in_addr faddr, u_int fport, struct in_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 (in_pcbgroup_enabled(pcbinfo)) { + pcbgroup = in_pcbgroup_byhash(pcbinfo, M_HASHTYPE_GET(m), + m->m_pkthdr.flowid); + if (pcbgroup != NULL) + return (in_pcblookup_group(pcbinfo, pcbgroup, faddr, + fport, laddr, lport, lookupflags, ifp)); + pcbgroup = in_pcbgroup_bytuple(pcbinfo, laddr, lport, faddr, + fport); + return (in_pcblookup_group(pcbinfo, pcbgroup, faddr, fport, + laddr, lport, lookupflags, ifp)); + } +#endif return (in_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport, lookupflags, ifp)); } #endif /* INET */ /* * Insert PCB onto various hash lists. */ -int -in_pcbinshash(struct inpcb *inp) +static int +in_pcbinshash_internal(struct inpcb *inp, int do_pcbgroup_update) { struct inpcbhead *pcbhash; struct inpcbporthead *pcbporthash; struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; struct inpcbport *phd; u_int32_t hashkey_faddr; INP_WLOCK_ASSERT(inp); INP_HASH_WLOCK_ASSERT(pcbinfo); KASSERT((inp->inp_flags & INP_INHASHLIST) == 0, ("in_pcbinshash: INP_INHASHLIST")); #ifdef INET6 if (inp->inp_vflag & INP_IPV6) hashkey_faddr = inp->in6p_faddr.s6_addr32[3] /* XXX */; else #endif /* INET6 */ hashkey_faddr = inp->inp_faddr.s_addr; pcbhash = &pcbinfo->ipi_hashbase[INP_PCBHASH(hashkey_faddr, inp->inp_lport, inp->inp_fport, pcbinfo->ipi_hashmask)]; pcbporthash = &pcbinfo->ipi_porthashbase[ INP_PCBPORTHASH(inp->inp_lport, pcbinfo->ipi_porthashmask)]; /* * Go through port list and look for a head for this lport. */ LIST_FOREACH(phd, pcbporthash, phd_hash) { if (phd->phd_port == inp->inp_lport) break; } /* * If none exists, malloc one and tack it on. */ if (phd == NULL) { phd = malloc(sizeof(struct inpcbport), M_PCB, M_NOWAIT); if (phd == NULL) { return (ENOBUFS); /* XXX */ } phd->phd_port = inp->inp_lport; LIST_INIT(&phd->phd_pcblist); LIST_INSERT_HEAD(pcbporthash, phd, phd_hash); } inp->inp_phd = phd; LIST_INSERT_HEAD(&phd->phd_pcblist, inp, inp_portlist); LIST_INSERT_HEAD(pcbhash, inp, inp_hash); inp->inp_flags |= INP_INHASHLIST; +#ifdef PCBGROUP + if (do_pcbgroup_update) + in_pcbgroup_update(inp); +#endif return (0); } /* + * For now, there are two public interfaces to insert an inpcb into the hash + * lists -- one that does update pcbgroups, and one that doesn't. The latter + * is used only in the TCP syncache, where in_pcbinshash is called before the + * full 4-tuple is set for the inpcb, and we don't want to install in the + * pcbgroup until later. + * + * XXXRW: This seems like a misfeature. in_pcbinshash should always update + * connection groups, and partially initialised inpcbs should not be exposed + * to either reservation hash tables or pcbgroups. + */ +int +in_pcbinshash(struct inpcb *inp) +{ + + return (in_pcbinshash_internal(inp, 1)); +} + +int +in_pcbinshash_nopcbgroup(struct inpcb *inp) +{ + + return (in_pcbinshash_internal(inp, 0)); +} + +/* * Move PCB to the proper hash bucket when { faddr, fport } have been * changed. NOTE: This does not handle the case of the lport changing (the * hashed port list would have to be updated as well), so the lport must * not change after in_pcbinshash() has been called. */ void in_pcbrehash_mbuf(struct inpcb *inp, struct mbuf *m) { struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; struct inpcbhead *head; u_int32_t hashkey_faddr; INP_WLOCK_ASSERT(inp); INP_HASH_WLOCK_ASSERT(pcbinfo); KASSERT(inp->inp_flags & INP_INHASHLIST, ("in_pcbrehash: !INP_INHASHLIST")); #ifdef INET6 if (inp->inp_vflag & INP_IPV6) hashkey_faddr = inp->in6p_faddr.s6_addr32[3] /* XXX */; else #endif /* INET6 */ hashkey_faddr = inp->inp_faddr.s_addr; head = &pcbinfo->ipi_hashbase[INP_PCBHASH(hashkey_faddr, inp->inp_lport, inp->inp_fport, pcbinfo->ipi_hashmask)]; LIST_REMOVE(inp, inp_hash); LIST_INSERT_HEAD(head, inp, inp_hash); + +#ifdef PCBGROUP + if (m != NULL) + in_pcbgroup_update_mbuf(inp, m); + else + in_pcbgroup_update(inp); +#endif } void in_pcbrehash(struct inpcb *inp) { in_pcbrehash_mbuf(inp, NULL); } /* * Remove PCB from various lists. */ static void in_pcbremlists(struct inpcb *inp) { struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; INP_INFO_WLOCK_ASSERT(pcbinfo); INP_WLOCK_ASSERT(inp); inp->inp_gencnt = ++pcbinfo->ipi_gencnt; if (inp->inp_flags & INP_INHASHLIST) { struct inpcbport *phd = inp->inp_phd; INP_HASH_WLOCK(pcbinfo); LIST_REMOVE(inp, inp_hash); LIST_REMOVE(inp, inp_portlist); if (LIST_FIRST(&phd->phd_pcblist) == NULL) { LIST_REMOVE(phd, phd_hash); free(phd, M_PCB); } INP_HASH_WUNLOCK(pcbinfo); inp->inp_flags &= ~INP_INHASHLIST; } LIST_REMOVE(inp, inp_list); pcbinfo->ipi_count--; +#ifdef PCBGROUP + in_pcbgroup_remove(inp); +#endif } /* * A set label operation has occurred at the socket layer, propagate the * label change into the in_pcb for the socket. */ void in_pcbsosetlabel(struct socket *so) { #ifdef MAC struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("in_pcbsosetlabel: so->so_pcb == NULL")); INP_WLOCK(inp); SOCK_LOCK(so); mac_inpcb_sosetlabel(so, inp); SOCK_UNLOCK(so); INP_WUNLOCK(inp); #endif } /* * ipport_tick runs once per second, determining if random port allocation * should be continued. If more than ipport_randomcps ports have been * allocated in the last second, then we return to sequential port * allocation. We return to random allocation only once we drop below * ipport_randomcps for at least ipport_randomtime seconds. */ static void ipport_tick(void *xtp) { VNET_ITERATOR_DECL(vnet_iter); VNET_LIST_RLOCK_NOSLEEP(); VNET_FOREACH(vnet_iter) { CURVNET_SET(vnet_iter); /* XXX appease INVARIANTS here */ if (V_ipport_tcpallocs <= V_ipport_tcplastcount + V_ipport_randomcps) { if (V_ipport_stoprandom > 0) V_ipport_stoprandom--; } else V_ipport_stoprandom = V_ipport_randomtime; V_ipport_tcplastcount = V_ipport_tcpallocs; CURVNET_RESTORE(); } VNET_LIST_RUNLOCK_NOSLEEP(); callout_reset(&ipport_tick_callout, hz, ipport_tick, NULL); } static void ip_fini(void *xtp) { callout_stop(&ipport_tick_callout); } /* * The ipport_callout should start running at about the time we attach the * inet or inet6 domains. */ static void ipport_tick_init(const void *unused __unused) { /* Start ipport_tick. */ callout_init(&ipport_tick_callout, CALLOUT_MPSAFE); callout_reset(&ipport_tick_callout, 1, ipport_tick, NULL); EVENTHANDLER_REGISTER(shutdown_pre_sync, ip_fini, NULL, SHUTDOWN_PRI_DEFAULT); } SYSINIT(ipport_tick_init, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ipport_tick_init, NULL); void inp_wlock(struct inpcb *inp) { INP_WLOCK(inp); } void inp_wunlock(struct inpcb *inp) { INP_WUNLOCK(inp); } void inp_rlock(struct inpcb *inp) { INP_RLOCK(inp); } void inp_runlock(struct inpcb *inp) { INP_RUNLOCK(inp); } #ifdef INVARIANTS void inp_lock_assert(struct inpcb *inp) { INP_WLOCK_ASSERT(inp); } void inp_unlock_assert(struct inpcb *inp) { INP_UNLOCK_ASSERT(inp); } #endif void inp_apply_all(void (*func)(struct inpcb *, void *), void *arg) { struct inpcb *inp; INP_INFO_RLOCK(&V_tcbinfo); LIST_FOREACH(inp, V_tcbinfo.ipi_listhead, inp_list) { INP_WLOCK(inp); func(inp, arg); INP_WUNLOCK(inp); } INP_INFO_RUNLOCK(&V_tcbinfo); } struct socket * inp_inpcbtosocket(struct inpcb *inp) { INP_WLOCK_ASSERT(inp); return (inp->inp_socket); } struct tcpcb * inp_inpcbtotcpcb(struct inpcb *inp) { INP_WLOCK_ASSERT(inp); return ((struct tcpcb *)inp->inp_ppcb); } int inp_ip_tos_get(const struct inpcb *inp) { return (inp->inp_ip_tos); } void inp_ip_tos_set(struct inpcb *inp, int val) { inp->inp_ip_tos = val; } void inp_4tuple_get(struct inpcb *inp, uint32_t *laddr, uint16_t *lp, uint32_t *faddr, uint16_t *fp) { INP_LOCK_ASSERT(inp); *laddr = inp->inp_laddr.s_addr; *faddr = inp->inp_faddr.s_addr; *lp = inp->inp_lport; *fp = inp->inp_fport; } struct inpcb * so_sotoinpcb(struct socket *so) { return (sotoinpcb(so)); } struct tcpcb * so_sototcpcb(struct socket *so) { return (sototcpcb(so)); } #ifdef DDB static void db_print_indent(int indent) { int i; for (i = 0; i < indent; i++) db_printf(" "); } static void db_print_inconninfo(struct in_conninfo *inc, const char *name, int indent) { char faddr_str[48], laddr_str[48]; db_print_indent(indent); db_printf("%s at %p\n", name, inc); indent += 2; #ifdef INET6 if (inc->inc_flags & INC_ISIPV6) { /* IPv6. */ ip6_sprintf(laddr_str, &inc->inc6_laddr); ip6_sprintf(faddr_str, &inc->inc6_faddr); } else { #endif /* IPv4. */ inet_ntoa_r(inc->inc_laddr, laddr_str); inet_ntoa_r(inc->inc_faddr, faddr_str); #ifdef INET6 } #endif db_print_indent(indent); db_printf("inc_laddr %s inc_lport %u\n", laddr_str, ntohs(inc->inc_lport)); db_print_indent(indent); db_printf("inc_faddr %s inc_fport %u\n", faddr_str, ntohs(inc->inc_fport)); } static void db_print_inpflags(int inp_flags) { int comma; comma = 0; if (inp_flags & INP_RECVOPTS) { db_printf("%sINP_RECVOPTS", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_RECVRETOPTS) { db_printf("%sINP_RECVRETOPTS", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_RECVDSTADDR) { db_printf("%sINP_RECVDSTADDR", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_HDRINCL) { db_printf("%sINP_HDRINCL", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_HIGHPORT) { db_printf("%sINP_HIGHPORT", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_LOWPORT) { db_printf("%sINP_LOWPORT", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_ANONPORT) { db_printf("%sINP_ANONPORT", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_RECVIF) { db_printf("%sINP_RECVIF", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_MTUDISC) { db_printf("%sINP_MTUDISC", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_FAITH) { db_printf("%sINP_FAITH", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_RECVTTL) { db_printf("%sINP_RECVTTL", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_DONTFRAG) { db_printf("%sINP_DONTFRAG", comma ? ", " : ""); comma = 1; } if (inp_flags & IN6P_IPV6_V6ONLY) { db_printf("%sIN6P_IPV6_V6ONLY", comma ? ", " : ""); comma = 1; } if (inp_flags & IN6P_PKTINFO) { db_printf("%sIN6P_PKTINFO", comma ? ", " : ""); comma = 1; } if (inp_flags & IN6P_HOPLIMIT) { db_printf("%sIN6P_HOPLIMIT", comma ? ", " : ""); comma = 1; } if (inp_flags & IN6P_HOPOPTS) { db_printf("%sIN6P_HOPOPTS", comma ? ", " : ""); comma = 1; } if (inp_flags & IN6P_DSTOPTS) { db_printf("%sIN6P_DSTOPTS", comma ? ", " : ""); comma = 1; } if (inp_flags & IN6P_RTHDR) { db_printf("%sIN6P_RTHDR", comma ? ", " : ""); comma = 1; } if (inp_flags & IN6P_RTHDRDSTOPTS) { db_printf("%sIN6P_RTHDRDSTOPTS", comma ? ", " : ""); comma = 1; } if (inp_flags & IN6P_TCLASS) { db_printf("%sIN6P_TCLASS", comma ? ", " : ""); comma = 1; } if (inp_flags & IN6P_AUTOFLOWLABEL) { db_printf("%sIN6P_AUTOFLOWLABEL", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_TIMEWAIT) { db_printf("%sINP_TIMEWAIT", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_ONESBCAST) { db_printf("%sINP_ONESBCAST", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_DROPPED) { db_printf("%sINP_DROPPED", comma ? ", " : ""); comma = 1; } if (inp_flags & INP_SOCKREF) { db_printf("%sINP_SOCKREF", comma ? ", " : ""); comma = 1; } if (inp_flags & IN6P_RFC2292) { db_printf("%sIN6P_RFC2292", comma ? ", " : ""); comma = 1; } if (inp_flags & IN6P_MTU) { db_printf("IN6P_MTU%s", comma ? ", " : ""); comma = 1; } } static void db_print_inpvflag(u_char inp_vflag) { int comma; comma = 0; if (inp_vflag & INP_IPV4) { db_printf("%sINP_IPV4", comma ? ", " : ""); comma = 1; } if (inp_vflag & INP_IPV6) { db_printf("%sINP_IPV6", comma ? ", " : ""); comma = 1; } if (inp_vflag & INP_IPV6PROTO) { db_printf("%sINP_IPV6PROTO", comma ? ", " : ""); comma = 1; } } static void db_print_inpcb(struct inpcb *inp, const char *name, int indent) { db_print_indent(indent); db_printf("%s at %p\n", name, inp); indent += 2; db_print_indent(indent); db_printf("inp_flow: 0x%x\n", inp->inp_flow); db_print_inconninfo(&inp->inp_inc, "inp_conninfo", indent); db_print_indent(indent); db_printf("inp_ppcb: %p inp_pcbinfo: %p inp_socket: %p\n", inp->inp_ppcb, inp->inp_pcbinfo, inp->inp_socket); db_print_indent(indent); db_printf("inp_label: %p inp_flags: 0x%x (", inp->inp_label, inp->inp_flags); db_print_inpflags(inp->inp_flags); db_printf(")\n"); db_print_indent(indent); db_printf("inp_sp: %p inp_vflag: 0x%x (", inp->inp_sp, inp->inp_vflag); db_print_inpvflag(inp->inp_vflag); db_printf(")\n"); db_print_indent(indent); db_printf("inp_ip_ttl: %d inp_ip_p: %d inp_ip_minttl: %d\n", inp->inp_ip_ttl, inp->inp_ip_p, inp->inp_ip_minttl); db_print_indent(indent); #ifdef INET6 if (inp->inp_vflag & INP_IPV6) { db_printf("in6p_options: %p in6p_outputopts: %p " "in6p_moptions: %p\n", inp->in6p_options, inp->in6p_outputopts, inp->in6p_moptions); db_printf("in6p_icmp6filt: %p in6p_cksum %d " "in6p_hops %u\n", inp->in6p_icmp6filt, inp->in6p_cksum, inp->in6p_hops); } else #endif { db_printf("inp_ip_tos: %d inp_ip_options: %p " "inp_ip_moptions: %p\n", inp->inp_ip_tos, inp->inp_options, inp->inp_moptions); } db_print_indent(indent); db_printf("inp_phd: %p inp_gencnt: %ju\n", inp->inp_phd, (uintmax_t)inp->inp_gencnt); } DB_SHOW_COMMAND(inpcb, db_show_inpcb) { struct inpcb *inp; if (!have_addr) { db_printf("usage: show inpcb \n"); return; } inp = (struct inpcb *)addr; db_print_inpcb(inp, "inpcb", 0); } #endif Index: projects/largeSMP/sys/netinet/in_pcb.h =================================================================== --- projects/largeSMP/sys/netinet/in_pcb.h (revision 222782) +++ projects/largeSMP/sys/netinet/in_pcb.h (revision 222783) @@ -1,579 +1,657 @@ /*- * Copyright (c) 1982, 1986, 1990, 1993 * The Regents of the University of California. * 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. * 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.h 8.1 (Berkeley) 6/10/93 * $FreeBSD$ */ #ifndef _NETINET_IN_PCB_H_ #define _NETINET_IN_PCB_H_ #include #include #include #include #ifdef _KERNEL #include #include #include #endif #define in6pcb inpcb /* for KAME src sync over BSD*'s */ #define in6p_sp inp_sp /* for KAME src sync over BSD*'s */ struct inpcbpolicy; /* * struct inpcb is the common protocol control block structure used in most * IP transport protocols. * * Pointers to local and foreign host table entries, local and foreign socket * numbers, and pointers up (to a socket structure) and down (to a * protocol-specific control block) are stored here. */ LIST_HEAD(inpcbhead, inpcb); LIST_HEAD(inpcbporthead, inpcbport); typedef u_quad_t inp_gen_t; /* * PCB with AF_INET6 null bind'ed laddr can receive AF_INET input packet. * So, AF_INET6 null laddr is also used as AF_INET null laddr, by utilizing * the following structure. */ struct in_addr_4in6 { u_int32_t ia46_pad32[3]; struct in_addr ia46_addr4; }; /* * NOTE: ipv6 addrs should be 64-bit aligned, per RFC 2553. in_conninfo has * some extra padding to accomplish this. */ struct in_endpoints { u_int16_t ie_fport; /* foreign port */ u_int16_t ie_lport; /* local port */ /* protocol dependent part, local and foreign addr */ union { /* foreign host table entry */ struct in_addr_4in6 ie46_foreign; struct in6_addr ie6_foreign; } ie_dependfaddr; union { /* local host table entry */ struct in_addr_4in6 ie46_local; struct in6_addr ie6_local; } ie_dependladdr; }; #define ie_faddr ie_dependfaddr.ie46_foreign.ia46_addr4 #define ie_laddr ie_dependladdr.ie46_local.ia46_addr4 #define ie6_faddr ie_dependfaddr.ie6_foreign #define ie6_laddr ie_dependladdr.ie6_local /* * XXX The defines for inc_* are hacks and should be changed to direct * references. */ struct in_conninfo { u_int8_t inc_flags; u_int8_t inc_len; u_int16_t inc_fibnum; /* XXX was pad, 16 bits is plenty */ /* protocol dependent part */ struct in_endpoints inc_ie; }; /* * Flags for inc_flags. */ #define INC_ISIPV6 0x01 #define inc_isipv6 inc_flags /* temp compatability */ #define inc_fport inc_ie.ie_fport #define inc_lport inc_ie.ie_lport #define inc_faddr inc_ie.ie_faddr #define inc_laddr inc_ie.ie_laddr #define inc6_faddr inc_ie.ie6_faddr #define inc6_laddr inc_ie.ie6_laddr struct icmp6_filter; /*- * struct inpcb captures the network layer state for TCP, UDP, and raw IPv4 * and IPv6 sockets. In the case of TCP, further per-connection state is * hung off of inp_ppcb most of the time. Almost all fields of struct inpcb * are static after creation or protected by a per-inpcb rwlock, inp_lock. A * few fields also require the global pcbinfo lock for the inpcb to be held, * when modified, such as the global connection lists and hashes, as well as * binding information (which affects which hash a connection is on). This * model means that connections can be looked up without holding the * per-connection lock, which is important for performance when attempting to * find the connection for a packet given its IP and port tuple. Writing to * these fields that write locks be held on both the inpcb and global locks. * * Key: * (c) - Constant after initialization + * (g) - Protected by the pcbgroup lock * (i) - Protected by the inpcb lock * (p) - Protected by the pcbinfo lock for the inpcb * (s) - Protected by another subsystem's locks * (x) - Undefined locking * * A few other notes: * * When a read lock is held, stability of the field is guaranteed; to write * to a field, a write lock must generally be held. * * netinet/netinet6-layer code should not assume that the inp_socket pointer * is safe to dereference without inp_lock being held, even for protocols * other than TCP (where the inpcb persists during TIMEWAIT even after the * socket has been freed), or there may be close(2)-related races. * * The inp_vflag field is overloaded, and would otherwise ideally be (c). */ struct inpcb { LIST_ENTRY(inpcb) inp_hash; /* (i/p) hash list */ + LIST_ENTRY(inpcb) inp_pcbgrouphash; /* (g/i) hash list */ LIST_ENTRY(inpcb) inp_list; /* (i/p) list for all PCBs for proto */ void *inp_ppcb; /* (i) pointer to per-protocol pcb */ struct inpcbinfo *inp_pcbinfo; /* (c) PCB list info */ + struct inpcbgroup *inp_pcbgroup; /* (g/i) PCB group list */ + LIST_ENTRY(inpcb) inp_pcbgroup_wild; /* (g/i/p) group wildcard entry */ struct socket *inp_socket; /* (i) back pointer to socket */ struct ucred *inp_cred; /* (c) cache of socket cred */ u_int32_t inp_flow; /* (i) IPv6 flow information */ int inp_flags; /* (i) generic IP/datagram flags */ int inp_flags2; /* (i) generic IP/datagram flags #2*/ u_char inp_vflag; /* (i) IP version flag (v4/v6) */ u_char inp_ip_ttl; /* (i) time to live proto */ u_char inp_ip_p; /* (c) protocol proto */ u_char inp_ip_minttl; /* (i) minimum TTL or drop */ uint32_t inp_flowid; /* (x) flow id / queue id */ u_int inp_refcount; /* (i) refcount */ void *inp_pspare[4]; /* (x) rtentry / general use */ u_int inp_ispare[4]; /* general use */ /* Local and foreign ports, local and foreign addr. */ struct in_conninfo inp_inc; /* (i/p) list for PCB's local port */ /* MAC and IPSEC policy information. */ struct label *inp_label; /* (i) MAC label */ struct inpcbpolicy *inp_sp; /* (s) for IPSEC */ /* Protocol-dependent part; options. */ struct { u_char inp4_ip_tos; /* (i) type of service proto */ struct mbuf *inp4_options; /* (i) IP options */ struct ip_moptions *inp4_moptions; /* (i) IP mcast options */ } inp_depend4; struct { /* (i) IP options */ struct mbuf *inp6_options; /* (i) IP6 options for outgoing packets */ struct ip6_pktopts *inp6_outputopts; /* (i) IP multicast options */ struct ip6_moptions *inp6_moptions; /* (i) ICMPv6 code type filter */ struct icmp6_filter *inp6_icmp6filt; /* (i) IPV6_CHECKSUM setsockopt */ int inp6_cksum; short inp6_hops; } inp_depend6; LIST_ENTRY(inpcb) inp_portlist; /* (i/p) */ struct inpcbport *inp_phd; /* (i/p) head of this list */ #define inp_zero_size offsetof(struct inpcb, inp_gencnt) inp_gen_t inp_gencnt; /* (c) generation count */ struct llentry *inp_lle; /* cached L2 information */ struct rtentry *inp_rt; /* cached L3 information */ struct rwlock inp_lock; }; #define inp_fport inp_inc.inc_fport #define inp_lport inp_inc.inc_lport #define inp_faddr inp_inc.inc_faddr #define inp_laddr inp_inc.inc_laddr #define inp_ip_tos inp_depend4.inp4_ip_tos #define inp_options inp_depend4.inp4_options #define inp_moptions inp_depend4.inp4_moptions #define in6p_faddr inp_inc.inc6_faddr #define in6p_laddr inp_inc.inc6_laddr #define in6p_hops inp_depend6.inp6_hops /* default hop limit */ #define in6p_flowinfo inp_flow #define in6p_options inp_depend6.inp6_options #define in6p_outputopts inp_depend6.inp6_outputopts #define in6p_moptions inp_depend6.inp6_moptions #define in6p_icmp6filt inp_depend6.inp6_icmp6filt #define in6p_cksum inp_depend6.inp6_cksum #define inp_vnet inp_pcbinfo->ipi_vnet /* * The range of the generation count, as used in this implementation, is 9e19. * We would have to create 300 billion connections per second for this number * to roll over in a year. This seems sufficiently unlikely that we simply * don't concern ourselves with that possibility. */ /* * Interface exported to userland by various protocols which use inpcbs. Hack * alert -- only define if struct xsocket is in scope. */ #ifdef _SYS_SOCKETVAR_H_ struct xinpcb { size_t xi_len; /* length of this structure */ struct inpcb xi_inp; struct xsocket xi_socket; u_quad_t xi_alignment_hack; }; struct xinpgen { size_t xig_len; /* length of this structure */ u_int xig_count; /* number of PCBs at this time */ inp_gen_t xig_gen; /* generation count at this time */ so_gen_t xig_sogen; /* socket generation count at this time */ }; #endif /* _SYS_SOCKETVAR_H_ */ struct inpcbport { LIST_ENTRY(inpcbport) phd_hash; struct inpcbhead phd_pcblist; u_short phd_port; }; /*- * Global data structure for each high-level protocol (UDP, TCP, ...) in both * IPv4 and IPv6. Holds inpcb lists and information for managing them. * * Each pcbinfo is protected by two locks: ipi_lock and ipi_hash_lock, * the former covering mutable global fields (such as the global pcb list), * and the latter covering the hashed lookup tables. The lock order is: * - * ipi_lock (before) inpcb locks (before) ipi_hash_lock + * ipi_lock (before) inpcb locks (before) {ipi_hash_lock, pcbgroup locks} * * Locking key: * * (c) Constant or nearly constant after initialisation * (g) Locked by ipi_lock - * (h) Read using either ipi_hash_lock or inpcb lock; write requires both. + * (h) Read using either ipi_hash_lock or inpcb lock; write requires both + * (p) Protected by one or more pcbgroup locks * (x) Synchronisation properties poorly defined */ struct inpcbinfo { /* * Global lock protecting global inpcb list, inpcb count, etc. */ struct rwlock ipi_lock; /* * Global list of inpcbs on the protocol. */ struct inpcbhead *ipi_listhead; /* (g) */ u_int ipi_count; /* (g) */ /* * Generation count -- incremented each time a connection is allocated * or freed. */ u_quad_t ipi_gencnt; /* (g) */ /* * Fields associated with port lookup and allocation. */ u_short ipi_lastport; /* (x) */ u_short ipi_lastlow; /* (x) */ u_short ipi_lasthi; /* (x) */ /* * UMA zone from which inpcbs are allocated for this protocol. */ struct uma_zone *ipi_zone; /* (c) */ /* - * Global lock protecting hash lookup tables. + * Connection groups associated with this protocol. These fields are + * constant, but pcbgroup structures themselves are protected by + * per-pcbgroup locks. */ + struct inpcbgroup *ipi_pcbgroups; /* (c) */ + u_int ipi_npcbgroups; /* (c) */ + u_int ipi_hashfields; /* (c) */ + + /* + * Global lock protecting non-pcbgroup hash lookup tables. + */ struct rwlock ipi_hash_lock; /* * Global hash of inpcbs, hashed by local and foreign addresses and * port numbers. */ struct inpcbhead *ipi_hashbase; /* (h) */ u_long ipi_hashmask; /* (h) */ /* * Global hash of inpcbs, hashed by only local port number. */ struct inpcbporthead *ipi_porthashbase; /* (h) */ u_long ipi_porthashmask; /* (h) */ /* + * List of wildcard inpcbs for use with pcbgroups. In the past, was + * per-pcbgroup but is now global. All pcbgroup locks must be held + * to modify the list, so any is sufficient to read it. + */ + struct inpcbhead *ipi_wildbase; /* (p) */ + u_long ipi_wildmask; /* (p) */ + + /* * Pointer to network stack instance */ struct vnet *ipi_vnet; /* (c) */ /* * general use 2 */ void *ipi_pspare[2]; }; +/* + * Connection groups hold sets of connections that have similar CPU/thread + * affinity. Each connection belongs to exactly one connection group. + */ +struct inpcbgroup { + /* + * Per-connection group hash of inpcbs, hashed by local and foreign + * addresses and port numbers. + */ + struct inpcbhead *ipg_hashbase; /* (c) */ + u_long ipg_hashmask; /* (c) */ + + /* + * Notional affinity of this pcbgroup. + */ + u_int ipg_cpu; /* (p) */ + + /* + * Per-connection group lock, not to be confused with ipi_lock. + * Protects the hash table hung off the group, but also the global + * wildcard list in inpcbinfo. + */ + struct mtx ipg_lock; +} __aligned(CACHE_LINE_SIZE); + #define INP_LOCK_INIT(inp, d, t) \ rw_init_flags(&(inp)->inp_lock, (t), RW_RECURSE | RW_DUPOK) #define INP_LOCK_DESTROY(inp) rw_destroy(&(inp)->inp_lock) #define INP_RLOCK(inp) rw_rlock(&(inp)->inp_lock) #define INP_WLOCK(inp) rw_wlock(&(inp)->inp_lock) #define INP_TRY_RLOCK(inp) rw_try_rlock(&(inp)->inp_lock) #define INP_TRY_WLOCK(inp) rw_try_wlock(&(inp)->inp_lock) #define INP_RUNLOCK(inp) rw_runlock(&(inp)->inp_lock) #define INP_WUNLOCK(inp) rw_wunlock(&(inp)->inp_lock) #define INP_TRY_UPGRADE(inp) rw_try_upgrade(&(inp)->inp_lock) #define INP_DOWNGRADE(inp) rw_downgrade(&(inp)->inp_lock) #define INP_WLOCKED(inp) rw_wowned(&(inp)->inp_lock) #define INP_LOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_LOCKED) #define INP_RLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_RLOCKED) #define INP_WLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_WLOCKED) #define INP_UNLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_UNLOCKED) #ifdef _KERNEL /* * These locking functions are for inpcb consumers outside of sys/netinet, * more specifically, they were added for the benefit of TOE drivers. The * macros are reserved for use by the stack. */ void inp_wlock(struct inpcb *); void inp_wunlock(struct inpcb *); void inp_rlock(struct inpcb *); void inp_runlock(struct inpcb *); #ifdef INVARIANTS void inp_lock_assert(struct inpcb *); void inp_unlock_assert(struct inpcb *); #else static __inline void inp_lock_assert(struct inpcb *inp __unused) { } static __inline void inp_unlock_assert(struct inpcb *inp __unused) { } #endif void inp_apply_all(void (*func)(struct inpcb *, void *), void *arg); int inp_ip_tos_get(const struct inpcb *inp); void inp_ip_tos_set(struct inpcb *inp, int val); struct socket * inp_inpcbtosocket(struct inpcb *inp); struct tcpcb * inp_inpcbtotcpcb(struct inpcb *inp); void inp_4tuple_get(struct inpcb *inp, uint32_t *laddr, uint16_t *lp, uint32_t *faddr, uint16_t *fp); #endif /* _KERNEL */ #define INP_INFO_LOCK_INIT(ipi, d) \ rw_init_flags(&(ipi)->ipi_lock, (d), RW_RECURSE) #define INP_INFO_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_lock) #define INP_INFO_RLOCK(ipi) rw_rlock(&(ipi)->ipi_lock) #define INP_INFO_WLOCK(ipi) rw_wlock(&(ipi)->ipi_lock) #define INP_INFO_TRY_RLOCK(ipi) rw_try_rlock(&(ipi)->ipi_lock) #define INP_INFO_TRY_WLOCK(ipi) rw_try_wlock(&(ipi)->ipi_lock) #define INP_INFO_TRY_UPGRADE(ipi) rw_try_upgrade(&(ipi)->ipi_lock) #define INP_INFO_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_lock) #define INP_INFO_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_lock) #define INP_INFO_LOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_LOCKED) #define INP_INFO_RLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_RLOCKED) #define INP_INFO_WLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_WLOCKED) #define INP_INFO_UNLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_UNLOCKED) #define INP_HASH_LOCK_INIT(ipi, d) \ rw_init_flags(&(ipi)->ipi_hash_lock, (d), 0) #define INP_HASH_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_hash_lock) #define INP_HASH_RLOCK(ipi) rw_rlock(&(ipi)->ipi_hash_lock) #define INP_HASH_WLOCK(ipi) rw_wlock(&(ipi)->ipi_hash_lock) #define INP_HASH_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_hash_lock) #define INP_HASH_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_hash_lock) #define INP_HASH_LOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_hash_lock, \ RA_LOCKED) #define INP_HASH_WLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_hash_lock, \ RA_WLOCKED) +#define INP_GROUP_LOCK_INIT(ipg, d) mtx_init(&(ipg)->ipg_lock, (d), NULL, \ + MTX_DEF | MTX_DUPOK) +#define INP_GROUP_LOCK_DESTROY(ipg) mtx_destroy(&(ipg)->ipg_lock) + +#define INP_GROUP_LOCK(ipg) mtx_lock(&(ipg)->ipg_lock) +#define INP_GROUP_LOCK_ASSERT(ipg) mtx_assert(&(ipg)->ipg_lock, MA_OWNED) +#define INP_GROUP_UNLOCK(ipg) mtx_unlock(&(ipg)->ipg_lock) + #define INP_PCBHASH(faddr, lport, fport, mask) \ (((faddr) ^ ((faddr) >> 16) ^ ntohs((lport) ^ (fport))) & (mask)) #define INP_PCBPORTHASH(lport, mask) \ (ntohs((lport)) & (mask)) /* * Flags for inp_vflags -- historically version flags only */ #define INP_IPV4 0x1 #define INP_IPV6 0x2 #define INP_IPV6PROTO 0x4 /* opened under IPv6 protocol */ /* * Flags for inp_flags. */ #define INP_RECVOPTS 0x00000001 /* receive incoming IP options */ #define INP_RECVRETOPTS 0x00000002 /* receive IP options for reply */ #define INP_RECVDSTADDR 0x00000004 /* receive IP dst address */ #define INP_HDRINCL 0x00000008 /* user supplies entire IP header */ #define INP_HIGHPORT 0x00000010 /* user wants "high" port binding */ #define INP_LOWPORT 0x00000020 /* user wants "low" port binding */ #define INP_ANONPORT 0x00000040 /* port chosen for user */ #define INP_RECVIF 0x00000080 /* receive incoming interface */ #define INP_MTUDISC 0x00000100 /* user can do MTU discovery */ #define INP_FAITH 0x00000200 /* accept FAITH'ed connections */ #define INP_RECVTTL 0x00000400 /* receive incoming IP TTL */ #define INP_DONTFRAG 0x00000800 /* don't fragment packet */ #define INP_BINDANY 0x00001000 /* allow bind to any address */ #define INP_INHASHLIST 0x00002000 /* in_pcbinshash() has been called */ #define IN6P_IPV6_V6ONLY 0x00008000 /* restrict AF_INET6 socket for v6 */ #define IN6P_PKTINFO 0x00010000 /* receive IP6 dst and I/F */ #define IN6P_HOPLIMIT 0x00020000 /* receive hoplimit */ #define IN6P_HOPOPTS 0x00040000 /* receive hop-by-hop options */ #define IN6P_DSTOPTS 0x00080000 /* receive dst options after rthdr */ #define IN6P_RTHDR 0x00100000 /* receive routing header */ #define IN6P_RTHDRDSTOPTS 0x00200000 /* receive dstoptions before rthdr */ #define IN6P_TCLASS 0x00400000 /* receive traffic class value */ #define IN6P_AUTOFLOWLABEL 0x00800000 /* attach flowlabel automatically */ #define INP_TIMEWAIT 0x01000000 /* in TIMEWAIT, ppcb is tcptw */ #define INP_ONESBCAST 0x02000000 /* send all-ones broadcast */ #define INP_DROPPED 0x04000000 /* protocol drop flag */ #define INP_SOCKREF 0x08000000 /* strong socket reference */ #define INP_SW_FLOWID 0x10000000 /* software generated flow id */ #define INP_HW_FLOWID 0x20000000 /* hardware generated flow id */ #define IN6P_RFC2292 0x40000000 /* used RFC2292 API on the socket */ #define IN6P_MTU 0x80000000 /* receive path MTU */ #define INP_CONTROLOPTS (INP_RECVOPTS|INP_RECVRETOPTS|INP_RECVDSTADDR|\ INP_RECVIF|INP_RECVTTL|\ IN6P_PKTINFO|IN6P_HOPLIMIT|IN6P_HOPOPTS|\ IN6P_DSTOPTS|IN6P_RTHDR|IN6P_RTHDRDSTOPTS|\ IN6P_TCLASS|IN6P_AUTOFLOWLABEL|IN6P_RFC2292|\ IN6P_MTU) /* * Flags for inp_flags2. */ #define INP_LLE_VALID 0x00000001 /* cached lle is valid */ #define INP_RT_VALID 0x00000002 /* cached rtentry is valid */ +#define INP_PCBGROUPWILD 0x00000004 /* in pcbgroup wildcard list */ /* * Flags passed to in_pcblookup*() functions. */ #define INPLOOKUP_WILDCARD 0x00000001 /* Allow wildcard sockets. */ #define INPLOOKUP_RLOCKPCB 0x00000002 /* Return inpcb read-locked. */ #define INPLOOKUP_WLOCKPCB 0x00000004 /* Return inpcb write-locked. */ #define INPLOOKUP_MASK (INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB | \ INPLOOKUP_WLOCKPCB) #define sotoinpcb(so) ((struct inpcb *)(so)->so_pcb) #define sotoin6pcb(so) sotoinpcb(so) /* for KAME src sync over BSD*'s */ #define INP_SOCKAF(so) so->so_proto->pr_domain->dom_family #define INP_CHECK_SOCKAF(so, af) (INP_SOCKAF(so) == af) +/* + * Constants for pcbinfo.ipi_hashfields. + */ +#define IPI_HASHFIELDS_NONE 0 +#define IPI_HASHFIELDS_2TUPLE 1 +#define IPI_HASHFIELDS_4TUPLE 2 + #ifdef _KERNEL VNET_DECLARE(int, ipport_reservedhigh); VNET_DECLARE(int, ipport_reservedlow); VNET_DECLARE(int, ipport_lowfirstauto); VNET_DECLARE(int, ipport_lowlastauto); VNET_DECLARE(int, ipport_firstauto); VNET_DECLARE(int, ipport_lastauto); VNET_DECLARE(int, ipport_hifirstauto); VNET_DECLARE(int, ipport_hilastauto); VNET_DECLARE(int, ipport_randomized); VNET_DECLARE(int, ipport_randomcps); VNET_DECLARE(int, ipport_randomtime); VNET_DECLARE(int, ipport_stoprandom); VNET_DECLARE(int, ipport_tcpallocs); #define V_ipport_reservedhigh VNET(ipport_reservedhigh) #define V_ipport_reservedlow VNET(ipport_reservedlow) #define V_ipport_lowfirstauto VNET(ipport_lowfirstauto) #define V_ipport_lowlastauto VNET(ipport_lowlastauto) #define V_ipport_firstauto VNET(ipport_firstauto) #define V_ipport_lastauto VNET(ipport_lastauto) #define V_ipport_hifirstauto VNET(ipport_hifirstauto) #define V_ipport_hilastauto VNET(ipport_hilastauto) #define V_ipport_randomized VNET(ipport_randomized) #define V_ipport_randomcps VNET(ipport_randomcps) #define V_ipport_randomtime VNET(ipport_randomtime) #define V_ipport_stoprandom VNET(ipport_stoprandom) #define V_ipport_tcpallocs VNET(ipport_tcpallocs) void in_pcbinfo_destroy(struct inpcbinfo *); void in_pcbinfo_init(struct inpcbinfo *, const char *, struct inpcbhead *, - int, int, char *, uma_init, uma_fini, uint32_t); + int, int, char *, uma_init, uma_fini, uint32_t, u_int); +struct inpcbgroup * + in_pcbgroup_byhash(struct inpcbinfo *, u_int, uint32_t); +struct inpcbgroup * + in_pcbgroup_byinpcb(struct inpcb *); +struct inpcbgroup * + in_pcbgroup_bytuple(struct inpcbinfo *, struct in_addr, u_short, + struct in_addr, u_short); +void in_pcbgroup_destroy(struct inpcbinfo *); +int in_pcbgroup_enabled(struct inpcbinfo *); +void in_pcbgroup_init(struct inpcbinfo *, u_int, int); +void in_pcbgroup_remove(struct inpcb *); +void in_pcbgroup_update(struct inpcb *); +void in_pcbgroup_update_mbuf(struct inpcb *, struct mbuf *); + void in_pcbpurgeif0(struct inpcbinfo *, struct ifnet *); int in_pcballoc(struct socket *, struct inpcbinfo *); int in_pcbbind(struct inpcb *, struct sockaddr *, struct ucred *); int in_pcb_lport(struct inpcb *, struct in_addr *, u_short *, struct ucred *, int); int in_pcbbind_setup(struct inpcb *, struct sockaddr *, in_addr_t *, u_short *, struct ucred *); int in_pcbconnect(struct inpcb *, struct sockaddr *, struct ucred *); int in_pcbconnect_mbuf(struct inpcb *, struct sockaddr *, struct ucred *, struct mbuf *); int in_pcbconnect_setup(struct inpcb *, struct sockaddr *, in_addr_t *, u_short *, in_addr_t *, u_short *, struct inpcb **, struct ucred *); void in_pcbdetach(struct inpcb *); void in_pcbdisconnect(struct inpcb *); void in_pcbdrop(struct inpcb *); void in_pcbfree(struct inpcb *); int in_pcbinshash(struct inpcb *); +int in_pcbinshash_nopcbgroup(struct inpcb *); struct inpcb * in_pcblookup_local(struct inpcbinfo *, struct in_addr, u_short, int, struct ucred *); struct inpcb * in_pcblookup(struct inpcbinfo *, struct in_addr, u_int, struct in_addr, u_int, int, struct ifnet *); struct inpcb * in_pcblookup_mbuf(struct inpcbinfo *, struct in_addr, u_int, struct in_addr, u_int, int, struct ifnet *, struct mbuf *); void in_pcbnotifyall(struct inpcbinfo *pcbinfo, struct in_addr, int, struct inpcb *(*)(struct inpcb *, int)); void in_pcbref(struct inpcb *); void in_pcbrehash(struct inpcb *); void in_pcbrehash_mbuf(struct inpcb *, struct mbuf *); int in_pcbrele(struct inpcb *); int in_pcbrele_rlocked(struct inpcb *); int in_pcbrele_wlocked(struct inpcb *); void in_pcbsetsolabel(struct socket *so); int in_getpeeraddr(struct socket *so, struct sockaddr **nam); int in_getsockaddr(struct socket *so, struct sockaddr **nam); struct sockaddr * in_sockaddr(in_port_t port, struct in_addr *addr); void in_pcbsosetlabel(struct socket *so); #endif /* _KERNEL */ #endif /* !_NETINET_IN_PCB_H_ */ Index: projects/largeSMP/sys/netinet/in_pcbgroup.c =================================================================== --- projects/largeSMP/sys/netinet/in_pcbgroup.c (nonexistent) +++ projects/largeSMP/sys/netinet/in_pcbgroup.c (revision 222783) @@ -0,0 +1,457 @@ +/*- + * Copyright (c) 2010-2011 Juniper Networks, Inc. + * All rights reserved. + * + * This software was 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. + * + * 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_inet6.h" + +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#ifdef INET6 +#include +#endif /* INET6 */ + +/* + * pcbgroups, or "connection groups" are based on Willman, Rixner, and Cox's + * 2006 USENIX paper, "An Evaluation of Network Stack Parallelization + * Strategies in Modern Operating Systems". This implementation differs + * significantly from that described in the paper, in that it attempts to + * introduce not just notions of affinity for connections and distribute work + * so as to reduce lock contention, but also align those notions with + * hardware work distribution strategies such as RSS. In this construction, + * connection groups supplement, rather than replace, existing reservation + * tables for protocol 4-tuples, offering CPU-affine lookup tables with + * minimal cache line migration and lock contention during steady state + * operation. + * + * Internet protocols, such as UDP and TCP, register to use connection groups + * by providing an ipi_hashfields value other than IPI_HASHFIELDS_NONE; this + * indicates to the connection group code whether a 2-tuple or 4-tuple is + * used as an argument to hashes that assign a connection to a particular + * group. This must be aligned with any hardware offloaded distribution + * model, such as RSS or similar approaches taken in embedded network boards. + * Wildcard sockets require special handling, as in Willman 2006, and are + * shared between connection groups -- while being protected by group-local + * locks. This means that connection establishment and teardown can be + * signficantly more expensive than without connection groups, but that + * steady-state processing can be significantly faster. + * + * Most of the implementation of connection groups is in this file; however, + * connection group lookup is implemented in in_pcb.c alongside reservation + * table lookups -- see in_pcblookup_group(). + * + * TODO: + * + * Implement dynamic rebalancing of buckets with connection groups; when + * load is unevenly distributed, search for more optimal balancing on + * demand. This might require scaling up the number of connection groups + * by <<1. + * + * Provide an IP 2-tuple or 4-tuple netisr m2cpu handler based on connection + * groups for ip_input and ip6_input, allowing non-offloaded work + * distribution. + * + * Expose effective CPU affinity of connections to userspace using socket + * options. + * + * Investigate per-connection affinity overrides based on socket options; an + * option could be set, certainly resulting in work being distributed + * differently in software, and possibly propagated to supporting hardware + * with TCAMs or hardware hash tables. This might require connections to + * exist in more than one connection group at a time. + * + * Hook netisr thread reconfiguration events, and propagate those to RSS so + * that rebalancing can occur when the thread pool grows or shrinks. + * + * Expose per-pcbgroup statistics to userspace monitoring tools such as + * netstat, in order to allow better debugging and profiling. + */ + +void +in_pcbgroup_init(struct inpcbinfo *pcbinfo, u_int hashfields, + int hash_nelements) +{ + struct inpcbgroup *pcbgroup; + u_int numpcbgroups, pgn; + + /* + * Only enable connection groups for a protocol if it has been + * specifically requested. + */ + if (hashfields == IPI_HASHFIELDS_NONE) + return; + + /* + * Connection groups are about multi-processor load distribution, + * lock contention, and connection CPU affinity. As such, no point + * in turning them on for a uniprocessor machine, it only wastes + * memory. + */ + if (mp_ncpus == 1) + return; + + /* + * Use one group per CPU for now. If we decide to do dynamic + * rebalancing a la RSS, we'll need to shift left by at least 1. + */ + numpcbgroups = mp_ncpus; + + pcbinfo->ipi_hashfields = hashfields; + pcbinfo->ipi_pcbgroups = malloc(numpcbgroups * + sizeof(*pcbinfo->ipi_pcbgroups), M_PCB, M_WAITOK | M_ZERO); + pcbinfo->ipi_npcbgroups = numpcbgroups; + pcbinfo->ipi_wildbase = hashinit(hash_nelements, M_PCB, + &pcbinfo->ipi_wildmask); + for (pgn = 0; pgn < pcbinfo->ipi_npcbgroups; pgn++) { + pcbgroup = &pcbinfo->ipi_pcbgroups[pgn]; + pcbgroup->ipg_hashbase = hashinit(hash_nelements, M_PCB, + &pcbgroup->ipg_hashmask); + INP_GROUP_LOCK_INIT(pcbgroup, "pcbgroup"); + + /* + * Initialise notional affinity of the pcbgroup -- for RSS, + * we want the same notion of affinity as NICs to be used. + * Just round robin for the time being. + */ + pcbgroup->ipg_cpu = (pgn % mp_ncpus); + } +} + +void +in_pcbgroup_destroy(struct inpcbinfo *pcbinfo) +{ + struct inpcbgroup *pcbgroup; + u_int pgn; + + if (pcbinfo->ipi_npcbgroups == 0) + return; + + for (pgn = 0; pgn < pcbinfo->ipi_npcbgroups; pgn++) { + pcbgroup = &pcbinfo->ipi_pcbgroups[pgn]; + KASSERT(LIST_EMPTY(pcbinfo->ipi_listhead), + ("in_pcbinfo_destroy: listhead not empty")); + INP_GROUP_LOCK_DESTROY(pcbgroup); + hashdestroy(pcbgroup->ipg_hashbase, M_PCB, + pcbgroup->ipg_hashmask); + } + hashdestroy(pcbinfo->ipi_wildbase, M_PCB, pcbinfo->ipi_wildmask); + free(pcbinfo->ipi_pcbgroups, M_PCB); + pcbinfo->ipi_pcbgroups = NULL; + pcbinfo->ipi_npcbgroups = 0; + pcbinfo->ipi_hashfields = 0; +} + +/* + * Given a hash of whatever the covered tuple might be, return a pcbgroup + * index. + */ +static __inline u_int +in_pcbgroup_getbucket(struct inpcbinfo *pcbinfo, uint32_t hash) +{ + + return (hash % pcbinfo->ipi_npcbgroups); +} + +/* + * Map a (hashtype, hash) tuple into a connection group, or NULL if the hash + * information is insufficient to identify the pcbgroup. + */ +struct inpcbgroup * +in_pcbgroup_byhash(struct inpcbinfo *pcbinfo, u_int hashtype, uint32_t hash) +{ + + return (NULL); +} + +static struct inpcbgroup * +in_pcbgroup_bymbuf(struct inpcbinfo *pcbinfo, struct mbuf *m) +{ + + return (in_pcbgroup_byhash(pcbinfo, M_HASHTYPE_GET(m), + m->m_pkthdr.flowid)); +} + +struct inpcbgroup * +in_pcbgroup_bytuple(struct inpcbinfo *pcbinfo, struct in_addr laddr, + u_short lport, struct in_addr faddr, u_short fport) +{ + uint32_t hash; + + switch (pcbinfo->ipi_hashfields) { + case IPI_HASHFIELDS_4TUPLE: + hash = faddr.s_addr ^ fport; + break; + + case IPI_HASHFIELDS_2TUPLE: + hash = faddr.s_addr ^ laddr.s_addr; + break; + + default: + hash = 0; + } + return (&pcbinfo->ipi_pcbgroups[in_pcbgroup_getbucket(pcbinfo, + hash)]); +} + +struct inpcbgroup * +in_pcbgroup_byinpcb(struct inpcb *inp) +{ + + return (in_pcbgroup_bytuple(inp->inp_pcbinfo, inp->inp_laddr, + inp->inp_lport, inp->inp_faddr, inp->inp_fport)); +} + +static void +in_pcbwild_add(struct inpcb *inp) +{ + struct inpcbinfo *pcbinfo; + struct inpcbhead *head; + u_int pgn; + + INP_WLOCK_ASSERT(inp); + KASSERT(!(inp->inp_flags2 & INP_PCBGROUPWILD), + ("%s: is wild",__func__)); + + pcbinfo = inp->inp_pcbinfo; + for (pgn = 0; pgn < pcbinfo->ipi_npcbgroups; pgn++) + INP_GROUP_LOCK(&pcbinfo->ipi_pcbgroups[pgn]); + head = &pcbinfo->ipi_wildbase[INP_PCBHASH(INADDR_ANY, inp->inp_lport, + 0, pcbinfo->ipi_wildmask)]; + LIST_INSERT_HEAD(head, inp, inp_pcbgroup_wild); + inp->inp_flags2 |= INP_PCBGROUPWILD; + for (pgn = 0; pgn < pcbinfo->ipi_npcbgroups; pgn++) + INP_GROUP_UNLOCK(&pcbinfo->ipi_pcbgroups[pgn]); +} + +static void +in_pcbwild_remove(struct inpcb *inp) +{ + struct inpcbinfo *pcbinfo; + u_int pgn; + + INP_WLOCK_ASSERT(inp); + KASSERT((inp->inp_flags2 & INP_PCBGROUPWILD), + ("%s: not wild", __func__)); + + pcbinfo = inp->inp_pcbinfo; + for (pgn = 0; pgn < pcbinfo->ipi_npcbgroups; pgn++) + INP_GROUP_LOCK(&pcbinfo->ipi_pcbgroups[pgn]); + LIST_REMOVE(inp, inp_pcbgroup_wild); + for (pgn = 0; pgn < pcbinfo->ipi_npcbgroups; pgn++) + INP_GROUP_UNLOCK(&pcbinfo->ipi_pcbgroups[pgn]); + inp->inp_flags2 &= ~INP_PCBGROUPWILD; +} + +static __inline int +in_pcbwild_needed(struct inpcb *inp) +{ + +#ifdef INET6 + if (inp->inp_vflag & INP_IPV6) + return (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)); + else +#endif + return (inp->inp_faddr.s_addr == htonl(INADDR_ANY)); +} + +static void +in_pcbwild_update_internal(struct inpcb *inp) +{ + int wildcard_needed; + + wildcard_needed = in_pcbwild_needed(inp); + if (wildcard_needed && !(inp->inp_flags2 & INP_PCBGROUPWILD)) + in_pcbwild_add(inp); + else if (!wildcard_needed && (inp->inp_flags2 & INP_PCBGROUPWILD)) + in_pcbwild_remove(inp); +} + +/* + * Update the pcbgroup of an inpcb, which might include removing an old + * pcbgroup reference and/or adding a new one. Wildcard processing is not + * performed here, although ideally we'll never install a pcbgroup for a + * wildcard inpcb (asserted below). + */ +static void +in_pcbgroup_update_internal(struct inpcbinfo *pcbinfo, + struct inpcbgroup *newpcbgroup, struct inpcb *inp) +{ + struct inpcbgroup *oldpcbgroup; + struct inpcbhead *pcbhash; + uint32_t hashkey_faddr; + + INP_WLOCK_ASSERT(inp); + + oldpcbgroup = inp->inp_pcbgroup; + if (oldpcbgroup != NULL && oldpcbgroup != newpcbgroup) { + INP_GROUP_LOCK(oldpcbgroup); + LIST_REMOVE(inp, inp_pcbgrouphash); + inp->inp_pcbgroup = NULL; + INP_GROUP_UNLOCK(oldpcbgroup); + } + if (newpcbgroup != NULL && oldpcbgroup != newpcbgroup) { +#ifdef INET6 + if (inp->inp_vflag & INP_IPV6) + hashkey_faddr = inp->in6p_faddr.s6_addr32[3]; /* XXX */ + else +#endif + hashkey_faddr = inp->inp_faddr.s_addr; + INP_GROUP_LOCK(newpcbgroup); + pcbhash = &newpcbgroup->ipg_hashbase[ + INP_PCBHASH(hashkey_faddr, inp->inp_lport, inp->inp_fport, + newpcbgroup->ipg_hashmask)]; + LIST_INSERT_HEAD(pcbhash, inp, inp_pcbgrouphash); + inp->inp_pcbgroup = newpcbgroup; + INP_GROUP_UNLOCK(newpcbgroup); + } + + KASSERT(!(newpcbgroup != NULL && in_pcbwild_needed(inp)), + ("%s: pcbgroup and wildcard!", __func__)); +} + +/* + * Two update paths: one in which the 4-tuple on an inpcb has been updated + * and therefore connection groups may need to change (or a wildcard entry + * may needed to be installed), and another in which the 4-tuple has been + * set as a result of a packet received, in which case we may be able to use + * the hash on the mbuf to avoid doing a software hash calculation for RSS. + * + * In each case: first, let the wildcard code have a go at placing it as a + * wildcard socket. If it was a wildcard, or if the connection has been + * dropped, then no pcbgroup is required (so potentially clear it); + * otherwise, calculate and update the pcbgroup for the inpcb. + */ +void +in_pcbgroup_update(struct inpcb *inp) +{ + struct inpcbinfo *pcbinfo; + struct inpcbgroup *newpcbgroup; + + INP_WLOCK_ASSERT(inp); + + pcbinfo = inp->inp_pcbinfo; + if (!in_pcbgroup_enabled(pcbinfo)) + return; + + in_pcbwild_update_internal(inp); + if (!(inp->inp_flags2 & INP_PCBGROUPWILD) && + !(inp->inp_flags & INP_DROPPED)) { +#ifdef INET6 + if (inp->inp_vflag & INP_IPV6) + newpcbgroup = in6_pcbgroup_byinpcb(inp); + else +#endif + newpcbgroup = in_pcbgroup_byinpcb(inp); + } else + newpcbgroup = NULL; + in_pcbgroup_update_internal(pcbinfo, newpcbgroup, inp); +} + +void +in_pcbgroup_update_mbuf(struct inpcb *inp, struct mbuf *m) +{ + struct inpcbinfo *pcbinfo; + struct inpcbgroup *newpcbgroup; + + INP_WLOCK_ASSERT(inp); + + pcbinfo = inp->inp_pcbinfo; + if (!in_pcbgroup_enabled(pcbinfo)) + return; + + /* + * Possibly should assert !INP_PCBGROUPWILD rather than testing for + * it; presumably this function should never be called for anything + * other than non-wildcard socket? + */ + in_pcbwild_update_internal(inp); + if (!(inp->inp_flags2 & INP_PCBGROUPWILD) && + !(inp->inp_flags & INP_DROPPED)) { + newpcbgroup = in_pcbgroup_bymbuf(pcbinfo, m); +#ifdef INET6 + if (inp->inp_vflag & INP_IPV6) { + if (newpcbgroup == NULL) + newpcbgroup = in6_pcbgroup_byinpcb(inp); + } else { +#endif + if (newpcbgroup == NULL) + newpcbgroup = in_pcbgroup_byinpcb(inp); +#ifdef INET6 + } +#endif + } else + newpcbgroup = NULL; + in_pcbgroup_update_internal(pcbinfo, newpcbgroup, inp); +} + +/* + * Remove pcbgroup entry and optional pcbgroup wildcard entry for this inpcb. + */ +void +in_pcbgroup_remove(struct inpcb *inp) +{ + struct inpcbgroup *pcbgroup; + + INP_WLOCK_ASSERT(inp); + + if (!in_pcbgroup_enabled(inp->inp_pcbinfo)) + return; + + if (inp->inp_flags2 & INP_PCBGROUPWILD) + in_pcbwild_remove(inp); + + pcbgroup = inp->inp_pcbgroup; + if (pcbgroup != NULL) { + INP_GROUP_LOCK(pcbgroup); + LIST_REMOVE(inp, inp_pcbgrouphash); + inp->inp_pcbgroup = NULL; + INP_GROUP_UNLOCK(pcbgroup); + } +} + +/* + * Query whether or not it is appropriate to use pcbgroups to look up inpcbs + * for a protocol. + */ +int +in_pcbgroup_enabled(struct inpcbinfo *pcbinfo) +{ + + return (pcbinfo->ipi_npcbgroups > 0); +} Property changes on: projects/largeSMP/sys/netinet/in_pcbgroup.c ___________________________________________________________________ Added: svn:eol-style ## -0,0 +1 ## +native \ No newline at end of property Added: svn:keywords ## -0,0 +1 ## +FreeBSD=%H \ No newline at end of property Added: svn:mime-type ## -0,0 +1 ## +text/plain \ No newline at end of property Index: projects/largeSMP/sys/netinet/ip_divert.c =================================================================== --- projects/largeSMP/sys/netinet/ip_divert.c (revision 222782) +++ projects/largeSMP/sys/netinet/ip_divert.c (revision 222783) @@ -1,796 +1,797 @@ /*- * 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. */ #include __FBSDID("$FreeBSD$"); #if !defined(KLD_MODULE) #include "opt_inet.h" #include "opt_sctp.h" #ifndef INET #error "IPDIVERT requires INET." #endif #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef SCTP #include #endif #include /* * Divert sockets */ /* * Allocate enough space to hold a full IP packet */ #define DIVSNDQ (65536 + 100) #define DIVRCVQ (65536 + 100) /* * Divert sockets work in conjunction with ipfw or other packet filters, * see the divert(4) manpage for features. * Packets are selected by the packet filter and tagged with an * MTAG_IPFW_RULE tag carrying the 'divert port' number (as set by * the packet filter) and information on the matching filter rule for * subsequent reinjection. The divert_port is used to put the packet * on the corresponding divert socket, while the rule number is passed * up (at least partially) as the sin_port in the struct sockaddr. * * Packets written to the divert socket carry in sin_addr a * destination address, and in sin_port the number of the filter rule * after which to continue processing. * If the destination address is INADDR_ANY, the packet is treated as * as outgoing and sent to ip_output(); otherwise it is treated as * incoming and sent to ip_input(). * Further, sin_zero carries some information on the interface, * which can be used in the reinject -- see comments in the code. * * On reinjection, processing in ip_input() and ip_output() * will be exactly the same as for the original packet, except that * packet filter processing will start at the rule number after the one * written in the sin_port (ipfw does not allow a rule #0, so sin_port=0 * will apply the entire ruleset to the packet). */ /* Internal variables. */ static VNET_DEFINE(struct inpcbhead, divcb); static VNET_DEFINE(struct inpcbinfo, divcbinfo); #define V_divcb VNET(divcb) #define V_divcbinfo VNET(divcbinfo) static u_long div_sendspace = DIVSNDQ; /* XXX sysctl ? */ static u_long div_recvspace = DIVRCVQ; /* XXX sysctl ? */ static eventhandler_tag ip_divert_event_tag; /* * Initialize divert connection block queue. */ static void div_zone_change(void *tag) { uma_zone_set_max(V_divcbinfo.ipi_zone, maxsockets); } static int div_inpcb_init(void *mem, int size, int flags) { struct inpcb *inp = mem; INP_LOCK_INIT(inp, "inp", "divinp"); return (0); } static void div_inpcb_fini(void *mem, int size) { struct inpcb *inp = mem; INP_LOCK_DESTROY(inp); } static void div_init(void) { /* * XXX We don't use the hash list for divert IP, but it's easier to * allocate one-entry hash lists than it is to check all over the * place for hashbase == NULL. */ in_pcbinfo_init(&V_divcbinfo, "div", &V_divcb, 1, 1, "divcb", - div_inpcb_init, div_inpcb_fini, UMA_ZONE_NOFREE); + div_inpcb_init, div_inpcb_fini, UMA_ZONE_NOFREE, + IPI_HASHFIELDS_NONE); } static void div_destroy(void) { in_pcbinfo_destroy(&V_divcbinfo); } /* * IPPROTO_DIVERT is not in the real IP protocol number space; this * function should never be called. Just in case, drop any packets. */ static void div_input(struct mbuf *m, int off) { KMOD_IPSTAT_INC(ips_noproto); m_freem(m); } /* * Divert a packet by passing it up to the divert socket at port 'port'. * * Setup generic address and protocol structures for div_input routine, * then pass them along with mbuf chain. */ static void divert_packet(struct mbuf *m, int incoming) { struct ip *ip; struct inpcb *inp; struct socket *sa; u_int16_t nport; struct sockaddr_in divsrc; struct m_tag *mtag; mtag = m_tag_locate(m, MTAG_IPFW_RULE, 0, NULL); if (mtag == NULL) { m_freem(m); return; } /* Assure header */ if (m->m_len < sizeof(struct ip) && (m = m_pullup(m, sizeof(struct ip))) == 0) return; ip = mtod(m, struct ip *); /* Delayed checksums are currently not compatible with divert. */ if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { ip->ip_len = ntohs(ip->ip_len); in_delayed_cksum(m); m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; ip->ip_len = htons(ip->ip_len); } #ifdef SCTP if (m->m_pkthdr.csum_flags & CSUM_SCTP) { ip->ip_len = ntohs(ip->ip_len); sctp_delayed_cksum(m, (uint32_t)(ip->ip_hl << 2)); m->m_pkthdr.csum_flags &= ~CSUM_SCTP; ip->ip_len = htons(ip->ip_len); } #endif bzero(&divsrc, sizeof(divsrc)); divsrc.sin_len = sizeof(divsrc); divsrc.sin_family = AF_INET; /* record matching rule, in host format */ divsrc.sin_port = ((struct ipfw_rule_ref *)(mtag+1))->rulenum; /* * Record receive interface address, if any. * But only for incoming packets. */ if (incoming) { struct ifaddr *ifa; struct ifnet *ifp; /* Sanity check */ M_ASSERTPKTHDR(m); /* Find IP address for receive interface */ ifp = m->m_pkthdr.rcvif; if_addr_rlock(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET) continue; divsrc.sin_addr = ((struct sockaddr_in *) ifa->ifa_addr)->sin_addr; break; } if_addr_runlock(ifp); } /* * Record the incoming interface name whenever we have one. */ if (m->m_pkthdr.rcvif) { /* * Hide the actual interface name in there in the * sin_zero array. XXX This needs to be moved to a * different sockaddr type for divert, e.g. * sockaddr_div with multiple fields like * sockaddr_dl. Presently we have only 7 bytes * but that will do for now as most interfaces * are 4 or less + 2 or less bytes for unit. * There is probably a faster way of doing this, * possibly taking it from the sockaddr_dl on the iface. * This solves the problem of a P2P link and a LAN interface * having the same address, which can result in the wrong * interface being assigned to the packet when fed back * into the divert socket. Theoretically if the daemon saves * and re-uses the sockaddr_in as suggested in the man pages, * this iface name will come along for the ride. * (see div_output for the other half of this.) */ strlcpy(divsrc.sin_zero, m->m_pkthdr.rcvif->if_xname, sizeof(divsrc.sin_zero)); } /* Put packet on socket queue, if any */ sa = NULL; nport = htons((u_int16_t)(((struct ipfw_rule_ref *)(mtag+1))->info)); INP_INFO_RLOCK(&V_divcbinfo); LIST_FOREACH(inp, &V_divcb, inp_list) { /* XXX why does only one socket match? */ if (inp->inp_lport == nport) { INP_RLOCK(inp); sa = inp->inp_socket; SOCKBUF_LOCK(&sa->so_rcv); if (sbappendaddr_locked(&sa->so_rcv, (struct sockaddr *)&divsrc, m, (struct mbuf *)0) == 0) { SOCKBUF_UNLOCK(&sa->so_rcv); sa = NULL; /* force mbuf reclaim below */ } else sorwakeup_locked(sa); INP_RUNLOCK(inp); break; } } INP_INFO_RUNLOCK(&V_divcbinfo); if (sa == NULL) { m_freem(m); KMOD_IPSTAT_INC(ips_noproto); KMOD_IPSTAT_DEC(ips_delivered); } } /* * Deliver packet back into the IP processing machinery. * * If no address specified, or address is 0.0.0.0, send to ip_output(); * otherwise, send to ip_input() and mark as having been received on * the interface with that address. */ static int div_output(struct socket *so, struct mbuf *m, struct sockaddr_in *sin, struct mbuf *control) { struct m_tag *mtag; struct ipfw_rule_ref *dt; int error = 0; struct mbuf *options; /* * An mbuf may hasn't come from userland, but we pretend * that it has. */ m->m_pkthdr.rcvif = NULL; m->m_nextpkt = NULL; M_SETFIB(m, so->so_fibnum); if (control) m_freem(control); /* XXX */ mtag = m_tag_locate(m, MTAG_IPFW_RULE, 0, NULL); if (mtag == NULL) { /* this should be normal */ mtag = m_tag_alloc(MTAG_IPFW_RULE, 0, sizeof(struct ipfw_rule_ref), M_NOWAIT | M_ZERO); if (mtag == NULL) { error = ENOBUFS; goto cantsend; } m_tag_prepend(m, mtag); } dt = (struct ipfw_rule_ref *)(mtag+1); /* Loopback avoidance and state recovery */ if (sin) { int i; /* set the starting point. We provide a non-zero slot, * but a non_matching chain_id to skip that info and use * the rulenum/rule_id. */ dt->slot = 1; /* dummy, chain_id is invalid */ dt->chain_id = 0; dt->rulenum = sin->sin_port+1; /* host format ? */ dt->rule_id = 0; /* * Find receive interface with the given name, stuffed * (if it exists) in the sin_zero[] field. * The name is user supplied data so don't trust its size * or that it is zero terminated. */ for (i = 0; i < sizeof(sin->sin_zero) && sin->sin_zero[i]; i++) ; if ( i > 0 && i < sizeof(sin->sin_zero)) m->m_pkthdr.rcvif = ifunit(sin->sin_zero); } /* Reinject packet into the system as incoming or outgoing */ if (!sin || sin->sin_addr.s_addr == 0) { struct ip *const ip = mtod(m, struct ip *); struct inpcb *inp; dt->info |= IPFW_IS_DIVERT | IPFW_INFO_OUT; inp = sotoinpcb(so); INP_RLOCK(inp); /* * Don't allow both user specified and setsockopt options, * and don't allow packet length sizes that will crash */ if (((ip->ip_hl != (sizeof (*ip) >> 2)) && inp->inp_options) || ((u_short)ntohs(ip->ip_len) > m->m_pkthdr.len)) { error = EINVAL; INP_RUNLOCK(inp); m_freem(m); } else { /* Convert fields to host order for ip_output() */ ip->ip_len = ntohs(ip->ip_len); ip->ip_off = ntohs(ip->ip_off); /* Send packet to output processing */ KMOD_IPSTAT_INC(ips_rawout); /* XXX */ #ifdef MAC mac_inpcb_create_mbuf(inp, m); #endif /* * Get ready to inject the packet into ip_output(). * Just in case socket options were specified on the * divert socket, we duplicate them. This is done * to avoid having to hold the PCB locks over the call * to ip_output(), as doing this results in a number of * lock ordering complexities. * * Note that we set the multicast options argument for * ip_output() to NULL since it should be invariant that * they are not present. */ KASSERT(inp->inp_moptions == NULL, ("multicast options set on a divert socket")); options = NULL; /* * XXXCSJP: It is unclear to me whether or not it makes * sense for divert sockets to have options. However, * for now we will duplicate them with the INP locks * held so we can use them in ip_output() without * requring a reference to the pcb. */ if (inp->inp_options != NULL) { options = m_dup(inp->inp_options, M_DONTWAIT); if (options == NULL) error = ENOBUFS; } INP_RUNLOCK(inp); if (error == ENOBUFS) { m_freem(m); return (error); } error = ip_output(m, options, NULL, ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) | IP_ALLOWBROADCAST | IP_RAWOUTPUT, NULL, NULL); if (options != NULL) m_freem(options); } } else { dt->info |= IPFW_IS_DIVERT | IPFW_INFO_IN; if (m->m_pkthdr.rcvif == NULL) { /* * No luck with the name, check by IP address. * Clear the port and the ifname to make sure * there are no distractions for ifa_ifwithaddr. */ struct ifaddr *ifa; bzero(sin->sin_zero, sizeof(sin->sin_zero)); sin->sin_port = 0; ifa = ifa_ifwithaddr((struct sockaddr *) sin); if (ifa == NULL) { error = EADDRNOTAVAIL; goto cantsend; } m->m_pkthdr.rcvif = ifa->ifa_ifp; ifa_free(ifa); } #ifdef MAC mac_socket_create_mbuf(so, m); #endif /* Send packet to input processing via netisr */ netisr_queue_src(NETISR_IP, (uintptr_t)so, m); } return error; cantsend: m_freem(m); return error; } static int div_attach(struct socket *so, int proto, struct thread *td) { struct inpcb *inp; int error; inp = sotoinpcb(so); KASSERT(inp == NULL, ("div_attach: inp != NULL")); if (td != NULL) { error = priv_check(td, PRIV_NETINET_DIVERT); if (error) return (error); } error = soreserve(so, div_sendspace, div_recvspace); if (error) return error; INP_INFO_WLOCK(&V_divcbinfo); error = in_pcballoc(so, &V_divcbinfo); if (error) { INP_INFO_WUNLOCK(&V_divcbinfo); return error; } inp = (struct inpcb *)so->so_pcb; INP_INFO_WUNLOCK(&V_divcbinfo); inp->inp_ip_p = proto; inp->inp_vflag |= INP_IPV4; inp->inp_flags |= INP_HDRINCL; INP_WUNLOCK(inp); return 0; } static void div_detach(struct socket *so) { struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("div_detach: inp == NULL")); INP_INFO_WLOCK(&V_divcbinfo); INP_WLOCK(inp); in_pcbdetach(inp); in_pcbfree(inp); INP_INFO_WUNLOCK(&V_divcbinfo); } static int div_bind(struct socket *so, struct sockaddr *nam, struct thread *td) { struct inpcb *inp; int error; inp = sotoinpcb(so); KASSERT(inp != NULL, ("div_bind: inp == NULL")); /* in_pcbbind assumes that nam is a sockaddr_in * and in_pcbbind requires a valid address. Since divert * sockets don't we need to make sure the address is * filled in properly. * XXX -- divert should not be abusing in_pcbind * and should probably have its own family. */ if (nam->sa_family != AF_INET) return EAFNOSUPPORT; ((struct sockaddr_in *)nam)->sin_addr.s_addr = INADDR_ANY; INP_INFO_WLOCK(&V_divcbinfo); INP_WLOCK(inp); INP_HASH_WLOCK(&V_divcbinfo); error = in_pcbbind(inp, nam, td->td_ucred); INP_HASH_WUNLOCK(&V_divcbinfo); INP_WUNLOCK(inp); INP_INFO_WUNLOCK(&V_divcbinfo); return error; } static int div_shutdown(struct socket *so) { struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("div_shutdown: inp == NULL")); INP_WLOCK(inp); socantsendmore(so); INP_WUNLOCK(inp); return 0; } static int div_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, struct mbuf *control, struct thread *td) { /* Packet must have a header (but that's about it) */ if (m->m_len < sizeof (struct ip) && (m = m_pullup(m, sizeof (struct ip))) == 0) { KMOD_IPSTAT_INC(ips_toosmall); m_freem(m); return EINVAL; } /* Send packet */ return div_output(so, m, (struct sockaddr_in *)nam, control); } static void div_ctlinput(int cmd, struct sockaddr *sa, void *vip) { struct in_addr faddr; faddr = ((struct sockaddr_in *)sa)->sin_addr; if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY) return; if (PRC_IS_REDIRECT(cmd)) return; } static int div_pcblist(SYSCTL_HANDLER_ARGS) { int error, i, n; struct inpcb *inp, **inp_list; inp_gen_t gencnt; struct xinpgen xig; /* * The process of preparing the TCB list is too time-consuming and * resource-intensive to repeat twice on every request. */ if (req->oldptr == 0) { n = V_divcbinfo.ipi_count; n += imax(n / 8, 10); req->oldidx = 2 * (sizeof xig) + n * sizeof(struct xinpcb); return 0; } if (req->newptr != 0) return EPERM; /* * OK, now we're committed to doing something. */ INP_INFO_RLOCK(&V_divcbinfo); gencnt = V_divcbinfo.ipi_gencnt; n = V_divcbinfo.ipi_count; INP_INFO_RUNLOCK(&V_divcbinfo); error = sysctl_wire_old_buffer(req, 2 * sizeof(xig) + n*sizeof(struct xinpcb)); if (error != 0) return (error); xig.xig_len = sizeof xig; xig.xig_count = n; xig.xig_gen = gencnt; xig.xig_sogen = so_gencnt; error = SYSCTL_OUT(req, &xig, sizeof xig); if (error) return error; inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); if (inp_list == 0) return ENOMEM; INP_INFO_RLOCK(&V_divcbinfo); for (inp = LIST_FIRST(V_divcbinfo.ipi_listhead), i = 0; inp && i < n; inp = LIST_NEXT(inp, inp_list)) { INP_WLOCK(inp); if (inp->inp_gencnt <= gencnt && cr_canseeinpcb(req->td->td_ucred, inp) == 0) { in_pcbref(inp); inp_list[i++] = inp; } INP_WUNLOCK(inp); } INP_INFO_RUNLOCK(&V_divcbinfo); n = i; error = 0; for (i = 0; i < n; i++) { inp = inp_list[i]; INP_RLOCK(inp); if (inp->inp_gencnt <= gencnt) { struct xinpcb xi; bzero(&xi, sizeof(xi)); xi.xi_len = sizeof xi; /* XXX should avoid extra copy */ bcopy(inp, &xi.xi_inp, sizeof *inp); if (inp->inp_socket) sotoxsocket(inp->inp_socket, &xi.xi_socket); INP_RUNLOCK(inp); error = SYSCTL_OUT(req, &xi, sizeof xi); } else INP_RUNLOCK(inp); } INP_INFO_WLOCK(&V_divcbinfo); for (i = 0; i < n; i++) { inp = inp_list[i]; INP_RLOCK(inp); if (!in_pcbrele_rlocked(inp)) INP_RUNLOCK(inp); } INP_INFO_WUNLOCK(&V_divcbinfo); if (!error) { /* * Give the user an updated idea of our state. * If the generation differs from what we told * her before, she knows that something happened * while we were processing this request, and it * might be necessary to retry. */ INP_INFO_RLOCK(&V_divcbinfo); xig.xig_gen = V_divcbinfo.ipi_gencnt; xig.xig_sogen = so_gencnt; xig.xig_count = V_divcbinfo.ipi_count; INP_INFO_RUNLOCK(&V_divcbinfo); error = SYSCTL_OUT(req, &xig, sizeof xig); } free(inp_list, M_TEMP); return error; } #ifdef SYSCTL_NODE SYSCTL_NODE(_net_inet, IPPROTO_DIVERT, divert, CTLFLAG_RW, 0, "IPDIVERT"); SYSCTL_PROC(_net_inet_divert, OID_AUTO, pcblist, CTLTYPE_OPAQUE | CTLFLAG_RD, NULL, 0, div_pcblist, "S,xinpcb", "List of active divert sockets"); #endif struct pr_usrreqs div_usrreqs = { .pru_attach = div_attach, .pru_bind = div_bind, .pru_control = in_control, .pru_detach = div_detach, .pru_peeraddr = in_getpeeraddr, .pru_send = div_send, .pru_shutdown = div_shutdown, .pru_sockaddr = in_getsockaddr, .pru_sosetlabel = in_pcbsosetlabel }; struct protosw div_protosw = { .pr_type = SOCK_RAW, .pr_protocol = IPPROTO_DIVERT, .pr_flags = PR_ATOMIC|PR_ADDR, .pr_input = div_input, .pr_ctlinput = div_ctlinput, .pr_ctloutput = ip_ctloutput, .pr_init = div_init, #ifdef VIMAGE .pr_destroy = div_destroy, #endif .pr_usrreqs = &div_usrreqs }; static int div_modevent(module_t mod, int type, void *unused) { int err = 0; #ifndef VIMAGE int n; #endif switch (type) { case MOD_LOAD: /* * Protocol will be initialized by pf_proto_register(). * We don't have to register ip_protox because we are not * a true IP protocol that goes over the wire. */ err = pf_proto_register(PF_INET, &div_protosw); if (err != 0) return (err); ip_divert_ptr = divert_packet; ip_divert_event_tag = EVENTHANDLER_REGISTER(maxsockets_change, div_zone_change, NULL, EVENTHANDLER_PRI_ANY); break; case MOD_QUIESCE: /* * IPDIVERT may normally not be unloaded because of the * potential race conditions. Tell kldunload we can't be * unloaded unless the unload is forced. */ err = EPERM; break; case MOD_UNLOAD: #ifdef VIMAGE err = EPERM; break; #else /* * Forced unload. * * Module ipdivert can only be unloaded if no sockets are * connected. Maybe this can be changed later to forcefully * disconnect any open sockets. * * XXXRW: Note that there is a slight race here, as a new * socket open request could be spinning on the lock and then * we destroy the lock. */ INP_INFO_WLOCK(&V_divcbinfo); n = V_divcbinfo.ipi_count; if (n != 0) { err = EBUSY; INP_INFO_WUNLOCK(&V_divcbinfo); break; } ip_divert_ptr = NULL; err = pf_proto_unregister(PF_INET, IPPROTO_DIVERT, SOCK_RAW); INP_INFO_WUNLOCK(&V_divcbinfo); div_destroy(); EVENTHANDLER_DEREGISTER(maxsockets_change, ip_divert_event_tag); break; #endif /* !VIMAGE */ default: err = EOPNOTSUPP; break; } return err; } static moduledata_t ipdivertmod = { "ipdivert", div_modevent, 0 }; DECLARE_MODULE(ipdivert, ipdivertmod, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY); MODULE_DEPEND(ipdivert, ipfw, 2, 2, 2); MODULE_VERSION(ipdivert, 1); Index: projects/largeSMP/sys/netinet/ipfw/ip_fw2.c =================================================================== --- projects/largeSMP/sys/netinet/ipfw/ip_fw2.c (revision 222782) +++ projects/largeSMP/sys/netinet/ipfw/ip_fw2.c (revision 222783) @@ -1,2559 +1,2563 @@ /*- * Copyright (c) 2002-2009 Luigi Rizzo, Universita` di Pisa * * 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("$FreeBSD$"); /* * The FreeBSD IP packet firewall, main file */ #if !defined(KLD_MODULE) #include "opt_ipfw.h" #include "opt_ipdivert.h" #include "opt_ipdn.h" #include "opt_inet.h" #ifndef INET #error IPFIREWALL requires INET. #endif /* INET */ #endif #include "opt_inet6.h" #include "opt_ipsec.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* for ETHERTYPE_IP */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef INET6 #include #include #endif #include /* XXX for in_cksum */ #ifdef MAC #include #endif /* * static variables followed by global ones. * All ipfw global variables are here. */ /* ipfw_vnet_ready controls when we are open for business */ static VNET_DEFINE(int, ipfw_vnet_ready) = 0; #define V_ipfw_vnet_ready VNET(ipfw_vnet_ready) static VNET_DEFINE(int, fw_deny_unknown_exthdrs); #define V_fw_deny_unknown_exthdrs VNET(fw_deny_unknown_exthdrs) #ifdef IPFIREWALL_DEFAULT_TO_ACCEPT static int default_to_accept = 1; #else static int default_to_accept; #endif VNET_DEFINE(int, autoinc_step); VNET_DEFINE(int, fw_one_pass) = 1; /* * Each rule belongs to one of 32 different sets (0..31). * The variable set_disable contains one bit per set. * If the bit is set, all rules in the corresponding set * are disabled. Set RESVD_SET(31) is reserved for the default rule * and rules that are not deleted by the flush command, * and CANNOT be disabled. * Rules in set RESVD_SET can only be deleted individually. */ VNET_DEFINE(u_int32_t, set_disable); #define V_set_disable VNET(set_disable) VNET_DEFINE(int, fw_verbose); /* counter for ipfw_log(NULL...) */ VNET_DEFINE(u_int64_t, norule_counter); VNET_DEFINE(int, verbose_limit); /* layer3_chain contains the list of rules for layer 3 */ VNET_DEFINE(struct ip_fw_chain, layer3_chain); ipfw_nat_t *ipfw_nat_ptr = NULL; struct cfg_nat *(*lookup_nat_ptr)(struct nat_list *, int); ipfw_nat_cfg_t *ipfw_nat_cfg_ptr; ipfw_nat_cfg_t *ipfw_nat_del_ptr; ipfw_nat_cfg_t *ipfw_nat_get_cfg_ptr; ipfw_nat_cfg_t *ipfw_nat_get_log_ptr; #ifdef SYSCTL_NODE uint32_t dummy_def = IPFW_DEFAULT_RULE; uint32_t dummy_tables_max = IPFW_TABLES_MAX; SYSBEGIN(f3) SYSCTL_NODE(_net_inet_ip, OID_AUTO, fw, CTLFLAG_RW, 0, "Firewall"); SYSCTL_VNET_INT(_net_inet_ip_fw, OID_AUTO, one_pass, CTLFLAG_RW | CTLFLAG_SECURE3, &VNET_NAME(fw_one_pass), 0, "Only do a single pass through ipfw when using dummynet(4)"); SYSCTL_VNET_INT(_net_inet_ip_fw, OID_AUTO, autoinc_step, CTLFLAG_RW, &VNET_NAME(autoinc_step), 0, "Rule number auto-increment step"); SYSCTL_VNET_INT(_net_inet_ip_fw, OID_AUTO, verbose, CTLFLAG_RW | CTLFLAG_SECURE3, &VNET_NAME(fw_verbose), 0, "Log matches to ipfw rules"); SYSCTL_VNET_INT(_net_inet_ip_fw, OID_AUTO, verbose_limit, CTLFLAG_RW, &VNET_NAME(verbose_limit), 0, "Set upper limit of matches of ipfw rules logged"); SYSCTL_UINT(_net_inet_ip_fw, OID_AUTO, default_rule, CTLFLAG_RD, &dummy_def, 0, "The default/max possible rule number."); SYSCTL_UINT(_net_inet_ip_fw, OID_AUTO, tables_max, CTLFLAG_RD, &dummy_tables_max, 0, "The maximum number of tables."); SYSCTL_INT(_net_inet_ip_fw, OID_AUTO, default_to_accept, CTLFLAG_RDTUN, &default_to_accept, 0, "Make the default rule accept all packets."); TUNABLE_INT("net.inet.ip.fw.default_to_accept", &default_to_accept); SYSCTL_VNET_INT(_net_inet_ip_fw, OID_AUTO, static_count, CTLFLAG_RD, &VNET_NAME(layer3_chain.n_rules), 0, "Number of static rules"); #ifdef INET6 SYSCTL_DECL(_net_inet6_ip6); SYSCTL_NODE(_net_inet6_ip6, OID_AUTO, fw, CTLFLAG_RW, 0, "Firewall"); SYSCTL_VNET_INT(_net_inet6_ip6_fw, OID_AUTO, deny_unknown_exthdrs, CTLFLAG_RW | CTLFLAG_SECURE, &VNET_NAME(fw_deny_unknown_exthdrs), 0, "Deny packets with unknown IPv6 Extension Headers"); #endif /* INET6 */ SYSEND #endif /* SYSCTL_NODE */ /* * Some macros used in the various matching options. * L3HDR maps an ipv4 pointer into a layer3 header pointer of type T * Other macros just cast void * into the appropriate type */ #define L3HDR(T, ip) ((T *)((u_int32_t *)(ip) + (ip)->ip_hl)) #define TCP(p) ((struct tcphdr *)(p)) #define SCTP(p) ((struct sctphdr *)(p)) #define UDP(p) ((struct udphdr *)(p)) #define ICMP(p) ((struct icmphdr *)(p)) #define ICMP6(p) ((struct icmp6_hdr *)(p)) static __inline int icmptype_match(struct icmphdr *icmp, ipfw_insn_u32 *cmd) { int type = icmp->icmp_type; return (type <= ICMP_MAXTYPE && (cmd->d[0] & (1<icmp_type; return (type <= ICMP_MAXTYPE && (TT & (1<arg1 or cmd->d[0]. * * We scan options and store the bits we find set. We succeed if * * (want_set & ~bits) == 0 && (want_clear & ~bits) == want_clear * * The code is sometimes optimized not to store additional variables. */ static int flags_match(ipfw_insn *cmd, u_int8_t bits) { u_char want_clear; bits = ~bits; if ( ((cmd->arg1 & 0xff) & bits) != 0) return 0; /* some bits we want set were clear */ want_clear = (cmd->arg1 >> 8) & 0xff; if ( (want_clear & bits) != want_clear) return 0; /* some bits we want clear were set */ return 1; } static int ipopts_match(struct ip *ip, ipfw_insn *cmd) { int optlen, bits = 0; u_char *cp = (u_char *)(ip + 1); int x = (ip->ip_hl << 2) - sizeof (struct ip); for (; x > 0; x -= optlen, cp += optlen) { int opt = cp[IPOPT_OPTVAL]; if (opt == IPOPT_EOL) break; if (opt == IPOPT_NOP) optlen = 1; else { optlen = cp[IPOPT_OLEN]; if (optlen <= 0 || optlen > x) return 0; /* invalid or truncated */ } switch (opt) { default: break; case IPOPT_LSRR: bits |= IP_FW_IPOPT_LSRR; break; case IPOPT_SSRR: bits |= IP_FW_IPOPT_SSRR; break; case IPOPT_RR: bits |= IP_FW_IPOPT_RR; break; case IPOPT_TS: bits |= IP_FW_IPOPT_TS; break; } } return (flags_match(cmd, bits)); } static int tcpopts_match(struct tcphdr *tcp, ipfw_insn *cmd) { int optlen, bits = 0; u_char *cp = (u_char *)(tcp + 1); int x = (tcp->th_off << 2) - sizeof(struct tcphdr); for (; x > 0; x -= optlen, cp += optlen) { int opt = cp[0]; if (opt == TCPOPT_EOL) break; if (opt == TCPOPT_NOP) optlen = 1; else { optlen = cp[1]; if (optlen <= 0) break; } switch (opt) { default: break; case TCPOPT_MAXSEG: bits |= IP_FW_TCPOPT_MSS; break; case TCPOPT_WINDOW: bits |= IP_FW_TCPOPT_WINDOW; break; case TCPOPT_SACK_PERMITTED: case TCPOPT_SACK: bits |= IP_FW_TCPOPT_SACK; break; case TCPOPT_TIMESTAMP: bits |= IP_FW_TCPOPT_TS; break; } } return (flags_match(cmd, bits)); } static int iface_match(struct ifnet *ifp, ipfw_insn_if *cmd) { if (ifp == NULL) /* no iface with this packet, match fails */ return 0; /* Check by name or by IP address */ if (cmd->name[0] != '\0') { /* match by name */ /* Check name */ if (cmd->p.glob) { if (fnmatch(cmd->name, ifp->if_xname, 0) == 0) return(1); } else { if (strncmp(ifp->if_xname, cmd->name, IFNAMSIZ) == 0) return(1); } } else { #ifdef __FreeBSD__ /* and OSX too ? */ struct ifaddr *ia; if_addr_rlock(ifp); TAILQ_FOREACH(ia, &ifp->if_addrhead, ifa_link) { if (ia->ifa_addr->sa_family != AF_INET) continue; if (cmd->p.ip.s_addr == ((struct sockaddr_in *) (ia->ifa_addr))->sin_addr.s_addr) { if_addr_runlock(ifp); return(1); /* match */ } } if_addr_runlock(ifp); #endif /* __FreeBSD__ */ } return(0); /* no match, fail ... */ } /* * The verify_path function checks if a route to the src exists and * if it is reachable via ifp (when provided). * * The 'verrevpath' option checks that the interface that an IP packet * arrives on is the same interface that traffic destined for the * packet's source address would be routed out of. * The 'versrcreach' option just checks that the source address is * reachable via any route (except default) in the routing table. * These two are a measure to block forged packets. This is also * commonly known as "anti-spoofing" or Unicast Reverse Path * Forwarding (Unicast RFP) in Cisco-ese. The name of the knobs * is purposely reminiscent of the Cisco IOS command, * * ip verify unicast reverse-path * ip verify unicast source reachable-via any * * which implements the same functionality. But note that the syntax * is misleading, and the check may be performed on all IP packets * whether unicast, multicast, or broadcast. */ static int verify_path(struct in_addr src, struct ifnet *ifp, u_int fib) { #ifndef __FreeBSD__ return 0; #else struct route ro; struct sockaddr_in *dst; bzero(&ro, sizeof(ro)); dst = (struct sockaddr_in *)&(ro.ro_dst); dst->sin_family = AF_INET; dst->sin_len = sizeof(*dst); dst->sin_addr = src; in_rtalloc_ign(&ro, 0, fib); if (ro.ro_rt == NULL) return 0; /* * If ifp is provided, check for equality with rtentry. * We should use rt->rt_ifa->ifa_ifp, instead of rt->rt_ifp, * in order to pass packets injected back by if_simloop(): * if useloopback == 1 routing entry (via lo0) for our own address * may exist, so we need to handle routing assymetry. */ if (ifp != NULL && ro.ro_rt->rt_ifa->ifa_ifp != ifp) { RTFREE(ro.ro_rt); return 0; } /* if no ifp provided, check if rtentry is not default route */ if (ifp == NULL && satosin(rt_key(ro.ro_rt))->sin_addr.s_addr == INADDR_ANY) { RTFREE(ro.ro_rt); return 0; } /* or if this is a blackhole/reject route */ if (ifp == NULL && ro.ro_rt->rt_flags & (RTF_REJECT|RTF_BLACKHOLE)) { RTFREE(ro.ro_rt); return 0; } /* found valid route */ RTFREE(ro.ro_rt); return 1; #endif /* __FreeBSD__ */ } #ifdef INET6 /* * ipv6 specific rules here... */ static __inline int icmp6type_match (int type, ipfw_insn_u32 *cmd) { return (type <= ICMP6_MAXTYPE && (cmd->d[type/32] & (1<<(type%32)) ) ); } static int flow6id_match( int curr_flow, ipfw_insn_u32 *cmd ) { int i; for (i=0; i <= cmd->o.arg1; ++i ) if (curr_flow == cmd->d[i] ) return 1; return 0; } /* support for IP6_*_ME opcodes */ static int search_ip6_addr_net (struct in6_addr * ip6_addr) { struct ifnet *mdc; struct ifaddr *mdc2; struct in6_ifaddr *fdm; struct in6_addr copia; TAILQ_FOREACH(mdc, &V_ifnet, if_link) { if_addr_rlock(mdc); TAILQ_FOREACH(mdc2, &mdc->if_addrhead, ifa_link) { if (mdc2->ifa_addr->sa_family == AF_INET6) { fdm = (struct in6_ifaddr *)mdc2; copia = fdm->ia_addr.sin6_addr; /* need for leaving scope_id in the sock_addr */ in6_clearscope(&copia); if (IN6_ARE_ADDR_EQUAL(ip6_addr, &copia)) { if_addr_runlock(mdc); return 1; } } } if_addr_runlock(mdc); } return 0; } static int verify_path6(struct in6_addr *src, struct ifnet *ifp) { struct route_in6 ro; struct sockaddr_in6 *dst; bzero(&ro, sizeof(ro)); dst = (struct sockaddr_in6 * )&(ro.ro_dst); dst->sin6_family = AF_INET6; dst->sin6_len = sizeof(*dst); dst->sin6_addr = *src; /* XXX MRT 0 for ipv6 at this time */ rtalloc_ign((struct route *)&ro, 0); if (ro.ro_rt == NULL) return 0; /* * if ifp is provided, check for equality with rtentry * We should use rt->rt_ifa->ifa_ifp, instead of rt->rt_ifp, * to support the case of sending packets to an address of our own. * (where the former interface is the first argument of if_simloop() * (=ifp), the latter is lo0) */ if (ifp != NULL && ro.ro_rt->rt_ifa->ifa_ifp != ifp) { RTFREE(ro.ro_rt); return 0; } /* if no ifp provided, check if rtentry is not default route */ if (ifp == NULL && IN6_IS_ADDR_UNSPECIFIED(&satosin6(rt_key(ro.ro_rt))->sin6_addr)) { RTFREE(ro.ro_rt); return 0; } /* or if this is a blackhole/reject route */ if (ifp == NULL && ro.ro_rt->rt_flags & (RTF_REJECT|RTF_BLACKHOLE)) { RTFREE(ro.ro_rt); return 0; } /* found valid route */ RTFREE(ro.ro_rt); return 1; } static int is_icmp6_query(int icmp6_type) { if ((icmp6_type <= ICMP6_MAXTYPE) && (icmp6_type == ICMP6_ECHO_REQUEST || icmp6_type == ICMP6_MEMBERSHIP_QUERY || icmp6_type == ICMP6_WRUREQUEST || icmp6_type == ICMP6_FQDN_QUERY || icmp6_type == ICMP6_NI_QUERY)) return (1); return (0); } static void send_reject6(struct ip_fw_args *args, int code, u_int hlen, struct ip6_hdr *ip6) { struct mbuf *m; m = args->m; if (code == ICMP6_UNREACH_RST && args->f_id.proto == IPPROTO_TCP) { struct tcphdr *tcp; tcp = (struct tcphdr *)((char *)ip6 + hlen); if ((tcp->th_flags & TH_RST) == 0) { struct mbuf *m0; m0 = ipfw_send_pkt(args->m, &(args->f_id), ntohl(tcp->th_seq), ntohl(tcp->th_ack), tcp->th_flags | TH_RST); if (m0 != NULL) ip6_output(m0, NULL, NULL, 0, NULL, NULL, NULL); } FREE_PKT(m); } else if (code != ICMP6_UNREACH_RST) { /* Send an ICMPv6 unreach. */ #if 0 /* * Unlike above, the mbufs need to line up with the ip6 hdr, * as the contents are read. We need to m_adj() the * needed amount. * The mbuf will however be thrown away so we can adjust it. * Remember we did an m_pullup on it already so we * can make some assumptions about contiguousness. */ if (args->L3offset) m_adj(m, args->L3offset); #endif icmp6_error(m, ICMP6_DST_UNREACH, code, 0); } else FREE_PKT(m); args->m = NULL; } #endif /* INET6 */ /* * sends a reject message, consuming the mbuf passed as an argument. */ static void send_reject(struct ip_fw_args *args, int code, int iplen, struct ip *ip) { #if 0 /* XXX When ip is not guaranteed to be at mtod() we will * need to account for this */ * The mbuf will however be thrown away so we can adjust it. * Remember we did an m_pullup on it already so we * can make some assumptions about contiguousness. */ if (args->L3offset) m_adj(m, args->L3offset); #endif if (code != ICMP_REJECT_RST) { /* Send an ICMP unreach */ /* We need the IP header in host order for icmp_error(). */ SET_HOST_IPLEN(ip); icmp_error(args->m, ICMP_UNREACH, code, 0L, 0); } else if (args->f_id.proto == IPPROTO_TCP) { struct tcphdr *const tcp = L3HDR(struct tcphdr, mtod(args->m, struct ip *)); if ( (tcp->th_flags & TH_RST) == 0) { struct mbuf *m; m = ipfw_send_pkt(args->m, &(args->f_id), ntohl(tcp->th_seq), ntohl(tcp->th_ack), tcp->th_flags | TH_RST); if (m != NULL) ip_output(m, NULL, NULL, 0, NULL, NULL); } FREE_PKT(args->m); } else FREE_PKT(args->m); args->m = NULL; } /* * Support for uid/gid/jail lookup. These tests are expensive * (because we may need to look into the list of active sockets) * so we cache the results. ugid_lookupp is 0 if we have not * yet done a lookup, 1 if we succeeded, and -1 if we tried * and failed. The function always returns the match value. * We could actually spare the variable and use *uc, setting * it to '(void *)check_uidgid if we have no info, NULL if * we tried and failed, or any other value if successful. */ static int check_uidgid(ipfw_insn_u32 *insn, int proto, struct ifnet *oif, struct in_addr dst_ip, u_int16_t dst_port, struct in_addr src_ip, u_int16_t src_port, int *ugid_lookupp, struct ucred **uc, struct inpcb *inp) { #ifndef __FreeBSD__ return cred_check(insn, proto, oif, dst_ip, dst_port, src_ip, src_port, (struct bsd_ucred *)uc, ugid_lookupp, ((struct mbuf *)inp)->m_skb); #else /* FreeBSD */ struct inpcbinfo *pi; int lookupflags; struct inpcb *pcb; int match; /* * Check to see if the UDP or TCP stack supplied us with * the PCB. If so, rather then holding a lock and looking * up the PCB, we can use the one that was supplied. */ if (inp && *ugid_lookupp == 0) { INP_LOCK_ASSERT(inp); if (inp->inp_socket != NULL) { *uc = crhold(inp->inp_cred); *ugid_lookupp = 1; } else *ugid_lookupp = -1; } /* * If we have already been here and the packet has no * PCB entry associated with it, then we can safely * assume that this is a no match. */ if (*ugid_lookupp == -1) return (0); if (proto == IPPROTO_TCP) { lookupflags = 0; pi = &V_tcbinfo; } else if (proto == IPPROTO_UDP) { lookupflags = INPLOOKUP_WILDCARD; pi = &V_udbinfo; } else return 0; lookupflags |= INPLOOKUP_RLOCKPCB; match = 0; if (*ugid_lookupp == 0) { + /* + * XXXRW: If we had the mbuf here, could use + * in_pcblookup_mbuf(). + */ pcb = (oif) ? in_pcblookup(pi, dst_ip, htons(dst_port), src_ip, htons(src_port), lookupflags, oif) : in_pcblookup(pi, src_ip, htons(src_port), dst_ip, htons(dst_port), lookupflags, NULL); if (pcb != NULL) { INP_RLOCK_ASSERT(pcb); *uc = crhold(pcb->inp_cred); *ugid_lookupp = 1; INP_RUNLOCK(pcb); } if (*ugid_lookupp == 0) { /* * We tried and failed, set the variable to -1 * so we will not try again on this packet. */ *ugid_lookupp = -1; return (0); } } if (insn->o.opcode == O_UID) match = ((*uc)->cr_uid == (uid_t)insn->d[0]); else if (insn->o.opcode == O_GID) match = groupmember((gid_t)insn->d[0], *uc); else if (insn->o.opcode == O_JAIL) match = ((*uc)->cr_prison->pr_id == (int)insn->d[0]); return match; #endif /* __FreeBSD__ */ } /* * Helper function to set args with info on the rule after the matching * one. slot is precise, whereas we guess rule_id as they are * assigned sequentially. */ static inline void set_match(struct ip_fw_args *args, int slot, struct ip_fw_chain *chain) { args->rule.chain_id = chain->id; args->rule.slot = slot + 1; /* we use 0 as a marker */ args->rule.rule_id = 1 + chain->map[slot]->id; args->rule.rulenum = chain->map[slot]->rulenum; } /* * The main check routine for the firewall. * * All arguments are in args so we can modify them and return them * back to the caller. * * Parameters: * * args->m (in/out) The packet; we set to NULL when/if we nuke it. * Starts with the IP header. * args->eh (in) Mac header if present, NULL for layer3 packet. * args->L3offset Number of bytes bypassed if we came from L2. * e.g. often sizeof(eh) ** NOTYET ** * args->oif Outgoing interface, NULL if packet is incoming. * The incoming interface is in the mbuf. (in) * args->divert_rule (in/out) * Skip up to the first rule past this rule number; * upon return, non-zero port number for divert or tee. * * args->rule Pointer to the last matching rule (in/out) * args->next_hop Socket we are forwarding to (out). * args->f_id Addresses grabbed from the packet (out) * args->rule.info a cookie depending on rule action * * Return value: * * IP_FW_PASS the packet must be accepted * IP_FW_DENY the packet must be dropped * IP_FW_DIVERT divert packet, port in m_tag * IP_FW_TEE tee packet, port in m_tag * IP_FW_DUMMYNET to dummynet, pipe in args->cookie * IP_FW_NETGRAPH into netgraph, cookie args->cookie * args->rule contains the matching rule, * args->rule.info has additional information. * */ int ipfw_chk(struct ip_fw_args *args) { /* * Local variables holding state while processing a packet: * * IMPORTANT NOTE: to speed up the processing of rules, there * are some assumption on the values of the variables, which * are documented here. Should you change them, please check * the implementation of the various instructions to make sure * that they still work. * * args->eh The MAC header. It is non-null for a layer2 * packet, it is NULL for a layer-3 packet. * **notyet** * args->L3offset Offset in the packet to the L3 (IP or equiv.) header. * * m | args->m Pointer to the mbuf, as received from the caller. * It may change if ipfw_chk() does an m_pullup, or if it * consumes the packet because it calls send_reject(). * XXX This has to change, so that ipfw_chk() never modifies * or consumes the buffer. * ip is the beginning of the ip(4 or 6) header. * Calculated by adding the L3offset to the start of data. * (Until we start using L3offset, the packet is * supposed to start with the ip header). */ struct mbuf *m = args->m; struct ip *ip = mtod(m, struct ip *); /* * For rules which contain uid/gid or jail constraints, cache * a copy of the users credentials after the pcb lookup has been * executed. This will speed up the processing of rules with * these types of constraints, as well as decrease contention * on pcb related locks. */ #ifndef __FreeBSD__ struct bsd_ucred ucred_cache; #else struct ucred *ucred_cache = NULL; #endif int ucred_lookup = 0; /* * oif | args->oif If NULL, ipfw_chk has been called on the * inbound path (ether_input, ip_input). * If non-NULL, ipfw_chk has been called on the outbound path * (ether_output, ip_output). */ struct ifnet *oif = args->oif; int f_pos = 0; /* index of current rule in the array */ int retval = 0; /* * hlen The length of the IP header. */ u_int hlen = 0; /* hlen >0 means we have an IP pkt */ /* * offset The offset of a fragment. offset != 0 means that * we have a fragment at this offset of an IPv4 packet. * offset == 0 means that (if this is an IPv4 packet) * this is the first or only fragment. * For IPv6 offset == 0 means there is no Fragment Header. * If offset != 0 for IPv6 always use correct mask to * get the correct offset because we add IP6F_MORE_FRAG * to be able to dectect the first fragment which would * otherwise have offset = 0. */ u_short offset = 0; /* * Local copies of addresses. They are only valid if we have * an IP packet. * * proto The protocol. Set to 0 for non-ip packets, * or to the protocol read from the packet otherwise. * proto != 0 means that we have an IPv4 packet. * * src_port, dst_port port numbers, in HOST format. Only * valid for TCP and UDP packets. * * src_ip, dst_ip ip addresses, in NETWORK format. * Only valid for IPv4 packets. */ uint8_t proto; uint16_t src_port = 0, dst_port = 0; /* NOTE: host format */ struct in_addr src_ip, dst_ip; /* NOTE: network format */ uint16_t iplen=0; int pktlen; uint16_t etype = 0; /* Host order stored ether type */ /* * dyn_dir = MATCH_UNKNOWN when rules unchecked, * MATCH_NONE when checked and not matched (q = NULL), * MATCH_FORWARD or MATCH_REVERSE otherwise (q != NULL) */ int dyn_dir = MATCH_UNKNOWN; ipfw_dyn_rule *q = NULL; struct ip_fw_chain *chain = &V_layer3_chain; /* * We store in ulp a pointer to the upper layer protocol header. * In the ipv4 case this is easy to determine from the header, * but for ipv6 we might have some additional headers in the middle. * ulp is NULL if not found. */ void *ulp = NULL; /* upper layer protocol pointer. */ /* XXX ipv6 variables */ int is_ipv6 = 0; uint8_t icmp6_type = 0; uint16_t ext_hd = 0; /* bits vector for extension header filtering */ /* end of ipv6 variables */ int is_ipv4 = 0; int done = 0; /* flag to exit the outer loop */ if (m->m_flags & M_SKIP_FIREWALL || (! V_ipfw_vnet_ready)) return (IP_FW_PASS); /* accept */ dst_ip.s_addr = 0; /* make sure it is initialized */ src_ip.s_addr = 0; /* make sure it is initialized */ pktlen = m->m_pkthdr.len; args->f_id.fib = M_GETFIB(m); /* note mbuf not altered) */ proto = args->f_id.proto = 0; /* mark f_id invalid */ /* XXX 0 is a valid proto: IP/IPv6 Hop-by-Hop Option */ /* * PULLUP_TO(len, p, T) makes sure that len + sizeof(T) is contiguous, * then it sets p to point at the offset "len" in the mbuf. WARNING: the * pointer might become stale after other pullups (but we never use it * this way). */ #define PULLUP_TO(_len, p, T) PULLUP_LEN(_len, p, sizeof(T)) #define PULLUP_LEN(_len, p, T) \ do { \ int x = (_len) + T; \ if ((m)->m_len < x) { \ args->m = m = m_pullup(m, x); \ if (m == NULL) \ goto pullup_failed; \ } \ p = (mtod(m, char *) + (_len)); \ } while (0) /* * if we have an ether header, */ if (args->eh) etype = ntohs(args->eh->ether_type); /* Identify IP packets and fill up variables. */ if (pktlen >= sizeof(struct ip6_hdr) && (args->eh == NULL || etype == ETHERTYPE_IPV6) && ip->ip_v == 6) { struct ip6_hdr *ip6 = (struct ip6_hdr *)ip; is_ipv6 = 1; args->f_id.addr_type = 6; hlen = sizeof(struct ip6_hdr); proto = ip6->ip6_nxt; /* Search extension headers to find upper layer protocols */ while (ulp == NULL) { switch (proto) { case IPPROTO_ICMPV6: PULLUP_TO(hlen, ulp, struct icmp6_hdr); icmp6_type = ICMP6(ulp)->icmp6_type; break; case IPPROTO_TCP: PULLUP_TO(hlen, ulp, struct tcphdr); dst_port = TCP(ulp)->th_dport; src_port = TCP(ulp)->th_sport; /* save flags for dynamic rules */ args->f_id._flags = TCP(ulp)->th_flags; break; case IPPROTO_SCTP: PULLUP_TO(hlen, ulp, struct sctphdr); src_port = SCTP(ulp)->src_port; dst_port = SCTP(ulp)->dest_port; break; case IPPROTO_UDP: PULLUP_TO(hlen, ulp, struct udphdr); dst_port = UDP(ulp)->uh_dport; src_port = UDP(ulp)->uh_sport; break; case IPPROTO_HOPOPTS: /* RFC 2460 */ PULLUP_TO(hlen, ulp, struct ip6_hbh); ext_hd |= EXT_HOPOPTS; hlen += (((struct ip6_hbh *)ulp)->ip6h_len + 1) << 3; proto = ((struct ip6_hbh *)ulp)->ip6h_nxt; ulp = NULL; break; case IPPROTO_ROUTING: /* RFC 2460 */ PULLUP_TO(hlen, ulp, struct ip6_rthdr); switch (((struct ip6_rthdr *)ulp)->ip6r_type) { case 0: ext_hd |= EXT_RTHDR0; break; case 2: ext_hd |= EXT_RTHDR2; break; default: printf("IPFW2: IPV6 - Unknown Routing " "Header type(%d)\n", ((struct ip6_rthdr *)ulp)->ip6r_type); if (V_fw_deny_unknown_exthdrs) return (IP_FW_DENY); break; } ext_hd |= EXT_ROUTING; hlen += (((struct ip6_rthdr *)ulp)->ip6r_len + 1) << 3; proto = ((struct ip6_rthdr *)ulp)->ip6r_nxt; ulp = NULL; break; case IPPROTO_FRAGMENT: /* RFC 2460 */ PULLUP_TO(hlen, ulp, struct ip6_frag); ext_hd |= EXT_FRAGMENT; hlen += sizeof (struct ip6_frag); proto = ((struct ip6_frag *)ulp)->ip6f_nxt; offset = ((struct ip6_frag *)ulp)->ip6f_offlg & IP6F_OFF_MASK; /* Add IP6F_MORE_FRAG for offset of first * fragment to be != 0. */ offset |= ((struct ip6_frag *)ulp)->ip6f_offlg & IP6F_MORE_FRAG; if (offset == 0) { printf("IPFW2: IPV6 - Invalid Fragment " "Header\n"); if (V_fw_deny_unknown_exthdrs) return (IP_FW_DENY); break; } args->f_id.extra = ntohl(((struct ip6_frag *)ulp)->ip6f_ident); ulp = NULL; break; case IPPROTO_DSTOPTS: /* RFC 2460 */ PULLUP_TO(hlen, ulp, struct ip6_hbh); ext_hd |= EXT_DSTOPTS; hlen += (((struct ip6_hbh *)ulp)->ip6h_len + 1) << 3; proto = ((struct ip6_hbh *)ulp)->ip6h_nxt; ulp = NULL; break; case IPPROTO_AH: /* RFC 2402 */ PULLUP_TO(hlen, ulp, struct ip6_ext); ext_hd |= EXT_AH; hlen += (((struct ip6_ext *)ulp)->ip6e_len + 2) << 2; proto = ((struct ip6_ext *)ulp)->ip6e_nxt; ulp = NULL; break; case IPPROTO_ESP: /* RFC 2406 */ PULLUP_TO(hlen, ulp, uint32_t); /* SPI, Seq# */ /* Anything past Seq# is variable length and * data past this ext. header is encrypted. */ ext_hd |= EXT_ESP; break; case IPPROTO_NONE: /* RFC 2460 */ /* * Packet ends here, and IPv6 header has * already been pulled up. If ip6e_len!=0 * then octets must be ignored. */ ulp = ip; /* non-NULL to get out of loop. */ break; case IPPROTO_OSPFIGP: /* XXX OSPF header check? */ PULLUP_TO(hlen, ulp, struct ip6_ext); break; case IPPROTO_PIM: /* XXX PIM header check? */ PULLUP_TO(hlen, ulp, struct pim); break; case IPPROTO_CARP: PULLUP_TO(hlen, ulp, struct carp_header); if (((struct carp_header *)ulp)->carp_version != CARP_VERSION) return (IP_FW_DENY); if (((struct carp_header *)ulp)->carp_type != CARP_ADVERTISEMENT) return (IP_FW_DENY); break; case IPPROTO_IPV6: /* RFC 2893 */ PULLUP_TO(hlen, ulp, struct ip6_hdr); break; case IPPROTO_IPV4: /* RFC 2893 */ PULLUP_TO(hlen, ulp, struct ip); break; default: printf("IPFW2: IPV6 - Unknown Extension " "Header(%d), ext_hd=%x\n", proto, ext_hd); if (V_fw_deny_unknown_exthdrs) return (IP_FW_DENY); PULLUP_TO(hlen, ulp, struct ip6_ext); break; } /*switch */ } ip = mtod(m, struct ip *); ip6 = (struct ip6_hdr *)ip; args->f_id.src_ip6 = ip6->ip6_src; args->f_id.dst_ip6 = ip6->ip6_dst; args->f_id.src_ip = 0; args->f_id.dst_ip = 0; args->f_id.flow_id6 = ntohl(ip6->ip6_flow); } else if (pktlen >= sizeof(struct ip) && (args->eh == NULL || etype == ETHERTYPE_IP) && ip->ip_v == 4) { is_ipv4 = 1; hlen = ip->ip_hl << 2; args->f_id.addr_type = 4; /* * Collect parameters into local variables for faster matching. */ proto = ip->ip_p; src_ip = ip->ip_src; dst_ip = ip->ip_dst; offset = ntohs(ip->ip_off) & IP_OFFMASK; iplen = ntohs(ip->ip_len); pktlen = iplen < pktlen ? iplen : pktlen; if (offset == 0) { switch (proto) { case IPPROTO_TCP: PULLUP_TO(hlen, ulp, struct tcphdr); dst_port = TCP(ulp)->th_dport; src_port = TCP(ulp)->th_sport; /* save flags for dynamic rules */ args->f_id._flags = TCP(ulp)->th_flags; break; case IPPROTO_SCTP: PULLUP_TO(hlen, ulp, struct sctphdr); src_port = SCTP(ulp)->src_port; dst_port = SCTP(ulp)->dest_port; break; case IPPROTO_UDP: PULLUP_TO(hlen, ulp, struct udphdr); dst_port = UDP(ulp)->uh_dport; src_port = UDP(ulp)->uh_sport; break; case IPPROTO_ICMP: PULLUP_TO(hlen, ulp, struct icmphdr); //args->f_id.flags = ICMP(ulp)->icmp_type; break; default: break; } } ip = mtod(m, struct ip *); args->f_id.src_ip = ntohl(src_ip.s_addr); args->f_id.dst_ip = ntohl(dst_ip.s_addr); } #undef PULLUP_TO if (proto) { /* we may have port numbers, store them */ args->f_id.proto = proto; args->f_id.src_port = src_port = ntohs(src_port); args->f_id.dst_port = dst_port = ntohs(dst_port); } IPFW_RLOCK(chain); if (! V_ipfw_vnet_ready) { /* shutting down, leave NOW. */ IPFW_RUNLOCK(chain); return (IP_FW_PASS); /* accept */ } if (args->rule.slot) { /* * Packet has already been tagged as a result of a previous * match on rule args->rule aka args->rule_id (PIPE, QUEUE, * REASS, NETGRAPH, DIVERT/TEE...) * Validate the slot and continue from the next one * if still present, otherwise do a lookup. */ f_pos = (args->rule.chain_id == chain->id) ? args->rule.slot : ipfw_find_rule(chain, args->rule.rulenum, args->rule.rule_id); } else { f_pos = 0; } /* * Now scan the rules, and parse microinstructions for each rule. * We have two nested loops and an inner switch. Sometimes we * need to break out of one or both loops, or re-enter one of * the loops with updated variables. Loop variables are: * * f_pos (outer loop) points to the current rule. * On output it points to the matching rule. * done (outer loop) is used as a flag to break the loop. * l (inner loop) residual length of current rule. * cmd points to the current microinstruction. * * We break the inner loop by setting l=0 and possibly * cmdlen=0 if we don't want to advance cmd. * We break the outer loop by setting done=1 * We can restart the inner loop by setting l>0 and f_pos, f, cmd * as needed. */ for (; f_pos < chain->n_rules; f_pos++) { ipfw_insn *cmd; uint32_t tablearg = 0; int l, cmdlen, skip_or; /* skip rest of OR block */ struct ip_fw *f; f = chain->map[f_pos]; if (V_set_disable & (1 << f->set) ) continue; skip_or = 0; for (l = f->cmd_len, cmd = f->cmd ; l > 0 ; l -= cmdlen, cmd += cmdlen) { int match; /* * check_body is a jump target used when we find a * CHECK_STATE, and need to jump to the body of * the target rule. */ /* check_body: */ cmdlen = F_LEN(cmd); /* * An OR block (insn_1 || .. || insn_n) has the * F_OR bit set in all but the last instruction. * The first match will set "skip_or", and cause * the following instructions to be skipped until * past the one with the F_OR bit clear. */ if (skip_or) { /* skip this instruction */ if ((cmd->len & F_OR) == 0) skip_or = 0; /* next one is good */ continue; } match = 0; /* set to 1 if we succeed */ switch (cmd->opcode) { /* * The first set of opcodes compares the packet's * fields with some pattern, setting 'match' if a * match is found. At the end of the loop there is * logic to deal with F_NOT and F_OR flags associated * with the opcode. */ case O_NOP: match = 1; break; case O_FORWARD_MAC: printf("ipfw: opcode %d unimplemented\n", cmd->opcode); break; case O_GID: case O_UID: case O_JAIL: /* * We only check offset == 0 && proto != 0, * as this ensures that we have a * packet with the ports info. */ if (offset!=0) break; if (is_ipv6) /* XXX to be fixed later */ break; if (proto == IPPROTO_TCP || proto == IPPROTO_UDP) match = check_uidgid( (ipfw_insn_u32 *)cmd, proto, oif, dst_ip, dst_port, src_ip, src_port, &ucred_lookup, #ifdef __FreeBSD__ &ucred_cache, args->inp); #else (void *)&ucred_cache, (struct inpcb *)args->m); #endif break; case O_RECV: match = iface_match(m->m_pkthdr.rcvif, (ipfw_insn_if *)cmd); break; case O_XMIT: match = iface_match(oif, (ipfw_insn_if *)cmd); break; case O_VIA: match = iface_match(oif ? oif : m->m_pkthdr.rcvif, (ipfw_insn_if *)cmd); break; case O_MACADDR2: if (args->eh != NULL) { /* have MAC header */ u_int32_t *want = (u_int32_t *) ((ipfw_insn_mac *)cmd)->addr; u_int32_t *mask = (u_int32_t *) ((ipfw_insn_mac *)cmd)->mask; u_int32_t *hdr = (u_int32_t *)args->eh; match = ( want[0] == (hdr[0] & mask[0]) && want[1] == (hdr[1] & mask[1]) && want[2] == (hdr[2] & mask[2]) ); } break; case O_MAC_TYPE: if (args->eh != NULL) { u_int16_t *p = ((ipfw_insn_u16 *)cmd)->ports; int i; for (i = cmdlen - 1; !match && i>0; i--, p += 2) match = (etype >= p[0] && etype <= p[1]); } break; case O_FRAG: match = (offset != 0); break; case O_IN: /* "out" is "not in" */ match = (oif == NULL); break; case O_LAYER2: match = (args->eh != NULL); break; case O_DIVERTED: { /* For diverted packets, args->rule.info * contains the divert port (in host format) * reason and direction. */ uint32_t i = args->rule.info; match = (i&IPFW_IS_MASK) == IPFW_IS_DIVERT && cmd->arg1 & ((i & IPFW_INFO_IN) ? 1 : 2); } break; case O_PROTO: /* * We do not allow an arg of 0 so the * check of "proto" only suffices. */ match = (proto == cmd->arg1); break; case O_IP_SRC: match = is_ipv4 && (((ipfw_insn_ip *)cmd)->addr.s_addr == src_ip.s_addr); break; case O_IP_SRC_LOOKUP: case O_IP_DST_LOOKUP: if (is_ipv4) { uint32_t key = (cmd->opcode == O_IP_DST_LOOKUP) ? dst_ip.s_addr : src_ip.s_addr; uint32_t v = 0; if (cmdlen > F_INSN_SIZE(ipfw_insn_u32)) { /* generic lookup. The key must be * in 32bit big-endian format. */ v = ((ipfw_insn_u32 *)cmd)->d[1]; if (v == 0) key = dst_ip.s_addr; else if (v == 1) key = src_ip.s_addr; else if (v == 6) /* dscp */ key = (ip->ip_tos >> 2) & 0x3f; else if (offset != 0) break; else if (proto != IPPROTO_TCP && proto != IPPROTO_UDP) break; else if (v == 2) key = htonl(dst_port); else if (v == 3) key = htonl(src_port); else if (v == 4 || v == 5) { check_uidgid( (ipfw_insn_u32 *)cmd, proto, oif, dst_ip, dst_port, src_ip, src_port, &ucred_lookup, #ifdef __FreeBSD__ &ucred_cache, args->inp); if (v == 4 /* O_UID */) key = ucred_cache->cr_uid; else if (v == 5 /* O_JAIL */) key = ucred_cache->cr_prison->pr_id; #else /* !__FreeBSD__ */ (void *)&ucred_cache, (struct inpcb *)args->m); if (v ==4 /* O_UID */) key = ucred_cache.uid; else if (v == 5 /* O_JAIL */) key = ucred_cache.xid; #endif /* !__FreeBSD__ */ key = htonl(key); } else break; } match = ipfw_lookup_table(chain, cmd->arg1, key, &v); if (!match) break; if (cmdlen == F_INSN_SIZE(ipfw_insn_u32)) match = ((ipfw_insn_u32 *)cmd)->d[0] == v; else tablearg = v; } break; case O_IP_SRC_MASK: case O_IP_DST_MASK: if (is_ipv4) { uint32_t a = (cmd->opcode == O_IP_DST_MASK) ? dst_ip.s_addr : src_ip.s_addr; uint32_t *p = ((ipfw_insn_u32 *)cmd)->d; int i = cmdlen-1; for (; !match && i>0; i-= 2, p+= 2) match = (p[0] == (a & p[1])); } break; case O_IP_SRC_ME: if (is_ipv4) { struct ifnet *tif; INADDR_TO_IFP(src_ip, tif); match = (tif != NULL); break; } #ifdef INET6 /* FALLTHROUGH */ case O_IP6_SRC_ME: match= is_ipv6 && search_ip6_addr_net(&args->f_id.src_ip6); #endif break; case O_IP_DST_SET: case O_IP_SRC_SET: if (is_ipv4) { u_int32_t *d = (u_int32_t *)(cmd+1); u_int32_t addr = cmd->opcode == O_IP_DST_SET ? args->f_id.dst_ip : args->f_id.src_ip; if (addr < d[0]) break; addr -= d[0]; /* subtract base */ match = (addr < cmd->arg1) && ( d[ 1 + (addr>>5)] & (1<<(addr & 0x1f)) ); } break; case O_IP_DST: match = is_ipv4 && (((ipfw_insn_ip *)cmd)->addr.s_addr == dst_ip.s_addr); break; case O_IP_DST_ME: if (is_ipv4) { struct ifnet *tif; INADDR_TO_IFP(dst_ip, tif); match = (tif != NULL); break; } #ifdef INET6 /* FALLTHROUGH */ case O_IP6_DST_ME: match= is_ipv6 && search_ip6_addr_net(&args->f_id.dst_ip6); #endif break; case O_IP_SRCPORT: case O_IP_DSTPORT: /* * offset == 0 && proto != 0 is enough * to guarantee that we have a * packet with port info. */ if ((proto==IPPROTO_UDP || proto==IPPROTO_TCP) && offset == 0) { u_int16_t x = (cmd->opcode == O_IP_SRCPORT) ? src_port : dst_port ; u_int16_t *p = ((ipfw_insn_u16 *)cmd)->ports; int i; for (i = cmdlen - 1; !match && i>0; i--, p += 2) match = (x>=p[0] && x<=p[1]); } break; case O_ICMPTYPE: match = (offset == 0 && proto==IPPROTO_ICMP && icmptype_match(ICMP(ulp), (ipfw_insn_u32 *)cmd) ); break; #ifdef INET6 case O_ICMP6TYPE: match = is_ipv6 && offset == 0 && proto==IPPROTO_ICMPV6 && icmp6type_match( ICMP6(ulp)->icmp6_type, (ipfw_insn_u32 *)cmd); break; #endif /* INET6 */ case O_IPOPT: match = (is_ipv4 && ipopts_match(ip, cmd) ); break; case O_IPVER: match = (is_ipv4 && cmd->arg1 == ip->ip_v); break; case O_IPID: case O_IPLEN: case O_IPTTL: if (is_ipv4) { /* only for IP packets */ uint16_t x; uint16_t *p; int i; if (cmd->opcode == O_IPLEN) x = iplen; else if (cmd->opcode == O_IPTTL) x = ip->ip_ttl; else /* must be IPID */ x = ntohs(ip->ip_id); if (cmdlen == 1) { match = (cmd->arg1 == x); break; } /* otherwise we have ranges */ p = ((ipfw_insn_u16 *)cmd)->ports; i = cmdlen - 1; for (; !match && i>0; i--, p += 2) match = (x >= p[0] && x <= p[1]); } break; case O_IPPRECEDENCE: match = (is_ipv4 && (cmd->arg1 == (ip->ip_tos & 0xe0)) ); break; case O_IPTOS: match = (is_ipv4 && flags_match(cmd, ip->ip_tos)); break; case O_TCPDATALEN: if (proto == IPPROTO_TCP && offset == 0) { struct tcphdr *tcp; uint16_t x; uint16_t *p; int i; tcp = TCP(ulp); x = iplen - ((ip->ip_hl + tcp->th_off) << 2); if (cmdlen == 1) { match = (cmd->arg1 == x); break; } /* otherwise we have ranges */ p = ((ipfw_insn_u16 *)cmd)->ports; i = cmdlen - 1; for (; !match && i>0; i--, p += 2) match = (x >= p[0] && x <= p[1]); } break; case O_TCPFLAGS: match = (proto == IPPROTO_TCP && offset == 0 && flags_match(cmd, TCP(ulp)->th_flags)); break; case O_TCPOPTS: PULLUP_LEN(hlen, ulp, (TCP(ulp)->th_off << 2)); match = (proto == IPPROTO_TCP && offset == 0 && tcpopts_match(TCP(ulp), cmd)); break; case O_TCPSEQ: match = (proto == IPPROTO_TCP && offset == 0 && ((ipfw_insn_u32 *)cmd)->d[0] == TCP(ulp)->th_seq); break; case O_TCPACK: match = (proto == IPPROTO_TCP && offset == 0 && ((ipfw_insn_u32 *)cmd)->d[0] == TCP(ulp)->th_ack); break; case O_TCPWIN: match = (proto == IPPROTO_TCP && offset == 0 && cmd->arg1 == TCP(ulp)->th_win); break; case O_ESTAB: /* reject packets which have SYN only */ /* XXX should i also check for TH_ACK ? */ match = (proto == IPPROTO_TCP && offset == 0 && (TCP(ulp)->th_flags & (TH_RST | TH_ACK | TH_SYN)) != TH_SYN); break; case O_ALTQ: { struct pf_mtag *at; ipfw_insn_altq *altq = (ipfw_insn_altq *)cmd; match = 1; at = pf_find_mtag(m); if (at != NULL && at->qid != 0) break; at = pf_get_mtag(m); if (at == NULL) { /* * Let the packet fall back to the * default ALTQ. */ break; } at->qid = altq->qid; if (is_ipv4) at->af = AF_INET; else at->af = AF_LINK; at->hdr = ip; break; } case O_LOG: ipfw_log(f, hlen, args, m, oif, offset, tablearg, ip); match = 1; break; case O_PROB: match = (random()<((ipfw_insn_u32 *)cmd)->d[0]); break; case O_VERREVPATH: /* Outgoing packets automatically pass/match */ match = ((oif != NULL) || (m->m_pkthdr.rcvif == NULL) || ( #ifdef INET6 is_ipv6 ? verify_path6(&(args->f_id.src_ip6), m->m_pkthdr.rcvif) : #endif verify_path(src_ip, m->m_pkthdr.rcvif, args->f_id.fib))); break; case O_VERSRCREACH: /* Outgoing packets automatically pass/match */ match = (hlen > 0 && ((oif != NULL) || #ifdef INET6 is_ipv6 ? verify_path6(&(args->f_id.src_ip6), NULL) : #endif verify_path(src_ip, NULL, args->f_id.fib))); break; case O_ANTISPOOF: /* Outgoing packets automatically pass/match */ if (oif == NULL && hlen > 0 && ( (is_ipv4 && in_localaddr(src_ip)) #ifdef INET6 || (is_ipv6 && in6_localaddr(&(args->f_id.src_ip6))) #endif )) match = #ifdef INET6 is_ipv6 ? verify_path6( &(args->f_id.src_ip6), m->m_pkthdr.rcvif) : #endif verify_path(src_ip, m->m_pkthdr.rcvif, args->f_id.fib); else match = 1; break; case O_IPSEC: #ifdef IPSEC match = (m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL) != NULL); #endif /* otherwise no match */ break; #ifdef INET6 case O_IP6_SRC: match = is_ipv6 && IN6_ARE_ADDR_EQUAL(&args->f_id.src_ip6, &((ipfw_insn_ip6 *)cmd)->addr6); break; case O_IP6_DST: match = is_ipv6 && IN6_ARE_ADDR_EQUAL(&args->f_id.dst_ip6, &((ipfw_insn_ip6 *)cmd)->addr6); break; case O_IP6_SRC_MASK: case O_IP6_DST_MASK: if (is_ipv6) { int i = cmdlen - 1; struct in6_addr p; struct in6_addr *d = &((ipfw_insn_ip6 *)cmd)->addr6; for (; !match && i > 0; d += 2, i -= F_INSN_SIZE(struct in6_addr) * 2) { p = (cmd->opcode == O_IP6_SRC_MASK) ? args->f_id.src_ip6: args->f_id.dst_ip6; APPLY_MASK(&p, &d[1]); match = IN6_ARE_ADDR_EQUAL(&d[0], &p); } } break; case O_FLOW6ID: match = is_ipv6 && flow6id_match(args->f_id.flow_id6, (ipfw_insn_u32 *) cmd); break; case O_EXT_HDR: match = is_ipv6 && (ext_hd & ((ipfw_insn *) cmd)->arg1); break; case O_IP6: match = is_ipv6; break; #endif case O_IP4: match = is_ipv4; break; case O_TAG: { struct m_tag *mtag; uint32_t tag = (cmd->arg1 == IP_FW_TABLEARG) ? tablearg : cmd->arg1; /* Packet is already tagged with this tag? */ mtag = m_tag_locate(m, MTAG_IPFW, tag, NULL); /* We have `untag' action when F_NOT flag is * present. And we must remove this mtag from * mbuf and reset `match' to zero (`match' will * be inversed later). * Otherwise we should allocate new mtag and * push it into mbuf. */ if (cmd->len & F_NOT) { /* `untag' action */ if (mtag != NULL) m_tag_delete(m, mtag); match = 0; } else { if (mtag == NULL) { mtag = m_tag_alloc( MTAG_IPFW, tag, 0, M_NOWAIT); if (mtag != NULL) m_tag_prepend(m, mtag); } match = 1; } break; } case O_FIB: /* try match the specified fib */ if (args->f_id.fib == cmd->arg1) match = 1; break; case O_SOCKARG: { struct inpcb *inp = args->inp; struct inpcbinfo *pi; if (is_ipv6) /* XXX can we remove this ? */ break; if (proto == IPPROTO_TCP) pi = &V_tcbinfo; else if (proto == IPPROTO_UDP) pi = &V_udbinfo; else break; /* * XXXRW: so_user_cookie should almost * certainly be inp_user_cookie? */ /* For incomming packet, lookup up the inpcb using the src/dest ip/port tuple */ if (inp == NULL) { inp = in_pcblookup(pi, src_ip, htons(src_port), dst_ip, htons(dst_port), INPLOOKUP_RLOCKPCB, NULL); if (inp != NULL) { tablearg = inp->inp_socket->so_user_cookie; if (tablearg) match = 1; INP_RUNLOCK(inp); } } else { if (inp->inp_socket) { tablearg = inp->inp_socket->so_user_cookie; if (tablearg) match = 1; } } break; } case O_TAGGED: { struct m_tag *mtag; uint32_t tag = (cmd->arg1 == IP_FW_TABLEARG) ? tablearg : cmd->arg1; if (cmdlen == 1) { match = m_tag_locate(m, MTAG_IPFW, tag, NULL) != NULL; break; } /* we have ranges */ for (mtag = m_tag_first(m); mtag != NULL && !match; mtag = m_tag_next(m, mtag)) { uint16_t *p; int i; if (mtag->m_tag_cookie != MTAG_IPFW) continue; p = ((ipfw_insn_u16 *)cmd)->ports; i = cmdlen - 1; for(; !match && i > 0; i--, p += 2) match = mtag->m_tag_id >= p[0] && mtag->m_tag_id <= p[1]; } break; } /* * The second set of opcodes represents 'actions', * i.e. the terminal part of a rule once the packet * matches all previous patterns. * Typically there is only one action for each rule, * and the opcode is stored at the end of the rule * (but there are exceptions -- see below). * * In general, here we set retval and terminate the * outer loop (would be a 'break 3' in some language, * but we need to set l=0, done=1) * * Exceptions: * O_COUNT and O_SKIPTO actions: * instead of terminating, we jump to the next rule * (setting l=0), or to the SKIPTO target (setting * f/f_len, cmd and l as needed), respectively. * * O_TAG, O_LOG and O_ALTQ action parameters: * perform some action and set match = 1; * * O_LIMIT and O_KEEP_STATE: these opcodes are * not real 'actions', and are stored right * before the 'action' part of the rule. * These opcodes try to install an entry in the * state tables; if successful, we continue with * the next opcode (match=1; break;), otherwise * the packet must be dropped (set retval, * break loops with l=0, done=1) * * O_PROBE_STATE and O_CHECK_STATE: these opcodes * cause a lookup of the state table, and a jump * to the 'action' part of the parent rule * if an entry is found, or * (CHECK_STATE only) a jump to the next rule if * the entry is not found. * The result of the lookup is cached so that * further instances of these opcodes become NOPs. * The jump to the next rule is done by setting * l=0, cmdlen=0. */ case O_LIMIT: case O_KEEP_STATE: if (ipfw_install_state(f, (ipfw_insn_limit *)cmd, args, tablearg)) { /* error or limit violation */ retval = IP_FW_DENY; l = 0; /* exit inner loop */ done = 1; /* exit outer loop */ } match = 1; break; case O_PROBE_STATE: case O_CHECK_STATE: /* * dynamic rules are checked at the first * keep-state or check-state occurrence, * with the result being stored in dyn_dir. * The compiler introduces a PROBE_STATE * instruction for us when we have a * KEEP_STATE (because PROBE_STATE needs * to be run first). */ if (dyn_dir == MATCH_UNKNOWN && (q = ipfw_lookup_dyn_rule(&args->f_id, &dyn_dir, proto == IPPROTO_TCP ? TCP(ulp) : NULL)) != NULL) { /* * Found dynamic entry, update stats * and jump to the 'action' part of * the parent rule by setting * f, cmd, l and clearing cmdlen. */ q->pcnt++; q->bcnt += pktlen; /* XXX we would like to have f_pos * readily accessible in the dynamic * rule, instead of having to * lookup q->rule. */ f = q->rule; f_pos = ipfw_find_rule(chain, f->rulenum, f->id); cmd = ACTION_PTR(f); l = f->cmd_len - f->act_ofs; ipfw_dyn_unlock(); cmdlen = 0; match = 1; break; } /* * Dynamic entry not found. If CHECK_STATE, * skip to next rule, if PROBE_STATE just * ignore and continue with next opcode. */ if (cmd->opcode == O_CHECK_STATE) l = 0; /* exit inner loop */ match = 1; break; case O_ACCEPT: retval = 0; /* accept */ l = 0; /* exit inner loop */ done = 1; /* exit outer loop */ break; case O_PIPE: case O_QUEUE: set_match(args, f_pos, chain); args->rule.info = (cmd->arg1 == IP_FW_TABLEARG) ? tablearg : cmd->arg1; if (cmd->opcode == O_PIPE) args->rule.info |= IPFW_IS_PIPE; if (V_fw_one_pass) args->rule.info |= IPFW_ONEPASS; retval = IP_FW_DUMMYNET; l = 0; /* exit inner loop */ done = 1; /* exit outer loop */ break; case O_DIVERT: case O_TEE: if (args->eh) /* not on layer 2 */ break; /* otherwise this is terminal */ l = 0; /* exit inner loop */ done = 1; /* exit outer loop */ retval = (cmd->opcode == O_DIVERT) ? IP_FW_DIVERT : IP_FW_TEE; set_match(args, f_pos, chain); args->rule.info = (cmd->arg1 == IP_FW_TABLEARG) ? tablearg : cmd->arg1; break; case O_COUNT: f->pcnt++; /* update stats */ f->bcnt += pktlen; f->timestamp = time_uptime; l = 0; /* exit inner loop */ break; case O_SKIPTO: f->pcnt++; /* update stats */ f->bcnt += pktlen; f->timestamp = time_uptime; /* If possible use cached f_pos (in f->next_rule), * whose version is written in f->next_rule * (horrible hacks to avoid changing the ABI). */ if (cmd->arg1 != IP_FW_TABLEARG && (uintptr_t)f->x_next == chain->id) { f_pos = (uintptr_t)f->next_rule; } else { int i = (cmd->arg1 == IP_FW_TABLEARG) ? tablearg : cmd->arg1; /* make sure we do not jump backward */ if (i <= f->rulenum) i = f->rulenum + 1; f_pos = ipfw_find_rule(chain, i, 0); /* update the cache */ if (cmd->arg1 != IP_FW_TABLEARG) { f->next_rule = (void *)(uintptr_t)f_pos; f->x_next = (void *)(uintptr_t)chain->id; } } /* * Skip disabled rules, and re-enter * the inner loop with the correct * f_pos, f, l and cmd. * Also clear cmdlen and skip_or */ for (; f_pos < chain->n_rules - 1 && (V_set_disable & (1 << chain->map[f_pos]->set)); f_pos++) ; /* Re-enter the inner loop at the skipto rule. */ f = chain->map[f_pos]; l = f->cmd_len; cmd = f->cmd; match = 1; cmdlen = 0; skip_or = 0; continue; break; /* not reached */ case O_REJECT: /* * Drop the packet and send a reject notice * if the packet is not ICMP (or is an ICMP * query), and it is not multicast/broadcast. */ if (hlen > 0 && is_ipv4 && offset == 0 && (proto != IPPROTO_ICMP || is_icmp_query(ICMP(ulp))) && !(m->m_flags & (M_BCAST|M_MCAST)) && !IN_MULTICAST(ntohl(dst_ip.s_addr))) { send_reject(args, cmd->arg1, iplen, ip); m = args->m; } /* FALLTHROUGH */ #ifdef INET6 case O_UNREACH6: if (hlen > 0 && is_ipv6 && ((offset & IP6F_OFF_MASK) == 0) && (proto != IPPROTO_ICMPV6 || (is_icmp6_query(icmp6_type) == 1)) && !(m->m_flags & (M_BCAST|M_MCAST)) && !IN6_IS_ADDR_MULTICAST(&args->f_id.dst_ip6)) { send_reject6( args, cmd->arg1, hlen, (struct ip6_hdr *)ip); m = args->m; } /* FALLTHROUGH */ #endif case O_DENY: retval = IP_FW_DENY; l = 0; /* exit inner loop */ done = 1; /* exit outer loop */ break; case O_FORWARD_IP: if (args->eh) /* not valid on layer2 pkts */ break; if (q == NULL || q->rule != f || dyn_dir == MATCH_FORWARD) { struct sockaddr_in *sa; sa = &(((ipfw_insn_sa *)cmd)->sa); if (sa->sin_addr.s_addr == INADDR_ANY) { bcopy(sa, &args->hopstore, sizeof(*sa)); args->hopstore.sin_addr.s_addr = htonl(tablearg); args->next_hop = &args->hopstore; } else { args->next_hop = sa; } } retval = IP_FW_PASS; l = 0; /* exit inner loop */ done = 1; /* exit outer loop */ break; case O_NETGRAPH: case O_NGTEE: set_match(args, f_pos, chain); args->rule.info = (cmd->arg1 == IP_FW_TABLEARG) ? tablearg : cmd->arg1; if (V_fw_one_pass) args->rule.info |= IPFW_ONEPASS; retval = (cmd->opcode == O_NETGRAPH) ? IP_FW_NETGRAPH : IP_FW_NGTEE; l = 0; /* exit inner loop */ done = 1; /* exit outer loop */ break; case O_SETFIB: { uint32_t fib; f->pcnt++; /* update stats */ f->bcnt += pktlen; f->timestamp = time_uptime; fib = (cmd->arg1 == IP_FW_TABLEARG) ? tablearg: cmd->arg1; if (fib >= rt_numfibs) fib = 0; M_SETFIB(m, fib); args->f_id.fib = fib; l = 0; /* exit inner loop */ break; } case O_NAT: if (!IPFW_NAT_LOADED) { retval = IP_FW_DENY; } else { struct cfg_nat *t; int nat_id; set_match(args, f_pos, chain); t = ((ipfw_insn_nat *)cmd)->nat; if (t == NULL) { nat_id = (cmd->arg1 == IP_FW_TABLEARG) ? tablearg : cmd->arg1; t = (*lookup_nat_ptr)(&chain->nat, nat_id); if (t == NULL) { retval = IP_FW_DENY; l = 0; /* exit inner loop */ done = 1; /* exit outer loop */ break; } if (cmd->arg1 != IP_FW_TABLEARG) ((ipfw_insn_nat *)cmd)->nat = t; } retval = ipfw_nat_ptr(args, t, m); } l = 0; /* exit inner loop */ done = 1; /* exit outer loop */ break; case O_REASS: { int ip_off; f->pcnt++; f->bcnt += pktlen; l = 0; /* in any case exit inner loop */ ip_off = ntohs(ip->ip_off); /* if not fragmented, go to next rule */ if ((ip_off & (IP_MF | IP_OFFMASK)) == 0) break; /* * ip_reass() expects len & off in host * byte order. */ SET_HOST_IPLEN(ip); args->m = m = ip_reass(m); /* * do IP header checksum fixup. */ if (m == NULL) { /* fragment got swallowed */ retval = IP_FW_DENY; } else { /* good, packet complete */ int hlen; ip = mtod(m, struct ip *); hlen = ip->ip_hl << 2; SET_NET_IPLEN(ip); ip->ip_sum = 0; if (hlen == sizeof(struct ip)) ip->ip_sum = in_cksum_hdr(ip); else ip->ip_sum = in_cksum(m, hlen); retval = IP_FW_REASS; set_match(args, f_pos, chain); } done = 1; /* exit outer loop */ break; } default: panic("-- unknown opcode %d\n", cmd->opcode); } /* end of switch() on opcodes */ /* * if we get here with l=0, then match is irrelevant. */ if (cmd->len & F_NOT) match = !match; if (match) { if (cmd->len & F_OR) skip_or = 1; } else { if (!(cmd->len & F_OR)) /* not an OR block, */ break; /* try next rule */ } } /* end of inner loop, scan opcodes */ #undef PULLUP_LEN if (done) break; /* next_rule:; */ /* try next rule */ } /* end of outer for, scan rules */ if (done) { struct ip_fw *rule = chain->map[f_pos]; /* Update statistics */ rule->pcnt++; rule->bcnt += pktlen; rule->timestamp = time_uptime; } else { retval = IP_FW_DENY; printf("ipfw: ouch!, skip past end of rules, denying packet\n"); } IPFW_RUNLOCK(chain); #ifdef __FreeBSD__ if (ucred_cache != NULL) crfree(ucred_cache); #endif return (retval); pullup_failed: if (V_fw_verbose) printf("ipfw: pullup failed\n"); return (IP_FW_DENY); } /* * Module and VNET glue */ /* * Stuff that must be initialised only on boot or module load */ static int ipfw_init(void) { int error = 0; ipfw_dyn_attach(); /* * Only print out this stuff the first time around, * when called from the sysinit code. */ printf("ipfw2 " #ifdef INET6 "(+ipv6) " #endif "initialized, divert %s, nat %s, " "rule-based forwarding " #ifdef IPFIREWALL_FORWARD "enabled, " #else "disabled, " #endif "default to %s, logging ", #ifdef IPDIVERT "enabled", #else "loadable", #endif #ifdef IPFIREWALL_NAT "enabled", #else "loadable", #endif default_to_accept ? "accept" : "deny"); /* * Note: V_xxx variables can be accessed here but the vnet specific * initializer may not have been called yet for the VIMAGE case. * Tuneables will have been processed. We will print out values for * the default vnet. * XXX This should all be rationalized AFTER 8.0 */ if (V_fw_verbose == 0) printf("disabled\n"); else if (V_verbose_limit == 0) printf("unlimited\n"); else printf("limited to %d packets/entry by default\n", V_verbose_limit); ipfw_log_bpf(1); /* init */ return (error); } /* * Called for the removal of the last instance only on module unload. */ static void ipfw_destroy(void) { ipfw_log_bpf(0); /* uninit */ ipfw_dyn_detach(); printf("IP firewall unloaded\n"); } /* * Stuff that must be initialized for every instance * (including the first of course). */ static int vnet_ipfw_init(const void *unused) { int error; struct ip_fw *rule = NULL; struct ip_fw_chain *chain; chain = &V_layer3_chain; /* First set up some values that are compile time options */ V_autoinc_step = 100; /* bounded to 1..1000 in add_rule() */ V_fw_deny_unknown_exthdrs = 1; #ifdef IPFIREWALL_VERBOSE V_fw_verbose = 1; #endif #ifdef IPFIREWALL_VERBOSE_LIMIT V_verbose_limit = IPFIREWALL_VERBOSE_LIMIT; #endif #ifdef IPFIREWALL_NAT LIST_INIT(&chain->nat); #endif /* insert the default rule and create the initial map */ chain->n_rules = 1; chain->static_len = sizeof(struct ip_fw); chain->map = malloc(sizeof(struct ip_fw *), M_IPFW, M_NOWAIT | M_ZERO); if (chain->map) rule = malloc(chain->static_len, M_IPFW, M_NOWAIT | M_ZERO); if (rule == NULL) { if (chain->map) free(chain->map, M_IPFW); printf("ipfw2: ENOSPC initializing default rule " "(support disabled)\n"); return (ENOSPC); } error = ipfw_init_tables(chain); if (error) { panic("init_tables"); /* XXX Marko fix this ! */ } /* fill and insert the default rule */ rule->act_ofs = 0; rule->rulenum = IPFW_DEFAULT_RULE; rule->cmd_len = 1; rule->set = RESVD_SET; rule->cmd[0].len = 1; rule->cmd[0].opcode = default_to_accept ? O_ACCEPT : O_DENY; chain->rules = chain->default_rule = chain->map[0] = rule; chain->id = rule->id = 1; IPFW_LOCK_INIT(chain); ipfw_dyn_init(); /* First set up some values that are compile time options */ V_ipfw_vnet_ready = 1; /* Open for business */ /* * Hook the sockopt handler, and the layer2 (V_ip_fw_chk_ptr) * and pfil hooks for ipv4 and ipv6. Even if the latter two fail * we still keep the module alive because the sockopt and * layer2 paths are still useful. * ipfw[6]_hook return 0 on success, ENOENT on failure, * so we can ignore the exact return value and just set a flag. * * Note that V_fw[6]_enable are manipulated by a SYSCTL_PROC so * changes in the underlying (per-vnet) variables trigger * immediate hook()/unhook() calls. * In layer2 we have the same behaviour, except that V_ether_ipfw * is checked on each packet because there are no pfil hooks. */ V_ip_fw_ctl_ptr = ipfw_ctl; V_ip_fw_chk_ptr = ipfw_chk; error = ipfw_attach_hooks(1); return (error); } /* * Called for the removal of each instance. */ static int vnet_ipfw_uninit(const void *unused) { struct ip_fw *reap, *rule; struct ip_fw_chain *chain = &V_layer3_chain; int i; V_ipfw_vnet_ready = 0; /* tell new callers to go away */ /* * disconnect from ipv4, ipv6, layer2 and sockopt. * Then grab, release and grab again the WLOCK so we make * sure the update is propagated and nobody will be in. */ (void)ipfw_attach_hooks(0 /* detach */); V_ip_fw_chk_ptr = NULL; V_ip_fw_ctl_ptr = NULL; IPFW_UH_WLOCK(chain); IPFW_UH_WUNLOCK(chain); IPFW_UH_WLOCK(chain); IPFW_WLOCK(chain); IPFW_WUNLOCK(chain); IPFW_WLOCK(chain); ipfw_dyn_uninit(0); /* run the callout_drain */ ipfw_destroy_tables(chain); reap = NULL; for (i = 0; i < chain->n_rules; i++) { rule = chain->map[i]; rule->x_next = reap; reap = rule; } if (chain->map) free(chain->map, M_IPFW); IPFW_WUNLOCK(chain); IPFW_UH_WUNLOCK(chain); if (reap != NULL) ipfw_reap_rules(reap); IPFW_LOCK_DESTROY(chain); ipfw_dyn_uninit(1); /* free the remaining parts */ return 0; } /* * Module event handler. * In general we have the choice of handling most of these events by the * event handler or by the (VNET_)SYS(UN)INIT handlers. I have chosen to * use the SYSINIT handlers as they are more capable of expressing the * flow of control during module and vnet operations, so this is just * a skeleton. Note there is no SYSINIT equivalent of the module * SHUTDOWN handler, but we don't have anything to do in that case anyhow. */ static int ipfw_modevent(module_t mod, int type, void *unused) { int err = 0; switch (type) { case MOD_LOAD: /* Called once at module load or * system boot if compiled in. */ break; case MOD_QUIESCE: /* Called before unload. May veto unloading. */ break; case MOD_UNLOAD: /* Called during unload. */ break; case MOD_SHUTDOWN: /* Called during system shutdown. */ break; default: err = EOPNOTSUPP; break; } return err; } static moduledata_t ipfwmod = { "ipfw", ipfw_modevent, 0 }; /* Define startup order. */ #define IPFW_SI_SUB_FIREWALL SI_SUB_PROTO_IFATTACHDOMAIN #define IPFW_MODEVENT_ORDER (SI_ORDER_ANY - 255) /* On boot slot in here. */ #define IPFW_MODULE_ORDER (IPFW_MODEVENT_ORDER + 1) /* A little later. */ #define IPFW_VNET_ORDER (IPFW_MODEVENT_ORDER + 2) /* Later still. */ DECLARE_MODULE(ipfw, ipfwmod, IPFW_SI_SUB_FIREWALL, IPFW_MODEVENT_ORDER); MODULE_VERSION(ipfw, 2); /* should declare some dependencies here */ /* * Starting up. Done in order after ipfwmod() has been called. * VNET_SYSINIT is also called for each existing vnet and each new vnet. */ SYSINIT(ipfw_init, IPFW_SI_SUB_FIREWALL, IPFW_MODULE_ORDER, ipfw_init, NULL); VNET_SYSINIT(vnet_ipfw_init, IPFW_SI_SUB_FIREWALL, IPFW_VNET_ORDER, vnet_ipfw_init, NULL); /* * Closing up shop. These are done in REVERSE ORDER, but still * after ipfwmod() has been called. Not called on reboot. * VNET_SYSUNINIT is also called for each exiting vnet as it exits. * or when the module is unloaded. */ SYSUNINIT(ipfw_destroy, IPFW_SI_SUB_FIREWALL, IPFW_MODULE_ORDER, ipfw_destroy, NULL); VNET_SYSUNINIT(vnet_ipfw_uninit, IPFW_SI_SUB_FIREWALL, IPFW_VNET_ORDER, vnet_ipfw_uninit, NULL); /* end of file */ Index: projects/largeSMP/sys/netinet/ipfw/ip_fw_sockopt.c =================================================================== --- projects/largeSMP/sys/netinet/ipfw/ip_fw_sockopt.c (revision 222782) +++ projects/largeSMP/sys/netinet/ipfw/ip_fw_sockopt.c (revision 222783) @@ -1,1345 +1,1346 @@ /*- * Copyright (c) 2002-2009 Luigi Rizzo, Universita` di Pisa * * Supported by: Valeria Paoli * * 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("$FreeBSD$"); /* * Sockopt support for ipfw. The routines here implement * the upper half of the ipfw code. */ #if !defined(KLD_MODULE) #include "opt_ipfw.h" #include "opt_ipdivert.h" #include "opt_ipdn.h" #include "opt_inet.h" #ifndef INET #error IPFIREWALL requires INET. #endif /* INET */ #endif #include "opt_inet6.h" #include "opt_ipsec.h" #include #include #include #include /* struct m_tag used by nested headers */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* hooks */ #include #include #ifdef MAC #include #endif MALLOC_DEFINE(M_IPFW, "IpFw/IpAcct", "IpFw/IpAcct chain's"); /* * static variables followed by global ones (none in this file) */ /* * Find the smallest rule >= key, id. * We could use bsearch but it is so simple that we code it directly */ int ipfw_find_rule(struct ip_fw_chain *chain, uint32_t key, uint32_t id) { int i, lo, hi; struct ip_fw *r; for (lo = 0, hi = chain->n_rules - 1; lo < hi;) { i = (lo + hi) / 2; r = chain->map[i]; if (r->rulenum < key) lo = i + 1; /* continue from the next one */ else if (r->rulenum > key) hi = i; /* this might be good */ else if (r->id < id) lo = i + 1; /* continue from the next one */ else /* r->id >= id */ hi = i; /* this might be good */ }; return hi; } /* * allocate a new map, returns the chain locked. extra is the number * of entries to add or delete. */ static struct ip_fw ** get_map(struct ip_fw_chain *chain, int extra, int locked) { for (;;) { struct ip_fw **map; int i; i = chain->n_rules + extra; map = malloc(i * sizeof(struct ip_fw *), M_IPFW, locked ? M_NOWAIT : M_WAITOK); if (map == NULL) { printf("%s: cannot allocate map\n", __FUNCTION__); return NULL; } if (!locked) IPFW_UH_WLOCK(chain); if (i >= chain->n_rules + extra) /* good */ return map; /* otherwise we lost the race, free and retry */ if (!locked) IPFW_UH_WUNLOCK(chain); free(map, M_IPFW); } } /* * swap the maps. It is supposed to be called with IPFW_UH_WLOCK */ static struct ip_fw ** swap_map(struct ip_fw_chain *chain, struct ip_fw **new_map, int new_len) { struct ip_fw **old_map; IPFW_WLOCK(chain); chain->id++; chain->n_rules = new_len; old_map = chain->map; chain->map = new_map; IPFW_WUNLOCK(chain); return old_map; } /* * Add a new rule to the list. Copy the rule into a malloc'ed area, then * possibly create a rule number and add the rule to the list. * Update the rule_number in the input struct so the caller knows it as well. * XXX DO NOT USE FOR THE DEFAULT RULE. * Must be called without IPFW_UH held */ int ipfw_add_rule(struct ip_fw_chain *chain, struct ip_fw *input_rule) { struct ip_fw *rule; int i, l, insert_before; struct ip_fw **map; /* the new array of pointers */ if (chain->rules == NULL || input_rule->rulenum > IPFW_DEFAULT_RULE-1) return (EINVAL); l = RULESIZE(input_rule); rule = malloc(l, M_IPFW, M_WAITOK | M_ZERO); if (rule == NULL) return (ENOSPC); /* get_map returns with IPFW_UH_WLOCK if successful */ map = get_map(chain, 1, 0 /* not locked */); if (map == NULL) { free(rule, M_IPFW); return ENOSPC; } bcopy(input_rule, rule, l); /* clear fields not settable from userland */ rule->x_next = NULL; rule->next_rule = NULL; rule->pcnt = 0; rule->bcnt = 0; rule->timestamp = 0; if (V_autoinc_step < 1) V_autoinc_step = 1; else if (V_autoinc_step > 1000) V_autoinc_step = 1000; /* find the insertion point, we will insert before */ insert_before = rule->rulenum ? rule->rulenum + 1 : IPFW_DEFAULT_RULE; i = ipfw_find_rule(chain, insert_before, 0); /* duplicate first part */ if (i > 0) bcopy(chain->map, map, i * sizeof(struct ip_fw *)); map[i] = rule; /* duplicate remaining part, we always have the default rule */ bcopy(chain->map + i, map + i + 1, sizeof(struct ip_fw *) *(chain->n_rules - i)); if (rule->rulenum == 0) { /* write back the number */ rule->rulenum = i > 0 ? map[i-1]->rulenum : 0; if (rule->rulenum < IPFW_DEFAULT_RULE - V_autoinc_step) rule->rulenum += V_autoinc_step; input_rule->rulenum = rule->rulenum; } rule->id = chain->id + 1; map = swap_map(chain, map, chain->n_rules + 1); chain->static_len += l; IPFW_UH_WUNLOCK(chain); if (map) free(map, M_IPFW); return (0); } /* * Reclaim storage associated with a list of rules. This is * typically the list created using remove_rule. * A NULL pointer on input is handled correctly. */ void ipfw_reap_rules(struct ip_fw *head) { struct ip_fw *rule; while ((rule = head) != NULL) { head = head->x_next; free(rule, M_IPFW); } } /* * Used by del_entry() to check if a rule should be kept. * Returns 1 if the rule must be kept, 0 otherwise. * * Called with cmd = {0,1,5}. * cmd == 0 matches on rule numbers, excludes rules in RESVD_SET if n == 0 ; * cmd == 1 matches on set numbers only, rule numbers are ignored; * cmd == 5 matches on rule and set numbers. * * n == 0 is a wildcard for rule numbers, there is no wildcard for sets. * * Rules to keep are * (default || reserved || !match_set || !match_number) * where * default ::= (rule->rulenum == IPFW_DEFAULT_RULE) * // the default rule is always protected * * reserved ::= (cmd == 0 && n == 0 && rule->set == RESVD_SET) * // RESVD_SET is protected only if cmd == 0 and n == 0 ("ipfw flush") * * match_set ::= (cmd == 0 || rule->set == set) * // set number is ignored for cmd == 0 * * match_number ::= (cmd == 1 || n == 0 || n == rule->rulenum) * // number is ignored for cmd == 1 or n == 0 * */ static int keep_rule(struct ip_fw *rule, uint8_t cmd, uint8_t set, uint32_t n) { return (rule->rulenum == IPFW_DEFAULT_RULE) || (cmd == 0 && n == 0 && rule->set == RESVD_SET) || !(cmd == 0 || rule->set == set) || !(cmd == 1 || n == 0 || n == rule->rulenum); } /** * Remove all rules with given number, or do set manipulation. * Assumes chain != NULL && *chain != NULL. * * The argument is an uint32_t. The low 16 bit are the rule or set number; * the next 8 bits are the new set; the top 8 bits indicate the command: * * 0 delete rules numbered "rulenum" * 1 delete rules in set "rulenum" * 2 move rules "rulenum" to set "new_set" * 3 move rules from set "rulenum" to set "new_set" * 4 swap sets "rulenum" and "new_set" * 5 delete rules "rulenum" and set "new_set" */ static int del_entry(struct ip_fw_chain *chain, uint32_t arg) { struct ip_fw *rule; uint32_t num; /* rule number or old_set */ uint8_t cmd, new_set; int start, end, i, ofs, n; struct ip_fw **map = NULL; int error = 0; num = arg & 0xffff; cmd = (arg >> 24) & 0xff; new_set = (arg >> 16) & 0xff; if (cmd > 5 || new_set > RESVD_SET) return EINVAL; if (cmd == 0 || cmd == 2 || cmd == 5) { if (num >= IPFW_DEFAULT_RULE) return EINVAL; } else { if (num > RESVD_SET) /* old_set */ return EINVAL; } IPFW_UH_WLOCK(chain); /* arbitrate writers */ chain->reap = NULL; /* prepare for deletions */ switch (cmd) { case 0: /* delete rules "num" (num == 0 matches all) */ case 1: /* delete all rules in set N */ case 5: /* delete rules with number N and set "new_set". */ /* * Locate first rule to delete (start), the rule after * the last one to delete (end), and count how many * rules to delete (n). Always use keep_rule() to * determine which rules to keep. */ n = 0; if (cmd == 1) { /* look for a specific set including RESVD_SET. * Must scan the entire range, ignore num. */ new_set = num; for (start = -1, end = i = 0; i < chain->n_rules; i++) { if (keep_rule(chain->map[i], cmd, new_set, 0)) continue; if (start < 0) start = i; end = i; n++; } end++; /* first non-matching */ } else { /* Optimized search on rule numbers */ start = ipfw_find_rule(chain, num, 0); for (end = start; end < chain->n_rules; end++) { rule = chain->map[end]; if (num > 0 && rule->rulenum != num) break; if (!keep_rule(rule, cmd, new_set, num)) n++; } } if (n == 0) { - /* A flush request (arg == 0) on empty ruleset - * returns with no error. On the contrary, + /* A flush request (arg == 0 or cmd == 1) on empty + * ruleset returns with no error. On the contrary, * if there is no match on a specific request, * we return EINVAL. */ - error = (arg == 0) ? 0 : EINVAL; + if (arg != 0 && cmd != 1) + error = EINVAL; break; } /* We have something to delete. Allocate the new map */ map = get_map(chain, -n, 1 /* locked */); if (map == NULL) { error = EINVAL; break; } /* 1. bcopy the initial part of the map */ if (start > 0) bcopy(chain->map, map, start * sizeof(struct ip_fw *)); /* 2. copy active rules between start and end */ for (i = ofs = start; i < end; i++) { rule = chain->map[i]; if (keep_rule(rule, cmd, new_set, num)) map[ofs++] = rule; } /* 3. copy the final part of the map */ bcopy(chain->map + end, map + ofs, (chain->n_rules - end) * sizeof(struct ip_fw *)); /* 4. swap the maps (under BH_LOCK) */ map = swap_map(chain, map, chain->n_rules - n); /* 5. now remove the rules deleted from the old map */ for (i = start; i < end; i++) { int l; rule = map[i]; if (keep_rule(rule, cmd, new_set, num)) continue; l = RULESIZE(rule); chain->static_len -= l; ipfw_remove_dyn_children(rule); rule->x_next = chain->reap; chain->reap = rule; } break; /* * In the next 3 cases the loop stops at (n_rules - 1) * because the default rule is never eligible.. */ case 2: /* move rules with given RULE number to new set */ for (i = 0; i < chain->n_rules - 1; i++) { rule = chain->map[i]; if (rule->rulenum == num) rule->set = new_set; } break; case 3: /* move rules with given SET number to new set */ for (i = 0; i < chain->n_rules - 1; i++) { rule = chain->map[i]; if (rule->set == num) rule->set = new_set; } break; case 4: /* swap two sets */ for (i = 0; i < chain->n_rules - 1; i++) { rule = chain->map[i]; if (rule->set == num) rule->set = new_set; else if (rule->set == new_set) rule->set = num; } break; } rule = chain->reap; chain->reap = NULL; IPFW_UH_WUNLOCK(chain); ipfw_reap_rules(rule); if (map) free(map, M_IPFW); return error; } /* * Clear counters for a specific rule. * Normally run under IPFW_UH_RLOCK, but these are idempotent ops * so we only care that rules do not disappear. */ static void clear_counters(struct ip_fw *rule, int log_only) { ipfw_insn_log *l = (ipfw_insn_log *)ACTION_PTR(rule); if (log_only == 0) { rule->bcnt = rule->pcnt = 0; rule->timestamp = 0; } if (l->o.opcode == O_LOG) l->log_left = l->max_log; } /** * Reset some or all counters on firewall rules. * The argument `arg' is an u_int32_t. The low 16 bit are the rule number, * the next 8 bits are the set number, the top 8 bits are the command: * 0 work with rules from all set's; * 1 work with rules only from specified set. * Specified rule number is zero if we want to clear all entries. * log_only is 1 if we only want to reset logs, zero otherwise. */ static int zero_entry(struct ip_fw_chain *chain, u_int32_t arg, int log_only) { struct ip_fw *rule; char *msg; int i; uint16_t rulenum = arg & 0xffff; uint8_t set = (arg >> 16) & 0xff; uint8_t cmd = (arg >> 24) & 0xff; if (cmd > 1) return (EINVAL); if (cmd == 1 && set > RESVD_SET) return (EINVAL); IPFW_UH_RLOCK(chain); if (rulenum == 0) { V_norule_counter = 0; for (i = 0; i < chain->n_rules; i++) { rule = chain->map[i]; /* Skip rules not in our set. */ if (cmd == 1 && rule->set != set) continue; clear_counters(rule, log_only); } msg = log_only ? "All logging counts reset" : "Accounting cleared"; } else { int cleared = 0; for (i = 0; i < chain->n_rules; i++) { rule = chain->map[i]; if (rule->rulenum == rulenum) { if (cmd == 0 || rule->set == set) clear_counters(rule, log_only); cleared = 1; } if (rule->rulenum > rulenum) break; } if (!cleared) { /* we did not find any matching rules */ IPFW_UH_RUNLOCK(chain); return (EINVAL); } msg = log_only ? "logging count reset" : "cleared"; } IPFW_UH_RUNLOCK(chain); if (V_fw_verbose) { int lev = LOG_SECURITY | LOG_NOTICE; if (rulenum) log(lev, "ipfw: Entry %d %s.\n", rulenum, msg); else log(lev, "ipfw: %s.\n", msg); } return (0); } /* * Check validity of the structure before insert. * Rules are simple, so this mostly need to check rule sizes. */ static int check_ipfw_struct(struct ip_fw *rule, int size) { int l, cmdlen = 0; int have_action=0; ipfw_insn *cmd; if (size < sizeof(*rule)) { printf("ipfw: rule too short\n"); return (EINVAL); } /* first, check for valid size */ l = RULESIZE(rule); if (l != size) { printf("ipfw: size mismatch (have %d want %d)\n", size, l); return (EINVAL); } if (rule->act_ofs >= rule->cmd_len) { printf("ipfw: bogus action offset (%u > %u)\n", rule->act_ofs, rule->cmd_len - 1); return (EINVAL); } /* * Now go for the individual checks. Very simple ones, basically only * instruction sizes. */ for (l = rule->cmd_len, cmd = rule->cmd ; l > 0 ; l -= cmdlen, cmd += cmdlen) { cmdlen = F_LEN(cmd); if (cmdlen > l) { printf("ipfw: opcode %d size truncated\n", cmd->opcode); return EINVAL; } switch (cmd->opcode) { case O_PROBE_STATE: case O_KEEP_STATE: case O_PROTO: case O_IP_SRC_ME: case O_IP_DST_ME: case O_LAYER2: case O_IN: case O_FRAG: case O_DIVERTED: case O_IPOPT: case O_IPTOS: case O_IPPRECEDENCE: case O_IPVER: case O_SOCKARG: case O_TCPWIN: case O_TCPFLAGS: case O_TCPOPTS: case O_ESTAB: case O_VERREVPATH: case O_VERSRCREACH: case O_ANTISPOOF: case O_IPSEC: #ifdef INET6 case O_IP6_SRC_ME: case O_IP6_DST_ME: case O_EXT_HDR: case O_IP6: #endif case O_IP4: case O_TAG: if (cmdlen != F_INSN_SIZE(ipfw_insn)) goto bad_size; break; case O_FIB: if (cmdlen != F_INSN_SIZE(ipfw_insn)) goto bad_size; if (cmd->arg1 >= rt_numfibs) { printf("ipfw: invalid fib number %d\n", cmd->arg1); return EINVAL; } break; case O_SETFIB: if (cmdlen != F_INSN_SIZE(ipfw_insn)) goto bad_size; if ((cmd->arg1 != IP_FW_TABLEARG) && (cmd->arg1 >= rt_numfibs)) { printf("ipfw: invalid fib number %d\n", cmd->arg1); return EINVAL; } goto check_action; case O_UID: case O_GID: case O_JAIL: case O_IP_SRC: case O_IP_DST: case O_TCPSEQ: case O_TCPACK: case O_PROB: case O_ICMPTYPE: if (cmdlen != F_INSN_SIZE(ipfw_insn_u32)) goto bad_size; break; case O_LIMIT: if (cmdlen != F_INSN_SIZE(ipfw_insn_limit)) goto bad_size; break; case O_LOG: if (cmdlen != F_INSN_SIZE(ipfw_insn_log)) goto bad_size; ((ipfw_insn_log *)cmd)->log_left = ((ipfw_insn_log *)cmd)->max_log; break; case O_IP_SRC_MASK: case O_IP_DST_MASK: /* only odd command lengths */ if ( !(cmdlen & 1) || cmdlen > 31) goto bad_size; break; case O_IP_SRC_SET: case O_IP_DST_SET: if (cmd->arg1 == 0 || cmd->arg1 > 256) { printf("ipfw: invalid set size %d\n", cmd->arg1); return EINVAL; } if (cmdlen != F_INSN_SIZE(ipfw_insn_u32) + (cmd->arg1+31)/32 ) goto bad_size; break; case O_IP_SRC_LOOKUP: case O_IP_DST_LOOKUP: if (cmd->arg1 >= IPFW_TABLES_MAX) { printf("ipfw: invalid table number %d\n", cmd->arg1); return (EINVAL); } if (cmdlen != F_INSN_SIZE(ipfw_insn) && cmdlen != F_INSN_SIZE(ipfw_insn_u32) + 1 && cmdlen != F_INSN_SIZE(ipfw_insn_u32)) goto bad_size; break; case O_MACADDR2: if (cmdlen != F_INSN_SIZE(ipfw_insn_mac)) goto bad_size; break; case O_NOP: case O_IPID: case O_IPTTL: case O_IPLEN: case O_TCPDATALEN: case O_TAGGED: if (cmdlen < 1 || cmdlen > 31) goto bad_size; break; case O_MAC_TYPE: case O_IP_SRCPORT: case O_IP_DSTPORT: /* XXX artificial limit, 30 port pairs */ if (cmdlen < 2 || cmdlen > 31) goto bad_size; break; case O_RECV: case O_XMIT: case O_VIA: if (cmdlen != F_INSN_SIZE(ipfw_insn_if)) goto bad_size; break; case O_ALTQ: if (cmdlen != F_INSN_SIZE(ipfw_insn_altq)) goto bad_size; break; case O_PIPE: case O_QUEUE: if (cmdlen != F_INSN_SIZE(ipfw_insn)) goto bad_size; goto check_action; case O_FORWARD_IP: #ifdef IPFIREWALL_FORWARD if (cmdlen != F_INSN_SIZE(ipfw_insn_sa)) goto bad_size; goto check_action; #else return EINVAL; #endif case O_DIVERT: case O_TEE: if (ip_divert_ptr == NULL) return EINVAL; else goto check_size; case O_NETGRAPH: case O_NGTEE: if (ng_ipfw_input_p == NULL) return EINVAL; else goto check_size; case O_NAT: if (!IPFW_NAT_LOADED) return EINVAL; if (cmdlen != F_INSN_SIZE(ipfw_insn_nat)) goto bad_size; goto check_action; case O_FORWARD_MAC: /* XXX not implemented yet */ case O_CHECK_STATE: case O_COUNT: case O_ACCEPT: case O_DENY: case O_REJECT: #ifdef INET6 case O_UNREACH6: #endif case O_SKIPTO: case O_REASS: check_size: if (cmdlen != F_INSN_SIZE(ipfw_insn)) goto bad_size; check_action: if (have_action) { printf("ipfw: opcode %d, multiple actions" " not allowed\n", cmd->opcode); return EINVAL; } have_action = 1; if (l != cmdlen) { printf("ipfw: opcode %d, action must be" " last opcode\n", cmd->opcode); return EINVAL; } break; #ifdef INET6 case O_IP6_SRC: case O_IP6_DST: if (cmdlen != F_INSN_SIZE(struct in6_addr) + F_INSN_SIZE(ipfw_insn)) goto bad_size; break; case O_FLOW6ID: if (cmdlen != F_INSN_SIZE(ipfw_insn_u32) + ((ipfw_insn_u32 *)cmd)->o.arg1) goto bad_size; break; case O_IP6_SRC_MASK: case O_IP6_DST_MASK: if ( !(cmdlen & 1) || cmdlen > 127) goto bad_size; break; case O_ICMP6TYPE: if( cmdlen != F_INSN_SIZE( ipfw_insn_icmp6 ) ) goto bad_size; break; #endif default: switch (cmd->opcode) { #ifndef INET6 case O_IP6_SRC_ME: case O_IP6_DST_ME: case O_EXT_HDR: case O_IP6: case O_UNREACH6: case O_IP6_SRC: case O_IP6_DST: case O_FLOW6ID: case O_IP6_SRC_MASK: case O_IP6_DST_MASK: case O_ICMP6TYPE: printf("ipfw: no IPv6 support in kernel\n"); return EPROTONOSUPPORT; #endif default: printf("ipfw: opcode %d, unknown opcode\n", cmd->opcode); return EINVAL; } } } if (have_action == 0) { printf("ipfw: missing action\n"); return EINVAL; } return 0; bad_size: printf("ipfw: opcode %d size %d wrong\n", cmd->opcode, cmdlen); return EINVAL; } /* * Translation of requests for compatibility with FreeBSD 7.2/8. * a static variable tells us if we have an old client from userland, * and if necessary we translate requests and responses between the * two formats. */ static int is7 = 0; struct ip_fw7 { struct ip_fw7 *next; /* linked list of rules */ struct ip_fw7 *next_rule; /* ptr to next [skipto] rule */ /* 'next_rule' is used to pass up 'set_disable' status */ uint16_t act_ofs; /* offset of action in 32-bit units */ uint16_t cmd_len; /* # of 32-bit words in cmd */ uint16_t rulenum; /* rule number */ uint8_t set; /* rule set (0..31) */ // #define RESVD_SET 31 /* set for default and persistent rules */ uint8_t _pad; /* padding */ // uint32_t id; /* rule id, only in v.8 */ /* These fields are present in all rules. */ uint64_t pcnt; /* Packet counter */ uint64_t bcnt; /* Byte counter */ uint32_t timestamp; /* tv_sec of last match */ ipfw_insn cmd[1]; /* storage for commands */ }; int convert_rule_to_7(struct ip_fw *rule); int convert_rule_to_8(struct ip_fw *rule); #ifndef RULESIZE7 #define RULESIZE7(rule) (sizeof(struct ip_fw7) + \ ((struct ip_fw7 *)(rule))->cmd_len * 4 - 4) #endif /* * Copy the static and dynamic rules to the supplied buffer * and return the amount of space actually used. * Must be run under IPFW_UH_RLOCK */ static size_t ipfw_getrules(struct ip_fw_chain *chain, void *buf, size_t space) { char *bp = buf; char *ep = bp + space; struct ip_fw *rule, *dst; int l, i; time_t boot_seconds; boot_seconds = boottime.tv_sec; for (i = 0; i < chain->n_rules; i++) { rule = chain->map[i]; if (is7) { /* Convert rule to FreeBSd 7.2 format */ l = RULESIZE7(rule); if (bp + l + sizeof(uint32_t) <= ep) { int error; bcopy(rule, bp, l + sizeof(uint32_t)); error = convert_rule_to_7((struct ip_fw *) bp); if (error) return 0; /*XXX correct? */ /* * XXX HACK. Store the disable mask in the "next" * pointer in a wild attempt to keep the ABI the same. * Why do we do this on EVERY rule? */ bcopy(&V_set_disable, &(((struct ip_fw7 *)bp)->next_rule), sizeof(V_set_disable)); if (((struct ip_fw7 *)bp)->timestamp) ((struct ip_fw7 *)bp)->timestamp += boot_seconds; bp += l; } continue; /* go to next rule */ } /* normal mode, don't touch rules */ l = RULESIZE(rule); if (bp + l > ep) { /* should not happen */ printf("overflow dumping static rules\n"); break; } dst = (struct ip_fw *)bp; bcopy(rule, dst, l); /* * XXX HACK. Store the disable mask in the "next" * pointer in a wild attempt to keep the ABI the same. * Why do we do this on EVERY rule? */ bcopy(&V_set_disable, &dst->next_rule, sizeof(V_set_disable)); if (dst->timestamp) dst->timestamp += boot_seconds; bp += l; } ipfw_get_dynamic(&bp, ep); /* protected by the dynamic lock */ return (bp - (char *)buf); } /** * {set|get}sockopt parser. */ int ipfw_ctl(struct sockopt *sopt) { #define RULE_MAXSIZE (256*sizeof(u_int32_t)) int error; size_t size; struct ip_fw *buf, *rule; struct ip_fw_chain *chain; u_int32_t rulenum[2]; error = priv_check(sopt->sopt_td, PRIV_NETINET_IPFW); if (error) return (error); /* * Disallow modifications in really-really secure mode, but still allow * the logging counters to be reset. */ if (sopt->sopt_name == IP_FW_ADD || (sopt->sopt_dir == SOPT_SET && sopt->sopt_name != IP_FW_RESETLOG)) { error = securelevel_ge(sopt->sopt_td->td_ucred, 3); if (error) return (error); } chain = &V_layer3_chain; error = 0; switch (sopt->sopt_name) { case IP_FW_GET: /* * pass up a copy of the current rules. Static rules * come first (the last of which has number IPFW_DEFAULT_RULE), * followed by a possibly empty list of dynamic rule. * The last dynamic rule has NULL in the "next" field. * * Note that the calculated size is used to bound the * amount of data returned to the user. The rule set may * change between calculating the size and returning the * data in which case we'll just return what fits. */ for (;;) { int len = 0, want; size = chain->static_len; size += ipfw_dyn_len(); if (size >= sopt->sopt_valsize) break; buf = malloc(size, M_TEMP, M_WAITOK); if (buf == NULL) break; IPFW_UH_RLOCK(chain); /* check again how much space we need */ want = chain->static_len + ipfw_dyn_len(); if (size >= want) len = ipfw_getrules(chain, buf, size); IPFW_UH_RUNLOCK(chain); if (size >= want) error = sooptcopyout(sopt, buf, len); free(buf, M_TEMP); if (size >= want) break; } break; case IP_FW_FLUSH: /* locking is done within del_entry() */ error = del_entry(chain, 0); /* special case, rule=0, cmd=0 means all */ break; case IP_FW_ADD: rule = malloc(RULE_MAXSIZE, M_TEMP, M_WAITOK); error = sooptcopyin(sopt, rule, RULE_MAXSIZE, sizeof(struct ip_fw7) ); /* * If the size of commands equals RULESIZE7 then we assume * a FreeBSD7.2 binary is talking to us (set is7=1). * is7 is persistent so the next 'ipfw list' command * will use this format. * NOTE: If wrong version is guessed (this can happen if * the first ipfw command is 'ipfw [pipe] list') * the ipfw binary may crash or loop infinitly... */ if (sopt->sopt_valsize == RULESIZE7(rule)) { is7 = 1; error = convert_rule_to_8(rule); if (error) return error; if (error == 0) error = check_ipfw_struct(rule, RULESIZE(rule)); } else { is7 = 0; if (error == 0) error = check_ipfw_struct(rule, sopt->sopt_valsize); } if (error == 0) { /* locking is done within ipfw_add_rule() */ error = ipfw_add_rule(chain, rule); size = RULESIZE(rule); if (!error && sopt->sopt_dir == SOPT_GET) { if (is7) { error = convert_rule_to_7(rule); size = RULESIZE7(rule); if (error) return error; } error = sooptcopyout(sopt, rule, size); } } free(rule, M_TEMP); break; case IP_FW_DEL: /* * IP_FW_DEL is used for deleting single rules or sets, * and (ab)used to atomically manipulate sets. Argument size * is used to distinguish between the two: * sizeof(u_int32_t) * delete single rule or set of rules, * or reassign rules (or sets) to a different set. * 2*sizeof(u_int32_t) * atomic disable/enable sets. * first u_int32_t contains sets to be disabled, * second u_int32_t contains sets to be enabled. */ error = sooptcopyin(sopt, rulenum, 2*sizeof(u_int32_t), sizeof(u_int32_t)); if (error) break; size = sopt->sopt_valsize; if (size == sizeof(u_int32_t) && rulenum[0] != 0) { /* delete or reassign, locking done in del_entry() */ error = del_entry(chain, rulenum[0]); } else if (size == 2*sizeof(u_int32_t)) { /* set enable/disable */ IPFW_UH_WLOCK(chain); V_set_disable = (V_set_disable | rulenum[0]) & ~rulenum[1] & ~(1<sopt_val != 0) { error = sooptcopyin(sopt, rulenum, sizeof(u_int32_t), sizeof(u_int32_t)); if (error) break; } error = zero_entry(chain, rulenum[0], sopt->sopt_name == IP_FW_RESETLOG); break; /*--- TABLE manipulations are protected by the IPFW_LOCK ---*/ case IP_FW_TABLE_ADD: { ipfw_table_entry ent; error = sooptcopyin(sopt, &ent, sizeof(ent), sizeof(ent)); if (error) break; error = ipfw_add_table_entry(chain, ent.tbl, ent.addr, ent.masklen, ent.value); } break; case IP_FW_TABLE_DEL: { ipfw_table_entry ent; error = sooptcopyin(sopt, &ent, sizeof(ent), sizeof(ent)); if (error) break; error = ipfw_del_table_entry(chain, ent.tbl, ent.addr, ent.masklen); } break; case IP_FW_TABLE_FLUSH: { u_int16_t tbl; error = sooptcopyin(sopt, &tbl, sizeof(tbl), sizeof(tbl)); if (error) break; IPFW_WLOCK(chain); error = ipfw_flush_table(chain, tbl); IPFW_WUNLOCK(chain); } break; case IP_FW_TABLE_GETSIZE: { u_int32_t tbl, cnt; if ((error = sooptcopyin(sopt, &tbl, sizeof(tbl), sizeof(tbl)))) break; IPFW_RLOCK(chain); error = ipfw_count_table(chain, tbl, &cnt); IPFW_RUNLOCK(chain); if (error) break; error = sooptcopyout(sopt, &cnt, sizeof(cnt)); } break; case IP_FW_TABLE_LIST: { ipfw_table *tbl; if (sopt->sopt_valsize < sizeof(*tbl)) { error = EINVAL; break; } size = sopt->sopt_valsize; tbl = malloc(size, M_TEMP, M_WAITOK); error = sooptcopyin(sopt, tbl, size, sizeof(*tbl)); if (error) { free(tbl, M_TEMP); break; } tbl->size = (size - sizeof(*tbl)) / sizeof(ipfw_table_entry); IPFW_RLOCK(chain); error = ipfw_dump_table(chain, tbl); IPFW_RUNLOCK(chain); if (error) { free(tbl, M_TEMP); break; } error = sooptcopyout(sopt, tbl, size); free(tbl, M_TEMP); } break; /*--- NAT operations are protected by the IPFW_LOCK ---*/ case IP_FW_NAT_CFG: if (IPFW_NAT_LOADED) error = ipfw_nat_cfg_ptr(sopt); else { printf("IP_FW_NAT_CFG: %s\n", "ipfw_nat not present, please load it"); error = EINVAL; } break; case IP_FW_NAT_DEL: if (IPFW_NAT_LOADED) error = ipfw_nat_del_ptr(sopt); else { printf("IP_FW_NAT_DEL: %s\n", "ipfw_nat not present, please load it"); error = EINVAL; } break; case IP_FW_NAT_GET_CONFIG: if (IPFW_NAT_LOADED) error = ipfw_nat_get_cfg_ptr(sopt); else { printf("IP_FW_NAT_GET_CFG: %s\n", "ipfw_nat not present, please load it"); error = EINVAL; } break; case IP_FW_NAT_GET_LOG: if (IPFW_NAT_LOADED) error = ipfw_nat_get_log_ptr(sopt); else { printf("IP_FW_NAT_GET_LOG: %s\n", "ipfw_nat not present, please load it"); error = EINVAL; } break; default: printf("ipfw: ipfw_ctl invalid option %d\n", sopt->sopt_name); error = EINVAL; } return (error); #undef RULE_MAXSIZE } #define RULE_MAXSIZE (256*sizeof(u_int32_t)) /* Functions to convert rules 7.2 <==> 8.0 */ int convert_rule_to_7(struct ip_fw *rule) { /* Used to modify original rule */ struct ip_fw7 *rule7 = (struct ip_fw7 *)rule; /* copy of original rule, version 8 */ struct ip_fw *tmp; /* Used to copy commands */ ipfw_insn *ccmd, *dst; int ll = 0, ccmdlen = 0; tmp = malloc(RULE_MAXSIZE, M_TEMP, M_NOWAIT | M_ZERO); if (tmp == NULL) { return 1; //XXX error } bcopy(rule, tmp, RULE_MAXSIZE); /* Copy fields */ rule7->_pad = tmp->_pad; rule7->set = tmp->set; rule7->rulenum = tmp->rulenum; rule7->cmd_len = tmp->cmd_len; rule7->act_ofs = tmp->act_ofs; rule7->next_rule = (struct ip_fw7 *)tmp->next_rule; rule7->next = (struct ip_fw7 *)tmp->x_next; rule7->cmd_len = tmp->cmd_len; rule7->pcnt = tmp->pcnt; rule7->bcnt = tmp->bcnt; rule7->timestamp = tmp->timestamp; /* Copy commands */ for (ll = tmp->cmd_len, ccmd = tmp->cmd, dst = rule7->cmd ; ll > 0 ; ll -= ccmdlen, ccmd += ccmdlen, dst += ccmdlen) { ccmdlen = F_LEN(ccmd); bcopy(ccmd, dst, F_LEN(ccmd)*sizeof(uint32_t)); if (dst->opcode > O_NAT) /* O_REASS doesn't exists in 7.2 version, so * decrement opcode if it is after O_REASS */ dst->opcode--; if (ccmdlen > ll) { printf("ipfw: opcode %d size truncated\n", ccmd->opcode); return EINVAL; } } free(tmp, M_TEMP); return 0; } int convert_rule_to_8(struct ip_fw *rule) { /* Used to modify original rule */ struct ip_fw7 *rule7 = (struct ip_fw7 *) rule; /* Used to copy commands */ ipfw_insn *ccmd, *dst; int ll = 0, ccmdlen = 0; /* Copy of original rule */ struct ip_fw7 *tmp = malloc(RULE_MAXSIZE, M_TEMP, M_NOWAIT | M_ZERO); if (tmp == NULL) { return 1; //XXX error } bcopy(rule7, tmp, RULE_MAXSIZE); for (ll = tmp->cmd_len, ccmd = tmp->cmd, dst = rule->cmd ; ll > 0 ; ll -= ccmdlen, ccmd += ccmdlen, dst += ccmdlen) { ccmdlen = F_LEN(ccmd); bcopy(ccmd, dst, F_LEN(ccmd)*sizeof(uint32_t)); if (dst->opcode > O_NAT) /* O_REASS doesn't exists in 7.2 version, so * increment opcode if it is after O_REASS */ dst->opcode++; if (ccmdlen > ll) { printf("ipfw: opcode %d size truncated\n", ccmd->opcode); return EINVAL; } } rule->_pad = tmp->_pad; rule->set = tmp->set; rule->rulenum = tmp->rulenum; rule->cmd_len = tmp->cmd_len; rule->act_ofs = tmp->act_ofs; rule->next_rule = (struct ip_fw *)tmp->next_rule; rule->x_next = (struct ip_fw *)tmp->next; rule->cmd_len = tmp->cmd_len; rule->id = 0; /* XXX see if is ok = 0 */ rule->pcnt = tmp->pcnt; rule->bcnt = tmp->bcnt; rule->timestamp = tmp->timestamp; free (tmp, M_TEMP); return 0; } /* end of file */ Index: projects/largeSMP/sys/netinet/raw_ip.c =================================================================== --- projects/largeSMP/sys/netinet/raw_ip.c (revision 222782) +++ projects/largeSMP/sys/netinet/raw_ip.c (revision 222783) @@ -1,1115 +1,1116 @@ /*- * 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. * * @(#)raw_ip.c 8.7 (Berkeley) 5/15/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.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 #ifdef IPSEC #include #endif /*IPSEC*/ #include VNET_DEFINE(int, ip_defttl) = IPDEFTTL; SYSCTL_VNET_INT(_net_inet_ip, IPCTL_DEFTTL, ttl, CTLFLAG_RW, &VNET_NAME(ip_defttl), 0, "Maximum TTL on IP packets"); VNET_DEFINE(struct inpcbhead, ripcb); VNET_DEFINE(struct inpcbinfo, ripcbinfo); #define V_ripcb VNET(ripcb) #define V_ripcbinfo VNET(ripcbinfo) /* * Control and data hooks for ipfw, dummynet, divert and so on. * The data hooks are not used here but it is convenient * to keep them all in one place. */ VNET_DEFINE(ip_fw_chk_ptr_t, ip_fw_chk_ptr) = NULL; VNET_DEFINE(ip_fw_ctl_ptr_t, ip_fw_ctl_ptr) = NULL; int (*ip_dn_ctl_ptr)(struct sockopt *); int (*ip_dn_io_ptr)(struct mbuf **, int, struct ip_fw_args *); void (*ip_divert_ptr)(struct mbuf *, int); int (*ng_ipfw_input_p)(struct mbuf **, int, struct ip_fw_args *, int); #ifdef INET /* * Hooks for multicast routing. They all default to NULL, so leave them not * initialized and rely on BSS being set to 0. */ /* * The socket used to communicate with the multicast routing daemon. */ VNET_DEFINE(struct socket *, ip_mrouter); /* * The various mrouter and rsvp functions. */ int (*ip_mrouter_set)(struct socket *, struct sockopt *); int (*ip_mrouter_get)(struct socket *, struct sockopt *); int (*ip_mrouter_done)(void); int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *, struct ip_moptions *); int (*mrt_ioctl)(u_long, caddr_t, int); int (*legal_vif_num)(int); u_long (*ip_mcast_src)(int); void (*rsvp_input_p)(struct mbuf *m, int off); int (*ip_rsvp_vif)(struct socket *, struct sockopt *); void (*ip_rsvp_force_done)(struct socket *); #endif /* INET */ u_long rip_sendspace = 9216; SYSCTL_ULONG(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW, &rip_sendspace, 0, "Maximum outgoing raw IP datagram size"); u_long rip_recvspace = 9216; SYSCTL_ULONG(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW, &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams"); /* * Hash functions */ #define INP_PCBHASH_RAW_SIZE 256 #define INP_PCBHASH_RAW(proto, laddr, faddr, mask) \ (((proto) + (laddr) + (faddr)) % (mask) + 1) #ifdef INET static void rip_inshash(struct inpcb *inp) { struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; struct inpcbhead *pcbhash; int hash; INP_INFO_WLOCK_ASSERT(pcbinfo); INP_WLOCK_ASSERT(inp); if (inp->inp_ip_p != 0 && inp->inp_laddr.s_addr != INADDR_ANY && inp->inp_faddr.s_addr != INADDR_ANY) { hash = INP_PCBHASH_RAW(inp->inp_ip_p, inp->inp_laddr.s_addr, inp->inp_faddr.s_addr, pcbinfo->ipi_hashmask); } else hash = 0; pcbhash = &pcbinfo->ipi_hashbase[hash]; LIST_INSERT_HEAD(pcbhash, inp, inp_hash); } static void rip_delhash(struct inpcb *inp) { INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); INP_WLOCK_ASSERT(inp); LIST_REMOVE(inp, inp_hash); } #endif /* INET */ /* * Raw interface to IP protocol. */ /* * Initialize raw connection block q. */ static void rip_zone_change(void *tag) { uma_zone_set_max(V_ripcbinfo.ipi_zone, maxsockets); } static int rip_inpcb_init(void *mem, int size, int flags) { struct inpcb *inp = mem; INP_LOCK_INIT(inp, "inp", "rawinp"); return (0); } void rip_init(void) { in_pcbinfo_init(&V_ripcbinfo, "rip", &V_ripcb, INP_PCBHASH_RAW_SIZE, - 1, "ripcb", rip_inpcb_init, NULL, UMA_ZONE_NOFREE); + 1, "ripcb", rip_inpcb_init, NULL, UMA_ZONE_NOFREE, + IPI_HASHFIELDS_NONE); EVENTHANDLER_REGISTER(maxsockets_change, rip_zone_change, NULL, EVENTHANDLER_PRI_ANY); } #ifdef VIMAGE void rip_destroy(void) { in_pcbinfo_destroy(&V_ripcbinfo); } #endif #ifdef INET static int rip_append(struct inpcb *last, struct ip *ip, struct mbuf *n, struct sockaddr_in *ripsrc) { int policyfail = 0; INP_LOCK_ASSERT(last); #ifdef IPSEC /* check AH/ESP integrity. */ if (ipsec4_in_reject(n, last)) { policyfail = 1; } #endif /* IPSEC */ #ifdef MAC if (!policyfail && mac_inpcb_check_deliver(last, n) != 0) policyfail = 1; #endif /* Check the minimum TTL for socket. */ if (last->inp_ip_minttl && last->inp_ip_minttl > ip->ip_ttl) policyfail = 1; if (!policyfail) { struct mbuf *opts = NULL; struct socket *so; so = last->inp_socket; if ((last->inp_flags & INP_CONTROLOPTS) || (so->so_options & (SO_TIMESTAMP | SO_BINTIME))) ip_savecontrol(last, &opts, ip, n); SOCKBUF_LOCK(&so->so_rcv); if (sbappendaddr_locked(&so->so_rcv, (struct sockaddr *)ripsrc, n, opts) == 0) { /* should notify about lost packet */ m_freem(n); if (opts) m_freem(opts); SOCKBUF_UNLOCK(&so->so_rcv); } else sorwakeup_locked(so); } else m_freem(n); return (policyfail); } /* * Setup generic address and protocol structures for raw_input routine, then * pass them along with mbuf chain. */ void rip_input(struct mbuf *m, int off) { struct ifnet *ifp; struct ip *ip = mtod(m, struct ip *); int proto = ip->ip_p; struct inpcb *inp, *last; struct sockaddr_in ripsrc; int hash; bzero(&ripsrc, sizeof(ripsrc)); ripsrc.sin_len = sizeof(ripsrc); ripsrc.sin_family = AF_INET; ripsrc.sin_addr = ip->ip_src; last = NULL; ifp = m->m_pkthdr.rcvif; hash = INP_PCBHASH_RAW(proto, ip->ip_src.s_addr, ip->ip_dst.s_addr, V_ripcbinfo.ipi_hashmask); INP_INFO_RLOCK(&V_ripcbinfo); LIST_FOREACH(inp, &V_ripcbinfo.ipi_hashbase[hash], inp_hash) { if (inp->inp_ip_p != proto) continue; #ifdef INET6 /* XXX inp locking */ if ((inp->inp_vflag & INP_IPV4) == 0) continue; #endif if (inp->inp_laddr.s_addr != ip->ip_dst.s_addr) continue; if (inp->inp_faddr.s_addr != ip->ip_src.s_addr) continue; if (jailed_without_vnet(inp->inp_cred)) { /* * XXX: If faddr was bound to multicast group, * jailed raw socket will drop datagram. */ if (prison_check_ip4(inp->inp_cred, &ip->ip_dst) != 0) continue; } if (last != NULL) { struct mbuf *n; n = m_copy(m, 0, (int)M_COPYALL); if (n != NULL) (void) rip_append(last, ip, n, &ripsrc); /* XXX count dropped packet */ INP_RUNLOCK(last); } INP_RLOCK(inp); last = inp; } LIST_FOREACH(inp, &V_ripcbinfo.ipi_hashbase[0], inp_hash) { if (inp->inp_ip_p && inp->inp_ip_p != proto) continue; #ifdef INET6 /* XXX inp locking */ if ((inp->inp_vflag & INP_IPV4) == 0) continue; #endif if (!in_nullhost(inp->inp_laddr) && !in_hosteq(inp->inp_laddr, ip->ip_dst)) continue; if (!in_nullhost(inp->inp_faddr) && !in_hosteq(inp->inp_faddr, ip->ip_src)) continue; if (jailed_without_vnet(inp->inp_cred)) { /* * Allow raw socket in jail to receive multicast; * assume process had PRIV_NETINET_RAW at attach, * and fall through into normal filter path if so. */ if (!IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) && prison_check_ip4(inp->inp_cred, &ip->ip_dst) != 0) continue; } /* * If this raw socket has multicast state, and we * have received a multicast, check if this socket * should receive it, as multicast filtering is now * the responsibility of the transport layer. */ if (inp->inp_moptions != NULL && IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { /* * If the incoming datagram is for IGMP, allow it * through unconditionally to the raw socket. * * In the case of IGMPv2, we may not have explicitly * joined the group, and may have set IFF_ALLMULTI * on the interface. imo_multi_filter() may discard * control traffic we actually need to see. * * Userland multicast routing daemons should continue * filter the control traffic appropriately. */ int blocked; blocked = MCAST_PASS; if (proto != IPPROTO_IGMP) { struct sockaddr_in group; bzero(&group, sizeof(struct sockaddr_in)); group.sin_len = sizeof(struct sockaddr_in); group.sin_family = AF_INET; group.sin_addr = ip->ip_dst; blocked = imo_multi_filter(inp->inp_moptions, ifp, (struct sockaddr *)&group, (struct sockaddr *)&ripsrc); } if (blocked != MCAST_PASS) { IPSTAT_INC(ips_notmember); continue; } } if (last != NULL) { struct mbuf *n; n = m_copy(m, 0, (int)M_COPYALL); if (n != NULL) (void) rip_append(last, ip, n, &ripsrc); /* XXX count dropped packet */ INP_RUNLOCK(last); } INP_RLOCK(inp); last = inp; } INP_INFO_RUNLOCK(&V_ripcbinfo); if (last != NULL) { if (rip_append(last, ip, m, &ripsrc) != 0) IPSTAT_INC(ips_delivered); INP_RUNLOCK(last); } else { m_freem(m); IPSTAT_INC(ips_noproto); IPSTAT_DEC(ips_delivered); } } /* * Generate IP header and pass packet to ip_output. Tack on options user may * have setup with control call. */ int rip_output(struct mbuf *m, struct socket *so, u_long dst) { struct ip *ip; int error; struct inpcb *inp = sotoinpcb(so); int flags = ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) | IP_ALLOWBROADCAST; /* * If the user handed us a complete IP packet, use it. Otherwise, * allocate an mbuf for a header and fill it in. */ if ((inp->inp_flags & INP_HDRINCL) == 0) { if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) { m_freem(m); return(EMSGSIZE); } M_PREPEND(m, sizeof(struct ip), M_DONTWAIT); if (m == NULL) return(ENOBUFS); INP_RLOCK(inp); ip = mtod(m, struct ip *); ip->ip_tos = inp->inp_ip_tos; if (inp->inp_flags & INP_DONTFRAG) ip->ip_off = IP_DF; else ip->ip_off = 0; ip->ip_p = inp->inp_ip_p; ip->ip_len = m->m_pkthdr.len; ip->ip_src = inp->inp_laddr; if (jailed(inp->inp_cred)) { /* * prison_local_ip4() would be good enough but would * let a source of INADDR_ANY pass, which we do not * want to see from jails. We do not go through the * pain of in_pcbladdr() for raw sockets. */ if (ip->ip_src.s_addr == INADDR_ANY) error = prison_get_ip4(inp->inp_cred, &ip->ip_src); else error = prison_local_ip4(inp->inp_cred, &ip->ip_src); if (error != 0) { INP_RUNLOCK(inp); m_freem(m); return (error); } } ip->ip_dst.s_addr = dst; ip->ip_ttl = inp->inp_ip_ttl; } else { if (m->m_pkthdr.len > IP_MAXPACKET) { m_freem(m); return(EMSGSIZE); } INP_RLOCK(inp); ip = mtod(m, struct ip *); error = prison_check_ip4(inp->inp_cred, &ip->ip_src); if (error != 0) { INP_RUNLOCK(inp); m_freem(m); return (error); } /* * Don't allow both user specified and setsockopt options, * and don't allow packet length sizes that will crash. */ if (((ip->ip_hl != (sizeof (*ip) >> 2)) && inp->inp_options) || (ip->ip_len > m->m_pkthdr.len) || (ip->ip_len < (ip->ip_hl << 2))) { INP_RUNLOCK(inp); m_freem(m); return (EINVAL); } if (ip->ip_id == 0) ip->ip_id = ip_newid(); /* * XXX prevent ip_output from overwriting header fields. */ flags |= IP_RAWOUTPUT; IPSTAT_INC(ips_rawout); } if (inp->inp_flags & INP_ONESBCAST) flags |= IP_SENDONES; #ifdef MAC mac_inpcb_create_mbuf(inp, m); #endif error = ip_output(m, inp->inp_options, NULL, flags, inp->inp_moptions, inp); INP_RUNLOCK(inp); return (error); } /* * Raw IP socket option processing. * * IMPORTANT NOTE regarding access control: Traditionally, raw sockets could * only be created by a privileged process, and as such, socket option * operations to manage system properties on any raw socket were allowed to * take place without explicit additional access control checks. However, * raw sockets can now also be created in jail(), and therefore explicit * checks are now required. Likewise, raw sockets can be used by a process * after it gives up privilege, so some caution is required. For options * passed down to the IP layer via ip_ctloutput(), checks are assumed to be * performed in ip_ctloutput() and therefore no check occurs here. * Unilaterally checking priv_check() here breaks normal IP socket option * operations on raw sockets. * * When adding new socket options here, make sure to add access control * checks here as necessary. */ int rip_ctloutput(struct socket *so, struct sockopt *sopt) { struct inpcb *inp = sotoinpcb(so); int error, optval; if (sopt->sopt_level != IPPROTO_IP) { if ((sopt->sopt_level == SOL_SOCKET) && (sopt->sopt_name == SO_SETFIB)) { inp->inp_inc.inc_fibnum = so->so_fibnum; return (0); } return (EINVAL); } error = 0; switch (sopt->sopt_dir) { case SOPT_GET: switch (sopt->sopt_name) { case IP_HDRINCL: optval = inp->inp_flags & INP_HDRINCL; error = sooptcopyout(sopt, &optval, sizeof optval); break; case IP_FW3: /* generic ipfw v.3 functions */ case IP_FW_ADD: /* ADD actually returns the body... */ case IP_FW_GET: case IP_FW_TABLE_GETSIZE: case IP_FW_TABLE_LIST: case IP_FW_NAT_GET_CONFIG: case IP_FW_NAT_GET_LOG: if (V_ip_fw_ctl_ptr != NULL) error = V_ip_fw_ctl_ptr(sopt); else error = ENOPROTOOPT; break; case IP_DUMMYNET3: /* generic dummynet v.3 functions */ case IP_DUMMYNET_GET: if (ip_dn_ctl_ptr != NULL) error = ip_dn_ctl_ptr(sopt); else error = ENOPROTOOPT; break ; case MRT_INIT: case MRT_DONE: case MRT_ADD_VIF: case MRT_DEL_VIF: case MRT_ADD_MFC: case MRT_DEL_MFC: case MRT_VERSION: case MRT_ASSERT: case MRT_API_SUPPORT: case MRT_API_CONFIG: case MRT_ADD_BW_UPCALL: case MRT_DEL_BW_UPCALL: error = priv_check(curthread, PRIV_NETINET_MROUTE); if (error != 0) return (error); error = ip_mrouter_get ? ip_mrouter_get(so, sopt) : EOPNOTSUPP; break; default: error = ip_ctloutput(so, sopt); break; } break; case SOPT_SET: switch (sopt->sopt_name) { case IP_HDRINCL: error = sooptcopyin(sopt, &optval, sizeof optval, sizeof optval); if (error) break; if (optval) inp->inp_flags |= INP_HDRINCL; else inp->inp_flags &= ~INP_HDRINCL; break; case IP_FW3: /* generic ipfw v.3 functions */ case IP_FW_ADD: case IP_FW_DEL: case IP_FW_FLUSH: case IP_FW_ZERO: case IP_FW_RESETLOG: case IP_FW_TABLE_ADD: case IP_FW_TABLE_DEL: case IP_FW_TABLE_FLUSH: case IP_FW_NAT_CFG: case IP_FW_NAT_DEL: if (V_ip_fw_ctl_ptr != NULL) error = V_ip_fw_ctl_ptr(sopt); else error = ENOPROTOOPT; break; case IP_DUMMYNET3: /* generic dummynet v.3 functions */ case IP_DUMMYNET_CONFIGURE: case IP_DUMMYNET_DEL: case IP_DUMMYNET_FLUSH: if (ip_dn_ctl_ptr != NULL) error = ip_dn_ctl_ptr(sopt); else error = ENOPROTOOPT ; break ; case IP_RSVP_ON: error = priv_check(curthread, PRIV_NETINET_MROUTE); if (error != 0) return (error); error = ip_rsvp_init(so); break; case IP_RSVP_OFF: error = priv_check(curthread, PRIV_NETINET_MROUTE); if (error != 0) return (error); error = ip_rsvp_done(); break; case IP_RSVP_VIF_ON: case IP_RSVP_VIF_OFF: error = priv_check(curthread, PRIV_NETINET_MROUTE); if (error != 0) return (error); error = ip_rsvp_vif ? ip_rsvp_vif(so, sopt) : EINVAL; break; case MRT_INIT: case MRT_DONE: case MRT_ADD_VIF: case MRT_DEL_VIF: case MRT_ADD_MFC: case MRT_DEL_MFC: case MRT_VERSION: case MRT_ASSERT: case MRT_API_SUPPORT: case MRT_API_CONFIG: case MRT_ADD_BW_UPCALL: case MRT_DEL_BW_UPCALL: error = priv_check(curthread, PRIV_NETINET_MROUTE); if (error != 0) return (error); error = ip_mrouter_set ? ip_mrouter_set(so, sopt) : EOPNOTSUPP; break; default: error = ip_ctloutput(so, sopt); break; } break; } return (error); } /* * This function exists solely to receive the PRC_IFDOWN messages which are * sent by if_down(). It looks for an ifaddr whose ifa_addr is sa, and calls * in_ifadown() to remove all routes corresponding to that address. It also * receives the PRC_IFUP messages from if_up() and reinstalls the interface * routes. */ void rip_ctlinput(int cmd, struct sockaddr *sa, void *vip) { struct in_ifaddr *ia; struct ifnet *ifp; int err; int flags; switch (cmd) { case PRC_IFDOWN: IN_IFADDR_RLOCK(); TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) { if (ia->ia_ifa.ifa_addr == sa && (ia->ia_flags & IFA_ROUTE)) { ifa_ref(&ia->ia_ifa); IN_IFADDR_RUNLOCK(); /* * in_ifscrub kills the interface route. */ in_ifscrub(ia->ia_ifp, ia, 0); /* * in_ifadown gets rid of all the rest of the * routes. This is not quite the right thing * to do, but at least if we are running a * routing process they will come back. */ in_ifadown(&ia->ia_ifa, 0); ifa_free(&ia->ia_ifa); break; } } if (ia == NULL) /* If ia matched, already unlocked. */ IN_IFADDR_RUNLOCK(); break; case PRC_IFUP: IN_IFADDR_RLOCK(); TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) { if (ia->ia_ifa.ifa_addr == sa) break; } if (ia == NULL || (ia->ia_flags & IFA_ROUTE)) { IN_IFADDR_RUNLOCK(); return; } ifa_ref(&ia->ia_ifa); IN_IFADDR_RUNLOCK(); flags = RTF_UP; ifp = ia->ia_ifa.ifa_ifp; if ((ifp->if_flags & IFF_LOOPBACK) || (ifp->if_flags & IFF_POINTOPOINT)) flags |= RTF_HOST; err = ifa_del_loopback_route((struct ifaddr *)ia, sa); if (err == 0) ia->ia_flags &= ~IFA_RTSELF; err = rtinit(&ia->ia_ifa, RTM_ADD, flags); if (err == 0) ia->ia_flags |= IFA_ROUTE; err = ifa_add_loopback_route((struct ifaddr *)ia, sa); if (err == 0) ia->ia_flags |= IFA_RTSELF; ifa_free(&ia->ia_ifa); break; } } static int rip_attach(struct socket *so, int proto, struct thread *td) { struct inpcb *inp; int error; inp = sotoinpcb(so); KASSERT(inp == NULL, ("rip_attach: inp != NULL")); error = priv_check(td, PRIV_NETINET_RAW); if (error) return (error); if (proto >= IPPROTO_MAX || proto < 0) return EPROTONOSUPPORT; error = soreserve(so, rip_sendspace, rip_recvspace); if (error) return (error); INP_INFO_WLOCK(&V_ripcbinfo); error = in_pcballoc(so, &V_ripcbinfo); if (error) { INP_INFO_WUNLOCK(&V_ripcbinfo); return (error); } inp = (struct inpcb *)so->so_pcb; inp->inp_vflag |= INP_IPV4; inp->inp_ip_p = proto; inp->inp_ip_ttl = V_ip_defttl; rip_inshash(inp); INP_INFO_WUNLOCK(&V_ripcbinfo); INP_WUNLOCK(inp); return (0); } static void rip_detach(struct socket *so) { struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("rip_detach: inp == NULL")); KASSERT(inp->inp_faddr.s_addr == INADDR_ANY, ("rip_detach: not closed")); INP_INFO_WLOCK(&V_ripcbinfo); INP_WLOCK(inp); rip_delhash(inp); if (so == V_ip_mrouter && ip_mrouter_done) ip_mrouter_done(); if (ip_rsvp_force_done) ip_rsvp_force_done(so); if (so == V_ip_rsvpd) ip_rsvp_done(); in_pcbdetach(inp); in_pcbfree(inp); INP_INFO_WUNLOCK(&V_ripcbinfo); } static void rip_dodisconnect(struct socket *so, struct inpcb *inp) { struct inpcbinfo *pcbinfo; pcbinfo = inp->inp_pcbinfo; INP_INFO_WLOCK(pcbinfo); INP_WLOCK(inp); rip_delhash(inp); inp->inp_faddr.s_addr = INADDR_ANY; rip_inshash(inp); SOCK_LOCK(so); so->so_state &= ~SS_ISCONNECTED; SOCK_UNLOCK(so); INP_WUNLOCK(inp); INP_INFO_WUNLOCK(pcbinfo); } static void rip_abort(struct socket *so) { struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("rip_abort: inp == NULL")); rip_dodisconnect(so, inp); } static void rip_close(struct socket *so) { struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("rip_close: inp == NULL")); rip_dodisconnect(so, inp); } static int rip_disconnect(struct socket *so) { struct inpcb *inp; if ((so->so_state & SS_ISCONNECTED) == 0) return (ENOTCONN); inp = sotoinpcb(so); KASSERT(inp != NULL, ("rip_disconnect: inp == NULL")); rip_dodisconnect(so, inp); return (0); } static int rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td) { struct sockaddr_in *addr = (struct sockaddr_in *)nam; struct inpcb *inp; int error; if (nam->sa_len != sizeof(*addr)) return (EINVAL); error = prison_check_ip4(td->td_ucred, &addr->sin_addr); if (error != 0) return (error); inp = sotoinpcb(so); KASSERT(inp != NULL, ("rip_bind: inp == NULL")); if (TAILQ_EMPTY(&V_ifnet) || (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) || (addr->sin_addr.s_addr && (inp->inp_flags & INP_BINDANY) == 0 && ifa_ifwithaddr_check((struct sockaddr *)addr) == 0)) return (EADDRNOTAVAIL); INP_INFO_WLOCK(&V_ripcbinfo); INP_WLOCK(inp); rip_delhash(inp); inp->inp_laddr = addr->sin_addr; rip_inshash(inp); INP_WUNLOCK(inp); INP_INFO_WUNLOCK(&V_ripcbinfo); return (0); } static int rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td) { struct sockaddr_in *addr = (struct sockaddr_in *)nam; struct inpcb *inp; if (nam->sa_len != sizeof(*addr)) return (EINVAL); if (TAILQ_EMPTY(&V_ifnet)) return (EADDRNOTAVAIL); if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) return (EAFNOSUPPORT); inp = sotoinpcb(so); KASSERT(inp != NULL, ("rip_connect: inp == NULL")); INP_INFO_WLOCK(&V_ripcbinfo); INP_WLOCK(inp); rip_delhash(inp); inp->inp_faddr = addr->sin_addr; rip_inshash(inp); soisconnected(so); INP_WUNLOCK(inp); INP_INFO_WUNLOCK(&V_ripcbinfo); return (0); } static int rip_shutdown(struct socket *so) { struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("rip_shutdown: inp == NULL")); INP_WLOCK(inp); socantsendmore(so); INP_WUNLOCK(inp); return (0); } static int rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, struct mbuf *control, struct thread *td) { struct inpcb *inp; u_long dst; inp = sotoinpcb(so); KASSERT(inp != NULL, ("rip_send: inp == NULL")); /* * Note: 'dst' reads below are unlocked. */ if (so->so_state & SS_ISCONNECTED) { if (nam) { m_freem(m); return (EISCONN); } dst = inp->inp_faddr.s_addr; /* Unlocked read. */ } else { if (nam == NULL) { m_freem(m); return (ENOTCONN); } dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr; } return (rip_output(m, so, dst)); } #endif /* INET */ static int rip_pcblist(SYSCTL_HANDLER_ARGS) { int error, i, n; struct inpcb *inp, **inp_list; inp_gen_t gencnt; struct xinpgen xig; /* * The process of preparing the TCB list is too time-consuming and * resource-intensive to repeat twice on every request. */ if (req->oldptr == 0) { n = V_ripcbinfo.ipi_count; n += imax(n / 8, 10); req->oldidx = 2 * (sizeof xig) + n * sizeof(struct xinpcb); return (0); } if (req->newptr != 0) return (EPERM); /* * OK, now we're committed to doing something. */ INP_INFO_RLOCK(&V_ripcbinfo); gencnt = V_ripcbinfo.ipi_gencnt; n = V_ripcbinfo.ipi_count; INP_INFO_RUNLOCK(&V_ripcbinfo); xig.xig_len = sizeof xig; xig.xig_count = n; xig.xig_gen = gencnt; xig.xig_sogen = so_gencnt; error = SYSCTL_OUT(req, &xig, sizeof xig); if (error) return (error); inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); if (inp_list == 0) return (ENOMEM); INP_INFO_RLOCK(&V_ripcbinfo); for (inp = LIST_FIRST(V_ripcbinfo.ipi_listhead), i = 0; inp && i < n; inp = LIST_NEXT(inp, inp_list)) { INP_WLOCK(inp); if (inp->inp_gencnt <= gencnt && cr_canseeinpcb(req->td->td_ucred, inp) == 0) { in_pcbref(inp); inp_list[i++] = inp; } INP_WUNLOCK(inp); } INP_INFO_RUNLOCK(&V_ripcbinfo); n = i; error = 0; for (i = 0; i < n; i++) { inp = inp_list[i]; INP_RLOCK(inp); if (inp->inp_gencnt <= gencnt) { struct xinpcb xi; bzero(&xi, sizeof(xi)); xi.xi_len = sizeof xi; /* XXX should avoid extra copy */ bcopy(inp, &xi.xi_inp, sizeof *inp); if (inp->inp_socket) sotoxsocket(inp->inp_socket, &xi.xi_socket); INP_RUNLOCK(inp); error = SYSCTL_OUT(req, &xi, sizeof xi); } else INP_RUNLOCK(inp); } INP_INFO_WLOCK(&V_ripcbinfo); for (i = 0; i < n; i++) { inp = inp_list[i]; INP_RLOCK(inp); if (!in_pcbrele_rlocked(inp)) INP_RUNLOCK(inp); } INP_INFO_WUNLOCK(&V_ripcbinfo); if (!error) { /* * Give the user an updated idea of our state. If the * generation differs from what we told her before, she knows * that something happened while we were processing this * request, and it might be necessary to retry. */ INP_INFO_RLOCK(&V_ripcbinfo); xig.xig_gen = V_ripcbinfo.ipi_gencnt; xig.xig_sogen = so_gencnt; xig.xig_count = V_ripcbinfo.ipi_count; INP_INFO_RUNLOCK(&V_ripcbinfo); error = SYSCTL_OUT(req, &xig, sizeof xig); } free(inp_list, M_TEMP); return (error); } SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLTYPE_OPAQUE | CTLFLAG_RD, NULL, 0, rip_pcblist, "S,xinpcb", "List of active raw IP sockets"); #ifdef INET struct pr_usrreqs rip_usrreqs = { .pru_abort = rip_abort, .pru_attach = rip_attach, .pru_bind = rip_bind, .pru_connect = rip_connect, .pru_control = in_control, .pru_detach = rip_detach, .pru_disconnect = rip_disconnect, .pru_peeraddr = in_getpeeraddr, .pru_send = rip_send, .pru_shutdown = rip_shutdown, .pru_sockaddr = in_getsockaddr, .pru_sosetlabel = in_pcbsosetlabel, .pru_close = rip_close, }; #endif /* INET */ Index: projects/largeSMP/sys/netinet/tcp_subr.c =================================================================== --- projects/largeSMP/sys/netinet/tcp_subr.c (revision 222782) +++ projects/largeSMP/sys/netinet/tcp_subr.c (revision 222783) @@ -1,2303 +1,2304 @@ /*- * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 * 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. * * @(#)tcp_subr.c 8.2 (Berkeley) 5/24/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_compat.h" #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_tcpdebug.h" #include #include #include #include #include #include #include #include #include #include #ifdef INET6 #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef INET6 #include #include #include #include #include #endif #include #include #include #include #include #include #ifdef INET6 #include #endif #include #ifdef TCPDEBUG #include #endif #ifdef INET6 #include #endif #ifdef IPSEC #include #include #ifdef INET6 #include #endif #include #include #endif /*IPSEC*/ #include #include #include VNET_DEFINE(int, tcp_mssdflt) = TCP_MSS; #ifdef INET6 VNET_DEFINE(int, tcp_v6mssdflt) = TCP6_MSS; #endif static int sysctl_net_inet_tcp_mss_check(SYSCTL_HANDLER_ARGS) { int error, new; new = V_tcp_mssdflt; error = sysctl_handle_int(oidp, &new, 0, req); if (error == 0 && req->newptr) { if (new < TCP_MINMSS) error = EINVAL; else V_tcp_mssdflt = new; } return (error); } SYSCTL_VNET_PROC(_net_inet_tcp, TCPCTL_MSSDFLT, mssdflt, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(tcp_mssdflt), 0, &sysctl_net_inet_tcp_mss_check, "I", "Default TCP Maximum Segment Size"); #ifdef INET6 static int sysctl_net_inet_tcp_mss_v6_check(SYSCTL_HANDLER_ARGS) { int error, new; new = V_tcp_v6mssdflt; error = sysctl_handle_int(oidp, &new, 0, req); if (error == 0 && req->newptr) { if (new < TCP_MINMSS) error = EINVAL; else V_tcp_v6mssdflt = new; } return (error); } SYSCTL_VNET_PROC(_net_inet_tcp, TCPCTL_V6MSSDFLT, v6mssdflt, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(tcp_v6mssdflt), 0, &sysctl_net_inet_tcp_mss_v6_check, "I", "Default TCP Maximum Segment Size for IPv6"); #endif /* INET6 */ /* * Minimum MSS we accept and use. This prevents DoS attacks where * we are forced to a ridiculous low MSS like 20 and send hundreds * of packets instead of one. The effect scales with the available * bandwidth and quickly saturates the CPU and network interface * with packet generation and sending. Set to zero to disable MINMSS * checking. This setting prevents us from sending too small packets. */ VNET_DEFINE(int, tcp_minmss) = TCP_MINMSS; SYSCTL_VNET_INT(_net_inet_tcp, OID_AUTO, minmss, CTLFLAG_RW, &VNET_NAME(tcp_minmss), 0, "Minmum TCP Maximum Segment Size"); VNET_DEFINE(int, tcp_do_rfc1323) = 1; SYSCTL_VNET_INT(_net_inet_tcp, TCPCTL_DO_RFC1323, rfc1323, CTLFLAG_RW, &VNET_NAME(tcp_do_rfc1323), 0, "Enable rfc1323 (high performance TCP) extensions"); static int tcp_log_debug = 0; SYSCTL_INT(_net_inet_tcp, OID_AUTO, log_debug, CTLFLAG_RW, &tcp_log_debug, 0, "Log errors caused by incoming TCP segments"); static int tcp_tcbhashsize = 0; SYSCTL_INT(_net_inet_tcp, OID_AUTO, tcbhashsize, CTLFLAG_RDTUN, &tcp_tcbhashsize, 0, "Size of TCP control-block hashtable"); static int do_tcpdrain = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, do_tcpdrain, CTLFLAG_RW, &do_tcpdrain, 0, "Enable tcp_drain routine for extra help when low on mbufs"); SYSCTL_VNET_UINT(_net_inet_tcp, OID_AUTO, pcbcount, CTLFLAG_RD, &VNET_NAME(tcbinfo.ipi_count), 0, "Number of active PCBs"); static VNET_DEFINE(int, icmp_may_rst) = 1; #define V_icmp_may_rst VNET(icmp_may_rst) SYSCTL_VNET_INT(_net_inet_tcp, OID_AUTO, icmp_may_rst, CTLFLAG_RW, &VNET_NAME(icmp_may_rst), 0, "Certain ICMP unreachable messages may abort connections in SYN_SENT"); static VNET_DEFINE(int, tcp_isn_reseed_interval) = 0; #define V_tcp_isn_reseed_interval VNET(tcp_isn_reseed_interval) SYSCTL_VNET_INT(_net_inet_tcp, OID_AUTO, isn_reseed_interval, CTLFLAG_RW, &VNET_NAME(tcp_isn_reseed_interval), 0, "Seconds between reseeding of ISN secret"); #ifdef TCP_SORECEIVE_STREAM static int tcp_soreceive_stream = 0; SYSCTL_INT(_net_inet_tcp, OID_AUTO, soreceive_stream, CTLFLAG_RDTUN, &tcp_soreceive_stream, 0, "Using soreceive_stream for TCP sockets"); #endif #ifdef TCP_SIGNATURE static int tcp_sig_checksigs = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, signature_verify_input, CTLFLAG_RW, &tcp_sig_checksigs, 0, "Verify RFC2385 digests on inbound traffic"); #endif VNET_DEFINE(uma_zone_t, sack_hole_zone); #define V_sack_hole_zone VNET(sack_hole_zone) VNET_DEFINE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST+1]); static struct inpcb *tcp_notify(struct inpcb *, int); static char * tcp_log_addr(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr); /* * Target size of TCP PCB hash tables. Must be a power of two. * * Note that this can be overridden by the kernel environment * variable net.inet.tcp.tcbhashsize */ #ifndef TCBHASHSIZE #define TCBHASHSIZE 512 #endif /* * XXX * Callouts should be moved into struct tcp directly. They are currently * separate because the tcpcb structure is exported to userland for sysctl * parsing purposes, which do not know about callouts. */ struct tcpcb_mem { struct tcpcb tcb; struct tcp_timer tt; struct cc_var ccv; struct osd osd; }; static VNET_DEFINE(uma_zone_t, tcpcb_zone); #define V_tcpcb_zone VNET(tcpcb_zone) MALLOC_DEFINE(M_TCPLOG, "tcplog", "TCP address and flags print buffers"); static struct mtx isn_mtx; #define ISN_LOCK_INIT() mtx_init(&isn_mtx, "isn_mtx", NULL, MTX_DEF) #define ISN_LOCK() mtx_lock(&isn_mtx) #define ISN_UNLOCK() mtx_unlock(&isn_mtx) /* * TCP initialization. */ static void tcp_zone_change(void *tag) { uma_zone_set_max(V_tcbinfo.ipi_zone, maxsockets); uma_zone_set_max(V_tcpcb_zone, maxsockets); tcp_tw_zone_change(); } static int tcp_inpcb_init(void *mem, int size, int flags) { struct inpcb *inp = mem; INP_LOCK_INIT(inp, "inp", "tcpinp"); return (0); } void tcp_init(void) { int hashsize; if (hhook_head_register(HHOOK_TYPE_TCP, HHOOK_TCP_EST_IN, &V_tcp_hhh[HHOOK_TCP_EST_IN], HHOOK_NOWAIT|HHOOK_HEADISINVNET) != 0) printf("%s: WARNING: unable to register helper hook\n", __func__); if (hhook_head_register(HHOOK_TYPE_TCP, HHOOK_TCP_EST_OUT, &V_tcp_hhh[HHOOK_TCP_EST_OUT], HHOOK_NOWAIT|HHOOK_HEADISINVNET) != 0) printf("%s: WARNING: unable to register helper hook\n", __func__); hashsize = TCBHASHSIZE; TUNABLE_INT_FETCH("net.inet.tcp.tcbhashsize", &hashsize); if (!powerof2(hashsize)) { printf("WARNING: TCB hash size not a power of 2\n"); hashsize = 512; /* safe default */ } in_pcbinfo_init(&V_tcbinfo, "tcp", &V_tcb, hashsize, hashsize, - "tcp_inpcb", tcp_inpcb_init, NULL, UMA_ZONE_NOFREE); + "tcp_inpcb", tcp_inpcb_init, NULL, UMA_ZONE_NOFREE, + IPI_HASHFIELDS_4TUPLE); /* * These have to be type stable for the benefit of the timers. */ V_tcpcb_zone = uma_zcreate("tcpcb", sizeof(struct tcpcb_mem), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); uma_zone_set_max(V_tcpcb_zone, maxsockets); tcp_tw_init(); syncache_init(); tcp_hc_init(); tcp_reass_init(); TUNABLE_INT_FETCH("net.inet.tcp.sack.enable", &V_tcp_do_sack); V_sack_hole_zone = uma_zcreate("sackhole", sizeof(struct sackhole), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); /* Skip initialization of globals for non-default instances. */ if (!IS_DEFAULT_VNET(curvnet)) return; /* XXX virtualize those bellow? */ tcp_delacktime = TCPTV_DELACK; tcp_keepinit = TCPTV_KEEP_INIT; tcp_keepidle = TCPTV_KEEP_IDLE; tcp_keepintvl = TCPTV_KEEPINTVL; tcp_maxpersistidle = TCPTV_KEEP_IDLE; tcp_msl = TCPTV_MSL; tcp_rexmit_min = TCPTV_MIN; if (tcp_rexmit_min < 1) tcp_rexmit_min = 1; tcp_rexmit_slop = TCPTV_CPU_VAR; tcp_finwait2_timeout = TCPTV_FINWAIT2_TIMEOUT; tcp_tcbhashsize = hashsize; #ifdef TCP_SORECEIVE_STREAM TUNABLE_INT_FETCH("net.inet.tcp.soreceive_stream", &tcp_soreceive_stream); if (tcp_soreceive_stream) { tcp_usrreqs.pru_soreceive = soreceive_stream; tcp6_usrreqs.pru_soreceive = soreceive_stream; } #endif #ifdef INET6 #define TCP_MINPROTOHDR (sizeof(struct ip6_hdr) + sizeof(struct tcphdr)) #else /* INET6 */ #define TCP_MINPROTOHDR (sizeof(struct tcpiphdr)) #endif /* INET6 */ if (max_protohdr < TCP_MINPROTOHDR) max_protohdr = TCP_MINPROTOHDR; if (max_linkhdr + TCP_MINPROTOHDR > MHLEN) panic("tcp_init"); #undef TCP_MINPROTOHDR ISN_LOCK_INIT(); EVENTHANDLER_REGISTER(shutdown_pre_sync, tcp_fini, NULL, SHUTDOWN_PRI_DEFAULT); EVENTHANDLER_REGISTER(maxsockets_change, tcp_zone_change, NULL, EVENTHANDLER_PRI_ANY); } #ifdef VIMAGE void tcp_destroy(void) { tcp_reass_destroy(); tcp_hc_destroy(); syncache_destroy(); tcp_tw_destroy(); in_pcbinfo_destroy(&V_tcbinfo); uma_zdestroy(V_sack_hole_zone); uma_zdestroy(V_tcpcb_zone); } #endif void tcp_fini(void *xtp) { } /* * Fill in the IP and TCP headers for an outgoing packet, given the tcpcb. * tcp_template used to store this data in mbufs, but we now recopy it out * of the tcpcb each time to conserve mbufs. */ void tcpip_fillheaders(struct inpcb *inp, void *ip_ptr, void *tcp_ptr) { struct tcphdr *th = (struct tcphdr *)tcp_ptr; INP_WLOCK_ASSERT(inp); #ifdef INET6 if ((inp->inp_vflag & INP_IPV6) != 0) { struct ip6_hdr *ip6; ip6 = (struct ip6_hdr *)ip_ptr; ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) | (inp->inp_flow & IPV6_FLOWINFO_MASK); ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) | (IPV6_VERSION & IPV6_VERSION_MASK); ip6->ip6_nxt = IPPROTO_TCP; ip6->ip6_plen = htons(sizeof(struct tcphdr)); ip6->ip6_src = inp->in6p_laddr; ip6->ip6_dst = inp->in6p_faddr; } #endif /* INET6 */ #if defined(INET6) && defined(INET) else #endif #ifdef INET { struct ip *ip; ip = (struct ip *)ip_ptr; ip->ip_v = IPVERSION; ip->ip_hl = 5; ip->ip_tos = inp->inp_ip_tos; ip->ip_len = 0; ip->ip_id = 0; ip->ip_off = 0; ip->ip_ttl = inp->inp_ip_ttl; ip->ip_sum = 0; ip->ip_p = IPPROTO_TCP; ip->ip_src = inp->inp_laddr; ip->ip_dst = inp->inp_faddr; } #endif /* INET */ th->th_sport = inp->inp_lport; th->th_dport = inp->inp_fport; th->th_seq = 0; th->th_ack = 0; th->th_x2 = 0; th->th_off = 5; th->th_flags = 0; th->th_win = 0; th->th_urp = 0; th->th_sum = 0; /* in_pseudo() is called later for ipv4 */ } /* * Create template to be used to send tcp packets on a connection. * Allocates an mbuf and fills in a skeletal tcp/ip header. The only * use for this function is in keepalives, which use tcp_respond. */ struct tcptemp * tcpip_maketemplate(struct inpcb *inp) { struct tcptemp *t; t = malloc(sizeof(*t), M_TEMP, M_NOWAIT); if (t == NULL) return (NULL); tcpip_fillheaders(inp, (void *)&t->tt_ipgen, (void *)&t->tt_t); return (t); } /* * Send a single message to the TCP at address specified by * the given TCP/IP header. If m == NULL, then we make a copy * of the tcpiphdr at ti and send directly to the addressed host. * This is used to force keep alive messages out using the TCP * template for a connection. If flags are given then we send * a message back to the TCP which originated the * segment ti, * and discard the mbuf containing it and any other attached mbufs. * * In any case the ack and sequence number of the transmitted * segment are as specified by the parameters. * * NOTE: If m != NULL, then ti must point to *inside* the mbuf. */ void tcp_respond(struct tcpcb *tp, void *ipgen, struct tcphdr *th, struct mbuf *m, tcp_seq ack, tcp_seq seq, int flags) { int tlen; int win = 0; struct ip *ip; struct tcphdr *nth; #ifdef INET6 struct ip6_hdr *ip6; int isipv6; #endif /* INET6 */ int ipflags = 0; struct inpcb *inp; KASSERT(tp != NULL || m != NULL, ("tcp_respond: tp and m both NULL")); #ifdef INET6 isipv6 = ((struct ip *)ipgen)->ip_v == (IPV6_VERSION >> 4); ip6 = ipgen; #endif /* INET6 */ ip = ipgen; if (tp != NULL) { inp = tp->t_inpcb; KASSERT(inp != NULL, ("tcp control block w/o inpcb")); INP_WLOCK_ASSERT(inp); } else inp = NULL; if (tp != NULL) { if (!(flags & TH_RST)) { win = sbspace(&inp->inp_socket->so_rcv); if (win > (long)TCP_MAXWIN << tp->rcv_scale) win = (long)TCP_MAXWIN << tp->rcv_scale; } } if (m == NULL) { m = m_gethdr(M_DONTWAIT, MT_DATA); if (m == NULL) return; tlen = 0; m->m_data += max_linkhdr; #ifdef INET6 if (isipv6) { bcopy((caddr_t)ip6, mtod(m, caddr_t), sizeof(struct ip6_hdr)); ip6 = mtod(m, struct ip6_hdr *); nth = (struct tcphdr *)(ip6 + 1); } else #endif /* INET6 */ { bcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); ip = mtod(m, struct ip *); nth = (struct tcphdr *)(ip + 1); } bcopy((caddr_t)th, (caddr_t)nth, sizeof(struct tcphdr)); flags = TH_ACK; } else { /* * reuse the mbuf. * XXX MRT We inherrit the FIB, which is lucky. */ m_freem(m->m_next); m->m_next = NULL; m->m_data = (caddr_t)ipgen; /* m_len is set later */ tlen = 0; #define xchg(a,b,type) { type t; t=a; a=b; b=t; } #ifdef INET6 if (isipv6) { xchg(ip6->ip6_dst, ip6->ip6_src, struct in6_addr); nth = (struct tcphdr *)(ip6 + 1); } else #endif /* INET6 */ { xchg(ip->ip_dst.s_addr, ip->ip_src.s_addr, uint32_t); nth = (struct tcphdr *)(ip + 1); } if (th != nth) { /* * this is usually a case when an extension header * exists between the IPv6 header and the * TCP header. */ nth->th_sport = th->th_sport; nth->th_dport = th->th_dport; } xchg(nth->th_dport, nth->th_sport, uint16_t); #undef xchg } #ifdef INET6 if (isipv6) { ip6->ip6_flow = 0; ip6->ip6_vfc = IPV6_VERSION; ip6->ip6_nxt = IPPROTO_TCP; ip6->ip6_plen = htons((u_short)(sizeof (struct tcphdr) + tlen)); tlen += sizeof (struct ip6_hdr) + sizeof (struct tcphdr); } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { tlen += sizeof (struct tcpiphdr); ip->ip_len = tlen; ip->ip_ttl = V_ip_defttl; if (V_path_mtu_discovery) ip->ip_off |= IP_DF; } #endif m->m_len = tlen; m->m_pkthdr.len = tlen; m->m_pkthdr.rcvif = NULL; #ifdef MAC if (inp != NULL) { /* * Packet is associated with a socket, so allow the * label of the response to reflect the socket label. */ INP_WLOCK_ASSERT(inp); mac_inpcb_create_mbuf(inp, m); } else { /* * Packet is not associated with a socket, so possibly * update the label in place. */ mac_netinet_tcp_reply(m); } #endif nth->th_seq = htonl(seq); nth->th_ack = htonl(ack); nth->th_x2 = 0; nth->th_off = sizeof (struct tcphdr) >> 2; nth->th_flags = flags; if (tp != NULL) nth->th_win = htons((u_short) (win >> tp->rcv_scale)); else nth->th_win = htons((u_short)win); nth->th_urp = 0; #ifdef INET6 if (isipv6) { nth->th_sum = 0; nth->th_sum = in6_cksum(m, IPPROTO_TCP, sizeof(struct ip6_hdr), tlen - sizeof(struct ip6_hdr)); ip6->ip6_hlim = in6_selecthlim(tp != NULL ? tp->t_inpcb : NULL, NULL); } #endif /* INET6 */ #if defined(INET6) && defined(INET) else #endif #ifdef INET { nth->th_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htons((u_short)(tlen - sizeof(struct ip) + ip->ip_p))); m->m_pkthdr.csum_flags = CSUM_TCP; m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum); } #endif /* INET */ #ifdef TCPDEBUG if (tp == NULL || (inp->inp_socket->so_options & SO_DEBUG)) tcp_trace(TA_OUTPUT, 0, tp, mtod(m, void *), th, 0); #endif #ifdef INET6 if (isipv6) (void) ip6_output(m, NULL, NULL, ipflags, NULL, NULL, inp); #endif /* INET6 */ #if defined(INET) && defined(INET6) else #endif #ifdef INET (void) ip_output(m, NULL, NULL, ipflags, NULL, inp); #endif } /* * Create a new TCP control block, making an * empty reassembly queue and hooking it to the argument * protocol control block. The `inp' parameter must have * come from the zone allocator set up in tcp_init(). */ struct tcpcb * tcp_newtcpcb(struct inpcb *inp) { struct tcpcb_mem *tm; struct tcpcb *tp; #ifdef INET6 int isipv6 = (inp->inp_vflag & INP_IPV6) != 0; #endif /* INET6 */ tm = uma_zalloc(V_tcpcb_zone, M_NOWAIT | M_ZERO); if (tm == NULL) return (NULL); tp = &tm->tcb; /* Initialise cc_var struct for this tcpcb. */ tp->ccv = &tm->ccv; tp->ccv->type = IPPROTO_TCP; tp->ccv->ccvc.tcp = tp; /* * Use the current system default CC algorithm. */ CC_LIST_RLOCK(); KASSERT(!STAILQ_EMPTY(&cc_list), ("cc_list is empty!")); CC_ALGO(tp) = CC_DEFAULT(); CC_LIST_RUNLOCK(); if (CC_ALGO(tp)->cb_init != NULL) if (CC_ALGO(tp)->cb_init(tp->ccv) > 0) { uma_zfree(V_tcpcb_zone, tm); return (NULL); } tp->osd = &tm->osd; if (khelp_init_osd(HELPER_CLASS_TCP, tp->osd)) { uma_zfree(V_tcpcb_zone, tm); return (NULL); } #ifdef VIMAGE tp->t_vnet = inp->inp_vnet; #endif tp->t_timers = &tm->tt; /* LIST_INIT(&tp->t_segq); */ /* XXX covered by M_ZERO */ tp->t_maxseg = tp->t_maxopd = #ifdef INET6 isipv6 ? V_tcp_v6mssdflt : #endif /* INET6 */ V_tcp_mssdflt; /* Set up our timeouts. */ callout_init(&tp->t_timers->tt_rexmt, CALLOUT_MPSAFE); callout_init(&tp->t_timers->tt_persist, CALLOUT_MPSAFE); callout_init(&tp->t_timers->tt_keep, CALLOUT_MPSAFE); callout_init(&tp->t_timers->tt_2msl, CALLOUT_MPSAFE); callout_init(&tp->t_timers->tt_delack, CALLOUT_MPSAFE); if (V_tcp_do_rfc1323) tp->t_flags = (TF_REQ_SCALE|TF_REQ_TSTMP); if (V_tcp_do_sack) tp->t_flags |= TF_SACK_PERMIT; TAILQ_INIT(&tp->snd_holes); tp->t_inpcb = inp; /* XXX */ /* * Init srtt to TCPTV_SRTTBASE (0), so we can tell that we have no * rtt estimate. Set rttvar so that srtt + 4 * rttvar gives * reasonable initial retransmit time. */ tp->t_srtt = TCPTV_SRTTBASE; tp->t_rttvar = ((TCPTV_RTOBASE - TCPTV_SRTTBASE) << TCP_RTTVAR_SHIFT) / 4; tp->t_rttmin = tcp_rexmit_min; tp->t_rxtcur = TCPTV_RTOBASE; tp->snd_cwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT; tp->snd_ssthresh = TCP_MAXWIN << TCP_MAX_WINSHIFT; tp->t_rcvtime = ticks; /* * IPv4 TTL initialization is necessary for an IPv6 socket as well, * because the socket may be bound to an IPv6 wildcard address, * which may match an IPv4-mapped IPv6 address. */ inp->inp_ip_ttl = V_ip_defttl; inp->inp_ppcb = tp; return (tp); /* XXX */ } /* * Switch the congestion control algorithm back to NewReno for any active * control blocks using an algorithm which is about to go away. * This ensures the CC framework can allow the unload to proceed without leaving * any dangling pointers which would trigger a panic. * Returning non-zero would inform the CC framework that something went wrong * and it would be unsafe to allow the unload to proceed. However, there is no * way for this to occur with this implementation so we always return zero. */ int tcp_ccalgounload(struct cc_algo *unload_algo) { struct cc_algo *tmpalgo; struct inpcb *inp; struct tcpcb *tp; VNET_ITERATOR_DECL(vnet_iter); /* * Check all active control blocks across all network stacks and change * any that are using "unload_algo" back to NewReno. If "unload_algo" * requires cleanup code to be run, call it. */ VNET_LIST_RLOCK(); VNET_FOREACH(vnet_iter) { CURVNET_SET(vnet_iter); INP_INFO_RLOCK(&V_tcbinfo); /* * New connections already part way through being initialised * with the CC algo we're removing will not race with this code * because the INP_INFO_WLOCK is held during initialisation. We * therefore don't enter the loop below until the connection * list has stabilised. */ LIST_FOREACH(inp, &V_tcb, inp_list) { INP_WLOCK(inp); /* Important to skip tcptw structs. */ if (!(inp->inp_flags & INP_TIMEWAIT) && (tp = intotcpcb(inp)) != NULL) { /* * By holding INP_WLOCK here, we are assured * that the connection is not currently * executing inside the CC module's functions * i.e. it is safe to make the switch back to * NewReno. */ if (CC_ALGO(tp) == unload_algo) { tmpalgo = CC_ALGO(tp); /* NewReno does not require any init. */ CC_ALGO(tp) = &newreno_cc_algo; if (tmpalgo->cb_destroy != NULL) tmpalgo->cb_destroy(tp->ccv); } } INP_WUNLOCK(inp); } INP_INFO_RUNLOCK(&V_tcbinfo); CURVNET_RESTORE(); } VNET_LIST_RUNLOCK(); return (0); } /* * Drop a TCP connection, reporting * the specified error. If connection is synchronized, * then send a RST to peer. */ struct tcpcb * tcp_drop(struct tcpcb *tp, int errno) { struct socket *so = tp->t_inpcb->inp_socket; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(tp->t_inpcb); if (TCPS_HAVERCVDSYN(tp->t_state)) { tp->t_state = TCPS_CLOSED; (void) tcp_output_reset(tp); TCPSTAT_INC(tcps_drops); } else TCPSTAT_INC(tcps_conndrops); if (errno == ETIMEDOUT && tp->t_softerror) errno = tp->t_softerror; so->so_error = errno; return (tcp_close(tp)); } void tcp_discardcb(struct tcpcb *tp) { struct inpcb *inp = tp->t_inpcb; struct socket *so = inp->inp_socket; #ifdef INET6 int isipv6 = (inp->inp_vflag & INP_IPV6) != 0; #endif /* INET6 */ INP_WLOCK_ASSERT(inp); /* * Make sure that all of our timers are stopped before we delete the * PCB. * * XXXRW: Really, we would like to use callout_drain() here in order * to avoid races experienced in tcp_timer.c where a timer is already * executing at this point. However, we can't, both because we're * running in a context where we can't sleep, and also because we * hold locks required by the timers. What we instead need to do is * test to see if callout_drain() is required, and if so, defer some * portion of the remainder of tcp_discardcb() to an asynchronous * context that can callout_drain() and then continue. Some care * will be required to ensure that no further processing takes place * on the tcpcb, even though it hasn't been freed (a flag?). */ callout_stop(&tp->t_timers->tt_rexmt); callout_stop(&tp->t_timers->tt_persist); callout_stop(&tp->t_timers->tt_keep); callout_stop(&tp->t_timers->tt_2msl); callout_stop(&tp->t_timers->tt_delack); /* * If we got enough samples through the srtt filter, * save the rtt and rttvar in the routing entry. * 'Enough' is arbitrarily defined as 4 rtt samples. * 4 samples is enough for the srtt filter to converge * to within enough % of the correct value; fewer samples * and we could save a bogus rtt. The danger is not high * as tcp quickly recovers from everything. * XXX: Works very well but needs some more statistics! */ if (tp->t_rttupdated >= 4) { struct hc_metrics_lite metrics; u_long ssthresh; bzero(&metrics, sizeof(metrics)); /* * Update the ssthresh always when the conditions below * are satisfied. This gives us better new start value * for the congestion avoidance for new connections. * ssthresh is only set if packet loss occured on a session. * * XXXRW: 'so' may be NULL here, and/or socket buffer may be * being torn down. Ideally this code would not use 'so'. */ ssthresh = tp->snd_ssthresh; if (ssthresh != 0 && ssthresh < so->so_snd.sb_hiwat / 2) { /* * convert the limit from user data bytes to * packets then to packet data bytes. */ ssthresh = (ssthresh + tp->t_maxseg / 2) / tp->t_maxseg; if (ssthresh < 2) ssthresh = 2; ssthresh *= (u_long)(tp->t_maxseg + #ifdef INET6 (isipv6 ? sizeof (struct ip6_hdr) + sizeof (struct tcphdr) : #endif sizeof (struct tcpiphdr) #ifdef INET6 ) #endif ); } else ssthresh = 0; metrics.rmx_ssthresh = ssthresh; metrics.rmx_rtt = tp->t_srtt; metrics.rmx_rttvar = tp->t_rttvar; metrics.rmx_cwnd = tp->snd_cwnd; metrics.rmx_sendpipe = 0; metrics.rmx_recvpipe = 0; tcp_hc_update(&inp->inp_inc, &metrics); } /* free the reassembly queue, if any */ tcp_reass_flush(tp); /* Disconnect offload device, if any. */ tcp_offload_detach(tp); tcp_free_sackholes(tp); /* Allow the CC algorithm to clean up after itself. */ if (CC_ALGO(tp)->cb_destroy != NULL) CC_ALGO(tp)->cb_destroy(tp->ccv); khelp_destroy_osd(tp->osd); CC_ALGO(tp) = NULL; inp->inp_ppcb = NULL; tp->t_inpcb = NULL; uma_zfree(V_tcpcb_zone, tp); } /* * Attempt to close a TCP control block, marking it as dropped, and freeing * the socket if we hold the only reference. */ struct tcpcb * tcp_close(struct tcpcb *tp) { struct inpcb *inp = tp->t_inpcb; struct socket *so; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(inp); /* Notify any offload devices of listener close */ if (tp->t_state == TCPS_LISTEN) tcp_offload_listen_close(tp); in_pcbdrop(inp); TCPSTAT_INC(tcps_closed); KASSERT(inp->inp_socket != NULL, ("tcp_close: inp_socket NULL")); so = inp->inp_socket; soisdisconnected(so); if (inp->inp_flags & INP_SOCKREF) { KASSERT(so->so_state & SS_PROTOREF, ("tcp_close: !SS_PROTOREF")); inp->inp_flags &= ~INP_SOCKREF; INP_WUNLOCK(inp); ACCEPT_LOCK(); SOCK_LOCK(so); so->so_state &= ~SS_PROTOREF; sofree(so); return (NULL); } return (tp); } void tcp_drain(void) { VNET_ITERATOR_DECL(vnet_iter); if (!do_tcpdrain) return; VNET_LIST_RLOCK_NOSLEEP(); VNET_FOREACH(vnet_iter) { CURVNET_SET(vnet_iter); struct inpcb *inpb; struct tcpcb *tcpb; /* * Walk the tcpbs, if existing, and flush the reassembly queue, * if there is one... * XXX: The "Net/3" implementation doesn't imply that the TCP * reassembly queue should be flushed, but in a situation * where we're really low on mbufs, this is potentially * usefull. */ INP_INFO_RLOCK(&V_tcbinfo); LIST_FOREACH(inpb, V_tcbinfo.ipi_listhead, inp_list) { if (inpb->inp_flags & INP_TIMEWAIT) continue; INP_WLOCK(inpb); if ((tcpb = intotcpcb(inpb)) != NULL) { tcp_reass_flush(tcpb); tcp_clean_sackreport(tcpb); } INP_WUNLOCK(inpb); } INP_INFO_RUNLOCK(&V_tcbinfo); CURVNET_RESTORE(); } VNET_LIST_RUNLOCK_NOSLEEP(); } /* * Notify a tcp user of an asynchronous error; * store error as soft error, but wake up user * (for now, won't do anything until can select for soft error). * * Do not wake up user since there currently is no mechanism for * reporting soft errors (yet - a kqueue filter may be added). */ static struct inpcb * tcp_notify(struct inpcb *inp, int error) { struct tcpcb *tp; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(inp); if ((inp->inp_flags & INP_TIMEWAIT) || (inp->inp_flags & INP_DROPPED)) return (inp); tp = intotcpcb(inp); KASSERT(tp != NULL, ("tcp_notify: tp == NULL")); /* * Ignore some errors if we are hooked up. * If connection hasn't completed, has retransmitted several times, * and receives a second error, give up now. This is better * than waiting a long time to establish a connection that * can never complete. */ if (tp->t_state == TCPS_ESTABLISHED && (error == EHOSTUNREACH || error == ENETUNREACH || error == EHOSTDOWN)) { return (inp); } else if (tp->t_state < TCPS_ESTABLISHED && tp->t_rxtshift > 3 && tp->t_softerror) { tp = tcp_drop(tp, error); if (tp != NULL) return (inp); else return (NULL); } else { tp->t_softerror = error; return (inp); } #if 0 wakeup( &so->so_timeo); sorwakeup(so); sowwakeup(so); #endif } static int tcp_pcblist(SYSCTL_HANDLER_ARGS) { int error, i, m, n, pcb_count; struct inpcb *inp, **inp_list; inp_gen_t gencnt; struct xinpgen xig; /* * The process of preparing the TCB list is too time-consuming and * resource-intensive to repeat twice on every request. */ if (req->oldptr == NULL) { n = V_tcbinfo.ipi_count + syncache_pcbcount(); n += imax(n / 8, 10); req->oldidx = 2 * (sizeof xig) + n * sizeof(struct xtcpcb); return (0); } if (req->newptr != NULL) return (EPERM); /* * OK, now we're committed to doing something. */ INP_INFO_RLOCK(&V_tcbinfo); gencnt = V_tcbinfo.ipi_gencnt; n = V_tcbinfo.ipi_count; INP_INFO_RUNLOCK(&V_tcbinfo); m = syncache_pcbcount(); error = sysctl_wire_old_buffer(req, 2 * (sizeof xig) + (n + m) * sizeof(struct xtcpcb)); if (error != 0) return (error); xig.xig_len = sizeof xig; xig.xig_count = n + m; xig.xig_gen = gencnt; xig.xig_sogen = so_gencnt; error = SYSCTL_OUT(req, &xig, sizeof xig); if (error) return (error); error = syncache_pcblist(req, m, &pcb_count); if (error) return (error); inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); if (inp_list == NULL) return (ENOMEM); INP_INFO_RLOCK(&V_tcbinfo); for (inp = LIST_FIRST(V_tcbinfo.ipi_listhead), i = 0; inp != NULL && i < n; inp = LIST_NEXT(inp, inp_list)) { INP_WLOCK(inp); if (inp->inp_gencnt <= gencnt) { /* * XXX: This use of cr_cansee(), introduced with * TCP state changes, is not quite right, but for * now, better than nothing. */ if (inp->inp_flags & INP_TIMEWAIT) { if (intotw(inp) != NULL) error = cr_cansee(req->td->td_ucred, intotw(inp)->tw_cred); else error = EINVAL; /* Skip this inp. */ } else error = cr_canseeinpcb(req->td->td_ucred, inp); if (error == 0) { in_pcbref(inp); inp_list[i++] = inp; } } INP_WUNLOCK(inp); } INP_INFO_RUNLOCK(&V_tcbinfo); n = i; error = 0; for (i = 0; i < n; i++) { inp = inp_list[i]; INP_RLOCK(inp); if (inp->inp_gencnt <= gencnt) { struct xtcpcb xt; void *inp_ppcb; bzero(&xt, sizeof(xt)); xt.xt_len = sizeof xt; /* XXX should avoid extra copy */ bcopy(inp, &xt.xt_inp, sizeof *inp); inp_ppcb = inp->inp_ppcb; if (inp_ppcb == NULL) bzero((char *) &xt.xt_tp, sizeof xt.xt_tp); else if (inp->inp_flags & INP_TIMEWAIT) { bzero((char *) &xt.xt_tp, sizeof xt.xt_tp); xt.xt_tp.t_state = TCPS_TIME_WAIT; } else { bcopy(inp_ppcb, &xt.xt_tp, sizeof xt.xt_tp); if (xt.xt_tp.t_timers) tcp_timer_to_xtimer(&xt.xt_tp, xt.xt_tp.t_timers, &xt.xt_timer); } if (inp->inp_socket != NULL) sotoxsocket(inp->inp_socket, &xt.xt_socket); else { bzero(&xt.xt_socket, sizeof xt.xt_socket); xt.xt_socket.xso_protocol = IPPROTO_TCP; } xt.xt_inp.inp_gencnt = inp->inp_gencnt; INP_RUNLOCK(inp); error = SYSCTL_OUT(req, &xt, sizeof xt); } else INP_RUNLOCK(inp); } INP_INFO_WLOCK(&V_tcbinfo); for (i = 0; i < n; i++) { inp = inp_list[i]; INP_RLOCK(inp); if (!in_pcbrele_rlocked(inp)) INP_RUNLOCK(inp); } INP_INFO_WUNLOCK(&V_tcbinfo); if (!error) { /* * Give the user an updated idea of our state. * If the generation differs from what we told * her before, she knows that something happened * while we were processing this request, and it * might be necessary to retry. */ INP_INFO_RLOCK(&V_tcbinfo); xig.xig_gen = V_tcbinfo.ipi_gencnt; xig.xig_sogen = so_gencnt; xig.xig_count = V_tcbinfo.ipi_count + pcb_count; INP_INFO_RUNLOCK(&V_tcbinfo); error = SYSCTL_OUT(req, &xig, sizeof xig); } free(inp_list, M_TEMP); return (error); } SYSCTL_PROC(_net_inet_tcp, TCPCTL_PCBLIST, pcblist, CTLTYPE_OPAQUE | CTLFLAG_RD, NULL, 0, tcp_pcblist, "S,xtcpcb", "List of active TCP connections"); #ifdef INET static int tcp_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in addrs[2]; struct inpcb *inp; int error; error = priv_check(req->td, PRIV_NETINET_GETCRED); if (error) return (error); error = SYSCTL_IN(req, addrs, sizeof(addrs)); if (error) return (error); inp = in_pcblookup(&V_tcbinfo, addrs[1].sin_addr, addrs[1].sin_port, addrs[0].sin_addr, addrs[0].sin_port, INPLOOKUP_RLOCKPCB, NULL); if (inp != NULL) { if (inp->inp_socket == NULL) error = ENOENT; if (error == 0) error = cr_canseeinpcb(req->td->td_ucred, inp); if (error == 0) cru2x(inp->inp_cred, &xuc); INP_RUNLOCK(inp); } else error = ENOENT; if (error == 0) error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred)); return (error); } SYSCTL_PROC(_net_inet_tcp, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW|CTLFLAG_PRISON, 0, 0, tcp_getcred, "S,xucred", "Get the xucred of a TCP connection"); #endif /* INET */ #ifdef INET6 static int tcp6_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in6 addrs[2]; struct inpcb *inp; int error; #ifdef INET int mapped = 0; #endif error = priv_check(req->td, PRIV_NETINET_GETCRED); if (error) return (error); error = SYSCTL_IN(req, addrs, sizeof(addrs)); if (error) return (error); if ((error = sa6_embedscope(&addrs[0], V_ip6_use_defzone)) != 0 || (error = sa6_embedscope(&addrs[1], V_ip6_use_defzone)) != 0) { return (error); } if (IN6_IS_ADDR_V4MAPPED(&addrs[0].sin6_addr)) { #ifdef INET if (IN6_IS_ADDR_V4MAPPED(&addrs[1].sin6_addr)) mapped = 1; else #endif return (EINVAL); } #ifdef INET if (mapped == 1) inp = in_pcblookup(&V_tcbinfo, *(struct in_addr *)&addrs[1].sin6_addr.s6_addr[12], addrs[1].sin6_port, *(struct in_addr *)&addrs[0].sin6_addr.s6_addr[12], addrs[0].sin6_port, INPLOOKUP_RLOCKPCB, NULL); else #endif inp = in6_pcblookup(&V_tcbinfo, &addrs[1].sin6_addr, addrs[1].sin6_port, &addrs[0].sin6_addr, addrs[0].sin6_port, INPLOOKUP_RLOCKPCB, NULL); if (inp != NULL) { if (inp->inp_socket == NULL) error = ENOENT; if (error == 0) error = cr_canseeinpcb(req->td->td_ucred, inp); if (error == 0) cru2x(inp->inp_cred, &xuc); INP_RUNLOCK(inp); } else error = ENOENT; if (error == 0) error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred)); return (error); } SYSCTL_PROC(_net_inet6_tcp6, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW|CTLFLAG_PRISON, 0, 0, tcp6_getcred, "S,xucred", "Get the xucred of a TCP6 connection"); #endif /* INET6 */ #ifdef INET void tcp_ctlinput(int cmd, struct sockaddr *sa, void *vip) { struct ip *ip = vip; struct tcphdr *th; struct in_addr faddr; struct inpcb *inp; struct tcpcb *tp; struct inpcb *(*notify)(struct inpcb *, int) = tcp_notify; struct icmp *icp; struct in_conninfo inc; tcp_seq icmp_tcp_seq; int mtu; faddr = ((struct sockaddr_in *)sa)->sin_addr; if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY) return; if (cmd == PRC_MSGSIZE) notify = tcp_mtudisc; else if (V_icmp_may_rst && (cmd == PRC_UNREACH_ADMIN_PROHIB || cmd == PRC_UNREACH_PORT || cmd == PRC_TIMXCEED_INTRANS) && ip) notify = tcp_drop_syn_sent; /* * Redirects don't need to be handled up here. */ else if (PRC_IS_REDIRECT(cmd)) return; /* * Source quench is depreciated. */ else if (cmd == PRC_QUENCH) return; /* * Hostdead is ugly because it goes linearly through all PCBs. * XXX: We never get this from ICMP, otherwise it makes an * excellent DoS attack on machines with many connections. */ else if (cmd == PRC_HOSTDEAD) ip = NULL; else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0) return; if (ip != NULL) { icp = (struct icmp *)((caddr_t)ip - offsetof(struct icmp, icmp_ip)); th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2)); INP_INFO_WLOCK(&V_tcbinfo); inp = in_pcblookup(&V_tcbinfo, faddr, th->th_dport, ip->ip_src, th->th_sport, INPLOOKUP_WLOCKPCB, NULL); if (inp != NULL) { if (!(inp->inp_flags & INP_TIMEWAIT) && !(inp->inp_flags & INP_DROPPED) && !(inp->inp_socket == NULL)) { icmp_tcp_seq = htonl(th->th_seq); tp = intotcpcb(inp); if (SEQ_GEQ(icmp_tcp_seq, tp->snd_una) && SEQ_LT(icmp_tcp_seq, tp->snd_max)) { if (cmd == PRC_MSGSIZE) { /* * MTU discovery: * If we got a needfrag set the MTU * in the route to the suggested new * value (if given) and then notify. */ bzero(&inc, sizeof(inc)); inc.inc_faddr = faddr; inc.inc_fibnum = inp->inp_inc.inc_fibnum; mtu = ntohs(icp->icmp_nextmtu); /* * If no alternative MTU was * proposed, try the next smaller * one. ip->ip_len has already * been swapped in icmp_input(). */ if (!mtu) mtu = ip_next_mtu(ip->ip_len, 1); if (mtu < V_tcp_minmss + sizeof(struct tcpiphdr)) mtu = V_tcp_minmss + sizeof(struct tcpiphdr); /* * Only cache the MTU if it * is smaller than the interface * or route MTU. tcp_mtudisc() * will do right thing by itself. */ if (mtu <= tcp_maxmtu(&inc, NULL)) tcp_hc_updatemtu(&inc, mtu); } inp = (*notify)(inp, inetctlerrmap[cmd]); } } if (inp != NULL) INP_WUNLOCK(inp); } else { bzero(&inc, sizeof(inc)); inc.inc_fport = th->th_dport; inc.inc_lport = th->th_sport; inc.inc_faddr = faddr; inc.inc_laddr = ip->ip_src; syncache_unreach(&inc, th); } INP_INFO_WUNLOCK(&V_tcbinfo); } else in_pcbnotifyall(&V_tcbinfo, faddr, inetctlerrmap[cmd], notify); } #endif /* INET */ #ifdef INET6 void tcp6_ctlinput(int cmd, struct sockaddr *sa, void *d) { struct tcphdr th; struct inpcb *(*notify)(struct inpcb *, int) = tcp_notify; struct ip6_hdr *ip6; struct mbuf *m; struct ip6ctlparam *ip6cp = NULL; const struct sockaddr_in6 *sa6_src = NULL; int off; struct tcp_portonly { u_int16_t th_sport; u_int16_t th_dport; } *thp; if (sa->sa_family != AF_INET6 || sa->sa_len != sizeof(struct sockaddr_in6)) return; if (cmd == PRC_MSGSIZE) notify = tcp_mtudisc; else if (!PRC_IS_REDIRECT(cmd) && ((unsigned)cmd >= PRC_NCMDS || inet6ctlerrmap[cmd] == 0)) return; /* Source quench is depreciated. */ else if (cmd == PRC_QUENCH) return; /* if the parameter is from icmp6, decode it. */ if (d != NULL) { ip6cp = (struct ip6ctlparam *)d; m = ip6cp->ip6c_m; ip6 = ip6cp->ip6c_ip6; off = ip6cp->ip6c_off; sa6_src = ip6cp->ip6c_src; } else { m = NULL; ip6 = NULL; off = 0; /* fool gcc */ sa6_src = &sa6_any; } if (ip6 != NULL) { struct in_conninfo inc; /* * XXX: We assume that when IPV6 is non NULL, * M and OFF are valid. */ /* check if we can safely examine src and dst ports */ if (m->m_pkthdr.len < off + sizeof(*thp)) return; bzero(&th, sizeof(th)); m_copydata(m, off, sizeof(*thp), (caddr_t)&th); in6_pcbnotify(&V_tcbinfo, sa, th.th_dport, (struct sockaddr *)ip6cp->ip6c_src, th.th_sport, cmd, NULL, notify); bzero(&inc, sizeof(inc)); inc.inc_fport = th.th_dport; inc.inc_lport = th.th_sport; inc.inc6_faddr = ((struct sockaddr_in6 *)sa)->sin6_addr; inc.inc6_laddr = ip6cp->ip6c_src->sin6_addr; inc.inc_flags |= INC_ISIPV6; INP_INFO_WLOCK(&V_tcbinfo); syncache_unreach(&inc, &th); INP_INFO_WUNLOCK(&V_tcbinfo); } else in6_pcbnotify(&V_tcbinfo, sa, 0, (const struct sockaddr *)sa6_src, 0, cmd, NULL, notify); } #endif /* INET6 */ /* * Following is where TCP initial sequence number generation occurs. * * There are two places where we must use initial sequence numbers: * 1. In SYN-ACK packets. * 2. In SYN packets. * * All ISNs for SYN-ACK packets are generated by the syncache. See * tcp_syncache.c for details. * * The ISNs in SYN packets must be monotonic; TIME_WAIT recycling * depends on this property. In addition, these ISNs should be * unguessable so as to prevent connection hijacking. To satisfy * the requirements of this situation, the algorithm outlined in * RFC 1948 is used, with only small modifications. * * Implementation details: * * Time is based off the system timer, and is corrected so that it * increases by one megabyte per second. This allows for proper * recycling on high speed LANs while still leaving over an hour * before rollover. * * As reading the *exact* system time is too expensive to be done * whenever setting up a TCP connection, we increment the time * offset in two ways. First, a small random positive increment * is added to isn_offset for each connection that is set up. * Second, the function tcp_isn_tick fires once per clock tick * and increments isn_offset as necessary so that sequence numbers * are incremented at approximately ISN_BYTES_PER_SECOND. The * random positive increments serve only to ensure that the same * exact sequence number is never sent out twice (as could otherwise * happen when a port is recycled in less than the system tick * interval.) * * net.inet.tcp.isn_reseed_interval controls the number of seconds * between seeding of isn_secret. This is normally set to zero, * as reseeding should not be necessary. * * Locking of the global variables isn_secret, isn_last_reseed, isn_offset, * isn_offset_old, and isn_ctx is performed using the TCP pcbinfo lock. In * general, this means holding an exclusive (write) lock. */ #define ISN_BYTES_PER_SECOND 1048576 #define ISN_STATIC_INCREMENT 4096 #define ISN_RANDOM_INCREMENT (4096 - 1) static VNET_DEFINE(u_char, isn_secret[32]); static VNET_DEFINE(int, isn_last); static VNET_DEFINE(int, isn_last_reseed); static VNET_DEFINE(u_int32_t, isn_offset); static VNET_DEFINE(u_int32_t, isn_offset_old); #define V_isn_secret VNET(isn_secret) #define V_isn_last VNET(isn_last) #define V_isn_last_reseed VNET(isn_last_reseed) #define V_isn_offset VNET(isn_offset) #define V_isn_offset_old VNET(isn_offset_old) tcp_seq tcp_new_isn(struct tcpcb *tp) { MD5_CTX isn_ctx; u_int32_t md5_buffer[4]; tcp_seq new_isn; u_int32_t projected_offset; INP_WLOCK_ASSERT(tp->t_inpcb); ISN_LOCK(); /* Seed if this is the first use, reseed if requested. */ if ((V_isn_last_reseed == 0) || ((V_tcp_isn_reseed_interval > 0) && (((u_int)V_isn_last_reseed + (u_int)V_tcp_isn_reseed_interval*hz) < (u_int)ticks))) { read_random(&V_isn_secret, sizeof(V_isn_secret)); V_isn_last_reseed = ticks; } /* Compute the md5 hash and return the ISN. */ MD5Init(&isn_ctx); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_fport, sizeof(u_short)); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_lport, sizeof(u_short)); #ifdef INET6 if ((tp->t_inpcb->inp_vflag & INP_IPV6) != 0) { MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->in6p_faddr, sizeof(struct in6_addr)); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->in6p_laddr, sizeof(struct in6_addr)); } else #endif { MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_faddr, sizeof(struct in_addr)); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_laddr, sizeof(struct in_addr)); } MD5Update(&isn_ctx, (u_char *) &V_isn_secret, sizeof(V_isn_secret)); MD5Final((u_char *) &md5_buffer, &isn_ctx); new_isn = (tcp_seq) md5_buffer[0]; V_isn_offset += ISN_STATIC_INCREMENT + (arc4random() & ISN_RANDOM_INCREMENT); if (ticks != V_isn_last) { projected_offset = V_isn_offset_old + ISN_BYTES_PER_SECOND / hz * (ticks - V_isn_last); if (SEQ_GT(projected_offset, V_isn_offset)) V_isn_offset = projected_offset; V_isn_offset_old = V_isn_offset; V_isn_last = ticks; } new_isn += V_isn_offset; ISN_UNLOCK(); return (new_isn); } /* * When a specific ICMP unreachable message is received and the * connection state is SYN-SENT, drop the connection. This behavior * is controlled by the icmp_may_rst sysctl. */ struct inpcb * tcp_drop_syn_sent(struct inpcb *inp, int errno) { struct tcpcb *tp; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(inp); if ((inp->inp_flags & INP_TIMEWAIT) || (inp->inp_flags & INP_DROPPED)) return (inp); tp = intotcpcb(inp); if (tp->t_state != TCPS_SYN_SENT) return (inp); tp = tcp_drop(tp, errno); if (tp != NULL) return (inp); else return (NULL); } /* * When `need fragmentation' ICMP is received, update our idea of the MSS * based on the new value in the route. Also nudge TCP to send something, * since we know the packet we just sent was dropped. * This duplicates some code in the tcp_mss() function in tcp_input.c. */ struct inpcb * tcp_mtudisc(struct inpcb *inp, int errno) { struct tcpcb *tp; struct socket *so; INP_WLOCK_ASSERT(inp); if ((inp->inp_flags & INP_TIMEWAIT) || (inp->inp_flags & INP_DROPPED)) return (inp); tp = intotcpcb(inp); KASSERT(tp != NULL, ("tcp_mtudisc: tp == NULL")); tcp_mss_update(tp, -1, NULL, NULL); so = inp->inp_socket; SOCKBUF_LOCK(&so->so_snd); /* If the mss is larger than the socket buffer, decrease the mss. */ if (so->so_snd.sb_hiwat < tp->t_maxseg) tp->t_maxseg = so->so_snd.sb_hiwat; SOCKBUF_UNLOCK(&so->so_snd); TCPSTAT_INC(tcps_mturesent); tp->t_rtttime = 0; tp->snd_nxt = tp->snd_una; tcp_free_sackholes(tp); tp->snd_recover = tp->snd_max; if (tp->t_flags & TF_SACK_PERMIT) EXIT_FASTRECOVERY(tp->t_flags); tcp_output_send(tp); return (inp); } #ifdef INET /* * Look-up the routing entry to the peer of this inpcb. If no route * is found and it cannot be allocated, then return 0. This routine * is called by TCP routines that access the rmx structure and by * tcp_mss_update to get the peer/interface MTU. */ u_long tcp_maxmtu(struct in_conninfo *inc, int *flags) { struct route sro; struct sockaddr_in *dst; struct ifnet *ifp; u_long maxmtu = 0; KASSERT(inc != NULL, ("tcp_maxmtu with NULL in_conninfo pointer")); bzero(&sro, sizeof(sro)); if (inc->inc_faddr.s_addr != INADDR_ANY) { dst = (struct sockaddr_in *)&sro.ro_dst; dst->sin_family = AF_INET; dst->sin_len = sizeof(*dst); dst->sin_addr = inc->inc_faddr; in_rtalloc_ign(&sro, 0, inc->inc_fibnum); } if (sro.ro_rt != NULL) { ifp = sro.ro_rt->rt_ifp; if (sro.ro_rt->rt_rmx.rmx_mtu == 0) maxmtu = ifp->if_mtu; else maxmtu = min(sro.ro_rt->rt_rmx.rmx_mtu, ifp->if_mtu); /* Report additional interface capabilities. */ if (flags != NULL) { if (ifp->if_capenable & IFCAP_TSO4 && ifp->if_hwassist & CSUM_TSO) *flags |= CSUM_TSO; } RTFREE(sro.ro_rt); } return (maxmtu); } #endif /* INET */ #ifdef INET6 u_long tcp_maxmtu6(struct in_conninfo *inc, int *flags) { struct route_in6 sro6; struct ifnet *ifp; u_long maxmtu = 0; KASSERT(inc != NULL, ("tcp_maxmtu6 with NULL in_conninfo pointer")); bzero(&sro6, sizeof(sro6)); if (!IN6_IS_ADDR_UNSPECIFIED(&inc->inc6_faddr)) { sro6.ro_dst.sin6_family = AF_INET6; sro6.ro_dst.sin6_len = sizeof(struct sockaddr_in6); sro6.ro_dst.sin6_addr = inc->inc6_faddr; rtalloc_ign((struct route *)&sro6, 0); } if (sro6.ro_rt != NULL) { ifp = sro6.ro_rt->rt_ifp; if (sro6.ro_rt->rt_rmx.rmx_mtu == 0) maxmtu = IN6_LINKMTU(sro6.ro_rt->rt_ifp); else maxmtu = min(sro6.ro_rt->rt_rmx.rmx_mtu, IN6_LINKMTU(sro6.ro_rt->rt_ifp)); /* Report additional interface capabilities. */ if (flags != NULL) { if (ifp->if_capenable & IFCAP_TSO6 && ifp->if_hwassist & CSUM_TSO) *flags |= CSUM_TSO; } RTFREE(sro6.ro_rt); } return (maxmtu); } #endif /* INET6 */ #ifdef IPSEC /* compute ESP/AH header size for TCP, including outer IP header. */ size_t ipsec_hdrsiz_tcp(struct tcpcb *tp) { struct inpcb *inp; struct mbuf *m; size_t hdrsiz; struct ip *ip; #ifdef INET6 struct ip6_hdr *ip6; #endif struct tcphdr *th; if ((tp == NULL) || ((inp = tp->t_inpcb) == NULL)) return (0); MGETHDR(m, M_DONTWAIT, MT_DATA); if (!m) return (0); #ifdef INET6 if ((inp->inp_vflag & INP_IPV6) != 0) { ip6 = mtod(m, struct ip6_hdr *); th = (struct tcphdr *)(ip6 + 1); m->m_pkthdr.len = m->m_len = sizeof(struct ip6_hdr) + sizeof(struct tcphdr); tcpip_fillheaders(inp, ip6, th); hdrsiz = ipsec_hdrsiz(m, IPSEC_DIR_OUTBOUND, inp); } else #endif /* INET6 */ { ip = mtod(m, struct ip *); th = (struct tcphdr *)(ip + 1); m->m_pkthdr.len = m->m_len = sizeof(struct tcpiphdr); tcpip_fillheaders(inp, ip, th); hdrsiz = ipsec_hdrsiz(m, IPSEC_DIR_OUTBOUND, inp); } m_free(m); return (hdrsiz); } #endif /* IPSEC */ #ifdef TCP_SIGNATURE /* * Callback function invoked by m_apply() to digest TCP segment data * contained within an mbuf chain. */ static int tcp_signature_apply(void *fstate, void *data, u_int len) { MD5Update(fstate, (u_char *)data, len); return (0); } /* * Compute TCP-MD5 hash of a TCP segment. (RFC2385) * * Parameters: * m pointer to head of mbuf chain * _unused * len length of TCP segment data, excluding options * optlen length of TCP segment options * buf pointer to storage for computed MD5 digest * direction direction of flow (IPSEC_DIR_INBOUND or OUTBOUND) * * We do this over ip, tcphdr, segment data, and the key in the SADB. * When called from tcp_input(), we can be sure that th_sum has been * zeroed out and verified already. * * Return 0 if successful, otherwise return -1. * * XXX The key is retrieved from the system's PF_KEY SADB, by keying a * search with the destination IP address, and a 'magic SPI' to be * determined by the application. This is hardcoded elsewhere to 1179 * right now. Another branch of this code exists which uses the SPD to * specify per-application flows but it is unstable. */ int tcp_signature_compute(struct mbuf *m, int _unused, int len, int optlen, u_char *buf, u_int direction) { union sockaddr_union dst; #ifdef INET struct ippseudo ippseudo; #endif MD5_CTX ctx; int doff; struct ip *ip; #ifdef INET struct ipovly *ipovly; #endif struct secasvar *sav; struct tcphdr *th; #ifdef INET6 struct ip6_hdr *ip6; struct in6_addr in6; char ip6buf[INET6_ADDRSTRLEN]; uint32_t plen; uint16_t nhdr; #endif u_short savecsum; KASSERT(m != NULL, ("NULL mbuf chain")); KASSERT(buf != NULL, ("NULL signature pointer")); /* Extract the destination from the IP header in the mbuf. */ bzero(&dst, sizeof(union sockaddr_union)); ip = mtod(m, struct ip *); #ifdef INET6 ip6 = NULL; /* Make the compiler happy. */ #endif switch (ip->ip_v) { #ifdef INET case IPVERSION: dst.sa.sa_len = sizeof(struct sockaddr_in); dst.sa.sa_family = AF_INET; dst.sin.sin_addr = (direction == IPSEC_DIR_INBOUND) ? ip->ip_src : ip->ip_dst; break; #endif #ifdef INET6 case (IPV6_VERSION >> 4): ip6 = mtod(m, struct ip6_hdr *); dst.sa.sa_len = sizeof(struct sockaddr_in6); dst.sa.sa_family = AF_INET6; dst.sin6.sin6_addr = (direction == IPSEC_DIR_INBOUND) ? ip6->ip6_src : ip6->ip6_dst; break; #endif default: return (EINVAL); /* NOTREACHED */ break; } /* Look up an SADB entry which matches the address of the peer. */ sav = KEY_ALLOCSA(&dst, IPPROTO_TCP, htonl(TCP_SIG_SPI)); if (sav == NULL) { ipseclog((LOG_ERR, "%s: SADB lookup failed for %s\n", __func__, (ip->ip_v == IPVERSION) ? inet_ntoa(dst.sin.sin_addr) : #ifdef INET6 (ip->ip_v == (IPV6_VERSION >> 4)) ? ip6_sprintf(ip6buf, &dst.sin6.sin6_addr) : #endif "(unsupported)")); return (EINVAL); } MD5Init(&ctx); /* * Step 1: Update MD5 hash with IP(v6) pseudo-header. * * XXX The ippseudo header MUST be digested in network byte order, * or else we'll fail the regression test. Assume all fields we've * been doing arithmetic on have been in host byte order. * XXX One cannot depend on ipovly->ih_len here. When called from * tcp_output(), the underlying ip_len member has not yet been set. */ switch (ip->ip_v) { #ifdef INET case IPVERSION: ipovly = (struct ipovly *)ip; ippseudo.ippseudo_src = ipovly->ih_src; ippseudo.ippseudo_dst = ipovly->ih_dst; ippseudo.ippseudo_pad = 0; ippseudo.ippseudo_p = IPPROTO_TCP; ippseudo.ippseudo_len = htons(len + sizeof(struct tcphdr) + optlen); MD5Update(&ctx, (char *)&ippseudo, sizeof(struct ippseudo)); th = (struct tcphdr *)((u_char *)ip + sizeof(struct ip)); doff = sizeof(struct ip) + sizeof(struct tcphdr) + optlen; break; #endif #ifdef INET6 /* * RFC 2385, 2.0 Proposal * For IPv6, the pseudo-header is as described in RFC 2460, namely the * 128-bit source IPv6 address, 128-bit destination IPv6 address, zero- * extended next header value (to form 32 bits), and 32-bit segment * length. * Note: Upper-Layer Packet Length comes before Next Header. */ case (IPV6_VERSION >> 4): in6 = ip6->ip6_src; in6_clearscope(&in6); MD5Update(&ctx, (char *)&in6, sizeof(struct in6_addr)); in6 = ip6->ip6_dst; in6_clearscope(&in6); MD5Update(&ctx, (char *)&in6, sizeof(struct in6_addr)); plen = htonl(len + sizeof(struct tcphdr) + optlen); MD5Update(&ctx, (char *)&plen, sizeof(uint32_t)); nhdr = 0; MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); nhdr = IPPROTO_TCP; MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); th = (struct tcphdr *)((u_char *)ip6 + sizeof(struct ip6_hdr)); doff = sizeof(struct ip6_hdr) + sizeof(struct tcphdr) + optlen; break; #endif default: return (EINVAL); /* NOTREACHED */ break; } /* * Step 2: Update MD5 hash with TCP header, excluding options. * The TCP checksum must be set to zero. */ savecsum = th->th_sum; th->th_sum = 0; MD5Update(&ctx, (char *)th, sizeof(struct tcphdr)); th->th_sum = savecsum; /* * Step 3: Update MD5 hash with TCP segment data. * Use m_apply() to avoid an early m_pullup(). */ if (len > 0) m_apply(m, doff, len, tcp_signature_apply, &ctx); /* * Step 4: Update MD5 hash with shared secret. */ MD5Update(&ctx, sav->key_auth->key_data, _KEYLEN(sav->key_auth)); MD5Final(buf, &ctx); key_sa_recordxfer(sav, m); KEY_FREESAV(&sav); return (0); } /* * Verify the TCP-MD5 hash of a TCP segment. (RFC2385) * * Parameters: * m pointer to head of mbuf chain * len length of TCP segment data, excluding options * optlen length of TCP segment options * buf pointer to storage for computed MD5 digest * direction direction of flow (IPSEC_DIR_INBOUND or OUTBOUND) * * Return 1 if successful, otherwise return 0. */ int tcp_signature_verify(struct mbuf *m, int off0, int tlen, int optlen, struct tcpopt *to, struct tcphdr *th, u_int tcpbflag) { char tmpdigest[TCP_SIGLEN]; if (tcp_sig_checksigs == 0) return (1); if ((tcpbflag & TF_SIGNATURE) == 0) { if ((to->to_flags & TOF_SIGNATURE) != 0) { /* * If this socket is not expecting signature but * the segment contains signature just fail. */ TCPSTAT_INC(tcps_sig_err_sigopt); TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } /* Signature is not expected, and not present in segment. */ return (1); } /* * If this socket is expecting signature but the segment does not * contain any just fail. */ if ((to->to_flags & TOF_SIGNATURE) == 0) { TCPSTAT_INC(tcps_sig_err_nosigopt); TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } if (tcp_signature_compute(m, off0, tlen, optlen, &tmpdigest[0], IPSEC_DIR_INBOUND) == -1) { TCPSTAT_INC(tcps_sig_err_buildsig); TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } if (bcmp(to->to_signature, &tmpdigest[0], TCP_SIGLEN) != 0) { TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } TCPSTAT_INC(tcps_sig_rcvgoodsig); return (1); } #endif /* TCP_SIGNATURE */ static int sysctl_drop(SYSCTL_HANDLER_ARGS) { /* addrs[0] is a foreign socket, addrs[1] is a local one. */ struct sockaddr_storage addrs[2]; struct inpcb *inp; struct tcpcb *tp; struct tcptw *tw; struct sockaddr_in *fin, *lin; #ifdef INET6 struct sockaddr_in6 *fin6, *lin6; #endif int error; inp = NULL; fin = lin = NULL; #ifdef INET6 fin6 = lin6 = NULL; #endif error = 0; if (req->oldptr != NULL || req->oldlen != 0) return (EINVAL); if (req->newptr == NULL) return (EPERM); if (req->newlen < sizeof(addrs)) return (ENOMEM); error = SYSCTL_IN(req, &addrs, sizeof(addrs)); if (error) return (error); switch (addrs[0].ss_family) { #ifdef INET6 case AF_INET6: fin6 = (struct sockaddr_in6 *)&addrs[0]; lin6 = (struct sockaddr_in6 *)&addrs[1]; if (fin6->sin6_len != sizeof(struct sockaddr_in6) || lin6->sin6_len != sizeof(struct sockaddr_in6)) return (EINVAL); if (IN6_IS_ADDR_V4MAPPED(&fin6->sin6_addr)) { if (!IN6_IS_ADDR_V4MAPPED(&lin6->sin6_addr)) return (EINVAL); in6_sin6_2_sin_in_sock((struct sockaddr *)&addrs[0]); in6_sin6_2_sin_in_sock((struct sockaddr *)&addrs[1]); fin = (struct sockaddr_in *)&addrs[0]; lin = (struct sockaddr_in *)&addrs[1]; break; } error = sa6_embedscope(fin6, V_ip6_use_defzone); if (error) return (error); error = sa6_embedscope(lin6, V_ip6_use_defzone); if (error) return (error); break; #endif #ifdef INET case AF_INET: fin = (struct sockaddr_in *)&addrs[0]; lin = (struct sockaddr_in *)&addrs[1]; if (fin->sin_len != sizeof(struct sockaddr_in) || lin->sin_len != sizeof(struct sockaddr_in)) return (EINVAL); break; #endif default: return (EINVAL); } INP_INFO_WLOCK(&V_tcbinfo); switch (addrs[0].ss_family) { #ifdef INET6 case AF_INET6: inp = in6_pcblookup(&V_tcbinfo, &fin6->sin6_addr, fin6->sin6_port, &lin6->sin6_addr, lin6->sin6_port, INPLOOKUP_WLOCKPCB, NULL); break; #endif #ifdef INET case AF_INET: inp = in_pcblookup(&V_tcbinfo, fin->sin_addr, fin->sin_port, lin->sin_addr, lin->sin_port, INPLOOKUP_WLOCKPCB, NULL); break; #endif } if (inp != NULL) { if (inp->inp_flags & INP_TIMEWAIT) { /* * XXXRW: There currently exists a state where an * inpcb is present, but its timewait state has been * discarded. For now, don't allow dropping of this * type of inpcb. */ tw = intotw(inp); if (tw != NULL) tcp_twclose(tw, 0); else INP_WUNLOCK(inp); } else if (!(inp->inp_flags & INP_DROPPED) && !(inp->inp_socket->so_options & SO_ACCEPTCONN)) { tp = intotcpcb(inp); tp = tcp_drop(tp, ECONNABORTED); if (tp != NULL) INP_WUNLOCK(inp); } else INP_WUNLOCK(inp); } else error = ESRCH; INP_INFO_WUNLOCK(&V_tcbinfo); return (error); } SYSCTL_PROC(_net_inet_tcp, TCPCTL_DROP, drop, CTLTYPE_STRUCT|CTLFLAG_WR|CTLFLAG_SKIP, NULL, 0, sysctl_drop, "", "Drop TCP connection"); /* * Generate a standardized TCP log line for use throughout the * tcp subsystem. Memory allocation is done with M_NOWAIT to * allow use in the interrupt context. * * NB: The caller MUST free(s, M_TCPLOG) the returned string. * NB: The function may return NULL if memory allocation failed. * * Due to header inclusion and ordering limitations the struct ip * and ip6_hdr pointers have to be passed as void pointers. */ char * tcp_log_vain(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr) { /* Is logging enabled? */ if (tcp_log_in_vain == 0) return (NULL); return (tcp_log_addr(inc, th, ip4hdr, ip6hdr)); } char * tcp_log_addrs(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr) { /* Is logging enabled? */ if (tcp_log_debug == 0) return (NULL); return (tcp_log_addr(inc, th, ip4hdr, ip6hdr)); } static char * tcp_log_addr(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr) { char *s, *sp; size_t size; struct ip *ip; #ifdef INET6 const struct ip6_hdr *ip6; ip6 = (const struct ip6_hdr *)ip6hdr; #endif /* INET6 */ ip = (struct ip *)ip4hdr; /* * The log line looks like this: * "TCP: [1.2.3.4]:50332 to [1.2.3.4]:80 tcpflags 0x2" */ size = sizeof("TCP: []:12345 to []:12345 tcpflags 0x2<>") + sizeof(PRINT_TH_FLAGS) + 1 + #ifdef INET6 2 * INET6_ADDRSTRLEN; #else 2 * INET_ADDRSTRLEN; #endif /* INET6 */ s = malloc(size, M_TCPLOG, M_ZERO|M_NOWAIT); if (s == NULL) return (NULL); strcat(s, "TCP: ["); sp = s + strlen(s); if (inc && ((inc->inc_flags & INC_ISIPV6) == 0)) { inet_ntoa_r(inc->inc_faddr, sp); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(inc->inc_fport)); sp = s + strlen(s); inet_ntoa_r(inc->inc_laddr, sp); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(inc->inc_lport)); #ifdef INET6 } else if (inc) { ip6_sprintf(sp, &inc->inc6_faddr); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(inc->inc_fport)); sp = s + strlen(s); ip6_sprintf(sp, &inc->inc6_laddr); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(inc->inc_lport)); } else if (ip6 && th) { ip6_sprintf(sp, &ip6->ip6_src); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(th->th_sport)); sp = s + strlen(s); ip6_sprintf(sp, &ip6->ip6_dst); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(th->th_dport)); #endif /* INET6 */ #ifdef INET } else if (ip && th) { inet_ntoa_r(ip->ip_src, sp); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(th->th_sport)); sp = s + strlen(s); inet_ntoa_r(ip->ip_dst, sp); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(th->th_dport)); #endif /* INET */ } else { free(s, M_TCPLOG); return (NULL); } sp = s + strlen(s); if (th) sprintf(sp, " tcpflags 0x%b", th->th_flags, PRINT_TH_FLAGS); if (*(s + size - 1) != '\0') panic("%s: string too long", __func__); return (s); } Index: projects/largeSMP/sys/netinet/tcp_syncache.c =================================================================== --- projects/largeSMP/sys/netinet/tcp_syncache.c (revision 222782) +++ projects/largeSMP/sys/netinet/tcp_syncache.c (revision 222783) @@ -1,1835 +1,1842 @@ /*- * Copyright (c) 2001 McAfee, Inc. * Copyright (c) 2006 Andre Oppermann, Internet Business Solutions AG * All rights reserved. * * This software was developed for the FreeBSD Project by Jonathan Lemon * and McAfee Research, the Security Research Division of McAfee, 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 "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" +#include "opt_pcbgroup.h" #include #include #include #include #include #include #include #include #include #include #include /* for proc0 declaration */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef INET6 #include #include #include #include #include #endif #include #include #include #include #include #include #include #ifdef INET6 #include #endif #ifdef IPSEC #include #ifdef INET6 #include #endif #include #endif /*IPSEC*/ #include #include static VNET_DEFINE(int, tcp_syncookies) = 1; #define V_tcp_syncookies VNET(tcp_syncookies) SYSCTL_VNET_INT(_net_inet_tcp, OID_AUTO, syncookies, CTLFLAG_RW, &VNET_NAME(tcp_syncookies), 0, "Use TCP SYN cookies if the syncache overflows"); static VNET_DEFINE(int, tcp_syncookiesonly) = 0; #define V_tcp_syncookiesonly VNET(tcp_syncookiesonly) SYSCTL_VNET_INT(_net_inet_tcp, OID_AUTO, syncookies_only, CTLFLAG_RW, &VNET_NAME(tcp_syncookiesonly), 0, "Use only TCP SYN cookies"); #ifdef TCP_OFFLOAD_DISABLE #define TOEPCB_ISSET(sc) (0) #else #define TOEPCB_ISSET(sc) ((sc)->sc_toepcb != NULL) #endif static void syncache_drop(struct syncache *, struct syncache_head *); static void syncache_free(struct syncache *); static void syncache_insert(struct syncache *, struct syncache_head *); struct syncache *syncache_lookup(struct in_conninfo *, struct syncache_head **); static int syncache_respond(struct syncache *); static struct socket *syncache_socket(struct syncache *, struct socket *, struct mbuf *m); static void syncache_timeout(struct syncache *sc, struct syncache_head *sch, int docallout); static void syncache_timer(void *); static void syncookie_generate(struct syncache_head *, struct syncache *, u_int32_t *); static struct syncache *syncookie_lookup(struct in_conninfo *, struct syncache_head *, struct syncache *, struct tcpopt *, struct tcphdr *, struct socket *); /* * Transmit the SYN,ACK fewer times than TCP_MAXRXTSHIFT specifies. * 3 retransmits corresponds to a timeout of 3 * (1 + 2 + 4 + 8) == 45 seconds, * the odds are that the user has given up attempting to connect by then. */ #define SYNCACHE_MAXREXMTS 3 /* Arbitrary values */ #define TCP_SYNCACHE_HASHSIZE 512 #define TCP_SYNCACHE_BUCKETLIMIT 30 static VNET_DEFINE(struct tcp_syncache, tcp_syncache); #define V_tcp_syncache VNET(tcp_syncache) SYSCTL_NODE(_net_inet_tcp, OID_AUTO, syncache, CTLFLAG_RW, 0, "TCP SYN cache"); SYSCTL_VNET_UINT(_net_inet_tcp_syncache, OID_AUTO, bucketlimit, CTLFLAG_RDTUN, &VNET_NAME(tcp_syncache.bucket_limit), 0, "Per-bucket hash limit for syncache"); SYSCTL_VNET_UINT(_net_inet_tcp_syncache, OID_AUTO, cachelimit, CTLFLAG_RDTUN, &VNET_NAME(tcp_syncache.cache_limit), 0, "Overall entry limit for syncache"); SYSCTL_VNET_UINT(_net_inet_tcp_syncache, OID_AUTO, count, CTLFLAG_RD, &VNET_NAME(tcp_syncache.cache_count), 0, "Current number of entries in syncache"); SYSCTL_VNET_UINT(_net_inet_tcp_syncache, OID_AUTO, hashsize, CTLFLAG_RDTUN, &VNET_NAME(tcp_syncache.hashsize), 0, "Size of TCP syncache hashtable"); SYSCTL_VNET_UINT(_net_inet_tcp_syncache, OID_AUTO, rexmtlimit, CTLFLAG_RW, &VNET_NAME(tcp_syncache.rexmt_limit), 0, "Limit on SYN/ACK retransmissions"); VNET_DEFINE(int, tcp_sc_rst_sock_fail) = 1; SYSCTL_VNET_INT(_net_inet_tcp_syncache, OID_AUTO, rst_on_sock_fail, CTLFLAG_RW, &VNET_NAME(tcp_sc_rst_sock_fail), 0, "Send reset on socket allocation failure"); static MALLOC_DEFINE(M_SYNCACHE, "syncache", "TCP syncache"); #define SYNCACHE_HASH(inc, mask) \ ((V_tcp_syncache.hash_secret ^ \ (inc)->inc_faddr.s_addr ^ \ ((inc)->inc_faddr.s_addr >> 16) ^ \ (inc)->inc_fport ^ (inc)->inc_lport) & mask) #define SYNCACHE_HASH6(inc, mask) \ ((V_tcp_syncache.hash_secret ^ \ (inc)->inc6_faddr.s6_addr32[0] ^ \ (inc)->inc6_faddr.s6_addr32[3] ^ \ (inc)->inc_fport ^ (inc)->inc_lport) & mask) #define ENDPTS_EQ(a, b) ( \ (a)->ie_fport == (b)->ie_fport && \ (a)->ie_lport == (b)->ie_lport && \ (a)->ie_faddr.s_addr == (b)->ie_faddr.s_addr && \ (a)->ie_laddr.s_addr == (b)->ie_laddr.s_addr \ ) #define ENDPTS6_EQ(a, b) (memcmp(a, b, sizeof(*a)) == 0) #define SCH_LOCK(sch) mtx_lock(&(sch)->sch_mtx) #define SCH_UNLOCK(sch) mtx_unlock(&(sch)->sch_mtx) #define SCH_LOCK_ASSERT(sch) mtx_assert(&(sch)->sch_mtx, MA_OWNED) /* * Requires the syncache entry to be already removed from the bucket list. */ static void syncache_free(struct syncache *sc) { if (sc->sc_ipopts) (void) m_free(sc->sc_ipopts); if (sc->sc_cred) crfree(sc->sc_cred); #ifdef MAC mac_syncache_destroy(&sc->sc_label); #endif uma_zfree(V_tcp_syncache.zone, sc); } void syncache_init(void) { int i; V_tcp_syncache.cache_count = 0; V_tcp_syncache.hashsize = TCP_SYNCACHE_HASHSIZE; V_tcp_syncache.bucket_limit = TCP_SYNCACHE_BUCKETLIMIT; V_tcp_syncache.rexmt_limit = SYNCACHE_MAXREXMTS; V_tcp_syncache.hash_secret = arc4random(); TUNABLE_INT_FETCH("net.inet.tcp.syncache.hashsize", &V_tcp_syncache.hashsize); TUNABLE_INT_FETCH("net.inet.tcp.syncache.bucketlimit", &V_tcp_syncache.bucket_limit); if (!powerof2(V_tcp_syncache.hashsize) || V_tcp_syncache.hashsize == 0) { printf("WARNING: syncache hash size is not a power of 2.\n"); V_tcp_syncache.hashsize = TCP_SYNCACHE_HASHSIZE; } V_tcp_syncache.hashmask = V_tcp_syncache.hashsize - 1; /* Set limits. */ V_tcp_syncache.cache_limit = V_tcp_syncache.hashsize * V_tcp_syncache.bucket_limit; TUNABLE_INT_FETCH("net.inet.tcp.syncache.cachelimit", &V_tcp_syncache.cache_limit); /* Allocate the hash table. */ V_tcp_syncache.hashbase = malloc(V_tcp_syncache.hashsize * sizeof(struct syncache_head), M_SYNCACHE, M_WAITOK | M_ZERO); /* Initialize the hash buckets. */ for (i = 0; i < V_tcp_syncache.hashsize; i++) { #ifdef VIMAGE V_tcp_syncache.hashbase[i].sch_vnet = curvnet; #endif TAILQ_INIT(&V_tcp_syncache.hashbase[i].sch_bucket); mtx_init(&V_tcp_syncache.hashbase[i].sch_mtx, "tcp_sc_head", NULL, MTX_DEF); callout_init_mtx(&V_tcp_syncache.hashbase[i].sch_timer, &V_tcp_syncache.hashbase[i].sch_mtx, 0); V_tcp_syncache.hashbase[i].sch_length = 0; } /* Create the syncache entry zone. */ V_tcp_syncache.zone = uma_zcreate("syncache", sizeof(struct syncache), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); uma_zone_set_max(V_tcp_syncache.zone, V_tcp_syncache.cache_limit); } #ifdef VIMAGE void syncache_destroy(void) { struct syncache_head *sch; struct syncache *sc, *nsc; int i; /* Cleanup hash buckets: stop timers, free entries, destroy locks. */ for (i = 0; i < V_tcp_syncache.hashsize; i++) { sch = &V_tcp_syncache.hashbase[i]; callout_drain(&sch->sch_timer); SCH_LOCK(sch); TAILQ_FOREACH_SAFE(sc, &sch->sch_bucket, sc_hash, nsc) syncache_drop(sc, sch); SCH_UNLOCK(sch); KASSERT(TAILQ_EMPTY(&sch->sch_bucket), ("%s: sch->sch_bucket not empty", __func__)); KASSERT(sch->sch_length == 0, ("%s: sch->sch_length %d not 0", __func__, sch->sch_length)); mtx_destroy(&sch->sch_mtx); } KASSERT(V_tcp_syncache.cache_count == 0, ("%s: cache_count %d not 0", __func__, V_tcp_syncache.cache_count)); /* Free the allocated global resources. */ uma_zdestroy(V_tcp_syncache.zone); free(V_tcp_syncache.hashbase, M_SYNCACHE); } #endif /* * Inserts a syncache entry into the specified bucket row. * Locks and unlocks the syncache_head autonomously. */ static void syncache_insert(struct syncache *sc, struct syncache_head *sch) { struct syncache *sc2; SCH_LOCK(sch); /* * Make sure that we don't overflow the per-bucket limit. * If the bucket is full, toss the oldest element. */ if (sch->sch_length >= V_tcp_syncache.bucket_limit) { KASSERT(!TAILQ_EMPTY(&sch->sch_bucket), ("sch->sch_length incorrect")); sc2 = TAILQ_LAST(&sch->sch_bucket, sch_head); syncache_drop(sc2, sch); TCPSTAT_INC(tcps_sc_bucketoverflow); } /* Put it into the bucket. */ TAILQ_INSERT_HEAD(&sch->sch_bucket, sc, sc_hash); sch->sch_length++; /* Reinitialize the bucket row's timer. */ if (sch->sch_length == 1) sch->sch_nextc = ticks + INT_MAX; syncache_timeout(sc, sch, 1); SCH_UNLOCK(sch); V_tcp_syncache.cache_count++; TCPSTAT_INC(tcps_sc_added); } /* * Remove and free entry from syncache bucket row. * Expects locked syncache head. */ static void syncache_drop(struct syncache *sc, struct syncache_head *sch) { SCH_LOCK_ASSERT(sch); TAILQ_REMOVE(&sch->sch_bucket, sc, sc_hash); sch->sch_length--; #ifndef TCP_OFFLOAD_DISABLE if (sc->sc_tu) sc->sc_tu->tu_syncache_event(TOE_SC_DROP, sc->sc_toepcb); #endif syncache_free(sc); V_tcp_syncache.cache_count--; } /* * Engage/reengage time on bucket row. */ static void syncache_timeout(struct syncache *sc, struct syncache_head *sch, int docallout) { sc->sc_rxttime = ticks + TCPTV_RTOBASE * (tcp_backoff[sc->sc_rxmits]); sc->sc_rxmits++; if (TSTMP_LT(sc->sc_rxttime, sch->sch_nextc)) { sch->sch_nextc = sc->sc_rxttime; if (docallout) callout_reset(&sch->sch_timer, sch->sch_nextc - ticks, syncache_timer, (void *)sch); } } /* * Walk the timer queues, looking for SYN,ACKs that need to be retransmitted. * If we have retransmitted an entry the maximum number of times, expire it. * One separate timer for each bucket row. */ static void syncache_timer(void *xsch) { struct syncache_head *sch = (struct syncache_head *)xsch; struct syncache *sc, *nsc; int tick = ticks; char *s; CURVNET_SET(sch->sch_vnet); /* NB: syncache_head has already been locked by the callout. */ SCH_LOCK_ASSERT(sch); /* * In the following cycle we may remove some entries and/or * advance some timeouts, so re-initialize the bucket timer. */ sch->sch_nextc = tick + INT_MAX; TAILQ_FOREACH_SAFE(sc, &sch->sch_bucket, sc_hash, nsc) { /* * We do not check if the listen socket still exists * and accept the case where the listen socket may be * gone by the time we resend the SYN/ACK. We do * not expect this to happens often. If it does, * then the RST will be sent by the time the remote * host does the SYN/ACK->ACK. */ if (TSTMP_GT(sc->sc_rxttime, tick)) { if (TSTMP_LT(sc->sc_rxttime, sch->sch_nextc)) sch->sch_nextc = sc->sc_rxttime; continue; } if (sc->sc_rxmits > V_tcp_syncache.rexmt_limit) { if ((s = tcp_log_addrs(&sc->sc_inc, NULL, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: Retransmits exhausted, " "giving up and removing syncache entry\n", s, __func__); free(s, M_TCPLOG); } syncache_drop(sc, sch); TCPSTAT_INC(tcps_sc_stale); continue; } if ((s = tcp_log_addrs(&sc->sc_inc, NULL, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: Response timeout, " "retransmitting (%u) SYN|ACK\n", s, __func__, sc->sc_rxmits); free(s, M_TCPLOG); } (void) syncache_respond(sc); TCPSTAT_INC(tcps_sc_retransmitted); syncache_timeout(sc, sch, 0); } if (!TAILQ_EMPTY(&(sch)->sch_bucket)) callout_reset(&(sch)->sch_timer, (sch)->sch_nextc - tick, syncache_timer, (void *)(sch)); CURVNET_RESTORE(); } /* * Find an entry in the syncache. * Returns always with locked syncache_head plus a matching entry or NULL. */ struct syncache * syncache_lookup(struct in_conninfo *inc, struct syncache_head **schp) { struct syncache *sc; struct syncache_head *sch; #ifdef INET6 if (inc->inc_flags & INC_ISIPV6) { sch = &V_tcp_syncache.hashbase[ SYNCACHE_HASH6(inc, V_tcp_syncache.hashmask)]; *schp = sch; SCH_LOCK(sch); /* Circle through bucket row to find matching entry. */ TAILQ_FOREACH(sc, &sch->sch_bucket, sc_hash) { if (ENDPTS6_EQ(&inc->inc_ie, &sc->sc_inc.inc_ie)) return (sc); } } else #endif { sch = &V_tcp_syncache.hashbase[ SYNCACHE_HASH(inc, V_tcp_syncache.hashmask)]; *schp = sch; SCH_LOCK(sch); /* Circle through bucket row to find matching entry. */ TAILQ_FOREACH(sc, &sch->sch_bucket, sc_hash) { #ifdef INET6 if (sc->sc_inc.inc_flags & INC_ISIPV6) continue; #endif if (ENDPTS_EQ(&inc->inc_ie, &sc->sc_inc.inc_ie)) return (sc); } } SCH_LOCK_ASSERT(*schp); return (NULL); /* always returns with locked sch */ } /* * This function is called when we get a RST for a * non-existent connection, so that we can see if the * connection is in the syn cache. If it is, zap it. */ void syncache_chkrst(struct in_conninfo *inc, struct tcphdr *th) { struct syncache *sc; struct syncache_head *sch; char *s = NULL; sc = syncache_lookup(inc, &sch); /* returns locked sch */ SCH_LOCK_ASSERT(sch); /* * Any RST to our SYN|ACK must not carry ACK, SYN or FIN flags. * See RFC 793 page 65, section SEGMENT ARRIVES. */ if (th->th_flags & (TH_ACK|TH_SYN|TH_FIN)) { if ((s = tcp_log_addrs(inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Spurious RST with ACK, SYN or " "FIN flag set, segment ignored\n", s, __func__); TCPSTAT_INC(tcps_badrst); goto done; } /* * No corresponding connection was found in syncache. * If syncookies are enabled and possibly exclusively * used, or we are under memory pressure, a valid RST * may not find a syncache entry. In that case we're * done and no SYN|ACK retransmissions will happen. * Otherwise the RST was misdirected or spoofed. */ if (sc == NULL) { if ((s = tcp_log_addrs(inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Spurious RST without matching " "syncache entry (possibly syncookie only), " "segment ignored\n", s, __func__); TCPSTAT_INC(tcps_badrst); goto done; } /* * If the RST bit is set, check the sequence number to see * if this is a valid reset segment. * RFC 793 page 37: * In all states except SYN-SENT, all reset (RST) segments * are validated by checking their SEQ-fields. A reset is * valid if its sequence number is in the window. * * The sequence number in the reset segment is normally an * echo of our outgoing acknowlegement numbers, but some hosts * send a reset with the sequence number at the rightmost edge * of our receive window, and we have to handle this case. */ if (SEQ_GEQ(th->th_seq, sc->sc_irs) && SEQ_LEQ(th->th_seq, sc->sc_irs + sc->sc_wnd)) { syncache_drop(sc, sch); if ((s = tcp_log_addrs(inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Our SYN|ACK was rejected, " "connection attempt aborted by remote endpoint\n", s, __func__); TCPSTAT_INC(tcps_sc_reset); } else { if ((s = tcp_log_addrs(inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: RST with invalid SEQ %u != " "IRS %u (+WND %u), segment ignored\n", s, __func__, th->th_seq, sc->sc_irs, sc->sc_wnd); TCPSTAT_INC(tcps_badrst); } done: if (s != NULL) free(s, M_TCPLOG); SCH_UNLOCK(sch); } void syncache_badack(struct in_conninfo *inc) { struct syncache *sc; struct syncache_head *sch; sc = syncache_lookup(inc, &sch); /* returns locked sch */ SCH_LOCK_ASSERT(sch); if (sc != NULL) { syncache_drop(sc, sch); TCPSTAT_INC(tcps_sc_badack); } SCH_UNLOCK(sch); } void syncache_unreach(struct in_conninfo *inc, struct tcphdr *th) { struct syncache *sc; struct syncache_head *sch; sc = syncache_lookup(inc, &sch); /* returns locked sch */ SCH_LOCK_ASSERT(sch); if (sc == NULL) goto done; /* If the sequence number != sc_iss, then it's a bogus ICMP msg */ if (ntohl(th->th_seq) != sc->sc_iss) goto done; /* * If we've rertransmitted 3 times and this is our second error, * we remove the entry. Otherwise, we allow it to continue on. * This prevents us from incorrectly nuking an entry during a * spurious network outage. * * See tcp_notify(). */ if ((sc->sc_flags & SCF_UNREACH) == 0 || sc->sc_rxmits < 3 + 1) { sc->sc_flags |= SCF_UNREACH; goto done; } syncache_drop(sc, sch); TCPSTAT_INC(tcps_sc_unreach); done: SCH_UNLOCK(sch); } /* * Build a new TCP socket structure from a syncache entry. */ static struct socket * syncache_socket(struct syncache *sc, struct socket *lso, struct mbuf *m) { struct inpcb *inp = NULL; struct socket *so; struct tcpcb *tp; int error; char *s; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); /* * Ok, create the full blown connection, and set things up * as they would have been set up if we had created the * connection when the SYN arrived. If we can't create * the connection, abort it. */ so = sonewconn(lso, SS_ISCONNECTED); if (so == NULL) { /* * Drop the connection; we will either send a RST or * have the peer retransmit its SYN again after its * RTO and try again. */ TCPSTAT_INC(tcps_listendrop); if ((s = tcp_log_addrs(&sc->sc_inc, NULL, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: Socket create failed " "due to limits or memory shortage\n", s, __func__); free(s, M_TCPLOG); } goto abort2; } #ifdef MAC mac_socketpeer_set_from_mbuf(m, so); #endif inp = sotoinpcb(so); inp->inp_inc.inc_fibnum = so->so_fibnum; INP_WLOCK(inp); INP_HASH_WLOCK(&V_tcbinfo); /* Insert new socket into PCB hash list. */ inp->inp_inc.inc_flags = sc->sc_inc.inc_flags; #ifdef INET6 if (sc->sc_inc.inc_flags & INC_ISIPV6) { inp->in6p_laddr = sc->sc_inc.inc6_laddr; } else { inp->inp_vflag &= ~INP_IPV6; inp->inp_vflag |= INP_IPV4; #endif inp->inp_laddr = sc->sc_inc.inc_laddr; #ifdef INET6 } #endif + + /* + * Install in the reservation hash table for now, but don't yet + * install a connection group since the full 4-tuple isn't yet + * configured. + */ inp->inp_lport = sc->sc_inc.inc_lport; - if ((error = in_pcbinshash(inp)) != 0) { + if ((error = in_pcbinshash_nopcbgroup(inp)) != 0) { /* * Undo the assignments above if we failed to * put the PCB on the hash lists. */ #ifdef INET6 if (sc->sc_inc.inc_flags & INC_ISIPV6) inp->in6p_laddr = in6addr_any; else #endif inp->inp_laddr.s_addr = INADDR_ANY; inp->inp_lport = 0; if ((s = tcp_log_addrs(&sc->sc_inc, NULL, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: in_pcbinshash failed " "with error %i\n", s, __func__, error); free(s, M_TCPLOG); } INP_HASH_WUNLOCK(&V_tcbinfo); goto abort; } #ifdef IPSEC /* Copy old policy into new socket's. */ if (ipsec_copy_policy(sotoinpcb(lso)->inp_sp, inp->inp_sp)) printf("syncache_socket: could not copy policy\n"); #endif #ifdef INET6 if (sc->sc_inc.inc_flags & INC_ISIPV6) { struct inpcb *oinp = sotoinpcb(lso); struct in6_addr laddr6; struct sockaddr_in6 sin6; /* * Inherit socket options from the listening socket. * Note that in6p_inputopts are not (and should not be) * copied, since it stores previously received options and is * used to detect if each new option is different than the * previous one and hence should be passed to a user. * If we copied in6p_inputopts, a user would not be able to * receive options just after calling the accept system call. */ inp->inp_flags |= oinp->inp_flags & INP_CONTROLOPTS; if (oinp->in6p_outputopts) inp->in6p_outputopts = ip6_copypktopts(oinp->in6p_outputopts, M_NOWAIT); sin6.sin6_family = AF_INET6; sin6.sin6_len = sizeof(sin6); sin6.sin6_addr = sc->sc_inc.inc6_faddr; sin6.sin6_port = sc->sc_inc.inc_fport; sin6.sin6_flowinfo = sin6.sin6_scope_id = 0; laddr6 = inp->in6p_laddr; if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) inp->in6p_laddr = sc->sc_inc.inc6_laddr; if ((error = in6_pcbconnect_mbuf(inp, (struct sockaddr *)&sin6, thread0.td_ucred, m)) != 0) { inp->in6p_laddr = laddr6; if ((s = tcp_log_addrs(&sc->sc_inc, NULL, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: in6_pcbconnect failed " "with error %i\n", s, __func__, error); free(s, M_TCPLOG); } INP_HASH_WUNLOCK(&V_tcbinfo); goto abort; } /* Override flowlabel from in6_pcbconnect. */ inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; inp->inp_flow |= sc->sc_flowlabel; } #endif /* INET6 */ #if defined(INET) && defined(INET6) else #endif #ifdef INET { struct in_addr laddr; struct sockaddr_in sin; inp->inp_options = (m) ? ip_srcroute(m) : NULL; if (inp->inp_options == NULL) { inp->inp_options = sc->sc_ipopts; sc->sc_ipopts = NULL; } sin.sin_family = AF_INET; sin.sin_len = sizeof(sin); sin.sin_addr = sc->sc_inc.inc_faddr; sin.sin_port = sc->sc_inc.inc_fport; bzero((caddr_t)sin.sin_zero, sizeof(sin.sin_zero)); laddr = inp->inp_laddr; if (inp->inp_laddr.s_addr == INADDR_ANY) inp->inp_laddr = sc->sc_inc.inc_laddr; if ((error = in_pcbconnect_mbuf(inp, (struct sockaddr *)&sin, thread0.td_ucred, m)) != 0) { inp->inp_laddr = laddr; if ((s = tcp_log_addrs(&sc->sc_inc, NULL, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: in_pcbconnect failed " "with error %i\n", s, __func__, error); free(s, M_TCPLOG); } INP_HASH_WUNLOCK(&V_tcbinfo); goto abort; } } #endif /* INET */ INP_HASH_WUNLOCK(&V_tcbinfo); tp = intotcpcb(inp); tp->t_state = TCPS_SYN_RECEIVED; tp->iss = sc->sc_iss; tp->irs = sc->sc_irs; tcp_rcvseqinit(tp); tcp_sendseqinit(tp); tp->snd_wl1 = sc->sc_irs; tp->snd_max = tp->iss + 1; tp->snd_nxt = tp->iss + 1; tp->rcv_up = sc->sc_irs + 1; tp->rcv_wnd = sc->sc_wnd; tp->rcv_adv += tp->rcv_wnd; tp->last_ack_sent = tp->rcv_nxt; tp->t_flags = sototcpcb(lso)->t_flags & (TF_NOPUSH|TF_NODELAY); if (sc->sc_flags & SCF_NOOPT) tp->t_flags |= TF_NOOPT; else { if (sc->sc_flags & SCF_WINSCALE) { tp->t_flags |= TF_REQ_SCALE|TF_RCVD_SCALE; tp->snd_scale = sc->sc_requested_s_scale; tp->request_r_scale = sc->sc_requested_r_scale; } if (sc->sc_flags & SCF_TIMESTAMP) { tp->t_flags |= TF_REQ_TSTMP|TF_RCVD_TSTMP; tp->ts_recent = sc->sc_tsreflect; tp->ts_recent_age = ticks; tp->ts_offset = sc->sc_tsoff; } #ifdef TCP_SIGNATURE if (sc->sc_flags & SCF_SIGNATURE) tp->t_flags |= TF_SIGNATURE; #endif if (sc->sc_flags & SCF_SACK) tp->t_flags |= TF_SACK_PERMIT; } if (sc->sc_flags & SCF_ECN) tp->t_flags |= TF_ECN_PERMIT; /* * Set up MSS and get cached values from tcp_hostcache. * This might overwrite some of the defaults we just set. */ tcp_mss(tp, sc->sc_peer_mss); /* * If the SYN,ACK was retransmitted, reset cwnd to 1 segment. * NB: sc_rxmits counts all SYN,ACK transmits, not just retransmits. */ if (sc->sc_rxmits > 1) tp->snd_cwnd = tp->t_maxseg; tcp_timer_activate(tp, TT_KEEP, tcp_keepinit); INP_WUNLOCK(inp); TCPSTAT_INC(tcps_accepts); return (so); abort: INP_WUNLOCK(inp); abort2: if (so != NULL) soabort(so); return (NULL); } /* * This function gets called when we receive an ACK for a * socket in the LISTEN state. We look up the connection * in the syncache, and if its there, we pull it out of * the cache and turn it into a full-blown connection in * the SYN-RECEIVED state. */ int syncache_expand(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, struct socket **lsop, struct mbuf *m) { struct syncache *sc; struct syncache_head *sch; struct syncache scs; char *s; /* * Global TCP locks are held because we manipulate the PCB lists * and create a new socket. */ INP_INFO_WLOCK_ASSERT(&V_tcbinfo); KASSERT((th->th_flags & (TH_RST|TH_ACK|TH_SYN)) == TH_ACK, ("%s: can handle only ACK", __func__)); sc = syncache_lookup(inc, &sch); /* returns locked sch */ SCH_LOCK_ASSERT(sch); if (sc == NULL) { /* * There is no syncache entry, so see if this ACK is * a returning syncookie. To do this, first: * A. See if this socket has had a syncache entry dropped in * the past. We don't want to accept a bogus syncookie * if we've never received a SYN. * B. check that the syncookie is valid. If it is, then * cobble up a fake syncache entry, and return. */ if (!V_tcp_syncookies) { SCH_UNLOCK(sch); if ((s = tcp_log_addrs(inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Spurious ACK, " "segment rejected (syncookies disabled)\n", s, __func__); goto failed; } bzero(&scs, sizeof(scs)); sc = syncookie_lookup(inc, sch, &scs, to, th, *lsop); SCH_UNLOCK(sch); if (sc == NULL) { if ((s = tcp_log_addrs(inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Segment failed " "SYNCOOKIE authentication, segment rejected " "(probably spoofed)\n", s, __func__); goto failed; } } else { /* Pull out the entry to unlock the bucket row. */ TAILQ_REMOVE(&sch->sch_bucket, sc, sc_hash); sch->sch_length--; V_tcp_syncache.cache_count--; SCH_UNLOCK(sch); } /* * Segment validation: * ACK must match our initial sequence number + 1 (the SYN|ACK). */ if (th->th_ack != sc->sc_iss + 1 && !TOEPCB_ISSET(sc)) { if ((s = tcp_log_addrs(inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: ACK %u != ISS+1 %u, segment " "rejected\n", s, __func__, th->th_ack, sc->sc_iss); goto failed; } /* * The SEQ must fall in the window starting at the received * initial receive sequence number + 1 (the SYN). */ if ((SEQ_LEQ(th->th_seq, sc->sc_irs) || SEQ_GT(th->th_seq, sc->sc_irs + sc->sc_wnd)) && !TOEPCB_ISSET(sc)) { if ((s = tcp_log_addrs(inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: SEQ %u != IRS+1 %u, segment " "rejected\n", s, __func__, th->th_seq, sc->sc_irs); goto failed; } if (!(sc->sc_flags & SCF_TIMESTAMP) && (to->to_flags & TOF_TS)) { if ((s = tcp_log_addrs(inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Timestamp not expected, " "segment rejected\n", s, __func__); goto failed; } /* * If timestamps were negotiated the reflected timestamp * must be equal to what we actually sent in the SYN|ACK. */ if ((to->to_flags & TOF_TS) && to->to_tsecr != sc->sc_ts && !TOEPCB_ISSET(sc)) { if ((s = tcp_log_addrs(inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: TSECR %u != TS %u, " "segment rejected\n", s, __func__, to->to_tsecr, sc->sc_ts); goto failed; } *lsop = syncache_socket(sc, *lsop, m); if (*lsop == NULL) TCPSTAT_INC(tcps_sc_aborted); else TCPSTAT_INC(tcps_sc_completed); /* how do we find the inp for the new socket? */ if (sc != &scs) syncache_free(sc); return (1); failed: if (sc != NULL && sc != &scs) syncache_free(sc); if (s != NULL) free(s, M_TCPLOG); *lsop = NULL; return (0); } int tcp_offload_syncache_expand(struct in_conninfo *inc, struct toeopt *toeo, struct tcphdr *th, struct socket **lsop, struct mbuf *m) { struct tcpopt to; int rc; bzero(&to, sizeof(struct tcpopt)); to.to_mss = toeo->to_mss; to.to_wscale = toeo->to_wscale; to.to_flags = toeo->to_flags; INP_INFO_WLOCK(&V_tcbinfo); rc = syncache_expand(inc, &to, th, lsop, m); INP_INFO_WUNLOCK(&V_tcbinfo); return (rc); } /* * Given a LISTEN socket and an inbound SYN request, add * this to the syn cache, and send back a segment: * * to the source. * * IMPORTANT NOTE: We do _NOT_ ACK data that might accompany the SYN. * Doing so would require that we hold onto the data and deliver it * to the application. However, if we are the target of a SYN-flood * DoS attack, an attacker could send data which would eventually * consume all available buffer space if it were ACKed. By not ACKing * the data, we avoid this DoS scenario. */ static void _syncache_add(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, struct inpcb *inp, struct socket **lsop, struct mbuf *m, struct toe_usrreqs *tu, void *toepcb) { struct tcpcb *tp; struct socket *so; struct syncache *sc = NULL; struct syncache_head *sch; struct mbuf *ipopts = NULL; u_int32_t flowtmp; u_int ltflags; int win, sb_hiwat, ip_ttl, ip_tos; char *s; #ifdef INET6 int autoflowlabel = 0; #endif #ifdef MAC struct label *maclabel; #endif struct syncache scs; struct ucred *cred; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(inp); /* listen socket */ KASSERT((th->th_flags & (TH_RST|TH_ACK|TH_SYN)) == TH_SYN, ("%s: unexpected tcp flags", __func__)); /* * Combine all so/tp operations very early to drop the INP lock as * soon as possible. */ so = *lsop; tp = sototcpcb(so); cred = crhold(so->so_cred); #ifdef INET6 if ((inc->inc_flags & INC_ISIPV6) && (inp->inp_flags & IN6P_AUTOFLOWLABEL)) autoflowlabel = 1; #endif ip_ttl = inp->inp_ip_ttl; ip_tos = inp->inp_ip_tos; win = sbspace(&so->so_rcv); sb_hiwat = so->so_rcv.sb_hiwat; ltflags = (tp->t_flags & (TF_NOOPT | TF_SIGNATURE)); /* By the time we drop the lock these should no longer be used. */ so = NULL; tp = NULL; #ifdef MAC if (mac_syncache_init(&maclabel) != 0) { INP_WUNLOCK(inp); INP_INFO_WUNLOCK(&V_tcbinfo); goto done; } else mac_syncache_create(maclabel, inp); #endif INP_WUNLOCK(inp); INP_INFO_WUNLOCK(&V_tcbinfo); /* * Remember the IP options, if any. */ #ifdef INET6 if (!(inc->inc_flags & INC_ISIPV6)) #endif #ifdef INET ipopts = (m) ? ip_srcroute(m) : NULL; #else ipopts = NULL; #endif /* * See if we already have an entry for this connection. * If we do, resend the SYN,ACK, and reset the retransmit timer. * * XXX: should the syncache be re-initialized with the contents * of the new SYN here (which may have different options?) * * XXX: We do not check the sequence number to see if this is a * real retransmit or a new connection attempt. The question is * how to handle such a case; either ignore it as spoofed, or * drop the current entry and create a new one? */ sc = syncache_lookup(inc, &sch); /* returns locked entry */ SCH_LOCK_ASSERT(sch); if (sc != NULL) { #ifndef TCP_OFFLOAD_DISABLE if (sc->sc_tu) sc->sc_tu->tu_syncache_event(TOE_SC_ENTRY_PRESENT, sc->sc_toepcb); #endif TCPSTAT_INC(tcps_sc_dupsyn); if (ipopts) { /* * If we were remembering a previous source route, * forget it and use the new one we've been given. */ if (sc->sc_ipopts) (void) m_free(sc->sc_ipopts); sc->sc_ipopts = ipopts; } /* * Update timestamp if present. */ if ((sc->sc_flags & SCF_TIMESTAMP) && (to->to_flags & TOF_TS)) sc->sc_tsreflect = to->to_tsval; else sc->sc_flags &= ~SCF_TIMESTAMP; #ifdef MAC /* * Since we have already unconditionally allocated label * storage, free it up. The syncache entry will already * have an initialized label we can use. */ mac_syncache_destroy(&maclabel); #endif /* Retransmit SYN|ACK and reset retransmit count. */ if ((s = tcp_log_addrs(&sc->sc_inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: Received duplicate SYN, " "resetting timer and retransmitting SYN|ACK\n", s, __func__); free(s, M_TCPLOG); } if (!TOEPCB_ISSET(sc) && syncache_respond(sc) == 0) { sc->sc_rxmits = 0; syncache_timeout(sc, sch, 1); TCPSTAT_INC(tcps_sndacks); TCPSTAT_INC(tcps_sndtotal); } SCH_UNLOCK(sch); goto done; } sc = uma_zalloc(V_tcp_syncache.zone, M_NOWAIT | M_ZERO); if (sc == NULL) { /* * The zone allocator couldn't provide more entries. * Treat this as if the cache was full; drop the oldest * entry and insert the new one. */ TCPSTAT_INC(tcps_sc_zonefail); if ((sc = TAILQ_LAST(&sch->sch_bucket, sch_head)) != NULL) syncache_drop(sc, sch); sc = uma_zalloc(V_tcp_syncache.zone, M_NOWAIT | M_ZERO); if (sc == NULL) { if (V_tcp_syncookies) { bzero(&scs, sizeof(scs)); sc = &scs; } else { SCH_UNLOCK(sch); if (ipopts) (void) m_free(ipopts); goto done; } } } /* * Fill in the syncache values. */ #ifdef MAC sc->sc_label = maclabel; #endif sc->sc_cred = cred; cred = NULL; sc->sc_ipopts = ipopts; bcopy(inc, &sc->sc_inc, sizeof(struct in_conninfo)); #ifdef INET6 if (!(inc->inc_flags & INC_ISIPV6)) #endif { sc->sc_ip_tos = ip_tos; sc->sc_ip_ttl = ip_ttl; } #ifndef TCP_OFFLOAD_DISABLE sc->sc_tu = tu; sc->sc_toepcb = toepcb; #endif sc->sc_irs = th->th_seq; sc->sc_iss = arc4random(); sc->sc_flags = 0; sc->sc_flowlabel = 0; /* * Initial receive window: clip sbspace to [0 .. TCP_MAXWIN]. * win was derived from socket earlier in the function. */ win = imax(win, 0); win = imin(win, TCP_MAXWIN); sc->sc_wnd = win; if (V_tcp_do_rfc1323) { /* * A timestamp received in a SYN makes * it ok to send timestamp requests and replies. */ if (to->to_flags & TOF_TS) { sc->sc_tsreflect = to->to_tsval; sc->sc_ts = ticks; sc->sc_flags |= SCF_TIMESTAMP; } if (to->to_flags & TOF_SCALE) { int wscale = 0; /* * Pick the smallest possible scaling factor that * will still allow us to scale up to sb_max, aka * kern.ipc.maxsockbuf. * * We do this because there are broken firewalls that * will corrupt the window scale option, leading to * the other endpoint believing that our advertised * window is unscaled. At scale factors larger than * 5 the unscaled window will drop below 1500 bytes, * leading to serious problems when traversing these * broken firewalls. * * With the default maxsockbuf of 256K, a scale factor * of 3 will be chosen by this algorithm. Those who * choose a larger maxsockbuf should watch out * for the compatiblity problems mentioned above. * * RFC1323: The Window field in a SYN (i.e., a * or ) segment itself is never scaled. */ while (wscale < TCP_MAX_WINSHIFT && (TCP_MAXWIN << wscale) < sb_max) wscale++; sc->sc_requested_r_scale = wscale; sc->sc_requested_s_scale = to->to_wscale; sc->sc_flags |= SCF_WINSCALE; } } #ifdef TCP_SIGNATURE /* * If listening socket requested TCP digests, and received SYN * contains the option, flag this in the syncache so that * syncache_respond() will do the right thing with the SYN+ACK. * XXX: Currently we always record the option by default and will * attempt to use it in syncache_respond(). */ if (to->to_flags & TOF_SIGNATURE || ltflags & TF_SIGNATURE) sc->sc_flags |= SCF_SIGNATURE; #endif if (to->to_flags & TOF_SACKPERM) sc->sc_flags |= SCF_SACK; if (to->to_flags & TOF_MSS) sc->sc_peer_mss = to->to_mss; /* peer mss may be zero */ if (ltflags & TF_NOOPT) sc->sc_flags |= SCF_NOOPT; if ((th->th_flags & (TH_ECE|TH_CWR)) && V_tcp_do_ecn) sc->sc_flags |= SCF_ECN; if (V_tcp_syncookies) { syncookie_generate(sch, sc, &flowtmp); #ifdef INET6 if (autoflowlabel) sc->sc_flowlabel = flowtmp; #endif } else { #ifdef INET6 if (autoflowlabel) sc->sc_flowlabel = (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK); #endif } SCH_UNLOCK(sch); /* * Do a standard 3-way handshake. */ if (TOEPCB_ISSET(sc) || syncache_respond(sc) == 0) { if (V_tcp_syncookies && V_tcp_syncookiesonly && sc != &scs) syncache_free(sc); else if (sc != &scs) syncache_insert(sc, sch); /* locks and unlocks sch */ TCPSTAT_INC(tcps_sndacks); TCPSTAT_INC(tcps_sndtotal); } else { if (sc != &scs) syncache_free(sc); TCPSTAT_INC(tcps_sc_dropped); } done: if (cred != NULL) crfree(cred); #ifdef MAC if (sc == &scs) mac_syncache_destroy(&maclabel); #endif if (m) { *lsop = NULL; m_freem(m); } } static int syncache_respond(struct syncache *sc) { struct ip *ip = NULL; struct mbuf *m; struct tcphdr *th = NULL; int optlen, error = 0; /* Make compiler happy */ u_int16_t hlen, tlen, mssopt; struct tcpopt to; #ifdef INET6 struct ip6_hdr *ip6 = NULL; #endif hlen = #ifdef INET6 (sc->sc_inc.inc_flags & INC_ISIPV6) ? sizeof(struct ip6_hdr) : #endif sizeof(struct ip); tlen = hlen + sizeof(struct tcphdr); /* Determine MSS we advertize to other end of connection. */ mssopt = tcp_mssopt(&sc->sc_inc); if (sc->sc_peer_mss) mssopt = max( min(sc->sc_peer_mss, mssopt), V_tcp_minmss); /* XXX: Assume that the entire packet will fit in a header mbuf. */ KASSERT(max_linkhdr + tlen + TCP_MAXOLEN <= MHLEN, ("syncache: mbuf too small")); /* Create the IP+TCP header from scratch. */ m = m_gethdr(M_DONTWAIT, MT_DATA); if (m == NULL) return (ENOBUFS); #ifdef MAC mac_syncache_create_mbuf(sc->sc_label, m); #endif m->m_data += max_linkhdr; m->m_len = tlen; m->m_pkthdr.len = tlen; m->m_pkthdr.rcvif = NULL; #ifdef INET6 if (sc->sc_inc.inc_flags & INC_ISIPV6) { ip6 = mtod(m, struct ip6_hdr *); ip6->ip6_vfc = IPV6_VERSION; ip6->ip6_nxt = IPPROTO_TCP; ip6->ip6_src = sc->sc_inc.inc6_laddr; ip6->ip6_dst = sc->sc_inc.inc6_faddr; ip6->ip6_plen = htons(tlen - hlen); /* ip6_hlim is set after checksum */ ip6->ip6_flow &= ~IPV6_FLOWLABEL_MASK; ip6->ip6_flow |= sc->sc_flowlabel; th = (struct tcphdr *)(ip6 + 1); } #endif #if defined(INET6) && defined(INET) else #endif #ifdef INET { ip = mtod(m, struct ip *); ip->ip_v = IPVERSION; ip->ip_hl = sizeof(struct ip) >> 2; ip->ip_len = tlen; ip->ip_id = 0; ip->ip_off = 0; ip->ip_sum = 0; ip->ip_p = IPPROTO_TCP; ip->ip_src = sc->sc_inc.inc_laddr; ip->ip_dst = sc->sc_inc.inc_faddr; ip->ip_ttl = sc->sc_ip_ttl; ip->ip_tos = sc->sc_ip_tos; /* * See if we should do MTU discovery. Route lookups are * expensive, so we will only unset the DF bit if: * * 1) path_mtu_discovery is disabled * 2) the SCF_UNREACH flag has been set */ if (V_path_mtu_discovery && ((sc->sc_flags & SCF_UNREACH) == 0)) ip->ip_off |= IP_DF; th = (struct tcphdr *)(ip + 1); } #endif /* INET */ th->th_sport = sc->sc_inc.inc_lport; th->th_dport = sc->sc_inc.inc_fport; th->th_seq = htonl(sc->sc_iss); th->th_ack = htonl(sc->sc_irs + 1); th->th_off = sizeof(struct tcphdr) >> 2; th->th_x2 = 0; th->th_flags = TH_SYN|TH_ACK; th->th_win = htons(sc->sc_wnd); th->th_urp = 0; if (sc->sc_flags & SCF_ECN) { th->th_flags |= TH_ECE; TCPSTAT_INC(tcps_ecn_shs); } /* Tack on the TCP options. */ if ((sc->sc_flags & SCF_NOOPT) == 0) { to.to_flags = 0; to.to_mss = mssopt; to.to_flags = TOF_MSS; if (sc->sc_flags & SCF_WINSCALE) { to.to_wscale = sc->sc_requested_r_scale; to.to_flags |= TOF_SCALE; } if (sc->sc_flags & SCF_TIMESTAMP) { /* Virgin timestamp or TCP cookie enhanced one. */ to.to_tsval = sc->sc_ts; to.to_tsecr = sc->sc_tsreflect; to.to_flags |= TOF_TS; } if (sc->sc_flags & SCF_SACK) to.to_flags |= TOF_SACKPERM; #ifdef TCP_SIGNATURE if (sc->sc_flags & SCF_SIGNATURE) to.to_flags |= TOF_SIGNATURE; #endif optlen = tcp_addoptions(&to, (u_char *)(th + 1)); /* Adjust headers by option size. */ th->th_off = (sizeof(struct tcphdr) + optlen) >> 2; m->m_len += optlen; m->m_pkthdr.len += optlen; #ifdef TCP_SIGNATURE if (sc->sc_flags & SCF_SIGNATURE) tcp_signature_compute(m, 0, 0, optlen, to.to_signature, IPSEC_DIR_OUTBOUND); #endif #ifdef INET6 if (sc->sc_inc.inc_flags & INC_ISIPV6) ip6->ip6_plen = htons(ntohs(ip6->ip6_plen) + optlen); else #endif ip->ip_len += optlen; } else optlen = 0; M_SETFIB(m, sc->sc_inc.inc_fibnum); #ifdef INET6 if (sc->sc_inc.inc_flags & INC_ISIPV6) { th->th_sum = 0; th->th_sum = in6_cksum(m, IPPROTO_TCP, hlen, tlen + optlen - hlen); ip6->ip6_hlim = in6_selecthlim(NULL, NULL); error = ip6_output(m, NULL, NULL, 0, NULL, NULL, NULL); } #endif #if defined(INET6) && defined(INET) else #endif #ifdef INET { th->th_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htons(tlen + optlen - hlen + IPPROTO_TCP)); m->m_pkthdr.csum_flags = CSUM_TCP; m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum); error = ip_output(m, sc->sc_ipopts, NULL, 0, NULL, NULL); } #endif return (error); } void syncache_add(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, struct inpcb *inp, struct socket **lsop, struct mbuf *m) { _syncache_add(inc, to, th, inp, lsop, m, NULL, NULL); } void tcp_offload_syncache_add(struct in_conninfo *inc, struct toeopt *toeo, struct tcphdr *th, struct inpcb *inp, struct socket **lsop, struct toe_usrreqs *tu, void *toepcb) { struct tcpopt to; bzero(&to, sizeof(struct tcpopt)); to.to_mss = toeo->to_mss; to.to_wscale = toeo->to_wscale; to.to_flags = toeo->to_flags; INP_INFO_WLOCK(&V_tcbinfo); INP_WLOCK(inp); _syncache_add(inc, &to, th, inp, lsop, NULL, tu, toepcb); } /* * The purpose of SYN cookies is to avoid keeping track of all SYN's we * receive and to be able to handle SYN floods from bogus source addresses * (where we will never receive any reply). SYN floods try to exhaust all * our memory and available slots in the SYN cache table to cause a denial * of service to legitimate users of the local host. * * The idea of SYN cookies is to encode and include all necessary information * about the connection setup state within the SYN-ACK we send back and thus * to get along without keeping any local state until the ACK to the SYN-ACK * arrives (if ever). Everything we need to know should be available from * the information we encoded in the SYN-ACK. * * More information about the theory behind SYN cookies and its first * discussion and specification can be found at: * http://cr.yp.to/syncookies.html (overview) * http://cr.yp.to/syncookies/archive (gory details) * * This implementation extends the orginal idea and first implementation * of FreeBSD by using not only the initial sequence number field to store * information but also the timestamp field if present. This way we can * keep track of the entire state we need to know to recreate the session in * its original form. Almost all TCP speakers implement RFC1323 timestamps * these days. For those that do not we still have to live with the known * shortcomings of the ISN only SYN cookies. * * Cookie layers: * * Initial sequence number we send: * 31|................................|0 * DDDDDDDDDDDDDDDDDDDDDDDDDMMMRRRP * D = MD5 Digest (first dword) * M = MSS index * R = Rotation of secret * P = Odd or Even secret * * The MD5 Digest is computed with over following parameters: * a) randomly rotated secret * b) struct in_conninfo containing the remote/local ip/port (IPv4&IPv6) * c) the received initial sequence number from remote host * d) the rotation offset and odd/even bit * * Timestamp we send: * 31|................................|0 * DDDDDDDDDDDDDDDDDDDDDDSSSSRRRRA5 * D = MD5 Digest (third dword) (only as filler) * S = Requested send window scale * R = Requested receive window scale * A = SACK allowed * 5 = TCP-MD5 enabled (not implemented yet) * XORed with MD5 Digest (forth dword) * * The timestamp isn't cryptographically secure and doesn't need to be. * The double use of the MD5 digest dwords ties it to a specific remote/ * local host/port, remote initial sequence number and our local time * limited secret. A received timestamp is reverted (XORed) and then * the contained MD5 dword is compared to the computed one to ensure the * timestamp belongs to the SYN-ACK we sent. The other parameters may * have been tampered with but this isn't different from supplying bogus * values in the SYN in the first place. * * Some problems with SYN cookies remain however: * Consider the problem of a recreated (and retransmitted) cookie. If the * original SYN was accepted, the connection is established. The second * SYN is inflight, and if it arrives with an ISN that falls within the * receive window, the connection is killed. * * Notes: * A heuristic to determine when to accept syn cookies is not necessary. * An ACK flood would cause the syncookie verification to be attempted, * but a SYN flood causes syncookies to be generated. Both are of equal * cost, so there's no point in trying to optimize the ACK flood case. * Also, if you don't process certain ACKs for some reason, then all someone * would have to do is launch a SYN and ACK flood at the same time, which * would stop cookie verification and defeat the entire purpose of syncookies. */ static int tcp_sc_msstab[] = { 0, 256, 468, 536, 996, 1452, 1460, 8960 }; static void syncookie_generate(struct syncache_head *sch, struct syncache *sc, u_int32_t *flowlabel) { MD5_CTX ctx; u_int32_t md5_buffer[MD5_DIGEST_LENGTH / sizeof(u_int32_t)]; u_int32_t data; u_int32_t *secbits; u_int off, pmss, mss; int i; SCH_LOCK_ASSERT(sch); /* Which of the two secrets to use. */ secbits = sch->sch_oddeven ? sch->sch_secbits_odd : sch->sch_secbits_even; /* Reseed secret if too old. */ if (sch->sch_reseed < time_uptime) { sch->sch_oddeven = sch->sch_oddeven ? 0 : 1; /* toggle */ secbits = sch->sch_oddeven ? sch->sch_secbits_odd : sch->sch_secbits_even; for (i = 0; i < SYNCOOKIE_SECRET_SIZE; i++) secbits[i] = arc4random(); sch->sch_reseed = time_uptime + SYNCOOKIE_LIFETIME; } /* Secret rotation offset. */ off = sc->sc_iss & 0x7; /* iss was randomized before */ /* Maximum segment size calculation. */ pmss = max( min(sc->sc_peer_mss, tcp_mssopt(&sc->sc_inc)), V_tcp_minmss); for (mss = sizeof(tcp_sc_msstab) / sizeof(int) - 1; mss > 0; mss--) if (tcp_sc_msstab[mss] <= pmss) break; /* Fold parameters and MD5 digest into the ISN we will send. */ data = sch->sch_oddeven;/* odd or even secret, 1 bit */ data |= off << 1; /* secret offset, derived from iss, 3 bits */ data |= mss << 4; /* mss, 3 bits */ MD5Init(&ctx); MD5Update(&ctx, ((u_int8_t *)secbits) + off, SYNCOOKIE_SECRET_SIZE * sizeof(*secbits) - off); MD5Update(&ctx, secbits, off); MD5Update(&ctx, &sc->sc_inc, sizeof(sc->sc_inc)); MD5Update(&ctx, &sc->sc_irs, sizeof(sc->sc_irs)); MD5Update(&ctx, &data, sizeof(data)); MD5Final((u_int8_t *)&md5_buffer, &ctx); data |= (md5_buffer[0] << 7); sc->sc_iss = data; #ifdef INET6 *flowlabel = md5_buffer[1] & IPV6_FLOWLABEL_MASK; #endif /* Additional parameters are stored in the timestamp if present. */ if (sc->sc_flags & SCF_TIMESTAMP) { data = ((sc->sc_flags & SCF_SIGNATURE) ? 1 : 0); /* TCP-MD5, 1 bit */ data |= ((sc->sc_flags & SCF_SACK) ? 1 : 0) << 1; /* SACK, 1 bit */ data |= sc->sc_requested_s_scale << 2; /* SWIN scale, 4 bits */ data |= sc->sc_requested_r_scale << 6; /* RWIN scale, 4 bits */ data |= md5_buffer[2] << 10; /* more digest bits */ data ^= md5_buffer[3]; sc->sc_ts = data; sc->sc_tsoff = data - ticks; /* after XOR */ } TCPSTAT_INC(tcps_sc_sendcookie); } static struct syncache * syncookie_lookup(struct in_conninfo *inc, struct syncache_head *sch, struct syncache *sc, struct tcpopt *to, struct tcphdr *th, struct socket *so) { MD5_CTX ctx; u_int32_t md5_buffer[MD5_DIGEST_LENGTH / sizeof(u_int32_t)]; u_int32_t data = 0; u_int32_t *secbits; tcp_seq ack, seq; int off, mss, wnd, flags; SCH_LOCK_ASSERT(sch); /* * Pull information out of SYN-ACK/ACK and * revert sequence number advances. */ ack = th->th_ack - 1; seq = th->th_seq - 1; off = (ack >> 1) & 0x7; mss = (ack >> 4) & 0x7; flags = ack & 0x7f; /* Which of the two secrets to use. */ secbits = (flags & 0x1) ? sch->sch_secbits_odd : sch->sch_secbits_even; /* * The secret wasn't updated for the lifetime of a syncookie, * so this SYN-ACK/ACK is either too old (replay) or totally bogus. */ if (sch->sch_reseed + SYNCOOKIE_LIFETIME < time_uptime) { return (NULL); } /* Recompute the digest so we can compare it. */ MD5Init(&ctx); MD5Update(&ctx, ((u_int8_t *)secbits) + off, SYNCOOKIE_SECRET_SIZE * sizeof(*secbits) - off); MD5Update(&ctx, secbits, off); MD5Update(&ctx, inc, sizeof(*inc)); MD5Update(&ctx, &seq, sizeof(seq)); MD5Update(&ctx, &flags, sizeof(flags)); MD5Final((u_int8_t *)&md5_buffer, &ctx); /* Does the digest part of or ACK'ed ISS match? */ if ((ack & (~0x7f)) != (md5_buffer[0] << 7)) return (NULL); /* Does the digest part of our reflected timestamp match? */ if (to->to_flags & TOF_TS) { data = md5_buffer[3] ^ to->to_tsecr; if ((data & (~0x3ff)) != (md5_buffer[2] << 10)) return (NULL); } /* Fill in the syncache values. */ bcopy(inc, &sc->sc_inc, sizeof(struct in_conninfo)); sc->sc_ipopts = NULL; sc->sc_irs = seq; sc->sc_iss = ack; #ifdef INET6 if (inc->inc_flags & INC_ISIPV6) { if (sotoinpcb(so)->inp_flags & IN6P_AUTOFLOWLABEL) sc->sc_flowlabel = md5_buffer[1] & IPV6_FLOWLABEL_MASK; } else #endif { sc->sc_ip_ttl = sotoinpcb(so)->inp_ip_ttl; sc->sc_ip_tos = sotoinpcb(so)->inp_ip_tos; } /* Additional parameters that were encoded in the timestamp. */ if (data) { sc->sc_flags |= SCF_TIMESTAMP; sc->sc_tsreflect = to->to_tsval; sc->sc_ts = to->to_tsecr; sc->sc_tsoff = to->to_tsecr - ticks; sc->sc_flags |= (data & 0x1) ? SCF_SIGNATURE : 0; sc->sc_flags |= ((data >> 1) & 0x1) ? SCF_SACK : 0; sc->sc_requested_s_scale = min((data >> 2) & 0xf, TCP_MAX_WINSHIFT); sc->sc_requested_r_scale = min((data >> 6) & 0xf, TCP_MAX_WINSHIFT); if (sc->sc_requested_s_scale || sc->sc_requested_r_scale) sc->sc_flags |= SCF_WINSCALE; } else sc->sc_flags |= SCF_NOOPT; wnd = sbspace(&so->so_rcv); wnd = imax(wnd, 0); wnd = imin(wnd, TCP_MAXWIN); sc->sc_wnd = wnd; sc->sc_rxmits = 0; sc->sc_peer_mss = tcp_sc_msstab[mss]; TCPSTAT_INC(tcps_sc_recvcookie); return (sc); } /* * Returns the current number of syncache entries. This number * will probably change before you get around to calling * syncache_pcblist. */ int syncache_pcbcount(void) { struct syncache_head *sch; int count, i; for (count = 0, i = 0; i < V_tcp_syncache.hashsize; i++) { /* No need to lock for a read. */ sch = &V_tcp_syncache.hashbase[i]; count += sch->sch_length; } return count; } /* * Exports the syncache entries to userland so that netstat can display * them alongside the other sockets. This function is intended to be * called only from tcp_pcblist. * * Due to concurrency on an active system, the number of pcbs exported * may have no relation to max_pcbs. max_pcbs merely indicates the * amount of space the caller allocated for this function to use. */ int syncache_pcblist(struct sysctl_req *req, int max_pcbs, int *pcbs_exported) { struct xtcpcb xt; struct syncache *sc; struct syncache_head *sch; int count, error, i; for (count = 0, error = 0, i = 0; i < V_tcp_syncache.hashsize; i++) { sch = &V_tcp_syncache.hashbase[i]; SCH_LOCK(sch); TAILQ_FOREACH(sc, &sch->sch_bucket, sc_hash) { if (count >= max_pcbs) { SCH_UNLOCK(sch); goto exit; } if (cr_cansee(req->td->td_ucred, sc->sc_cred) != 0) continue; bzero(&xt, sizeof(xt)); xt.xt_len = sizeof(xt); if (sc->sc_inc.inc_flags & INC_ISIPV6) xt.xt_inp.inp_vflag = INP_IPV6; else xt.xt_inp.inp_vflag = INP_IPV4; bcopy(&sc->sc_inc, &xt.xt_inp.inp_inc, sizeof (struct in_conninfo)); xt.xt_tp.t_inpcb = &xt.xt_inp; xt.xt_tp.t_state = TCPS_SYN_RECEIVED; xt.xt_socket.xso_protocol = IPPROTO_TCP; xt.xt_socket.xso_len = sizeof (struct xsocket); xt.xt_socket.so_type = SOCK_STREAM; xt.xt_socket.so_state = SS_ISCONNECTING; error = SYSCTL_OUT(req, &xt, sizeof xt); if (error) { SCH_UNLOCK(sch); goto exit; } count++; } SCH_UNLOCK(sch); } exit: *pcbs_exported = count; return error; } Index: projects/largeSMP/sys/netinet/udp_usrreq.c =================================================================== --- projects/largeSMP/sys/netinet/udp_usrreq.c (revision 222782) +++ projects/largeSMP/sys/netinet/udp_usrreq.c (revision 222783) @@ -1,1604 +1,1605 @@ /*- * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 * The Regents of the University of California. * Copyright (c) 2008 Robert N. M. Watson * 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. * 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. * * @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_ipfw.h" #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.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 #ifdef INET6 #include #endif #include #include #include #include #ifdef INET6 #include #endif #include #include #ifdef IPSEC #include #include #endif #include #include /* * UDP protocol implementation. * Per RFC 768, August, 1980. */ /* * BSD 4.2 defaulted the udp checksum to be off. Turning off udp checksums * removes the only data integrity mechanism for packets and malformed * packets that would otherwise be discarded due to bad checksums, and may * cause problems (especially for NFS data blocks). */ static int udp_cksum = 1; SYSCTL_INT(_net_inet_udp, UDPCTL_CHECKSUM, checksum, CTLFLAG_RW, &udp_cksum, 0, "compute udp checksum"); int udp_log_in_vain = 0; SYSCTL_INT(_net_inet_udp, OID_AUTO, log_in_vain, CTLFLAG_RW, &udp_log_in_vain, 0, "Log all incoming UDP packets"); VNET_DEFINE(int, udp_blackhole) = 0; SYSCTL_VNET_INT(_net_inet_udp, OID_AUTO, blackhole, CTLFLAG_RW, &VNET_NAME(udp_blackhole), 0, "Do not send port unreachables for refused connects"); u_long udp_sendspace = 9216; /* really max datagram size */ /* 40 1K datagrams */ SYSCTL_ULONG(_net_inet_udp, UDPCTL_MAXDGRAM, maxdgram, CTLFLAG_RW, &udp_sendspace, 0, "Maximum outgoing UDP datagram size"); u_long udp_recvspace = 40 * (1024 + #ifdef INET6 sizeof(struct sockaddr_in6) #else sizeof(struct sockaddr_in) #endif ); SYSCTL_ULONG(_net_inet_udp, UDPCTL_RECVSPACE, recvspace, CTLFLAG_RW, &udp_recvspace, 0, "Maximum space for incoming UDP datagrams"); VNET_DEFINE(struct inpcbhead, udb); /* from udp_var.h */ VNET_DEFINE(struct inpcbinfo, udbinfo); static VNET_DEFINE(uma_zone_t, udpcb_zone); #define V_udpcb_zone VNET(udpcb_zone) #ifndef UDBHASHSIZE #define UDBHASHSIZE 128 #endif VNET_DEFINE(struct udpstat, udpstat); /* from udp_var.h */ SYSCTL_VNET_STRUCT(_net_inet_udp, UDPCTL_STATS, stats, CTLFLAG_RW, &VNET_NAME(udpstat), udpstat, "UDP statistics (struct udpstat, netinet/udp_var.h)"); #ifdef INET static void udp_detach(struct socket *so); static int udp_output(struct inpcb *, struct mbuf *, struct sockaddr *, struct mbuf *, struct thread *); #endif #ifdef IPSEC #ifdef IPSEC_NAT_T #define UF_ESPINUDP_ALL (UF_ESPINUDP_NON_IKE|UF_ESPINUDP) #ifdef INET static struct mbuf *udp4_espdecap(struct inpcb *, struct mbuf *, int); #endif #endif /* IPSEC_NAT_T */ #endif /* IPSEC */ static void udp_zone_change(void *tag) { uma_zone_set_max(V_udbinfo.ipi_zone, maxsockets); uma_zone_set_max(V_udpcb_zone, maxsockets); } static int udp_inpcb_init(void *mem, int size, int flags) { struct inpcb *inp; inp = mem; INP_LOCK_INIT(inp, "inp", "udpinp"); return (0); } void udp_init(void) { in_pcbinfo_init(&V_udbinfo, "udp", &V_udb, UDBHASHSIZE, UDBHASHSIZE, - "udp_inpcb", udp_inpcb_init, NULL, UMA_ZONE_NOFREE); + "udp_inpcb", udp_inpcb_init, NULL, UMA_ZONE_NOFREE, + IPI_HASHFIELDS_2TUPLE); V_udpcb_zone = uma_zcreate("udpcb", sizeof(struct udpcb), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); uma_zone_set_max(V_udpcb_zone, maxsockets); EVENTHANDLER_REGISTER(maxsockets_change, udp_zone_change, NULL, EVENTHANDLER_PRI_ANY); } /* * Kernel module interface for updating udpstat. The argument is an index * into udpstat treated as an array of u_long. While this encodes the * general layout of udpstat into the caller, it doesn't encode its location, * so that future changes to add, for example, per-CPU stats support won't * cause binary compatibility problems for kernel modules. */ void kmod_udpstat_inc(int statnum) { (*((u_long *)&V_udpstat + statnum))++; } int udp_newudpcb(struct inpcb *inp) { struct udpcb *up; up = uma_zalloc(V_udpcb_zone, M_NOWAIT | M_ZERO); if (up == NULL) return (ENOBUFS); inp->inp_ppcb = up; return (0); } void udp_discardcb(struct udpcb *up) { uma_zfree(V_udpcb_zone, up); } #ifdef VIMAGE void udp_destroy(void) { in_pcbinfo_destroy(&V_udbinfo); uma_zdestroy(V_udpcb_zone); } #endif #ifdef INET /* * Subroutine of udp_input(), which appends the provided mbuf chain to the * passed pcb/socket. The caller must provide a sockaddr_in via udp_in that * contains the source address. If the socket ends up being an IPv6 socket, * udp_append() will convert to a sockaddr_in6 before passing the address * into the socket code. */ static void udp_append(struct inpcb *inp, struct ip *ip, struct mbuf *n, int off, struct sockaddr_in *udp_in) { struct sockaddr *append_sa; struct socket *so; struct mbuf *opts = 0; #ifdef INET6 struct sockaddr_in6 udp_in6; #endif struct udpcb *up; INP_LOCK_ASSERT(inp); /* * Engage the tunneling protocol. */ up = intoudpcb(inp); if (up->u_tun_func != NULL) { (*up->u_tun_func)(n, off, inp); return; } if (n == NULL) return; off += sizeof(struct udphdr); #ifdef IPSEC /* Check AH/ESP integrity. */ if (ipsec4_in_reject(n, inp)) { m_freem(n); V_ipsec4stat.in_polvio++; return; } #ifdef IPSEC_NAT_T up = intoudpcb(inp); KASSERT(up != NULL, ("%s: udpcb NULL", __func__)); if (up->u_flags & UF_ESPINUDP_ALL) { /* IPSec UDP encaps. */ n = udp4_espdecap(inp, n, off); if (n == NULL) /* Consumed. */ return; } #endif /* IPSEC_NAT_T */ #endif /* IPSEC */ #ifdef MAC if (mac_inpcb_check_deliver(inp, n) != 0) { m_freem(n); return; } #endif /* MAC */ if (inp->inp_flags & INP_CONTROLOPTS || inp->inp_socket->so_options & (SO_TIMESTAMP | SO_BINTIME)) { #ifdef INET6 if (inp->inp_vflag & INP_IPV6) (void)ip6_savecontrol_v4(inp, n, &opts, NULL); else #endif /* INET6 */ ip_savecontrol(inp, &opts, ip, n); } #ifdef INET6 if (inp->inp_vflag & INP_IPV6) { bzero(&udp_in6, sizeof(udp_in6)); udp_in6.sin6_len = sizeof(udp_in6); udp_in6.sin6_family = AF_INET6; in6_sin_2_v4mapsin6(udp_in, &udp_in6); append_sa = (struct sockaddr *)&udp_in6; } else #endif /* INET6 */ append_sa = (struct sockaddr *)udp_in; m_adj(n, off); so = inp->inp_socket; SOCKBUF_LOCK(&so->so_rcv); if (sbappendaddr_locked(&so->so_rcv, append_sa, n, opts) == 0) { SOCKBUF_UNLOCK(&so->so_rcv); m_freem(n); if (opts) m_freem(opts); UDPSTAT_INC(udps_fullsock); } else sorwakeup_locked(so); } void udp_input(struct mbuf *m, int off) { int iphlen = off; struct ip *ip; struct udphdr *uh; struct ifnet *ifp; struct inpcb *inp; int len; struct ip save_ip; struct sockaddr_in udp_in; #ifdef IPFIREWALL_FORWARD struct m_tag *fwd_tag; #endif ifp = m->m_pkthdr.rcvif; UDPSTAT_INC(udps_ipackets); /* * Strip IP options, if any; should skip this, make available to * user, and use on returned packets, but we don't yet have a way to * check the checksum with options still present. */ if (iphlen > sizeof (struct ip)) { ip_stripoptions(m, (struct mbuf *)0); iphlen = sizeof(struct ip); } /* * Get IP and UDP header together in first mbuf. */ ip = mtod(m, struct ip *); if (m->m_len < iphlen + sizeof(struct udphdr)) { if ((m = m_pullup(m, iphlen + sizeof(struct udphdr))) == 0) { UDPSTAT_INC(udps_hdrops); return; } ip = mtod(m, struct ip *); } uh = (struct udphdr *)((caddr_t)ip + iphlen); /* * Destination port of 0 is illegal, based on RFC768. */ if (uh->uh_dport == 0) goto badunlocked; /* * Construct sockaddr format source address. Stuff source address * and datagram in user buffer. */ bzero(&udp_in, sizeof(udp_in)); udp_in.sin_len = sizeof(udp_in); udp_in.sin_family = AF_INET; udp_in.sin_port = uh->uh_sport; udp_in.sin_addr = ip->ip_src; /* * Make mbuf data length reflect UDP length. If not enough data to * reflect UDP length, drop. */ len = ntohs((u_short)uh->uh_ulen); if (ip->ip_len != len) { if (len > ip->ip_len || len < sizeof(struct udphdr)) { UDPSTAT_INC(udps_badlen); goto badunlocked; } m_adj(m, len - ip->ip_len); /* ip->ip_len = len; */ } /* * Save a copy of the IP header in case we want restore it for * sending an ICMP error message in response. */ if (!V_udp_blackhole) save_ip = *ip; else memset(&save_ip, 0, sizeof(save_ip)); /* * Checksum extended UDP header and data. */ if (uh->uh_sum) { u_short uh_sum; if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) { if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR) uh_sum = m->m_pkthdr.csum_data; else uh_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htonl((u_short)len + m->m_pkthdr.csum_data + IPPROTO_UDP)); uh_sum ^= 0xffff; } else { char b[9]; bcopy(((struct ipovly *)ip)->ih_x1, b, 9); bzero(((struct ipovly *)ip)->ih_x1, 9); ((struct ipovly *)ip)->ih_len = uh->uh_ulen; uh_sum = in_cksum(m, len + sizeof (struct ip)); bcopy(b, ((struct ipovly *)ip)->ih_x1, 9); } if (uh_sum) { UDPSTAT_INC(udps_badsum); m_freem(m); return; } } else UDPSTAT_INC(udps_nosum); #ifdef IPFIREWALL_FORWARD /* * Grab info from PACKET_TAG_IPFORWARD tag prepended to the chain. */ fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL); if (fwd_tag != NULL) { struct sockaddr_in *next_hop; /* * Do the hack. */ next_hop = (struct sockaddr_in *)(fwd_tag + 1); ip->ip_dst = next_hop->sin_addr; uh->uh_dport = ntohs(next_hop->sin_port); /* * Remove the tag from the packet. We don't need it anymore. */ m_tag_delete(m, fwd_tag); } #endif if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) || in_broadcast(ip->ip_dst, ifp)) { struct inpcb *last; struct ip_moptions *imo; INP_INFO_RLOCK(&V_udbinfo); last = NULL; LIST_FOREACH(inp, &V_udb, inp_list) { if (inp->inp_lport != uh->uh_dport) continue; #ifdef INET6 if ((inp->inp_vflag & INP_IPV4) == 0) continue; #endif if (inp->inp_laddr.s_addr != INADDR_ANY && inp->inp_laddr.s_addr != ip->ip_dst.s_addr) continue; if (inp->inp_faddr.s_addr != INADDR_ANY && inp->inp_faddr.s_addr != ip->ip_src.s_addr) continue; if (inp->inp_fport != 0 && inp->inp_fport != uh->uh_sport) continue; INP_RLOCK(inp); /* * XXXRW: Because we weren't holding either the inpcb * or the hash lock when we checked for a match * before, we should probably recheck now that the * inpcb lock is held. */ /* * Handle socket delivery policy for any-source * and source-specific multicast. [RFC3678] */ imo = inp->inp_moptions; if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { struct sockaddr_in group; int blocked; if (imo == NULL) { INP_RUNLOCK(inp); continue; } bzero(&group, sizeof(struct sockaddr_in)); group.sin_len = sizeof(struct sockaddr_in); group.sin_family = AF_INET; group.sin_addr = ip->ip_dst; blocked = imo_multi_filter(imo, ifp, (struct sockaddr *)&group, (struct sockaddr *)&udp_in); if (blocked != MCAST_PASS) { if (blocked == MCAST_NOTGMEMBER) IPSTAT_INC(ips_notmember); if (blocked == MCAST_NOTSMEMBER || blocked == MCAST_MUTED) UDPSTAT_INC(udps_filtermcast); INP_RUNLOCK(inp); continue; } } if (last != NULL) { struct mbuf *n; n = m_copy(m, 0, M_COPYALL); udp_append(last, ip, n, iphlen, &udp_in); INP_RUNLOCK(last); } last = inp; /* * Don't look for additional matches if this one does * not have either the SO_REUSEPORT or SO_REUSEADDR * socket options set. This heuristic avoids * searching through all pcbs in the common case of a * non-shared port. It assumes that an application * will never clear these options after setting them. */ if ((last->inp_socket->so_options & (SO_REUSEPORT|SO_REUSEADDR)) == 0) break; } if (last == NULL) { /* * No matching pcb found; discard datagram. (No need * to send an ICMP Port Unreachable for a broadcast * or multicast datgram.) */ UDPSTAT_INC(udps_noportbcast); if (inp) INP_RUNLOCK(inp); INP_INFO_RUNLOCK(&V_udbinfo); goto badunlocked; } udp_append(last, ip, m, iphlen, &udp_in); INP_RUNLOCK(last); INP_INFO_RUNLOCK(&V_udbinfo); return; } /* * Locate pcb for datagram. */ inp = in_pcblookup_mbuf(&V_udbinfo, ip->ip_src, uh->uh_sport, ip->ip_dst, uh->uh_dport, INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, ifp, m); if (inp == NULL) { if (udp_log_in_vain) { char buf[4*sizeof "123"]; strcpy(buf, inet_ntoa(ip->ip_dst)); log(LOG_INFO, "Connection attempt to UDP %s:%d from %s:%d\n", buf, ntohs(uh->uh_dport), inet_ntoa(ip->ip_src), ntohs(uh->uh_sport)); } UDPSTAT_INC(udps_noport); if (m->m_flags & (M_BCAST | M_MCAST)) { UDPSTAT_INC(udps_noportbcast); goto badunlocked; } if (V_udp_blackhole) goto badunlocked; if (badport_bandlim(BANDLIM_ICMP_UNREACH) < 0) goto badunlocked; *ip = save_ip; ip->ip_len += iphlen; icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0); return; } /* * Check the minimum TTL for socket. */ INP_RLOCK_ASSERT(inp); if (inp->inp_ip_minttl && inp->inp_ip_minttl > ip->ip_ttl) { INP_RUNLOCK(inp); m_freem(m); return; } udp_append(inp, ip, m, iphlen, &udp_in); INP_RUNLOCK(inp); return; badunlocked: m_freem(m); } #endif /* INET */ /* * Notify a udp user of an asynchronous error; just wake up so that they can * collect error status. */ struct inpcb * udp_notify(struct inpcb *inp, int errno) { /* * While udp_ctlinput() always calls udp_notify() with a read lock * when invoking it directly, in_pcbnotifyall() currently uses write * locks due to sharing code with TCP. For now, accept either a read * or a write lock, but a read lock is sufficient. */ INP_LOCK_ASSERT(inp); inp->inp_socket->so_error = errno; sorwakeup(inp->inp_socket); sowwakeup(inp->inp_socket); return (inp); } #ifdef INET void udp_ctlinput(int cmd, struct sockaddr *sa, void *vip) { struct ip *ip = vip; struct udphdr *uh; struct in_addr faddr; struct inpcb *inp; faddr = ((struct sockaddr_in *)sa)->sin_addr; if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY) return; /* * Redirects don't need to be handled up here. */ if (PRC_IS_REDIRECT(cmd)) return; /* * Hostdead is ugly because it goes linearly through all PCBs. * * XXX: We never get this from ICMP, otherwise it makes an excellent * DoS attack on machines with many connections. */ if (cmd == PRC_HOSTDEAD) ip = NULL; else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0) return; if (ip != NULL) { uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2)); inp = in_pcblookup(&V_udbinfo, faddr, uh->uh_dport, ip->ip_src, uh->uh_sport, INPLOOKUP_RLOCKPCB, NULL); if (inp != NULL) { INP_RLOCK_ASSERT(inp); if (inp->inp_socket != NULL) { udp_notify(inp, inetctlerrmap[cmd]); } INP_RUNLOCK(inp); } } else in_pcbnotifyall(&V_udbinfo, faddr, inetctlerrmap[cmd], udp_notify); } #endif /* INET */ static int udp_pcblist(SYSCTL_HANDLER_ARGS) { int error, i, n; struct inpcb *inp, **inp_list; inp_gen_t gencnt; struct xinpgen xig; /* * The process of preparing the PCB list is too time-consuming and * resource-intensive to repeat twice on every request. */ if (req->oldptr == 0) { n = V_udbinfo.ipi_count; n += imax(n / 8, 10); req->oldidx = 2 * (sizeof xig) + n * sizeof(struct xinpcb); return (0); } if (req->newptr != 0) return (EPERM); /* * OK, now we're committed to doing something. */ INP_INFO_RLOCK(&V_udbinfo); gencnt = V_udbinfo.ipi_gencnt; n = V_udbinfo.ipi_count; INP_INFO_RUNLOCK(&V_udbinfo); error = sysctl_wire_old_buffer(req, 2 * (sizeof xig) + n * sizeof(struct xinpcb)); if (error != 0) return (error); xig.xig_len = sizeof xig; xig.xig_count = n; xig.xig_gen = gencnt; xig.xig_sogen = so_gencnt; error = SYSCTL_OUT(req, &xig, sizeof xig); if (error) return (error); inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); if (inp_list == 0) return (ENOMEM); INP_INFO_RLOCK(&V_udbinfo); for (inp = LIST_FIRST(V_udbinfo.ipi_listhead), i = 0; inp && i < n; inp = LIST_NEXT(inp, inp_list)) { INP_WLOCK(inp); if (inp->inp_gencnt <= gencnt && cr_canseeinpcb(req->td->td_ucred, inp) == 0) { in_pcbref(inp); inp_list[i++] = inp; } INP_WUNLOCK(inp); } INP_INFO_RUNLOCK(&V_udbinfo); n = i; error = 0; for (i = 0; i < n; i++) { inp = inp_list[i]; INP_RLOCK(inp); if (inp->inp_gencnt <= gencnt) { struct xinpcb xi; bzero(&xi, sizeof(xi)); xi.xi_len = sizeof xi; /* XXX should avoid extra copy */ bcopy(inp, &xi.xi_inp, sizeof *inp); if (inp->inp_socket) sotoxsocket(inp->inp_socket, &xi.xi_socket); xi.xi_inp.inp_gencnt = inp->inp_gencnt; INP_RUNLOCK(inp); error = SYSCTL_OUT(req, &xi, sizeof xi); } else INP_RUNLOCK(inp); } INP_INFO_WLOCK(&V_udbinfo); for (i = 0; i < n; i++) { inp = inp_list[i]; INP_RLOCK(inp); if (!in_pcbrele_rlocked(inp)) INP_RUNLOCK(inp); } INP_INFO_WUNLOCK(&V_udbinfo); if (!error) { /* * Give the user an updated idea of our state. If the * generation differs from what we told her before, she knows * that something happened while we were processing this * request, and it might be necessary to retry. */ INP_INFO_RLOCK(&V_udbinfo); xig.xig_gen = V_udbinfo.ipi_gencnt; xig.xig_sogen = so_gencnt; xig.xig_count = V_udbinfo.ipi_count; INP_INFO_RUNLOCK(&V_udbinfo); error = SYSCTL_OUT(req, &xig, sizeof xig); } free(inp_list, M_TEMP); return (error); } SYSCTL_PROC(_net_inet_udp, UDPCTL_PCBLIST, pcblist, CTLTYPE_OPAQUE | CTLFLAG_RD, NULL, 0, udp_pcblist, "S,xinpcb", "List of active UDP sockets"); #ifdef INET static int udp_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in addrs[2]; struct inpcb *inp; int error; error = priv_check(req->td, PRIV_NETINET_GETCRED); if (error) return (error); error = SYSCTL_IN(req, addrs, sizeof(addrs)); if (error) return (error); inp = in_pcblookup(&V_udbinfo, addrs[1].sin_addr, addrs[1].sin_port, addrs[0].sin_addr, addrs[0].sin_port, INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, NULL); if (inp != NULL) { INP_RLOCK_ASSERT(inp); if (inp->inp_socket == NULL) error = ENOENT; if (error == 0) error = cr_canseeinpcb(req->td->td_ucred, inp); if (error == 0) cru2x(inp->inp_cred, &xuc); INP_RUNLOCK(inp); } else error = ENOENT; if (error == 0) error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred)); return (error); } SYSCTL_PROC(_net_inet_udp, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW|CTLFLAG_PRISON, 0, 0, udp_getcred, "S,xucred", "Get the xucred of a UDP connection"); #endif /* INET */ int udp_ctloutput(struct socket *so, struct sockopt *sopt) { int error = 0, optval; struct inpcb *inp; #ifdef IPSEC_NAT_T struct udpcb *up; #endif inp = sotoinpcb(so); KASSERT(inp != NULL, ("%s: inp == NULL", __func__)); INP_WLOCK(inp); if (sopt->sopt_level != IPPROTO_UDP) { #ifdef INET6 if (INP_CHECK_SOCKAF(so, AF_INET6)) { INP_WUNLOCK(inp); error = ip6_ctloutput(so, sopt); } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { INP_WUNLOCK(inp); error = ip_ctloutput(so, sopt); } #endif return (error); } switch (sopt->sopt_dir) { case SOPT_SET: switch (sopt->sopt_name) { case UDP_ENCAP: INP_WUNLOCK(inp); error = sooptcopyin(sopt, &optval, sizeof optval, sizeof optval); if (error) break; inp = sotoinpcb(so); KASSERT(inp != NULL, ("%s: inp == NULL", __func__)); INP_WLOCK(inp); #ifdef IPSEC_NAT_T up = intoudpcb(inp); KASSERT(up != NULL, ("%s: up == NULL", __func__)); #endif switch (optval) { case 0: /* Clear all UDP encap. */ #ifdef IPSEC_NAT_T up->u_flags &= ~UF_ESPINUDP_ALL; #endif break; #ifdef IPSEC_NAT_T case UDP_ENCAP_ESPINUDP: case UDP_ENCAP_ESPINUDP_NON_IKE: up->u_flags &= ~UF_ESPINUDP_ALL; if (optval == UDP_ENCAP_ESPINUDP) up->u_flags |= UF_ESPINUDP; else if (optval == UDP_ENCAP_ESPINUDP_NON_IKE) up->u_flags |= UF_ESPINUDP_NON_IKE; break; #endif default: error = EINVAL; break; } INP_WUNLOCK(inp); break; default: INP_WUNLOCK(inp); error = ENOPROTOOPT; break; } break; case SOPT_GET: switch (sopt->sopt_name) { #ifdef IPSEC_NAT_T case UDP_ENCAP: up = intoudpcb(inp); KASSERT(up != NULL, ("%s: up == NULL", __func__)); optval = up->u_flags & UF_ESPINUDP_ALL; INP_WUNLOCK(inp); error = sooptcopyout(sopt, &optval, sizeof optval); break; #endif default: INP_WUNLOCK(inp); error = ENOPROTOOPT; break; } break; } return (error); } #ifdef INET #define UH_WLOCKED 2 #define UH_RLOCKED 1 #define UH_UNLOCKED 0 static int udp_output(struct inpcb *inp, struct mbuf *m, struct sockaddr *addr, struct mbuf *control, struct thread *td) { struct udpiphdr *ui; int len = m->m_pkthdr.len; struct in_addr faddr, laddr; struct cmsghdr *cm; struct sockaddr_in *sin, src; int error = 0; int ipflags; u_short fport, lport; int unlock_udbinfo; /* * udp_output() may need to temporarily bind or connect the current * inpcb. As such, we don't know up front whether we will need the * pcbinfo lock or not. Do any work to decide what is needed up * front before acquiring any locks. */ if (len + sizeof(struct udpiphdr) > IP_MAXPACKET) { if (control) m_freem(control); m_freem(m); return (EMSGSIZE); } src.sin_family = 0; if (control != NULL) { /* * XXX: Currently, we assume all the optional information is * stored in a single mbuf. */ if (control->m_next) { m_freem(control); m_freem(m); return (EINVAL); } for (; control->m_len > 0; control->m_data += CMSG_ALIGN(cm->cmsg_len), control->m_len -= CMSG_ALIGN(cm->cmsg_len)) { cm = mtod(control, struct cmsghdr *); if (control->m_len < sizeof(*cm) || cm->cmsg_len == 0 || cm->cmsg_len > control->m_len) { error = EINVAL; break; } if (cm->cmsg_level != IPPROTO_IP) continue; switch (cm->cmsg_type) { case IP_SENDSRCADDR: if (cm->cmsg_len != CMSG_LEN(sizeof(struct in_addr))) { error = EINVAL; break; } bzero(&src, sizeof(src)); src.sin_family = AF_INET; src.sin_len = sizeof(src); src.sin_port = inp->inp_lport; src.sin_addr = *(struct in_addr *)CMSG_DATA(cm); break; default: error = ENOPROTOOPT; break; } if (error) break; } m_freem(control); } if (error) { m_freem(m); return (error); } /* * Depending on whether or not the application has bound or connected * the socket, we may have to do varying levels of work. The optimal * case is for a connected UDP socket, as a global lock isn't * required at all. * * In order to decide which we need, we require stability of the * inpcb binding, which we ensure by acquiring a read lock on the * inpcb. This doesn't strictly follow the lock order, so we play * the trylock and retry game; note that we may end up with more * conservative locks than required the second time around, so later * assertions have to accept that. Further analysis of the number of * misses under contention is required. * * XXXRW: Check that hash locking update here is correct. */ sin = (struct sockaddr_in *)addr; INP_RLOCK(inp); if (sin != NULL && (inp->inp_laddr.s_addr == INADDR_ANY && inp->inp_lport == 0)) { INP_RUNLOCK(inp); INP_WLOCK(inp); INP_HASH_WLOCK(&V_udbinfo); unlock_udbinfo = UH_WLOCKED; } else if ((sin != NULL && ( (sin->sin_addr.s_addr == INADDR_ANY) || (sin->sin_addr.s_addr == INADDR_BROADCAST) || (inp->inp_laddr.s_addr == INADDR_ANY) || (inp->inp_lport == 0))) || (src.sin_family == AF_INET)) { INP_HASH_RLOCK(&V_udbinfo); unlock_udbinfo = UH_RLOCKED; } else unlock_udbinfo = UH_UNLOCKED; /* * If the IP_SENDSRCADDR control message was specified, override the * source address for this datagram. Its use is invalidated if the * address thus specified is incomplete or clobbers other inpcbs. */ laddr = inp->inp_laddr; lport = inp->inp_lport; if (src.sin_family == AF_INET) { INP_HASH_LOCK_ASSERT(&V_udbinfo); if ((lport == 0) || (laddr.s_addr == INADDR_ANY && src.sin_addr.s_addr == INADDR_ANY)) { error = EINVAL; goto release; } error = in_pcbbind_setup(inp, (struct sockaddr *)&src, &laddr.s_addr, &lport, td->td_ucred); if (error) goto release; } /* * If a UDP socket has been connected, then a local address/port will * have been selected and bound. * * If a UDP socket has not been connected to, then an explicit * destination address must be used, in which case a local * address/port may not have been selected and bound. */ if (sin != NULL) { INP_LOCK_ASSERT(inp); if (inp->inp_faddr.s_addr != INADDR_ANY) { error = EISCONN; goto release; } /* * Jail may rewrite the destination address, so let it do * that before we use it. */ error = prison_remote_ip4(td->td_ucred, &sin->sin_addr); if (error) goto release; /* * If a local address or port hasn't yet been selected, or if * the destination address needs to be rewritten due to using * a special INADDR_ constant, invoke in_pcbconnect_setup() * to do the heavy lifting. Once a port is selected, we * commit the binding back to the socket; we also commit the * binding of the address if in jail. * * If we already have a valid binding and we're not * requesting a destination address rewrite, use a fast path. */ if (inp->inp_laddr.s_addr == INADDR_ANY || inp->inp_lport == 0 || sin->sin_addr.s_addr == INADDR_ANY || sin->sin_addr.s_addr == INADDR_BROADCAST) { INP_HASH_LOCK_ASSERT(&V_udbinfo); error = in_pcbconnect_setup(inp, addr, &laddr.s_addr, &lport, &faddr.s_addr, &fport, NULL, td->td_ucred); if (error) goto release; /* * XXXRW: Why not commit the port if the address is * !INADDR_ANY? */ /* Commit the local port if newly assigned. */ if (inp->inp_laddr.s_addr == INADDR_ANY && inp->inp_lport == 0) { INP_WLOCK_ASSERT(inp); INP_HASH_WLOCK_ASSERT(&V_udbinfo); /* * Remember addr if jailed, to prevent * rebinding. */ if (prison_flag(td->td_ucred, PR_IP4)) inp->inp_laddr = laddr; inp->inp_lport = lport; if (in_pcbinshash(inp) != 0) { inp->inp_lport = 0; error = EAGAIN; goto release; } inp->inp_flags |= INP_ANONPORT; } } else { faddr = sin->sin_addr; fport = sin->sin_port; } } else { INP_LOCK_ASSERT(inp); faddr = inp->inp_faddr; fport = inp->inp_fport; if (faddr.s_addr == INADDR_ANY) { error = ENOTCONN; goto release; } } /* * Calculate data length and get a mbuf for UDP, IP, and possible * link-layer headers. Immediate slide the data pointer back forward * since we won't use that space at this layer. */ M_PREPEND(m, sizeof(struct udpiphdr) + max_linkhdr, M_DONTWAIT); if (m == NULL) { error = ENOBUFS; goto release; } m->m_data += max_linkhdr; m->m_len -= max_linkhdr; m->m_pkthdr.len -= max_linkhdr; /* * Fill in mbuf with extended UDP header and addresses and length put * into network format. */ ui = mtod(m, struct udpiphdr *); bzero(ui->ui_x1, sizeof(ui->ui_x1)); /* XXX still needed? */ ui->ui_pr = IPPROTO_UDP; ui->ui_src = laddr; ui->ui_dst = faddr; ui->ui_sport = lport; ui->ui_dport = fport; ui->ui_ulen = htons((u_short)len + sizeof(struct udphdr)); /* * Set the Don't Fragment bit in the IP header. */ if (inp->inp_flags & INP_DONTFRAG) { struct ip *ip; ip = (struct ip *)&ui->ui_i; ip->ip_off |= IP_DF; } ipflags = 0; if (inp->inp_socket->so_options & SO_DONTROUTE) ipflags |= IP_ROUTETOIF; if (inp->inp_socket->so_options & SO_BROADCAST) ipflags |= IP_ALLOWBROADCAST; if (inp->inp_flags & INP_ONESBCAST) ipflags |= IP_SENDONES; #ifdef MAC mac_inpcb_create_mbuf(inp, m); #endif /* * Set up checksum and output datagram. */ if (udp_cksum) { if (inp->inp_flags & INP_ONESBCAST) faddr.s_addr = INADDR_BROADCAST; ui->ui_sum = in_pseudo(ui->ui_src.s_addr, faddr.s_addr, htons((u_short)len + sizeof(struct udphdr) + IPPROTO_UDP)); m->m_pkthdr.csum_flags = CSUM_UDP; m->m_pkthdr.csum_data = offsetof(struct udphdr, uh_sum); } else ui->ui_sum = 0; ((struct ip *)ui)->ip_len = sizeof (struct udpiphdr) + len; ((struct ip *)ui)->ip_ttl = inp->inp_ip_ttl; /* XXX */ ((struct ip *)ui)->ip_tos = inp->inp_ip_tos; /* XXX */ UDPSTAT_INC(udps_opackets); if (unlock_udbinfo == UH_WLOCKED) INP_HASH_WUNLOCK(&V_udbinfo); else if (unlock_udbinfo == UH_RLOCKED) INP_HASH_RUNLOCK(&V_udbinfo); error = ip_output(m, inp->inp_options, NULL, ipflags, inp->inp_moptions, inp); if (unlock_udbinfo == UH_WLOCKED) INP_WUNLOCK(inp); else INP_RUNLOCK(inp); return (error); release: if (unlock_udbinfo == UH_WLOCKED) { INP_HASH_WUNLOCK(&V_udbinfo); INP_WUNLOCK(inp); } else if (unlock_udbinfo == UH_RLOCKED) { INP_HASH_RUNLOCK(&V_udbinfo); INP_RUNLOCK(inp); } else INP_RUNLOCK(inp); m_freem(m); return (error); } #if defined(IPSEC) && defined(IPSEC_NAT_T) /* * Potentially decap ESP in UDP frame. Check for an ESP header * and optional marker; if present, strip the UDP header and * push the result through IPSec. * * Returns mbuf to be processed (potentially re-allocated) or * NULL if consumed and/or processed. */ static struct mbuf * udp4_espdecap(struct inpcb *inp, struct mbuf *m, int off) { size_t minlen, payload, skip, iphlen; caddr_t data; struct udpcb *up; struct m_tag *tag; struct udphdr *udphdr; struct ip *ip; INP_RLOCK_ASSERT(inp); /* * Pull up data so the longest case is contiguous: * IP/UDP hdr + non ESP marker + ESP hdr. */ minlen = off + sizeof(uint64_t) + sizeof(struct esp); if (minlen > m->m_pkthdr.len) minlen = m->m_pkthdr.len; if ((m = m_pullup(m, minlen)) == NULL) { V_ipsec4stat.in_inval++; return (NULL); /* Bypass caller processing. */ } data = mtod(m, caddr_t); /* Points to ip header. */ payload = m->m_len - off; /* Size of payload. */ if (payload == 1 && data[off] == '\xff') return (m); /* NB: keepalive packet, no decap. */ up = intoudpcb(inp); KASSERT(up != NULL, ("%s: udpcb NULL", __func__)); KASSERT((up->u_flags & UF_ESPINUDP_ALL) != 0, ("u_flags 0x%x", up->u_flags)); /* * Check that the payload is large enough to hold an * ESP header and compute the amount of data to remove. * * NB: the caller has already done a pullup for us. * XXX can we assume alignment and eliminate bcopys? */ if (up->u_flags & UF_ESPINUDP_NON_IKE) { /* * draft-ietf-ipsec-nat-t-ike-0[01].txt and * draft-ietf-ipsec-udp-encaps-(00/)01.txt, ignoring * possible AH mode non-IKE marker+non-ESP marker * from draft-ietf-ipsec-udp-encaps-00.txt. */ uint64_t marker; if (payload <= sizeof(uint64_t) + sizeof(struct esp)) return (m); /* NB: no decap. */ bcopy(data + off, &marker, sizeof(uint64_t)); if (marker != 0) /* Non-IKE marker. */ return (m); /* NB: no decap. */ skip = sizeof(uint64_t) + sizeof(struct udphdr); } else { uint32_t spi; if (payload <= sizeof(struct esp)) { V_ipsec4stat.in_inval++; m_freem(m); return (NULL); /* Discard. */ } bcopy(data + off, &spi, sizeof(uint32_t)); if (spi == 0) /* Non-ESP marker. */ return (m); /* NB: no decap. */ skip = sizeof(struct udphdr); } /* * Setup a PACKET_TAG_IPSEC_NAT_T_PORT tag to remember * the UDP ports. This is required if we want to select * the right SPD for multiple hosts behind same NAT. * * NB: ports are maintained in network byte order everywhere * in the NAT-T code. */ tag = m_tag_get(PACKET_TAG_IPSEC_NAT_T_PORTS, 2 * sizeof(uint16_t), M_NOWAIT); if (tag == NULL) { V_ipsec4stat.in_nomem++; m_freem(m); return (NULL); /* Discard. */ } iphlen = off - sizeof(struct udphdr); udphdr = (struct udphdr *)(data + iphlen); ((uint16_t *)(tag + 1))[0] = udphdr->uh_sport; ((uint16_t *)(tag + 1))[1] = udphdr->uh_dport; m_tag_prepend(m, tag); /* * Remove the UDP header (and possibly the non ESP marker) * IP header length is iphlen * Before: * <--- off ---> * +----+------+-----+ * | IP | UDP | ESP | * +----+------+-----+ * <-skip-> * After: * +----+-----+ * | IP | ESP | * +----+-----+ * <-skip-> */ ovbcopy(data, data + skip, iphlen); m_adj(m, skip); ip = mtod(m, struct ip *); ip->ip_len -= skip; ip->ip_p = IPPROTO_ESP; /* * We cannot yet update the cksums so clear any * h/w cksum flags as they are no longer valid. */ if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) m->m_pkthdr.csum_flags &= ~(CSUM_DATA_VALID|CSUM_PSEUDO_HDR); (void) ipsec4_common_input(m, iphlen, ip->ip_p); return (NULL); /* NB: consumed, bypass processing. */ } #endif /* defined(IPSEC) && defined(IPSEC_NAT_T) */ static void udp_abort(struct socket *so) { struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_abort: inp == NULL")); INP_WLOCK(inp); if (inp->inp_faddr.s_addr != INADDR_ANY) { INP_HASH_WLOCK(&V_udbinfo); in_pcbdisconnect(inp); inp->inp_laddr.s_addr = INADDR_ANY; INP_HASH_WUNLOCK(&V_udbinfo); soisdisconnected(so); } INP_WUNLOCK(inp); } static int udp_attach(struct socket *so, int proto, struct thread *td) { struct inpcb *inp; int error; inp = sotoinpcb(so); KASSERT(inp == NULL, ("udp_attach: inp != NULL")); error = soreserve(so, udp_sendspace, udp_recvspace); if (error) return (error); INP_INFO_WLOCK(&V_udbinfo); error = in_pcballoc(so, &V_udbinfo); if (error) { INP_INFO_WUNLOCK(&V_udbinfo); return (error); } inp = sotoinpcb(so); inp->inp_vflag |= INP_IPV4; inp->inp_ip_ttl = V_ip_defttl; error = udp_newudpcb(inp); if (error) { in_pcbdetach(inp); in_pcbfree(inp); INP_INFO_WUNLOCK(&V_udbinfo); return (error); } INP_WUNLOCK(inp); INP_INFO_WUNLOCK(&V_udbinfo); return (0); } #endif /* INET */ int udp_set_kernel_tunneling(struct socket *so, udp_tun_func_t f) { struct inpcb *inp; struct udpcb *up; KASSERT(so->so_type == SOCK_DGRAM, ("udp_set_kernel_tunneling: !dgram")); inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_set_kernel_tunneling: inp == NULL")); INP_WLOCK(inp); up = intoudpcb(inp); if (up->u_tun_func != NULL) { INP_WUNLOCK(inp); return (EBUSY); } up->u_tun_func = f; INP_WUNLOCK(inp); return (0); } #ifdef INET static int udp_bind(struct socket *so, struct sockaddr *nam, struct thread *td) { struct inpcb *inp; int error; inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_bind: inp == NULL")); INP_WLOCK(inp); INP_HASH_WLOCK(&V_udbinfo); error = in_pcbbind(inp, nam, td->td_ucred); INP_HASH_WUNLOCK(&V_udbinfo); INP_WUNLOCK(inp); return (error); } static void udp_close(struct socket *so) { struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_close: inp == NULL")); INP_WLOCK(inp); if (inp->inp_faddr.s_addr != INADDR_ANY) { INP_HASH_WLOCK(&V_udbinfo); in_pcbdisconnect(inp); inp->inp_laddr.s_addr = INADDR_ANY; INP_HASH_WUNLOCK(&V_udbinfo); soisdisconnected(so); } INP_WUNLOCK(inp); } static int udp_connect(struct socket *so, struct sockaddr *nam, struct thread *td) { struct inpcb *inp; int error; struct sockaddr_in *sin; inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_connect: inp == NULL")); INP_WLOCK(inp); if (inp->inp_faddr.s_addr != INADDR_ANY) { INP_WUNLOCK(inp); return (EISCONN); } sin = (struct sockaddr_in *)nam; error = prison_remote_ip4(td->td_ucred, &sin->sin_addr); if (error != 0) { INP_WUNLOCK(inp); return (error); } INP_HASH_WLOCK(&V_udbinfo); error = in_pcbconnect(inp, nam, td->td_ucred); INP_HASH_WUNLOCK(&V_udbinfo); if (error == 0) soisconnected(so); INP_WUNLOCK(inp); return (error); } static void udp_detach(struct socket *so) { struct inpcb *inp; struct udpcb *up; inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_detach: inp == NULL")); KASSERT(inp->inp_faddr.s_addr == INADDR_ANY, ("udp_detach: not disconnected")); INP_INFO_WLOCK(&V_udbinfo); INP_WLOCK(inp); up = intoudpcb(inp); KASSERT(up != NULL, ("%s: up == NULL", __func__)); inp->inp_ppcb = NULL; in_pcbdetach(inp); in_pcbfree(inp); INP_INFO_WUNLOCK(&V_udbinfo); udp_discardcb(up); } static int udp_disconnect(struct socket *so) { struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_disconnect: inp == NULL")); INP_WLOCK(inp); if (inp->inp_faddr.s_addr == INADDR_ANY) { INP_WUNLOCK(inp); return (ENOTCONN); } INP_HASH_WLOCK(&V_udbinfo); in_pcbdisconnect(inp); inp->inp_laddr.s_addr = INADDR_ANY; INP_HASH_WUNLOCK(&V_udbinfo); SOCK_LOCK(so); so->so_state &= ~SS_ISCONNECTED; /* XXX */ SOCK_UNLOCK(so); INP_WUNLOCK(inp); return (0); } static int udp_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, struct mbuf *control, struct thread *td) { struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_send: inp == NULL")); return (udp_output(inp, m, addr, control, td)); } #endif /* INET */ int udp_shutdown(struct socket *so) { struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_shutdown: inp == NULL")); INP_WLOCK(inp); socantsendmore(so); INP_WUNLOCK(inp); return (0); } #ifdef INET struct pr_usrreqs udp_usrreqs = { .pru_abort = udp_abort, .pru_attach = udp_attach, .pru_bind = udp_bind, .pru_connect = udp_connect, .pru_control = in_control, .pru_detach = udp_detach, .pru_disconnect = udp_disconnect, .pru_peeraddr = in_getpeeraddr, .pru_send = udp_send, .pru_soreceive = soreceive_dgram, .pru_sosend = sosend_dgram, .pru_shutdown = udp_shutdown, .pru_sockaddr = in_getsockaddr, .pru_sosetlabel = in_pcbsosetlabel, .pru_close = udp_close, }; #endif /* INET */ Index: projects/largeSMP/sys/netinet6/in6.c =================================================================== --- projects/largeSMP/sys/netinet6/in6.c (revision 222782) +++ projects/largeSMP/sys/netinet6/in6.c (revision 222783) @@ -1,2700 +1,2724 @@ /*- * 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 /* * 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_lifaddr_ioctl __P((struct socket *, u_long, caddr_t, struct ifnet *, struct thread *)); static int in6_ifinit __P((struct ifnet *, struct in6_ifaddr *, struct sockaddr_in6 *, int)); static void in6_unlink_ifa(struct in6_ifaddr *, struct ifnet *); int (*faithprefix_p)(struct in6_addr *); 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; } #define ifa2ia6(ifa) ((struct in6_ifaddr *)(ifa)) #define ia62ifa(ia6) (&((ia6)->ia_ifa)) #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 error; switch (cmd) { case SIOCGETSGCNT_IN6: case SIOCGETMIFCNT_IN6: 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); } return (scope6_set(ifp, (struct scope6_id *)ifr->ifr_ifru.ifru_scope_id)); case SIOCGSCOPE6: return (scope6_get(ifp, (struct scope6_id *)ifr->ifr_ifru.ifru_scope_id)); case SIOCGSCOPE6DEF: return (scope6_get_default((struct scope6_id *) ifr->ifr_ifru.ifru_scope_id)); } switch (cmd) { case SIOCALIFADDR: if (td != NULL) { error = priv_check(td, PRIV_NET_ADDIFADDR); if (error) return (error); } return in6_lifaddr_ioctl(so, cmd, data, ifp, td); case SIOCDLIFADDR: if (td != NULL) { error = priv_check(td, PRIV_NET_DELIFADDR); if (error) return (error); } /* FALLTHROUGH */ case SIOCGLIFADDR: return in6_lifaddr_ioctl(so, cmd, data, ifp, td); } /* * 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; 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; } 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_second < time_second) { error = EINVAL; goto out; } if (lt->ia6t_pltime != ND6_INFINITE_LIFETIME && lt->ia6t_pltime + time_second < time_second) { 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: if (ifp == NULL) { error = EINVAL; goto out; } bzero(&ifr->ifr_ifru.ifru_stat, sizeof(ifr->ifr_ifru.ifru_stat)); ifr->ifr_ifru.ifru_stat = *((struct in6_ifextra *)ifp->if_afdata[AF_INET6])->in6_ifstat; break; case SIOCGIFSTAT_ICMP6: if (ifp == NULL) { error = EINVAL; goto out; } bzero(&ifr->ifr_ifru.ifru_icmp6stat, sizeof(ifr->ifr_ifru.ifru_icmp6stat)); ifr->ifr_ifru.ifru_icmp6stat = *((struct in6_ifextra *)ifp->if_afdata[AF_INET6])->icmp6_ifstat; 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_second + 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_second + ia->ia6_lifetime.ia6t_pltime; } else ia->ia6_lifetime.ia6t_preferred = 0; break; case SIOCAIFADDR_IN6: { int i; 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; } /* * 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) { break; /* we don't need to install a host route. */ } pr0.ndpr_prefix = ifra->ifra_addr; /* apply the mask for safety. */ for (i = 0; i < 4; i++) { pr0.ndpr_prefix.sin6_addr.s6_addr32[i] &= ifra->ifra_prefixmask.sin6_addr.s6_addr32[i]; } /* * 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) goto out; if (pr == NULL) { 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(); - if (error == 0 && ia) + if (error == 0 && ia) { + if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) { + /* + * Try to clear the flag when a new + * IPv6 address is added onto an + * IFDISABLED interface and it + * succeeds. + */ + 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 == NULL || ifp->if_ioctl == 0) { 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); } /* * 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(). * XXX: should this be performed under splnet()? */ int in6_update_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra, struct in6_ifaddr *ia, int flags) { int error = 0, hostIsNew = 0, plen = -1; struct sockaddr_in6 dst6; struct in6_addrlifetime *lt; struct in6_multi_mship *imm; struct in6_multi *in6m_sol; struct rtentry *rt; int delay; 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 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 */ } /* * 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 */ } /* * If this is a new address, allocate a new ifaddr and link it * into chains. */ if (ia == NULL) { hostIsNew = 1; /* * When in6_update_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 *) malloc(sizeof(*ia), M_IFADDR, M_NOWAIT); if (ia == NULL) return (ENOBUFS); bzero((caddr_t)ia, sizeof(*ia)); ifa_init(&ia->ia_ifa); 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); ia->ia6_createtime = time_second; if ((ifp->if_flags & (IFF_POINTOPOINT | IFF_LOOPBACK)) != 0) { /* * XXX: 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; } ia->ia_ifa.ifa_netmask = (struct sockaddr *)&ia->ia_prefixmask; ia->ia_ifp = ifp; ifa_ref(&ia->ia_ifa); /* if_addrhead */ IF_ADDR_LOCK(ifp); TAILQ_INSERT_TAIL(&ifp->if_addrhead, &ia->ia_ifa, ifa_link); IF_ADDR_UNLOCK(ifp); ifa_ref(&ia->ia_ifa); /* in6_ifaddrhead */ IN6_IFADDR_WLOCK(); TAILQ_INSERT_TAIL(&V_in6_ifaddrhead, ia, ia_link); IN6_IFADDR_WUNLOCK(); } /* update timestamp */ ia->ia6_updatetime = time_second; /* set prefix mask */ if (ifra->ifra_prefixmask.sin6_len) { /* * 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 && in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL) != plen) { nd6log((LOG_INFO, "in6_update_ifa: the prefix length of an" " existing (%s) address should not be changed\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr))); error = EINVAL; goto unlink; } ia->ia_prefixmask = ifra->ifra_prefixmask; } /* * 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 (see the check above.) */ if (dst6.sin6_family == AF_INET6 && !IN6_ARE_ADDR_EQUAL(&dst6.sin6_addr, &ia->ia_dstaddr.sin6_addr)) { int e; if ((ia->ia_flags & IFA_ROUTE) != 0 && (e = rtinit(&(ia->ia_ifa), (int)RTM_DELETE, RTF_HOST)) != 0) { nd6log((LOG_ERR, "in6_update_ifa: 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 = dst6; } /* * 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_second + 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_second + ia->ia6_lifetime.ia6t_pltime; } else ia->ia6_lifetime.ia6t_preferred = 0; /* reset the interface and routing table appropriately. */ if ((error = in6_ifinit(ifp, ia, &ifra->ifra_addr, hostIsNew)) != 0) goto unlink; /* * configure address flags. */ ia->ia6_flags = ifra->ifra_flags; /* * 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_second; } /* * 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; /* * We are done if we have simply modified an existing address. */ if (!hostIsNew) return (error); /* * Beyond this point, we should call in6_purgeaddr upon an error, * not just go to unlink. */ /* Join necessary multicast groups */ in6m_sol = NULL; if ((ifp->if_flags & IFF_MULTICAST) != 0) { struct sockaddr_in6 mltaddr, mltmask; struct in6_addr llsol; /* join solicited multicast addr for new host id */ bzero(&llsol, sizeof(struct in6_addr)); llsol.s6_addr32[0] = IPV6_ADDR_INT32_MLL; llsol.s6_addr32[1] = 0; llsol.s6_addr32[2] = htonl(1); llsol.s6_addr32[3] = ifra->ifra_addr.sin6_addr.s6_addr32[3]; llsol.s6_addr8[12] = 0xff; if ((error = in6_setscope(&llsol, ifp, NULL)) != 0) { /* XXX: should not happen */ log(LOG_ERR, "in6_update_ifa: " "in6_setscope failed\n"); goto cleanup; } delay = 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, &llsol, &error, delay); if (imm == NULL) { nd6log((LOG_WARNING, "in6_update_ifa: addmulti failed for " "%s on %s (errno=%d)\n", ip6_sprintf(ip6buf, &llsol), if_name(ifp), error)); goto cleanup; } LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain); in6m_sol = imm->i6mm_maddr; bzero(&mltmask, sizeof(mltmask)); mltmask.sin6_len = sizeof(struct sockaddr_in6); mltmask.sin6_family = AF_INET6; mltmask.sin6_addr = in6mask32; #define MLTMASK_LEN 4 /* mltmask's masklen (=32bit=4octet) */ /* * join link-local all-nodes address */ bzero(&mltaddr, sizeof(mltaddr)); mltaddr.sin6_len = sizeof(struct sockaddr_in6); mltaddr.sin6_family = AF_INET6; mltaddr.sin6_addr = in6addr_linklocal_allnodes; if ((error = in6_setscope(&mltaddr.sin6_addr, ifp, NULL)) != 0) goto cleanup; /* XXX: should not fail */ /* * XXX: do we really need this automatic routes? * We should probably reconsider this stuff. Most applications * actually do not need the routes, since they usually specify * the outgoing interface. */ rt = rtalloc1((struct sockaddr *)&mltaddr, 0, 0UL); if (rt) { /* XXX: only works in !SCOPEDROUTING case. */ if (memcmp(&mltaddr.sin6_addr, &((struct sockaddr_in6 *)rt_key(rt))->sin6_addr, MLTMASK_LEN)) { RTFREE_LOCKED(rt); rt = NULL; } } if (!rt) { error = rtrequest(RTM_ADD, (struct sockaddr *)&mltaddr, (struct sockaddr *)&ia->ia_addr, (struct sockaddr *)&mltmask, RTF_UP, (struct rtentry **)0); if (error) goto cleanup; } else { RTFREE_LOCKED(rt); } imm = in6_joingroup(ifp, &mltaddr.sin6_addr, &error, 0); if (!imm) { nd6log((LOG_WARNING, "in6_update_ifa: addmulti failed for " "%s on %s (errno=%d)\n", ip6_sprintf(ip6buf, &mltaddr.sin6_addr), 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 doesn't 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.sin6_addr) == 0) { imm = in6_joingroup(ifp, &mltaddr.sin6_addr, &error, delay); /* XXX jinmei */ if (!imm) { nd6log((LOG_WARNING, "in6_update_ifa: " "addmulti failed for %s on %s " "(errno=%d)\n", ip6_sprintf(ip6buf, &mltaddr.sin6_addr), 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.sin6_addr = in6addr_nodelocal_allnodes; if ((error = in6_setscope(&mltaddr.sin6_addr, ifp, NULL)) != 0) goto cleanup; /* XXX: should not fail */ /* XXX: again, do we really need the route? */ rt = rtalloc1((struct sockaddr *)&mltaddr, 0, 0UL); if (rt) { if (memcmp(&mltaddr.sin6_addr, &((struct sockaddr_in6 *)rt_key(rt))->sin6_addr, MLTMASK_LEN)) { RTFREE_LOCKED(rt); rt = NULL; } } if (!rt) { error = rtrequest(RTM_ADD, (struct sockaddr *)&mltaddr, (struct sockaddr *)&ia->ia_addr, (struct sockaddr *)&mltmask, RTF_UP, (struct rtentry **)0); if (error) goto cleanup; } else RTFREE_LOCKED(rt); imm = in6_joingroup(ifp, &mltaddr.sin6_addr, &error, 0); if (!imm) { nd6log((LOG_WARNING, "in6_update_ifa: " "addmulti failed for %s on %s " "(errno=%d)\n", ip6_sprintf(ip6buf, &mltaddr.sin6_addr), if_name(ifp), error)); goto cleanup; } LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain); #undef MLTMASK_LEN } /* * 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 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); } KASSERT(hostIsNew, ("in6_update_ifa: !hostIsNew")); ifa_free(&ia->ia_ifa); return (error); unlink: /* * XXX: if a change of an existing address failed, keep the entry * anyway. */ if (hostIsNew) { in6_unlink_ifa(ia, ifp); ifa_free(&ia->ia_ifa); } return (error); cleanup: KASSERT(hostIsNew, ("in6_update_ifa: cleanup: !hostIsNew")); ifa_free(&ia->ia_ifa); in6_purgeaddr(&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; struct sockaddr_in6 mltaddr, mltmask; struct rtentry rt0; struct sockaddr_dl gateway; struct sockaddr_in6 mask, addr; int plen, error; struct rtentry *rt; struct ifaddr *ifa0, *nifa; /* * find another IPv6 address as the gateway for the * link-local and node-local all-nodes multicast * address routes */ IF_ADDR_LOCK(ifp); TAILQ_FOREACH_SAFE(ifa0, &ifp->if_addrhead, ifa_link, nifa) { if ((ifa0->ifa_addr->sa_family != AF_INET6) || memcmp(&satosin6(ifa0->ifa_addr)->sin6_addr, &ia->ia_addr.sin6_addr, sizeof(struct in6_addr)) == 0) continue; else break; } if (ifa0 != NULL) ifa_ref(ifa0); IF_ADDR_UNLOCK(ifp); /* * 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); IF_AFDATA_LOCK(ifp); lla_lookup(LLTABLE6(ifp), (LLE_DELETE | LLE_IFADDR), (struct sockaddr *)&ia->ia_addr); IF_AFDATA_UNLOCK(ifp); /* * initialize for rtmsg generation */ bzero(&gateway, sizeof(gateway)); gateway.sdl_len = sizeof(gateway); gateway.sdl_family = AF_LINK; gateway.sdl_nlen = 0; gateway.sdl_alen = ifp->if_addrlen; /* */ bzero(&rt0, sizeof(rt0)); rt0.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(&rt0) = (struct sockaddr *)&mask; rt_key(&rt0) = (struct sockaddr *)&addr; rt0.rt_flags = RTF_HOST | RTF_STATIC; rt_newaddrmsg(RTM_DELETE, ifa, 0, &rt0); /* * leave from multicast groups we have joined for the interface */ while ((imm = ia->ia6_memberships.lh_first) != NULL) { LIST_REMOVE(imm, i6mm_chain); in6_leavegroup(imm); } /* * remove the link-local all-nodes address */ bzero(&mltmask, sizeof(mltmask)); mltmask.sin6_len = sizeof(struct sockaddr_in6); mltmask.sin6_family = AF_INET6; mltmask.sin6_addr = in6mask32; bzero(&mltaddr, sizeof(mltaddr)); mltaddr.sin6_len = sizeof(struct sockaddr_in6); mltaddr.sin6_family = AF_INET6; mltaddr.sin6_addr = in6addr_linklocal_allnodes; if ((error = in6_setscope(&mltaddr.sin6_addr, ifp, NULL)) != 0) goto cleanup; rt = rtalloc1((struct sockaddr *)&mltaddr, 0, 0UL); if (rt != NULL && rt->rt_gateway != NULL && (memcmp(&satosin6(rt->rt_gateway)->sin6_addr, &ia->ia_addr.sin6_addr, sizeof(ia->ia_addr.sin6_addr)) == 0)) { /* * if no more IPv6 address exists on this interface * then remove the multicast address route */ if (ifa0 == NULL) { memcpy(&mltaddr.sin6_addr, &satosin6(rt_key(rt))->sin6_addr, sizeof(mltaddr.sin6_addr)); RTFREE_LOCKED(rt); error = rtrequest(RTM_DELETE, (struct sockaddr *)&mltaddr, (struct sockaddr *)&ia->ia_addr, (struct sockaddr *)&mltmask, RTF_UP, (struct rtentry **)0); if (error) log(LOG_INFO, "in6_purgeaddr: link-local all-nodes" "multicast address deletion error\n"); } else { /* * replace the gateway of the route */ struct sockaddr_in6 sa; bzero(&sa, sizeof(sa)); sa.sin6_len = sizeof(struct sockaddr_in6); sa.sin6_family = AF_INET6; memcpy(&sa.sin6_addr, &satosin6(ifa0->ifa_addr)->sin6_addr, sizeof(sa.sin6_addr)); in6_setscope(&sa.sin6_addr, ifa0->ifa_ifp, NULL); memcpy(rt->rt_gateway, &sa, sizeof(sa)); RTFREE_LOCKED(rt); } } else { if (rt != NULL) RTFREE_LOCKED(rt); } /* * remove the node-local all-nodes address */ mltaddr.sin6_addr = in6addr_nodelocal_allnodes; if ((error = in6_setscope(&mltaddr.sin6_addr, ifp, NULL)) != 0) goto cleanup; rt = rtalloc1((struct sockaddr *)&mltaddr, 0, 0UL); if (rt != NULL && rt->rt_gateway != NULL && (memcmp(&satosin6(rt->rt_gateway)->sin6_addr, &ia->ia_addr.sin6_addr, sizeof(ia->ia_addr.sin6_addr)) == 0)) { /* * if no more IPv6 address exists on this interface * then remove the multicast address route */ if (ifa0 == NULL) { memcpy(&mltaddr.sin6_addr, &satosin6(rt_key(rt))->sin6_addr, sizeof(mltaddr.sin6_addr)); RTFREE_LOCKED(rt); error = rtrequest(RTM_DELETE, (struct sockaddr *)&mltaddr, (struct sockaddr *)&ia->ia_addr, (struct sockaddr *)&mltmask, RTF_UP, (struct rtentry **)0); if (error) log(LOG_INFO, "in6_purgeaddr: node-local all-nodes" "multicast address deletion error\n"); } else { /* * replace the gateway of the route */ struct sockaddr_in6 sa; bzero(&sa, sizeof(sa)); sa.sin6_len = sizeof(struct sockaddr_in6); sa.sin6_family = AF_INET6; memcpy(&sa.sin6_addr, &satosin6(ifa0->ifa_addr)->sin6_addr, sizeof(sa.sin6_addr)); in6_setscope(&sa.sin6_addr, ifa0->ifa_ifp, NULL); memcpy(rt->rt_gateway, &sa, sizeof(sa)); RTFREE_LOCKED(rt); } } else { if (rt != NULL) RTFREE_LOCKED(rt); } cleanup: plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); /* XXX */ if ((ia->ia_flags & IFA_ROUTE) && plen == 128) { int error; struct sockaddr *dstaddr; /* * use the interface address if configuring an * interface address with a /128 prefix len */ if (ia->ia_dstaddr.sin6_family == AF_INET6) dstaddr = (struct sockaddr *)&ia->ia_dstaddr; else dstaddr = (struct sockaddr *)&ia->ia_addr; error = rtrequest(RTM_DELETE, (struct sockaddr *)dstaddr, (struct sockaddr *)&ia->ia_addr, (struct sockaddr *)&ia->ia_prefixmask, ia->ia_flags | RTF_HOST, NULL); if (error != 0) return; ia->ia_flags &= ~IFA_ROUTE; } if (ifa0 != NULL) ifa_free(ifa0); in6_unlink_ifa(ia, ifp); } static void in6_unlink_ifa(struct in6_ifaddr *ia, struct ifnet *ifp) { int s = splnet(); IF_ADDR_LOCK(ifp); TAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifa_link); IF_ADDR_UNLOCK(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); 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 " "%p has no prefix\n", 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 */ splx(s); } void in6_purgeif(struct ifnet *ifp) { struct ifaddr *ifa, *nifa; TAILQ_FOREACH_SAFE(ifa, &ifp->if_addrhead, ifa_link, nifa) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; in6_purgeaddr(ifa); } in6_ifdetach(ifp); } /* * SIOC[GAD]LIFADDR. * SIOCGLIFADDR: get first address. (?) * SIOCGLIFADDR with IFLR_PREFIX: * get first address that matches the specified prefix. * SIOCALIFADDR: add the specified address. * SIOCALIFADDR with IFLR_PREFIX: * add the specified prefix, filling hostid part from * the first link-local address. prefixlen must be <= 64. * SIOCDLIFADDR: delete the specified address. * SIOCDLIFADDR with IFLR_PREFIX: * delete the first address that matches the specified prefix. * return values: * EINVAL on invalid parameters * EADDRNOTAVAIL on prefix match failed/specified address not found * other values may be returned from in6_ioctl() * * NOTE: SIOCALIFADDR(with IFLR_PREFIX set) allows prefixlen less than 64. * this is to accomodate address naming scheme other than RFC2374, * in the future. * RFC2373 defines interface id to be 64bit, but it allows non-RFC2374 * address encoding scheme. (see figure on page 8) */ static int in6_lifaddr_ioctl(struct socket *so, u_long cmd, caddr_t data, struct ifnet *ifp, struct thread *td) { struct if_laddrreq *iflr = (struct if_laddrreq *)data; struct ifaddr *ifa; struct sockaddr *sa; /* sanity checks */ if (!data || !ifp) { panic("invalid argument to in6_lifaddr_ioctl"); /* NOTREACHED */ } switch (cmd) { case SIOCGLIFADDR: /* address must be specified on GET with IFLR_PREFIX */ if ((iflr->flags & IFLR_PREFIX) == 0) break; /* FALLTHROUGH */ case SIOCALIFADDR: case SIOCDLIFADDR: /* address must be specified on ADD and DELETE */ sa = (struct sockaddr *)&iflr->addr; if (sa->sa_family != AF_INET6) return EINVAL; if (sa->sa_len != sizeof(struct sockaddr_in6)) return EINVAL; /* XXX need improvement */ sa = (struct sockaddr *)&iflr->dstaddr; if (sa->sa_family && sa->sa_family != AF_INET6) return EINVAL; if (sa->sa_len && sa->sa_len != sizeof(struct sockaddr_in6)) return EINVAL; break; default: /* shouldn't happen */ #if 0 panic("invalid cmd to in6_lifaddr_ioctl"); /* NOTREACHED */ #else return EOPNOTSUPP; #endif } if (sizeof(struct in6_addr) * 8 < iflr->prefixlen) return EINVAL; switch (cmd) { case SIOCALIFADDR: { struct in6_aliasreq ifra; struct in6_addr *hostid = NULL; int prefixlen; ifa = NULL; if ((iflr->flags & IFLR_PREFIX) != 0) { struct sockaddr_in6 *sin6; /* * hostid is to fill in the hostid part of the * address. hostid points to the first link-local * address attached to the interface. */ ifa = (struct ifaddr *)in6ifa_ifpforlinklocal(ifp, 0); if (!ifa) return EADDRNOTAVAIL; hostid = IFA_IN6(ifa); /* prefixlen must be <= 64. */ if (64 < iflr->prefixlen) return EINVAL; prefixlen = iflr->prefixlen; /* hostid part must be zero. */ sin6 = (struct sockaddr_in6 *)&iflr->addr; if (sin6->sin6_addr.s6_addr32[2] != 0 || sin6->sin6_addr.s6_addr32[3] != 0) { return EINVAL; } } else prefixlen = iflr->prefixlen; /* copy args to in6_aliasreq, perform ioctl(SIOCAIFADDR_IN6). */ bzero(&ifra, sizeof(ifra)); bcopy(iflr->iflr_name, ifra.ifra_name, sizeof(ifra.ifra_name)); bcopy(&iflr->addr, &ifra.ifra_addr, ((struct sockaddr *)&iflr->addr)->sa_len); if (hostid) { /* fill in hostid part */ ifra.ifra_addr.sin6_addr.s6_addr32[2] = hostid->s6_addr32[2]; ifra.ifra_addr.sin6_addr.s6_addr32[3] = hostid->s6_addr32[3]; } if (((struct sockaddr *)&iflr->dstaddr)->sa_family) { /* XXX */ bcopy(&iflr->dstaddr, &ifra.ifra_dstaddr, ((struct sockaddr *)&iflr->dstaddr)->sa_len); if (hostid) { ifra.ifra_dstaddr.sin6_addr.s6_addr32[2] = hostid->s6_addr32[2]; ifra.ifra_dstaddr.sin6_addr.s6_addr32[3] = hostid->s6_addr32[3]; } } if (ifa != NULL) ifa_free(ifa); ifra.ifra_prefixmask.sin6_len = sizeof(struct sockaddr_in6); in6_prefixlen2mask(&ifra.ifra_prefixmask.sin6_addr, prefixlen); ifra.ifra_flags = iflr->flags & ~IFLR_PREFIX; return in6_control(so, SIOCAIFADDR_IN6, (caddr_t)&ifra, ifp, td); } case SIOCGLIFADDR: case SIOCDLIFADDR: { struct in6_ifaddr *ia; struct in6_addr mask, candidate, match; struct sockaddr_in6 *sin6; int cmp; bzero(&mask, sizeof(mask)); if (iflr->flags & IFLR_PREFIX) { /* lookup a prefix rather than address. */ in6_prefixlen2mask(&mask, iflr->prefixlen); sin6 = (struct sockaddr_in6 *)&iflr->addr; bcopy(&sin6->sin6_addr, &match, sizeof(match)); match.s6_addr32[0] &= mask.s6_addr32[0]; match.s6_addr32[1] &= mask.s6_addr32[1]; match.s6_addr32[2] &= mask.s6_addr32[2]; match.s6_addr32[3] &= mask.s6_addr32[3]; /* if you set extra bits, that's wrong */ if (bcmp(&match, &sin6->sin6_addr, sizeof(match))) return EINVAL; cmp = 1; } else { if (cmd == SIOCGLIFADDR) { /* on getting an address, take the 1st match */ cmp = 0; /* XXX */ } else { /* on deleting an address, do exact match */ in6_prefixlen2mask(&mask, 128); sin6 = (struct sockaddr_in6 *)&iflr->addr; bcopy(&sin6->sin6_addr, &match, sizeof(match)); cmp = 1; } } IF_ADDR_LOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; if (!cmp) break; /* * XXX: this is adhoc, but is necessary to allow * a user to specify fe80::/64 (not /10) for a * link-local address. */ bcopy(IFA_IN6(ifa), &candidate, sizeof(candidate)); in6_clearscope(&candidate); candidate.s6_addr32[0] &= mask.s6_addr32[0]; candidate.s6_addr32[1] &= mask.s6_addr32[1]; candidate.s6_addr32[2] &= mask.s6_addr32[2]; candidate.s6_addr32[3] &= mask.s6_addr32[3]; if (IN6_ARE_ADDR_EQUAL(&candidate, &match)) break; } IF_ADDR_UNLOCK(ifp); if (!ifa) return EADDRNOTAVAIL; ia = ifa2ia6(ifa); if (cmd == SIOCGLIFADDR) { int error; /* fill in the if_laddrreq structure */ bcopy(&ia->ia_addr, &iflr->addr, ia->ia_addr.sin6_len); error = sa6_recoverscope( (struct sockaddr_in6 *)&iflr->addr); if (error != 0) return (error); if ((ifp->if_flags & IFF_POINTOPOINT) != 0) { bcopy(&ia->ia_dstaddr, &iflr->dstaddr, ia->ia_dstaddr.sin6_len); error = sa6_recoverscope( (struct sockaddr_in6 *)&iflr->dstaddr); if (error != 0) return (error); } else bzero(&iflr->dstaddr, sizeof(iflr->dstaddr)); iflr->prefixlen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); iflr->flags = ia->ia6_flags; /* XXX */ return 0; } else { struct in6_aliasreq ifra; /* fill in6_aliasreq and do ioctl(SIOCDIFADDR_IN6) */ bzero(&ifra, sizeof(ifra)); bcopy(iflr->iflr_name, ifra.ifra_name, sizeof(ifra.ifra_name)); bcopy(&ia->ia_addr, &ifra.ifra_addr, ia->ia_addr.sin6_len); if ((ifp->if_flags & IFF_POINTOPOINT) != 0) { bcopy(&ia->ia_dstaddr, &ifra.ifra_dstaddr, ia->ia_dstaddr.sin6_len); } else { bzero(&ifra.ifra_dstaddr, sizeof(ifra.ifra_dstaddr)); } bcopy(&ia->ia_prefixmask, &ifra.ifra_dstaddr, ia->ia_prefixmask.sin6_len); ifra.ifra_flags = ia->ia6_flags; return in6_control(so, SIOCDIFADDR_IN6, (caddr_t)&ifra, ifp, td); } } } return EOPNOTSUPP; /* just for safety */ } /* * Initialize an interface's intetnet6 address * and routing table entry. */ static int in6_ifinit(struct ifnet *ifp, struct in6_ifaddr *ia, struct sockaddr_in6 *sin6, int newhost) { int error = 0, plen, ifacount = 0; int s = splimp(); struct ifaddr *ifa; /* * Give the interface a chance to initialize * if this is its first address, * and to validate the address if necessary. */ IF_ADDR_LOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; ifacount++; } IF_ADDR_UNLOCK(ifp); ia->ia_addr = *sin6; if (ifacount <= 1 && ifp->if_ioctl) { error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia); if (error) { splx(s); return (error); } } splx(s); ia->ia_ifa.ifa_metric = ifp->if_metric; /* we could do in(6)_socktrim here, but just omit it at this moment. */ /* * Special case: * 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; error = rtrequest(RTM_ADD, (struct sockaddr *)&ia->ia_dstaddr, (struct sockaddr *)&ia->ia_addr, (struct sockaddr *)&ia->ia_prefixmask, ia->ia_flags | rtflags, NULL); if (error != 0) return (error); ia->ia_flags |= IFA_ROUTE; } /* * add a loopback route to self */ if (!(ia->ia_flags & IFA_ROUTE) && (V_nd6_useloopback || (ifp->if_flags & IFF_LOOPBACK))) { error = ifa_add_loopback_route((struct ifaddr *)ia, (struct sockaddr *)&ia->ia_addr); if (error == 0) ia->ia_flags |= IFA_RTSELF; } /* Add ownaddr as loopback rtentry, if necessary (ex. on p2p link). */ if (newhost) { struct llentry *ln; struct rtentry rt; struct sockaddr_dl gateway; struct sockaddr_in6 mask, addr; IF_AFDATA_LOCK(ifp); ia->ia_ifa.ifa_rtrequest = NULL; /* XXX QL * we need to report rt_newaddrmsg */ ln = lla_lookup(LLTABLE6(ifp), (LLE_CREATE | LLE_IFADDR | LLE_EXCLUSIVE), (struct sockaddr *)&ia->ia_addr); IF_AFDATA_UNLOCK(ifp); if (ln != NULL) { ln->la_expire = 0; /* for IPv6 this means permanent */ ln->ln_state = ND6_LLINFO_REACHABLE; /* * initialize for rtmsg generation */ bzero(&gateway, sizeof(gateway)); gateway.sdl_len = sizeof(gateway); gateway.sdl_family = AF_LINK; gateway.sdl_nlen = 0; gateway.sdl_alen = 6; memcpy(gateway.sdl_data, &ln->ll_addr.mac_aligned, sizeof(ln->ll_addr)); /* */ LLE_WUNLOCK(ln); } 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_UP | RTF_HOST | RTF_STATIC; rt_newaddrmsg(RTM_ADD, &ia->ia_ifa, 0, &rt); } 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_LOCK(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_UNLOCK(ifp); return ((struct in6_ifaddr *)ifa); } /* * 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_LOCK(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_UNLOCK(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); } int in6_is_addr_deprecated(struct sockaddr_in6 *sa6) { struct in6_ifaddr *ia; IN6_IFADDR_RLOCK(); TAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) { if (IN6_ARE_ADDR_EQUAL(&ia->ia_addr.sin6_addr, &sa6->sin6_addr) && (ia->ia6_flags & IN6_IFF_DEPRECATED) != 0) { IN6_IFADDR_RUNLOCK(); return (1); /* true */ } /* XXX: do we still have to go thru the rest of the list? */ } 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_LOCK(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_UNLOCK(ifp); return (besta); } IF_ADDR_UNLOCK(ifp); IN6_IFADDR_RLOCK(); 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); IN6_IFADDR_RUNLOCK(); return (struct in6_ifaddr *)ifa; } IN6_IFADDR_RUNLOCK(); /* use the last-resort values, that are, deprecated addresses */ if (dep[0]) return dep[0]; if (dep[1]) return dep[1]; 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_LOCK(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_UNLOCK(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) return (0); switch (ifp->if_type) { #ifdef IFT_DUMMY case IFT_DUMMY: #endif case IFT_FAITH: /* * These interfaces do not have the IFF_LOOPBACK flag, * but loop packets back. We do not have to do DAD on such * interfaces. We should even omit it, because loop-backed * NS would confuse the DAD procedure. */ return (0); default: /* * 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(); for (ifp = TAILQ_FIRST(&V_ifnet); ifp; ifp = TAILQ_NEXT(ifp, if_list)) { /* 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; }; 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_DONTWAIT | 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_LOCK_INIT(&lle->base); callout_init_rw(&lle->base.ln_timer_ch, &lle->base.lle_lock, CALLOUT_RETURNUNLOCKED); return &lle->base; } /* * 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 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; register int i; /* * (flags & LLE_STATIC) means deleting all entries * including static ND6 entries */ 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( &((struct sockaddr_in6 *)L3_ADDR(lle))->sin6_addr, &pfx->sin6_addr, &msk->sin6_addr) && ((flags & LLE_STATIC) || !(lle->la_flags & LLE_STATIC))) { int canceled; canceled = callout_drain(&lle->la_timer); LLE_WLOCK(lle); if (canceled) LLE_REMREF(lle); llentry_free(lle); } } } } 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)); /* XXX rtalloc1 should take a const param */ rt = rtalloc1(__DECONST(struct sockaddr *, l3addr), 0, 0); 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); /* * 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; 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; LLE_WUNLOCK(lle); #ifdef DIAGNOSTIC log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle); #endif } 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)); /* 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; ext = (struct in6_ifextra *)malloc(sizeof(*ext), M_IFADDR, M_WAITOK); bzero(ext, sizeof(*ext)); ext->in6_ifstat = (struct in6_ifstat *)malloc(sizeof(struct in6_ifstat), M_IFADDR, M_WAITOK); bzero(ext->in6_ifstat, sizeof(*ext->in6_ifstat)); ext->icmp6_ifstat = (struct icmp6_ifstat *)malloc(sizeof(struct icmp6_ifstat), M_IFADDR, M_WAITOK); bzero(ext->icmp6_ifstat, sizeof(*ext->icmp6_ifstat)); 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_new = in6_lltable_new; ext->lltable->llt_free = in6_lltable_free; ext->lltable->llt_prefix_free = in6_lltable_prefix_free; ext->lltable->llt_rtcheck = in6_lltable_rtcheck; ext->lltable->llt_lookup = in6_lltable_lookup; ext->lltable->llt_dump = in6_lltable_dump; } ext->mld_ifinfo = mld_domifattach(ifp); return ext; } 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); free(ext->in6_ifstat, M_IFADDR); 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/largeSMP/sys/netinet6/in6.h =================================================================== --- projects/largeSMP/sys/netinet6/in6.h (revision 222782) +++ projects/largeSMP/sys/netinet6/in6.h (revision 222783) @@ -1,708 +1,711 @@ /*- * 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.h,v 1.89 2001/05/27 13:28:35 itojun Exp $ */ /*- * Copyright (c) 1982, 1986, 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. * * @(#)in.h 8.3 (Berkeley) 1/3/94 * $FreeBSD$ */ #ifndef __KAME_NETINET_IN_H_INCLUDED_ #error "do not include netinet6/in6.h directly, include netinet/in.h. see RFC2553" #endif #ifndef _NETINET6_IN6_H_ #define _NETINET6_IN6_H_ /* * Identification of the network protocol stack * for *BSD-current/release: http://www.kame.net/dev/cvsweb.cgi/kame/COVERAGE * has the table of implementation/integration differences. */ #define __KAME__ #define __KAME_VERSION "FreeBSD" /* * IPv6 port allocation rules should mirror the IPv4 rules and are controlled * by the net.inet.ip.portrange sysctl tree. The following defines exist * for compatibility with userland applications that need them. */ #if __BSD_VISIBLE #define IPV6PORT_RESERVED 1024 #define IPV6PORT_ANONMIN 49152 #define IPV6PORT_ANONMAX 65535 #define IPV6PORT_RESERVEDMIN 600 #define IPV6PORT_RESERVEDMAX (IPV6PORT_RESERVED-1) #endif /* * IPv6 address */ struct in6_addr { union { uint8_t __u6_addr8[16]; uint16_t __u6_addr16[8]; uint32_t __u6_addr32[4]; } __u6_addr; /* 128-bit IP6 address */ }; #define s6_addr __u6_addr.__u6_addr8 #ifdef _KERNEL /* XXX nonstandard */ #define s6_addr8 __u6_addr.__u6_addr8 #define s6_addr16 __u6_addr.__u6_addr16 #define s6_addr32 __u6_addr.__u6_addr32 #endif #define INET6_ADDRSTRLEN 46 /* * XXX missing POSIX.1-2001 macro IPPROTO_IPV6. */ /* * Socket address for IPv6 */ #if __BSD_VISIBLE #define SIN6_LEN #endif struct sockaddr_in6 { uint8_t sin6_len; /* length of this struct */ sa_family_t sin6_family; /* AF_INET6 */ in_port_t sin6_port; /* Transport layer port # */ uint32_t sin6_flowinfo; /* IP6 flow information */ struct in6_addr sin6_addr; /* IP6 address */ uint32_t sin6_scope_id; /* scope zone index */ }; /* * Local definition for masks */ #ifdef _KERNEL /* XXX nonstandard */ #define IN6MASK0 {{{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }}} #define IN6MASK32 {{{ 0xff, 0xff, 0xff, 0xff, 0x00, 0x00, 0x00, 0x00, \ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 }}} #define IN6MASK64 {{{ 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, \ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 }}} #define IN6MASK96 {{{ 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, \ 0xff, 0xff, 0xff, 0xff, 0x00, 0x00, 0x00, 0x00 }}} #define IN6MASK128 {{{ 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, \ 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }}} #endif #ifdef _KERNEL extern const struct sockaddr_in6 sa6_any; extern const struct in6_addr in6mask0; extern const struct in6_addr in6mask32; extern const struct in6_addr in6mask64; extern const struct in6_addr in6mask96; extern const struct in6_addr in6mask128; #endif /* _KERNEL */ /* * Macros started with IPV6_ADDR is KAME local */ #ifdef _KERNEL /* XXX nonstandard */ #if _BYTE_ORDER == _BIG_ENDIAN #define IPV6_ADDR_INT32_ONE 1 #define IPV6_ADDR_INT32_TWO 2 #define IPV6_ADDR_INT32_MNL 0xff010000 #define IPV6_ADDR_INT32_MLL 0xff020000 #define IPV6_ADDR_INT32_SMP 0x0000ffff #define IPV6_ADDR_INT16_ULL 0xfe80 #define IPV6_ADDR_INT16_USL 0xfec0 #define IPV6_ADDR_INT16_MLL 0xff02 #elif _BYTE_ORDER == _LITTLE_ENDIAN #define IPV6_ADDR_INT32_ONE 0x01000000 #define IPV6_ADDR_INT32_TWO 0x02000000 #define IPV6_ADDR_INT32_MNL 0x000001ff #define IPV6_ADDR_INT32_MLL 0x000002ff #define IPV6_ADDR_INT32_SMP 0xffff0000 #define IPV6_ADDR_INT16_ULL 0x80fe #define IPV6_ADDR_INT16_USL 0xc0fe #define IPV6_ADDR_INT16_MLL 0x02ff #endif #endif /* * Definition of some useful macros to handle IP6 addresses */ #if __BSD_VISIBLE #define IN6ADDR_ANY_INIT \ {{{ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, \ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 }}} #define IN6ADDR_LOOPBACK_INIT \ {{{ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, \ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01 }}} #define IN6ADDR_NODELOCAL_ALLNODES_INIT \ {{{ 0xff, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, \ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01 }}} #define IN6ADDR_INTFACELOCAL_ALLNODES_INIT \ {{{ 0xff, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, \ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01 }}} #define IN6ADDR_LINKLOCAL_ALLNODES_INIT \ {{{ 0xff, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, \ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01 }}} #define IN6ADDR_LINKLOCAL_ALLROUTERS_INIT \ {{{ 0xff, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, \ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02 }}} #define IN6ADDR_LINKLOCAL_ALLV2ROUTERS_INIT \ {{{ 0xff, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, \ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x16 }}} #endif extern const struct in6_addr in6addr_any; extern const struct in6_addr in6addr_loopback; #if __BSD_VISIBLE extern const struct in6_addr in6addr_nodelocal_allnodes; extern const struct in6_addr in6addr_linklocal_allnodes; extern const struct in6_addr in6addr_linklocal_allrouters; extern const struct in6_addr in6addr_linklocal_allv2routers; #endif /* * Equality * NOTE: Some of kernel programming environment (for example, openbsd/sparc) * does not supply memcmp(). For userland memcmp() is preferred as it is * in ANSI standard. */ #ifdef _KERNEL #define IN6_ARE_ADDR_EQUAL(a, b) \ (bcmp(&(a)->s6_addr[0], &(b)->s6_addr[0], sizeof(struct in6_addr)) == 0) #else #if __BSD_VISIBLE #define IN6_ARE_ADDR_EQUAL(a, b) \ (memcmp(&(a)->s6_addr[0], &(b)->s6_addr[0], sizeof(struct in6_addr)) == 0) #endif #endif /* * Unspecified */ #define IN6_IS_ADDR_UNSPECIFIED(a) \ ((*(const u_int32_t *)(const void *)(&(a)->s6_addr[0]) == 0) && \ (*(const u_int32_t *)(const void *)(&(a)->s6_addr[4]) == 0) && \ (*(const u_int32_t *)(const void *)(&(a)->s6_addr[8]) == 0) && \ (*(const u_int32_t *)(const void *)(&(a)->s6_addr[12]) == 0)) /* * Loopback */ #define IN6_IS_ADDR_LOOPBACK(a) \ ((*(const u_int32_t *)(const void *)(&(a)->s6_addr[0]) == 0) && \ (*(const u_int32_t *)(const void *)(&(a)->s6_addr[4]) == 0) && \ (*(const u_int32_t *)(const void *)(&(a)->s6_addr[8]) == 0) && \ (*(const u_int32_t *)(const void *)(&(a)->s6_addr[12]) == ntohl(1))) /* * IPv4 compatible */ #define IN6_IS_ADDR_V4COMPAT(a) \ ((*(const u_int32_t *)(const void *)(&(a)->s6_addr[0]) == 0) && \ (*(const u_int32_t *)(const void *)(&(a)->s6_addr[4]) == 0) && \ (*(const u_int32_t *)(const void *)(&(a)->s6_addr[8]) == 0) && \ (*(const u_int32_t *)(const void *)(&(a)->s6_addr[12]) != 0) && \ (*(const u_int32_t *)(const void *)(&(a)->s6_addr[12]) != ntohl(1))) /* * Mapped */ #define IN6_IS_ADDR_V4MAPPED(a) \ ((*(const u_int32_t *)(const void *)(&(a)->s6_addr[0]) == 0) && \ (*(const u_int32_t *)(const void *)(&(a)->s6_addr[4]) == 0) && \ (*(const u_int32_t *)(const void *)(&(a)->s6_addr[8]) == ntohl(0x0000ffff))) /* * KAME Scope Values */ #ifdef _KERNEL /* XXX nonstandard */ #define IPV6_ADDR_SCOPE_NODELOCAL 0x01 #define IPV6_ADDR_SCOPE_INTFACELOCAL 0x01 #define IPV6_ADDR_SCOPE_LINKLOCAL 0x02 #define IPV6_ADDR_SCOPE_SITELOCAL 0x05 #define IPV6_ADDR_SCOPE_ORGLOCAL 0x08 /* just used in this file */ #define IPV6_ADDR_SCOPE_GLOBAL 0x0e #else #define __IPV6_ADDR_SCOPE_NODELOCAL 0x01 #define __IPV6_ADDR_SCOPE_INTFACELOCAL 0x01 #define __IPV6_ADDR_SCOPE_LINKLOCAL 0x02 #define __IPV6_ADDR_SCOPE_SITELOCAL 0x05 #define __IPV6_ADDR_SCOPE_ORGLOCAL 0x08 /* just used in this file */ #define __IPV6_ADDR_SCOPE_GLOBAL 0x0e #endif /* * Unicast Scope * Note that we must check topmost 10 bits only, not 16 bits (see RFC2373). */ #define IN6_IS_ADDR_LINKLOCAL(a) \ (((a)->s6_addr[0] == 0xfe) && (((a)->s6_addr[1] & 0xc0) == 0x80)) #define IN6_IS_ADDR_SITELOCAL(a) \ (((a)->s6_addr[0] == 0xfe) && (((a)->s6_addr[1] & 0xc0) == 0xc0)) /* * Multicast */ #define IN6_IS_ADDR_MULTICAST(a) ((a)->s6_addr[0] == 0xff) #ifdef _KERNEL /* XXX nonstandard */ #define IPV6_ADDR_MC_SCOPE(a) ((a)->s6_addr[1] & 0x0f) #else #define __IPV6_ADDR_MC_SCOPE(a) ((a)->s6_addr[1] & 0x0f) #endif /* * Multicast Scope */ #ifdef _KERNEL /* refers nonstandard items */ #define IN6_IS_ADDR_MC_NODELOCAL(a) \ (IN6_IS_ADDR_MULTICAST(a) && \ (IPV6_ADDR_MC_SCOPE(a) == IPV6_ADDR_SCOPE_NODELOCAL)) #define IN6_IS_ADDR_MC_INTFACELOCAL(a) \ (IN6_IS_ADDR_MULTICAST(a) && \ (IPV6_ADDR_MC_SCOPE(a) == IPV6_ADDR_SCOPE_INTFACELOCAL)) #define IN6_IS_ADDR_MC_LINKLOCAL(a) \ (IN6_IS_ADDR_MULTICAST(a) && \ (IPV6_ADDR_MC_SCOPE(a) == IPV6_ADDR_SCOPE_LINKLOCAL)) #define IN6_IS_ADDR_MC_SITELOCAL(a) \ (IN6_IS_ADDR_MULTICAST(a) && \ (IPV6_ADDR_MC_SCOPE(a) == IPV6_ADDR_SCOPE_SITELOCAL)) #define IN6_IS_ADDR_MC_ORGLOCAL(a) \ (IN6_IS_ADDR_MULTICAST(a) && \ (IPV6_ADDR_MC_SCOPE(a) == IPV6_ADDR_SCOPE_ORGLOCAL)) #define IN6_IS_ADDR_MC_GLOBAL(a) \ (IN6_IS_ADDR_MULTICAST(a) && \ (IPV6_ADDR_MC_SCOPE(a) == IPV6_ADDR_SCOPE_GLOBAL)) #else #define IN6_IS_ADDR_MC_NODELOCAL(a) \ (IN6_IS_ADDR_MULTICAST(a) && \ (__IPV6_ADDR_MC_SCOPE(a) == __IPV6_ADDR_SCOPE_NODELOCAL)) #define IN6_IS_ADDR_MC_LINKLOCAL(a) \ (IN6_IS_ADDR_MULTICAST(a) && \ (__IPV6_ADDR_MC_SCOPE(a) == __IPV6_ADDR_SCOPE_LINKLOCAL)) #define IN6_IS_ADDR_MC_SITELOCAL(a) \ (IN6_IS_ADDR_MULTICAST(a) && \ (__IPV6_ADDR_MC_SCOPE(a) == __IPV6_ADDR_SCOPE_SITELOCAL)) #define IN6_IS_ADDR_MC_ORGLOCAL(a) \ (IN6_IS_ADDR_MULTICAST(a) && \ (__IPV6_ADDR_MC_SCOPE(a) == __IPV6_ADDR_SCOPE_ORGLOCAL)) #define IN6_IS_ADDR_MC_GLOBAL(a) \ (IN6_IS_ADDR_MULTICAST(a) && \ (__IPV6_ADDR_MC_SCOPE(a) == __IPV6_ADDR_SCOPE_GLOBAL)) #endif #ifdef _KERNEL /* nonstandard */ /* * KAME Scope */ #define IN6_IS_SCOPE_LINKLOCAL(a) \ ((IN6_IS_ADDR_LINKLOCAL(a)) || \ (IN6_IS_ADDR_MC_LINKLOCAL(a))) #define IN6_IS_SCOPE_EMBED(a) \ ((IN6_IS_ADDR_LINKLOCAL(a)) || \ (IN6_IS_ADDR_MC_LINKLOCAL(a)) || \ (IN6_IS_ADDR_MC_INTFACELOCAL(a))) #define IFA6_IS_DEPRECATED(a) \ ((a)->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME && \ (u_int32_t)((time_second - (a)->ia6_updatetime)) > \ (a)->ia6_lifetime.ia6t_pltime) #define IFA6_IS_INVALID(a) \ ((a)->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME && \ (u_int32_t)((time_second - (a)->ia6_updatetime)) > \ (a)->ia6_lifetime.ia6t_vltime) #endif /* _KERNEL */ /* * IP6 route structure */ #if __BSD_VISIBLE struct route_in6 { struct rtentry *ro_rt; struct llentry *ro_lle; struct sockaddr_in6 ro_dst; }; #endif /* * Options for use with [gs]etsockopt at the IPV6 level. * First word of comment is data type; bool is stored in int. */ /* no hdrincl */ #if 0 /* the followings are relic in IPv4 and hence are disabled */ #define IPV6_OPTIONS 1 /* buf/ip6_opts; set/get IP6 options */ #define IPV6_RECVOPTS 5 /* bool; receive all IP6 opts w/dgram */ #define IPV6_RECVRETOPTS 6 /* bool; receive IP6 opts for response */ #define IPV6_RECVDSTADDR 7 /* bool; receive IP6 dst addr w/dgram */ #define IPV6_RETOPTS 8 /* ip6_opts; set/get IP6 options */ #endif #define IPV6_SOCKOPT_RESERVED1 3 /* reserved for future use */ #define IPV6_UNICAST_HOPS 4 /* int; IP6 hops */ #define IPV6_MULTICAST_IF 9 /* u_int; set/get IP6 multicast i/f */ #define IPV6_MULTICAST_HOPS 10 /* int; set/get IP6 multicast hops */ #define IPV6_MULTICAST_LOOP 11 /* u_int; set/get IP6 multicast loopback */ #define IPV6_JOIN_GROUP 12 /* ipv6_mreq; join a group membership */ #define IPV6_LEAVE_GROUP 13 /* ipv6_mreq; leave a group membership */ #define IPV6_PORTRANGE 14 /* int; range to choose for unspec port */ #define ICMP6_FILTER 18 /* icmp6_filter; icmp6 filter */ /* RFC2292 options */ #ifdef _KERNEL #define IPV6_2292PKTINFO 19 /* bool; send/recv if, src/dst addr */ #define IPV6_2292HOPLIMIT 20 /* bool; hop limit */ #define IPV6_2292NEXTHOP 21 /* bool; next hop addr */ #define IPV6_2292HOPOPTS 22 /* bool; hop-by-hop option */ #define IPV6_2292DSTOPTS 23 /* bool; destinaion option */ #define IPV6_2292RTHDR 24 /* bool; routing header */ #define IPV6_2292PKTOPTIONS 25 /* buf/cmsghdr; set/get IPv6 options */ #endif #define IPV6_CHECKSUM 26 /* int; checksum offset for raw socket */ #define IPV6_V6ONLY 27 /* bool; make AF_INET6 sockets v6 only */ #ifndef _KERNEL #define IPV6_BINDV6ONLY IPV6_V6ONLY #endif #if 1 /* IPSEC */ #define IPV6_IPSEC_POLICY 28 /* struct; get/set security policy */ #endif /* IPSEC */ #define IPV6_FAITH 29 /* bool; accept FAITH'ed connections */ #if 1 /* IPV6FIREWALL */ #define IPV6_FW_ADD 30 /* add a firewall rule to chain */ #define IPV6_FW_DEL 31 /* delete a firewall rule from chain */ #define IPV6_FW_FLUSH 32 /* flush firewall rule chain */ #define IPV6_FW_ZERO 33 /* clear single/all firewall counter(s) */ #define IPV6_FW_GET 34 /* get entire firewall rule chain */ #endif /* new socket options introduced in RFC3542 */ #define IPV6_RTHDRDSTOPTS 35 /* ip6_dest; send dst option before rthdr */ #define IPV6_RECVPKTINFO 36 /* bool; recv if, dst addr */ #define IPV6_RECVHOPLIMIT 37 /* bool; recv hop limit */ #define IPV6_RECVRTHDR 38 /* bool; recv routing header */ #define IPV6_RECVHOPOPTS 39 /* bool; recv hop-by-hop option */ #define IPV6_RECVDSTOPTS 40 /* bool; recv dst option after rthdr */ #ifdef _KERNEL #define IPV6_RECVRTHDRDSTOPTS 41 /* bool; recv dst option before rthdr */ #endif #define IPV6_USE_MIN_MTU 42 /* bool; send packets at the minimum MTU */ #define IPV6_RECVPATHMTU 43 /* bool; notify an according MTU */ #define IPV6_PATHMTU 44 /* mtuinfo; get the current path MTU (sopt), 4 bytes int; MTU notification (cmsg) */ #if 0 /*obsoleted during 2292bis -> 3542*/ #define IPV6_REACHCONF 45 /* no data; ND reachability confirm (cmsg only/not in of RFC3542) */ #endif /* more new socket options introduced in RFC3542 */ #define IPV6_PKTINFO 46 /* in6_pktinfo; send if, src addr */ #define IPV6_HOPLIMIT 47 /* int; send hop limit */ #define IPV6_NEXTHOP 48 /* sockaddr; next hop addr */ #define IPV6_HOPOPTS 49 /* ip6_hbh; send hop-by-hop option */ #define IPV6_DSTOPTS 50 /* ip6_dest; send dst option befor rthdr */ #define IPV6_RTHDR 51 /* ip6_rthdr; send routing header */ #if 0 #define IPV6_PKTOPTIONS 52 /* buf/cmsghdr; set/get IPv6 options */ /* obsoleted by RFC3542 */ #endif #define IPV6_RECVTCLASS 57 /* bool; recv traffic class values */ #define IPV6_AUTOFLOWLABEL 59 /* bool; attach flowlabel automagically */ #define IPV6_TCLASS 61 /* int; send traffic class value */ #define IPV6_DONTFRAG 62 /* bool; disable IPv6 fragmentation */ #define IPV6_PREFER_TEMPADDR 63 /* int; prefer temporary addresses as * the source address. */ #define IPV6_BINDANY 64 /* bool: allow bind to any address */ /* * The following option is private; do not use it from user applications. * It is deliberately defined to the same value as IP_MSFILTER. */ #define IPV6_MSFILTER 74 /* struct __msfilterreq; * set/get multicast source filter list. */ /* to define items, should talk with KAME guys first, for *BSD compatibility */ #define IPV6_RTHDR_LOOSE 0 /* this hop need not be a neighbor. XXX old spec */ #define IPV6_RTHDR_STRICT 1 /* this hop must be a neighbor. XXX old spec */ #define IPV6_RTHDR_TYPE_0 0 /* IPv6 routing header type 0 */ /* * Defaults and limits for options */ #define IPV6_DEFAULT_MULTICAST_HOPS 1 /* normally limit m'casts to 1 hop */ #define IPV6_DEFAULT_MULTICAST_LOOP 1 /* normally hear sends if a member */ /* * The im6o_membership vector for each socket is now dynamically allocated at * run-time, bounded by USHRT_MAX, and is reallocated when needed, sized * according to a power-of-two increment. */ #define IPV6_MIN_MEMBERSHIPS 31 #define IPV6_MAX_MEMBERSHIPS 4095 /* * Default resource limits for IPv6 multicast source filtering. * These may be modified by sysctl. */ #define IPV6_MAX_GROUP_SRC_FILTER 512 /* sources per group */ #define IPV6_MAX_SOCK_SRC_FILTER 128 /* sources per socket/group */ /* * Argument structure for IPV6_JOIN_GROUP and IPV6_LEAVE_GROUP. */ struct ipv6_mreq { struct in6_addr ipv6mr_multiaddr; unsigned int ipv6mr_interface; }; /* * IPV6_PKTINFO: Packet information(RFC2292 sec 5) */ struct in6_pktinfo { struct in6_addr ipi6_addr; /* src/dst IPv6 address */ unsigned int ipi6_ifindex; /* send/recv interface index */ }; /* * Control structure for IPV6_RECVPATHMTU socket option. */ struct ip6_mtuinfo { struct sockaddr_in6 ip6m_addr; /* or sockaddr_storage? */ uint32_t ip6m_mtu; }; /* * Argument for IPV6_PORTRANGE: * - which range to search when port is unspecified at bind() or connect() */ #define IPV6_PORTRANGE_DEFAULT 0 /* default range */ #define IPV6_PORTRANGE_HIGH 1 /* "high" - request firewall bypass */ #define IPV6_PORTRANGE_LOW 2 /* "low" - vouchsafe security */ #if __BSD_VISIBLE /* * Definitions for inet6 sysctl operations. * * Third level is protocol number. * Fourth level is desired variable within that protocol. */ #define IPV6PROTO_MAXID (IPPROTO_PIM + 1) /* don't list to IPV6PROTO_MAX */ /* * Names for IP sysctl objects */ #define IPV6CTL_FORWARDING 1 /* act as router */ #define IPV6CTL_SENDREDIRECTS 2 /* may send redirects when forwarding*/ #define IPV6CTL_DEFHLIM 3 /* default Hop-Limit */ #ifdef notyet #define IPV6CTL_DEFMTU 4 /* default MTU */ #endif #define IPV6CTL_FORWSRCRT 5 /* forward source-routed dgrams */ #define IPV6CTL_STATS 6 /* stats */ #define IPV6CTL_MRTSTATS 7 /* multicast forwarding stats */ #define IPV6CTL_MRTPROTO 8 /* multicast routing protocol */ #define IPV6CTL_MAXFRAGPACKETS 9 /* max packets reassembly queue */ #define IPV6CTL_SOURCECHECK 10 /* verify source route and intf */ #define IPV6CTL_SOURCECHECK_LOGINT 11 /* minimume logging interval */ #define IPV6CTL_ACCEPT_RTADV 12 #define IPV6CTL_KEEPFAITH 13 #define IPV6CTL_LOG_INTERVAL 14 #define IPV6CTL_HDRNESTLIMIT 15 #define IPV6CTL_DAD_COUNT 16 #define IPV6CTL_AUTO_FLOWLABEL 17 #define IPV6CTL_DEFMCASTHLIM 18 #define IPV6CTL_GIF_HLIM 19 /* default HLIM for gif encap packet */ #define IPV6CTL_KAME_VERSION 20 #define IPV6CTL_USE_DEPRECATED 21 /* use deprecated addr (RFC2462 5.5.4) */ #define IPV6CTL_RR_PRUNE 22 /* walk timer for router renumbering */ #if 0 /* obsolete */ #define IPV6CTL_MAPPED_ADDR 23 #endif #define IPV6CTL_V6ONLY 24 #define IPV6CTL_RTEXPIRE 25 /* cloned route expiration time */ #define IPV6CTL_RTMINEXPIRE 26 /* min value for expiration time */ #define IPV6CTL_RTMAXCACHE 27 /* trigger level for dynamic expire */ #define IPV6CTL_USETEMPADDR 32 /* use temporary addresses (RFC3041) */ #define IPV6CTL_TEMPPLTIME 33 /* preferred lifetime for tmpaddrs */ #define IPV6CTL_TEMPVLTIME 34 /* valid lifetime for tmpaddrs */ #define IPV6CTL_AUTO_LINKLOCAL 35 /* automatic link-local addr assign */ #define IPV6CTL_RIP6STATS 36 /* raw_ip6 stats */ #define IPV6CTL_PREFER_TEMPADDR 37 /* prefer temporary addr as src */ #define IPV6CTL_ADDRCTLPOLICY 38 /* get/set address selection policy */ #define IPV6CTL_USE_DEFAULTZONE 39 /* use default scope zone */ #define IPV6CTL_MAXFRAGS 41 /* max fragments */ #if 0 #define IPV6CTL_IFQ 42 /* ip6intrq node */ #define IPV6CTL_ISATAPRTR 43 /* isatap router */ #endif #define IPV6CTL_MCAST_PMTU 44 /* enable pMTU discovery for multicast? */ /* New entries should be added here from current IPV6CTL_MAXID value. */ /* to define items, should talk with KAME guys first, for *BSD compatibility */ #define IPV6CTL_STEALTH 45 #define ICMPV6CTL_ND6_ONLINKNSRFC4861 47 -#define IPV6CTL_MAXID 48 +#define IPV6CTL_NO_RADR 48 /* No defroute from RA */ +#define IPV6CTL_NORBIT_RAIF 49 /* Disable R-bit in NA on RA + * receiving IF. */ +#define IPV6CTL_MAXID 50 #endif /* __BSD_VISIBLE */ /* * Redefinition of mbuf flags */ #define M_AUTHIPHDR M_PROTO2 #define M_DECRYPTED M_PROTO3 #define M_LOOP M_PROTO4 #define M_AUTHIPDGM M_PROTO5 #define M_RTALERT_MLD M_PROTO6 #ifdef _KERNEL struct cmsghdr; int in6_cksum __P((struct mbuf *, u_int8_t, u_int32_t, u_int32_t)); int in6_localaddr __P((struct in6_addr *)); int in6_addrscope __P((struct in6_addr *)); struct in6_ifaddr *in6_ifawithifp __P((struct ifnet *, struct in6_addr *)); extern void in6_if_up __P((struct ifnet *)); struct sockaddr; extern u_char ip6_protox[]; void in6_sin6_2_sin __P((struct sockaddr_in *sin, struct sockaddr_in6 *sin6)); void in6_sin_2_v4mapsin6 __P((struct sockaddr_in *sin, struct sockaddr_in6 *sin6)); void in6_sin6_2_sin_in_sock __P((struct sockaddr *nam)); void in6_sin_2_v4mapsin6_in_sock __P((struct sockaddr **nam)); extern void addrsel_policy_init __P((void)); #define satosin6(sa) ((struct sockaddr_in6 *)(sa)) #define sin6tosa(sin6) ((struct sockaddr *)(sin6)) #define ifatoia6(ifa) ((struct in6_ifaddr *)(ifa)) extern int (*faithprefix_p)(struct in6_addr *); #endif /* _KERNEL */ #ifndef _SIZE_T_DECLARED typedef __size_t size_t; #define _SIZE_T_DECLARED #endif #ifndef _SOCKLEN_T_DECLARED typedef __socklen_t socklen_t; #define _SOCKLEN_T_DECLARED #endif #if __BSD_VISIBLE __BEGIN_DECLS struct cmsghdr; extern int inet6_option_space __P((int)); extern int inet6_option_init __P((void *, struct cmsghdr **, int)); extern int inet6_option_append __P((struct cmsghdr *, const uint8_t *, int, int)); extern uint8_t *inet6_option_alloc __P((struct cmsghdr *, int, int, int)); extern int inet6_option_next __P((const struct cmsghdr *, uint8_t **)); extern int inet6_option_find __P((const struct cmsghdr *, uint8_t **, int)); extern size_t inet6_rthdr_space __P((int, int)); extern struct cmsghdr *inet6_rthdr_init __P((void *, int)); extern int inet6_rthdr_add __P((struct cmsghdr *, const struct in6_addr *, unsigned int)); extern int inet6_rthdr_lasthop __P((struct cmsghdr *, unsigned int)); #if 0 /* not implemented yet */ extern int inet6_rthdr_reverse __P((const struct cmsghdr *, struct cmsghdr *)); #endif extern int inet6_rthdr_segments __P((const struct cmsghdr *)); extern struct in6_addr *inet6_rthdr_getaddr __P((struct cmsghdr *, int)); extern int inet6_rthdr_getflags __P((const struct cmsghdr *, int)); extern int inet6_opt_init __P((void *, socklen_t)); extern int inet6_opt_append __P((void *, socklen_t, int, uint8_t, socklen_t, uint8_t, void **)); extern int inet6_opt_finish __P((void *, socklen_t, int)); extern int inet6_opt_set_val __P((void *, int, void *, socklen_t)); extern int inet6_opt_next __P((void *, socklen_t, int, uint8_t *, socklen_t *, void **)); extern int inet6_opt_find __P((void *, socklen_t, int, uint8_t, socklen_t *, void **)); extern int inet6_opt_get_val __P((void *, int, void *, socklen_t)); extern socklen_t inet6_rth_space __P((int, int)); extern void *inet6_rth_init __P((void *, socklen_t, int, int)); extern int inet6_rth_add __P((void *, const struct in6_addr *)); extern int inet6_rth_reverse __P((const void *, void *)); extern int inet6_rth_segments __P((const void *)); extern struct in6_addr *inet6_rth_getaddr __P((const void *, int)); __END_DECLS #endif /* __BSD_VISIBLE */ #endif /* !_NETINET6_IN6_H_ */ Index: projects/largeSMP/sys/netinet6/in6_pcb.c =================================================================== --- projects/largeSMP/sys/netinet6/in6_pcb.c (revision 222782) +++ projects/largeSMP/sys/netinet6/in6_pcb.c (revision 222783) @@ -1,1028 +1,1195 @@ /*- * 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 #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 struct in6_addr zeroin6_addr; 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) 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; /* 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 && ((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_socket->so_options & SO_REUSEPORT) == 0) && (inp->inp_cred->cr_uid != t->inp_cred->cr_uid)) 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 && ((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); } #endif } t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr, lport, lookupflags, cred); if (t && (reuseport & ((t->inp_flags & INP_TIMEWAIT) ? intotw(t)->tw_so_options : t->inp_socket->so_options)) == 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) { if ((reuseport & intotw(t)->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 & t->inp_socket->so_options) == 0 && (ntohl(t->inp_laddr.s_addr) != INADDR_ANY || INP_SOCKAF(so) == INP_SOCKAF(t->inp_socket))) 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). */ 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. * 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))) { ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst, (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(INADDR_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; + int faith; + + if (faithprefix_p != NULL) + faith = (*faithprefix_p)(laddr); + else + faith = 0; + + /* + * First look for an exact match. + */ + tmpinp = NULL; + INP_GROUP_LOCK(pcbgroup); + head = &pcbgroup->ipg_hashbase[ + INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, 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, 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(INADDR_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; + } + + /* XXX inp locking */ + if (faith && (inp->inp_flags & INP_FAITH) == 0) + 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. */ 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; int faith; KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0, ("%s: invalid lookup flags %d", __func__, lookupflags)); INP_HASH_LOCK_ASSERT(pcbinfo); if (faithprefix_p != NULL) faith = (*faithprefix_p)(laddr); else faith = 0; /* * First look for an exact match. */ tmpinp = NULL; head = &pcbinfo->ipi_hashbase[ INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, 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(INADDR_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; } /* XXX inp locking */ if (faith && (inp->inp_flags & INP_FAITH) == 0) 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) + 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__)); +#if defined(PCBGROUP) + 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 (in_pcbgroup_enabled(pcbinfo)) { + 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)); + 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)); } 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/largeSMP/sys/netinet6/in6_pcb.h =================================================================== --- projects/largeSMP/sys/netinet6/in6_pcb.h (revision 222782) +++ projects/largeSMP/sys/netinet6/in6_pcb.h (revision 222783) @@ -1,114 +1,124 @@ /*- * 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_pcb.h,v 1.13 2001/02/06 09:16:53 itojun Exp $ */ /*- * Copyright (c) 1982, 1986, 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. * * @(#)in_pcb.h 8.1 (Berkeley) 6/10/93 * $FreeBSD$ */ #ifndef _NETINET6_IN6_PCB_H_ #define _NETINET6_IN6_PCB_H_ #ifdef _KERNEL #define satosin6(sa) ((struct sockaddr_in6 *)(sa)) #define sin6tosa(sin6) ((struct sockaddr *)(sin6)) #define ifatoia6(ifa) ((struct in6_ifaddr *)(ifa)) +struct inpcbgroup * + in6_pcbgroup_byhash(struct inpcbinfo *, u_int, uint32_t); +struct inpcbgroup * + in6_pcbgroup_byinpcb __P((struct inpcb *)); +struct inpcbgroup * + in6_pcbgroup_bymbuf(struct inpcbinfo *, struct mbuf *); +struct inpcbgroup * + in6_pcbgroup_bytuple __P((struct inpcbinfo *, const struct in6_addr *, + u_short, const struct in6_addr *, u_short)); + void in6_pcbpurgeif0 __P((struct inpcbinfo *, struct ifnet *)); void in6_losing __P((struct inpcb *)); int in6_pcbbind __P((struct inpcb *, struct sockaddr *, struct ucred *)); int in6_pcbconnect __P((struct inpcb *, struct sockaddr *, struct ucred *)); int in6_pcbconnect_mbuf __P((struct inpcb *, struct sockaddr *, struct ucred *, struct mbuf *)); void in6_pcbdisconnect __P((struct inpcb *)); int in6_pcbladdr(struct inpcb *, struct sockaddr *, struct in6_addr *); struct inpcb * in6_pcblookup_local __P((struct inpcbinfo *, struct in6_addr *, u_short, int, struct ucred *)); struct inpcb * in6_pcblookup __P((struct inpcbinfo *, struct in6_addr *, u_int, struct in6_addr *, u_int, int, struct ifnet *)); struct inpcb * in6_pcblookup_hash_locked __P((struct inpcbinfo *, struct in6_addr *, u_int, struct in6_addr *, u_int, int, struct ifnet *)); struct inpcb * in6_pcblookup_mbuf __P((struct inpcbinfo *, struct in6_addr *, u_int, struct in6_addr *, u_int, int, struct ifnet *ifp, struct mbuf *)); void in6_pcbnotify __P((struct inpcbinfo *, struct sockaddr *, u_int, const struct sockaddr *, u_int, int, void *, struct inpcb *(*)(struct inpcb *, int))); struct inpcb * in6_rtchange __P((struct inpcb *, int)); struct sockaddr * in6_sockaddr __P((in_port_t port, struct in6_addr *addr_p)); struct sockaddr * in6_v4mapsin6_sockaddr __P((in_port_t port, struct in_addr *addr_p)); int in6_getpeeraddr __P((struct socket *so, struct sockaddr **nam)); int in6_getsockaddr __P((struct socket *so, struct sockaddr **nam)); int in6_mapped_sockaddr __P((struct socket *so, struct sockaddr **nam)); int in6_mapped_peeraddr __P((struct socket *so, struct sockaddr **nam)); int in6_selecthlim __P((struct in6pcb *, struct ifnet *)); int in6_pcbsetport __P((struct in6_addr *, struct inpcb *, struct ucred *)); void init_sin6 __P((struct sockaddr_in6 *sin6, struct mbuf *m)); #endif /* _KERNEL */ #endif /* !_NETINET6_IN6_PCB_H_ */ Index: projects/largeSMP/sys/netinet6/in6_pcbgroup.c =================================================================== --- projects/largeSMP/sys/netinet6/in6_pcbgroup.c (nonexistent) +++ projects/largeSMP/sys/netinet6/in6_pcbgroup.c (revision 222783) @@ -0,0 +1,103 @@ +/*- + * Copyright (c) 2010-2011 Juniper Networks, Inc. + * All rights reserved. + * + * This software was 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. + * + * 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_inet6.h" + +#include +#include + +#include +#include +#ifdef INET6 +#include +#endif /* INET6 */ + +/* + * Given a hash of whatever the covered tuple might be, return a pcbgroup + * index. + */ +static __inline u_int +in6_pcbgroup_getbucket(struct inpcbinfo *pcbinfo, uint32_t hash) +{ + + return (hash % pcbinfo->ipi_npcbgroups); +} + +/* + * Map a (hashtype, hash) tuple into a connection group, or NULL if the hash + * information is insufficient to identify the pcbgroup. + */ +struct inpcbgroup * +in6_pcbgroup_byhash(struct inpcbinfo *pcbinfo, u_int hashtype, uint32_t hash) +{ + + return (NULL); +} + +struct inpcbgroup * +in6_pcbgroup_bymbuf(struct inpcbinfo *pcbinfo, struct mbuf *m) +{ + + return (in6_pcbgroup_byhash(pcbinfo, M_HASHTYPE_GET(m), + m->m_pkthdr.flowid)); +} + +struct inpcbgroup * +in6_pcbgroup_bytuple(struct inpcbinfo *pcbinfo, const struct in6_addr *laddrp, + u_short lport, const struct in6_addr *faddrp, u_short fport) +{ + uint32_t hash; + + switch (pcbinfo->ipi_hashfields) { + case IPI_HASHFIELDS_4TUPLE: + hash = faddrp->s6_addr32[3] ^ fport; + break; + + case IPI_HASHFIELDS_2TUPLE: + hash = faddrp->s6_addr32[3] ^ laddrp->s6_addr32[3]; + break; + + default: + hash = 0; + } + return (&pcbinfo->ipi_pcbgroups[in6_pcbgroup_getbucket(pcbinfo, + hash)]); +} + +struct inpcbgroup * +in6_pcbgroup_byinpcb(struct inpcb *inp) +{ + + return (in6_pcbgroup_bytuple(inp->inp_pcbinfo, &inp->in6p_laddr, + inp->inp_lport, &inp->in6p_faddr, inp->inp_fport)); +} Property changes on: projects/largeSMP/sys/netinet6/in6_pcbgroup.c ___________________________________________________________________ Added: svn:eol-style ## -0,0 +1 ## +native \ No newline at end of property Added: svn:keywords ## -0,0 +1 ## +FreeBSD=%H \ No newline at end of property Added: svn:mime-type ## -0,0 +1 ## +text/plain \ No newline at end of property Index: projects/largeSMP/sys/netinet6/in6_proto.c =================================================================== --- projects/largeSMP/sys/netinet6/in6_proto.c (revision 222782) +++ projects/largeSMP/sys/netinet6/in6_proto.c (revision 222783) @@ -1,624 +1,635 @@ /*- * 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_proto.c,v 1.91 2001/05/27 13:28:35 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. * * @(#)in_proto.c 8.1 (Berkeley) 6/10/93 */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_ipstealth.h" #include "opt_sctp.h" #include "opt_mpath.h" #include "opt_route.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef RADIX_MPATH #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef SCTP #include #include #include #include #include #endif /* SCTP */ #ifdef IPSEC #include #include #endif /* IPSEC */ #include #ifdef FLOWTABLE #include #endif /* * TCP/IP protocol family: IP6, ICMP6, UDP, TCP. */ FEATURE(inet6, "Internet Protocol version 6"); extern struct domain inet6domain; static struct pr_usrreqs nousrreqs; #define PR_LISTEN 0 #define PR_ABRTACPTDIS 0 /* Spacer for loadable protocols. */ #define IP6PROTOSPACER \ { \ .pr_domain = &inet6domain, \ .pr_protocol = PROTO_SPACER, \ .pr_usrreqs = &nousrreqs \ } struct ip6protosw inet6sw[] = { { .pr_type = 0, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_IPV6, .pr_init = ip6_init, #ifdef VIMAGE .pr_destroy = ip6_destroy, #endif .pr_slowtimo = frag6_slowtimo, .pr_drain = frag6_drain, .pr_usrreqs = &nousrreqs, }, { .pr_type = SOCK_DGRAM, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_UDP, .pr_flags = PR_ATOMIC|PR_ADDR, .pr_input = udp6_input, .pr_ctlinput = udp6_ctlinput, .pr_ctloutput = ip6_ctloutput, #ifndef INET /* Do not call initialization twice. */ .pr_init = udp_init, #endif .pr_usrreqs = &udp6_usrreqs, }, { .pr_type = SOCK_STREAM, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_TCP, .pr_flags = PR_CONNREQUIRED|PR_WANTRCVD|PR_LISTEN, .pr_input = tcp6_input, .pr_ctlinput = tcp6_ctlinput, .pr_ctloutput = tcp_ctloutput, #ifndef INET /* don't call initialization and timeout routines twice */ .pr_init = tcp_init, .pr_slowtimo = tcp_slowtimo, #endif .pr_drain = tcp_drain, .pr_usrreqs = &tcp6_usrreqs, }, #ifdef SCTP { .pr_type = SOCK_DGRAM, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_SCTP, .pr_flags = PR_WANTRCVD, .pr_input = sctp6_input, .pr_ctlinput = sctp6_ctlinput, .pr_ctloutput = sctp_ctloutput, .pr_drain = sctp_drain, #ifndef INET /* Do not call initialization twice. */ .pr_init = sctp_init, #endif .pr_usrreqs = &sctp6_usrreqs }, { .pr_type = SOCK_SEQPACKET, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_SCTP, .pr_flags = PR_WANTRCVD, .pr_input = sctp6_input, .pr_ctlinput = sctp6_ctlinput, .pr_ctloutput = sctp_ctloutput, .pr_drain = sctp_drain, .pr_usrreqs = &sctp6_usrreqs }, { .pr_type = SOCK_STREAM, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_SCTP, .pr_flags = PR_WANTRCVD, .pr_input = sctp6_input, .pr_ctlinput = sctp6_ctlinput, .pr_ctloutput = sctp_ctloutput, .pr_drain = sctp_drain, .pr_usrreqs = &sctp6_usrreqs }, #endif /* SCTP */ { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_RAW, .pr_flags = PR_ATOMIC|PR_ADDR, .pr_input = rip6_input, .pr_output = rip6_output, .pr_ctlinput = rip6_ctlinput, .pr_ctloutput = rip6_ctloutput, #ifndef INET /* Do not call initialization twice. */ .pr_init = rip_init, #endif .pr_usrreqs = &rip6_usrreqs }, { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_ICMPV6, .pr_flags = PR_ATOMIC|PR_ADDR|PR_LASTHDR, .pr_input = icmp6_input, .pr_output = rip6_output, .pr_ctlinput = rip6_ctlinput, .pr_ctloutput = rip6_ctloutput, .pr_fasttimo = icmp6_fasttimo, .pr_slowtimo = icmp6_slowtimo, .pr_usrreqs = &rip6_usrreqs }, { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_DSTOPTS, .pr_flags = PR_ATOMIC|PR_ADDR, .pr_input = dest6_input, .pr_usrreqs = &nousrreqs }, { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_ROUTING, .pr_flags = PR_ATOMIC|PR_ADDR, .pr_input = route6_input, .pr_usrreqs = &nousrreqs }, { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_FRAGMENT, .pr_flags = PR_ATOMIC|PR_ADDR, .pr_input = frag6_input, .pr_usrreqs = &nousrreqs }, #ifdef IPSEC { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_AH, .pr_flags = PR_ATOMIC|PR_ADDR, .pr_input = ipsec6_common_input, .pr_usrreqs = &nousrreqs, }, { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_ESP, .pr_flags = PR_ATOMIC|PR_ADDR, .pr_input = ipsec6_common_input, .pr_ctlinput = esp6_ctlinput, .pr_usrreqs = &nousrreqs, }, { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_IPCOMP, .pr_flags = PR_ATOMIC|PR_ADDR, .pr_input = ipsec6_common_input, .pr_usrreqs = &nousrreqs, }, #endif /* IPSEC */ #ifdef INET { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_IPV4, .pr_flags = PR_ATOMIC|PR_ADDR|PR_LASTHDR, .pr_input = encap6_input, .pr_output = rip6_output, .pr_ctloutput = rip6_ctloutput, .pr_init = encap_init, .pr_usrreqs = &rip6_usrreqs }, #endif /* INET */ { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_IPV6, .pr_flags = PR_ATOMIC|PR_ADDR|PR_LASTHDR, .pr_input = encap6_input, .pr_output = rip6_output, .pr_ctloutput = rip6_ctloutput, .pr_init = encap_init, .pr_usrreqs = &rip6_usrreqs }, { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, .pr_protocol = IPPROTO_PIM, .pr_flags = PR_ATOMIC|PR_ADDR|PR_LASTHDR, .pr_input = encap6_input, .pr_output = rip6_output, .pr_ctloutput = rip6_ctloutput, .pr_usrreqs = &rip6_usrreqs }, /* Spacer n-times for loadable protocols. */ IP6PROTOSPACER, IP6PROTOSPACER, IP6PROTOSPACER, IP6PROTOSPACER, IP6PROTOSPACER, IP6PROTOSPACER, IP6PROTOSPACER, IP6PROTOSPACER, /* raw wildcard */ { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, .pr_flags = PR_ATOMIC|PR_ADDR, .pr_input = rip6_input, .pr_output = rip6_output, .pr_ctloutput = rip6_ctloutput, .pr_usrreqs = &rip6_usrreqs }, }; extern int in6_inithead(void **, int); #ifdef VIMAGE extern int in6_detachhead(void **, int); #endif struct domain inet6domain = { .dom_family = AF_INET6, .dom_name = "internet6", .dom_protosw = (struct protosw *)inet6sw, .dom_protoswNPROTOSW = (struct protosw *) &inet6sw[sizeof(inet6sw)/sizeof(inet6sw[0])], #ifdef RADIX_MPATH .dom_rtattach = rn6_mpath_inithead, #else .dom_rtattach = in6_inithead, #endif #ifdef VIMAGE .dom_rtdetach = in6_detachhead, #endif .dom_rtoffset = offsetof(struct sockaddr_in6, sin6_addr) << 3, .dom_maxrtkey = sizeof(struct sockaddr_in6), .dom_ifattach = in6_domifattach, .dom_ifdetach = in6_domifdetach }; VNET_DOMAIN_SET(inet6); /* * Internet configuration info */ #ifndef IPV6FORWARDING #ifdef GATEWAY6 #define IPV6FORWARDING 1 /* forward IP6 packets not for us */ #else #define IPV6FORWARDING 0 /* don't forward IP6 packets not for us */ #endif /* GATEWAY6 */ #endif /* !IPV6FORWARDING */ #ifndef IPV6_SENDREDIRECTS #define IPV6_SENDREDIRECTS 1 #endif VNET_DEFINE(int, ip6_forwarding) = IPV6FORWARDING; /* act as router? */ VNET_DEFINE(int, ip6_sendredirects) = IPV6_SENDREDIRECTS; VNET_DEFINE(int, ip6_defhlim) = IPV6_DEFHLIM; VNET_DEFINE(int, ip6_defmcasthlim) = IPV6_DEFAULT_MULTICAST_HOPS; VNET_DEFINE(int, ip6_accept_rtadv) = 0; +VNET_DEFINE(int, ip6_no_radr) = 0; +VNET_DEFINE(int, ip6_norbit_raif) = 0; VNET_DEFINE(int, ip6_maxfragpackets); /* initialized in frag6.c:frag6_init() */ VNET_DEFINE(int, ip6_maxfrags); /* initialized in frag6.c:frag6_init() */ VNET_DEFINE(int, ip6_log_interval) = 5; VNET_DEFINE(int, ip6_hdrnestlimit) = 15;/* How many header options will we * process? */ VNET_DEFINE(int, ip6_dad_count) = 1; /* DupAddrDetectionTransmits */ VNET_DEFINE(int, ip6_auto_flowlabel) = 1; VNET_DEFINE(int, ip6_use_deprecated) = 1;/* allow deprecated addr * (RFC2462 5.5.4) */ VNET_DEFINE(int, ip6_rr_prune) = 5; /* router renumbering prefix * walk list every 5 sec. */ VNET_DEFINE(int, ip6_mcast_pmtu) = 0; /* enable pMTU discovery for multicast? */ VNET_DEFINE(int, ip6_v6only) = 1; VNET_DEFINE(int, ip6_keepfaith) = 0; VNET_DEFINE(time_t, ip6_log_time) = (time_t)0L; #ifdef IPSTEALTH VNET_DEFINE(int, ip6stealth) = 0; #endif VNET_DEFINE(int, nd6_onlink_ns_rfc4861) = 0;/* allow 'on-link' nd6 NS * (RFC 4861) */ /* icmp6 */ /* * BSDI4 defines these variables in in_proto.c... * XXX: what if we don't define INET? Should we define pmtu6_expire * or so? (jinmei@kame.net 19990310) */ VNET_DEFINE(int, pmtu_expire) = 60*10; VNET_DEFINE(int, pmtu_probe) = 60*2; /* raw IP6 parameters */ /* * Nominal space allocated to a raw ip socket. */ #define RIPV6SNDQ 8192 #define RIPV6RCVQ 8192 VNET_DEFINE(u_long, rip6_sendspace) = RIPV6SNDQ; VNET_DEFINE(u_long, rip6_recvspace) = RIPV6RCVQ; /* ICMPV6 parameters */ VNET_DEFINE(int, icmp6_rediraccept) = 1;/* accept and process redirects */ VNET_DEFINE(int, icmp6_redirtimeout) = 10 * 60; /* 10 minutes */ VNET_DEFINE(int, icmp6errppslim) = 100; /* 100pps */ /* control how to respond to NI queries */ VNET_DEFINE(int, icmp6_nodeinfo) = (ICMP6_NODEINFO_FQDNOK|ICMP6_NODEINFO_NODEADDROK); /* UDP on IP6 parameters */ VNET_DEFINE(int, udp6_sendspace) = 9216;/* really max datagram size */ VNET_DEFINE(int, udp6_recvspace) = 40 * (1024 + sizeof(struct sockaddr_in6)); /* 40 1K datagrams */ /* * sysctl related items. */ SYSCTL_NODE(_net, PF_INET6, inet6, CTLFLAG_RW, 0, "Internet6 Family"); /* net.inet6 */ SYSCTL_NODE(_net_inet6, IPPROTO_IPV6, ip6, CTLFLAG_RW, 0, "IP6"); SYSCTL_NODE(_net_inet6, IPPROTO_ICMPV6, icmp6, CTLFLAG_RW, 0, "ICMP6"); SYSCTL_NODE(_net_inet6, IPPROTO_UDP, udp6, CTLFLAG_RW, 0, "UDP6"); SYSCTL_NODE(_net_inet6, IPPROTO_TCP, tcp6, CTLFLAG_RW, 0, "TCP6"); #ifdef SCTP SYSCTL_NODE(_net_inet6, IPPROTO_SCTP, sctp6, CTLFLAG_RW, 0, "SCTP6"); #endif #ifdef IPSEC SYSCTL_NODE(_net_inet6, IPPROTO_ESP, ipsec6, CTLFLAG_RW, 0, "IPSEC6"); #endif /* IPSEC */ /* net.inet6.ip6 */ static int sysctl_ip6_temppltime(SYSCTL_HANDLER_ARGS) { int error = 0; int old; VNET_SYSCTL_ARG(req, arg1); error = SYSCTL_OUT(req, arg1, sizeof(int)); if (error || !req->newptr) return (error); old = V_ip6_temp_preferred_lifetime; error = SYSCTL_IN(req, arg1, sizeof(int)); if (V_ip6_temp_preferred_lifetime < V_ip6_desync_factor + V_ip6_temp_regen_advance) { V_ip6_temp_preferred_lifetime = old; return (EINVAL); } return (error); } static int sysctl_ip6_tempvltime(SYSCTL_HANDLER_ARGS) { int error = 0; int old; VNET_SYSCTL_ARG(req, arg1); error = SYSCTL_OUT(req, arg1, sizeof(int)); if (error || !req->newptr) return (error); old = V_ip6_temp_valid_lifetime; error = SYSCTL_IN(req, arg1, sizeof(int)); if (V_ip6_temp_valid_lifetime < V_ip6_temp_preferred_lifetime) { V_ip6_temp_preferred_lifetime = old; return (EINVAL); } return (error); } SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_FORWARDING, forwarding, CTLFLAG_RW, &VNET_NAME(ip6_forwarding), 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_SENDREDIRECTS, redirect, CTLFLAG_RW, &VNET_NAME(ip6_sendredirects), 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_DEFHLIM, hlim, CTLFLAG_RW, &VNET_NAME(ip6_defhlim), 0, ""); SYSCTL_VNET_STRUCT(_net_inet6_ip6, IPV6CTL_STATS, stats, CTLFLAG_RD, &VNET_NAME(ip6stat), ip6stat, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_MAXFRAGPACKETS, maxfragpackets, CTLFLAG_RW, &VNET_NAME(ip6_maxfragpackets), 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_ACCEPT_RTADV, accept_rtadv, CTLFLAG_RW, &VNET_NAME(ip6_accept_rtadv), 0, "Default value of per-interface flag for accepting ICMPv6 Router" "Advertisement messages"); +SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_NO_RADR, no_radr, + CTLFLAG_RW, &VNET_NAME(ip6_no_radr), 0, + "Default value of per-interface flag to control whether routers " + "sending ICMPv6 RA messages on that interface are added into the " + "default router list."); +SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_NORBIT_RAIF, norbit_raif, CTLFLAG_RW, + &VNET_NAME(ip6_norbit_raif), 0, + "Always set 0 to R flag in ICMPv6 NA messages when accepting RA" + " on the interface."); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_KEEPFAITH, keepfaith, CTLFLAG_RW, &VNET_NAME(ip6_keepfaith), 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_LOG_INTERVAL, log_interval, CTLFLAG_RW, &VNET_NAME(ip6_log_interval), 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_HDRNESTLIMIT, hdrnestlimit, CTLFLAG_RW, &VNET_NAME(ip6_hdrnestlimit), 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_DAD_COUNT, dad_count, CTLFLAG_RW, &VNET_NAME(ip6_dad_count), 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_AUTO_FLOWLABEL, auto_flowlabel, CTLFLAG_RW, &VNET_NAME(ip6_auto_flowlabel), 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_DEFMCASTHLIM, defmcasthlim, CTLFLAG_RW, &VNET_NAME(ip6_defmcasthlim), 0, ""); SYSCTL_STRING(_net_inet6_ip6, IPV6CTL_KAME_VERSION, kame_version, CTLFLAG_RD, __KAME_VERSION, 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_USE_DEPRECATED, use_deprecated, CTLFLAG_RW, &VNET_NAME(ip6_use_deprecated), 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_RR_PRUNE, rr_prune, CTLFLAG_RW, &VNET_NAME(ip6_rr_prune), 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_USETEMPADDR, use_tempaddr, CTLFLAG_RW, &VNET_NAME(ip6_use_tempaddr), 0, ""); SYSCTL_VNET_PROC(_net_inet6_ip6, IPV6CTL_TEMPPLTIME, temppltime, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(ip6_temp_preferred_lifetime), 0, sysctl_ip6_temppltime, "I", ""); SYSCTL_VNET_PROC(_net_inet6_ip6, IPV6CTL_TEMPVLTIME, tempvltime, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(ip6_temp_valid_lifetime), 0, sysctl_ip6_tempvltime, "I", ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_V6ONLY, v6only, CTLFLAG_RW, &VNET_NAME(ip6_v6only), 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_AUTO_LINKLOCAL, auto_linklocal, CTLFLAG_RW, &VNET_NAME(ip6_auto_linklocal), 0, "Default value of per-interface flag for automatically adding an IPv6" " link-local address to interfaces when attached"); SYSCTL_VNET_STRUCT(_net_inet6_ip6, IPV6CTL_RIP6STATS, rip6stats, CTLFLAG_RD, &VNET_NAME(rip6stat), rip6stat, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_PREFER_TEMPADDR, prefer_tempaddr, CTLFLAG_RW, &VNET_NAME(ip6_prefer_tempaddr), 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_USE_DEFAULTZONE, use_defaultzone, CTLFLAG_RW, &VNET_NAME(ip6_use_defzone), 0,""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_MAXFRAGS, maxfrags, CTLFLAG_RW, &VNET_NAME(ip6_maxfrags), 0, ""); SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_MCAST_PMTU, mcast_pmtu, CTLFLAG_RW, &VNET_NAME(ip6_mcast_pmtu), 0, ""); #ifdef IPSTEALTH SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_STEALTH, stealth, CTLFLAG_RW, &VNET_NAME(ip6stealth), 0, ""); #endif #ifdef FLOWTABLE VNET_DEFINE(int, ip6_output_flowtable_size) = 2048; VNET_DEFINE(struct flowtable *, ip6_ft); #define V_ip6_output_flowtable_size VNET(ip6_output_flowtable_size) SYSCTL_VNET_INT(_net_inet6_ip6, OID_AUTO, output_flowtable_size, CTLFLAG_RDTUN, &VNET_NAME(ip6_output_flowtable_size), 2048, "number of entries in the per-cpu output flow caches"); #endif /* net.inet6.icmp6 */ SYSCTL_VNET_INT(_net_inet6_icmp6, ICMPV6CTL_REDIRACCEPT, rediraccept, CTLFLAG_RW, &VNET_NAME(icmp6_rediraccept), 0, ""); SYSCTL_VNET_INT(_net_inet6_icmp6, ICMPV6CTL_REDIRTIMEOUT, redirtimeout, CTLFLAG_RW, &VNET_NAME(icmp6_redirtimeout), 0, ""); SYSCTL_VNET_STRUCT(_net_inet6_icmp6, ICMPV6CTL_STATS, stats, CTLFLAG_RD, &VNET_NAME(icmp6stat), icmp6stat, ""); SYSCTL_VNET_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_PRUNE, nd6_prune, CTLFLAG_RW, &VNET_NAME(nd6_prune), 0, ""); SYSCTL_VNET_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_DELAY, nd6_delay, CTLFLAG_RW, &VNET_NAME(nd6_delay), 0, ""); SYSCTL_VNET_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_UMAXTRIES, nd6_umaxtries, CTLFLAG_RW, &VNET_NAME(nd6_umaxtries), 0, ""); SYSCTL_VNET_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_MMAXTRIES, nd6_mmaxtries, CTLFLAG_RW, &VNET_NAME(nd6_mmaxtries), 0, ""); SYSCTL_VNET_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_USELOOPBACK, nd6_useloopback, CTLFLAG_RW, &VNET_NAME(nd6_useloopback), 0, ""); SYSCTL_VNET_INT(_net_inet6_icmp6, ICMPV6CTL_NODEINFO, nodeinfo, CTLFLAG_RW, &VNET_NAME(icmp6_nodeinfo), 0, ""); SYSCTL_VNET_INT(_net_inet6_icmp6, ICMPV6CTL_ERRPPSLIMIT, errppslimit, CTLFLAG_RW, &VNET_NAME(icmp6errppslim), 0, ""); SYSCTL_VNET_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_MAXNUDHINT, nd6_maxnudhint, CTLFLAG_RW, &VNET_NAME(nd6_maxnudhint), 0, ""); SYSCTL_VNET_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_DEBUG, nd6_debug, CTLFLAG_RW, &VNET_NAME(nd6_debug), 0, ""); SYSCTL_VNET_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_ONLINKNSRFC4861, nd6_onlink_ns_rfc4861, CTLFLAG_RW, &VNET_NAME(nd6_onlink_ns_rfc4861), 0, "Accept 'on-link' nd6 NS in compliance with RFC 4861."); Index: projects/largeSMP/sys/netinet6/ip6_var.h =================================================================== --- projects/largeSMP/sys/netinet6/ip6_var.h (revision 222782) +++ projects/largeSMP/sys/netinet6/ip6_var.h (revision 222783) @@ -1,444 +1,449 @@ /*- * 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 { u_quad_t ip6s_total; /* total packets received */ u_quad_t ip6s_tooshort; /* packet too short */ u_quad_t ip6s_toosmall; /* not enough data */ u_quad_t ip6s_fragments; /* fragments received */ u_quad_t ip6s_fragdropped; /* frags dropped(dups, out of space) */ u_quad_t ip6s_fragtimeout; /* fragments timed out */ u_quad_t ip6s_fragoverflow; /* fragments that exceeded limit */ u_quad_t ip6s_forward; /* packets forwarded */ u_quad_t ip6s_cantforward; /* packets rcvd for unreachable dest */ u_quad_t ip6s_redirectsent; /* packets forwarded on same net */ u_quad_t ip6s_delivered; /* datagrams delivered to upper level*/ u_quad_t ip6s_localout; /* total ip packets generated here */ u_quad_t ip6s_odropped; /* lost packets due to nobufs, etc. */ u_quad_t ip6s_reassembled; /* total packets reassembled ok */ u_quad_t ip6s_fragmented; /* datagrams successfully fragmented */ u_quad_t ip6s_ofragments; /* output fragments created */ u_quad_t ip6s_cantfrag; /* don't fragment flag was set, etc. */ u_quad_t ip6s_badoptions; /* error in option processing */ u_quad_t ip6s_noroute; /* packets discarded due to no route */ u_quad_t ip6s_badvers; /* ip6 version != 6 */ u_quad_t ip6s_rawout; /* total raw ip packets generated */ u_quad_t ip6s_badscope; /* scope error */ u_quad_t ip6s_notmember; /* don't join this multicast group */ u_quad_t ip6s_nxthist[256]; /* next header history */ u_quad_t ip6s_m1; /* one mbuf */ u_quad_t ip6s_m2m[32]; /* two or more mbuf */ u_quad_t ip6s_mext1; /* one ext mbuf */ u_quad_t ip6s_mext2m; /* two or more ext mbuf */ u_quad_t ip6s_exthdrtoolong; /* ext hdr are not contiguous */ u_quad_t ip6s_nogif; /* no match gif found */ u_quad_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 */ /* number of times that address selection fails */ u_quad_t ip6s_sources_none; /* number of times that an address on the outgoing I/F is chosen */ u_quad_t ip6s_sources_sameif[16]; /* number of times that an address on a non-outgoing I/F is chosen */ u_quad_t ip6s_sources_otherif[16]; /* * number of times that an address that has the same scope * from the destination is chosen. */ u_quad_t ip6s_sources_samescope[16]; /* * number of times that an address that has a different scope * from the destination is chosen. */ u_quad_t ip6s_sources_otherscope[16]; /* number of times that a deprecated address is chosen */ u_quad_t ip6s_sources_deprecated[16]; /* number of times that each rule of source selection is applied. */ u_quad_t ip6s_sources_rule[16]; }; #ifdef _KERNEL #define IP6STAT_ADD(name, val) V_ip6stat.name += (val) #define IP6STAT_SUB(name, val) V_ip6stat.name -= (val) #define IP6STAT_INC(name) IP6STAT_ADD(name, 1) #define IP6STAT_DEC(name) IP6STAT_SUB(name, 1) #endif #ifdef _KERNEL /* * IPv6 onion peeling state. * it will be initialized when we come into ip6_input(). * XXX do not make it a kitchen sink! */ struct ip6aux { u_int32_t ip6a_flags; #define IP6A_SWAP 0x01 /* swapped home/care-of on packet */ #define IP6A_HASEEN 0x02 /* HA was present */ #define IP6A_BRUID 0x04 /* BR Unique Identifier was present */ #define IP6A_RTALERTSEEN 0x08 /* rtalert present */ /* ip6.ip6_src */ struct in6_addr ip6a_careof; /* care-of address of the peer */ struct in6_addr ip6a_home; /* home address of the peer */ u_int16_t ip6a_bruid; /* BR unique identifier */ /* ip6.ip6_dst */ struct in6_ifaddr *ip6a_dstia6; /* my ifaddr that matches ip6_dst */ /* rtalert */ u_int16_t ip6a_rtalert; /* rtalert option value */ /* * decapsulation history will be here. * with IPsec it may not be accurate. */ }; #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(struct ip6stat, ip6stat); /* statistics */ 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_ip6stat VNET(ip6stat) #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_keepfaith); /* Firewall Aided Internet Translator */ 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_keepfaith VNET(ip6_keepfaith) #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 __P((struct socket *, struct sockopt *sopt)); struct in6_ifaddr; void ip6_init __P((void)); #ifdef VIMAGE void ip6_destroy __P((void)); #endif int ip6proto_register(short); int ip6proto_unregister(short); void ip6_input __P((struct mbuf *)); struct in6_ifaddr *ip6_getdstifaddr __P((struct mbuf *)); void ip6_freepcbopts __P((struct ip6_pktopts *)); int ip6_unknown_opt __P((u_int8_t *, struct mbuf *, int)); char * ip6_get_prevhdr __P((struct mbuf *, int)); int ip6_nexthdr __P((struct mbuf *, int, int, int *)); int ip6_lasthdr __P((struct mbuf *, int, int, int *)); struct ip6aux *ip6_addaux __P((struct mbuf *)); struct ip6aux *ip6_findaux __P((struct mbuf *)); void ip6_delaux __P((struct mbuf *)); extern int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *); int ip6_process_hopopts __P((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 __P((struct inpcb *, struct mbuf *, struct mbuf **)); void ip6_notify_pmtu __P((struct inpcb *, struct sockaddr_in6 *, u_int32_t *)); int ip6_sysctl __P((int *, u_int, void *, size_t *, void *, size_t)); void ip6_forward __P((struct mbuf *, int)); void ip6_mloopback __P((struct ifnet *, struct mbuf *, struct sockaddr_in6 *)); int ip6_output __P((struct mbuf *, struct ip6_pktopts *, struct route_in6 *, int, struct ip6_moptions *, struct ifnet **, struct inpcb *)); int ip6_ctloutput __P((struct socket *, struct sockopt *)); int ip6_raw_ctloutput __P((struct socket *, struct sockopt *)); void ip6_initpktopts __P((struct ip6_pktopts *)); int ip6_setpktopts __P((struct mbuf *, struct ip6_pktopts *, struct ip6_pktopts *, struct ucred *, int)); void ip6_clearpktopts __P((struct ip6_pktopts *, int)); struct ip6_pktopts *ip6_copypktopts __P((struct ip6_pktopts *, int)); int ip6_optlen __P((struct inpcb *)); int route6_input __P((struct mbuf **, int *, int)); void frag6_init __P((void)); int frag6_input __P((struct mbuf **, int *, int)); void frag6_slowtimo __P((void)); void frag6_drain __P((void)); void rip6_init __P((void)); int rip6_input __P((struct mbuf **, int *, int)); void rip6_ctlinput __P((int, struct sockaddr *, void *)); int rip6_ctloutput __P((struct socket *, struct sockopt *)); int rip6_output __P((struct mbuf *, ...)); int rip6_usrreq __P((struct socket *, int, struct mbuf *, struct mbuf *, struct mbuf *, struct thread *)); int dest6_input __P((struct mbuf **, int *, int)); int none_input __P((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 __P((struct sockaddr_in6 *, struct ip6_pktopts *, struct ip6_moptions *, struct route_in6 *, struct ifnet **, struct rtentry **)); u_int32_t ip6_randomid __P((void)); u_int32_t ip6_randomflowlabel __P((void)); #endif /* _KERNEL */ #endif /* !_NETINET6_IP6_VAR_H_ */ Index: projects/largeSMP/sys/netinet6/nd6.c =================================================================== --- projects/largeSMP/sys/netinet6/nd6.c (revision 222782) +++ projects/largeSMP/sys/netinet6/nd6.c (revision 222783) @@ -1,2341 +1,2373 @@ /*- * 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.c,v 1.144 2001/05/24 07:44:00 itojun Exp $ */ #include __FBSDID("$FreeBSD$"); #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 #define L3_ADDR_SIN6(le) ((struct sockaddr_in6 *) L3_ADDR(le)) #include #include #include #include #include #include #include #include #include #include #include #define ND6_SLOWTIMER_INTERVAL (60 * 60) /* 1 hour */ #define ND6_RECALC_REACHTM_INTERVAL (60 * 120) /* 2 hours */ #define SIN6(s) ((struct sockaddr_in6 *)s) /* timer values */ VNET_DEFINE(int, nd6_prune) = 1; /* walk list every 1 seconds */ VNET_DEFINE(int, nd6_delay) = 5; /* delay first probe time 5 second */ VNET_DEFINE(int, nd6_umaxtries) = 3; /* maximum unicast query */ VNET_DEFINE(int, nd6_mmaxtries) = 3; /* maximum multicast query */ VNET_DEFINE(int, nd6_useloopback) = 1; /* use loopback interface for * local traffic */ VNET_DEFINE(int, nd6_gctimer) = (60 * 60 * 24); /* 1 day: garbage * collection timer */ /* preventing too many loops in ND option parsing */ static VNET_DEFINE(int, nd6_maxndopt) = 10; /* max # of ND options allowed */ VNET_DEFINE(int, nd6_maxnudhint) = 0; /* max # of subsequent upper * layer hints */ static VNET_DEFINE(int, nd6_maxqueuelen) = 1; /* max pkts cached in unresolved * ND entries */ #define V_nd6_maxndopt VNET(nd6_maxndopt) #define V_nd6_maxqueuelen VNET(nd6_maxqueuelen) #ifdef ND6_DEBUG VNET_DEFINE(int, nd6_debug) = 1; #else VNET_DEFINE(int, nd6_debug) = 0; #endif /* for debugging? */ #if 0 static int nd6_inuse, nd6_allocated; #endif VNET_DEFINE(struct nd_drhead, nd_defrouter); VNET_DEFINE(struct nd_prhead, nd_prefix); VNET_DEFINE(int, nd6_recalc_reachtm_interval) = ND6_RECALC_REACHTM_INTERVAL; #define V_nd6_recalc_reachtm_interval VNET(nd6_recalc_reachtm_interval) static struct sockaddr_in6 all1_sa; int (*send_sendso_input_hook)(struct mbuf *, struct ifnet *, int, int); static int nd6_is_new_addr_neighbor __P((struct sockaddr_in6 *, struct ifnet *)); static void nd6_setmtu0(struct ifnet *, struct nd_ifinfo *); static void nd6_slowtimo(void *); static int regen_tmpaddr(struct in6_ifaddr *); static struct llentry *nd6_free(struct llentry *, int); static void nd6_llinfo_timer(void *); static void clear_llinfo_pqueue(struct llentry *); static VNET_DEFINE(struct callout, nd6_slowtimo_ch); #define V_nd6_slowtimo_ch VNET(nd6_slowtimo_ch) VNET_DEFINE(struct callout, nd6_timer_ch); void nd6_init(void) { int i; LIST_INIT(&V_nd_prefix); all1_sa.sin6_family = AF_INET6; all1_sa.sin6_len = sizeof(struct sockaddr_in6); for (i = 0; i < sizeof(all1_sa.sin6_addr); i++) all1_sa.sin6_addr.s6_addr[i] = 0xff; /* initialization of the default router list */ TAILQ_INIT(&V_nd_defrouter); /* start timer */ callout_init(&V_nd6_slowtimo_ch, 0); callout_reset(&V_nd6_slowtimo_ch, ND6_SLOWTIMER_INTERVAL * hz, nd6_slowtimo, curvnet); } #ifdef VIMAGE void nd6_destroy() { callout_drain(&V_nd6_slowtimo_ch); callout_drain(&V_nd6_timer_ch); } #endif struct nd_ifinfo * nd6_ifattach(struct ifnet *ifp) { struct nd_ifinfo *nd; nd = (struct nd_ifinfo *)malloc(sizeof(*nd), M_IP6NDP, M_WAITOK); bzero(nd, sizeof(*nd)); nd->initialized = 1; nd->chlim = IPV6_DEFHLIM; nd->basereachable = REACHABLE_TIME; nd->reachable = ND_COMPUTE_RTIME(nd->basereachable); nd->retrans = RETRANS_TIMER; nd->flags = ND6_IFF_PERFORMNUD; /* A loopback interface always has ND6_IFF_AUTO_LINKLOCAL. */ if (V_ip6_auto_linklocal || (ifp->if_flags & IFF_LOOPBACK)) nd->flags |= ND6_IFF_AUTO_LINKLOCAL; /* A loopback interface does not need to accept RTADV. */ if (V_ip6_accept_rtadv && !(ifp->if_flags & IFF_LOOPBACK)) nd->flags |= ND6_IFF_ACCEPT_RTADV; + if (V_ip6_no_radr && !(ifp->if_flags & IFF_LOOPBACK)) + nd->flags |= ND6_IFF_NO_RADR; /* XXX: we cannot call nd6_setmtu since ifp is not fully initialized */ nd6_setmtu0(ifp, nd); return nd; } void nd6_ifdetach(struct nd_ifinfo *nd) { free(nd, M_IP6NDP); } /* * Reset ND level link MTU. This function is called when the physical MTU * changes, which means we might have to adjust the ND level MTU. */ void nd6_setmtu(struct ifnet *ifp) { nd6_setmtu0(ifp, ND_IFINFO(ifp)); } /* XXX todo: do not maintain copy of ifp->if_mtu in ndi->maxmtu */ void nd6_setmtu0(struct ifnet *ifp, struct nd_ifinfo *ndi) { u_int32_t omaxmtu; omaxmtu = ndi->maxmtu; switch (ifp->if_type) { case IFT_ARCNET: ndi->maxmtu = MIN(ARC_PHDS_MAXMTU, ifp->if_mtu); /* RFC2497 */ break; case IFT_FDDI: ndi->maxmtu = MIN(FDDIIPMTU, ifp->if_mtu); /* RFC2467 */ break; case IFT_ISO88025: ndi->maxmtu = MIN(ISO88025_MAX_MTU, ifp->if_mtu); break; default: ndi->maxmtu = ifp->if_mtu; break; } /* * Decreasing the interface MTU under IPV6 minimum MTU may cause * undesirable situation. We thus notify the operator of the change * explicitly. The check for omaxmtu is necessary to restrict the * log to the case of changing the MTU, not initializing it. */ if (omaxmtu >= IPV6_MMTU && ndi->maxmtu < IPV6_MMTU) { log(LOG_NOTICE, "nd6_setmtu0: " "new link MTU on %s (%lu) is too small for IPv6\n", if_name(ifp), (unsigned long)ndi->maxmtu); } if (ndi->maxmtu > V_in6_maxmtu) in6_setmaxmtu(); /* check all interfaces just in case */ } void nd6_option_init(void *opt, int icmp6len, union nd_opts *ndopts) { bzero(ndopts, sizeof(*ndopts)); ndopts->nd_opts_search = (struct nd_opt_hdr *)opt; ndopts->nd_opts_last = (struct nd_opt_hdr *)(((u_char *)opt) + icmp6len); if (icmp6len == 0) { ndopts->nd_opts_done = 1; ndopts->nd_opts_search = NULL; } } /* * Take one ND option. */ struct nd_opt_hdr * nd6_option(union nd_opts *ndopts) { struct nd_opt_hdr *nd_opt; int olen; if (ndopts == NULL) panic("ndopts == NULL in nd6_option"); if (ndopts->nd_opts_last == NULL) panic("uninitialized ndopts in nd6_option"); if (ndopts->nd_opts_search == NULL) return NULL; if (ndopts->nd_opts_done) return NULL; nd_opt = ndopts->nd_opts_search; /* make sure nd_opt_len is inside the buffer */ if ((caddr_t)&nd_opt->nd_opt_len >= (caddr_t)ndopts->nd_opts_last) { bzero(ndopts, sizeof(*ndopts)); return NULL; } olen = nd_opt->nd_opt_len << 3; if (olen == 0) { /* * Message validation requires that all included * options have a length that is greater than zero. */ bzero(ndopts, sizeof(*ndopts)); return NULL; } ndopts->nd_opts_search = (struct nd_opt_hdr *)((caddr_t)nd_opt + olen); if (ndopts->nd_opts_search > ndopts->nd_opts_last) { /* option overruns the end of buffer, invalid */ bzero(ndopts, sizeof(*ndopts)); return NULL; } else if (ndopts->nd_opts_search == ndopts->nd_opts_last) { /* reached the end of options chain */ ndopts->nd_opts_done = 1; ndopts->nd_opts_search = NULL; } return nd_opt; } /* * Parse multiple ND options. * This function is much easier to use, for ND routines that do not need * multiple options of the same type. */ int nd6_options(union nd_opts *ndopts) { struct nd_opt_hdr *nd_opt; int i = 0; if (ndopts == NULL) panic("ndopts == NULL in nd6_options"); if (ndopts->nd_opts_last == NULL) panic("uninitialized ndopts in nd6_options"); if (ndopts->nd_opts_search == NULL) return 0; while (1) { nd_opt = nd6_option(ndopts); if (nd_opt == NULL && ndopts->nd_opts_last == NULL) { /* * Message validation requires that all included * options have a length that is greater than zero. */ ICMP6STAT_INC(icp6s_nd_badopt); bzero(ndopts, sizeof(*ndopts)); return -1; } if (nd_opt == NULL) goto skip1; switch (nd_opt->nd_opt_type) { case ND_OPT_SOURCE_LINKADDR: case ND_OPT_TARGET_LINKADDR: case ND_OPT_MTU: case ND_OPT_REDIRECTED_HEADER: if (ndopts->nd_opt_array[nd_opt->nd_opt_type]) { nd6log((LOG_INFO, "duplicated ND6 option found (type=%d)\n", nd_opt->nd_opt_type)); /* XXX bark? */ } else { ndopts->nd_opt_array[nd_opt->nd_opt_type] = nd_opt; } break; case ND_OPT_PREFIX_INFORMATION: if (ndopts->nd_opt_array[nd_opt->nd_opt_type] == 0) { ndopts->nd_opt_array[nd_opt->nd_opt_type] = nd_opt; } ndopts->nd_opts_pi_end = (struct nd_opt_prefix_info *)nd_opt; break; default: /* * Unknown options must be silently ignored, * to accomodate future extension to the protocol. */ nd6log((LOG_DEBUG, "nd6_options: unsupported option %d - " "option ignored\n", nd_opt->nd_opt_type)); } skip1: i++; if (i > V_nd6_maxndopt) { ICMP6STAT_INC(icp6s_nd_toomanyopt); nd6log((LOG_INFO, "too many loop in nd opt\n")); break; } if (ndopts->nd_opts_done) break; } return 0; } /* * ND6 timer routine to handle ND6 entries */ void nd6_llinfo_settimer_locked(struct llentry *ln, long tick) { int canceled; LLE_WLOCK_ASSERT(ln); if (tick < 0) { ln->la_expire = 0; ln->ln_ntick = 0; canceled = callout_stop(&ln->ln_timer_ch); } else { ln->la_expire = time_second + tick / hz; LLE_ADDREF(ln); if (tick > INT_MAX) { ln->ln_ntick = tick - INT_MAX; canceled = callout_reset(&ln->ln_timer_ch, INT_MAX, nd6_llinfo_timer, ln); } else { ln->ln_ntick = 0; canceled = callout_reset(&ln->ln_timer_ch, tick, nd6_llinfo_timer, ln); } } if (canceled) LLE_REMREF(ln); } void nd6_llinfo_settimer(struct llentry *ln, long tick) { LLE_WLOCK(ln); nd6_llinfo_settimer_locked(ln, tick); LLE_WUNLOCK(ln); } static void nd6_llinfo_timer(void *arg) { struct llentry *ln; struct in6_addr *dst; struct ifnet *ifp; struct nd_ifinfo *ndi = NULL; KASSERT(arg != NULL, ("%s: arg NULL", __func__)); ln = (struct llentry *)arg; LLE_WLOCK_ASSERT(ln); ifp = ln->lle_tbl->llt_ifp; CURVNET_SET(ifp->if_vnet); if (ln->ln_ntick > 0) { if (ln->ln_ntick > INT_MAX) { ln->ln_ntick -= INT_MAX; nd6_llinfo_settimer_locked(ln, INT_MAX); } else { ln->ln_ntick = 0; nd6_llinfo_settimer_locked(ln, ln->ln_ntick); } goto done; } ndi = ND_IFINFO(ifp); dst = &L3_ADDR_SIN6(ln)->sin6_addr; if (ln->la_flags & LLE_STATIC) { goto done; } if (ln->la_flags & LLE_DELETED) { (void)nd6_free(ln, 0); ln = NULL; goto done; } switch (ln->ln_state) { case ND6_LLINFO_INCOMPLETE: if (ln->la_asked < V_nd6_mmaxtries) { ln->la_asked++; nd6_llinfo_settimer_locked(ln, (long)ndi->retrans * hz / 1000); LLE_WUNLOCK(ln); nd6_ns_output(ifp, NULL, dst, ln, 0); LLE_WLOCK(ln); } else { struct mbuf *m = ln->la_hold; if (m) { struct mbuf *m0; /* * assuming every packet in la_hold has the * same IP header. Send error after unlock. */ m0 = m->m_nextpkt; m->m_nextpkt = NULL; ln->la_hold = m0; clear_llinfo_pqueue(ln); } (void)nd6_free(ln, 0); ln = NULL; if (m != NULL) icmp6_error2(m, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_ADDR, 0, ifp); } break; case ND6_LLINFO_REACHABLE: if (!ND6_LLINFO_PERMANENT(ln)) { ln->ln_state = ND6_LLINFO_STALE; nd6_llinfo_settimer_locked(ln, (long)V_nd6_gctimer * hz); } break; case ND6_LLINFO_STALE: /* Garbage Collection(RFC 2461 5.3) */ if (!ND6_LLINFO_PERMANENT(ln)) { (void)nd6_free(ln, 1); ln = NULL; } break; case ND6_LLINFO_DELAY: if (ndi && (ndi->flags & ND6_IFF_PERFORMNUD) != 0) { /* We need NUD */ ln->la_asked = 1; ln->ln_state = ND6_LLINFO_PROBE; nd6_llinfo_settimer_locked(ln, (long)ndi->retrans * hz / 1000); LLE_WUNLOCK(ln); nd6_ns_output(ifp, dst, dst, ln, 0); LLE_WLOCK(ln); } else { ln->ln_state = ND6_LLINFO_STALE; /* XXX */ nd6_llinfo_settimer_locked(ln, (long)V_nd6_gctimer * hz); } break; case ND6_LLINFO_PROBE: if (ln->la_asked < V_nd6_umaxtries) { ln->la_asked++; nd6_llinfo_settimer_locked(ln, (long)ndi->retrans * hz / 1000); LLE_WUNLOCK(ln); nd6_ns_output(ifp, dst, dst, ln, 0); LLE_WLOCK(ln); } else { (void)nd6_free(ln, 0); ln = NULL; } break; default: panic("%s: paths in a dark night can be confusing: %d", __func__, ln->ln_state); } done: if (ln != NULL) LLE_FREE_LOCKED(ln); CURVNET_RESTORE(); } /* * ND6 timer routine to expire default route list and prefix list */ void nd6_timer(void *arg) { CURVNET_SET((struct vnet *) arg); int s; struct nd_defrouter *dr; struct nd_prefix *pr; struct in6_ifaddr *ia6, *nia6; struct in6_addrlifetime *lt6; callout_reset(&V_nd6_timer_ch, V_nd6_prune * hz, nd6_timer, curvnet); /* expire default router list */ s = splnet(); dr = TAILQ_FIRST(&V_nd_defrouter); while (dr) { if (dr->expire && dr->expire < time_second) { struct nd_defrouter *t; t = TAILQ_NEXT(dr, dr_entry); defrtrlist_del(dr); dr = t; } else { dr = TAILQ_NEXT(dr, dr_entry); } } /* * expire interface addresses. * in the past the loop was inside prefix expiry processing. * However, from a stricter speci-confrmance standpoint, we should * rather separate address lifetimes and prefix lifetimes. * * XXXRW: in6_ifaddrhead locking. */ addrloop: TAILQ_FOREACH_SAFE(ia6, &V_in6_ifaddrhead, ia_link, nia6) { /* check address lifetime */ lt6 = &ia6->ia6_lifetime; if (IFA6_IS_INVALID(ia6)) { int regen = 0; /* * If the expiring address is temporary, try * regenerating a new one. This would be useful when * we suspended a laptop PC, then turned it on after a * period that could invalidate all temporary * addresses. Although we may have to restart the * loop (see below), it must be after purging the * address. Otherwise, we'd see an infinite loop of * regeneration. */ if (V_ip6_use_tempaddr && (ia6->ia6_flags & IN6_IFF_TEMPORARY) != 0) { if (regen_tmpaddr(ia6) == 0) regen = 1; } in6_purgeaddr(&ia6->ia_ifa); if (regen) goto addrloop; /* XXX: see below */ } else if (IFA6_IS_DEPRECATED(ia6)) { int oldflags = ia6->ia6_flags; ia6->ia6_flags |= IN6_IFF_DEPRECATED; /* * If a temporary address has just become deprecated, * regenerate a new one if possible. */ if (V_ip6_use_tempaddr && (ia6->ia6_flags & IN6_IFF_TEMPORARY) != 0 && (oldflags & IN6_IFF_DEPRECATED) == 0) { if (regen_tmpaddr(ia6) == 0) { /* * A new temporary address is * generated. * XXX: this means the address chain * has changed while we are still in * the loop. Although the change * would not cause disaster (because * it's not a deletion, but an * addition,) we'd rather restart the * loop just for safety. Or does this * significantly reduce performance?? */ goto addrloop; } } } else { /* * A new RA might have made a deprecated address * preferred. */ ia6->ia6_flags &= ~IN6_IFF_DEPRECATED; } } /* expire prefix list */ pr = V_nd_prefix.lh_first; while (pr) { /* * check prefix lifetime. * since pltime is just for autoconf, pltime processing for * prefix is not necessary. */ if (pr->ndpr_vltime != ND6_INFINITE_LIFETIME && time_second - pr->ndpr_lastupdate > pr->ndpr_vltime) { struct nd_prefix *t; t = pr->ndpr_next; /* * address expiration and prefix expiration are * separate. NEVER perform in6_purgeaddr here. */ prelist_remove(pr); pr = t; } else pr = pr->ndpr_next; } splx(s); CURVNET_RESTORE(); } /* * ia6 - deprecated/invalidated temporary address */ static int regen_tmpaddr(struct in6_ifaddr *ia6) { struct ifaddr *ifa; struct ifnet *ifp; struct in6_ifaddr *public_ifa6 = NULL; ifp = ia6->ia_ifa.ifa_ifp; IF_ADDR_LOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { struct in6_ifaddr *it6; if (ifa->ifa_addr->sa_family != AF_INET6) continue; it6 = (struct in6_ifaddr *)ifa; /* ignore no autoconf addresses. */ if ((it6->ia6_flags & IN6_IFF_AUTOCONF) == 0) continue; /* ignore autoconf addresses with different prefixes. */ if (it6->ia6_ndpr == NULL || it6->ia6_ndpr != ia6->ia6_ndpr) continue; /* * Now we are looking at an autoconf address with the same * prefix as ours. If the address is temporary and is still * preferred, do not create another one. It would be rare, but * could happen, for example, when we resume a laptop PC after * a long period. */ if ((it6->ia6_flags & IN6_IFF_TEMPORARY) != 0 && !IFA6_IS_DEPRECATED(it6)) { public_ifa6 = NULL; break; } /* * This is a public autoconf address that has the same prefix * as ours. If it is preferred, keep it. We can't break the * loop here, because there may be a still-preferred temporary * address with the prefix. */ if (!IFA6_IS_DEPRECATED(it6)) public_ifa6 = it6; if (public_ifa6 != NULL) ifa_ref(&public_ifa6->ia_ifa); } IF_ADDR_UNLOCK(ifp); if (public_ifa6 != NULL) { int e; if ((e = in6_tmpifadd(public_ifa6, 0, 0)) != 0) { ifa_free(&public_ifa6->ia_ifa); log(LOG_NOTICE, "regen_tmpaddr: failed to create a new" " tmp addr,errno=%d\n", e); return (-1); } ifa_free(&public_ifa6->ia_ifa); return (0); } return (-1); } /* * Nuke neighbor cache/prefix/default router management table, right before * ifp goes away. */ void nd6_purge(struct ifnet *ifp) { struct nd_defrouter *dr, *ndr; struct nd_prefix *pr, *npr; /* * Nuke default router list entries toward ifp. * We defer removal of default router list entries that is installed * in the routing table, in order to keep additional side effects as * small as possible. */ for (dr = TAILQ_FIRST(&V_nd_defrouter); dr; dr = ndr) { ndr = TAILQ_NEXT(dr, dr_entry); if (dr->installed) continue; if (dr->ifp == ifp) defrtrlist_del(dr); } for (dr = TAILQ_FIRST(&V_nd_defrouter); dr; dr = ndr) { ndr = TAILQ_NEXT(dr, dr_entry); if (!dr->installed) continue; if (dr->ifp == ifp) defrtrlist_del(dr); } /* Nuke prefix list entries toward ifp */ for (pr = V_nd_prefix.lh_first; pr; pr = npr) { npr = pr->ndpr_next; if (pr->ndpr_ifp == ifp) { /* * Because if_detach() does *not* release prefixes * while purging addresses the reference count will * still be above zero. We therefore reset it to * make sure that the prefix really gets purged. */ pr->ndpr_refcnt = 0; /* * Previously, pr->ndpr_addr is removed as well, * but I strongly believe we don't have to do it. * nd6_purge() is only called from in6_ifdetach(), * which removes all the associated interface addresses * by itself. * (jinmei@kame.net 20010129) */ prelist_remove(pr); } } /* cancel default outgoing interface setting */ if (V_nd6_defifindex == ifp->if_index) nd6_setdefaultiface(0); - if (!V_ip6_forwarding && ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV) { + if (ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV) { /* Refresh default router list. */ defrouter_select(); } /* XXXXX * We do not nuke the neighbor cache entries here any more * because the neighbor cache is kept in if_afdata[AF_INET6]. * nd6_purge() is invoked by in6_ifdetach() which is called * from if_detach() where everything gets purged. So let * in6_domifdetach() do the actual L2 table purging work. */ } /* * the caller acquires and releases the lock on the lltbls * Returns the llentry locked */ struct llentry * nd6_lookup(struct in6_addr *addr6, int flags, struct ifnet *ifp) { struct sockaddr_in6 sin6; struct llentry *ln; int llflags; bzero(&sin6, sizeof(sin6)); sin6.sin6_len = sizeof(struct sockaddr_in6); sin6.sin6_family = AF_INET6; sin6.sin6_addr = *addr6; IF_AFDATA_LOCK_ASSERT(ifp); llflags = 0; if (flags & ND6_CREATE) llflags |= LLE_CREATE; if (flags & ND6_EXCLUSIVE) llflags |= LLE_EXCLUSIVE; ln = lla_lookup(LLTABLE6(ifp), llflags, (struct sockaddr *)&sin6); if ((ln != NULL) && (llflags & LLE_CREATE)) ln->ln_state = ND6_LLINFO_NOSTATE; return (ln); } /* * Test whether a given IPv6 address is a neighbor or not, ignoring * the actual neighbor cache. The neighbor cache is ignored in order * to not reenter the routing code from within itself. */ static int nd6_is_new_addr_neighbor(struct sockaddr_in6 *addr, struct ifnet *ifp) { struct nd_prefix *pr; struct ifaddr *dstaddr; /* * A link-local address is always a neighbor. * XXX: a link does not necessarily specify a single interface. */ if (IN6_IS_ADDR_LINKLOCAL(&addr->sin6_addr)) { struct sockaddr_in6 sin6_copy; u_int32_t zone; /* * We need sin6_copy since sa6_recoverscope() may modify the * content (XXX). */ sin6_copy = *addr; if (sa6_recoverscope(&sin6_copy)) return (0); /* XXX: should be impossible */ if (in6_setscope(&sin6_copy.sin6_addr, ifp, &zone)) return (0); if (sin6_copy.sin6_scope_id == zone) return (1); else return (0); } /* * If the address matches one of our addresses, * it should be a neighbor. * If the address matches one of our on-link prefixes, it should be a * neighbor. */ for (pr = V_nd_prefix.lh_first; pr; pr = pr->ndpr_next) { if (pr->ndpr_ifp != ifp) continue; if (!(pr->ndpr_stateflags & NDPRF_ONLINK)) { struct rtentry *rt; rt = rtalloc1((struct sockaddr *)&pr->ndpr_prefix, 0, 0); if (rt == NULL) continue; /* * This is the case where multiple interfaces * have the same prefix, but only one is installed * into the routing table and that prefix entry * is not the one being examined here. In the case * where RADIX_MPATH is enabled, multiple route * entries (of the same rt_key value) will be * installed because the interface addresses all * differ. */ if (!IN6_ARE_ADDR_EQUAL(&pr->ndpr_prefix.sin6_addr, &((struct sockaddr_in6 *)rt_key(rt))->sin6_addr)) { RTFREE_LOCKED(rt); continue; } RTFREE_LOCKED(rt); } if (IN6_ARE_MASKED_ADDR_EQUAL(&pr->ndpr_prefix.sin6_addr, &addr->sin6_addr, &pr->ndpr_mask)) return (1); } /* * If the address is assigned on the node of the other side of * a p2p interface, the address should be a neighbor. */ dstaddr = ifa_ifwithdstaddr((struct sockaddr *)addr); if (dstaddr != NULL) { if (dstaddr->ifa_ifp == ifp) { ifa_free(dstaddr); return (1); } ifa_free(dstaddr); } /* * If the default router list is empty, all addresses are regarded * as on-link, and thus, as a neighbor. - * XXX: we restrict the condition to hosts, because routers usually do - * not have the "default router list". */ - if (!V_ip6_forwarding && TAILQ_FIRST(&V_nd_defrouter) == NULL && + if (ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV && + TAILQ_FIRST(&V_nd_defrouter) == NULL && V_nd6_defifindex == ifp->if_index) { return (1); } return (0); } /* * Detect if a given IPv6 address identifies a neighbor on a given link. * XXX: should take care of the destination of a p2p link? */ int nd6_is_addr_neighbor(struct sockaddr_in6 *addr, struct ifnet *ifp) { struct llentry *lle; int rc = 0; IF_AFDATA_UNLOCK_ASSERT(ifp); if (nd6_is_new_addr_neighbor(addr, ifp)) return (1); /* * Even if the address matches none of our addresses, it might be * in the neighbor cache. */ IF_AFDATA_LOCK(ifp); if ((lle = nd6_lookup(&addr->sin6_addr, 0, ifp)) != NULL) { LLE_RUNLOCK(lle); rc = 1; } IF_AFDATA_UNLOCK(ifp); return (rc); } /* * Free an nd6 llinfo entry. * Since the function would cause significant changes in the kernel, DO NOT * make it global, unless you have a strong reason for the change, and are sure * that the change is safe. */ static struct llentry * nd6_free(struct llentry *ln, int gc) { struct llentry *next; struct nd_defrouter *dr; struct ifnet *ifp; LLE_WLOCK_ASSERT(ln); /* * we used to have pfctlinput(PRC_HOSTDEAD) here. * even though it is not harmful, it was not really necessary. */ /* cancel timer */ nd6_llinfo_settimer_locked(ln, -1); ifp = ln->lle_tbl->llt_ifp; - if (!V_ip6_forwarding) { - + if (ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV) { dr = defrouter_lookup(&L3_ADDR_SIN6(ln)->sin6_addr, ifp); if (dr != NULL && dr->expire && ln->ln_state == ND6_LLINFO_STALE && gc) { /* * If the reason for the deletion is just garbage * collection, and the neighbor is an active default * router, do not delete it. Instead, reset the GC * timer using the router's lifetime. * Simply deleting the entry would affect default * router selection, which is not necessarily a good * thing, especially when we're using router preference * values. * XXX: the check for ln_state would be redundant, * but we intentionally keep it just in case. */ if (dr->expire > time_second) nd6_llinfo_settimer_locked(ln, (dr->expire - time_second) * hz); else nd6_llinfo_settimer_locked(ln, (long)V_nd6_gctimer * hz); next = LIST_NEXT(ln, lle_next); LLE_REMREF(ln); LLE_WUNLOCK(ln); return (next); } if (dr) { /* * Unreachablity of a router might affect the default * router selection and on-link detection of advertised * prefixes. */ /* * Temporarily fake the state to choose a new default * router and to perform on-link determination of * prefixes correctly. * Below the state will be set correctly, * or the entry itself will be deleted. */ ln->ln_state = ND6_LLINFO_INCOMPLETE; } if (ln->ln_router || dr) { /* * We need to unlock to avoid a LOR with rt6_flush() with the * rnh and for the calls to pfxlist_onlink_check() and * defrouter_select() in the block further down for calls * into nd6_lookup(). We still hold a ref. */ LLE_WUNLOCK(ln); /* * rt6_flush must be called whether or not the neighbor * is in the Default Router List. * See a corresponding comment in nd6_na_input(). */ rt6_flush(&L3_ADDR_SIN6(ln)->sin6_addr, ifp); } if (dr) { /* * Since defrouter_select() does not affect the * on-link determination and MIP6 needs the check * before the default router selection, we perform * the check now. */ pfxlist_onlink_check(); /* * Refresh default router list. */ defrouter_select(); } if (ln->ln_router || dr) LLE_WLOCK(ln); } /* * Before deleting the entry, remember the next entry as the * return value. We need this because pfxlist_onlink_check() above * might have freed other entries (particularly the old next entry) as * a side effect (XXX). */ next = LIST_NEXT(ln, lle_next); /* * Save to unlock. We still hold an extra reference and will not * free(9) in llentry_free() if someone else holds one as well. */ LLE_WUNLOCK(ln); IF_AFDATA_LOCK(ifp); LLE_WLOCK(ln); LLE_REMREF(ln); llentry_free(ln); IF_AFDATA_UNLOCK(ifp); return (next); } /* * Upper-layer reachability hint for Neighbor Unreachability Detection. * * XXX cost-effective methods? */ void nd6_nud_hint(struct rtentry *rt, struct in6_addr *dst6, int force) { struct llentry *ln; struct ifnet *ifp; if ((dst6 == NULL) || (rt == NULL)) return; ifp = rt->rt_ifp; IF_AFDATA_LOCK(ifp); ln = nd6_lookup(dst6, ND6_EXCLUSIVE, NULL); IF_AFDATA_UNLOCK(ifp); if (ln == NULL) return; if (ln->ln_state < ND6_LLINFO_REACHABLE) goto done; /* * if we get upper-layer reachability confirmation many times, * it is possible we have false information. */ if (!force) { ln->ln_byhint++; if (ln->ln_byhint > V_nd6_maxnudhint) { goto done; } } ln->ln_state = ND6_LLINFO_REACHABLE; if (!ND6_LLINFO_PERMANENT(ln)) { nd6_llinfo_settimer_locked(ln, (long)ND_IFINFO(rt->rt_ifp)->reachable * hz); } done: LLE_WUNLOCK(ln); } int nd6_ioctl(u_long cmd, caddr_t data, struct ifnet *ifp) { struct in6_drlist *drl = (struct in6_drlist *)data; struct in6_oprlist *oprl = (struct in6_oprlist *)data; struct in6_ndireq *ndi = (struct in6_ndireq *)data; struct in6_nbrinfo *nbi = (struct in6_nbrinfo *)data; struct in6_ndifreq *ndif = (struct in6_ndifreq *)data; struct nd_defrouter *dr; struct nd_prefix *pr; int i = 0, error = 0; int s; switch (cmd) { case SIOCGDRLST_IN6: /* * obsolete API, use sysctl under net.inet6.icmp6 */ bzero(drl, sizeof(*drl)); s = splnet(); dr = TAILQ_FIRST(&V_nd_defrouter); while (dr && i < DRLSTSIZ) { drl->defrouter[i].rtaddr = dr->rtaddr; in6_clearscope(&drl->defrouter[i].rtaddr); drl->defrouter[i].flags = dr->flags; drl->defrouter[i].rtlifetime = dr->rtlifetime; drl->defrouter[i].expire = dr->expire; drl->defrouter[i].if_index = dr->ifp->if_index; i++; dr = TAILQ_NEXT(dr, dr_entry); } splx(s); break; case SIOCGPRLST_IN6: /* * obsolete API, use sysctl under net.inet6.icmp6 * * XXX the structure in6_prlist was changed in backward- * incompatible manner. in6_oprlist is used for SIOCGPRLST_IN6, * in6_prlist is used for nd6_sysctl() - fill_prlist(). */ /* * XXX meaning of fields, especialy "raflags", is very * differnet between RA prefix list and RR/static prefix list. * how about separating ioctls into two? */ bzero(oprl, sizeof(*oprl)); s = splnet(); pr = V_nd_prefix.lh_first; while (pr && i < PRLSTSIZ) { struct nd_pfxrouter *pfr; int j; oprl->prefix[i].prefix = pr->ndpr_prefix.sin6_addr; oprl->prefix[i].raflags = pr->ndpr_raf; oprl->prefix[i].prefixlen = pr->ndpr_plen; oprl->prefix[i].vltime = pr->ndpr_vltime; oprl->prefix[i].pltime = pr->ndpr_pltime; oprl->prefix[i].if_index = pr->ndpr_ifp->if_index; if (pr->ndpr_vltime == ND6_INFINITE_LIFETIME) oprl->prefix[i].expire = 0; else { time_t maxexpire; /* XXX: we assume time_t is signed. */ maxexpire = (-1) & ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1)); if (pr->ndpr_vltime < maxexpire - pr->ndpr_lastupdate) { oprl->prefix[i].expire = pr->ndpr_lastupdate + pr->ndpr_vltime; } else oprl->prefix[i].expire = maxexpire; } pfr = pr->ndpr_advrtrs.lh_first; j = 0; while (pfr) { if (j < DRLSTSIZ) { #define RTRADDR oprl->prefix[i].advrtr[j] RTRADDR = pfr->router->rtaddr; in6_clearscope(&RTRADDR); #undef RTRADDR } j++; pfr = pfr->pfr_next; } oprl->prefix[i].advrtrs = j; oprl->prefix[i].origin = PR_ORIG_RA; i++; pr = pr->ndpr_next; } splx(s); break; case OSIOCGIFINFO_IN6: #define ND ndi->ndi /* XXX: old ndp(8) assumes a positive value for linkmtu. */ bzero(&ND, sizeof(ND)); ND.linkmtu = IN6_LINKMTU(ifp); ND.maxmtu = ND_IFINFO(ifp)->maxmtu; ND.basereachable = ND_IFINFO(ifp)->basereachable; ND.reachable = ND_IFINFO(ifp)->reachable; ND.retrans = ND_IFINFO(ifp)->retrans; ND.flags = ND_IFINFO(ifp)->flags; ND.recalctm = ND_IFINFO(ifp)->recalctm; ND.chlim = ND_IFINFO(ifp)->chlim; break; case SIOCGIFINFO_IN6: ND = *ND_IFINFO(ifp); break; case SIOCSIFINFO_IN6: /* * used to change host variables from userland. * intented for a use on router to reflect RA configurations. */ /* 0 means 'unspecified' */ if (ND.linkmtu != 0) { if (ND.linkmtu < IPV6_MMTU || ND.linkmtu > IN6_LINKMTU(ifp)) { error = EINVAL; break; } ND_IFINFO(ifp)->linkmtu = ND.linkmtu; } if (ND.basereachable != 0) { int obasereachable = ND_IFINFO(ifp)->basereachable; ND_IFINFO(ifp)->basereachable = ND.basereachable; if (ND.basereachable != obasereachable) ND_IFINFO(ifp)->reachable = ND_COMPUTE_RTIME(ND.basereachable); } if (ND.retrans != 0) ND_IFINFO(ifp)->retrans = ND.retrans; if (ND.chlim != 0) ND_IFINFO(ifp)->chlim = ND.chlim; /* FALLTHROUGH */ case SIOCSIFINFO_FLAGS: { struct ifaddr *ifa; struct in6_ifaddr *ia; + /* + * Try to clear ifdisabled flag when enabling + * accept_rtadv or auto_linklocal. + */ if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) && + !(ND.flags & ND6_IFF_IFDISABLED) && + (ND.flags & (ND6_IFF_ACCEPT_RTADV | + ND6_IFF_AUTO_LINKLOCAL))) + ND.flags &= ~ND6_IFF_IFDISABLED; + + if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) && !(ND.flags & ND6_IFF_IFDISABLED)) { /* ifdisabled 1->0 transision */ /* * If the interface is marked as ND6_IFF_IFDISABLED and * has an link-local address with IN6_IFF_DUPLICATED, * do not clear ND6_IFF_IFDISABLED. * See RFC 4862, Section 5.4.5. */ int duplicated_linklocal = 0; IF_ADDR_LOCK(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_DUPLICATED) && - IN6_IS_ADDR_LINKLOCAL(&ia->ia_addr.sin6_addr)) { + IN6_IS_ADDR_LINKLOCAL(IA6_IN6(ia))) { duplicated_linklocal = 1; break; } } IF_ADDR_UNLOCK(ifp); if (duplicated_linklocal) { ND.flags |= ND6_IFF_IFDISABLED; log(LOG_ERR, "Cannot enable an interface" " with a link-local address marked" " duplicate.\n"); } else { ND_IFINFO(ifp)->flags &= ~ND6_IFF_IFDISABLED; in6_if_up(ifp); } } else if (!(ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) && (ND.flags & ND6_IFF_IFDISABLED)) { /* ifdisabled 0->1 transision */ /* Mark all IPv6 address as tentative. */ ND_IFINFO(ifp)->flags |= ND6_IFF_IFDISABLED; IF_ADDR_LOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; ia = (struct in6_ifaddr *)ifa; ia->ia6_flags |= IN6_IFF_TENTATIVE; } IF_ADDR_UNLOCK(ifp); } if (!(ND_IFINFO(ifp)->flags & ND6_IFF_AUTO_LINKLOCAL) && (ND.flags & ND6_IFF_AUTO_LINKLOCAL)) { /* auto_linklocal 0->1 transision */ /* If no link-local address on ifp, configure */ ND_IFINFO(ifp)->flags |= ND6_IFF_AUTO_LINKLOCAL; in6_ifattach(ifp, NULL); + } else if ((ND_IFINFO(ifp)->flags & ND6_IFF_AUTO_LINKLOCAL) && + !(ND.flags & ND6_IFF_IFDISABLED)) { + /* + * When the IF already has + * ND6_IFF_AUTO_LINKLOCAL and no link-local + * address is assigned, try to assign one. + */ + int haslinklocal = 0; + + IF_ADDR_LOCK(ifp); + TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { + if (ifa->ifa_addr->sa_family != AF_INET6) + continue; + ia = (struct in6_ifaddr *)ifa; + if (IN6_IS_ADDR_LINKLOCAL(IA6_IN6(ia))) { + haslinklocal = 1; + break; + } + } + IF_ADDR_UNLOCK(ifp); + if (!haslinklocal) + in6_ifattach(ifp, NULL); } } ND_IFINFO(ifp)->flags = ND.flags; break; #undef ND case SIOCSNDFLUSH_IN6: /* XXX: the ioctl name is confusing... */ /* sync kernel routing table with the default router list */ defrouter_reset(); defrouter_select(); break; case SIOCSPFXFLUSH_IN6: { /* flush all the prefix advertised by routers */ struct nd_prefix *pr, *next; s = splnet(); for (pr = V_nd_prefix.lh_first; pr; pr = next) { struct in6_ifaddr *ia, *ia_next; next = pr->ndpr_next; if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr)) continue; /* XXX */ /* do we really have to remove addresses as well? */ /* XXXRW: in6_ifaddrhead locking. */ TAILQ_FOREACH_SAFE(ia, &V_in6_ifaddrhead, ia_link, ia_next) { if ((ia->ia6_flags & IN6_IFF_AUTOCONF) == 0) continue; if (ia->ia6_ndpr == pr) in6_purgeaddr(&ia->ia_ifa); } prelist_remove(pr); } splx(s); break; } case SIOCSRTRFLUSH_IN6: { /* flush all the default routers */ struct nd_defrouter *dr, *next; s = splnet(); defrouter_reset(); for (dr = TAILQ_FIRST(&V_nd_defrouter); dr; dr = next) { next = TAILQ_NEXT(dr, dr_entry); defrtrlist_del(dr); } defrouter_select(); splx(s); break; } case SIOCGNBRINFO_IN6: { struct llentry *ln; struct in6_addr nb_addr = nbi->addr; /* make local for safety */ if ((error = in6_setscope(&nb_addr, ifp, NULL)) != 0) return (error); IF_AFDATA_LOCK(ifp); ln = nd6_lookup(&nb_addr, 0, ifp); IF_AFDATA_UNLOCK(ifp); if (ln == NULL) { error = EINVAL; break; } nbi->state = ln->ln_state; nbi->asked = ln->la_asked; nbi->isrouter = ln->ln_router; nbi->expire = ln->la_expire; LLE_RUNLOCK(ln); break; } case SIOCGDEFIFACE_IN6: /* XXX: should be implemented as a sysctl? */ ndif->ifindex = V_nd6_defifindex; break; case SIOCSDEFIFACE_IN6: /* XXX: should be implemented as a sysctl? */ return (nd6_setdefaultiface(ndif->ifindex)); } return (error); } /* * Create neighbor cache entry and cache link-layer address, * on reception of inbound ND6 packets. (RS/RA/NS/redirect) * * type - ICMP6 type * code - type dependent information * * XXXXX * The caller of this function already acquired the ndp * cache table lock because the cache entry is returned. */ struct llentry * nd6_cache_lladdr(struct ifnet *ifp, struct in6_addr *from, char *lladdr, int lladdrlen, int type, int code) { struct llentry *ln = NULL; int is_newentry; int do_update; int olladdr; int llchange; int flags; int newstate = 0; uint16_t router = 0; struct sockaddr_in6 sin6; struct mbuf *chain = NULL; int static_route = 0; IF_AFDATA_UNLOCK_ASSERT(ifp); if (ifp == NULL) panic("ifp == NULL in nd6_cache_lladdr"); if (from == NULL) panic("from == NULL in nd6_cache_lladdr"); /* nothing must be updated for unspecified address */ if (IN6_IS_ADDR_UNSPECIFIED(from)) return NULL; /* * Validation about ifp->if_addrlen and lladdrlen must be done in * the caller. * * XXX If the link does not have link-layer adderss, what should * we do? (ifp->if_addrlen == 0) * Spec says nothing in sections for RA, RS and NA. There's small * description on it in NS section (RFC 2461 7.2.3). */ flags = lladdr ? ND6_EXCLUSIVE : 0; IF_AFDATA_LOCK(ifp); ln = nd6_lookup(from, flags, ifp); if (ln == NULL) { flags |= ND6_EXCLUSIVE; ln = nd6_lookup(from, flags | ND6_CREATE, ifp); IF_AFDATA_UNLOCK(ifp); is_newentry = 1; } else { IF_AFDATA_UNLOCK(ifp); /* do nothing if static ndp is set */ if (ln->la_flags & LLE_STATIC) { static_route = 1; goto done; } is_newentry = 0; } if (ln == NULL) return (NULL); olladdr = (ln->la_flags & LLE_VALID) ? 1 : 0; if (olladdr && lladdr) { llchange = bcmp(lladdr, &ln->ll_addr, ifp->if_addrlen); } else llchange = 0; /* * newentry olladdr lladdr llchange (*=record) * 0 n n -- (1) * 0 y n -- (2) * 0 n y -- (3) * STALE * 0 y y n (4) * * 0 y y y (5) * STALE * 1 -- n -- (6) NOSTATE(= PASSIVE) * 1 -- y -- (7) * STALE */ if (lladdr) { /* (3-5) and (7) */ /* * Record source link-layer address * XXX is it dependent to ifp->if_type? */ bcopy(lladdr, &ln->ll_addr, ifp->if_addrlen); ln->la_flags |= LLE_VALID; } if (!is_newentry) { if ((!olladdr && lladdr != NULL) || /* (3) */ (olladdr && lladdr != NULL && llchange)) { /* (5) */ do_update = 1; newstate = ND6_LLINFO_STALE; } else /* (1-2,4) */ do_update = 0; } else { do_update = 1; if (lladdr == NULL) /* (6) */ newstate = ND6_LLINFO_NOSTATE; else /* (7) */ newstate = ND6_LLINFO_STALE; } if (do_update) { /* * Update the state of the neighbor cache. */ ln->ln_state = newstate; if (ln->ln_state == ND6_LLINFO_STALE) { /* * XXX: since nd6_output() below will cause * state tansition to DELAY and reset the timer, * we must set the timer now, although it is actually * meaningless. */ nd6_llinfo_settimer_locked(ln, (long)V_nd6_gctimer * hz); if (ln->la_hold) { struct mbuf *m_hold, *m_hold_next; /* * reset the la_hold in advance, to explicitly * prevent a la_hold lookup in nd6_output() * (wouldn't happen, though...) */ for (m_hold = ln->la_hold, ln->la_hold = NULL; m_hold; m_hold = m_hold_next) { m_hold_next = m_hold->m_nextpkt; m_hold->m_nextpkt = NULL; /* * we assume ifp is not a p2p here, so * just set the 2nd argument as the * 1st one. */ nd6_output_lle(ifp, ifp, m_hold, L3_ADDR_SIN6(ln), NULL, ln, &chain); } /* * If we have mbufs in the chain we need to do * deferred transmit. Copy the address from the * llentry before dropping the lock down below. */ if (chain != NULL) memcpy(&sin6, L3_ADDR_SIN6(ln), sizeof(sin6)); } } else if (ln->ln_state == ND6_LLINFO_INCOMPLETE) { /* probe right away */ nd6_llinfo_settimer_locked((void *)ln, 0); } } /* * ICMP6 type dependent behavior. * * NS: clear IsRouter if new entry * RS: clear IsRouter * RA: set IsRouter if there's lladdr * redir: clear IsRouter if new entry * * RA case, (1): * The spec says that we must set IsRouter in the following cases: * - If lladdr exist, set IsRouter. This means (1-5). * - If it is old entry (!newentry), set IsRouter. This means (7). * So, based on the spec, in (1-5) and (7) cases we must set IsRouter. * A quetion arises for (1) case. (1) case has no lladdr in the * neighbor cache, this is similar to (6). * This case is rare but we figured that we MUST NOT set IsRouter. * * newentry olladdr lladdr llchange NS RS RA redir * D R * 0 n n -- (1) c ? s * 0 y n -- (2) c s s * 0 n y -- (3) c s s * 0 y y n (4) c s s * 0 y y y (5) c s s * 1 -- n -- (6) c c c s * 1 -- y -- (7) c c s c s * * (c=clear s=set) */ switch (type & 0xff) { case ND_NEIGHBOR_SOLICIT: /* * New entry must have is_router flag cleared. */ if (is_newentry) /* (6-7) */ ln->ln_router = 0; break; case ND_REDIRECT: /* * If the icmp is a redirect to a better router, always set the * is_router flag. Otherwise, if the entry is newly created, * clear the flag. [RFC 2461, sec 8.3] */ if (code == ND_REDIRECT_ROUTER) ln->ln_router = 1; else if (is_newentry) /* (6-7) */ ln->ln_router = 0; break; case ND_ROUTER_SOLICIT: /* * is_router flag must always be cleared. */ ln->ln_router = 0; break; case ND_ROUTER_ADVERT: /* * Mark an entry with lladdr as a router. */ if ((!is_newentry && (olladdr || lladdr)) || /* (2-5) */ (is_newentry && lladdr)) { /* (7) */ ln->ln_router = 1; } break; } if (ln != NULL) { static_route = (ln->la_flags & LLE_STATIC); router = ln->ln_router; if (flags & ND6_EXCLUSIVE) LLE_WUNLOCK(ln); else LLE_RUNLOCK(ln); if (static_route) ln = NULL; } if (chain) nd6_output_flush(ifp, ifp, chain, &sin6, NULL); /* * When the link-layer address of a router changes, select the * best router again. In particular, when the neighbor entry is newly * created, it might affect the selection policy. * Question: can we restrict the first condition to the "is_newentry" * case? * XXX: when we hear an RA from a new router with the link-layer * address option, defrouter_select() is called twice, since * defrtrlist_update called the function as well. However, I believe * we can compromise the overhead, since it only happens the first * time. * XXX: although defrouter_select() should not have a bad effect * for those are not autoconfigured hosts, we explicitly avoid such * cases for safety. */ - if (do_update && router && !V_ip6_forwarding && + if (do_update && router && ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV) { /* * guaranteed recursion */ defrouter_select(); } return (ln); done: if (ln != NULL) { if (flags & ND6_EXCLUSIVE) LLE_WUNLOCK(ln); else LLE_RUNLOCK(ln); if (static_route) ln = NULL; } return (ln); } static void nd6_slowtimo(void *arg) { CURVNET_SET((struct vnet *) arg); struct nd_ifinfo *nd6if; struct ifnet *ifp; callout_reset(&V_nd6_slowtimo_ch, ND6_SLOWTIMER_INTERVAL * hz, nd6_slowtimo, curvnet); IFNET_RLOCK_NOSLEEP(); for (ifp = TAILQ_FIRST(&V_ifnet); ifp; ifp = TAILQ_NEXT(ifp, if_list)) { nd6if = ND_IFINFO(ifp); if (nd6if->basereachable && /* already initialized */ (nd6if->recalctm -= ND6_SLOWTIMER_INTERVAL) <= 0) { /* * Since reachable time rarely changes by router * advertisements, we SHOULD insure that a new random * value gets recomputed at least once every few hours. * (RFC 2461, 6.3.4) */ nd6if->recalctm = V_nd6_recalc_reachtm_interval; nd6if->reachable = ND_COMPUTE_RTIME(nd6if->basereachable); } } IFNET_RUNLOCK_NOSLEEP(); CURVNET_RESTORE(); } int nd6_output(struct ifnet *ifp, struct ifnet *origifp, struct mbuf *m0, struct sockaddr_in6 *dst, struct rtentry *rt0) { return (nd6_output_lle(ifp, origifp, m0, dst, rt0, NULL, NULL)); } /* * Note that I'm not enforcing any global serialization * lle state or asked changes here as the logic is too * complicated to avoid having to always acquire an exclusive * lock * KMM * */ #define senderr(e) { error = (e); goto bad;} int nd6_output_lle(struct ifnet *ifp, struct ifnet *origifp, struct mbuf *m0, struct sockaddr_in6 *dst, struct rtentry *rt0, struct llentry *lle, struct mbuf **chain) { struct mbuf *m = m0; struct m_tag *mtag; struct llentry *ln = lle; struct ip6_hdr *ip6; int error = 0; int flags = 0; int ip6len; #ifdef INVARIANTS if (lle != NULL) { LLE_WLOCK_ASSERT(lle); KASSERT(chain != NULL, (" lle locked but no mbuf chain pointer passed")); } #endif if (IN6_IS_ADDR_MULTICAST(&dst->sin6_addr)) goto sendpkt; if (nd6_need_cache(ifp) == 0) goto sendpkt; /* * next hop determination. This routine is derived from ether_output. */ /* * Address resolution or Neighbor Unreachability Detection * for the next hop. * At this point, the destination of the packet must be a unicast * or an anycast address(i.e. not a multicast). */ flags = ((m != NULL) || (lle != NULL)) ? LLE_EXCLUSIVE : 0; if (ln == NULL) { retry: IF_AFDATA_LOCK(ifp); ln = lla_lookup(LLTABLE6(ifp), flags, (struct sockaddr *)dst); IF_AFDATA_UNLOCK(ifp); if ((ln == NULL) && nd6_is_addr_neighbor(dst, ifp)) { /* * Since nd6_is_addr_neighbor() internally calls nd6_lookup(), * the condition below is not very efficient. But we believe * it is tolerable, because this should be a rare case. */ flags = ND6_CREATE | (m ? ND6_EXCLUSIVE : 0); IF_AFDATA_LOCK(ifp); ln = nd6_lookup(&dst->sin6_addr, flags, ifp); IF_AFDATA_UNLOCK(ifp); } } if (ln == NULL) { if ((ifp->if_flags & IFF_POINTOPOINT) == 0 && !(ND_IFINFO(ifp)->flags & ND6_IFF_PERFORMNUD)) { char ip6buf[INET6_ADDRSTRLEN]; log(LOG_DEBUG, "nd6_output: can't allocate llinfo for %s " "(ln=%p)\n", ip6_sprintf(ip6buf, &dst->sin6_addr), ln); senderr(EIO); /* XXX: good error? */ } goto sendpkt; /* send anyway */ } /* We don't have to do link-layer address resolution on a p2p link. */ if ((ifp->if_flags & IFF_POINTOPOINT) != 0 && ln->ln_state < ND6_LLINFO_REACHABLE) { if ((flags & LLE_EXCLUSIVE) == 0) { flags |= LLE_EXCLUSIVE; goto retry; } ln->ln_state = ND6_LLINFO_STALE; nd6_llinfo_settimer_locked(ln, (long)V_nd6_gctimer * hz); } /* * The first time we send a packet to a neighbor whose entry is * STALE, we have to change the state to DELAY and a sets a timer to * expire in DELAY_FIRST_PROBE_TIME seconds to ensure do * neighbor unreachability detection on expiration. * (RFC 2461 7.3.3) */ if (ln->ln_state == ND6_LLINFO_STALE) { if ((flags & LLE_EXCLUSIVE) == 0) { flags |= LLE_EXCLUSIVE; LLE_RUNLOCK(ln); goto retry; } ln->la_asked = 0; ln->ln_state = ND6_LLINFO_DELAY; nd6_llinfo_settimer_locked(ln, (long)V_nd6_delay * hz); } /* * If the neighbor cache entry has a state other than INCOMPLETE * (i.e. its link-layer address is already resolved), just * send the packet. */ if (ln->ln_state > ND6_LLINFO_INCOMPLETE) goto sendpkt; /* * There is a neighbor cache entry, but no ethernet address * response yet. Append this latest packet to the end of the * packet queue in the mbuf, unless the number of the packet * does not exceed nd6_maxqueuelen. When it exceeds nd6_maxqueuelen, * the oldest packet in the queue will be removed. */ if (ln->ln_state == ND6_LLINFO_NOSTATE) ln->ln_state = ND6_LLINFO_INCOMPLETE; if ((flags & LLE_EXCLUSIVE) == 0) { flags |= LLE_EXCLUSIVE; LLE_RUNLOCK(ln); goto retry; } LLE_WLOCK_ASSERT(ln); if (ln->la_hold) { struct mbuf *m_hold; int i; i = 0; for (m_hold = ln->la_hold; m_hold; m_hold = m_hold->m_nextpkt) { i++; if (m_hold->m_nextpkt == NULL) { m_hold->m_nextpkt = m; break; } } while (i >= V_nd6_maxqueuelen) { m_hold = ln->la_hold; ln->la_hold = ln->la_hold->m_nextpkt; m_freem(m_hold); i--; } } else { ln->la_hold = m; } /* * If there has been no NS for the neighbor after entering the * INCOMPLETE state, send the first solicitation. */ if (!ND6_LLINFO_PERMANENT(ln) && ln->la_asked == 0) { ln->la_asked++; nd6_llinfo_settimer_locked(ln, (long)ND_IFINFO(ifp)->retrans * hz / 1000); LLE_WUNLOCK(ln); nd6_ns_output(ifp, NULL, &dst->sin6_addr, ln, 0); if (lle != NULL && ln == lle) LLE_WLOCK(lle); } else if (lle == NULL || ln != lle) { /* * We did the lookup (no lle arg) so we * need to do the unlock here. */ LLE_WUNLOCK(ln); } return (0); sendpkt: /* discard the packet if IPv6 operation is disabled on the interface */ if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)) { error = ENETDOWN; /* better error? */ goto bad; } /* * ln is valid and the caller did not pass in * an llentry */ if ((ln != NULL) && (lle == NULL)) { if (flags & LLE_EXCLUSIVE) LLE_WUNLOCK(ln); else LLE_RUNLOCK(ln); } #ifdef MAC mac_netinet6_nd6_send(ifp, m); #endif /* * If called from nd6_ns_output() (NS), nd6_na_output() (NA), * icmp6_redirect_output() (REDIRECT) or from rip6_output() (RS, RA * as handled by rtsol and rtadvd), mbufs will be tagged for SeND * to be diverted to user space. When re-injected into the kernel, * send_output() will directly dispatch them to the outgoing interface. */ if (send_sendso_input_hook != NULL) { mtag = m_tag_find(m, PACKET_TAG_ND_OUTGOING, NULL); if (mtag != NULL) { ip6 = mtod(m, struct ip6_hdr *); ip6len = sizeof(struct ip6_hdr) + ntohs(ip6->ip6_plen); /* Use the SEND socket */ error = send_sendso_input_hook(m, ifp, SND_OUT, ip6len); /* -1 == no app on SEND socket */ if (error == 0 || error != -1) return (error); } } /* * We were passed in a pointer to an lle with the lock held * this means that we can't call if_output as we will * recurse on the lle lock - so what we do is we create * a list of mbufs to send and transmit them in the caller * after the lock is dropped */ if (lle != NULL) { if (*chain == NULL) *chain = m; else { struct mbuf *m = *chain; /* * append mbuf to end of deferred chain */ while (m->m_nextpkt != NULL) m = m->m_nextpkt; m->m_nextpkt = m; } return (error); } if ((ifp->if_flags & IFF_LOOPBACK) != 0) { return ((*ifp->if_output)(origifp, m, (struct sockaddr *)dst, NULL)); } error = (*ifp->if_output)(ifp, m, (struct sockaddr *)dst, NULL); return (error); bad: /* * ln is valid and the caller did not pass in * an llentry */ if ((ln != NULL) && (lle == NULL)) { if (flags & LLE_EXCLUSIVE) LLE_WUNLOCK(ln); else LLE_RUNLOCK(ln); } if (m) m_freem(m); return (error); } #undef senderr int nd6_output_flush(struct ifnet *ifp, struct ifnet *origifp, struct mbuf *chain, struct sockaddr_in6 *dst, struct route *ro) { struct mbuf *m, *m_head; struct ifnet *outifp; int error = 0; m_head = chain; if ((ifp->if_flags & IFF_LOOPBACK) != 0) outifp = origifp; else outifp = ifp; while (m_head) { m = m_head; m_head = m_head->m_nextpkt; error = (*ifp->if_output)(ifp, m, (struct sockaddr *)dst, ro); } /* * XXX * note that intermediate errors are blindly ignored - but this is * the same convention as used with nd6_output when called by * nd6_cache_lladdr */ return (error); } int nd6_need_cache(struct ifnet *ifp) { /* * XXX: we currently do not make neighbor cache on any interface * other than ARCnet, Ethernet, FDDI and GIF. * * RFC2893 says: * - unidirectional tunnels needs no ND */ 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 #ifdef IFT_CARP case IFT_CARP: #endif case IFT_INFINIBAND: case IFT_GIF: /* XXX need more cases? */ case IFT_PPP: case IFT_TUNNEL: case IFT_BRIDGE: case IFT_PROPVIRTUAL: return (1); default: return (0); } } /* * the callers of this function need to be re-worked to drop * the lle lock, drop here for now */ int nd6_storelladdr(struct ifnet *ifp, struct mbuf *m, struct sockaddr *dst, u_char *desten, struct llentry **lle) { struct llentry *ln; *lle = NULL; IF_AFDATA_UNLOCK_ASSERT(ifp); if (m->m_flags & M_MCAST) { int i; switch (ifp->if_type) { case IFT_ETHER: case IFT_FDDI: #ifdef IFT_L2VLAN case IFT_L2VLAN: #endif #ifdef IFT_IEEE80211 case IFT_IEEE80211: #endif case IFT_BRIDGE: case IFT_ISO88025: ETHER_MAP_IPV6_MULTICAST(&SIN6(dst)->sin6_addr, desten); return (0); case IFT_IEEE1394: /* * netbsd can use if_broadcastaddr, but we don't do so * to reduce # of ifdef. */ for (i = 0; i < ifp->if_addrlen; i++) desten[i] = ~0; return (0); case IFT_ARCNET: *desten = 0; return (0); default: m_freem(m); return (EAFNOSUPPORT); } } /* * the entry should have been created in nd6_store_lladdr */ IF_AFDATA_LOCK(ifp); ln = lla_lookup(LLTABLE6(ifp), 0, dst); IF_AFDATA_UNLOCK(ifp); if ((ln == NULL) || !(ln->la_flags & LLE_VALID)) { if (ln != NULL) LLE_RUNLOCK(ln); /* this could happen, if we could not allocate memory */ m_freem(m); return (1); } bcopy(&ln->ll_addr, desten, ifp->if_addrlen); *lle = ln; LLE_RUNLOCK(ln); /* * A *small* use after free race exists here */ return (0); } static void clear_llinfo_pqueue(struct llentry *ln) { struct mbuf *m_hold, *m_hold_next; for (m_hold = ln->la_hold; m_hold; m_hold = m_hold_next) { m_hold_next = m_hold->m_nextpkt; m_hold->m_nextpkt = NULL; m_freem(m_hold); } ln->la_hold = NULL; return; } static int nd6_sysctl_drlist(SYSCTL_HANDLER_ARGS); static int nd6_sysctl_prlist(SYSCTL_HANDLER_ARGS); #ifdef SYSCTL_DECL SYSCTL_DECL(_net_inet6_icmp6); #endif SYSCTL_NODE(_net_inet6_icmp6, ICMPV6CTL_ND6_DRLIST, nd6_drlist, CTLFLAG_RD, nd6_sysctl_drlist, ""); SYSCTL_NODE(_net_inet6_icmp6, ICMPV6CTL_ND6_PRLIST, nd6_prlist, CTLFLAG_RD, nd6_sysctl_prlist, ""); SYSCTL_VNET_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_MAXQLEN, nd6_maxqueuelen, CTLFLAG_RW, &VNET_NAME(nd6_maxqueuelen), 1, ""); static int nd6_sysctl_drlist(SYSCTL_HANDLER_ARGS) { int error; char buf[1024] __aligned(4); struct in6_defrouter *d, *de; struct nd_defrouter *dr; if (req->newptr) return EPERM; error = 0; for (dr = TAILQ_FIRST(&V_nd_defrouter); dr; dr = TAILQ_NEXT(dr, dr_entry)) { d = (struct in6_defrouter *)buf; de = (struct in6_defrouter *)(buf + sizeof(buf)); if (d + 1 <= de) { bzero(d, sizeof(*d)); d->rtaddr.sin6_family = AF_INET6; d->rtaddr.sin6_len = sizeof(d->rtaddr); d->rtaddr.sin6_addr = dr->rtaddr; error = sa6_recoverscope(&d->rtaddr); if (error != 0) return (error); d->flags = dr->flags; d->rtlifetime = dr->rtlifetime; d->expire = dr->expire; d->if_index = dr->ifp->if_index; } else panic("buffer too short"); error = SYSCTL_OUT(req, buf, sizeof(*d)); if (error) break; } return (error); } static int nd6_sysctl_prlist(SYSCTL_HANDLER_ARGS) { int error; char buf[1024] __aligned(4); struct in6_prefix *p, *pe; struct nd_prefix *pr; char ip6buf[INET6_ADDRSTRLEN]; if (req->newptr) return EPERM; error = 0; for (pr = V_nd_prefix.lh_first; pr; pr = pr->ndpr_next) { u_short advrtrs; size_t advance; struct sockaddr_in6 *sin6, *s6; struct nd_pfxrouter *pfr; p = (struct in6_prefix *)buf; pe = (struct in6_prefix *)(buf + sizeof(buf)); if (p + 1 <= pe) { bzero(p, sizeof(*p)); sin6 = (struct sockaddr_in6 *)(p + 1); p->prefix = pr->ndpr_prefix; if (sa6_recoverscope(&p->prefix)) { log(LOG_ERR, "scope error in prefix list (%s)\n", ip6_sprintf(ip6buf, &p->prefix.sin6_addr)); /* XXX: press on... */ } p->raflags = pr->ndpr_raf; p->prefixlen = pr->ndpr_plen; p->vltime = pr->ndpr_vltime; p->pltime = pr->ndpr_pltime; p->if_index = pr->ndpr_ifp->if_index; if (pr->ndpr_vltime == ND6_INFINITE_LIFETIME) p->expire = 0; else { time_t maxexpire; /* XXX: we assume time_t is signed. */ maxexpire = (-1) & ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1)); if (pr->ndpr_vltime < maxexpire - pr->ndpr_lastupdate) { p->expire = pr->ndpr_lastupdate + pr->ndpr_vltime; } else p->expire = maxexpire; } p->refcnt = pr->ndpr_refcnt; p->flags = pr->ndpr_stateflags; p->origin = PR_ORIG_RA; advrtrs = 0; for (pfr = pr->ndpr_advrtrs.lh_first; pfr; pfr = pfr->pfr_next) { if ((void *)&sin6[advrtrs + 1] > (void *)pe) { advrtrs++; continue; } s6 = &sin6[advrtrs]; bzero(s6, sizeof(*s6)); s6->sin6_family = AF_INET6; s6->sin6_len = sizeof(*sin6); s6->sin6_addr = pfr->router->rtaddr; if (sa6_recoverscope(s6)) { log(LOG_ERR, "scope error in " "prefix list (%s)\n", ip6_sprintf(ip6buf, &pfr->router->rtaddr)); } advrtrs++; } p->advrtrs = advrtrs; } else panic("buffer too short"); advance = sizeof(*p) + sizeof(*sin6) * advrtrs; error = SYSCTL_OUT(req, buf, advance); if (error) break; } return (error); } Index: projects/largeSMP/sys/netinet6/nd6.h =================================================================== --- projects/largeSMP/sys/netinet6/nd6.h (revision 222782) +++ projects/largeSMP/sys/netinet6/nd6.h (revision 222783) @@ -1,456 +1,457 @@ /*- * 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 /* XXX: not related to ND. */ #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_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_next ndpr_entry.le_next #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; #define pfr_next pfr_entry.le_next 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[8]; /* max = target address list */ 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 *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_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 __P((void)); #ifdef VIMAGE void nd6_destroy __P((void)); #endif struct nd_ifinfo *nd6_ifattach __P((struct ifnet *)); void nd6_ifdetach __P((struct nd_ifinfo *)); int nd6_is_addr_neighbor __P((struct sockaddr_in6 *, struct ifnet *)); void nd6_option_init __P((void *, int, union nd_opts *)); struct nd_opt_hdr *nd6_option __P((union nd_opts *)); int nd6_options __P((union nd_opts *)); struct llentry *nd6_lookup __P((struct in6_addr *, int, struct ifnet *)); void nd6_setmtu __P((struct ifnet *)); void nd6_llinfo_settimer __P((struct llentry *, long)); void nd6_llinfo_settimer_locked __P((struct llentry *, long)); void nd6_timer __P((void *)); void nd6_purge __P((struct ifnet *)); void nd6_nud_hint __P((struct rtentry *, struct in6_addr *, int)); int nd6_resolve __P((struct ifnet *, struct rtentry *, struct mbuf *, struct sockaddr *, u_char *)); int nd6_ioctl __P((u_long, caddr_t, struct ifnet *)); struct llentry *nd6_cache_lladdr __P((struct ifnet *, struct in6_addr *, char *, int, int, int)); int nd6_output __P((struct ifnet *, struct ifnet *, struct mbuf *, struct sockaddr_in6 *, struct rtentry *)); int nd6_output_lle __P((struct ifnet *, struct ifnet *, struct mbuf *, struct sockaddr_in6 *, struct rtentry *, struct llentry *, struct mbuf **)); int nd6_output_flush __P((struct ifnet *, struct ifnet *, struct mbuf *, struct sockaddr_in6 *, struct route *)); int nd6_need_cache __P((struct ifnet *)); int nd6_storelladdr __P((struct ifnet *, struct mbuf *, struct sockaddr *, u_char *, struct llentry **)); /* nd6_nbr.c */ void nd6_na_input __P((struct mbuf *, int, int)); void nd6_na_output __P((struct ifnet *, const struct in6_addr *, const struct in6_addr *, u_long, int, struct sockaddr *)); void nd6_ns_input __P((struct mbuf *, int, int)); void nd6_ns_output __P((struct ifnet *, const struct in6_addr *, const struct in6_addr *, struct llentry *, int)); caddr_t nd6_ifptomac __P((struct ifnet *)); void nd6_dad_start __P((struct ifaddr *, int)); void nd6_dad_stop __P((struct ifaddr *)); void nd6_dad_duplicated __P((struct ifaddr *)); /* nd6_rtr.c */ void nd6_rs_input __P((struct mbuf *, int, int)); void nd6_ra_input __P((struct mbuf *, int, int)); void prelist_del __P((struct nd_prefix *)); void defrouter_addreq __P((struct nd_defrouter *)); void defrouter_reset __P((void)); void defrouter_select __P((void)); void defrtrlist_del __P((struct nd_defrouter *)); void prelist_remove __P((struct nd_prefix *)); int nd6_prelist_add __P((struct nd_prefixctl *, struct nd_defrouter *, struct nd_prefix **)); int nd6_prefix_onlink __P((struct nd_prefix *)); int nd6_prefix_offlink __P((struct nd_prefix *)); void pfxlist_onlink_check __P((void)); struct nd_defrouter *defrouter_lookup __P((struct in6_addr *, struct ifnet *)); struct nd_prefix *nd6_prefix_lookup __P((struct nd_prefixctl *)); void rt6_flush __P((struct in6_addr *, struct ifnet *)); int nd6_setdefaultiface __P((int)); int in6_tmpifadd __P((const struct in6_ifaddr *, int, int)); #endif /* _KERNEL */ #endif /* _NETINET6_ND6_H_ */ Index: projects/largeSMP/sys/netinet6/nd6_nbr.c =================================================================== --- projects/largeSMP/sys/netinet6/nd6_nbr.c (revision 222782) +++ projects/largeSMP/sys/netinet6/nd6_nbr.c (revision 222783) @@ -1,1554 +1,1558 @@ /*- * 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 #ifdef RADIX_MPATH #include #endif #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 *); 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_ns_output(struct dadq *, struct ifaddr *); static void nd6_dad_ns_input(struct ifaddr *); static void nd6_dad_na_input(struct ifaddr *); VNET_DEFINE(int, dad_ignore_ns) = 0; /* ignore NS in DAD - specwise incorrect*/ VNET_DEFINE(int, dad_maxtry) = 15; /* max # of *tries* to transmit DAD packet */ #define V_dad_ignore_ns VNET(dad_ignore_ns) #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); if (ifa == NULL) ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp, &taddr6); /* (2) check. */ if (ifa == NULL) { struct rtentry *rt; struct sockaddr_in6 tsin6; int need_proxy; #ifdef RADIX_MPATH struct route_in6 ro; #endif bzero(&tsin6, sizeof tsin6); tsin6.sin6_len = sizeof(struct sockaddr_in6); tsin6.sin6_family = AF_INET6; tsin6.sin6_addr = taddr6; #ifdef RADIX_MPATH bzero(&ro, sizeof(ro)); ro.ro_dst = tsin6; rtalloc_mpath((struct route *)&ro, RTF_ANNOUNCE); rt = ro.ro_rt; #else rt = rtalloc1((struct sockaddr *)&tsin6, 0, 0); #endif need_proxy = (rt && (rt->rt_flags & RTF_ANNOUNCE) != 0 && rt->rt_gateway->sa_family == AF_LINK); if (rt != NULL) { /* * Make a copy while we can be sure that rt_gateway * is still stable before unlocking to avoid lock * order problems. proxydl will only be used if * proxy will be set in the next block. */ if (need_proxy) proxydl = *SDL(rt->rt_gateway); RTFREE_LOCKED(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); 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(ifp, &in6_all, &taddr6, ((anycast || proxy || !tlladdr) ? 0 : ND_NA_FLAG_OVERRIDE) | - (V_ip6_forwarding ? ND_NA_FLAG_ROUTER : 0), - tlladdr, proxy ? (struct sockaddr *)&proxydl : NULL); + rflag, tlladdr, proxy ? (struct sockaddr *)&proxydl : NULL); goto freeit; } nd6_cache_lladdr(ifp, &saddr6, lladdr, lladdrlen, ND_NEIGHBOR_SOLICIT, 0); nd6_na_output(ifp, &saddr6, &taddr6, ((anycast || proxy || !tlladdr) ? 0 : ND_NA_FLAG_OVERRIDE) | - (V_ip6_forwarding ? ND_NA_FLAG_ROUTER : 0) | ND_NA_FLAG_SOLICITED, - tlladdr, proxy ? (struct sockaddr *)&proxydl : NULL); + rflag | ND_NA_FLAG_SOLICITED, tlladdr, + proxy ? (struct sockaddr *)&proxydl : NULL); 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 * dad - duplicate address detection */ void nd6_ns_output(struct ifnet *ifp, const struct in6_addr *daddr6, const struct in6_addr *taddr6, struct llentry *ln, int dad) { 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); #endif return; } MGETHDR(m, M_DONTWAIT, MT_DATA); if (m && max_linkhdr + maxlen >= MHLEN) { MCLGET(m, M_DONTWAIT); if ((m->m_flags & M_EXT) == 0) { m_free(m); m = NULL; } } if (m == NULL) return; m->m_pkthdr.rcvif = NULL; 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 MH_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 (!dad) { 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; 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_selectsrc(&dst_sa, NULL, NULL, &ro, NULL, NULL, &src_in); if (error) { char ip6buf[INET6_ADDRSTRLEN]; nd6log((LOG_DEBUG, "nd6_ns_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_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 (!dad && (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); } 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, dad ? 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]); if (ro.ro_rt) { /* we don't cache this route. */ RTFREE(ro.ro_rt); } return; bad: if (ro.ro_rt) { RTFREE(ro.ro_rt); } m_freem(m); return; } /* * 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 m_tag *mtag; 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); 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; } 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_LOCK(ifp); ln = nd6_lookup(&taddr6, LLE_EXCLUSIVE, ifp); IF_AFDATA_UNLOCK(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; 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; } /* * 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 (!V_ip6_forwarding) { + 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) { struct mbuf *m_hold, *m_hold_next; /* * reset the la_hold in advance, to explicitly * prevent a la_hold lookup in nd6_output() * (wouldn't happen, though...) */ for (m_hold = ln->la_hold, ln->la_hold = NULL; m_hold; m_hold = m_hold_next) { m_hold_next = m_hold->m_nextpkt; m_hold->m_nextpkt = NULL; /* * we assume ifp is not a loopback here, so just set * the 2nd argument as the 1st one. */ if (send_sendso_input_hook != NULL) { mtag = m_tag_get(PACKET_TAG_ND_OUTGOING, sizeof(unsigned short), M_NOWAIT); if (mtag == NULL) goto bad; m_tag_prepend(m, mtag); } nd6_output_lle(ifp, ifp, m_hold, L3_ADDR_SIN6(ln), NULL, ln, &chain); } } freeit: if (ln != NULL) { if (chain) memcpy(&sin6, L3_ADDR_SIN6(ln), sizeof(sin6)); LLE_WUNLOCK(ln); if (chain) nd6_output_flush(ifp, ifp, chain, &sin6, NULL); } 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 */ 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) { struct mbuf *m; struct m_tag *mtag; 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; } MGETHDR(m, M_DONTWAIT, MT_DATA); if (m && max_linkhdr + maxlen >= MHLEN) { MCLGET(m, M_DONTWAIT); if ((m->m_flags & M_EXT) == 0) { m_free(m); m = NULL; } } if (m == NULL) return; m->m_pkthdr.rcvif = NULL; 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 MH_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)); error = in6_selectsrc(&dst_sa, NULL, NULL, &ro, NULL, NULL, &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]); if (ro.ro_rt) { /* we don't cache this route. */ RTFREE(ro.ro_rt); } return; bad: if (ro.ro_rt) { RTFREE(ro.ro_rt); } m_freem(m); return; } 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 #ifdef IFT_CARP case IFT_CARP: #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; struct callout dad_timer_ch; struct vnet *dad_vnet; }; static VNET_DEFINE(TAILQ_HEAD(, dadq), dadq); VNET_DEFINE(int, dad_init) = 0; #define V_dadq VNET(dadq) #define V_dad_init VNET(dad_init) static struct dadq * nd6_dad_find(struct ifaddr *ifa) { struct dadq *dp; for (dp = V_dadq.tqh_first; dp; dp = dp->dad_list.tqe_next) { if (dp->dad_ifa == ifa) return dp; } return NULL; } 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_stop(&dp->dad_timer_ch); } /* * 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 (!V_dad_init) { TAILQ_INIT(&V_dadq); V_dad_init++; } /* * 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 (nd6_dad_find(ifa) != NULL) { /* DAD already in progress */ return; } dp = malloc(sizeof(*dp), M_IP6NDP, M_NOWAIT); 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; } bzero(dp, sizeof(*dp)); callout_init(&dp->dad_timer_ch, 0); #ifdef VIMAGE dp->dad_vnet = curvnet; #endif TAILQ_INSERT_TAIL(&V_dadq, (struct dadq *)dp, dad_list); 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(ifa); /* just for safety */ 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; 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; if (!V_dad_init) return; dp = nd6_dad_find(ifa); if (!dp) { /* DAD wasn't started yet */ return; } nd6_dad_stoptimer(dp); TAILQ_REMOVE(&V_dadq, (struct dadq *)dp, dad_list); free(dp, M_IP6NDP); dp = NULL; ifa_free(ifa); } static void nd6_dad_timer(struct dadq *dp) { CURVNET_SET(dp->dad_vnet); int s; struct ifaddr *ifa = dp->dad_ifa; struct in6_ifaddr *ia = (struct in6_ifaddr *)ifa; char ip6buf[INET6_ADDRSTRLEN]; s = splnet(); /* XXX */ /* Sanity check */ if (ia == NULL) { log(LOG_ERR, "nd6_dad_timer: called with null parameter\n"); goto done; } 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 done; } 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 done; } /* timeouted with IFF_{RUNNING,UP} check */ if (dp->dad_ns_tcount > V_dad_maxtry) { nd6log((LOG_INFO, "%s: could not run DAD, driver problem?\n", if_name(ifa->ifa_ifp))); TAILQ_REMOVE(&V_dadq, (struct dadq *)dp, dad_list); free(dp, M_IP6NDP); dp = NULL; ifa_free(ifa); goto done; } /* 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); } else { /* * We have transmitted sufficient number of DAD packets. * See what we've got. */ int duplicate; duplicate = 0; if (dp->dad_na_icount) { /* * the check is in nd6_dad_na_input(), * but just in case */ duplicate++; } if (dp->dad_ns_icount) { /* We've seen NS, means DAD has failed. */ duplicate++; } if (duplicate) { /* (*dp) will be freed in nd6_dad_duplicated() */ dp = NULL; nd6_dad_duplicated(ifa); } else { /* * We are done with DAD. No NA came, no NS came. * No duplicate address found. */ 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))); TAILQ_REMOVE(&V_dadq, (struct dadq *)dp, dad_list); free(dp, M_IP6NDP); dp = NULL; ifa_free(ifa); } } done: splx(s); CURVNET_RESTORE(); } void nd6_dad_duplicated(struct ifaddr *ifa) { struct in6_ifaddr *ia = (struct in6_ifaddr *)ifa; struct ifnet *ifp; struct dadq *dp; char ip6buf[INET6_ADDRSTRLEN]; dp = nd6_dad_find(ifa); if (dp == NULL) { log(LOG_ERR, "nd6_dad_duplicated: DAD structure not found\n"); return; } log(LOG_ERR, "%s: DAD detected duplicate IPv6 address %s: " "NS in/out=%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_na_icount); ia->ia6_flags &= ~IN6_IFF_TENTATIVE; ia->ia6_flags |= IN6_IFF_DUPLICATED; /* We are done with DAD, with duplicate address found. (failure) */ nd6_dad_stoptimer(dp); 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; } } TAILQ_REMOVE(&V_dadq, (struct dadq *)dp, dad_list); free(dp, M_IP6NDP); dp = NULL; ifa_free(ifa); } 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; 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++; nd6_ns_output(ifp, NULL, &ia->ia_addr.sin6_addr, NULL, 1); } static void nd6_dad_ns_input(struct ifaddr *ifa) { struct in6_ifaddr *ia; struct ifnet *ifp; const struct in6_addr *taddr6; struct dadq *dp; int duplicate; 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; duplicate = 0; dp = nd6_dad_find(ifa); /* Quickhack - completely ignore DAD NS packets */ if (V_dad_ignore_ns) { char ip6buf[INET6_ADDRSTRLEN]; nd6log((LOG_INFO, "nd6_dad_ns_input: ignoring DAD NS packet for " "address %s(%s)\n", ip6_sprintf(ip6buf, taddr6), if_name(ifa->ifa_ifp))); return; } /* * if I'm yet to start DAD, someone else started using this address * first. I have a duplicate and you win. */ if (dp == NULL || dp->dad_ns_ocount == 0) duplicate++; /* XXX more checks for loopback situation - see nd6_dad_timer too */ if (duplicate) { dp = NULL; /* will be freed in nd6_dad_duplicated() */ nd6_dad_duplicated(ifa); } else { /* * not sure if I got a duplicate. * increment ns count and see what happens. */ if (dp) dp->dad_ns_icount++; } } 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); if (dp) dp->dad_na_icount++; /* remove the address. */ nd6_dad_duplicated(ifa); } Index: projects/largeSMP/sys/netinet6/nd6_rtr.c =================================================================== --- projects/largeSMP/sys/netinet6/nd6_rtr.c (revision 222782) +++ projects/largeSMP/sys/netinet6/nd6_rtr.c (revision 222783) @@ -1,2158 +1,2154 @@ /*- * 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_rtr.c,v 1.111 2001/04/27 01:37:15 jinmei Exp $ */ #include __FBSDID("$FreeBSD$"); #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 static int rtpref(struct nd_defrouter *); static struct nd_defrouter *defrtrlist_update(struct nd_defrouter *); static int prelist_update __P((struct nd_prefixctl *, struct nd_defrouter *, struct mbuf *, int)); static struct in6_ifaddr *in6_ifadd(struct nd_prefixctl *, int); static struct nd_pfxrouter *pfxrtr_lookup __P((struct nd_prefix *, struct nd_defrouter *)); static void pfxrtr_add(struct nd_prefix *, struct nd_defrouter *); static void pfxrtr_del(struct nd_pfxrouter *); static struct nd_pfxrouter *find_pfxlist_reachable_router (struct nd_prefix *); static void defrouter_delreq(struct nd_defrouter *); static void nd6_rtmsg(int, struct rtentry *); static int in6_init_prefix_ltimes(struct nd_prefix *); static void in6_init_address_ltimes __P((struct nd_prefix *, struct in6_addrlifetime *)); static int rt6_deleteroute(struct radix_node *, void *); VNET_DECLARE(int, nd6_recalc_reachtm_interval); #define V_nd6_recalc_reachtm_interval VNET(nd6_recalc_reachtm_interval) static VNET_DEFINE(struct ifnet *, nd6_defifp); VNET_DEFINE(int, nd6_defifindex); #define V_nd6_defifp VNET(nd6_defifp) VNET_DEFINE(int, ip6_use_tempaddr) = 0; VNET_DEFINE(int, ip6_desync_factor); VNET_DEFINE(u_int32_t, ip6_temp_preferred_lifetime) = DEF_TEMP_PREFERRED_LIFETIME; VNET_DEFINE(u_int32_t, ip6_temp_valid_lifetime) = DEF_TEMP_VALID_LIFETIME; VNET_DEFINE(int, ip6_temp_regen_advance) = TEMPADDR_REGEN_ADVANCE; /* RTPREF_MEDIUM has to be 0! */ #define RTPREF_HIGH 1 #define RTPREF_MEDIUM 0 #define RTPREF_LOW (-1) #define RTPREF_RESERVED (-2) #define RTPREF_INVALID (-3) /* internal */ /* * Receive Router Solicitation Message - just for routers. * Router solicitation/advertisement is mostly managed by userland program * (rtadvd) so here we have no function like nd6_ra_output(). * * Based on RFC 2461 */ void nd6_rs_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_router_solicit *nd_rs; struct in6_addr saddr6 = ip6->ip6_src; char *lladdr = NULL; int lladdrlen = 0; union nd_opts ndopts; char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; - /* If I'm not a router, ignore it. */ - if (!V_ip6_forwarding) + /* + * Accept RS only when V_ip6_forwarding=1 and the interface has + * no ND6_IFF_ACCEPT_RTADV. + */ + if (!V_ip6_forwarding || ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV) goto freeit; /* Sanity checks */ if (ip6->ip6_hlim != 255) { nd6log((LOG_ERR, "nd6_rs_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; } /* * Don't update the neighbor cache, if src = ::. * This indicates that the src has no IP address assigned yet. */ if (IN6_IS_ADDR_UNSPECIFIED(&saddr6)) goto freeit; #ifndef PULLDOWN_TEST IP6_EXTHDR_CHECK(m, off, icmp6len,); nd_rs = (struct nd_router_solicit *)((caddr_t)ip6 + off); #else IP6_EXTHDR_GET(nd_rs, struct nd_router_solicit *, m, off, icmp6len); if (nd_rs == NULL) { ICMP6STAT_INC(icp6s_tooshort); return; } #endif icmp6len -= sizeof(*nd_rs); nd6_option_init(nd_rs + 1, icmp6len, &ndopts); if (nd6_options(&ndopts) < 0) { nd6log((LOG_INFO, "nd6_rs_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 (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) { nd6log((LOG_INFO, "nd6_rs_input: lladdrlen mismatch for %s " "(if %d, RS packet %d)\n", ip6_sprintf(ip6bufs, &saddr6), ifp->if_addrlen, lladdrlen - 2)); goto bad; } nd6_cache_lladdr(ifp, &saddr6, lladdr, lladdrlen, ND_ROUTER_SOLICIT, 0); freeit: m_freem(m); return; bad: ICMP6STAT_INC(icp6s_badrs); m_freem(m); } /* * Receive Router Advertisement Message. * * Based on RFC 2461 * TODO: on-link bit on prefix information * TODO: ND_RA_FLAG_{OTHER,MANAGED} processing */ void nd6_ra_input(struct mbuf *m, int off, int icmp6len) { struct ifnet *ifp = m->m_pkthdr.rcvif; struct nd_ifinfo *ndi = ND_IFINFO(ifp); struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); struct nd_router_advert *nd_ra; struct in6_addr saddr6 = ip6->ip6_src; int mcast = 0; union nd_opts ndopts; struct nd_defrouter *dr; char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; /* - * We only accept RAs only when - * the node is not a router and - * per-interface variable allows RAs on the receiving interface. + * We only accept RAs only when the per-interface flag + * ND6_IFF_ACCEPT_RTADV is on the receiving interface. */ - if (V_ip6_forwarding || !(ndi->flags & ND6_IFF_ACCEPT_RTADV)) + if (!(ndi->flags & ND6_IFF_ACCEPT_RTADV)) goto freeit; if (ip6->ip6_hlim != 255) { nd6log((LOG_ERR, "nd6_ra_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_LINKLOCAL(&saddr6)) { nd6log((LOG_ERR, "nd6_ra_input: src %s is not link-local\n", ip6_sprintf(ip6bufs, &saddr6))); goto bad; } #ifndef PULLDOWN_TEST IP6_EXTHDR_CHECK(m, off, icmp6len,); nd_ra = (struct nd_router_advert *)((caddr_t)ip6 + off); #else IP6_EXTHDR_GET(nd_ra, struct nd_router_advert *, m, off, icmp6len); if (nd_ra == NULL) { ICMP6STAT_INC(icp6s_tooshort); return; } #endif icmp6len -= sizeof(*nd_ra); nd6_option_init(nd_ra + 1, icmp6len, &ndopts); if (nd6_options(&ndopts) < 0) { nd6log((LOG_INFO, "nd6_ra_input: invalid ND option, ignored\n")); /* nd6_options have incremented stats */ goto freeit; } { struct nd_defrouter dr0; u_int32_t advreachable = nd_ra->nd_ra_reachable; /* remember if this is a multicasted advertisement */ if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) mcast = 1; bzero(&dr0, sizeof(dr0)); dr0.rtaddr = saddr6; dr0.flags = nd_ra->nd_ra_flags_reserved; - dr0.rtlifetime = ntohs(nd_ra->nd_ra_router_lifetime); + /* + * Effectively-disable the route in the RA packet + * when ND6_IFF_NO_RADR on the receiving interface or + * ip6.forwarding=1. + */ + if (ndi->flags & ND6_IFF_NO_RADR || V_ip6_forwarding) + dr0.rtlifetime = 0; + else + dr0.rtlifetime = ntohs(nd_ra->nd_ra_router_lifetime); dr0.expire = time_second + dr0.rtlifetime; dr0.ifp = ifp; /* unspecified or not? (RFC 2461 6.3.4) */ if (advreachable) { advreachable = ntohl(advreachable); if (advreachable <= MAX_REACHABLE_TIME && ndi->basereachable != advreachable) { ndi->basereachable = advreachable; ndi->reachable = ND_COMPUTE_RTIME(ndi->basereachable); ndi->recalctm = V_nd6_recalc_reachtm_interval; /* reset */ } } if (nd_ra->nd_ra_retransmit) ndi->retrans = ntohl(nd_ra->nd_ra_retransmit); if (nd_ra->nd_ra_curhoplimit) ndi->chlim = nd_ra->nd_ra_curhoplimit; dr = defrtrlist_update(&dr0); } /* * prefix */ if (ndopts.nd_opts_pi) { struct nd_opt_hdr *pt; struct nd_opt_prefix_info *pi = NULL; struct nd_prefixctl pr; for (pt = (struct nd_opt_hdr *)ndopts.nd_opts_pi; pt <= (struct nd_opt_hdr *)ndopts.nd_opts_pi_end; pt = (struct nd_opt_hdr *)((caddr_t)pt + (pt->nd_opt_len << 3))) { if (pt->nd_opt_type != ND_OPT_PREFIX_INFORMATION) continue; pi = (struct nd_opt_prefix_info *)pt; if (pi->nd_opt_pi_len != 4) { nd6log((LOG_INFO, "nd6_ra_input: invalid option " "len %d for prefix information option, " "ignored\n", pi->nd_opt_pi_len)); continue; } if (128 < pi->nd_opt_pi_prefix_len) { nd6log((LOG_INFO, "nd6_ra_input: invalid prefix " "len %d for prefix information option, " "ignored\n", pi->nd_opt_pi_prefix_len)); continue; } if (IN6_IS_ADDR_MULTICAST(&pi->nd_opt_pi_prefix) || IN6_IS_ADDR_LINKLOCAL(&pi->nd_opt_pi_prefix)) { nd6log((LOG_INFO, "nd6_ra_input: invalid prefix " "%s, ignored\n", ip6_sprintf(ip6bufs, &pi->nd_opt_pi_prefix))); continue; } bzero(&pr, sizeof(pr)); pr.ndpr_prefix.sin6_family = AF_INET6; pr.ndpr_prefix.sin6_len = sizeof(pr.ndpr_prefix); pr.ndpr_prefix.sin6_addr = pi->nd_opt_pi_prefix; pr.ndpr_ifp = (struct ifnet *)m->m_pkthdr.rcvif; pr.ndpr_raf_onlink = (pi->nd_opt_pi_flags_reserved & ND_OPT_PI_FLAG_ONLINK) ? 1 : 0; pr.ndpr_raf_auto = (pi->nd_opt_pi_flags_reserved & ND_OPT_PI_FLAG_AUTO) ? 1 : 0; pr.ndpr_plen = pi->nd_opt_pi_prefix_len; pr.ndpr_vltime = ntohl(pi->nd_opt_pi_valid_time); pr.ndpr_pltime = ntohl(pi->nd_opt_pi_preferred_time); (void)prelist_update(&pr, dr, m, mcast); } } /* * MTU */ if (ndopts.nd_opts_mtu && ndopts.nd_opts_mtu->nd_opt_mtu_len == 1) { u_long mtu; u_long maxmtu; mtu = (u_long)ntohl(ndopts.nd_opts_mtu->nd_opt_mtu_mtu); /* lower bound */ if (mtu < IPV6_MMTU) { nd6log((LOG_INFO, "nd6_ra_input: bogus mtu option " "mtu=%lu sent from %s, ignoring\n", mtu, ip6_sprintf(ip6bufs, &ip6->ip6_src))); goto skip; } /* upper bound */ maxmtu = (ndi->maxmtu && ndi->maxmtu < ifp->if_mtu) ? ndi->maxmtu : ifp->if_mtu; if (mtu <= maxmtu) { int change = (ndi->linkmtu != mtu); ndi->linkmtu = mtu; if (change) /* in6_maxmtu may change */ in6_setmaxmtu(); } else { nd6log((LOG_INFO, "nd6_ra_input: bogus mtu " "mtu=%lu sent from %s; " "exceeds maxmtu %lu, ignoring\n", mtu, ip6_sprintf(ip6bufs, &ip6->ip6_src), maxmtu)); } } skip: /* * Source link layer address */ { char *lladdr = NULL; int lladdrlen = 0; 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 (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) { nd6log((LOG_INFO, "nd6_ra_input: lladdrlen mismatch for %s " "(if %d, RA packet %d)\n", ip6_sprintf(ip6bufs, &saddr6), ifp->if_addrlen, lladdrlen - 2)); goto bad; } nd6_cache_lladdr(ifp, &saddr6, lladdr, lladdrlen, ND_ROUTER_ADVERT, 0); /* * Installing a link-layer address might change the state of the * router's neighbor cache, which might also affect our on-link * detection of adveritsed prefixes. */ pfxlist_onlink_check(); } freeit: m_freem(m); return; bad: ICMP6STAT_INC(icp6s_badra); m_freem(m); } /* * default router list proccessing sub routines */ /* tell the change to user processes watching the routing socket. */ static void nd6_rtmsg(int cmd, struct rtentry *rt) { struct rt_addrinfo info; struct ifnet *ifp; struct ifaddr *ifa; bzero((caddr_t)&info, sizeof(info)); info.rti_info[RTAX_DST] = rt_key(rt); info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; info.rti_info[RTAX_NETMASK] = rt_mask(rt); ifp = rt->rt_ifp; if (ifp != NULL) { IF_ADDR_LOCK(ifp); ifa = TAILQ_FIRST(&ifp->if_addrhead); info.rti_info[RTAX_IFP] = ifa->ifa_addr; ifa_ref(ifa); IF_ADDR_UNLOCK(ifp); info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr; } else ifa = NULL; rt_missmsg(cmd, &info, rt->rt_flags, 0); if (ifa != NULL) ifa_free(ifa); } void defrouter_addreq(struct nd_defrouter *new) { struct sockaddr_in6 def, mask, gate; struct rtentry *newrt = NULL; int s; int error; bzero(&def, sizeof(def)); bzero(&mask, sizeof(mask)); bzero(&gate, sizeof(gate)); def.sin6_len = mask.sin6_len = gate.sin6_len = sizeof(struct sockaddr_in6); def.sin6_family = gate.sin6_family = AF_INET6; gate.sin6_addr = new->rtaddr; s = splnet(); error = rtrequest(RTM_ADD, (struct sockaddr *)&def, (struct sockaddr *)&gate, (struct sockaddr *)&mask, RTF_GATEWAY, &newrt); if (newrt) { nd6_rtmsg(RTM_ADD, newrt); /* tell user process */ RTFREE(newrt); } if (error == 0) new->installed = 1; splx(s); return; } struct nd_defrouter * defrouter_lookup(struct in6_addr *addr, struct ifnet *ifp) { struct nd_defrouter *dr; for (dr = TAILQ_FIRST(&V_nd_defrouter); dr; dr = TAILQ_NEXT(dr, dr_entry)) { if (dr->ifp == ifp && IN6_ARE_ADDR_EQUAL(addr, &dr->rtaddr)) return (dr); } return (NULL); /* search failed */ } /* * Remove the default route for a given router. * This is just a subroutine function for defrouter_select(), and should * not be called from anywhere else. */ static void defrouter_delreq(struct nd_defrouter *dr) { struct sockaddr_in6 def, mask, gate; struct rtentry *oldrt = NULL; bzero(&def, sizeof(def)); bzero(&mask, sizeof(mask)); bzero(&gate, sizeof(gate)); def.sin6_len = mask.sin6_len = gate.sin6_len = sizeof(struct sockaddr_in6); def.sin6_family = gate.sin6_family = AF_INET6; gate.sin6_addr = dr->rtaddr; rtrequest(RTM_DELETE, (struct sockaddr *)&def, (struct sockaddr *)&gate, (struct sockaddr *)&mask, RTF_GATEWAY, &oldrt); if (oldrt) { nd6_rtmsg(RTM_DELETE, oldrt); RTFREE(oldrt); } dr->installed = 0; } /* * remove all default routes from default router list */ void defrouter_reset(void) { struct nd_defrouter *dr; for (dr = TAILQ_FIRST(&V_nd_defrouter); dr; dr = TAILQ_NEXT(dr, dr_entry)) defrouter_delreq(dr); /* * XXX should we also nuke any default routers in the kernel, by * going through them by rtalloc1()? */ } void defrtrlist_del(struct nd_defrouter *dr) { struct nd_defrouter *deldr = NULL; struct nd_prefix *pr; /* * Flush all the routing table entries that use the router * as a next hop. */ - if (!V_ip6_forwarding) + if (ND_IFINFO(dr->ifp)->flags & ND6_IFF_ACCEPT_RTADV) rt6_flush(&dr->rtaddr, dr->ifp); if (dr->installed) { deldr = dr; defrouter_delreq(dr); } TAILQ_REMOVE(&V_nd_defrouter, dr, dr_entry); /* * Also delete all the pointers to the router in each prefix lists. */ for (pr = V_nd_prefix.lh_first; pr; pr = pr->ndpr_next) { struct nd_pfxrouter *pfxrtr; if ((pfxrtr = pfxrtr_lookup(pr, dr)) != NULL) pfxrtr_del(pfxrtr); } pfxlist_onlink_check(); /* * If the router is the primary one, choose a new one. * Note that defrouter_select() will remove the current gateway * from the routing table. */ if (deldr) defrouter_select(); free(dr, M_IP6NDP); } /* * Default Router Selection according to Section 6.3.6 of RFC 2461 and * draft-ietf-ipngwg-router-selection: * 1) Routers that are reachable or probably reachable should be preferred. * If we have more than one (probably) reachable router, prefer ones * with the highest router preference. * 2) When no routers on the list are known to be reachable or * probably reachable, routers SHOULD be selected in a round-robin * fashion, regardless of router preference values. * 3) If the Default Router List is empty, assume that all * destinations are on-link. * * We assume nd_defrouter is sorted by router preference value. * Since the code below covers both with and without router preference cases, * we do not need to classify the cases by ifdef. * * At this moment, we do not try to install more than one default router, * even when the multipath routing is available, because we're not sure about * the benefits for stub hosts comparing to the risk of making the code * complicated and the possibility of introducing bugs. */ void defrouter_select(void) { int s = splnet(); struct nd_defrouter *dr, *selected_dr = NULL, *installed_dr = NULL; struct llentry *ln = NULL; - - /* - * This function should be called only when acting as an autoconfigured - * host. Although the remaining part of this function is not effective - * if the node is not an autoconfigured host, we explicitly exclude - * such cases here for safety. - */ - if (V_ip6_forwarding) { - nd6log((LOG_WARNING, - "defrouter_select: called unexpectedly (forwarding=%d)\n", - V_ip6_forwarding)); - splx(s); - return; - } /* * Let's handle easy case (3) first: * If default router list is empty, there's nothing to be done. */ if (!TAILQ_FIRST(&V_nd_defrouter)) { splx(s); return; } /* * Search for a (probably) reachable router from the list. * We just pick up the first reachable one (if any), assuming that * the ordering rule of the list described in defrtrlist_update(). */ for (dr = TAILQ_FIRST(&V_nd_defrouter); dr; dr = TAILQ_NEXT(dr, dr_entry)) { IF_AFDATA_LOCK(dr->ifp); if (selected_dr == NULL && (ln = nd6_lookup(&dr->rtaddr, 0, dr->ifp)) && ND6_IS_LLINFO_PROBREACH(ln)) { selected_dr = dr; } IF_AFDATA_UNLOCK(dr->ifp); if (ln != NULL) { LLE_RUNLOCK(ln); ln = NULL; } if (dr->installed && installed_dr == NULL) installed_dr = dr; else if (dr->installed && installed_dr) { /* this should not happen. warn for diagnosis. */ log(LOG_ERR, "defrouter_select: more than one router" " is installed\n"); } } /* * If none of the default routers was found to be reachable, * round-robin the list regardless of preference. * Otherwise, if we have an installed router, check if the selected * (reachable) router should really be preferred to the installed one. * We only prefer the new router when the old one is not reachable * or when the new one has a really higher preference value. */ if (selected_dr == NULL) { if (installed_dr == NULL || !TAILQ_NEXT(installed_dr, dr_entry)) selected_dr = TAILQ_FIRST(&V_nd_defrouter); else selected_dr = TAILQ_NEXT(installed_dr, dr_entry); } else if (installed_dr) { IF_AFDATA_LOCK(installed_dr->ifp); if ((ln = nd6_lookup(&installed_dr->rtaddr, 0, installed_dr->ifp)) && ND6_IS_LLINFO_PROBREACH(ln) && rtpref(selected_dr) <= rtpref(installed_dr)) { selected_dr = installed_dr; } IF_AFDATA_UNLOCK(installed_dr->ifp); if (ln != NULL) LLE_RUNLOCK(ln); } /* * If the selected router is different than the installed one, * remove the installed router and install the selected one. * Note that the selected router is never NULL here. */ if (installed_dr != selected_dr) { if (installed_dr) defrouter_delreq(installed_dr); defrouter_addreq(selected_dr); } splx(s); return; } /* * for default router selection * regards router-preference field as a 2-bit signed integer */ static int rtpref(struct nd_defrouter *dr) { switch (dr->flags & ND_RA_FLAG_RTPREF_MASK) { case ND_RA_FLAG_RTPREF_HIGH: return (RTPREF_HIGH); case ND_RA_FLAG_RTPREF_MEDIUM: case ND_RA_FLAG_RTPREF_RSV: return (RTPREF_MEDIUM); case ND_RA_FLAG_RTPREF_LOW: return (RTPREF_LOW); default: /* * This case should never happen. If it did, it would mean a * serious bug of kernel internal. We thus always bark here. * Or, can we even panic? */ log(LOG_ERR, "rtpref: impossible RA flag %x\n", dr->flags); return (RTPREF_INVALID); } /* NOTREACHED */ } static struct nd_defrouter * defrtrlist_update(struct nd_defrouter *new) { struct nd_defrouter *dr, *n; int s = splnet(); if ((dr = defrouter_lookup(&new->rtaddr, new->ifp)) != NULL) { /* entry exists */ if (new->rtlifetime == 0) { defrtrlist_del(dr); dr = NULL; } else { int oldpref = rtpref(dr); /* override */ dr->flags = new->flags; /* xxx flag check */ dr->rtlifetime = new->rtlifetime; dr->expire = new->expire; /* * If the preference does not change, there's no need * to sort the entries. */ if (rtpref(new) == oldpref) { splx(s); return (dr); } /* * preferred router may be changed, so relocate * this router. * XXX: calling TAILQ_REMOVE directly is a bad manner. * However, since defrtrlist_del() has many side * effects, we intentionally do so here. * defrouter_select() below will handle routing * changes later. */ TAILQ_REMOVE(&V_nd_defrouter, dr, dr_entry); n = dr; goto insert; } splx(s); return (dr); } /* entry does not exist */ if (new->rtlifetime == 0) { splx(s); return (NULL); } n = (struct nd_defrouter *)malloc(sizeof(*n), M_IP6NDP, M_NOWAIT); if (n == NULL) { splx(s); return (NULL); } bzero(n, sizeof(*n)); *n = *new; insert: /* * Insert the new router in the Default Router List; * The Default Router List should be in the descending order * of router-preferece. Routers with the same preference are * sorted in the arriving time order. */ /* insert at the end of the group */ for (dr = TAILQ_FIRST(&V_nd_defrouter); dr; dr = TAILQ_NEXT(dr, dr_entry)) { if (rtpref(n) > rtpref(dr)) break; } if (dr) TAILQ_INSERT_BEFORE(dr, n, dr_entry); else TAILQ_INSERT_TAIL(&V_nd_defrouter, n, dr_entry); defrouter_select(); splx(s); return (n); } static struct nd_pfxrouter * pfxrtr_lookup(struct nd_prefix *pr, struct nd_defrouter *dr) { struct nd_pfxrouter *search; for (search = pr->ndpr_advrtrs.lh_first; search; search = search->pfr_next) { if (search->router == dr) break; } return (search); } static void pfxrtr_add(struct nd_prefix *pr, struct nd_defrouter *dr) { struct nd_pfxrouter *new; new = (struct nd_pfxrouter *)malloc(sizeof(*new), M_IP6NDP, M_NOWAIT); if (new == NULL) return; bzero(new, sizeof(*new)); new->router = dr; LIST_INSERT_HEAD(&pr->ndpr_advrtrs, new, pfr_entry); pfxlist_onlink_check(); } static void pfxrtr_del(struct nd_pfxrouter *pfr) { LIST_REMOVE(pfr, pfr_entry); free(pfr, M_IP6NDP); } struct nd_prefix * nd6_prefix_lookup(struct nd_prefixctl *key) { struct nd_prefix *search; for (search = V_nd_prefix.lh_first; search; search = search->ndpr_next) { if (key->ndpr_ifp == search->ndpr_ifp && key->ndpr_plen == search->ndpr_plen && in6_are_prefix_equal(&key->ndpr_prefix.sin6_addr, &search->ndpr_prefix.sin6_addr, key->ndpr_plen)) { break; } } return (search); } int nd6_prelist_add(struct nd_prefixctl *pr, struct nd_defrouter *dr, struct nd_prefix **newp) { struct nd_prefix *new = NULL; int error = 0; int i, s; char ip6buf[INET6_ADDRSTRLEN]; new = (struct nd_prefix *)malloc(sizeof(*new), M_IP6NDP, M_NOWAIT); if (new == NULL) return(ENOMEM); bzero(new, sizeof(*new)); new->ndpr_ifp = pr->ndpr_ifp; new->ndpr_prefix = pr->ndpr_prefix; new->ndpr_plen = pr->ndpr_plen; new->ndpr_vltime = pr->ndpr_vltime; new->ndpr_pltime = pr->ndpr_pltime; new->ndpr_flags = pr->ndpr_flags; if ((error = in6_init_prefix_ltimes(new)) != 0) { free(new, M_IP6NDP); return(error); } new->ndpr_lastupdate = time_second; if (newp != NULL) *newp = new; /* initialization */ LIST_INIT(&new->ndpr_advrtrs); in6_prefixlen2mask(&new->ndpr_mask, new->ndpr_plen); /* make prefix in the canonical form */ for (i = 0; i < 4; i++) new->ndpr_prefix.sin6_addr.s6_addr32[i] &= new->ndpr_mask.s6_addr32[i]; s = splnet(); /* link ndpr_entry to nd_prefix list */ LIST_INSERT_HEAD(&V_nd_prefix, new, ndpr_entry); splx(s); /* ND_OPT_PI_FLAG_ONLINK processing */ if (new->ndpr_raf_onlink) { int e; if ((e = nd6_prefix_onlink(new)) != 0) { nd6log((LOG_ERR, "nd6_prelist_add: failed to make " "the prefix %s/%d on-link on %s (errno=%d)\n", ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr), pr->ndpr_plen, if_name(pr->ndpr_ifp), e)); /* proceed anyway. XXX: is it correct? */ } } if (dr) pfxrtr_add(new, dr); return 0; } void prelist_remove(struct nd_prefix *pr) { struct nd_pfxrouter *pfr, *next; int e, s; char ip6buf[INET6_ADDRSTRLEN]; /* make sure to invalidate the prefix until it is really freed. */ pr->ndpr_vltime = 0; pr->ndpr_pltime = 0; /* * Though these flags are now meaningless, we'd rather keep the value * of pr->ndpr_raf_onlink and pr->ndpr_raf_auto not to confuse users * when executing "ndp -p". */ if ((pr->ndpr_stateflags & NDPRF_ONLINK) != 0 && (e = nd6_prefix_offlink(pr)) != 0) { nd6log((LOG_ERR, "prelist_remove: failed to make %s/%d offlink " "on %s, errno=%d\n", ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr), pr->ndpr_plen, if_name(pr->ndpr_ifp), e)); /* what should we do? */ } if (pr->ndpr_refcnt > 0) return; /* notice here? */ s = splnet(); /* unlink ndpr_entry from nd_prefix list */ LIST_REMOVE(pr, ndpr_entry); /* free list of routers that adversed the prefix */ for (pfr = pr->ndpr_advrtrs.lh_first; pfr; pfr = next) { next = pfr->pfr_next; free(pfr, M_IP6NDP); } splx(s); free(pr, M_IP6NDP); pfxlist_onlink_check(); } /* * dr - may be NULL */ static int prelist_update(struct nd_prefixctl *new, struct nd_defrouter *dr, struct mbuf *m, int mcast) { struct in6_ifaddr *ia6 = NULL, *ia6_match = NULL; struct ifaddr *ifa; struct ifnet *ifp = new->ndpr_ifp; struct nd_prefix *pr; int s = splnet(); int error = 0; int newprefix = 0; int auth; struct in6_addrlifetime lt6_tmp; char ip6buf[INET6_ADDRSTRLEN]; auth = 0; if (m) { /* * Authenticity for NA consists authentication for * both IP header and IP datagrams, doesn't it ? */ #if defined(M_AUTHIPHDR) && defined(M_AUTHIPDGM) auth = ((m->m_flags & M_AUTHIPHDR) && (m->m_flags & M_AUTHIPDGM)); #endif } if ((pr = nd6_prefix_lookup(new)) != NULL) { /* * nd6_prefix_lookup() ensures that pr and new have the same * prefix on a same interface. */ /* * Update prefix information. Note that the on-link (L) bit * and the autonomous (A) bit should NOT be changed from 1 * to 0. */ if (new->ndpr_raf_onlink == 1) pr->ndpr_raf_onlink = 1; if (new->ndpr_raf_auto == 1) pr->ndpr_raf_auto = 1; if (new->ndpr_raf_onlink) { pr->ndpr_vltime = new->ndpr_vltime; pr->ndpr_pltime = new->ndpr_pltime; (void)in6_init_prefix_ltimes(pr); /* XXX error case? */ pr->ndpr_lastupdate = time_second; } if (new->ndpr_raf_onlink && (pr->ndpr_stateflags & NDPRF_ONLINK) == 0) { int e; if ((e = nd6_prefix_onlink(pr)) != 0) { nd6log((LOG_ERR, "prelist_update: failed to make " "the prefix %s/%d on-link on %s " "(errno=%d)\n", ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr), pr->ndpr_plen, if_name(pr->ndpr_ifp), e)); /* proceed anyway. XXX: is it correct? */ } } if (dr && pfxrtr_lookup(pr, dr) == NULL) pfxrtr_add(pr, dr); } else { struct nd_prefix *newpr = NULL; newprefix = 1; if (new->ndpr_vltime == 0) goto end; if (new->ndpr_raf_onlink == 0 && new->ndpr_raf_auto == 0) goto end; error = nd6_prelist_add(new, dr, &newpr); if (error != 0 || newpr == NULL) { nd6log((LOG_NOTICE, "prelist_update: " "nd6_prelist_add failed for %s/%d on %s " "errno=%d, returnpr=%p\n", ip6_sprintf(ip6buf, &new->ndpr_prefix.sin6_addr), new->ndpr_plen, if_name(new->ndpr_ifp), error, newpr)); goto end; /* we should just give up in this case. */ } /* * XXX: from the ND point of view, we can ignore a prefix * with the on-link bit being zero. However, we need a * prefix structure for references from autoconfigured * addresses. Thus, we explicitly make sure that the prefix * itself expires now. */ if (newpr->ndpr_raf_onlink == 0) { newpr->ndpr_vltime = 0; newpr->ndpr_pltime = 0; in6_init_prefix_ltimes(newpr); } pr = newpr; } /* * Address autoconfiguration based on Section 5.5.3 of RFC 2462. * Note that pr must be non NULL at this point. */ /* 5.5.3 (a). Ignore the prefix without the A bit set. */ if (!new->ndpr_raf_auto) goto end; /* * 5.5.3 (b). the link-local prefix should have been ignored in * nd6_ra_input. */ /* 5.5.3 (c). Consistency check on lifetimes: pltime <= vltime. */ if (new->ndpr_pltime > new->ndpr_vltime) { error = EINVAL; /* XXX: won't be used */ goto end; } /* * 5.5.3 (d). If the prefix advertised is not equal to the prefix of * an address configured by stateless autoconfiguration already in the * list of addresses associated with the interface, and the Valid * Lifetime is not 0, form an address. We first check if we have * a matching prefix. * Note: we apply a clarification in rfc2462bis-02 here. We only * consider autoconfigured addresses while RFC2462 simply said * "address". */ IF_ADDR_LOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { struct in6_ifaddr *ifa6; u_int32_t remaininglifetime; if (ifa->ifa_addr->sa_family != AF_INET6) continue; ifa6 = (struct in6_ifaddr *)ifa; /* * We only consider autoconfigured addresses as per rfc2462bis. */ if (!(ifa6->ia6_flags & IN6_IFF_AUTOCONF)) continue; /* * Spec is not clear here, but I believe we should concentrate * on unicast (i.e. not anycast) addresses. * XXX: other ia6_flags? detached or duplicated? */ if ((ifa6->ia6_flags & IN6_IFF_ANYCAST) != 0) continue; /* * Ignore the address if it is not associated with a prefix * or is associated with a prefix that is different from this * one. (pr is never NULL here) */ if (ifa6->ia6_ndpr != pr) continue; if (ia6_match == NULL) /* remember the first one */ ia6_match = ifa6; /* * An already autoconfigured address matched. Now that we * are sure there is at least one matched address, we can * proceed to 5.5.3. (e): update the lifetimes according to the * "two hours" rule and the privacy extension. * We apply some clarifications in rfc2462bis: * - use remaininglifetime instead of storedlifetime as a * variable name * - remove the dead code in the "two-hour" rule */ #define TWOHOUR (120*60) lt6_tmp = ifa6->ia6_lifetime; if (lt6_tmp.ia6t_vltime == ND6_INFINITE_LIFETIME) remaininglifetime = ND6_INFINITE_LIFETIME; else if (time_second - ifa6->ia6_updatetime > lt6_tmp.ia6t_vltime) { /* * The case of "invalid" address. We should usually * not see this case. */ remaininglifetime = 0; } else remaininglifetime = lt6_tmp.ia6t_vltime - (time_second - ifa6->ia6_updatetime); /* when not updating, keep the current stored lifetime. */ lt6_tmp.ia6t_vltime = remaininglifetime; if (TWOHOUR < new->ndpr_vltime || remaininglifetime < new->ndpr_vltime) { lt6_tmp.ia6t_vltime = new->ndpr_vltime; } else if (remaininglifetime <= TWOHOUR) { if (auth) { lt6_tmp.ia6t_vltime = new->ndpr_vltime; } } else { /* * new->ndpr_vltime <= TWOHOUR && * TWOHOUR < remaininglifetime */ lt6_tmp.ia6t_vltime = TWOHOUR; } /* The 2 hour rule is not imposed for preferred lifetime. */ lt6_tmp.ia6t_pltime = new->ndpr_pltime; in6_init_address_ltimes(pr, <6_tmp); /* * We need to treat lifetimes for temporary addresses * differently, according to * draft-ietf-ipv6-privacy-addrs-v2-01.txt 3.3 (1); * we only update the lifetimes when they are in the maximum * intervals. */ if ((ifa6->ia6_flags & IN6_IFF_TEMPORARY) != 0) { u_int32_t maxvltime, maxpltime; if (V_ip6_temp_valid_lifetime > (u_int32_t)((time_second - ifa6->ia6_createtime) + V_ip6_desync_factor)) { maxvltime = V_ip6_temp_valid_lifetime - (time_second - ifa6->ia6_createtime) - V_ip6_desync_factor; } else maxvltime = 0; if (V_ip6_temp_preferred_lifetime > (u_int32_t)((time_second - ifa6->ia6_createtime) + V_ip6_desync_factor)) { maxpltime = V_ip6_temp_preferred_lifetime - (time_second - ifa6->ia6_createtime) - V_ip6_desync_factor; } else maxpltime = 0; if (lt6_tmp.ia6t_vltime == ND6_INFINITE_LIFETIME || lt6_tmp.ia6t_vltime > maxvltime) { lt6_tmp.ia6t_vltime = maxvltime; } if (lt6_tmp.ia6t_pltime == ND6_INFINITE_LIFETIME || lt6_tmp.ia6t_pltime > maxpltime) { lt6_tmp.ia6t_pltime = maxpltime; } } ifa6->ia6_lifetime = lt6_tmp; ifa6->ia6_updatetime = time_second; } IF_ADDR_UNLOCK(ifp); if (ia6_match == NULL && new->ndpr_vltime) { int ifidlen; /* * 5.5.3 (d) (continued) * No address matched and the valid lifetime is non-zero. * Create a new address. */ /* * Prefix Length check: * If the sum of the prefix length and interface identifier * length does not equal 128 bits, the Prefix Information * option MUST be ignored. The length of the interface * identifier is defined in a separate link-type specific * document. */ ifidlen = in6_if2idlen(ifp); if (ifidlen < 0) { /* this should not happen, so we always log it. */ log(LOG_ERR, "prelist_update: IFID undefined (%s)\n", if_name(ifp)); goto end; } if (ifidlen + pr->ndpr_plen != 128) { nd6log((LOG_INFO, "prelist_update: invalid prefixlen " "%d for %s, ignored\n", pr->ndpr_plen, if_name(ifp))); goto end; } if ((ia6 = in6_ifadd(new, mcast)) != NULL) { /* * note that we should use pr (not new) for reference. */ pr->ndpr_refcnt++; ia6->ia6_ndpr = pr; /* * RFC 3041 3.3 (2). * When a new public address is created as described * in RFC2462, also create a new temporary address. * * RFC 3041 3.5. * When an interface connects to a new link, a new * randomized interface identifier should be generated * immediately together with a new set of temporary * addresses. Thus, we specifiy 1 as the 2nd arg of * in6_tmpifadd(). */ if (V_ip6_use_tempaddr) { int e; if ((e = in6_tmpifadd(ia6, 1, 1)) != 0) { nd6log((LOG_NOTICE, "prelist_update: " "failed to create a temporary " "address, errno=%d\n", e)); } } ifa_free(&ia6->ia_ifa); /* * A newly added address might affect the status * of other addresses, so we check and update it. * XXX: what if address duplication happens? */ pfxlist_onlink_check(); } else { /* just set an error. do not bark here. */ error = EADDRNOTAVAIL; /* XXX: might be unused. */ } } end: splx(s); return error; } /* * A supplement function used in the on-link detection below; * detect if a given prefix has a (probably) reachable advertising router. * XXX: lengthy function name... */ static struct nd_pfxrouter * find_pfxlist_reachable_router(struct nd_prefix *pr) { struct nd_pfxrouter *pfxrtr; struct llentry *ln; int canreach; for (pfxrtr = LIST_FIRST(&pr->ndpr_advrtrs); pfxrtr != NULL; pfxrtr = LIST_NEXT(pfxrtr, pfr_entry)) { IF_AFDATA_LOCK(pfxrtr->router->ifp); ln = nd6_lookup(&pfxrtr->router->rtaddr, 0, pfxrtr->router->ifp); IF_AFDATA_UNLOCK(pfxrtr->router->ifp); if (ln == NULL) continue; canreach = ND6_IS_LLINFO_PROBREACH(ln); LLE_RUNLOCK(ln); if (canreach) break; } return (pfxrtr); } /* * Check if each prefix in the prefix list has at least one available router * that advertised the prefix (a router is "available" if its neighbor cache * entry is reachable or probably reachable). * If the check fails, the prefix may be off-link, because, for example, * we have moved from the network but the lifetime of the prefix has not * expired yet. So we should not use the prefix if there is another prefix * that has an available router. * But, if there is no prefix that has an available router, we still regards * all the prefixes as on-link. This is because we can't tell if all the * routers are simply dead or if we really moved from the network and there * is no router around us. */ void pfxlist_onlink_check() { struct nd_prefix *pr; struct in6_ifaddr *ifa; struct nd_defrouter *dr; struct nd_pfxrouter *pfxrtr = NULL; /* * Check if there is a prefix that has a reachable advertising * router. */ for (pr = V_nd_prefix.lh_first; pr; pr = pr->ndpr_next) { if (pr->ndpr_raf_onlink && find_pfxlist_reachable_router(pr)) break; } /* * If we have no such prefix, check whether we still have a router * that does not advertise any prefixes. */ if (pr == NULL) { for (dr = TAILQ_FIRST(&V_nd_defrouter); dr; dr = TAILQ_NEXT(dr, dr_entry)) { struct nd_prefix *pr0; for (pr0 = V_nd_prefix.lh_first; pr0; pr0 = pr0->ndpr_next) { if ((pfxrtr = pfxrtr_lookup(pr0, dr)) != NULL) break; } if (pfxrtr != NULL) break; } } if (pr != NULL || (TAILQ_FIRST(&V_nd_defrouter) && pfxrtr == NULL)) { /* * There is at least one prefix that has a reachable router, * or at least a router which probably does not advertise * any prefixes. The latter would be the case when we move * to a new link where we have a router that does not provide * prefixes and we configure an address by hand. * Detach prefixes which have no reachable advertising * router, and attach other prefixes. */ for (pr = V_nd_prefix.lh_first; pr; pr = pr->ndpr_next) { /* XXX: a link-local prefix should never be detached */ if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr)) continue; /* * we aren't interested in prefixes without the L bit * set. */ if (pr->ndpr_raf_onlink == 0) continue; if (pr->ndpr_raf_auto == 0) continue; if ((pr->ndpr_stateflags & NDPRF_DETACHED) == 0 && find_pfxlist_reachable_router(pr) == NULL) pr->ndpr_stateflags |= NDPRF_DETACHED; if ((pr->ndpr_stateflags & NDPRF_DETACHED) != 0 && find_pfxlist_reachable_router(pr) != 0) pr->ndpr_stateflags &= ~NDPRF_DETACHED; } } else { /* there is no prefix that has a reachable router */ for (pr = V_nd_prefix.lh_first; pr; pr = pr->ndpr_next) { if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr)) continue; if (pr->ndpr_raf_onlink == 0) continue; if (pr->ndpr_raf_auto == 0) continue; if ((pr->ndpr_stateflags & NDPRF_DETACHED) != 0) pr->ndpr_stateflags &= ~NDPRF_DETACHED; } } /* * Remove each interface route associated with a (just) detached * prefix, and reinstall the interface route for a (just) attached * prefix. Note that all attempt of reinstallation does not * necessarily success, when a same prefix is shared among multiple * interfaces. Such cases will be handled in nd6_prefix_onlink, * so we don't have to care about them. */ for (pr = V_nd_prefix.lh_first; pr; pr = pr->ndpr_next) { int e; char ip6buf[INET6_ADDRSTRLEN]; if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr)) continue; if (pr->ndpr_raf_onlink == 0) continue; if (pr->ndpr_raf_auto == 0) continue; if ((pr->ndpr_stateflags & NDPRF_DETACHED) != 0 && (pr->ndpr_stateflags & NDPRF_ONLINK) != 0) { if ((e = nd6_prefix_offlink(pr)) != 0) { nd6log((LOG_ERR, "pfxlist_onlink_check: failed to " "make %s/%d offlink, errno=%d\n", ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr), pr->ndpr_plen, e)); } } if ((pr->ndpr_stateflags & NDPRF_DETACHED) == 0 && (pr->ndpr_stateflags & NDPRF_ONLINK) == 0 && pr->ndpr_raf_onlink) { if ((e = nd6_prefix_onlink(pr)) != 0) { nd6log((LOG_ERR, "pfxlist_onlink_check: failed to " "make %s/%d onlink, errno=%d\n", ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr), pr->ndpr_plen, e)); } } } /* * Changes on the prefix status might affect address status as well. * Make sure that all addresses derived from an attached prefix are * attached, and that all addresses derived from a detached prefix are * detached. Note, however, that a manually configured address should * always be attached. * The precise detection logic is same as the one for prefixes. * * XXXRW: in6_ifaddrhead locking. */ TAILQ_FOREACH(ifa, &V_in6_ifaddrhead, ia_link) { if (!(ifa->ia6_flags & IN6_IFF_AUTOCONF)) continue; if (ifa->ia6_ndpr == NULL) { /* * This can happen when we first configure the address * (i.e. the address exists, but the prefix does not). * XXX: complicated relationships... */ continue; } if (find_pfxlist_reachable_router(ifa->ia6_ndpr)) break; } if (ifa) { TAILQ_FOREACH(ifa, &V_in6_ifaddrhead, ia_link) { if ((ifa->ia6_flags & IN6_IFF_AUTOCONF) == 0) continue; if (ifa->ia6_ndpr == NULL) /* XXX: see above. */ continue; if (find_pfxlist_reachable_router(ifa->ia6_ndpr)) { if (ifa->ia6_flags & IN6_IFF_DETACHED) { ifa->ia6_flags &= ~IN6_IFF_DETACHED; ifa->ia6_flags |= IN6_IFF_TENTATIVE; nd6_dad_start((struct ifaddr *)ifa, 0); } } else { ifa->ia6_flags |= IN6_IFF_DETACHED; } } } else { TAILQ_FOREACH(ifa, &V_in6_ifaddrhead, ia_link) { if ((ifa->ia6_flags & IN6_IFF_AUTOCONF) == 0) continue; if (ifa->ia6_flags & IN6_IFF_DETACHED) { ifa->ia6_flags &= ~IN6_IFF_DETACHED; ifa->ia6_flags |= IN6_IFF_TENTATIVE; /* Do we need a delay in this case? */ nd6_dad_start((struct ifaddr *)ifa, 0); } } } } int nd6_prefix_onlink(struct nd_prefix *pr) { struct ifaddr *ifa; struct ifnet *ifp = pr->ndpr_ifp; struct sockaddr_in6 mask6; struct nd_prefix *opr; u_long rtflags; int error = 0; struct radix_node_head *rnh; struct rtentry *rt = NULL; char ip6buf[INET6_ADDRSTRLEN]; struct sockaddr_dl null_sdl = {sizeof(null_sdl), AF_LINK}; /* sanity check */ if ((pr->ndpr_stateflags & NDPRF_ONLINK) != 0) { nd6log((LOG_ERR, "nd6_prefix_onlink: %s/%d is already on-link\n", ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr), pr->ndpr_plen)); return (EEXIST); } /* * Add the interface route associated with the prefix. Before * installing the route, check if there's the same prefix on another * interface, and the prefix has already installed the interface route. * Although such a configuration is expected to be rare, we explicitly * allow it. */ for (opr = V_nd_prefix.lh_first; opr; opr = opr->ndpr_next) { if (opr == pr) continue; if ((opr->ndpr_stateflags & NDPRF_ONLINK) == 0) continue; if (opr->ndpr_plen == pr->ndpr_plen && in6_are_prefix_equal(&pr->ndpr_prefix.sin6_addr, &opr->ndpr_prefix.sin6_addr, pr->ndpr_plen)) return (0); } /* * We prefer link-local addresses as the associated interface address. */ /* search for a link-local addr */ ifa = (struct ifaddr *)in6ifa_ifpforlinklocal(ifp, IN6_IFF_NOTREADY | IN6_IFF_ANYCAST); if (ifa == NULL) { /* XXX: freebsd does not have ifa_ifwithaf */ IF_ADDR_LOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family == AF_INET6) break; } if (ifa != NULL) ifa_ref(ifa); IF_ADDR_UNLOCK(ifp); /* should we care about ia6_flags? */ } if (ifa == NULL) { /* * This can still happen, when, for example, we receive an RA * containing a prefix with the L bit set and the A bit clear, * after removing all IPv6 addresses on the receiving * interface. This should, of course, be rare though. */ nd6log((LOG_NOTICE, "nd6_prefix_onlink: failed to find any ifaddr" " to add route for a prefix(%s/%d) on %s\n", ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr), pr->ndpr_plen, if_name(ifp))); return (0); } /* * in6_ifinit() sets nd6_rtrequest to ifa_rtrequest for all ifaddrs. * ifa->ifa_rtrequest = nd6_rtrequest; */ bzero(&mask6, sizeof(mask6)); mask6.sin6_len = sizeof(mask6); mask6.sin6_addr = pr->ndpr_mask; rtflags = (ifa->ifa_flags & ~IFA_RTSELF) | RTF_UP; error = rtrequest(RTM_ADD, (struct sockaddr *)&pr->ndpr_prefix, ifa->ifa_addr, (struct sockaddr *)&mask6, rtflags, &rt); if (error == 0) { if (rt != NULL) /* this should be non NULL, though */ { rnh = rt_tables_get_rnh(rt->rt_fibnum, AF_INET6); /* XXX what if rhn == NULL? */ RADIX_NODE_HEAD_LOCK(rnh); RT_LOCK(rt); if (!rt_setgate(rt, rt_key(rt), (struct sockaddr *)&null_sdl)) { ((struct sockaddr_dl *)rt->rt_gateway)->sdl_type = rt->rt_ifp->if_type; ((struct sockaddr_dl *)rt->rt_gateway)->sdl_index = rt->rt_ifp->if_index; } RADIX_NODE_HEAD_UNLOCK(rnh); nd6_rtmsg(RTM_ADD, rt); RT_UNLOCK(rt); } pr->ndpr_stateflags |= NDPRF_ONLINK; } else { char ip6bufg[INET6_ADDRSTRLEN], ip6bufm[INET6_ADDRSTRLEN]; nd6log((LOG_ERR, "nd6_prefix_onlink: failed to add route for a" " prefix (%s/%d) on %s, gw=%s, mask=%s, flags=%lx " "errno = %d\n", ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr), pr->ndpr_plen, if_name(ifp), ip6_sprintf(ip6bufg, &((struct sockaddr_in6 *)ifa->ifa_addr)->sin6_addr), ip6_sprintf(ip6bufm, &mask6.sin6_addr), rtflags, error)); } if (rt != NULL) { RT_LOCK(rt); RT_REMREF(rt); RT_UNLOCK(rt); } if (ifa != NULL) ifa_free(ifa); return (error); } int nd6_prefix_offlink(struct nd_prefix *pr) { int error = 0; struct ifnet *ifp = pr->ndpr_ifp; struct nd_prefix *opr; struct sockaddr_in6 sa6, mask6; struct rtentry *rt = NULL; char ip6buf[INET6_ADDRSTRLEN]; /* sanity check */ if ((pr->ndpr_stateflags & NDPRF_ONLINK) == 0) { nd6log((LOG_ERR, "nd6_prefix_offlink: %s/%d is already off-link\n", ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr), pr->ndpr_plen)); return (EEXIST); } bzero(&sa6, sizeof(sa6)); sa6.sin6_family = AF_INET6; sa6.sin6_len = sizeof(sa6); bcopy(&pr->ndpr_prefix.sin6_addr, &sa6.sin6_addr, sizeof(struct in6_addr)); bzero(&mask6, sizeof(mask6)); mask6.sin6_family = AF_INET6; mask6.sin6_len = sizeof(sa6); bcopy(&pr->ndpr_mask, &mask6.sin6_addr, sizeof(struct in6_addr)); error = rtrequest(RTM_DELETE, (struct sockaddr *)&sa6, NULL, (struct sockaddr *)&mask6, 0, &rt); if (error == 0) { pr->ndpr_stateflags &= ~NDPRF_ONLINK; /* report the route deletion to the routing socket. */ if (rt != NULL) nd6_rtmsg(RTM_DELETE, rt); /* * There might be the same prefix on another interface, * the prefix which could not be on-link just because we have * the interface route (see comments in nd6_prefix_onlink). * If there's one, try to make the prefix on-link on the * interface. */ for (opr = V_nd_prefix.lh_first; opr; opr = opr->ndpr_next) { if (opr == pr) continue; if ((opr->ndpr_stateflags & NDPRF_ONLINK) != 0) continue; /* * KAME specific: detached prefixes should not be * on-link. */ if ((opr->ndpr_stateflags & NDPRF_DETACHED) != 0) continue; if (opr->ndpr_plen == pr->ndpr_plen && in6_are_prefix_equal(&pr->ndpr_prefix.sin6_addr, &opr->ndpr_prefix.sin6_addr, pr->ndpr_plen)) { int e; if ((e = nd6_prefix_onlink(opr)) != 0) { nd6log((LOG_ERR, "nd6_prefix_offlink: failed to " "recover a prefix %s/%d from %s " "to %s (errno = %d)\n", ip6_sprintf(ip6buf, &opr->ndpr_prefix.sin6_addr), opr->ndpr_plen, if_name(ifp), if_name(opr->ndpr_ifp), e)); } } } } else { /* XXX: can we still set the NDPRF_ONLINK flag? */ nd6log((LOG_ERR, "nd6_prefix_offlink: failed to delete route: " "%s/%d on %s (errno = %d)\n", ip6_sprintf(ip6buf, &sa6.sin6_addr), pr->ndpr_plen, if_name(ifp), error)); } if (rt != NULL) { RTFREE(rt); } return (error); } static struct in6_ifaddr * in6_ifadd(struct nd_prefixctl *pr, int mcast) { struct ifnet *ifp = pr->ndpr_ifp; struct ifaddr *ifa; struct in6_aliasreq ifra; struct in6_ifaddr *ia, *ib; int error, plen0; struct in6_addr mask; int prefixlen = pr->ndpr_plen; int updateflags; char ip6buf[INET6_ADDRSTRLEN]; in6_prefixlen2mask(&mask, prefixlen); /* * find a link-local address (will be interface ID). * Is it really mandatory? Theoretically, a global or a site-local * address can be configured without a link-local address, if we * have a unique interface identifier... * * it is not mandatory to have a link-local address, we can generate * interface identifier on the fly. we do this because: * (1) it should be the easiest way to find interface identifier. * (2) RFC2462 5.4 suggesting the use of the same interface identifier * for multiple addresses on a single interface, and possible shortcut * of DAD. we omitted DAD for this reason in the past. * (3) a user can prevent autoconfiguration of global address * by removing link-local address by hand (this is partly because we * don't have other way to control the use of IPv6 on an interface. * this has been our design choice - cf. NRL's "ifconfig auto"). * (4) it is easier to manage when an interface has addresses * with the same interface identifier, than to have multiple addresses * with different interface identifiers. */ ifa = (struct ifaddr *)in6ifa_ifpforlinklocal(ifp, 0); /* 0 is OK? */ if (ifa) ib = (struct in6_ifaddr *)ifa; else return NULL; /* prefixlen + ifidlen must be equal to 128 */ plen0 = in6_mask2len(&ib->ia_prefixmask.sin6_addr, NULL); if (prefixlen != plen0) { ifa_free(ifa); nd6log((LOG_INFO, "in6_ifadd: wrong prefixlen for %s " "(prefix=%d ifid=%d)\n", if_name(ifp), prefixlen, 128 - plen0)); return NULL; } /* make ifaddr */ bzero(&ifra, sizeof(ifra)); /* * in6_update_ifa() does not use ifra_name, but we accurately set it * for safety. */ strncpy(ifra.ifra_name, if_name(ifp), sizeof(ifra.ifra_name)); ifra.ifra_addr.sin6_family = AF_INET6; ifra.ifra_addr.sin6_len = sizeof(struct sockaddr_in6); /* prefix */ ifra.ifra_addr.sin6_addr = pr->ndpr_prefix.sin6_addr; ifra.ifra_addr.sin6_addr.s6_addr32[0] &= mask.s6_addr32[0]; ifra.ifra_addr.sin6_addr.s6_addr32[1] &= mask.s6_addr32[1]; ifra.ifra_addr.sin6_addr.s6_addr32[2] &= mask.s6_addr32[2]; ifra.ifra_addr.sin6_addr.s6_addr32[3] &= mask.s6_addr32[3]; /* interface ID */ ifra.ifra_addr.sin6_addr.s6_addr32[0] |= (ib->ia_addr.sin6_addr.s6_addr32[0] & ~mask.s6_addr32[0]); ifra.ifra_addr.sin6_addr.s6_addr32[1] |= (ib->ia_addr.sin6_addr.s6_addr32[1] & ~mask.s6_addr32[1]); ifra.ifra_addr.sin6_addr.s6_addr32[2] |= (ib->ia_addr.sin6_addr.s6_addr32[2] & ~mask.s6_addr32[2]); ifra.ifra_addr.sin6_addr.s6_addr32[3] |= (ib->ia_addr.sin6_addr.s6_addr32[3] & ~mask.s6_addr32[3]); ifa_free(ifa); /* new prefix mask. */ ifra.ifra_prefixmask.sin6_len = sizeof(struct sockaddr_in6); ifra.ifra_prefixmask.sin6_family = AF_INET6; bcopy(&mask, &ifra.ifra_prefixmask.sin6_addr, sizeof(ifra.ifra_prefixmask.sin6_addr)); /* lifetimes. */ ifra.ifra_lifetime.ia6t_vltime = pr->ndpr_vltime; ifra.ifra_lifetime.ia6t_pltime = pr->ndpr_pltime; /* XXX: scope zone ID? */ ifra.ifra_flags |= IN6_IFF_AUTOCONF; /* obey autoconf */ /* * Make sure that we do not have this address already. This should * usually not happen, but we can still see this case, e.g., if we * have manually configured the exact address to be configured. */ ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp, &ifra.ifra_addr.sin6_addr); if (ifa != NULL) { ifa_free(ifa); /* this should be rare enough to make an explicit log */ log(LOG_INFO, "in6_ifadd: %s is already configured\n", ip6_sprintf(ip6buf, &ifra.ifra_addr.sin6_addr)); return (NULL); } /* * Allocate ifaddr structure, link into chain, etc. * If we are going to create a new address upon receiving a multicasted * RA, we need to impose a random delay before starting DAD. * [draft-ietf-ipv6-rfc2462bis-02.txt, Section 5.4.2] */ updateflags = 0; if (mcast) updateflags |= IN6_IFAUPDATE_DADDELAY; if ((error = in6_update_ifa(ifp, &ifra, NULL, updateflags)) != 0) { nd6log((LOG_ERR, "in6_ifadd: failed to make ifaddr %s on %s (errno=%d)\n", ip6_sprintf(ip6buf, &ifra.ifra_addr.sin6_addr), if_name(ifp), error)); return (NULL); /* ifaddr must not have been allocated. */ } ia = in6ifa_ifpwithaddr(ifp, &ifra.ifra_addr.sin6_addr); /* * XXXRW: Assumption of non-NULLness here might not be true with * fine-grained locking -- should we validate it? Or just return * earlier ifa rather than looking it up again? */ return (ia); /* this is always non-NULL and referenced. */ } /* * ia0 - corresponding public address */ int in6_tmpifadd(const struct in6_ifaddr *ia0, int forcegen, int delay) { struct ifnet *ifp = ia0->ia_ifa.ifa_ifp; struct in6_ifaddr *newia, *ia; struct in6_aliasreq ifra; int i, error; int trylimit = 3; /* XXX: adhoc value */ int updateflags; u_int32_t randid[2]; time_t vltime0, pltime0; bzero(&ifra, sizeof(ifra)); strncpy(ifra.ifra_name, if_name(ifp), sizeof(ifra.ifra_name)); ifra.ifra_addr = ia0->ia_addr; /* copy prefix mask */ ifra.ifra_prefixmask = ia0->ia_prefixmask; /* clear the old IFID */ for (i = 0; i < 4; i++) { ifra.ifra_addr.sin6_addr.s6_addr32[i] &= ifra.ifra_prefixmask.sin6_addr.s6_addr32[i]; } again: if (in6_get_tmpifid(ifp, (u_int8_t *)randid, (const u_int8_t *)&ia0->ia_addr.sin6_addr.s6_addr[8], forcegen)) { nd6log((LOG_NOTICE, "in6_tmpifadd: failed to find a good " "random IFID\n")); return (EINVAL); } ifra.ifra_addr.sin6_addr.s6_addr32[2] |= (randid[0] & ~(ifra.ifra_prefixmask.sin6_addr.s6_addr32[2])); ifra.ifra_addr.sin6_addr.s6_addr32[3] |= (randid[1] & ~(ifra.ifra_prefixmask.sin6_addr.s6_addr32[3])); /* * in6_get_tmpifid() quite likely provided a unique interface ID. * However, we may still have a chance to see collision, because * there may be a time lag between generation of the ID and generation * of the address. So, we'll do one more sanity check. */ IN6_IFADDR_RLOCK(); TAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) { if (IN6_ARE_ADDR_EQUAL(&ia->ia_addr.sin6_addr, &ifra.ifra_addr.sin6_addr)) { if (trylimit-- == 0) { IN6_IFADDR_RUNLOCK(); /* * Give up. Something strange should have * happened. */ nd6log((LOG_NOTICE, "in6_tmpifadd: failed to " "find a unique random IFID\n")); return (EEXIST); } IN6_IFADDR_RUNLOCK(); forcegen = 1; goto again; } } IN6_IFADDR_RUNLOCK(); /* * The Valid Lifetime is the lower of the Valid Lifetime of the * public address or TEMP_VALID_LIFETIME. * The Preferred Lifetime is the lower of the Preferred Lifetime * of the public address or TEMP_PREFERRED_LIFETIME - * DESYNC_FACTOR. */ if (ia0->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) { vltime0 = IFA6_IS_INVALID(ia0) ? 0 : (ia0->ia6_lifetime.ia6t_vltime - (time_second - ia0->ia6_updatetime)); if (vltime0 > V_ip6_temp_valid_lifetime) vltime0 = V_ip6_temp_valid_lifetime; } else vltime0 = V_ip6_temp_valid_lifetime; if (ia0->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) { pltime0 = IFA6_IS_DEPRECATED(ia0) ? 0 : (ia0->ia6_lifetime.ia6t_pltime - (time_second - ia0->ia6_updatetime)); if (pltime0 > V_ip6_temp_preferred_lifetime - V_ip6_desync_factor){ pltime0 = V_ip6_temp_preferred_lifetime - V_ip6_desync_factor; } } else pltime0 = V_ip6_temp_preferred_lifetime - V_ip6_desync_factor; ifra.ifra_lifetime.ia6t_vltime = vltime0; ifra.ifra_lifetime.ia6t_pltime = pltime0; /* * A temporary address is created only if this calculated Preferred * Lifetime is greater than REGEN_ADVANCE time units. */ if (ifra.ifra_lifetime.ia6t_pltime <= V_ip6_temp_regen_advance) return (0); /* XXX: scope zone ID? */ ifra.ifra_flags |= (IN6_IFF_AUTOCONF|IN6_IFF_TEMPORARY); /* allocate ifaddr structure, link into chain, etc. */ updateflags = 0; if (delay) updateflags |= IN6_IFAUPDATE_DADDELAY; if ((error = in6_update_ifa(ifp, &ifra, NULL, updateflags)) != 0) return (error); newia = in6ifa_ifpwithaddr(ifp, &ifra.ifra_addr.sin6_addr); if (newia == NULL) { /* XXX: can it happen? */ nd6log((LOG_ERR, "in6_tmpifadd: ifa update succeeded, but we got " "no ifaddr\n")); return (EINVAL); /* XXX */ } newia->ia6_ndpr = ia0->ia6_ndpr; newia->ia6_ndpr->ndpr_refcnt++; ifa_free(&newia->ia_ifa); /* * A newly added address might affect the status of other addresses. * XXX: when the temporary address is generated with a new public * address, the onlink check is redundant. However, it would be safe * to do the check explicitly everywhere a new address is generated, * and, in fact, we surely need the check when we create a new * temporary address due to deprecation of an old temporary address. */ pfxlist_onlink_check(); return (0); } static int in6_init_prefix_ltimes(struct nd_prefix *ndpr) { if (ndpr->ndpr_pltime == ND6_INFINITE_LIFETIME) ndpr->ndpr_preferred = 0; else ndpr->ndpr_preferred = time_second + ndpr->ndpr_pltime; if (ndpr->ndpr_vltime == ND6_INFINITE_LIFETIME) ndpr->ndpr_expire = 0; else ndpr->ndpr_expire = time_second + ndpr->ndpr_vltime; return 0; } static void in6_init_address_ltimes(struct nd_prefix *new, struct in6_addrlifetime *lt6) { /* init ia6t_expire */ if (lt6->ia6t_vltime == ND6_INFINITE_LIFETIME) lt6->ia6t_expire = 0; else { lt6->ia6t_expire = time_second; lt6->ia6t_expire += lt6->ia6t_vltime; } /* init ia6t_preferred */ if (lt6->ia6t_pltime == ND6_INFINITE_LIFETIME) lt6->ia6t_preferred = 0; else { lt6->ia6t_preferred = time_second; lt6->ia6t_preferred += lt6->ia6t_pltime; } } /* * Delete all the routing table entries that use the specified gateway. * XXX: this function causes search through all entries of routing table, so * it shouldn't be called when acting as a router. */ void rt6_flush(struct in6_addr *gateway, struct ifnet *ifp) { struct radix_node_head *rnh; int s = splnet(); /* We'll care only link-local addresses */ if (!IN6_IS_ADDR_LINKLOCAL(gateway)) { splx(s); return; } rnh = rt_tables_get_rnh(0, AF_INET6); if (rnh == NULL) return; RADIX_NODE_HEAD_LOCK(rnh); rnh->rnh_walktree(rnh, rt6_deleteroute, (void *)gateway); RADIX_NODE_HEAD_UNLOCK(rnh); splx(s); } static int rt6_deleteroute(struct radix_node *rn, void *arg) { #define SIN6(s) ((struct sockaddr_in6 *)s) struct rtentry *rt = (struct rtentry *)rn; struct in6_addr *gate = (struct in6_addr *)arg; if (rt->rt_gateway == NULL || rt->rt_gateway->sa_family != AF_INET6) return (0); if (!IN6_ARE_ADDR_EQUAL(gate, &SIN6(rt->rt_gateway)->sin6_addr)) { return (0); } /* * Do not delete a static route. * XXX: this seems to be a bit ad-hoc. Should we consider the * 'cloned' bit instead? */ if ((rt->rt_flags & RTF_STATIC) != 0) return (0); /* * We delete only host route. This means, in particular, we don't * delete default route. */ if ((rt->rt_flags & RTF_HOST) == 0) return (0); return (rtrequest(RTM_DELETE, rt_key(rt), rt->rt_gateway, rt_mask(rt), rt->rt_flags, 0)); #undef SIN6 } int nd6_setdefaultiface(int ifindex) { int error = 0; if (ifindex < 0 || V_if_index < ifindex) return (EINVAL); if (ifindex != 0 && !ifnet_byindex(ifindex)) return (EINVAL); if (V_nd6_defifindex != ifindex) { V_nd6_defifindex = ifindex; if (V_nd6_defifindex > 0) V_nd6_defifp = ifnet_byindex(V_nd6_defifindex); else V_nd6_defifp = NULL; /* * Our current implementation assumes one-to-one maping between * interfaces and links, so it would be natural to use the * default interface as the default link. */ scope6_setdefault(V_nd6_defifp); } return (error); } Index: projects/largeSMP/sys/netinet6/send.h =================================================================== --- projects/largeSMP/sys/netinet6/send.h (revision 222782) +++ projects/largeSMP/sys/netinet6/send.h (revision 222783) @@ -1,45 +1,45 @@ /*- * Copyright (c) 2009-2010 Ana Kukec * 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 _NETINET6_SEND_H_ #define _NETINET6_SEND_H_ #define SND_OUT 0 /* Outgoing traffic */ #define SND_IN 1 /* Incoming traffic. */ struct sockaddr_send { - unsigned char send_len; /* total length */ + uint8_t send_len; /* total length */ sa_family_t send_family; /* address family */ int send_direction; int send_ifidx; char send_zero[8]; }; extern int (*send_sendso_input_hook)(struct mbuf *, struct ifnet *, int, int); #endif /* _NETINET6_SEND_H_ */ Index: projects/largeSMP/sys/sys/mbuf.h =================================================================== --- projects/largeSMP/sys/sys/mbuf.h (revision 222782) +++ projects/largeSMP/sys/sys/mbuf.h (revision 222783) @@ -1,1034 +1,1070 @@ /*- * 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. * 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. * * @(#)mbuf.h 8.5 (Berkeley) 2/19/95 * $FreeBSD$ */ #ifndef _SYS_MBUF_H_ #define _SYS_MBUF_H_ /* XXX: These includes suck. Sorry! */ #include #ifdef _KERNEL #include #include #ifdef WITNESS #include #endif #endif /* * Mbufs are of a single size, MSIZE (sys/param.h), which includes overhead. * An mbuf may add a single "mbuf cluster" of size MCLBYTES (also in * sys/param.h), which has no additional overhead and is used instead of the * internal data area; this is done when at least MINCLSIZE of data must be * stored. Additionally, it is possible to allocate a separate buffer * externally and attach it to the mbuf in a way similar to that of mbuf * clusters. */ #define MLEN (MSIZE - sizeof(struct m_hdr)) /* normal data len */ #define MHLEN (MLEN - sizeof(struct pkthdr)) /* data len w/pkthdr */ #define MINCLSIZE (MHLEN + 1) /* smallest amount to put in cluster */ #define M_MAXCOMPRESS (MHLEN / 2) /* max amount to copy for compression */ #ifdef _KERNEL /*- * Macro for type conversion: convert mbuf pointer to data pointer of correct * type: * * mtod(m, t) -- Convert mbuf pointer to data pointer of correct type. */ #define mtod(m, t) ((t)((m)->m_data)) /* * Argument structure passed to UMA routines during mbuf and packet * allocations. */ struct mb_args { int flags; /* Flags for mbuf being allocated */ short type; /* Type of mbuf being allocated */ }; #endif /* _KERNEL */ #if defined(__LP64__) #define M_HDR_PAD 6 #else #define M_HDR_PAD 2 #endif /* * Header present at the beginning of every mbuf. */ struct m_hdr { struct mbuf *mh_next; /* next buffer in chain */ struct mbuf *mh_nextpkt; /* next chain in queue/record */ caddr_t mh_data; /* location of data */ int mh_len; /* amount of data in this mbuf */ int mh_flags; /* flags; see below */ short mh_type; /* type of data in this mbuf */ uint8_t pad[M_HDR_PAD];/* word align */ }; /* * Packet tag structure (see below for details). */ struct m_tag { SLIST_ENTRY(m_tag) m_tag_link; /* List of packet tags */ u_int16_t m_tag_id; /* Tag ID */ u_int16_t m_tag_len; /* Length of data */ u_int32_t m_tag_cookie; /* ABI/Module ID */ void (*m_tag_free)(struct m_tag *); }; /* * Record/packet header in first mbuf of chain; valid only if M_PKTHDR is set. */ struct pkthdr { struct ifnet *rcvif; /* rcv interface */ /* variables for ip and tcp reassembly */ void *header; /* pointer to packet header */ int len; /* total packet length */ uint32_t flowid; /* packet's 4-tuple system * flow identifier */ /* variables for hardware checksum */ int csum_flags; /* flags regarding checksum */ int csum_data; /* data field used by csum routines */ u_int16_t tso_segsz; /* TSO segment size */ union { u_int16_t vt_vtag; /* Ethernet 802.1p+q vlan tag */ u_int16_t vt_nrecs; /* # of IGMPv3 records in this chain */ } PH_vt; SLIST_HEAD(packet_tags, m_tag) tags; /* list of packet tags */ }; #define ether_vtag PH_vt.vt_vtag /* * Description of external storage mapped into mbuf; valid only if M_EXT is * set. */ struct m_ext { caddr_t ext_buf; /* start of buffer */ void (*ext_free) /* free routine if not the usual */ (void *, void *); void *ext_arg1; /* optional argument pointer */ void *ext_arg2; /* optional argument pointer */ u_int ext_size; /* size of buffer, for ext_free */ volatile u_int *ref_cnt; /* pointer to ref count info */ int ext_type; /* type of external storage */ }; /* * The core of the mbuf object along with some shortcut defines for practical * purposes. */ struct mbuf { struct m_hdr m_hdr; union { struct { struct pkthdr MH_pkthdr; /* M_PKTHDR set */ union { struct m_ext MH_ext; /* M_EXT set */ char MH_databuf[MHLEN]; } MH_dat; } MH; char M_databuf[MLEN]; /* !M_PKTHDR, !M_EXT */ } M_dat; }; #define m_next m_hdr.mh_next #define m_len m_hdr.mh_len #define m_data m_hdr.mh_data #define m_type m_hdr.mh_type #define m_flags m_hdr.mh_flags #define m_nextpkt m_hdr.mh_nextpkt #define m_act m_nextpkt #define m_pkthdr M_dat.MH.MH_pkthdr #define m_ext M_dat.MH.MH_dat.MH_ext #define m_pktdat M_dat.MH.MH_dat.MH_databuf #define m_dat M_dat.M_databuf /* * mbuf flags. */ #define M_EXT 0x00000001 /* has associated external storage */ #define M_PKTHDR 0x00000002 /* start of record */ #define M_EOR 0x00000004 /* end of record */ #define M_RDONLY 0x00000008 /* associated data is marked read-only */ #define M_PROTO1 0x00000010 /* protocol-specific */ #define M_PROTO2 0x00000020 /* protocol-specific */ #define M_PROTO3 0x00000040 /* protocol-specific */ #define M_PROTO4 0x00000080 /* protocol-specific */ #define M_PROTO5 0x00000100 /* protocol-specific */ #define M_BCAST 0x00000200 /* send/received as link-level broadcast */ #define M_MCAST 0x00000400 /* send/received as link-level multicast */ #define M_FRAG 0x00000800 /* packet is a fragment of a larger packet */ #define M_FIRSTFRAG 0x00001000 /* packet is first fragment */ #define M_LASTFRAG 0x00002000 /* packet is last fragment */ #define M_SKIP_FIREWALL 0x00004000 /* skip firewall processing */ #define M_FREELIST 0x00008000 /* mbuf is on the free list */ #define M_VLANTAG 0x00010000 /* ether_vtag is valid */ #define M_PROMISC 0x00020000 /* packet was not for us */ #define M_NOFREE 0x00040000 /* do not free mbuf, embedded in cluster */ #define M_PROTO6 0x00080000 /* protocol-specific */ #define M_PROTO7 0x00100000 /* protocol-specific */ #define M_PROTO8 0x00200000 /* protocol-specific */ -#define M_FLOWID 0x00400000 /* flowid is valid */ +#define M_FLOWID 0x00400000 /* deprecated: flowid is valid */ +#define M_HASHTYPEBITS 0x0F000000 /* mask of bits holding flowid hash type */ + /* * For RELENG_{6,7} steal these flags for limited multiple routing table * support. In RELENG_8 and beyond, use just one flag and a tag. */ #define M_FIB 0xF0000000 /* steal some bits to store fib number. */ #define M_NOTIFICATION M_PROTO5 /* SCTP notification */ /* * Flags to purge when crossing layers. */ #define M_PROTOFLAGS \ (M_PROTO1|M_PROTO2|M_PROTO3|M_PROTO4|M_PROTO5|M_PROTO6|M_PROTO7|M_PROTO8) /* + * Network interface cards are able to hash protocol fields (such as IPv4 + * addresses and TCP port numbers) classify packets into flows. These flows + * can then be used to maintain ordering while delivering packets to the OS + * via parallel input queues, as well as to provide a stateless affinity + * model. NIC drivers can pass up the hash via m->m_pkthdr.flowid, and set + * m_flag fields to indicate how the hash should be interpreted by the + * network stack. + * + * Most NICs support RSS, which provides ordering and explicit affinity, and + * use the hash m_flag bits to indicate what header fields were covered by + * the hash. M_HASHTYPE_OPAQUE can be set by non-RSS cards or configurations + * that provide an opaque flow identifier, allowing for ordering and + * distribution without explicit affinity. + */ +#define M_HASHTYPE_SHIFT 24 +#define M_HASHTYPE_NONE 0x0 +#define M_HASHTYPE_RSS_IPV4 0x1 /* IPv4 2-tuple */ +#define M_HASHTYPE_RSS_TCP_IPV4 0x2 /* TCPv4 4-tuple */ +#define M_HASHTYPE_RSS_IPV6 0x3 /* IPv6 2-tuple */ +#define M_HASHTYPE_RSS_TCP_IPV6 0x4 /* TCPv6 4-tuple */ +#define M_HASHTYPE_RSS_IPV6_EX 0x5 /* IPv6 2-tuple + ext hdrs */ +#define M_HASHTYPE_RSS_TCP_IPV6_EX 0x6 /* TCPv6 4-tiple + ext hdrs */ +#define M_HASHTYPE_OPAQUE 0xf /* ordering, not affinity */ + +#define M_HASHTYPE_CLEAR(m) (m)->m_flags &= ~(M_HASHTYPEBITS) +#define M_HASHTYPE_GET(m) (((m)->m_flags & M_HASHTYPEBITS) >> \ + M_HASHTYPE_SHIFT) +#define M_HASHTYPE_SET(m, v) do { \ + (m)->m_flags &= ~M_HASHTYPEBITS; \ + (m)->m_flags |= ((v) << M_HASHTYPE_SHIFT); \ +} while (0) +#define M_HASHTYPE_TEST(m, v) (M_HASHTYPE_GET(m) == (v)) + +/* * Flags preserved when copying m_pkthdr. */ #define M_COPYFLAGS \ (M_PKTHDR|M_EOR|M_RDONLY|M_PROTOFLAGS|M_SKIP_FIREWALL|M_BCAST|M_MCAST|\ - M_FRAG|M_FIRSTFRAG|M_LASTFRAG|M_VLANTAG|M_PROMISC|M_FIB) + M_FRAG|M_FIRSTFRAG|M_LASTFRAG|M_VLANTAG|M_PROMISC|M_FIB|M_HASHTYPEBITS) /* * External buffer types: identify ext_buf type. */ #define EXT_CLUSTER 1 /* mbuf cluster */ #define EXT_SFBUF 2 /* sendfile(2)'s sf_bufs */ #define EXT_JUMBOP 3 /* jumbo cluster 4096 bytes */ #define EXT_JUMBO9 4 /* jumbo cluster 9216 bytes */ #define EXT_JUMBO16 5 /* jumbo cluster 16184 bytes */ #define EXT_PACKET 6 /* mbuf+cluster from packet zone */ #define EXT_MBUF 7 /* external mbuf reference (M_IOVEC) */ #define EXT_NET_DRV 100 /* custom ext_buf provided by net driver(s) */ #define EXT_MOD_TYPE 200 /* custom module's ext_buf type */ #define EXT_DISPOSABLE 300 /* can throw this buffer away w/page flipping */ #define EXT_EXTREF 400 /* has externally maintained ref_cnt ptr */ /* * Flags indicating hw checksum support and sw checksum requirements. This * field can be directly tested against if_data.ifi_hwassist. */ #define CSUM_IP 0x0001 /* will csum IP */ #define CSUM_TCP 0x0002 /* will csum TCP */ #define CSUM_UDP 0x0004 /* will csum UDP */ #define CSUM_IP_FRAGS 0x0008 /* will csum IP fragments */ #define CSUM_FRAGMENT 0x0010 /* will do IP fragmentation */ #define CSUM_TSO 0x0020 /* will do TSO */ #define CSUM_SCTP 0x0040 /* will csum SCTP */ #define CSUM_IP_CHECKED 0x0100 /* did csum IP */ #define CSUM_IP_VALID 0x0200 /* ... the csum is valid */ #define CSUM_DATA_VALID 0x0400 /* csum_data field is valid */ #define CSUM_PSEUDO_HDR 0x0800 /* csum_data has pseudo hdr */ #define CSUM_SCTP_VALID 0x1000 /* SCTP checksum is valid */ #define CSUM_DELAY_DATA (CSUM_TCP | CSUM_UDP) #define CSUM_DELAY_IP (CSUM_IP) /* XXX add ipv6 here too? */ /* * mbuf types. */ #define MT_NOTMBUF 0 /* USED INTERNALLY ONLY! Object is not mbuf */ #define MT_DATA 1 /* dynamic (data) allocation */ #define MT_HEADER MT_DATA /* packet header, use M_PKTHDR instead */ #define MT_SONAME 8 /* socket name */ #define MT_CONTROL 14 /* extra-data protocol message */ #define MT_OOBDATA 15 /* expedited data */ #define MT_NTYPES 16 /* number of mbuf types for mbtypes[] */ #define MT_NOINIT 255 /* Not a type but a flag to allocate a non-initialized mbuf */ #define MB_NOTAGS 0x1UL /* no tags attached to mbuf */ /* * General mbuf allocator statistics structure. * * Many of these statistics are no longer used; we instead track many * allocator statistics through UMA's built in statistics mechanism. */ struct mbstat { u_long m_mbufs; /* XXX */ u_long m_mclusts; /* XXX */ u_long m_drain; /* times drained protocols for space */ u_long m_mcfail; /* XXX: times m_copym failed */ u_long m_mpfail; /* XXX: times m_pullup failed */ u_long m_msize; /* length of an mbuf */ u_long m_mclbytes; /* length of an mbuf cluster */ u_long m_minclsize; /* min length of data to allocate a cluster */ u_long m_mlen; /* length of data in an mbuf */ u_long m_mhlen; /* length of data in a header mbuf */ /* Number of mbtypes (gives # elems in mbtypes[] array) */ short m_numtypes; /* XXX: Sendfile stats should eventually move to their own struct */ u_long sf_iocnt; /* times sendfile had to do disk I/O */ u_long sf_allocfail; /* times sfbuf allocation failed */ u_long sf_allocwait; /* times sfbuf allocation had to wait */ }; /* * Flags specifying how an allocation should be made. * * The flag to use is as follows: * - M_DONTWAIT or M_NOWAIT from an interrupt handler to not block allocation. * - M_WAIT or M_WAITOK from wherever it is safe to block. * * M_DONTWAIT/M_NOWAIT means that we will not block the thread explicitly and * if we cannot allocate immediately we may return NULL, whereas * M_WAIT/M_WAITOK means that if we cannot allocate resources we * will block until they are available, and thus never return NULL. * * XXX Eventually just phase this out to use M_WAITOK/M_NOWAIT. */ #define MBTOM(how) (how) #define M_DONTWAIT M_NOWAIT #define M_TRYWAIT M_WAITOK #define M_WAIT M_WAITOK /* * String names of mbuf-related UMA(9) and malloc(9) types. Exposed to * !_KERNEL so that monitoring tools can look up the zones with * libmemstat(3). */ #define MBUF_MEM_NAME "mbuf" #define MBUF_CLUSTER_MEM_NAME "mbuf_cluster" #define MBUF_PACKET_MEM_NAME "mbuf_packet" #define MBUF_JUMBOP_MEM_NAME "mbuf_jumbo_page" #define MBUF_JUMBO9_MEM_NAME "mbuf_jumbo_9k" #define MBUF_JUMBO16_MEM_NAME "mbuf_jumbo_16k" #define MBUF_TAG_MEM_NAME "mbuf_tag" #define MBUF_EXTREFCNT_MEM_NAME "mbuf_ext_refcnt" #ifdef _KERNEL #ifdef WITNESS #define MBUF_CHECKSLEEP(how) do { \ if (how == M_WAITOK) \ WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL, \ "Sleeping in \"%s\"", __func__); \ } while (0) #else #define MBUF_CHECKSLEEP(how) #endif /* * Network buffer allocation API * * The rest of it is defined in kern/kern_mbuf.c */ extern uma_zone_t zone_mbuf; extern uma_zone_t zone_clust; extern uma_zone_t zone_pack; extern uma_zone_t zone_jumbop; extern uma_zone_t zone_jumbo9; extern uma_zone_t zone_jumbo16; extern uma_zone_t zone_ext_refcnt; static __inline struct mbuf *m_getcl(int how, short type, int flags); static __inline struct mbuf *m_get(int how, short type); static __inline struct mbuf *m_gethdr(int how, short type); static __inline struct mbuf *m_getjcl(int how, short type, int flags, int size); static __inline struct mbuf *m_getclr(int how, short type); /* XXX */ static __inline int m_init(struct mbuf *m, uma_zone_t zone, int size, int how, short type, int flags); static __inline struct mbuf *m_free(struct mbuf *m); static __inline void m_clget(struct mbuf *m, int how); static __inline void *m_cljget(struct mbuf *m, int how, int size); static __inline void m_chtype(struct mbuf *m, short new_type); void mb_free_ext(struct mbuf *); static __inline struct mbuf *m_last(struct mbuf *m); int m_pkthdr_init(struct mbuf *m, int how); static __inline int m_gettype(int size) { int type; switch (size) { case MSIZE: type = EXT_MBUF; break; case MCLBYTES: type = EXT_CLUSTER; break; #if MJUMPAGESIZE != MCLBYTES case MJUMPAGESIZE: type = EXT_JUMBOP; break; #endif case MJUM9BYTES: type = EXT_JUMBO9; break; case MJUM16BYTES: type = EXT_JUMBO16; break; default: panic("%s: m_getjcl: invalid cluster size", __func__); } return (type); } static __inline uma_zone_t m_getzone(int size) { uma_zone_t zone; switch (size) { case MSIZE: zone = zone_mbuf; break; case MCLBYTES: zone = zone_clust; break; #if MJUMPAGESIZE != MCLBYTES case MJUMPAGESIZE: zone = zone_jumbop; break; #endif case MJUM9BYTES: zone = zone_jumbo9; break; case MJUM16BYTES: zone = zone_jumbo16; break; default: panic("%s: m_getjcl: invalid cluster type", __func__); } return (zone); } /* * Initialize an mbuf with linear storage. * * Inline because the consumer text overhead will be roughly the same to * initialize or call a function with this many parameters and M_PKTHDR * should go away with constant propagation for !MGETHDR. */ static __inline int m_init(struct mbuf *m, uma_zone_t zone, int size, int how, short type, int flags) { int error; m->m_next = NULL; m->m_nextpkt = NULL; m->m_data = m->m_dat; m->m_len = 0; m->m_flags = flags; m->m_type = type; if (flags & M_PKTHDR) { if ((error = m_pkthdr_init(m, how)) != 0) return (error); } return (0); } static __inline struct mbuf * m_get(int how, short type) { struct mb_args args; args.flags = 0; args.type = type; return ((struct mbuf *)(uma_zalloc_arg(zone_mbuf, &args, how))); } /* * XXX This should be deprecated, very little use. */ static __inline struct mbuf * m_getclr(int how, short type) { struct mbuf *m; struct mb_args args; args.flags = 0; args.type = type; m = uma_zalloc_arg(zone_mbuf, &args, how); if (m != NULL) bzero(m->m_data, MLEN); return (m); } static __inline struct mbuf * m_gethdr(int how, short type) { struct mb_args args; args.flags = M_PKTHDR; args.type = type; return ((struct mbuf *)(uma_zalloc_arg(zone_mbuf, &args, how))); } static __inline struct mbuf * m_getcl(int how, short type, int flags) { struct mb_args args; args.flags = flags; args.type = type; return ((struct mbuf *)(uma_zalloc_arg(zone_pack, &args, how))); } /* * m_getjcl() returns an mbuf with a cluster of the specified size attached. * For size it takes MCLBYTES, MJUMPAGESIZE, MJUM9BYTES, MJUM16BYTES. * * XXX: This is rather large, should be real function maybe. */ static __inline struct mbuf * m_getjcl(int how, short type, int flags, int size) { struct mb_args args; struct mbuf *m, *n; uma_zone_t zone; if (size == MCLBYTES) return m_getcl(how, type, flags); args.flags = flags; args.type = type; m = uma_zalloc_arg(zone_mbuf, &args, how); if (m == NULL) return (NULL); zone = m_getzone(size); n = uma_zalloc_arg(zone, m, how); if (n == NULL) { uma_zfree(zone_mbuf, m); return (NULL); } return (m); } static __inline void m_free_fast(struct mbuf *m) { #ifdef INVARIANTS if (m->m_flags & M_PKTHDR) KASSERT(SLIST_EMPTY(&m->m_pkthdr.tags), ("doing fast free of mbuf with tags")); #endif uma_zfree_arg(zone_mbuf, m, (void *)MB_NOTAGS); } static __inline struct mbuf * m_free(struct mbuf *m) { struct mbuf *n = m->m_next; if (m->m_flags & M_EXT) mb_free_ext(m); else if ((m->m_flags & M_NOFREE) == 0) uma_zfree(zone_mbuf, m); return (n); } static __inline void m_clget(struct mbuf *m, int how) { if (m->m_flags & M_EXT) printf("%s: %p mbuf already has cluster\n", __func__, m); m->m_ext.ext_buf = (char *)NULL; uma_zalloc_arg(zone_clust, m, how); /* * On a cluster allocation failure, drain the packet zone and retry, * we might be able to loosen a few clusters up on the drain. */ if ((how & M_NOWAIT) && (m->m_ext.ext_buf == NULL)) { zone_drain(zone_pack); uma_zalloc_arg(zone_clust, m, how); } } /* * m_cljget() is different from m_clget() as it can allocate clusters without * attaching them to an mbuf. In that case the return value is the pointer * to the cluster of the requested size. If an mbuf was specified, it gets * the cluster attached to it and the return value can be safely ignored. * For size it takes MCLBYTES, MJUMPAGESIZE, MJUM9BYTES, MJUM16BYTES. */ static __inline void * m_cljget(struct mbuf *m, int how, int size) { uma_zone_t zone; if (m && m->m_flags & M_EXT) printf("%s: %p mbuf already has cluster\n", __func__, m); if (m != NULL) m->m_ext.ext_buf = NULL; zone = m_getzone(size); return (uma_zalloc_arg(zone, m, how)); } static __inline void m_cljset(struct mbuf *m, void *cl, int type) { uma_zone_t zone; int size; switch (type) { case EXT_CLUSTER: size = MCLBYTES; zone = zone_clust; break; #if MJUMPAGESIZE != MCLBYTES case EXT_JUMBOP: size = MJUMPAGESIZE; zone = zone_jumbop; break; #endif case EXT_JUMBO9: size = MJUM9BYTES; zone = zone_jumbo9; break; case EXT_JUMBO16: size = MJUM16BYTES; zone = zone_jumbo16; break; default: panic("unknown cluster type"); break; } m->m_data = m->m_ext.ext_buf = cl; m->m_ext.ext_free = m->m_ext.ext_arg1 = m->m_ext.ext_arg2 = NULL; m->m_ext.ext_size = size; m->m_ext.ext_type = type; m->m_ext.ref_cnt = uma_find_refcnt(zone, cl); m->m_flags |= M_EXT; } static __inline void m_chtype(struct mbuf *m, short new_type) { m->m_type = new_type; } static __inline struct mbuf * m_last(struct mbuf *m) { while (m->m_next) m = m->m_next; return (m); } /* * mbuf, cluster, and external object allocation macros (for compatibility * purposes). */ #define M_MOVE_PKTHDR(to, from) m_move_pkthdr((to), (from)) #define MGET(m, how, type) ((m) = m_get((how), (type))) #define MGETHDR(m, how, type) ((m) = m_gethdr((how), (type))) #define MCLGET(m, how) m_clget((m), (how)) #define MEXTADD(m, buf, size, free, arg1, arg2, flags, type) \ m_extadd((m), (caddr_t)(buf), (size), (free),(arg1),(arg2),(flags), (type)) #define m_getm(m, len, how, type) \ m_getm2((m), (len), (how), (type), M_PKTHDR) /* * Evaluate TRUE if it's safe to write to the mbuf m's data region (this can * be both the local data payload, or an external buffer area, depending on * whether M_EXT is set). */ #define M_WRITABLE(m) (!((m)->m_flags & M_RDONLY) && \ (!(((m)->m_flags & M_EXT)) || \ (*((m)->m_ext.ref_cnt) == 1)) ) \ /* Check if the supplied mbuf has a packet header, or else panic. */ #define M_ASSERTPKTHDR(m) \ KASSERT((m) != NULL && (m)->m_flags & M_PKTHDR, \ ("%s: no mbuf packet header!", __func__)) /* * Ensure that the supplied mbuf is a valid, non-free mbuf. * * XXX: Broken at the moment. Need some UMA magic to make it work again. */ #define M_ASSERTVALID(m) \ KASSERT((((struct mbuf *)m)->m_flags & 0) == 0, \ ("%s: attempted use of a free mbuf!", __func__)) /* * Set the m_data pointer of a newly-allocated mbuf (m_get/MGET) to place an * object of the specified size at the end of the mbuf, longword aligned. */ #define M_ALIGN(m, len) do { \ KASSERT(!((m)->m_flags & (M_PKTHDR|M_EXT)), \ ("%s: M_ALIGN not normal mbuf", __func__)); \ KASSERT((m)->m_data == (m)->m_dat, \ ("%s: M_ALIGN not a virgin mbuf", __func__)); \ (m)->m_data += (MLEN - (len)) & ~(sizeof(long) - 1); \ } while (0) /* * As above, for mbufs allocated with m_gethdr/MGETHDR or initialized by * M_DUP/MOVE_PKTHDR. */ #define MH_ALIGN(m, len) do { \ KASSERT((m)->m_flags & M_PKTHDR && !((m)->m_flags & M_EXT), \ ("%s: MH_ALIGN not PKTHDR mbuf", __func__)); \ KASSERT((m)->m_data == (m)->m_pktdat, \ ("%s: MH_ALIGN not a virgin mbuf", __func__)); \ (m)->m_data += (MHLEN - (len)) & ~(sizeof(long) - 1); \ } while (0) /* * Compute the amount of space available before the current start of data in * an mbuf. * * The M_WRITABLE() is a temporary, conservative safety measure: the burden * of checking writability of the mbuf data area rests solely with the caller. */ #define M_LEADINGSPACE(m) \ ((m)->m_flags & M_EXT ? \ (M_WRITABLE(m) ? (m)->m_data - (m)->m_ext.ext_buf : 0): \ (m)->m_flags & M_PKTHDR ? (m)->m_data - (m)->m_pktdat : \ (m)->m_data - (m)->m_dat) /* * Compute the amount of space available after the end of data in an mbuf. * * The M_WRITABLE() is a temporary, conservative safety measure: the burden * of checking writability of the mbuf data area rests solely with the caller. */ #define M_TRAILINGSPACE(m) \ ((m)->m_flags & M_EXT ? \ (M_WRITABLE(m) ? (m)->m_ext.ext_buf + (m)->m_ext.ext_size \ - ((m)->m_data + (m)->m_len) : 0) : \ &(m)->m_dat[MLEN] - ((m)->m_data + (m)->m_len)) /* * Arrange to prepend space of size plen to mbuf m. If a new mbuf must be * allocated, how specifies whether to wait. If the allocation fails, the * original mbuf chain is freed and m is set to NULL. */ #define M_PREPEND(m, plen, how) do { \ struct mbuf **_mmp = &(m); \ struct mbuf *_mm = *_mmp; \ int _mplen = (plen); \ int __mhow = (how); \ \ MBUF_CHECKSLEEP(how); \ if (M_LEADINGSPACE(_mm) >= _mplen) { \ _mm->m_data -= _mplen; \ _mm->m_len += _mplen; \ } else \ _mm = m_prepend(_mm, _mplen, __mhow); \ if (_mm != NULL && _mm->m_flags & M_PKTHDR) \ _mm->m_pkthdr.len += _mplen; \ *_mmp = _mm; \ } while (0) /* * Change mbuf to new type. This is a relatively expensive operation and * should be avoided. */ #define MCHTYPE(m, t) m_chtype((m), (t)) /* Length to m_copy to copy all. */ #define M_COPYALL 1000000000 /* Compatibility with 4.3. */ #define m_copy(m, o, l) m_copym((m), (o), (l), M_DONTWAIT) extern int max_datalen; /* MHLEN - max_hdr */ extern int max_hdr; /* Largest link + protocol header */ extern int max_linkhdr; /* Largest link-level header */ extern int max_protohdr; /* Largest protocol header */ extern struct mbstat mbstat; /* General mbuf stats/infos */ extern int nmbclusters; /* Maximum number of clusters */ struct uio; void m_adj(struct mbuf *, int); void m_align(struct mbuf *, int); int m_apply(struct mbuf *, int, int, int (*)(void *, void *, u_int), void *); int m_append(struct mbuf *, int, c_caddr_t); void m_cat(struct mbuf *, struct mbuf *); void m_extadd(struct mbuf *, caddr_t, u_int, void (*)(void *, void *), void *, void *, int, int); struct mbuf *m_collapse(struct mbuf *, int, int); void m_copyback(struct mbuf *, int, int, c_caddr_t); void m_copydata(const struct mbuf *, int, int, caddr_t); struct mbuf *m_copym(struct mbuf *, int, int, int); struct mbuf *m_copymdata(struct mbuf *, struct mbuf *, int, int, int, int); struct mbuf *m_copypacket(struct mbuf *, int); void m_copy_pkthdr(struct mbuf *, struct mbuf *); struct mbuf *m_copyup(struct mbuf *n, int len, int dstoff); struct mbuf *m_defrag(struct mbuf *, int); void m_demote(struct mbuf *, int); struct mbuf *m_devget(char *, int, int, struct ifnet *, void (*)(char *, caddr_t, u_int)); struct mbuf *m_dup(struct mbuf *, int); int m_dup_pkthdr(struct mbuf *, struct mbuf *, int); u_int m_fixhdr(struct mbuf *); struct mbuf *m_fragment(struct mbuf *, int, int); void m_freem(struct mbuf *); struct mbuf *m_getm2(struct mbuf *, int, int, short, int); struct mbuf *m_getptr(struct mbuf *, int, int *); u_int m_length(struct mbuf *, struct mbuf **); int m_mbuftouio(struct uio *, struct mbuf *, int); void m_move_pkthdr(struct mbuf *, struct mbuf *); struct mbuf *m_prepend(struct mbuf *, int, int); void m_print(const struct mbuf *, int); struct mbuf *m_pulldown(struct mbuf *, int, int, int *); struct mbuf *m_pullup(struct mbuf *, int); int m_sanity(struct mbuf *, int); struct mbuf *m_split(struct mbuf *, int, int); struct mbuf *m_uiotombuf(struct uio *, int, int, int, int); struct mbuf *m_unshare(struct mbuf *, int how); /*- * Network packets may have annotations attached by affixing a list of * "packet tags" to the pkthdr structure. Packet tags are dynamically * allocated semi-opaque data structures that have a fixed header * (struct m_tag) that specifies the size of the memory block and a * pair that identifies it. The cookie is a 32-bit unique * unsigned value used to identify a module or ABI. By convention this value * is chosen as the date+time that the module is created, expressed as the * number of seconds since the epoch (e.g., using date -u +'%s'). The type * value is an ABI/module-specific value that identifies a particular * annotation and is private to the module. For compatibility with systems * like OpenBSD that define packet tags w/o an ABI/module cookie, the value * PACKET_ABI_COMPAT is used to implement m_tag_get and m_tag_find * compatibility shim functions and several tag types are defined below. * Users that do not require compatibility should use a private cookie value * so that packet tag-related definitions can be maintained privately. * * Note that the packet tag returned by m_tag_alloc has the default memory * alignment implemented by malloc. To reference private data one can use a * construct like: * * struct m_tag *mtag = m_tag_alloc(...); * struct foo *p = (struct foo *)(mtag+1); * * if the alignment of struct m_tag is sufficient for referencing members of * struct foo. Otherwise it is necessary to embed struct m_tag within the * private data structure to insure proper alignment; e.g., * * struct foo { * struct m_tag tag; * ... * }; * struct foo *p = (struct foo *) m_tag_alloc(...); * struct m_tag *mtag = &p->tag; */ /* * Persistent tags stay with an mbuf until the mbuf is reclaimed. Otherwise * tags are expected to ``vanish'' when they pass through a network * interface. For most interfaces this happens normally as the tags are * reclaimed when the mbuf is free'd. However in some special cases * reclaiming must be done manually. An example is packets that pass through * the loopback interface. Also, one must be careful to do this when * ``turning around'' packets (e.g., icmp_reflect). * * To mark a tag persistent bit-or this flag in when defining the tag id. * The tag will then be treated as described above. */ #define MTAG_PERSISTENT 0x800 #define PACKET_TAG_NONE 0 /* Nadda */ /* Packet tags for use with PACKET_ABI_COMPAT. */ #define PACKET_TAG_IPSEC_IN_DONE 1 /* IPsec applied, in */ #define PACKET_TAG_IPSEC_OUT_DONE 2 /* IPsec applied, out */ #define PACKET_TAG_IPSEC_IN_CRYPTO_DONE 3 /* NIC IPsec crypto done */ #define PACKET_TAG_IPSEC_OUT_CRYPTO_NEEDED 4 /* NIC IPsec crypto req'ed */ #define PACKET_TAG_IPSEC_IN_COULD_DO_CRYPTO 5 /* NIC notifies IPsec */ #define PACKET_TAG_IPSEC_PENDING_TDB 6 /* Reminder to do IPsec */ #define PACKET_TAG_BRIDGE 7 /* Bridge processing done */ #define PACKET_TAG_GIF 8 /* GIF processing done */ #define PACKET_TAG_GRE 9 /* GRE processing done */ #define PACKET_TAG_IN_PACKET_CHECKSUM 10 /* NIC checksumming done */ #define PACKET_TAG_ENCAP 11 /* Encap. processing */ #define PACKET_TAG_IPSEC_SOCKET 12 /* IPSEC socket ref */ #define PACKET_TAG_IPSEC_HISTORY 13 /* IPSEC history */ #define PACKET_TAG_IPV6_INPUT 14 /* IPV6 input processing */ #define PACKET_TAG_DUMMYNET 15 /* dummynet info */ #define PACKET_TAG_DIVERT 17 /* divert info */ #define PACKET_TAG_IPFORWARD 18 /* ipforward info */ #define PACKET_TAG_MACLABEL (19 | MTAG_PERSISTENT) /* MAC label */ #define PACKET_TAG_PF 21 /* PF + ALTQ information */ #define PACKET_TAG_RTSOCKFAM 25 /* rtsock sa family */ #define PACKET_TAG_IPOPTIONS 27 /* Saved IP options */ #define PACKET_TAG_CARP 28 /* CARP info */ #define PACKET_TAG_IPSEC_NAT_T_PORTS 29 /* two uint16_t */ #define PACKET_TAG_ND_OUTGOING 30 /* ND outgoing */ /* Specific cookies and tags. */ /* Packet tag routines. */ struct m_tag *m_tag_alloc(u_int32_t, int, int, int); void m_tag_delete(struct mbuf *, struct m_tag *); void m_tag_delete_chain(struct mbuf *, struct m_tag *); void m_tag_free_default(struct m_tag *); struct m_tag *m_tag_locate(struct mbuf *, u_int32_t, int, struct m_tag *); struct m_tag *m_tag_copy(struct m_tag *, int); int m_tag_copy_chain(struct mbuf *, struct mbuf *, int); void m_tag_delete_nonpersistent(struct mbuf *); /* * Initialize the list of tags associated with an mbuf. */ static __inline void m_tag_init(struct mbuf *m) { SLIST_INIT(&m->m_pkthdr.tags); } /* * Set up the contents of a tag. Note that this does not fill in the free * method; the caller is expected to do that. * * XXX probably should be called m_tag_init, but that was already taken. */ static __inline void m_tag_setup(struct m_tag *t, u_int32_t cookie, int type, int len) { t->m_tag_id = type; t->m_tag_len = len; t->m_tag_cookie = cookie; } /* * Reclaim resources associated with a tag. */ static __inline void m_tag_free(struct m_tag *t) { (*t->m_tag_free)(t); } /* * Return the first tag associated with an mbuf. */ static __inline struct m_tag * m_tag_first(struct mbuf *m) { return (SLIST_FIRST(&m->m_pkthdr.tags)); } /* * Return the next tag in the list of tags associated with an mbuf. */ static __inline struct m_tag * m_tag_next(struct mbuf *m, struct m_tag *t) { return (SLIST_NEXT(t, m_tag_link)); } /* * Prepend a tag to the list of tags associated with an mbuf. */ static __inline void m_tag_prepend(struct mbuf *m, struct m_tag *t) { SLIST_INSERT_HEAD(&m->m_pkthdr.tags, t, m_tag_link); } /* * Unlink a tag from the list of tags associated with an mbuf. */ static __inline void m_tag_unlink(struct mbuf *m, struct m_tag *t) { SLIST_REMOVE(&m->m_pkthdr.tags, t, m_tag, m_tag_link); } /* These are for OpenBSD compatibility. */ #define MTAG_ABI_COMPAT 0 /* compatibility ABI */ static __inline struct m_tag * m_tag_get(int type, int length, int wait) { return (m_tag_alloc(MTAG_ABI_COMPAT, type, length, wait)); } static __inline struct m_tag * m_tag_find(struct mbuf *m, int type, struct m_tag *start) { return (SLIST_EMPTY(&m->m_pkthdr.tags) ? (struct m_tag *)NULL : m_tag_locate(m, MTAG_ABI_COMPAT, type, start)); } /* XXX temporary FIB methods probably eventually use tags.*/ #define M_FIBSHIFT 28 #define M_FIBMASK 0x0F /* get the fib from an mbuf and if it is not set, return the default */ #define M_GETFIB(_m) \ ((((_m)->m_flags & M_FIB) >> M_FIBSHIFT) & M_FIBMASK) #define M_SETFIB(_m, _fib) do { \ _m->m_flags &= ~M_FIB; \ _m->m_flags |= (((_fib) << M_FIBSHIFT) & M_FIB); \ } while (0) #endif /* _KERNEL */ #ifdef MBUF_PROFILING void m_profile(struct mbuf *m); #define M_PROFILE(m) m_profile(m) #else #define M_PROFILE(m) #endif #endif /* !_SYS_MBUF_H_ */ Index: projects/largeSMP/sys/sys/soundcard.h =================================================================== --- projects/largeSMP/sys/sys/soundcard.h (revision 222782) +++ projects/largeSMP/sys/sys/soundcard.h (revision 222783) @@ -1,2003 +1,2004 @@ /* * soundcard.h */ /*- * Copyright by Hannu Savolainen 1993 / 4Front Technologies 1993-2006 * Modified for the new FreeBSD sound driver by Luigi Rizzo, 1997 * * 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. * * $FreeBSD$ */ /* * Unless coordinating changes with 4Front Technologies, do NOT make any * modifications to ioctl commands, types, etc. that would break * compatibility with the OSS API. */ #ifndef _SYS_SOUNDCARD_H_ #define _SYS_SOUNDCARD_H_ /* * If you make modifications to this file, please contact me before * distributing the modified version. There is already enough * diversity in the world. * * Regards, * Hannu Savolainen * hannu@voxware.pp.fi * ********************************************************************** * PS. The Hacker's Guide to VoxWare available from * nic.funet.fi:pub/Linux/ALPHA/sound. The file is * snd-sdk-doc-0.1.ps.gz (gzipped postscript). It contains * some useful information about programming with VoxWare. * (NOTE! The pub/Linux/ALPHA/ directories are hidden. You have * to cd inside them before the files are accessible.) ********************************************************************** */ /* * SOUND_VERSION is only used by the voxware driver. Hopefully apps * should not depend on it, but rather look at the capabilities * of the driver in the kernel! */ #define SOUND_VERSION 301 #define VOXWARE /* does this have any use ? */ /* * Supported card ID numbers (Should be somewhere else? We keep * them here just for compativility with the old driver, but these * constants are of little or no use). */ #define SNDCARD_ADLIB 1 #define SNDCARD_SB 2 #define SNDCARD_PAS 3 #define SNDCARD_GUS 4 #define SNDCARD_MPU401 5 #define SNDCARD_SB16 6 #define SNDCARD_SB16MIDI 7 #define SNDCARD_UART6850 8 #define SNDCARD_GUS16 9 #define SNDCARD_MSS 10 #define SNDCARD_PSS 11 #define SNDCARD_SSCAPE 12 #define SNDCARD_PSS_MPU 13 #define SNDCARD_PSS_MSS 14 #define SNDCARD_SSCAPE_MSS 15 #define SNDCARD_TRXPRO 16 #define SNDCARD_TRXPRO_SB 17 #define SNDCARD_TRXPRO_MPU 18 #define SNDCARD_MAD16 19 #define SNDCARD_MAD16_MPU 20 #define SNDCARD_CS4232 21 #define SNDCARD_CS4232_MPU 22 #define SNDCARD_MAUI 23 #define SNDCARD_PSEUDO_MSS 24 #define SNDCARD_AWE32 25 #define SNDCARD_NSS 26 #define SNDCARD_UART16550 27 #define SNDCARD_OPL 28 #include #include #ifndef _IOWR #include #endif /* !_IOWR */ /* * The first part of this file contains the new FreeBSD sound ioctl * interface. Tries to minimize the number of different ioctls, and * to be reasonably general. * * 970821: some of the new calls have not been implemented yet. */ /* * the following three calls extend the generic file descriptor * interface. AIONWRITE is the dual of FIONREAD, i.e. returns the max * number of bytes for a write operation to be non-blocking. * * AIOGSIZE/AIOSSIZE are used to change the behaviour of the device, * from a character device (default) to a block device. In block mode, * (not to be confused with blocking mode) the main difference for the * application is that select() will return only when a complete * block can be read/written to the device, whereas in character mode * select will return true when one byte can be exchanged. For audio * devices, character mode makes select almost useless since one byte * will always be ready by the next sample time (which is often only a * handful of microseconds away). * Use a size of 0 or 1 to return to character mode. */ #define AIONWRITE _IOR('A', 10, int) /* get # bytes to write */ struct snd_size { int play_size; int rec_size; }; #define AIOGSIZE _IOR('A', 11, struct snd_size)/* read current blocksize */ #define AIOSSIZE _IOWR('A', 11, struct snd_size) /* sets blocksize */ /* * The following constants define supported audio formats. The * encoding follows voxware conventions, i.e. 1 bit for each supported * format. We extend it by using bit 31 (RO) to indicate full-duplex * capability, and bit 29 (RO) to indicate that the card supports/ * needs different formats on capture & playback channels. * Bit 29 (RW) is used to indicate/ask stereo. * * The number of bits required to store the sample is: * o 4 bits for the IDA ADPCM format, * o 8 bits for 8-bit formats, mu-law and A-law, * o 16 bits for the 16-bit formats, and * o 32 bits for the 24/32-bit formats. * o undefined for the MPEG audio format. */ #define AFMT_QUERY 0x00000000 /* Return current format */ #define AFMT_MU_LAW 0x00000001 /* Logarithmic mu-law */ #define AFMT_A_LAW 0x00000002 /* Logarithmic A-law */ #define AFMT_IMA_ADPCM 0x00000004 /* A 4:1 compressed format where 16-bit * squence represented using the * the average 4 bits per sample */ #define AFMT_U8 0x00000008 /* Unsigned 8-bit */ #define AFMT_S16_LE 0x00000010 /* Little endian signed 16-bit */ #define AFMT_S16_BE 0x00000020 /* Big endian signed 16-bit */ #define AFMT_S8 0x00000040 /* Signed 8-bit */ #define AFMT_U16_LE 0x00000080 /* Little endian unsigned 16-bit */ #define AFMT_U16_BE 0x00000100 /* Big endian unsigned 16-bit */ #define AFMT_MPEG 0x00000200 /* MPEG MP2/MP3 audio */ #define AFMT_AC3 0x00000400 /* Dolby Digital AC3 */ /* * 32-bit formats below used for 24-bit audio data where the data is stored * in the 24 most significant bits and the least significant bits are not used * (should be set to 0). */ #define AFMT_S32_LE 0x00001000 /* Little endian signed 32-bit */ #define AFMT_S32_BE 0x00002000 /* Big endian signed 32-bit */ #define AFMT_U32_LE 0x00004000 /* Little endian unsigned 32-bit */ #define AFMT_U32_BE 0x00008000 /* Big endian unsigned 32-bit */ #define AFMT_S24_LE 0x00010000 /* Little endian signed 24-bit */ #define AFMT_S24_BE 0x00020000 /* Big endian signed 24-bit */ #define AFMT_U24_LE 0x00040000 /* Little endian unsigned 24-bit */ #define AFMT_U24_BE 0x00080000 /* Big endian unsigned 24-bit */ /* Machine dependant AFMT_* definitions. */ #if BYTE_ORDER == LITTLE_ENDIAN #define AFMT_S16_NE AFMT_S16_LE #define AFMT_S24_NE AFMT_S24_LE #define AFMT_S32_NE AFMT_S32_LE #define AFMT_U16_NE AFMT_U16_LE #define AFMT_U24_NE AFMT_U24_LE #define AFMT_U32_NE AFMT_U32_LE #define AFMT_S16_OE AFMT_S16_BE #define AFMT_S24_OE AFMT_S24_BE #define AFMT_S32_OE AFMT_S32_BE #define AFMT_U16_OE AFMT_U16_BE #define AFMT_U24_OE AFMT_U24_BE #define AFMT_U32_OE AFMT_U32_BE #else #define AFMT_S16_OE AFMT_S16_LE #define AFMT_S24_OE AFMT_S24_LE #define AFMT_S32_OE AFMT_S32_LE #define AFMT_U16_OE AFMT_U16_LE #define AFMT_U24_OE AFMT_U24_LE #define AFMT_U32_OE AFMT_U32_LE #define AFMT_S16_NE AFMT_S16_BE #define AFMT_S24_NE AFMT_S24_BE #define AFMT_S32_NE AFMT_S32_BE #define AFMT_U16_NE AFMT_U16_BE #define AFMT_U24_NE AFMT_U24_BE #define AFMT_U32_NE AFMT_U32_BE #endif #define AFMT_STEREO 0x10000000 /* can do/want stereo */ /* * the following are really capabilities */ #define AFMT_WEIRD 0x20000000 /* weird hardware... */ /* * AFMT_WEIRD reports that the hardware might need to operate * with different formats in the playback and capture * channels when operating in full duplex. * As an example, SoundBlaster16 cards only support U8 in one * direction and S16 in the other one, and applications should * be aware of this limitation. */ #define AFMT_FULLDUPLEX 0x80000000 /* can do full duplex */ /* * The following structure is used to get/set format and sampling rate. * While it would be better to have things such as stereo, bits per * sample, endiannes, etc split in different variables, it turns out * that formats are not that many, and not all combinations are possible. * So we followed the Voxware approach of associating one bit to each * format. */ typedef struct _snd_chan_param { u_long play_rate; /* sampling rate */ u_long rec_rate; /* sampling rate */ u_long play_format; /* everything describing the format */ u_long rec_format; /* everything describing the format */ } snd_chan_param; #define AIOGFMT _IOR('f', 12, snd_chan_param) /* get format */ #define AIOSFMT _IOWR('f', 12, snd_chan_param) /* sets format */ /* * The following structure is used to get/set the mixer setting. * Up to 32 mixers are supported, each one with up to 32 channels. */ typedef struct _snd_mix_param { u_char subdev; /* which output */ u_char line; /* which input */ u_char left,right; /* volumes, 0..255, 0 = mute */ } snd_mix_param ; /* XXX AIOGMIX, AIOSMIX not implemented yet */ #define AIOGMIX _IOWR('A', 13, snd_mix_param) /* return mixer status */ #define AIOSMIX _IOWR('A', 14, snd_mix_param) /* sets mixer status */ /* * channel specifiers used in AIOSTOP and AIOSYNC */ #define AIOSYNC_PLAY 0x1 /* play chan */ #define AIOSYNC_CAPTURE 0x2 /* capture chan */ /* AIOSTOP stop & flush a channel, returns the residual count */ #define AIOSTOP _IOWR ('A', 15, int) /* alternate method used to notify the sync condition */ #define AIOSYNC_SIGNAL 0x100 #define AIOSYNC_SELECT 0x200 /* what the 'pos' field refers to */ #define AIOSYNC_READY 0x400 #define AIOSYNC_FREE 0x800 typedef struct _snd_sync_parm { long chan ; /* play or capture channel, plus modifier */ long pos; } snd_sync_parm; #define AIOSYNC _IOWR ('A', 15, snd_sync_parm) /* misc. synchronization */ /* * The following is used to return device capabilities. If the structure * passed to the ioctl is zeroed, default values are returned for rate * and formats, a bitmap of available mixers is returned, and values * (inputs, different levels) for the first one are returned. * * If formats, mixers, inputs are instantiated, then detailed info * are returned depending on the call. */ typedef struct _snd_capabilities { u_long rate_min, rate_max; /* min-max sampling rate */ u_long formats; u_long bufsize; /* DMA buffer size */ u_long mixers; /* bitmap of available mixers */ u_long inputs; /* bitmap of available inputs (per mixer) */ u_short left, right; /* how many levels are supported */ } snd_capabilities; #define AIOGCAP _IOWR('A', 15, snd_capabilities) /* get capabilities */ /* * here is the old (Voxware) ioctl interface */ /* * IOCTL Commands for /dev/sequencer */ -#define SNDCTL_SEQ_RESET _IO ('Q', 0) +#define SNDCTL_SEQ_HALT _IO ('Q', 0) +#define SNDCTL_SEQ_RESET SNDCTL_SEQ_HALT /* Historic interface */ #define SNDCTL_SEQ_SYNC _IO ('Q', 1) #define SNDCTL_SYNTH_INFO _IOWR('Q', 2, struct synth_info) #define SNDCTL_SEQ_CTRLRATE _IOWR('Q', 3, int) /* Set/get timer res.(hz) */ #define SNDCTL_SEQ_GETOUTCOUNT _IOR ('Q', 4, int) #define SNDCTL_SEQ_GETINCOUNT _IOR ('Q', 5, int) #define SNDCTL_SEQ_PERCMODE _IOW ('Q', 6, int) #define SNDCTL_FM_LOAD_INSTR _IOW ('Q', 7, struct sbi_instrument) /* Valid for FM only */ #define SNDCTL_SEQ_TESTMIDI _IOW ('Q', 8, int) #define SNDCTL_SEQ_RESETSAMPLES _IOW ('Q', 9, int) #define SNDCTL_SEQ_NRSYNTHS _IOR ('Q',10, int) #define SNDCTL_SEQ_NRMIDIS _IOR ('Q',11, int) #define SNDCTL_MIDI_INFO _IOWR('Q',12, struct midi_info) #define SNDCTL_SEQ_THRESHOLD _IOW ('Q',13, int) #define SNDCTL_SEQ_TRESHOLD SNDCTL_SEQ_THRESHOLD /* there was once a typo */ #define SNDCTL_SYNTH_MEMAVL _IOWR('Q',14, int) /* in=dev#, out=memsize */ #define SNDCTL_FM_4OP_ENABLE _IOW ('Q',15, int) /* in=dev# */ #define SNDCTL_PMGR_ACCESS _IOWR('Q',16, struct patmgr_info) #define SNDCTL_SEQ_PANIC _IO ('Q',17) #define SNDCTL_SEQ_OUTOFBAND _IOW ('Q',18, struct seq_event_rec) #define SNDCTL_SEQ_GETTIME _IOR ('Q',19, int) struct seq_event_rec { u_char arr[8]; }; #define SNDCTL_TMR_TIMEBASE _IOWR('T', 1, int) #define SNDCTL_TMR_START _IO ('T', 2) #define SNDCTL_TMR_STOP _IO ('T', 3) #define SNDCTL_TMR_CONTINUE _IO ('T', 4) #define SNDCTL_TMR_TEMPO _IOWR('T', 5, int) #define SNDCTL_TMR_SOURCE _IOWR('T', 6, int) # define TMR_INTERNAL 0x00000001 # define TMR_EXTERNAL 0x00000002 # define TMR_MODE_MIDI 0x00000010 # define TMR_MODE_FSK 0x00000020 # define TMR_MODE_CLS 0x00000040 # define TMR_MODE_SMPTE 0x00000080 #define SNDCTL_TMR_METRONOME _IOW ('T', 7, int) #define SNDCTL_TMR_SELECT _IOW ('T', 8, int) /* * Endian aware patch key generation algorithm. */ #if defined(_AIX) || defined(AIX) # define _PATCHKEY(id) (0xfd00|id) #else # define _PATCHKEY(id) ((id<<8)|0xfd) #endif /* * Sample loading mechanism for internal synthesizers (/dev/sequencer) * The following patch_info structure has been designed to support * Gravis UltraSound. It tries to be universal format for uploading * sample based patches but is probably too limited. */ struct patch_info { /* u_short key; Use GUS_PATCH here */ short key; /* Use GUS_PATCH here */ #define GUS_PATCH _PATCHKEY(0x04) #define OBSOLETE_GUS_PATCH _PATCHKEY(0x02) short device_no; /* Synthesizer number */ short instr_no; /* Midi pgm# */ u_long mode; /* * The least significant byte has the same format than the GUS .PAT * files */ #define WAVE_16_BITS 0x01 /* bit 0 = 8 or 16 bit wave data. */ #define WAVE_UNSIGNED 0x02 /* bit 1 = Signed - Unsigned data. */ #define WAVE_LOOPING 0x04 /* bit 2 = looping enabled-1. */ #define WAVE_BIDIR_LOOP 0x08 /* bit 3 = Set is bidirectional looping. */ #define WAVE_LOOP_BACK 0x10 /* bit 4 = Set is looping backward. */ #define WAVE_SUSTAIN_ON 0x20 /* bit 5 = Turn sustaining on. (Env. pts. 3)*/ #define WAVE_ENVELOPES 0x40 /* bit 6 = Enable envelopes - 1 */ /* (use the env_rate/env_offs fields). */ /* Linux specific bits */ #define WAVE_VIBRATO 0x00010000 /* The vibrato info is valid */ #define WAVE_TREMOLO 0x00020000 /* The tremolo info is valid */ #define WAVE_SCALE 0x00040000 /* The scaling info is valid */ /* Other bits must be zeroed */ long len; /* Size of the wave data in bytes */ long loop_start, loop_end; /* Byte offsets from the beginning */ /* * The base_freq and base_note fields are used when computing the * playback speed for a note. The base_note defines the tone frequency * which is heard if the sample is played using the base_freq as the * playback speed. * * The low_note and high_note fields define the minimum and maximum note * frequencies for which this sample is valid. It is possible to define * more than one samples for an instrument number at the same time. The * low_note and high_note fields are used to select the most suitable one. * * The fields base_note, high_note and low_note should contain * the note frequency multiplied by 1000. For example value for the * middle A is 440*1000. */ u_int base_freq; u_long base_note; u_long high_note; u_long low_note; int panning; /* -128=left, 127=right */ int detuning; /* New fields introduced in version 1.99.5 */ /* Envelope. Enabled by mode bit WAVE_ENVELOPES */ u_char env_rate[ 6 ]; /* GUS HW ramping rate */ u_char env_offset[ 6 ]; /* 255 == 100% */ /* * The tremolo, vibrato and scale info are not supported yet. * Enable by setting the mode bits WAVE_TREMOLO, WAVE_VIBRATO or * WAVE_SCALE */ u_char tremolo_sweep; u_char tremolo_rate; u_char tremolo_depth; u_char vibrato_sweep; u_char vibrato_rate; u_char vibrato_depth; int scale_frequency; u_int scale_factor; /* from 0 to 2048 or 0 to 2 */ int volume; int spare[4]; char data[1]; /* The waveform data starts here */ }; struct sysex_info { short key; /* Use GUS_PATCH here */ #define SYSEX_PATCH _PATCHKEY(0x05) #define MAUI_PATCH _PATCHKEY(0x06) short device_no; /* Synthesizer number */ long len; /* Size of the sysex data in bytes */ u_char data[1]; /* Sysex data starts here */ }; /* * Patch management interface (/dev/sequencer, /dev/patmgr#) * Don't use these calls if you want to maintain compatibility with * the future versions of the driver. */ #define PS_NO_PATCHES 0 /* No patch support on device */ #define PS_MGR_NOT_OK 1 /* Plain patch support (no mgr) */ #define PS_MGR_OK 2 /* Patch manager supported */ #define PS_MANAGED 3 /* Patch manager running */ #define SNDCTL_PMGR_IFACE _IOWR('P', 1, struct patmgr_info) /* * The patmgr_info is a fixed size structure which is used for two * different purposes. The intended use is for communication between * the application using /dev/sequencer and the patch manager daemon * associated with a synthesizer device (ioctl(SNDCTL_PMGR_ACCESS)). * * This structure is also used with ioctl(SNDCTL_PGMR_IFACE) which allows * a patch manager daemon to read and write device parameters. This * ioctl available through /dev/sequencer also. Avoid using it since it's * extremely hardware dependent. In addition access trough /dev/sequencer * may confuse the patch manager daemon. */ struct patmgr_info { /* Note! size must be < 4k since kmalloc() is used */ u_long key; /* Don't worry. Reserved for communication between the patch manager and the driver. */ #define PM_K_EVENT 1 /* Event from the /dev/sequencer driver */ #define PM_K_COMMAND 2 /* Request from an application */ #define PM_K_RESPONSE 3 /* From patmgr to application */ #define PM_ERROR 4 /* Error returned by the patmgr */ int device; int command; /* * Commands 0x000 to 0xfff reserved for patch manager programs */ #define PM_GET_DEVTYPE 1 /* Returns type of the patch mgr interface of dev */ #define PMTYPE_FM2 1 /* 2 OP fm */ #define PMTYPE_FM4 2 /* Mixed 4 or 2 op FM (OPL-3) */ #define PMTYPE_WAVE 3 /* Wave table synthesizer (GUS) */ #define PM_GET_NRPGM 2 /* Returns max # of midi programs in parm1 */ #define PM_GET_PGMMAP 3 /* Returns map of loaded midi programs in data8 */ #define PM_GET_PGM_PATCHES 4 /* Return list of patches of a program (parm1) */ #define PM_GET_PATCH 5 /* Return patch header of patch parm1 */ #define PM_SET_PATCH 6 /* Set patch header of patch parm1 */ #define PM_READ_PATCH 7 /* Read patch (wave) data */ #define PM_WRITE_PATCH 8 /* Write patch (wave) data */ /* * Commands 0x1000 to 0xffff are for communication between the patch manager * and the client */ #define _PM_LOAD_PATCH 0x100 /* * Commands above 0xffff reserved for device specific use */ long parm1; long parm2; long parm3; union { u_char data8[4000]; u_short data16[2000]; u_long data32[1000]; struct patch_info patch; } data; }; /* * When a patch manager daemon is present, it will be informed by the * driver when something important happens. For example when the * /dev/sequencer is opened or closed. A record with key == PM_K_EVENT is * returned. The command field contains the event type: */ #define PM_E_OPENED 1 /* /dev/sequencer opened */ #define PM_E_CLOSED 2 /* /dev/sequencer closed */ #define PM_E_PATCH_RESET 3 /* SNDCTL_RESETSAMPLES called */ #define PM_E_PATCH_LOADED 4 /* A patch has been loaded by appl */ /* * /dev/sequencer input events. * * The data written to the /dev/sequencer is a stream of events. Events * are records of 4 or 8 bytes. The first byte defines the size. * Any number of events can be written with a write call. There * is a set of macros for sending these events. Use these macros if you * want to maximize portability of your program. * * Events SEQ_WAIT, SEQ_MIDIPUTC and SEQ_ECHO. Are also input events. * (All input events are currently 4 bytes long. Be prepared to support * 8 byte events also. If you receive any event having first byte >= 128, * it's a 8 byte event. * * The events are documented at the end of this file. * * Normal events (4 bytes) * There is also a 8 byte version of most of the 4 byte events. The * 8 byte one is recommended. */ #define SEQ_NOTEOFF 0 #define SEQ_FMNOTEOFF SEQ_NOTEOFF /* Just old name */ #define SEQ_NOTEON 1 #define SEQ_FMNOTEON SEQ_NOTEON #define SEQ_WAIT TMR_WAIT_ABS #define SEQ_PGMCHANGE 3 #define SEQ_FMPGMCHANGE SEQ_PGMCHANGE #define SEQ_SYNCTIMER TMR_START #define SEQ_MIDIPUTC 5 #define SEQ_DRUMON 6 /*** OBSOLETE ***/ #define SEQ_DRUMOFF 7 /*** OBSOLETE ***/ #define SEQ_ECHO TMR_ECHO /* For synching programs with output */ #define SEQ_AFTERTOUCH 9 #define SEQ_CONTROLLER 10 /* * Midi controller numbers * * Controllers 0 to 31 (0x00 to 0x1f) and 32 to 63 (0x20 to 0x3f) * are continuous controllers. * In the MIDI 1.0 these controllers are sent using two messages. * Controller numbers 0 to 31 are used to send the MSB and the * controller numbers 32 to 63 are for the LSB. Note that just 7 bits * are used in MIDI bytes. */ #define CTL_BANK_SELECT 0x00 #define CTL_MODWHEEL 0x01 #define CTL_BREATH 0x02 /* undefined 0x03 */ #define CTL_FOOT 0x04 #define CTL_PORTAMENTO_TIME 0x05 #define CTL_DATA_ENTRY 0x06 #define CTL_MAIN_VOLUME 0x07 #define CTL_BALANCE 0x08 /* undefined 0x09 */ #define CTL_PAN 0x0a #define CTL_EXPRESSION 0x0b /* undefined 0x0c - 0x0f */ #define CTL_GENERAL_PURPOSE1 0x10 #define CTL_GENERAL_PURPOSE2 0x11 #define CTL_GENERAL_PURPOSE3 0x12 #define CTL_GENERAL_PURPOSE4 0x13 /* undefined 0x14 - 0x1f */ /* undefined 0x20 */ /* * The controller numbers 0x21 to 0x3f are reserved for the * least significant bytes of the controllers 0x00 to 0x1f. * These controllers are not recognised by the driver. * * Controllers 64 to 69 (0x40 to 0x45) are on/off switches. * 0=OFF and 127=ON (intermediate values are possible) */ #define CTL_DAMPER_PEDAL 0x40 #define CTL_SUSTAIN CTL_DAMPER_PEDAL /* Alias */ #define CTL_HOLD CTL_DAMPER_PEDAL /* Alias */ #define CTL_PORTAMENTO 0x41 #define CTL_SOSTENUTO 0x42 #define CTL_SOFT_PEDAL 0x43 /* undefined 0x44 */ #define CTL_HOLD2 0x45 /* undefined 0x46 - 0x4f */ #define CTL_GENERAL_PURPOSE5 0x50 #define CTL_GENERAL_PURPOSE6 0x51 #define CTL_GENERAL_PURPOSE7 0x52 #define CTL_GENERAL_PURPOSE8 0x53 /* undefined 0x54 - 0x5a */ #define CTL_EXT_EFF_DEPTH 0x5b #define CTL_TREMOLO_DEPTH 0x5c #define CTL_CHORUS_DEPTH 0x5d #define CTL_DETUNE_DEPTH 0x5e #define CTL_CELESTE_DEPTH CTL_DETUNE_DEPTH /* Alias for the above one */ #define CTL_PHASER_DEPTH 0x5f #define CTL_DATA_INCREMENT 0x60 #define CTL_DATA_DECREMENT 0x61 #define CTL_NONREG_PARM_NUM_LSB 0x62 #define CTL_NONREG_PARM_NUM_MSB 0x63 #define CTL_REGIST_PARM_NUM_LSB 0x64 #define CTL_REGIST_PARM_NUM_MSB 0x65 /* undefined 0x66 - 0x78 */ /* reserved 0x79 - 0x7f */ /* Pseudo controllers (not midi compatible) */ #define CTRL_PITCH_BENDER 255 #define CTRL_PITCH_BENDER_RANGE 254 #define CTRL_EXPRESSION 253 /* Obsolete */ #define CTRL_MAIN_VOLUME 252 /* Obsolete */ #define SEQ_BALANCE 11 #define SEQ_VOLMODE 12 /* * Volume mode decides how volumes are used */ #define VOL_METHOD_ADAGIO 1 #define VOL_METHOD_LINEAR 2 /* * Note! SEQ_WAIT, SEQ_MIDIPUTC and SEQ_ECHO are used also as * input events. */ /* * Event codes 0xf0 to 0xfc are reserved for future extensions. */ #define SEQ_FULLSIZE 0xfd /* Long events */ /* * SEQ_FULLSIZE events are used for loading patches/samples to the * synthesizer devices. These events are passed directly to the driver * of the associated synthesizer device. There is no limit to the size * of the extended events. These events are not queued but executed * immediately when the write() is called (execution can take several * seconds of time). * * When a SEQ_FULLSIZE message is written to the device, it must * be written using exactly one write() call. Other events cannot * be mixed to the same write. * * For FM synths (YM3812/OPL3) use struct sbi_instrument and write * it to the /dev/sequencer. Don't write other data together with * the instrument structure Set the key field of the structure to * FM_PATCH. The device field is used to route the patch to the * corresponding device. * * For Gravis UltraSound use struct patch_info. Initialize the key field * to GUS_PATCH. */ #define SEQ_PRIVATE 0xfe /* Low level HW dependent events (8 bytes) */ #define SEQ_EXTENDED 0xff /* Extended events (8 bytes) OBSOLETE */ /* * Record for FM patches */ typedef u_char sbi_instr_data[32]; struct sbi_instrument { u_short key; /* FM_PATCH or OPL3_PATCH */ #define FM_PATCH _PATCHKEY(0x01) #define OPL3_PATCH _PATCHKEY(0x03) short device; /* Synth# (0-4) */ int channel; /* Program# to be initialized */ sbi_instr_data operators; /* Reg. settings for operator cells * (.SBI format) */ }; struct synth_info { /* Read only */ char name[30]; int device; /* 0-N. INITIALIZE BEFORE CALLING */ int synth_type; #define SYNTH_TYPE_FM 0 #define SYNTH_TYPE_SAMPLE 1 #define SYNTH_TYPE_MIDI 2 /* Midi interface */ int synth_subtype; #define FM_TYPE_ADLIB 0x00 #define FM_TYPE_OPL3 0x01 #define MIDI_TYPE_MPU401 0x401 #define SAMPLE_TYPE_BASIC 0x10 #define SAMPLE_TYPE_GUS SAMPLE_TYPE_BASIC #define SAMPLE_TYPE_AWE32 0x20 int perc_mode; /* No longer supported */ int nr_voices; int nr_drums; /* Obsolete field */ int instr_bank_size; u_long capabilities; #define SYNTH_CAP_PERCMODE 0x00000001 /* No longer used */ #define SYNTH_CAP_OPL3 0x00000002 /* Set if OPL3 supported */ #define SYNTH_CAP_INPUT 0x00000004 /* Input (MIDI) device */ int dummies[19]; /* Reserve space */ }; struct sound_timer_info { char name[32]; int caps; }; struct midi_info { char name[30]; int device; /* 0-N. INITIALIZE BEFORE CALLING */ u_long capabilities; /* To be defined later */ int dev_type; int dummies[18]; /* Reserve space */ }; /* * ioctl commands for the /dev/midi## */ typedef struct { u_char cmd; char nr_args, nr_returns; u_char data[30]; } mpu_command_rec; #define SNDCTL_MIDI_PRETIME _IOWR('m', 0, int) #define SNDCTL_MIDI_MPUMODE _IOWR('m', 1, int) #define SNDCTL_MIDI_MPUCMD _IOWR('m', 2, mpu_command_rec) #define MIOSPASSTHRU _IOWR('m', 3, int) #define MIOGPASSTHRU _IOWR('m', 4, int) /* * IOCTL commands for /dev/dsp and /dev/audio */ #define SNDCTL_DSP_RESET _IO ('P', 0) #define SNDCTL_DSP_SYNC _IO ('P', 1) #define SNDCTL_DSP_SPEED _IOWR('P', 2, int) #define SNDCTL_DSP_STEREO _IOWR('P', 3, int) #define SNDCTL_DSP_GETBLKSIZE _IOR('P', 4, int) #define SNDCTL_DSP_SETBLKSIZE _IOW('P', 4, int) #define SNDCTL_DSP_SETFMT _IOWR('P',5, int) /* Selects ONE fmt*/ /* * SOUND_PCM_WRITE_CHANNELS is not that different * from SNDCTL_DSP_STEREO */ #define SOUND_PCM_WRITE_CHANNELS _IOWR('P', 6, int) #define SNDCTL_DSP_CHANNELS SOUND_PCM_WRITE_CHANNELS #define SOUND_PCM_WRITE_FILTER _IOWR('P', 7, int) #define SNDCTL_DSP_POST _IO ('P', 8) /* * SNDCTL_DSP_SETBLKSIZE and the following two calls mostly do * the same thing, i.e. set the block size used in DMA transfers. */ #define SNDCTL_DSP_SUBDIVIDE _IOWR('P', 9, int) #define SNDCTL_DSP_SETFRAGMENT _IOWR('P',10, int) #define SNDCTL_DSP_GETFMTS _IOR ('P',11, int) /* Returns a mask */ /* * Buffer status queries. */ typedef struct audio_buf_info { int fragments; /* # of avail. frags (partly used ones not counted) */ int fragstotal; /* Total # of fragments allocated */ int fragsize; /* Size of a fragment in bytes */ int bytes; /* Avail. space in bytes (includes partly used fragments) */ /* Note! 'bytes' could be more than fragments*fragsize */ } audio_buf_info; #define SNDCTL_DSP_GETOSPACE _IOR ('P',12, audio_buf_info) #define SNDCTL_DSP_GETISPACE _IOR ('P',13, audio_buf_info) /* * SNDCTL_DSP_NONBLOCK is the same (but less powerful, since the * action cannot be undone) of FIONBIO. The same can be achieved * by opening the device with O_NDELAY */ #define SNDCTL_DSP_NONBLOCK _IO ('P',14) #define SNDCTL_DSP_GETCAPS _IOR ('P',15, int) # define PCM_CAP_REVISION 0x000000ff /* Bits for revision level (0 to 255) */ # define PCM_CAP_DUPLEX 0x00000100 /* Full duplex record/playback */ # define PCM_CAP_REALTIME 0x00000200 /* Not in use */ # define PCM_CAP_BATCH 0x00000400 /* Device has some kind of */ /* internal buffers which may */ /* cause some delays and */ /* decrease precision of timing */ # define PCM_CAP_COPROC 0x00000800 /* Has a coprocessor */ /* Sometimes it's a DSP */ /* but usually not */ # define PCM_CAP_TRIGGER 0x00001000 /* Supports SETTRIGGER */ # define PCM_CAP_MMAP 0x00002000 /* Supports mmap() */ # define PCM_CAP_MULTI 0x00004000 /* Supports multiple open */ # define PCM_CAP_BIND 0x00008000 /* Supports binding to front/rear/center/lfe */ # define PCM_CAP_INPUT 0x00010000 /* Supports recording */ # define PCM_CAP_OUTPUT 0x00020000 /* Supports playback */ # define PCM_CAP_VIRTUAL 0x00040000 /* Virtual device */ /* 0x00040000 and 0x00080000 reserved for future use */ /* Analog/digital control capabilities */ # define PCM_CAP_ANALOGOUT 0x00100000 # define PCM_CAP_ANALOGIN 0x00200000 # define PCM_CAP_DIGITALOUT 0x00400000 # define PCM_CAP_DIGITALIN 0x00800000 # define PCM_CAP_ADMASK 0x00f00000 /* * NOTE! (capabilities & PCM_CAP_ADMASK)==0 means just that the * digital/analog interface control features are not supported by the * device/driver. However the device still supports analog, digital or * both inputs/outputs (depending on the device). See the OSS Programmer's * Guide for full details. */ # define PCM_CAP_SPECIAL 0x01000000 /* Not for ordinary "multimedia" use */ # define PCM_CAP_SHADOW 0x00000000 /* OBSOLETE */ /* * Preferred channel usage. These bits can be used to * give recommendations to the application. Used by few drivers. * For example if ((caps & DSP_CH_MASK) == DSP_CH_MONO) means that * the device works best in mono mode. However it doesn't necessarily mean * that the device cannot be used in stereo. These bits should only be used * by special applications such as multi track hard disk recorders to find * out the initial setup. However the user should be able to override this * selection. * * To find out which modes are actually supported the application should * try to select them using SNDCTL_DSP_CHANNELS. */ # define DSP_CH_MASK 0x06000000 /* Mask */ # define DSP_CH_ANY 0x00000000 /* No preferred mode */ # define DSP_CH_MONO 0x02000000 # define DSP_CH_STEREO 0x04000000 # define DSP_CH_MULTI 0x06000000 /* More than two channels */ # define PCM_CAP_HIDDEN 0x08000000 /* Hidden device */ # define PCM_CAP_FREERATE 0x10000000 # define PCM_CAP_MODEM 0x20000000 /* Modem device */ # define PCM_CAP_DEFAULT 0x40000000 /* "Default" device */ /* * The PCM_CAP_* capability names were known as DSP_CAP_* prior OSS 4.0 * so it's necessary to define the older names too. */ #define DSP_CAP_ADMASK PCM_CAP_ADMASK #define DSP_CAP_ANALOGIN PCM_CAP_ANALOGIN #define DSP_CAP_ANALOGOUT PCM_CAP_ANALOGOUT #define DSP_CAP_BATCH PCM_CAP_BATCH #define DSP_CAP_BIND PCM_CAP_BIND #define DSP_CAP_COPROC PCM_CAP_COPROC #define DSP_CAP_DEFAULT PCM_CAP_DEFAULT #define DSP_CAP_DIGITALIN PCM_CAP_DIGITALIN #define DSP_CAP_DIGITALOUT PCM_CAP_DIGITALOUT #define DSP_CAP_DUPLEX PCM_CAP_DUPLEX #define DSP_CAP_FREERATE PCM_CAP_FREERATE #define DSP_CAP_HIDDEN PCM_CAP_HIDDEN #define DSP_CAP_INPUT PCM_CAP_INPUT #define DSP_CAP_MMAP PCM_CAP_MMAP #define DSP_CAP_MODEM PCM_CAP_MODEM #define DSP_CAP_MULTI PCM_CAP_MULTI #define DSP_CAP_OUTPUT PCM_CAP_OUTPUT #define DSP_CAP_REALTIME PCM_CAP_REALTIME #define DSP_CAP_REVISION PCM_CAP_REVISION #define DSP_CAP_SHADOW PCM_CAP_SHADOW #define DSP_CAP_TRIGGER PCM_CAP_TRIGGER #define DSP_CAP_VIRTUAL PCM_CAP_VIRTUAL /* * What do these function do ? */ #define SNDCTL_DSP_GETTRIGGER _IOR ('P',16, int) #define SNDCTL_DSP_SETTRIGGER _IOW ('P',16, int) #define PCM_ENABLE_INPUT 0x00000001 #define PCM_ENABLE_OUTPUT 0x00000002 typedef struct count_info { int bytes; /* Total # of bytes processed */ int blocks; /* # of fragment transitions since last time */ int ptr; /* Current DMA pointer value */ } count_info; /* * GETIPTR and GETISPACE are not that different... same for out. */ #define SNDCTL_DSP_GETIPTR _IOR ('P',17, count_info) #define SNDCTL_DSP_GETOPTR _IOR ('P',18, count_info) typedef struct buffmem_desc { caddr_t buffer; int size; } buffmem_desc; #define SNDCTL_DSP_MAPINBUF _IOR ('P', 19, buffmem_desc) #define SNDCTL_DSP_MAPOUTBUF _IOR ('P', 20, buffmem_desc) #define SNDCTL_DSP_SETSYNCRO _IO ('P', 21) #define SNDCTL_DSP_SETDUPLEX _IO ('P', 22) #define SNDCTL_DSP_GETODELAY _IOR ('P', 23, int) /* * I guess these are the readonly version of the same * functions that exist above as SNDCTL_DSP_... */ #define SOUND_PCM_READ_RATE _IOR ('P', 2, int) #define SOUND_PCM_READ_CHANNELS _IOR ('P', 6, int) #define SOUND_PCM_READ_BITS _IOR ('P', 5, int) #define SOUND_PCM_READ_FILTER _IOR ('P', 7, int) /* * ioctl calls to be used in communication with coprocessors and * DSP chips. */ typedef struct copr_buffer { int command; /* Set to 0 if not used */ int flags; #define CPF_NONE 0x0000 #define CPF_FIRST 0x0001 /* First block */ #define CPF_LAST 0x0002 /* Last block */ int len; int offs; /* If required by the device (0 if not used) */ u_char data[4000]; /* NOTE! 4000 is not 4k */ } copr_buffer; typedef struct copr_debug_buf { int command; /* Used internally. Set to 0 */ int parm1; int parm2; int flags; int len; /* Length of data in bytes */ } copr_debug_buf; typedef struct copr_msg { int len; u_char data[4000]; } copr_msg; #define SNDCTL_COPR_RESET _IO ('C', 0) #define SNDCTL_COPR_LOAD _IOWR('C', 1, copr_buffer) #define SNDCTL_COPR_RDATA _IOWR('C', 2, copr_debug_buf) #define SNDCTL_COPR_RCODE _IOWR('C', 3, copr_debug_buf) #define SNDCTL_COPR_WDATA _IOW ('C', 4, copr_debug_buf) #define SNDCTL_COPR_WCODE _IOW ('C', 5, copr_debug_buf) #define SNDCTL_COPR_RUN _IOWR('C', 6, copr_debug_buf) #define SNDCTL_COPR_HALT _IOWR('C', 7, copr_debug_buf) #define SNDCTL_COPR_SENDMSG _IOW ('C', 8, copr_msg) #define SNDCTL_COPR_RCVMSG _IOR ('C', 9, copr_msg) /* * IOCTL commands for /dev/mixer */ /* * Mixer devices * * There can be up to 20 different analog mixer channels. The * SOUND_MIXER_NRDEVICES gives the currently supported maximum. * The SOUND_MIXER_READ_DEVMASK returns a bitmask which tells * the devices supported by the particular mixer. */ #define SOUND_MIXER_NRDEVICES 25 #define SOUND_MIXER_VOLUME 0 /* Master output level */ #define SOUND_MIXER_BASS 1 /* Treble level of all output channels */ #define SOUND_MIXER_TREBLE 2 /* Bass level of all output channels */ #define SOUND_MIXER_SYNTH 3 /* Volume of synthesier input */ #define SOUND_MIXER_PCM 4 /* Output level for the audio device */ #define SOUND_MIXER_SPEAKER 5 /* Output level for the PC speaker * signals */ #define SOUND_MIXER_LINE 6 /* Volume level for the line in jack */ #define SOUND_MIXER_MIC 7 /* Volume for the signal coming from * the microphone jack */ #define SOUND_MIXER_CD 8 /* Volume level for the input signal * connected to the CD audio input */ #define SOUND_MIXER_IMIX 9 /* Recording monitor. It controls the * output volume of the selected * recording sources while recording */ #define SOUND_MIXER_ALTPCM 10 /* Volume of the alternative codec * device */ #define SOUND_MIXER_RECLEV 11 /* Global recording level */ #define SOUND_MIXER_IGAIN 12 /* Input gain */ #define SOUND_MIXER_OGAIN 13 /* Output gain */ /* * The AD1848 codec and compatibles have three line level inputs * (line, aux1 and aux2). Since each card manufacturer have assigned * different meanings to these inputs, it's inpractical to assign * specific meanings (line, cd, synth etc.) to them. */ #define SOUND_MIXER_LINE1 14 /* Input source 1 (aux1) */ #define SOUND_MIXER_LINE2 15 /* Input source 2 (aux2) */ #define SOUND_MIXER_LINE3 16 /* Input source 3 (line) */ #define SOUND_MIXER_DIGITAL1 17 /* Digital (input) 1 */ #define SOUND_MIXER_DIGITAL2 18 /* Digital (input) 2 */ #define SOUND_MIXER_DIGITAL3 19 /* Digital (input) 3 */ #define SOUND_MIXER_PHONEIN 20 /* Phone input */ #define SOUND_MIXER_PHONEOUT 21 /* Phone output */ #define SOUND_MIXER_VIDEO 22 /* Video/TV (audio) in */ #define SOUND_MIXER_RADIO 23 /* Radio in */ #define SOUND_MIXER_MONITOR 24 /* Monitor (usually mic) volume */ /* * Some on/off settings (SOUND_SPECIAL_MIN - SOUND_SPECIAL_MAX) * Not counted to SOUND_MIXER_NRDEVICES, but use the same number space */ #define SOUND_ONOFF_MIN 28 #define SOUND_ONOFF_MAX 30 #define SOUND_MIXER_MUTE 28 /* 0 or 1 */ #define SOUND_MIXER_ENHANCE 29 /* Enhanced stereo (0, 40, 60 or 80) */ #define SOUND_MIXER_LOUD 30 /* 0 or 1 */ /* Note! Number 31 cannot be used since the sign bit is reserved */ #define SOUND_MIXER_NONE 31 #define SOUND_DEVICE_LABELS { \ "Vol ", "Bass ", "Trebl", "Synth", "Pcm ", "Spkr ", "Line ", \ "Mic ", "CD ", "Mix ", "Pcm2 ", "Rec ", "IGain", "OGain", \ "Line1", "Line2", "Line3", "Digital1", "Digital2", "Digital3", \ "PhoneIn", "PhoneOut", "Video", "Radio", "Monitor"} #define SOUND_DEVICE_NAMES { \ "vol", "bass", "treble", "synth", "pcm", "speaker", "line", \ "mic", "cd", "mix", "pcm2", "rec", "igain", "ogain", \ "line1", "line2", "line3", "dig1", "dig2", "dig3", \ "phin", "phout", "video", "radio", "monitor"} /* Device bitmask identifiers */ #define SOUND_MIXER_RECSRC 0xff /* 1 bit per recording source */ #define SOUND_MIXER_DEVMASK 0xfe /* 1 bit per supported device */ #define SOUND_MIXER_RECMASK 0xfd /* 1 bit per supp. recording source */ #define SOUND_MIXER_CAPS 0xfc #define SOUND_CAP_EXCL_INPUT 0x00000001 /* Only 1 rec. src at a time */ #define SOUND_MIXER_STEREODEVS 0xfb /* Mixer channels supporting stereo */ /* Device mask bits */ #define SOUND_MASK_VOLUME (1 << SOUND_MIXER_VOLUME) #define SOUND_MASK_BASS (1 << SOUND_MIXER_BASS) #define SOUND_MASK_TREBLE (1 << SOUND_MIXER_TREBLE) #define SOUND_MASK_SYNTH (1 << SOUND_MIXER_SYNTH) #define SOUND_MASK_PCM (1 << SOUND_MIXER_PCM) #define SOUND_MASK_SPEAKER (1 << SOUND_MIXER_SPEAKER) #define SOUND_MASK_LINE (1 << SOUND_MIXER_LINE) #define SOUND_MASK_MIC (1 << SOUND_MIXER_MIC) #define SOUND_MASK_CD (1 << SOUND_MIXER_CD) #define SOUND_MASK_IMIX (1 << SOUND_MIXER_IMIX) #define SOUND_MASK_ALTPCM (1 << SOUND_MIXER_ALTPCM) #define SOUND_MASK_RECLEV (1 << SOUND_MIXER_RECLEV) #define SOUND_MASK_IGAIN (1 << SOUND_MIXER_IGAIN) #define SOUND_MASK_OGAIN (1 << SOUND_MIXER_OGAIN) #define SOUND_MASK_LINE1 (1 << SOUND_MIXER_LINE1) #define SOUND_MASK_LINE2 (1 << SOUND_MIXER_LINE2) #define SOUND_MASK_LINE3 (1 << SOUND_MIXER_LINE3) #define SOUND_MASK_DIGITAL1 (1 << SOUND_MIXER_DIGITAL1) #define SOUND_MASK_DIGITAL2 (1 << SOUND_MIXER_DIGITAL2) #define SOUND_MASK_DIGITAL3 (1 << SOUND_MIXER_DIGITAL3) #define SOUND_MASK_PHONEIN (1 << SOUND_MIXER_PHONEIN) #define SOUND_MASK_PHONEOUT (1 << SOUND_MIXER_PHONEOUT) #define SOUND_MASK_RADIO (1 << SOUND_MIXER_RADIO) #define SOUND_MASK_VIDEO (1 << SOUND_MIXER_VIDEO) #define SOUND_MASK_MONITOR (1 << SOUND_MIXER_MONITOR) /* Obsolete macros */ #define SOUND_MASK_MUTE (1 << SOUND_MIXER_MUTE) #define SOUND_MASK_ENHANCE (1 << SOUND_MIXER_ENHANCE) #define SOUND_MASK_LOUD (1 << SOUND_MIXER_LOUD) #define MIXER_READ(dev) _IOR('M', dev, int) #define SOUND_MIXER_READ_VOLUME MIXER_READ(SOUND_MIXER_VOLUME) #define SOUND_MIXER_READ_BASS MIXER_READ(SOUND_MIXER_BASS) #define SOUND_MIXER_READ_TREBLE MIXER_READ(SOUND_MIXER_TREBLE) #define SOUND_MIXER_READ_SYNTH MIXER_READ(SOUND_MIXER_SYNTH) #define SOUND_MIXER_READ_PCM MIXER_READ(SOUND_MIXER_PCM) #define SOUND_MIXER_READ_SPEAKER MIXER_READ(SOUND_MIXER_SPEAKER) #define SOUND_MIXER_READ_LINE MIXER_READ(SOUND_MIXER_LINE) #define SOUND_MIXER_READ_MIC MIXER_READ(SOUND_MIXER_MIC) #define SOUND_MIXER_READ_CD MIXER_READ(SOUND_MIXER_CD) #define SOUND_MIXER_READ_IMIX MIXER_READ(SOUND_MIXER_IMIX) #define SOUND_MIXER_READ_ALTPCM MIXER_READ(SOUND_MIXER_ALTPCM) #define SOUND_MIXER_READ_RECLEV MIXER_READ(SOUND_MIXER_RECLEV) #define SOUND_MIXER_READ_IGAIN MIXER_READ(SOUND_MIXER_IGAIN) #define SOUND_MIXER_READ_OGAIN MIXER_READ(SOUND_MIXER_OGAIN) #define SOUND_MIXER_READ_LINE1 MIXER_READ(SOUND_MIXER_LINE1) #define SOUND_MIXER_READ_LINE2 MIXER_READ(SOUND_MIXER_LINE2) #define SOUND_MIXER_READ_LINE3 MIXER_READ(SOUND_MIXER_LINE3) #define SOUND_MIXER_READ_DIGITAL1 MIXER_READ(SOUND_MIXER_DIGITAL1) #define SOUND_MIXER_READ_DIGITAL2 MIXER_READ(SOUND_MIXER_DIGITAL2) #define SOUND_MIXER_READ_DIGITAL3 MIXER_READ(SOUND_MIXER_DIGITAL3) #define SOUND_MIXER_READ_PHONEIN MIXER_READ(SOUND_MIXER_PHONEIN) #define SOUND_MIXER_READ_PHONEOUT MIXER_READ(SOUND_MIXER_PHONEOUT) #define SOUND_MIXER_READ_RADIO MIXER_READ(SOUND_MIXER_RADIO) #define SOUND_MIXER_READ_VIDEO MIXER_READ(SOUND_MIXER_VIDEO) #define SOUND_MIXER_READ_MONITOR MIXER_READ(SOUND_MIXER_MONITOR) /* Obsolete macros */ #define SOUND_MIXER_READ_MUTE MIXER_READ(SOUND_MIXER_MUTE) #define SOUND_MIXER_READ_ENHANCE MIXER_READ(SOUND_MIXER_ENHANCE) #define SOUND_MIXER_READ_LOUD MIXER_READ(SOUND_MIXER_LOUD) #define SOUND_MIXER_READ_RECSRC MIXER_READ(SOUND_MIXER_RECSRC) #define SOUND_MIXER_READ_DEVMASK MIXER_READ(SOUND_MIXER_DEVMASK) #define SOUND_MIXER_READ_RECMASK MIXER_READ(SOUND_MIXER_RECMASK) #define SOUND_MIXER_READ_STEREODEVS MIXER_READ(SOUND_MIXER_STEREODEVS) #define SOUND_MIXER_READ_CAPS MIXER_READ(SOUND_MIXER_CAPS) #define MIXER_WRITE(dev) _IOWR('M', dev, int) #define SOUND_MIXER_WRITE_VOLUME MIXER_WRITE(SOUND_MIXER_VOLUME) #define SOUND_MIXER_WRITE_BASS MIXER_WRITE(SOUND_MIXER_BASS) #define SOUND_MIXER_WRITE_TREBLE MIXER_WRITE(SOUND_MIXER_TREBLE) #define SOUND_MIXER_WRITE_SYNTH MIXER_WRITE(SOUND_MIXER_SYNTH) #define SOUND_MIXER_WRITE_PCM MIXER_WRITE(SOUND_MIXER_PCM) #define SOUND_MIXER_WRITE_SPEAKER MIXER_WRITE(SOUND_MIXER_SPEAKER) #define SOUND_MIXER_WRITE_LINE MIXER_WRITE(SOUND_MIXER_LINE) #define SOUND_MIXER_WRITE_MIC MIXER_WRITE(SOUND_MIXER_MIC) #define SOUND_MIXER_WRITE_CD MIXER_WRITE(SOUND_MIXER_CD) #define SOUND_MIXER_WRITE_IMIX MIXER_WRITE(SOUND_MIXER_IMIX) #define SOUND_MIXER_WRITE_ALTPCM MIXER_WRITE(SOUND_MIXER_ALTPCM) #define SOUND_MIXER_WRITE_RECLEV MIXER_WRITE(SOUND_MIXER_RECLEV) #define SOUND_MIXER_WRITE_IGAIN MIXER_WRITE(SOUND_MIXER_IGAIN) #define SOUND_MIXER_WRITE_OGAIN MIXER_WRITE(SOUND_MIXER_OGAIN) #define SOUND_MIXER_WRITE_LINE1 MIXER_WRITE(SOUND_MIXER_LINE1) #define SOUND_MIXER_WRITE_LINE2 MIXER_WRITE(SOUND_MIXER_LINE2) #define SOUND_MIXER_WRITE_LINE3 MIXER_WRITE(SOUND_MIXER_LINE3) #define SOUND_MIXER_WRITE_DIGITAL1 MIXER_WRITE(SOUND_MIXER_DIGITAL1) #define SOUND_MIXER_WRITE_DIGITAL2 MIXER_WRITE(SOUND_MIXER_DIGITAL2) #define SOUND_MIXER_WRITE_DIGITAL3 MIXER_WRITE(SOUND_MIXER_DIGITAL3) #define SOUND_MIXER_WRITE_PHONEIN MIXER_WRITE(SOUND_MIXER_PHONEIN) #define SOUND_MIXER_WRITE_PHONEOUT MIXER_WRITE(SOUND_MIXER_PHONEOUT) #define SOUND_MIXER_WRITE_RADIO MIXER_WRITE(SOUND_MIXER_RADIO) #define SOUND_MIXER_WRITE_VIDEO MIXER_WRITE(SOUND_MIXER_VIDEO) #define SOUND_MIXER_WRITE_MONITOR MIXER_WRITE(SOUND_MIXER_MONITOR) #define SOUND_MIXER_WRITE_MUTE MIXER_WRITE(SOUND_MIXER_MUTE) #define SOUND_MIXER_WRITE_ENHANCE MIXER_WRITE(SOUND_MIXER_ENHANCE) #define SOUND_MIXER_WRITE_LOUD MIXER_WRITE(SOUND_MIXER_LOUD) #define SOUND_MIXER_WRITE_RECSRC MIXER_WRITE(SOUND_MIXER_RECSRC) typedef struct mixer_info { char id[16]; char name[32]; int modify_counter; int fillers[10]; } mixer_info; #define SOUND_MIXER_INFO _IOR('M', 101, mixer_info) #define LEFT_CHN 0 #define RIGHT_CHN 1 /* * Level 2 event types for /dev/sequencer */ /* * The 4 most significant bits of byte 0 specify the class of * the event: * * 0x8X = system level events, * 0x9X = device/port specific events, event[1] = device/port, * The last 4 bits give the subtype: * 0x02 = Channel event (event[3] = chn). * 0x01 = note event (event[4] = note). * (0x01 is not used alone but always with bit 0x02). * event[2] = MIDI message code (0x80=note off etc.) * */ #define EV_SEQ_LOCAL 0x80 #define EV_TIMING 0x81 #define EV_CHN_COMMON 0x92 #define EV_CHN_VOICE 0x93 #define EV_SYSEX 0x94 /* * Event types 200 to 220 are reserved for application use. * These numbers will not be used by the driver. */ /* * Events for event type EV_CHN_VOICE */ #define MIDI_NOTEOFF 0x80 #define MIDI_NOTEON 0x90 #define MIDI_KEY_PRESSURE 0xA0 /* * Events for event type EV_CHN_COMMON */ #define MIDI_CTL_CHANGE 0xB0 #define MIDI_PGM_CHANGE 0xC0 #define MIDI_CHN_PRESSURE 0xD0 #define MIDI_PITCH_BEND 0xE0 #define MIDI_SYSTEM_PREFIX 0xF0 /* * Timer event types */ #define TMR_WAIT_REL 1 /* Time relative to the prev time */ #define TMR_WAIT_ABS 2 /* Absolute time since TMR_START */ #define TMR_STOP 3 #define TMR_START 4 #define TMR_CONTINUE 5 #define TMR_TEMPO 6 #define TMR_ECHO 8 #define TMR_CLOCK 9 /* MIDI clock */ #define TMR_SPP 10 /* Song position pointer */ #define TMR_TIMESIG 11 /* Time signature */ /* * Local event types */ #define LOCL_STARTAUDIO 1 #if (!defined(_KERNEL) && !defined(INKERNEL)) || defined(USE_SEQ_MACROS) /* * Some convenience macros to simplify programming of the * /dev/sequencer interface * * These macros define the API which should be used when possible. */ #ifndef USE_SIMPLE_MACROS void seqbuf_dump(void); /* This function must be provided by programs */ /* Sample seqbuf_dump() implementation: * * SEQ_DEFINEBUF (2048); -- Defines a buffer for 2048 bytes * * int seqfd; -- The file descriptor for /dev/sequencer. * * void * seqbuf_dump () * { * if (_seqbufptr) * if (write (seqfd, _seqbuf, _seqbufptr) == -1) * { * perror ("write /dev/sequencer"); * exit (-1); * } * _seqbufptr = 0; * } */ #define SEQ_DEFINEBUF(len) \ u_char _seqbuf[len]; int _seqbuflen = len;int _seqbufptr = 0 #define SEQ_USE_EXTBUF() \ extern u_char _seqbuf[]; \ extern int _seqbuflen;extern int _seqbufptr #define SEQ_DECLAREBUF() SEQ_USE_EXTBUF() #define SEQ_PM_DEFINES struct patmgr_info _pm_info #define _SEQ_NEEDBUF(len) \ if ((_seqbufptr+(len)) > _seqbuflen) \ seqbuf_dump() #define _SEQ_ADVBUF(len) _seqbufptr += len #define SEQ_DUMPBUF seqbuf_dump #else /* * This variation of the sequencer macros is used just to format one event * using fixed buffer. * * The program using the macro library must define the following macros before * using this library. * * #define _seqbuf name of the buffer (u_char[]) * #define _SEQ_ADVBUF(len) If the applic needs to know the exact * size of the event, this macro can be used. * Otherwise this must be defined as empty. * #define _seqbufptr Define the name of index variable or 0 if * not required. */ #define _SEQ_NEEDBUF(len) /* empty */ #endif #define PM_LOAD_PATCH(dev, bank, pgm) \ (SEQ_DUMPBUF(), _pm_info.command = _PM_LOAD_PATCH, \ _pm_info.device=dev, _pm_info.data.data8[0]=pgm, \ _pm_info.parm1 = bank, _pm_info.parm2 = 1, \ ioctl(seqfd, SNDCTL_PMGR_ACCESS, &_pm_info)) #define PM_LOAD_PATCHES(dev, bank, pgm) \ (SEQ_DUMPBUF(), _pm_info.command = _PM_LOAD_PATCH, \ _pm_info.device=dev, bcopy( pgm, _pm_info.data.data8, 128), \ _pm_info.parm1 = bank, _pm_info.parm2 = 128, \ ioctl(seqfd, SNDCTL_PMGR_ACCESS, &_pm_info)) #define SEQ_VOLUME_MODE(dev, mode) { \ _SEQ_NEEDBUF(8);\ _seqbuf[_seqbufptr] = SEQ_EXTENDED;\ _seqbuf[_seqbufptr+1] = SEQ_VOLMODE;\ _seqbuf[_seqbufptr+2] = (dev);\ _seqbuf[_seqbufptr+3] = (mode);\ _seqbuf[_seqbufptr+4] = 0;\ _seqbuf[_seqbufptr+5] = 0;\ _seqbuf[_seqbufptr+6] = 0;\ _seqbuf[_seqbufptr+7] = 0;\ _SEQ_ADVBUF(8);} /* * Midi voice messages */ #define _CHN_VOICE(dev, event, chn, note, parm) { \ _SEQ_NEEDBUF(8);\ _seqbuf[_seqbufptr] = EV_CHN_VOICE;\ _seqbuf[_seqbufptr+1] = (dev);\ _seqbuf[_seqbufptr+2] = (event);\ _seqbuf[_seqbufptr+3] = (chn);\ _seqbuf[_seqbufptr+4] = (note);\ _seqbuf[_seqbufptr+5] = (parm);\ _seqbuf[_seqbufptr+6] = (0);\ _seqbuf[_seqbufptr+7] = 0;\ _SEQ_ADVBUF(8);} #define SEQ_START_NOTE(dev, chn, note, vol) \ _CHN_VOICE(dev, MIDI_NOTEON, chn, note, vol) #define SEQ_STOP_NOTE(dev, chn, note, vol) \ _CHN_VOICE(dev, MIDI_NOTEOFF, chn, note, vol) #define SEQ_KEY_PRESSURE(dev, chn, note, pressure) \ _CHN_VOICE(dev, MIDI_KEY_PRESSURE, chn, note, pressure) /* * Midi channel messages */ #define _CHN_COMMON(dev, event, chn, p1, p2, w14) { \ _SEQ_NEEDBUF(8);\ _seqbuf[_seqbufptr] = EV_CHN_COMMON;\ _seqbuf[_seqbufptr+1] = (dev);\ _seqbuf[_seqbufptr+2] = (event);\ _seqbuf[_seqbufptr+3] = (chn);\ _seqbuf[_seqbufptr+4] = (p1);\ _seqbuf[_seqbufptr+5] = (p2);\ *(short *)&_seqbuf[_seqbufptr+6] = (w14);\ _SEQ_ADVBUF(8);} /* * SEQ_SYSEX permits sending of sysex messages. (It may look that it permits * sending any MIDI bytes but it's absolutely not possible. Trying to do * so _will_ cause problems with MPU401 intelligent mode). * * Sysex messages are sent in blocks of 1 to 6 bytes. Longer messages must be * sent by calling SEQ_SYSEX() several times (there must be no other events * between them). First sysex fragment must have 0xf0 in the first byte * and the last byte (buf[len-1] of the last fragment must be 0xf7. No byte * between these sysex start and end markers cannot be larger than 0x7f. Also * lengths of each fragments (except the last one) must be 6. * * Breaking the above rules may work with some MIDI ports but is likely to * cause fatal problems with some other devices (such as MPU401). */ #define SEQ_SYSEX(dev, buf, len) { \ int i, l=(len); if (l>6)l=6;\ _SEQ_NEEDBUF(8);\ _seqbuf[_seqbufptr] = EV_SYSEX;\ for(i=0;i __FBSDID("$FreeBSD$"); #include "opt_quota.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 typedef ufs2_daddr_t allocfcn_t(struct inode *ip, u_int cg, ufs2_daddr_t bpref, int size, int rsize); static ufs2_daddr_t ffs_alloccg(struct inode *, u_int, ufs2_daddr_t, int, int); static ufs2_daddr_t ffs_alloccgblk(struct inode *, struct buf *, ufs2_daddr_t, int); static void ffs_blkfree_cg(struct ufsmount *, struct fs *, struct vnode *, ufs2_daddr_t, long, ino_t, struct workhead *); static void ffs_blkfree_trim_completed(struct bio *); static void ffs_blkfree_trim_task(void *ctx, int pending __unused); #ifdef INVARIANTS static int ffs_checkblk(struct inode *, ufs2_daddr_t, long); #endif static ufs2_daddr_t ffs_clusteralloc(struct inode *, u_int, ufs2_daddr_t, int, int); static ino_t ffs_dirpref(struct inode *); static ufs2_daddr_t ffs_fragextend(struct inode *, u_int, ufs2_daddr_t, int, int); static void ffs_fserr(struct fs *, ino_t, char *); static ufs2_daddr_t ffs_hashalloc (struct inode *, u_int, ufs2_daddr_t, int, int, allocfcn_t *); static ufs2_daddr_t ffs_nodealloccg(struct inode *, u_int, ufs2_daddr_t, int, int); static ufs1_daddr_t ffs_mapsearch(struct fs *, struct cg *, ufs2_daddr_t, int); static int ffs_reallocblks_ufs1(struct vop_reallocblks_args *); static int ffs_reallocblks_ufs2(struct vop_reallocblks_args *); /* * Allocate a block in the filesystem. * * The size of the requested block is given, which must be some * multiple of fs_fsize and <= fs_bsize. * A preference may be optionally specified. If a preference is given * the following hierarchy is used to allocate a block: * 1) allocate the requested block. * 2) allocate a rotationally optimal block in the same cylinder. * 3) allocate a block in the same cylinder group. * 4) quadradically rehash into other cylinder groups, until an * available block is located. * If no block preference is given the following hierarchy is used * to allocate a block: * 1) allocate a block in the cylinder group that contains the * inode for the file. * 2) quadradically rehash into other cylinder groups, until an * available block is located. */ int ffs_alloc(ip, lbn, bpref, size, flags, cred, bnp) struct inode *ip; ufs2_daddr_t lbn, bpref; int size, flags; struct ucred *cred; ufs2_daddr_t *bnp; { struct fs *fs; struct ufsmount *ump; ufs2_daddr_t bno; u_int cg, reclaimed; static struct timeval lastfail; static int curfail; int64_t delta; #ifdef QUOTA int error; #endif *bnp = 0; fs = ip->i_fs; ump = ip->i_ump; mtx_assert(UFS_MTX(ump), MA_OWNED); #ifdef INVARIANTS if ((u_int)size > fs->fs_bsize || fragoff(fs, size) != 0) { printf("dev = %s, bsize = %ld, size = %d, fs = %s\n", devtoname(ip->i_dev), (long)fs->fs_bsize, size, fs->fs_fsmnt); panic("ffs_alloc: bad size"); } if (cred == NOCRED) panic("ffs_alloc: missing credential"); #endif /* INVARIANTS */ reclaimed = 0; retry: #ifdef QUOTA UFS_UNLOCK(ump); error = chkdq(ip, btodb(size), cred, 0); if (error) return (error); UFS_LOCK(ump); #endif if (size == fs->fs_bsize && fs->fs_cstotal.cs_nbfree == 0) goto nospace; if (priv_check_cred(cred, PRIV_VFS_BLOCKRESERVE, 0) && freespace(fs, fs->fs_minfree) - numfrags(fs, size) < 0) goto nospace; if (bpref >= fs->fs_size) bpref = 0; if (bpref == 0) cg = ino_to_cg(fs, ip->i_number); else cg = dtog(fs, bpref); bno = ffs_hashalloc(ip, cg, bpref, size, size, ffs_alloccg); if (bno > 0) { delta = btodb(size); DIP_SET(ip, i_blocks, DIP(ip, i_blocks) + delta); if (flags & IO_EXT) ip->i_flag |= IN_CHANGE; else ip->i_flag |= IN_CHANGE | IN_UPDATE; *bnp = bno; return (0); } nospace: #ifdef QUOTA UFS_UNLOCK(ump); /* * Restore user's disk quota because allocation failed. */ (void) chkdq(ip, -btodb(size), cred, FORCE); UFS_LOCK(ump); #endif if (reclaimed == 0) { reclaimed = 1; softdep_request_cleanup(fs, ITOV(ip), cred, FLUSH_BLOCKS_WAIT); goto retry; } UFS_UNLOCK(ump); if (ppsratecheck(&lastfail, &curfail, 1)) { ffs_fserr(fs, ip->i_number, "filesystem full"); uprintf("\n%s: write failed, filesystem is full\n", fs->fs_fsmnt); } return (ENOSPC); } /* * Reallocate a fragment to a bigger size * * The number and size of the old block is given, and a preference * and new size is also specified. The allocator attempts to extend * the original block. Failing that, the regular block allocator is * invoked to get an appropriate block. */ int ffs_realloccg(ip, lbprev, bprev, bpref, osize, nsize, flags, cred, bpp) struct inode *ip; ufs2_daddr_t lbprev; ufs2_daddr_t bprev; ufs2_daddr_t bpref; int osize, nsize, flags; struct ucred *cred; struct buf **bpp; { struct vnode *vp; struct fs *fs; struct buf *bp; struct ufsmount *ump; u_int cg, request, reclaimed; int error; ufs2_daddr_t bno; static struct timeval lastfail; static int curfail; int64_t delta; *bpp = 0; vp = ITOV(ip); fs = ip->i_fs; bp = NULL; ump = ip->i_ump; mtx_assert(UFS_MTX(ump), MA_OWNED); #ifdef INVARIANTS if (vp->v_mount->mnt_kern_flag & MNTK_SUSPENDED) panic("ffs_realloccg: allocation on suspended filesystem"); if ((u_int)osize > fs->fs_bsize || fragoff(fs, osize) != 0 || (u_int)nsize > fs->fs_bsize || fragoff(fs, nsize) != 0) { printf( "dev = %s, bsize = %ld, osize = %d, nsize = %d, fs = %s\n", devtoname(ip->i_dev), (long)fs->fs_bsize, osize, nsize, fs->fs_fsmnt); panic("ffs_realloccg: bad size"); } if (cred == NOCRED) panic("ffs_realloccg: missing credential"); #endif /* INVARIANTS */ reclaimed = 0; retry: if (priv_check_cred(cred, PRIV_VFS_BLOCKRESERVE, 0) && freespace(fs, fs->fs_minfree) - numfrags(fs, nsize - osize) < 0) { goto nospace; } if (bprev == 0) { printf("dev = %s, bsize = %ld, bprev = %jd, fs = %s\n", devtoname(ip->i_dev), (long)fs->fs_bsize, (intmax_t)bprev, fs->fs_fsmnt); panic("ffs_realloccg: bad bprev"); } UFS_UNLOCK(ump); /* * Allocate the extra space in the buffer. */ error = bread(vp, lbprev, osize, NOCRED, &bp); if (error) { brelse(bp); return (error); } if (bp->b_blkno == bp->b_lblkno) { if (lbprev >= NDADDR) panic("ffs_realloccg: lbprev out of range"); bp->b_blkno = fsbtodb(fs, bprev); } #ifdef QUOTA error = chkdq(ip, btodb(nsize - osize), cred, 0); if (error) { brelse(bp); return (error); } #endif /* * Check for extension in the existing location. */ cg = dtog(fs, bprev); UFS_LOCK(ump); bno = ffs_fragextend(ip, cg, bprev, osize, nsize); if (bno) { if (bp->b_blkno != fsbtodb(fs, bno)) panic("ffs_realloccg: bad blockno"); delta = btodb(nsize - osize); DIP_SET(ip, i_blocks, DIP(ip, i_blocks) + delta); if (flags & IO_EXT) ip->i_flag |= IN_CHANGE; else ip->i_flag |= IN_CHANGE | IN_UPDATE; allocbuf(bp, nsize); bp->b_flags |= B_DONE; bzero(bp->b_data + osize, nsize - osize); if ((bp->b_flags & (B_MALLOC | B_VMIO)) == B_VMIO) vfs_bio_set_valid(bp, osize, nsize - osize); *bpp = bp; return (0); } /* * Allocate a new disk location. */ if (bpref >= fs->fs_size) bpref = 0; switch ((int)fs->fs_optim) { case FS_OPTSPACE: /* * Allocate an exact sized fragment. Although this makes * best use of space, we will waste time relocating it if * the file continues to grow. If the fragmentation is * less than half of the minimum free reserve, we choose * to begin optimizing for time. */ request = nsize; if (fs->fs_minfree <= 5 || fs->fs_cstotal.cs_nffree > (off_t)fs->fs_dsize * fs->fs_minfree / (2 * 100)) break; log(LOG_NOTICE, "%s: optimization changed from SPACE to TIME\n", fs->fs_fsmnt); fs->fs_optim = FS_OPTTIME; break; case FS_OPTTIME: /* * At this point we have discovered a file that is trying to * grow a small fragment to a larger fragment. To save time, * we allocate a full sized block, then free the unused portion. * If the file continues to grow, the `ffs_fragextend' call * above will be able to grow it in place without further * copying. If aberrant programs cause disk fragmentation to * grow within 2% of the free reserve, we choose to begin * optimizing for space. */ request = fs->fs_bsize; if (fs->fs_cstotal.cs_nffree < (off_t)fs->fs_dsize * (fs->fs_minfree - 2) / 100) break; log(LOG_NOTICE, "%s: optimization changed from TIME to SPACE\n", fs->fs_fsmnt); fs->fs_optim = FS_OPTSPACE; break; default: printf("dev = %s, optim = %ld, fs = %s\n", devtoname(ip->i_dev), (long)fs->fs_optim, fs->fs_fsmnt); panic("ffs_realloccg: bad optim"); /* NOTREACHED */ } bno = ffs_hashalloc(ip, cg, bpref, request, nsize, ffs_alloccg); if (bno > 0) { bp->b_blkno = fsbtodb(fs, bno); if (!DOINGSOFTDEP(vp)) ffs_blkfree(ump, fs, ip->i_devvp, bprev, (long)osize, ip->i_number, NULL); delta = btodb(nsize - osize); DIP_SET(ip, i_blocks, DIP(ip, i_blocks) + delta); if (flags & IO_EXT) ip->i_flag |= IN_CHANGE; else ip->i_flag |= IN_CHANGE | IN_UPDATE; allocbuf(bp, nsize); bp->b_flags |= B_DONE; bzero(bp->b_data + osize, nsize - osize); if ((bp->b_flags & (B_MALLOC | B_VMIO)) == B_VMIO) vfs_bio_set_valid(bp, osize, nsize - osize); *bpp = bp; return (0); } #ifdef QUOTA UFS_UNLOCK(ump); /* * Restore user's disk quota because allocation failed. */ (void) chkdq(ip, -btodb(nsize - osize), cred, FORCE); UFS_LOCK(ump); #endif nospace: /* * no space available */ if (reclaimed == 0) { reclaimed = 1; - softdep_request_cleanup(fs, vp, cred, FLUSH_BLOCKS_WAIT); UFS_UNLOCK(ump); if (bp) { brelse(bp); bp = NULL; } UFS_LOCK(ump); + softdep_request_cleanup(fs, vp, cred, FLUSH_BLOCKS_WAIT); goto retry; } UFS_UNLOCK(ump); if (bp) brelse(bp); if (ppsratecheck(&lastfail, &curfail, 1)) { ffs_fserr(fs, ip->i_number, "filesystem full"); uprintf("\n%s: write failed, filesystem is full\n", fs->fs_fsmnt); } return (ENOSPC); } /* * Reallocate a sequence of blocks into a contiguous sequence of blocks. * * The vnode and an array of buffer pointers for a range of sequential * logical blocks to be made contiguous is given. The allocator attempts * to find a range of sequential blocks starting as close as possible * from the end of the allocation for the logical block immediately * preceding the current range. If successful, the physical block numbers * in the buffer pointers and in the inode are changed to reflect the new * allocation. If unsuccessful, the allocation is left unchanged. The * success in doing the reallocation is returned. Note that the error * return is not reflected back to the user. Rather the previous block * allocation will be used. */ SYSCTL_NODE(_vfs, OID_AUTO, ffs, CTLFLAG_RW, 0, "FFS filesystem"); static int doasyncfree = 1; SYSCTL_INT(_vfs_ffs, OID_AUTO, doasyncfree, CTLFLAG_RW, &doasyncfree, 0, ""); static int doreallocblks = 1; SYSCTL_INT(_vfs_ffs, OID_AUTO, doreallocblks, CTLFLAG_RW, &doreallocblks, 0, ""); #ifdef DEBUG static volatile int prtrealloc = 0; #endif int ffs_reallocblks(ap) struct vop_reallocblks_args /* { struct vnode *a_vp; struct cluster_save *a_buflist; } */ *ap; { if (doreallocblks == 0) return (ENOSPC); /* * We can't wait in softdep prealloc as it may fsync and recurse * here. Instead we simply fail to reallocate blocks if this * rare condition arises. */ if (DOINGSOFTDEP(ap->a_vp)) if (softdep_prealloc(ap->a_vp, MNT_NOWAIT) != 0) return (ENOSPC); if (VTOI(ap->a_vp)->i_ump->um_fstype == UFS1) return (ffs_reallocblks_ufs1(ap)); return (ffs_reallocblks_ufs2(ap)); } static int ffs_reallocblks_ufs1(ap) struct vop_reallocblks_args /* { struct vnode *a_vp; struct cluster_save *a_buflist; } */ *ap; { struct fs *fs; struct inode *ip; struct vnode *vp; struct buf *sbp, *ebp; ufs1_daddr_t *bap, *sbap, *ebap = 0; struct cluster_save *buflist; struct ufsmount *ump; ufs_lbn_t start_lbn, end_lbn; ufs1_daddr_t soff, newblk, blkno; ufs2_daddr_t pref; struct indir start_ap[NIADDR + 1], end_ap[NIADDR + 1], *idp; int i, len, start_lvl, end_lvl, ssize; vp = ap->a_vp; ip = VTOI(vp); fs = ip->i_fs; ump = ip->i_ump; if (fs->fs_contigsumsize <= 0) return (ENOSPC); buflist = ap->a_buflist; len = buflist->bs_nchildren; start_lbn = buflist->bs_children[0]->b_lblkno; end_lbn = start_lbn + len - 1; #ifdef INVARIANTS for (i = 0; i < len; i++) if (!ffs_checkblk(ip, dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize)) panic("ffs_reallocblks: unallocated block 1"); for (i = 1; i < len; i++) if (buflist->bs_children[i]->b_lblkno != start_lbn + i) panic("ffs_reallocblks: non-logical cluster"); blkno = buflist->bs_children[0]->b_blkno; ssize = fsbtodb(fs, fs->fs_frag); for (i = 1; i < len - 1; i++) if (buflist->bs_children[i]->b_blkno != blkno + (i * ssize)) panic("ffs_reallocblks: non-physical cluster %d", i); #endif /* * If the latest allocation is in a new cylinder group, assume that * the filesystem has decided to move and do not force it back to * the previous cylinder group. */ if (dtog(fs, dbtofsb(fs, buflist->bs_children[0]->b_blkno)) != dtog(fs, dbtofsb(fs, buflist->bs_children[len - 1]->b_blkno))) return (ENOSPC); if (ufs_getlbns(vp, start_lbn, start_ap, &start_lvl) || ufs_getlbns(vp, end_lbn, end_ap, &end_lvl)) return (ENOSPC); /* * Get the starting offset and block map for the first block. */ if (start_lvl == 0) { sbap = &ip->i_din1->di_db[0]; soff = start_lbn; } else { idp = &start_ap[start_lvl - 1]; if (bread(vp, idp->in_lbn, (int)fs->fs_bsize, NOCRED, &sbp)) { brelse(sbp); return (ENOSPC); } sbap = (ufs1_daddr_t *)sbp->b_data; soff = idp->in_off; } /* * If the block range spans two block maps, get the second map. */ if (end_lvl == 0 || (idp = &end_ap[end_lvl - 1])->in_off + 1 >= len) { ssize = len; } else { #ifdef INVARIANTS if (start_lvl > 0 && start_ap[start_lvl - 1].in_lbn == idp->in_lbn) panic("ffs_reallocblk: start == end"); #endif ssize = len - (idp->in_off + 1); if (bread(vp, idp->in_lbn, (int)fs->fs_bsize, NOCRED, &ebp)) goto fail; ebap = (ufs1_daddr_t *)ebp->b_data; } /* * Find the preferred location for the cluster. */ UFS_LOCK(ump); pref = ffs_blkpref_ufs1(ip, start_lbn, soff, sbap); /* * Search the block map looking for an allocation of the desired size. */ if ((newblk = ffs_hashalloc(ip, dtog(fs, pref), pref, len, len, ffs_clusteralloc)) == 0) { UFS_UNLOCK(ump); goto fail; } /* * We have found a new contiguous block. * * First we have to replace the old block pointers with the new * block pointers in the inode and indirect blocks associated * with the file. */ #ifdef DEBUG if (prtrealloc) printf("realloc: ino %d, lbns %jd-%jd\n\told:", ip->i_number, (intmax_t)start_lbn, (intmax_t)end_lbn); #endif blkno = newblk; for (bap = &sbap[soff], i = 0; i < len; i++, blkno += fs->fs_frag) { if (i == ssize) { bap = ebap; soff = -i; } #ifdef INVARIANTS if (!ffs_checkblk(ip, dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize)) panic("ffs_reallocblks: unallocated block 2"); if (dbtofsb(fs, buflist->bs_children[i]->b_blkno) != *bap) panic("ffs_reallocblks: alloc mismatch"); #endif #ifdef DEBUG if (prtrealloc) printf(" %d,", *bap); #endif if (DOINGSOFTDEP(vp)) { if (sbap == &ip->i_din1->di_db[0] && i < ssize) softdep_setup_allocdirect(ip, start_lbn + i, blkno, *bap, fs->fs_bsize, fs->fs_bsize, buflist->bs_children[i]); else softdep_setup_allocindir_page(ip, start_lbn + i, i < ssize ? sbp : ebp, soff + i, blkno, *bap, buflist->bs_children[i]); } *bap++ = blkno; } /* * Next we must write out the modified inode and indirect blocks. * For strict correctness, the writes should be synchronous since * the old block values may have been written to disk. In practise * they are almost never written, but if we are concerned about * strict correctness, the `doasyncfree' flag should be set to zero. * * The test on `doasyncfree' should be changed to test a flag * that shows whether the associated buffers and inodes have * been written. The flag should be set when the cluster is * started and cleared whenever the buffer or inode is flushed. * We can then check below to see if it is set, and do the * synchronous write only when it has been cleared. */ if (sbap != &ip->i_din1->di_db[0]) { if (doasyncfree) bdwrite(sbp); else bwrite(sbp); } else { ip->i_flag |= IN_CHANGE | IN_UPDATE; if (!doasyncfree) ffs_update(vp, 1); } if (ssize < len) { if (doasyncfree) bdwrite(ebp); else bwrite(ebp); } /* * Last, free the old blocks and assign the new blocks to the buffers. */ #ifdef DEBUG if (prtrealloc) printf("\n\tnew:"); #endif for (blkno = newblk, i = 0; i < len; i++, blkno += fs->fs_frag) { if (!DOINGSOFTDEP(vp)) ffs_blkfree(ump, fs, ip->i_devvp, dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize, ip->i_number, NULL); buflist->bs_children[i]->b_blkno = fsbtodb(fs, blkno); #ifdef INVARIANTS if (!ffs_checkblk(ip, dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize)) panic("ffs_reallocblks: unallocated block 3"); #endif #ifdef DEBUG if (prtrealloc) printf(" %d,", blkno); #endif } #ifdef DEBUG if (prtrealloc) { prtrealloc--; printf("\n"); } #endif return (0); fail: if (ssize < len) brelse(ebp); if (sbap != &ip->i_din1->di_db[0]) brelse(sbp); return (ENOSPC); } static int ffs_reallocblks_ufs2(ap) struct vop_reallocblks_args /* { struct vnode *a_vp; struct cluster_save *a_buflist; } */ *ap; { struct fs *fs; struct inode *ip; struct vnode *vp; struct buf *sbp, *ebp; ufs2_daddr_t *bap, *sbap, *ebap = 0; struct cluster_save *buflist; struct ufsmount *ump; ufs_lbn_t start_lbn, end_lbn; ufs2_daddr_t soff, newblk, blkno, pref; struct indir start_ap[NIADDR + 1], end_ap[NIADDR + 1], *idp; int i, len, start_lvl, end_lvl, ssize; vp = ap->a_vp; ip = VTOI(vp); fs = ip->i_fs; ump = ip->i_ump; if (fs->fs_contigsumsize <= 0) return (ENOSPC); buflist = ap->a_buflist; len = buflist->bs_nchildren; start_lbn = buflist->bs_children[0]->b_lblkno; end_lbn = start_lbn + len - 1; #ifdef INVARIANTS for (i = 0; i < len; i++) if (!ffs_checkblk(ip, dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize)) panic("ffs_reallocblks: unallocated block 1"); for (i = 1; i < len; i++) if (buflist->bs_children[i]->b_lblkno != start_lbn + i) panic("ffs_reallocblks: non-logical cluster"); blkno = buflist->bs_children[0]->b_blkno; ssize = fsbtodb(fs, fs->fs_frag); for (i = 1; i < len - 1; i++) if (buflist->bs_children[i]->b_blkno != blkno + (i * ssize)) panic("ffs_reallocblks: non-physical cluster %d", i); #endif /* * If the latest allocation is in a new cylinder group, assume that * the filesystem has decided to move and do not force it back to * the previous cylinder group. */ if (dtog(fs, dbtofsb(fs, buflist->bs_children[0]->b_blkno)) != dtog(fs, dbtofsb(fs, buflist->bs_children[len - 1]->b_blkno))) return (ENOSPC); if (ufs_getlbns(vp, start_lbn, start_ap, &start_lvl) || ufs_getlbns(vp, end_lbn, end_ap, &end_lvl)) return (ENOSPC); /* * Get the starting offset and block map for the first block. */ if (start_lvl == 0) { sbap = &ip->i_din2->di_db[0]; soff = start_lbn; } else { idp = &start_ap[start_lvl - 1]; if (bread(vp, idp->in_lbn, (int)fs->fs_bsize, NOCRED, &sbp)) { brelse(sbp); return (ENOSPC); } sbap = (ufs2_daddr_t *)sbp->b_data; soff = idp->in_off; } /* * If the block range spans two block maps, get the second map. */ if (end_lvl == 0 || (idp = &end_ap[end_lvl - 1])->in_off + 1 >= len) { ssize = len; } else { #ifdef INVARIANTS if (start_lvl > 0 && start_ap[start_lvl - 1].in_lbn == idp->in_lbn) panic("ffs_reallocblk: start == end"); #endif ssize = len - (idp->in_off + 1); if (bread(vp, idp->in_lbn, (int)fs->fs_bsize, NOCRED, &ebp)) goto fail; ebap = (ufs2_daddr_t *)ebp->b_data; } /* * Find the preferred location for the cluster. */ UFS_LOCK(ump); pref = ffs_blkpref_ufs2(ip, start_lbn, soff, sbap); /* * Search the block map looking for an allocation of the desired size. */ if ((newblk = ffs_hashalloc(ip, dtog(fs, pref), pref, len, len, ffs_clusteralloc)) == 0) { UFS_UNLOCK(ump); goto fail; } /* * We have found a new contiguous block. * * First we have to replace the old block pointers with the new * block pointers in the inode and indirect blocks associated * with the file. */ #ifdef DEBUG if (prtrealloc) printf("realloc: ino %d, lbns %jd-%jd\n\told:", ip->i_number, (intmax_t)start_lbn, (intmax_t)end_lbn); #endif blkno = newblk; for (bap = &sbap[soff], i = 0; i < len; i++, blkno += fs->fs_frag) { if (i == ssize) { bap = ebap; soff = -i; } #ifdef INVARIANTS if (!ffs_checkblk(ip, dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize)) panic("ffs_reallocblks: unallocated block 2"); if (dbtofsb(fs, buflist->bs_children[i]->b_blkno) != *bap) panic("ffs_reallocblks: alloc mismatch"); #endif #ifdef DEBUG if (prtrealloc) printf(" %jd,", (intmax_t)*bap); #endif if (DOINGSOFTDEP(vp)) { if (sbap == &ip->i_din2->di_db[0] && i < ssize) softdep_setup_allocdirect(ip, start_lbn + i, blkno, *bap, fs->fs_bsize, fs->fs_bsize, buflist->bs_children[i]); else softdep_setup_allocindir_page(ip, start_lbn + i, i < ssize ? sbp : ebp, soff + i, blkno, *bap, buflist->bs_children[i]); } *bap++ = blkno; } /* * Next we must write out the modified inode and indirect blocks. * For strict correctness, the writes should be synchronous since * the old block values may have been written to disk. In practise * they are almost never written, but if we are concerned about * strict correctness, the `doasyncfree' flag should be set to zero. * * The test on `doasyncfree' should be changed to test a flag * that shows whether the associated buffers and inodes have * been written. The flag should be set when the cluster is * started and cleared whenever the buffer or inode is flushed. * We can then check below to see if it is set, and do the * synchronous write only when it has been cleared. */ if (sbap != &ip->i_din2->di_db[0]) { if (doasyncfree) bdwrite(sbp); else bwrite(sbp); } else { ip->i_flag |= IN_CHANGE | IN_UPDATE; if (!doasyncfree) ffs_update(vp, 1); } if (ssize < len) { if (doasyncfree) bdwrite(ebp); else bwrite(ebp); } /* * Last, free the old blocks and assign the new blocks to the buffers. */ #ifdef DEBUG if (prtrealloc) printf("\n\tnew:"); #endif for (blkno = newblk, i = 0; i < len; i++, blkno += fs->fs_frag) { if (!DOINGSOFTDEP(vp)) ffs_blkfree(ump, fs, ip->i_devvp, dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize, ip->i_number, NULL); buflist->bs_children[i]->b_blkno = fsbtodb(fs, blkno); #ifdef INVARIANTS if (!ffs_checkblk(ip, dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize)) panic("ffs_reallocblks: unallocated block 3"); #endif #ifdef DEBUG if (prtrealloc) printf(" %jd,", (intmax_t)blkno); #endif } #ifdef DEBUG if (prtrealloc) { prtrealloc--; printf("\n"); } #endif return (0); fail: if (ssize < len) brelse(ebp); if (sbap != &ip->i_din2->di_db[0]) brelse(sbp); return (ENOSPC); } /* * Allocate an inode in the filesystem. * * If allocating a directory, use ffs_dirpref to select the inode. * If allocating in a directory, the following hierarchy is followed: * 1) allocate the preferred inode. * 2) allocate an inode in the same cylinder group. * 3) quadradically rehash into other cylinder groups, until an * available inode is located. * If no inode preference is given the following hierarchy is used * to allocate an inode: * 1) allocate an inode in cylinder group 0. * 2) quadradically rehash into other cylinder groups, until an * available inode is located. */ int ffs_valloc(pvp, mode, cred, vpp) struct vnode *pvp; int mode; struct ucred *cred; struct vnode **vpp; { struct inode *pip; struct fs *fs; struct inode *ip; struct timespec ts; struct ufsmount *ump; ino_t ino, ipref; u_int cg; int error, error1, reclaimed; static struct timeval lastfail; static int curfail; *vpp = NULL; pip = VTOI(pvp); fs = pip->i_fs; ump = pip->i_ump; UFS_LOCK(ump); reclaimed = 0; retry: if (fs->fs_cstotal.cs_nifree == 0) goto noinodes; if ((mode & IFMT) == IFDIR) ipref = ffs_dirpref(pip); else ipref = pip->i_number; if (ipref >= fs->fs_ncg * fs->fs_ipg) ipref = 0; cg = ino_to_cg(fs, ipref); /* * Track number of dirs created one after another * in a same cg without intervening by files. */ if ((mode & IFMT) == IFDIR) { if (fs->fs_contigdirs[cg] < 255) fs->fs_contigdirs[cg]++; } else { if (fs->fs_contigdirs[cg] > 0) fs->fs_contigdirs[cg]--; } ino = (ino_t)ffs_hashalloc(pip, cg, ipref, mode, 0, (allocfcn_t *)ffs_nodealloccg); if (ino == 0) goto noinodes; error = ffs_vget(pvp->v_mount, ino, LK_EXCLUSIVE, vpp); if (error) { error1 = ffs_vgetf(pvp->v_mount, ino, LK_EXCLUSIVE, vpp, FFSV_FORCEINSMQ); ffs_vfree(pvp, ino, mode); if (error1 == 0) { ip = VTOI(*vpp); if (ip->i_mode) goto dup_alloc; ip->i_flag |= IN_MODIFIED; vput(*vpp); } return (error); } ip = VTOI(*vpp); if (ip->i_mode) { dup_alloc: printf("mode = 0%o, inum = %lu, fs = %s\n", ip->i_mode, (u_long)ip->i_number, fs->fs_fsmnt); panic("ffs_valloc: dup alloc"); } if (DIP(ip, i_blocks) && (fs->fs_flags & FS_UNCLEAN) == 0) { /* XXX */ printf("free inode %s/%lu had %ld blocks\n", fs->fs_fsmnt, (u_long)ino, (long)DIP(ip, i_blocks)); DIP_SET(ip, i_blocks, 0); } ip->i_flags = 0; DIP_SET(ip, i_flags, 0); /* * Set up a new generation number for this inode. */ if (ip->i_gen == 0 || ++ip->i_gen == 0) ip->i_gen = arc4random() / 2 + 1; DIP_SET(ip, i_gen, ip->i_gen); if (fs->fs_magic == FS_UFS2_MAGIC) { vfs_timestamp(&ts); ip->i_din2->di_birthtime = ts.tv_sec; ip->i_din2->di_birthnsec = ts.tv_nsec; } ufs_prepare_reclaim(*vpp); ip->i_flag = 0; (*vpp)->v_vflag = 0; (*vpp)->v_type = VNON; if (fs->fs_magic == FS_UFS2_MAGIC) (*vpp)->v_op = &ffs_vnodeops2; else (*vpp)->v_op = &ffs_vnodeops1; return (0); noinodes: if (reclaimed == 0) { reclaimed = 1; softdep_request_cleanup(fs, pvp, cred, FLUSH_INODES_WAIT); goto retry; } UFS_UNLOCK(ump); if (ppsratecheck(&lastfail, &curfail, 1)) { ffs_fserr(fs, pip->i_number, "out of inodes"); uprintf("\n%s: create/symlink failed, no inodes free\n", fs->fs_fsmnt); } return (ENOSPC); } /* * Find a cylinder group to place a directory. * * The policy implemented by this algorithm is to allocate a * directory inode in the same cylinder group as its parent * directory, but also to reserve space for its files inodes * and data. Restrict the number of directories which may be * allocated one after another in the same cylinder group * without intervening allocation of files. * * If we allocate a first level directory then force allocation * in another cylinder group. */ static ino_t ffs_dirpref(pip) struct inode *pip; { struct fs *fs; u_int cg, prefcg, dirsize, cgsize; u_int avgifree, avgbfree, avgndir, curdirsize; u_int minifree, minbfree, maxndir; u_int mincg, minndir; u_int maxcontigdirs; mtx_assert(UFS_MTX(pip->i_ump), MA_OWNED); fs = pip->i_fs; avgifree = fs->fs_cstotal.cs_nifree / fs->fs_ncg; avgbfree = fs->fs_cstotal.cs_nbfree / fs->fs_ncg; avgndir = fs->fs_cstotal.cs_ndir / fs->fs_ncg; /* * Force allocation in another cg if creating a first level dir. */ ASSERT_VOP_LOCKED(ITOV(pip), "ffs_dirpref"); if (ITOV(pip)->v_vflag & VV_ROOT) { prefcg = arc4random() % fs->fs_ncg; mincg = prefcg; minndir = fs->fs_ipg; for (cg = prefcg; cg < fs->fs_ncg; cg++) if (fs->fs_cs(fs, cg).cs_ndir < minndir && fs->fs_cs(fs, cg).cs_nifree >= avgifree && fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) { mincg = cg; minndir = fs->fs_cs(fs, cg).cs_ndir; } for (cg = 0; cg < prefcg; cg++) if (fs->fs_cs(fs, cg).cs_ndir < minndir && fs->fs_cs(fs, cg).cs_nifree >= avgifree && fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) { mincg = cg; minndir = fs->fs_cs(fs, cg).cs_ndir; } return ((ino_t)(fs->fs_ipg * mincg)); } /* * Count various limits which used for * optimal allocation of a directory inode. */ maxndir = min(avgndir + fs->fs_ipg / 16, fs->fs_ipg); minifree = avgifree - avgifree / 4; if (minifree < 1) minifree = 1; minbfree = avgbfree - avgbfree / 4; if (minbfree < 1) minbfree = 1; cgsize = fs->fs_fsize * fs->fs_fpg; dirsize = fs->fs_avgfilesize * fs->fs_avgfpdir; curdirsize = avgndir ? (cgsize - avgbfree * fs->fs_bsize) / avgndir : 0; if (dirsize < curdirsize) dirsize = curdirsize; if (dirsize <= 0) maxcontigdirs = 0; /* dirsize overflowed */ else maxcontigdirs = min((avgbfree * fs->fs_bsize) / dirsize, 255); if (fs->fs_avgfpdir > 0) maxcontigdirs = min(maxcontigdirs, fs->fs_ipg / fs->fs_avgfpdir); if (maxcontigdirs == 0) maxcontigdirs = 1; /* * Limit number of dirs in one cg and reserve space for * regular files, but only if we have no deficit in * inodes or space. */ prefcg = ino_to_cg(fs, pip->i_number); for (cg = prefcg; cg < fs->fs_ncg; cg++) if (fs->fs_cs(fs, cg).cs_ndir < maxndir && fs->fs_cs(fs, cg).cs_nifree >= minifree && fs->fs_cs(fs, cg).cs_nbfree >= minbfree) { if (fs->fs_contigdirs[cg] < maxcontigdirs) return ((ino_t)(fs->fs_ipg * cg)); } for (cg = 0; cg < prefcg; cg++) if (fs->fs_cs(fs, cg).cs_ndir < maxndir && fs->fs_cs(fs, cg).cs_nifree >= minifree && fs->fs_cs(fs, cg).cs_nbfree >= minbfree) { if (fs->fs_contigdirs[cg] < maxcontigdirs) return ((ino_t)(fs->fs_ipg * cg)); } /* * This is a backstop when we have deficit in space. */ for (cg = prefcg; cg < fs->fs_ncg; cg++) if (fs->fs_cs(fs, cg).cs_nifree >= avgifree) return ((ino_t)(fs->fs_ipg * cg)); for (cg = 0; cg < prefcg; cg++) if (fs->fs_cs(fs, cg).cs_nifree >= avgifree) break; return ((ino_t)(fs->fs_ipg * cg)); } /* * Select the desired position for the next block in a file. The file is * logically divided into sections. The first section is composed of the * direct blocks. Each additional section contains fs_maxbpg blocks. * * If no blocks have been allocated in the first section, the policy is to * request a block in the same cylinder group as the inode that describes * the file. If no blocks have been allocated in any other section, the * policy is to place the section in a cylinder group with a greater than * average number of free blocks. An appropriate cylinder group is found * by using a rotor that sweeps the cylinder groups. When a new group of * blocks is needed, the sweep begins in the cylinder group following the * cylinder group from which the previous allocation was made. The sweep * continues until a cylinder group with greater than the average number * of free blocks is found. If the allocation is for the first block in an * indirect block, the information on the previous allocation is unavailable; * here a best guess is made based upon the logical block number being * allocated. * * If a section is already partially allocated, the policy is to * contiguously allocate fs_maxcontig blocks. The end of one of these * contiguous blocks and the beginning of the next is laid out * contiguously if possible. */ ufs2_daddr_t ffs_blkpref_ufs1(ip, lbn, indx, bap) struct inode *ip; ufs_lbn_t lbn; int indx; ufs1_daddr_t *bap; { struct fs *fs; u_int cg; u_int avgbfree, startcg; mtx_assert(UFS_MTX(ip->i_ump), MA_OWNED); fs = ip->i_fs; if (indx % fs->fs_maxbpg == 0 || bap[indx - 1] == 0) { if (lbn < NDADDR + NINDIR(fs)) { cg = ino_to_cg(fs, ip->i_number); return (cgbase(fs, cg) + fs->fs_frag); } /* * Find a cylinder with greater than average number of * unused data blocks. */ if (indx == 0 || bap[indx - 1] == 0) startcg = ino_to_cg(fs, ip->i_number) + lbn / fs->fs_maxbpg; else startcg = dtog(fs, bap[indx - 1]) + 1; startcg %= fs->fs_ncg; avgbfree = fs->fs_cstotal.cs_nbfree / fs->fs_ncg; for (cg = startcg; cg < fs->fs_ncg; cg++) if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) { fs->fs_cgrotor = cg; return (cgbase(fs, cg) + fs->fs_frag); } for (cg = 0; cg <= startcg; cg++) if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) { fs->fs_cgrotor = cg; return (cgbase(fs, cg) + fs->fs_frag); } return (0); } /* * We just always try to lay things out contiguously. */ return (bap[indx - 1] + fs->fs_frag); } /* * Same as above, but for UFS2 */ ufs2_daddr_t ffs_blkpref_ufs2(ip, lbn, indx, bap) struct inode *ip; ufs_lbn_t lbn; int indx; ufs2_daddr_t *bap; { struct fs *fs; u_int cg; u_int avgbfree, startcg; mtx_assert(UFS_MTX(ip->i_ump), MA_OWNED); fs = ip->i_fs; if (indx % fs->fs_maxbpg == 0 || bap[indx - 1] == 0) { if (lbn < NDADDR + NINDIR(fs)) { cg = ino_to_cg(fs, ip->i_number); return (cgbase(fs, cg) + fs->fs_frag); } /* * Find a cylinder with greater than average number of * unused data blocks. */ if (indx == 0 || bap[indx - 1] == 0) startcg = ino_to_cg(fs, ip->i_number) + lbn / fs->fs_maxbpg; else startcg = dtog(fs, bap[indx - 1]) + 1; startcg %= fs->fs_ncg; avgbfree = fs->fs_cstotal.cs_nbfree / fs->fs_ncg; for (cg = startcg; cg < fs->fs_ncg; cg++) if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) { fs->fs_cgrotor = cg; return (cgbase(fs, cg) + fs->fs_frag); } for (cg = 0; cg <= startcg; cg++) if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) { fs->fs_cgrotor = cg; return (cgbase(fs, cg) + fs->fs_frag); } return (0); } /* * We just always try to lay things out contiguously. */ return (bap[indx - 1] + fs->fs_frag); } /* * Implement the cylinder overflow algorithm. * * The policy implemented by this algorithm is: * 1) allocate the block in its requested cylinder group. * 2) quadradically rehash on the cylinder group number. * 3) brute force search for a free block. * * Must be called with the UFS lock held. Will release the lock on success * and return with it held on failure. */ /*VARARGS5*/ static ufs2_daddr_t ffs_hashalloc(ip, cg, pref, size, rsize, allocator) struct inode *ip; u_int cg; ufs2_daddr_t pref; int size; /* Search size for data blocks, mode for inodes */ int rsize; /* Real allocated size. */ allocfcn_t *allocator; { struct fs *fs; ufs2_daddr_t result; u_int i, icg = cg; mtx_assert(UFS_MTX(ip->i_ump), MA_OWNED); #ifdef INVARIANTS if (ITOV(ip)->v_mount->mnt_kern_flag & MNTK_SUSPENDED) panic("ffs_hashalloc: allocation on suspended filesystem"); #endif fs = ip->i_fs; /* * 1: preferred cylinder group */ result = (*allocator)(ip, cg, pref, size, rsize); if (result) return (result); /* * 2: quadratic rehash */ for (i = 1; i < fs->fs_ncg; i *= 2) { cg += i; if (cg >= fs->fs_ncg) cg -= fs->fs_ncg; result = (*allocator)(ip, cg, 0, size, rsize); if (result) return (result); } /* * 3: brute force search * Note that we start at i == 2, since 0 was checked initially, * and 1 is always checked in the quadratic rehash. */ cg = (icg + 2) % fs->fs_ncg; for (i = 2; i < fs->fs_ncg; i++) { result = (*allocator)(ip, cg, 0, size, rsize); if (result) return (result); cg++; if (cg == fs->fs_ncg) cg = 0; } return (0); } /* * Determine whether a fragment can be extended. * * Check to see if the necessary fragments are available, and * if they are, allocate them. */ static ufs2_daddr_t ffs_fragextend(ip, cg, bprev, osize, nsize) struct inode *ip; u_int cg; ufs2_daddr_t bprev; int osize, nsize; { struct fs *fs; struct cg *cgp; struct buf *bp; struct ufsmount *ump; int nffree; long bno; int frags, bbase; int i, error; u_int8_t *blksfree; ump = ip->i_ump; fs = ip->i_fs; if (fs->fs_cs(fs, cg).cs_nffree < numfrags(fs, nsize - osize)) return (0); frags = numfrags(fs, nsize); bbase = fragnum(fs, bprev); if (bbase > fragnum(fs, (bprev + frags - 1))) { /* cannot extend across a block boundary */ return (0); } UFS_UNLOCK(ump); error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)), (int)fs->fs_cgsize, NOCRED, &bp); if (error) goto fail; cgp = (struct cg *)bp->b_data; if (!cg_chkmagic(cgp)) goto fail; bp->b_xflags |= BX_BKGRDWRITE; cgp->cg_old_time = cgp->cg_time = time_second; bno = dtogd(fs, bprev); blksfree = cg_blksfree(cgp); for (i = numfrags(fs, osize); i < frags; i++) if (isclr(blksfree, bno + i)) goto fail; /* * the current fragment can be extended * deduct the count on fragment being extended into * increase the count on the remaining fragment (if any) * allocate the extended piece */ for (i = frags; i < fs->fs_frag - bbase; i++) if (isclr(blksfree, bno + i)) break; cgp->cg_frsum[i - numfrags(fs, osize)]--; if (i != frags) cgp->cg_frsum[i - frags]++; for (i = numfrags(fs, osize), nffree = 0; i < frags; i++) { clrbit(blksfree, bno + i); cgp->cg_cs.cs_nffree--; nffree++; } UFS_LOCK(ump); fs->fs_cstotal.cs_nffree -= nffree; fs->fs_cs(fs, cg).cs_nffree -= nffree; fs->fs_fmod = 1; ACTIVECLEAR(fs, cg); UFS_UNLOCK(ump); if (DOINGSOFTDEP(ITOV(ip))) softdep_setup_blkmapdep(bp, UFSTOVFS(ump), bprev, frags, numfrags(fs, osize)); bdwrite(bp); return (bprev); fail: brelse(bp); UFS_LOCK(ump); return (0); } /* * Determine whether a block can be allocated. * * Check to see if a block of the appropriate size is available, * and if it is, allocate it. */ static ufs2_daddr_t ffs_alloccg(ip, cg, bpref, size, rsize) struct inode *ip; u_int cg; ufs2_daddr_t bpref; int size; int rsize; { struct fs *fs; struct cg *cgp; struct buf *bp; struct ufsmount *ump; ufs1_daddr_t bno; ufs2_daddr_t blkno; int i, allocsiz, error, frags; u_int8_t *blksfree; ump = ip->i_ump; fs = ip->i_fs; if (fs->fs_cs(fs, cg).cs_nbfree == 0 && size == fs->fs_bsize) return (0); UFS_UNLOCK(ump); error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)), (int)fs->fs_cgsize, NOCRED, &bp); if (error) goto fail; cgp = (struct cg *)bp->b_data; if (!cg_chkmagic(cgp) || (cgp->cg_cs.cs_nbfree == 0 && size == fs->fs_bsize)) goto fail; bp->b_xflags |= BX_BKGRDWRITE; cgp->cg_old_time = cgp->cg_time = time_second; if (size == fs->fs_bsize) { UFS_LOCK(ump); blkno = ffs_alloccgblk(ip, bp, bpref, rsize); ACTIVECLEAR(fs, cg); UFS_UNLOCK(ump); bdwrite(bp); return (blkno); } /* * check to see if any fragments are already available * allocsiz is the size which will be allocated, hacking * it down to a smaller size if necessary */ blksfree = cg_blksfree(cgp); frags = numfrags(fs, size); for (allocsiz = frags; allocsiz < fs->fs_frag; allocsiz++) if (cgp->cg_frsum[allocsiz] != 0) break; if (allocsiz == fs->fs_frag) { /* * no fragments were available, so a block will be * allocated, and hacked up */ if (cgp->cg_cs.cs_nbfree == 0) goto fail; UFS_LOCK(ump); blkno = ffs_alloccgblk(ip, bp, bpref, rsize); ACTIVECLEAR(fs, cg); UFS_UNLOCK(ump); bdwrite(bp); return (blkno); } KASSERT(size == rsize, ("ffs_alloccg: size(%d) != rsize(%d)", size, rsize)); bno = ffs_mapsearch(fs, cgp, bpref, allocsiz); if (bno < 0) goto fail; for (i = 0; i < frags; i++) clrbit(blksfree, bno + i); cgp->cg_cs.cs_nffree -= frags; cgp->cg_frsum[allocsiz]--; if (frags != allocsiz) cgp->cg_frsum[allocsiz - frags]++; UFS_LOCK(ump); fs->fs_cstotal.cs_nffree -= frags; fs->fs_cs(fs, cg).cs_nffree -= frags; fs->fs_fmod = 1; blkno = cgbase(fs, cg) + bno; ACTIVECLEAR(fs, cg); UFS_UNLOCK(ump); if (DOINGSOFTDEP(ITOV(ip))) softdep_setup_blkmapdep(bp, UFSTOVFS(ump), blkno, frags, 0); bdwrite(bp); return (blkno); fail: brelse(bp); UFS_LOCK(ump); return (0); } /* * Allocate a block in a cylinder group. * * This algorithm implements the following policy: * 1) allocate the requested block. * 2) allocate a rotationally optimal block in the same cylinder. * 3) allocate the next available block on the block rotor for the * specified cylinder group. * Note that this routine only allocates fs_bsize blocks; these * blocks may be fragmented by the routine that allocates them. */ static ufs2_daddr_t ffs_alloccgblk(ip, bp, bpref, size) struct inode *ip; struct buf *bp; ufs2_daddr_t bpref; int size; { struct fs *fs; struct cg *cgp; struct ufsmount *ump; ufs1_daddr_t bno; ufs2_daddr_t blkno; u_int8_t *blksfree; int i; fs = ip->i_fs; ump = ip->i_ump; mtx_assert(UFS_MTX(ump), MA_OWNED); cgp = (struct cg *)bp->b_data; blksfree = cg_blksfree(cgp); if (bpref == 0 || dtog(fs, bpref) != cgp->cg_cgx) { bpref = cgp->cg_rotor; } else { bpref = blknum(fs, bpref); bno = dtogd(fs, bpref); /* * if the requested block is available, use it */ if (ffs_isblock(fs, blksfree, fragstoblks(fs, bno))) goto gotit; } /* * Take the next available block in this cylinder group. */ bno = ffs_mapsearch(fs, cgp, bpref, (int)fs->fs_frag); if (bno < 0) return (0); cgp->cg_rotor = bno; gotit: blkno = fragstoblks(fs, bno); ffs_clrblock(fs, blksfree, (long)blkno); ffs_clusteracct(fs, cgp, blkno, -1); cgp->cg_cs.cs_nbfree--; fs->fs_cstotal.cs_nbfree--; fs->fs_cs(fs, cgp->cg_cgx).cs_nbfree--; fs->fs_fmod = 1; blkno = cgbase(fs, cgp->cg_cgx) + bno; /* * If the caller didn't want the whole block free the frags here. */ size = numfrags(fs, size); if (size != fs->fs_frag) { bno = dtogd(fs, blkno); for (i = size; i < fs->fs_frag; i++) setbit(blksfree, bno + i); i = fs->fs_frag - size; cgp->cg_cs.cs_nffree += i; fs->fs_cstotal.cs_nffree += i; fs->fs_cs(fs, cgp->cg_cgx).cs_nffree += i; fs->fs_fmod = 1; cgp->cg_frsum[i]++; } /* XXX Fixme. */ UFS_UNLOCK(ump); if (DOINGSOFTDEP(ITOV(ip))) softdep_setup_blkmapdep(bp, UFSTOVFS(ump), blkno, size, 0); UFS_LOCK(ump); return (blkno); } /* * Determine whether a cluster can be allocated. * * We do not currently check for optimal rotational layout if there * are multiple choices in the same cylinder group. Instead we just * take the first one that we find following bpref. */ static ufs2_daddr_t ffs_clusteralloc(ip, cg, bpref, len, unused) struct inode *ip; u_int cg; ufs2_daddr_t bpref; int len; int unused; { struct fs *fs; struct cg *cgp; struct buf *bp; struct ufsmount *ump; int i, run, bit, map, got; ufs2_daddr_t bno; u_char *mapp; int32_t *lp; u_int8_t *blksfree; fs = ip->i_fs; ump = ip->i_ump; if (fs->fs_maxcluster[cg] < len) return (0); UFS_UNLOCK(ump); if (bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)), (int)fs->fs_cgsize, NOCRED, &bp)) goto fail_lock; cgp = (struct cg *)bp->b_data; if (!cg_chkmagic(cgp)) goto fail_lock; bp->b_xflags |= BX_BKGRDWRITE; /* * Check to see if a cluster of the needed size (or bigger) is * available in this cylinder group. */ lp = &cg_clustersum(cgp)[len]; for (i = len; i <= fs->fs_contigsumsize; i++) if (*lp++ > 0) break; if (i > fs->fs_contigsumsize) { /* * This is the first time looking for a cluster in this * cylinder group. Update the cluster summary information * to reflect the true maximum sized cluster so that * future cluster allocation requests can avoid reading * the cylinder group map only to find no clusters. */ lp = &cg_clustersum(cgp)[len - 1]; for (i = len - 1; i > 0; i--) if (*lp-- > 0) break; UFS_LOCK(ump); fs->fs_maxcluster[cg] = i; goto fail; } /* * Search the cluster map to find a big enough cluster. * We take the first one that we find, even if it is larger * than we need as we prefer to get one close to the previous * block allocation. We do not search before the current * preference point as we do not want to allocate a block * that is allocated before the previous one (as we will * then have to wait for another pass of the elevator * algorithm before it will be read). We prefer to fail and * be recalled to try an allocation in the next cylinder group. */ if (dtog(fs, bpref) != cg) bpref = 0; else bpref = fragstoblks(fs, dtogd(fs, blknum(fs, bpref))); mapp = &cg_clustersfree(cgp)[bpref / NBBY]; map = *mapp++; bit = 1 << (bpref % NBBY); for (run = 0, got = bpref; got < cgp->cg_nclusterblks; got++) { if ((map & bit) == 0) { run = 0; } else { run++; if (run == len) break; } if ((got & (NBBY - 1)) != (NBBY - 1)) { bit <<= 1; } else { map = *mapp++; bit = 1; } } if (got >= cgp->cg_nclusterblks) goto fail_lock; /* * Allocate the cluster that we have found. */ blksfree = cg_blksfree(cgp); for (i = 1; i <= len; i++) if (!ffs_isblock(fs, blksfree, got - run + i)) panic("ffs_clusteralloc: map mismatch"); bno = cgbase(fs, cg) + blkstofrags(fs, got - run + 1); if (dtog(fs, bno) != cg) panic("ffs_clusteralloc: allocated out of group"); len = blkstofrags(fs, len); UFS_LOCK(ump); for (i = 0; i < len; i += fs->fs_frag) if (ffs_alloccgblk(ip, bp, bno + i, fs->fs_bsize) != bno + i) panic("ffs_clusteralloc: lost block"); ACTIVECLEAR(fs, cg); UFS_UNLOCK(ump); bdwrite(bp); return (bno); fail_lock: UFS_LOCK(ump); fail: brelse(bp); return (0); } /* * Determine whether an inode can be allocated. * * Check to see if an inode is available, and if it is, * allocate it using the following policy: * 1) allocate the requested inode. * 2) allocate the next available inode after the requested * inode in the specified cylinder group. */ static ufs2_daddr_t ffs_nodealloccg(ip, cg, ipref, mode, unused) struct inode *ip; u_int cg; ufs2_daddr_t ipref; int mode; int unused; { struct fs *fs; struct cg *cgp; struct buf *bp, *ibp; struct ufsmount *ump; u_int8_t *inosused; struct ufs2_dinode *dp2; int error, start, len, loc, map, i; fs = ip->i_fs; ump = ip->i_ump; if (fs->fs_cs(fs, cg).cs_nifree == 0) return (0); UFS_UNLOCK(ump); error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)), (int)fs->fs_cgsize, NOCRED, &bp); if (error) { brelse(bp); UFS_LOCK(ump); return (0); } cgp = (struct cg *)bp->b_data; if (!cg_chkmagic(cgp) || cgp->cg_cs.cs_nifree == 0) { brelse(bp); UFS_LOCK(ump); return (0); } bp->b_xflags |= BX_BKGRDWRITE; cgp->cg_old_time = cgp->cg_time = time_second; inosused = cg_inosused(cgp); if (ipref) { ipref %= fs->fs_ipg; if (isclr(inosused, ipref)) goto gotit; } start = cgp->cg_irotor / NBBY; len = howmany(fs->fs_ipg - cgp->cg_irotor, NBBY); loc = skpc(0xff, len, &inosused[start]); if (loc == 0) { len = start + 1; start = 0; loc = skpc(0xff, len, &inosused[0]); if (loc == 0) { printf("cg = %d, irotor = %ld, fs = %s\n", cg, (long)cgp->cg_irotor, fs->fs_fsmnt); panic("ffs_nodealloccg: map corrupted"); /* NOTREACHED */ } } i = start + len - loc; map = inosused[i] ^ 0xff; if (map == 0) { printf("fs = %s\n", fs->fs_fsmnt); panic("ffs_nodealloccg: block not in map"); } ipref = i * NBBY + ffs(map) - 1; cgp->cg_irotor = ipref; gotit: /* * Check to see if we need to initialize more inodes. */ ibp = NULL; if (fs->fs_magic == FS_UFS2_MAGIC && ipref + INOPB(fs) > cgp->cg_initediblk && cgp->cg_initediblk < cgp->cg_niblk) { ibp = getblk(ip->i_devvp, fsbtodb(fs, ino_to_fsba(fs, cg * fs->fs_ipg + cgp->cg_initediblk)), (int)fs->fs_bsize, 0, 0, 0); bzero(ibp->b_data, (int)fs->fs_bsize); dp2 = (struct ufs2_dinode *)(ibp->b_data); for (i = 0; i < INOPB(fs); i++) { dp2->di_gen = arc4random() / 2 + 1; dp2++; } cgp->cg_initediblk += INOPB(fs); } UFS_LOCK(ump); ACTIVECLEAR(fs, cg); setbit(inosused, ipref); cgp->cg_cs.cs_nifree--; fs->fs_cstotal.cs_nifree--; fs->fs_cs(fs, cg).cs_nifree--; fs->fs_fmod = 1; if ((mode & IFMT) == IFDIR) { cgp->cg_cs.cs_ndir++; fs->fs_cstotal.cs_ndir++; fs->fs_cs(fs, cg).cs_ndir++; } UFS_UNLOCK(ump); if (DOINGSOFTDEP(ITOV(ip))) softdep_setup_inomapdep(bp, ip, cg * fs->fs_ipg + ipref); bdwrite(bp); if (ibp != NULL) bawrite(ibp); return ((ino_t)(cg * fs->fs_ipg + ipref)); } /* * Free a block or fragment. * * The specified block or fragment is placed back in the * free map. If a fragment is deallocated, a possible * block reassembly is checked. */ static void ffs_blkfree_cg(ump, fs, devvp, bno, size, inum, dephd) struct ufsmount *ump; struct fs *fs; struct vnode *devvp; ufs2_daddr_t bno; long size; ino_t inum; struct workhead *dephd; { struct mount *mp; struct cg *cgp; struct buf *bp; ufs1_daddr_t fragno, cgbno; ufs2_daddr_t cgblkno; int i, blk, frags, bbase; u_int cg; u_int8_t *blksfree; struct cdev *dev; cg = dtog(fs, bno); if (devvp->v_type == VREG) { /* devvp is a snapshot */ dev = VTOI(devvp)->i_devvp->v_rdev; cgblkno = fragstoblks(fs, cgtod(fs, cg)); } else { /* devvp is a normal disk device */ dev = devvp->v_rdev; cgblkno = fsbtodb(fs, cgtod(fs, cg)); ASSERT_VOP_LOCKED(devvp, "ffs_blkfree_cg"); } #ifdef INVARIANTS if ((u_int)size > fs->fs_bsize || fragoff(fs, size) != 0 || fragnum(fs, bno) + numfrags(fs, size) > fs->fs_frag) { printf("dev=%s, bno = %jd, bsize = %ld, size = %ld, fs = %s\n", devtoname(dev), (intmax_t)bno, (long)fs->fs_bsize, size, fs->fs_fsmnt); panic("ffs_blkfree: bad size"); } #endif if ((u_int)bno >= fs->fs_size) { printf("bad block %jd, ino %lu\n", (intmax_t)bno, (u_long)inum); ffs_fserr(fs, inum, "bad block"); return; } if (bread(devvp, cgblkno, (int)fs->fs_cgsize, NOCRED, &bp)) { brelse(bp); return; } cgp = (struct cg *)bp->b_data; if (!cg_chkmagic(cgp)) { brelse(bp); return; } bp->b_xflags |= BX_BKGRDWRITE; cgp->cg_old_time = cgp->cg_time = time_second; cgbno = dtogd(fs, bno); blksfree = cg_blksfree(cgp); UFS_LOCK(ump); if (size == fs->fs_bsize) { fragno = fragstoblks(fs, cgbno); if (!ffs_isfreeblock(fs, blksfree, fragno)) { if (devvp->v_type == VREG) { UFS_UNLOCK(ump); /* devvp is a snapshot */ brelse(bp); return; } printf("dev = %s, block = %jd, fs = %s\n", devtoname(dev), (intmax_t)bno, fs->fs_fsmnt); panic("ffs_blkfree: freeing free block"); } ffs_setblock(fs, blksfree, fragno); ffs_clusteracct(fs, cgp, fragno, 1); cgp->cg_cs.cs_nbfree++; fs->fs_cstotal.cs_nbfree++; fs->fs_cs(fs, cg).cs_nbfree++; } else { bbase = cgbno - fragnum(fs, cgbno); /* * decrement the counts associated with the old frags */ blk = blkmap(fs, blksfree, bbase); ffs_fragacct(fs, blk, cgp->cg_frsum, -1); /* * deallocate the fragment */ frags = numfrags(fs, size); for (i = 0; i < frags; i++) { if (isset(blksfree, cgbno + i)) { printf("dev = %s, block = %jd, fs = %s\n", devtoname(dev), (intmax_t)(bno + i), fs->fs_fsmnt); panic("ffs_blkfree: freeing free frag"); } setbit(blksfree, cgbno + i); } cgp->cg_cs.cs_nffree += i; fs->fs_cstotal.cs_nffree += i; fs->fs_cs(fs, cg).cs_nffree += i; /* * add back in counts associated with the new frags */ blk = blkmap(fs, blksfree, bbase); ffs_fragacct(fs, blk, cgp->cg_frsum, 1); /* * if a complete block has been reassembled, account for it */ fragno = fragstoblks(fs, bbase); if (ffs_isblock(fs, blksfree, fragno)) { cgp->cg_cs.cs_nffree -= fs->fs_frag; fs->fs_cstotal.cs_nffree -= fs->fs_frag; fs->fs_cs(fs, cg).cs_nffree -= fs->fs_frag; ffs_clusteracct(fs, cgp, fragno, 1); cgp->cg_cs.cs_nbfree++; fs->fs_cstotal.cs_nbfree++; fs->fs_cs(fs, cg).cs_nbfree++; } } fs->fs_fmod = 1; ACTIVECLEAR(fs, cg); UFS_UNLOCK(ump); mp = UFSTOVFS(ump); if (mp->mnt_flag & MNT_SOFTDEP && devvp->v_type != VREG) softdep_setup_blkfree(UFSTOVFS(ump), bp, bno, numfrags(fs, size), dephd); bdwrite(bp); } TASKQUEUE_DEFINE_THREAD(ffs_trim); struct ffs_blkfree_trim_params { struct task task; struct ufsmount *ump; struct vnode *devvp; ufs2_daddr_t bno; long size; ino_t inum; struct workhead *pdephd; struct workhead dephd; }; static void ffs_blkfree_trim_task(ctx, pending) void *ctx; int pending; { struct ffs_blkfree_trim_params *tp; tp = ctx; ffs_blkfree_cg(tp->ump, tp->ump->um_fs, tp->devvp, tp->bno, tp->size, tp->inum, tp->pdephd); vn_finished_secondary_write(UFSTOVFS(tp->ump)); free(tp, M_TEMP); } static void ffs_blkfree_trim_completed(bip) struct bio *bip; { struct ffs_blkfree_trim_params *tp; tp = bip->bio_caller2; g_destroy_bio(bip); TASK_INIT(&tp->task, 0, ffs_blkfree_trim_task, tp); taskqueue_enqueue(taskqueue_ffs_trim, &tp->task); } void ffs_blkfree(ump, fs, devvp, bno, size, inum, dephd) struct ufsmount *ump; struct fs *fs; struct vnode *devvp; ufs2_daddr_t bno; long size; ino_t inum; struct workhead *dephd; { struct mount *mp; struct bio *bip; struct ffs_blkfree_trim_params *tp; /* * Check to see if a snapshot wants to claim the block. * Check that devvp is a normal disk device, not a snapshot, * it has a snapshot(s) associated with it, and one of the * snapshots wants to claim the block. */ if (devvp->v_type != VREG && (devvp->v_vflag & VV_COPYONWRITE) && ffs_snapblkfree(fs, devvp, bno, size, inum)) { return; } if (!ump->um_candelete) { ffs_blkfree_cg(ump, fs, devvp, bno, size, inum, dephd); return; } /* * Postpone the set of the free bit in the cg bitmap until the * BIO_DELETE is completed. Otherwise, due to disk queue * reordering, TRIM might be issued after we reuse the block * and write some new data into it. */ tp = malloc(sizeof(struct ffs_blkfree_trim_params), M_TEMP, M_WAITOK); tp->ump = ump; tp->devvp = devvp; tp->bno = bno; tp->size = size; tp->inum = inum; if (dephd != NULL) { LIST_INIT(&tp->dephd); LIST_SWAP(dephd, &tp->dephd, worklist, wk_list); tp->pdephd = &tp->dephd; } else tp->pdephd = NULL; bip = g_alloc_bio(); bip->bio_cmd = BIO_DELETE; bip->bio_offset = dbtob(fsbtodb(fs, bno)); bip->bio_done = ffs_blkfree_trim_completed; bip->bio_length = size; bip->bio_caller2 = tp; mp = UFSTOVFS(ump); vn_start_secondary_write(NULL, &mp, 0); g_io_request(bip, (struct g_consumer *)devvp->v_bufobj.bo_private); } #ifdef INVARIANTS /* * Verify allocation of a block or fragment. Returns true if block or * fragment is allocated, false if it is free. */ static int ffs_checkblk(ip, bno, size) struct inode *ip; ufs2_daddr_t bno; long size; { struct fs *fs; struct cg *cgp; struct buf *bp; ufs1_daddr_t cgbno; int i, error, frags, free; u_int8_t *blksfree; fs = ip->i_fs; if ((u_int)size > fs->fs_bsize || fragoff(fs, size) != 0) { printf("bsize = %ld, size = %ld, fs = %s\n", (long)fs->fs_bsize, size, fs->fs_fsmnt); panic("ffs_checkblk: bad size"); } if ((u_int)bno >= fs->fs_size) panic("ffs_checkblk: bad block %jd", (intmax_t)bno); error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, dtog(fs, bno))), (int)fs->fs_cgsize, NOCRED, &bp); if (error) panic("ffs_checkblk: cg bread failed"); cgp = (struct cg *)bp->b_data; if (!cg_chkmagic(cgp)) panic("ffs_checkblk: cg magic mismatch"); bp->b_xflags |= BX_BKGRDWRITE; blksfree = cg_blksfree(cgp); cgbno = dtogd(fs, bno); if (size == fs->fs_bsize) { free = ffs_isblock(fs, blksfree, fragstoblks(fs, cgbno)); } else { frags = numfrags(fs, size); for (free = 0, i = 0; i < frags; i++) if (isset(blksfree, cgbno + i)) free++; if (free != 0 && free != frags) panic("ffs_checkblk: partially free fragment"); } brelse(bp); return (!free); } #endif /* INVARIANTS */ /* * Free an inode. */ int ffs_vfree(pvp, ino, mode) struct vnode *pvp; ino_t ino; int mode; { struct inode *ip; if (DOINGSOFTDEP(pvp)) { softdep_freefile(pvp, ino, mode); return (0); } ip = VTOI(pvp); return (ffs_freefile(ip->i_ump, ip->i_fs, ip->i_devvp, ino, mode, NULL)); } /* * Do the actual free operation. * The specified inode is placed back in the free map. */ int ffs_freefile(ump, fs, devvp, ino, mode, wkhd) struct ufsmount *ump; struct fs *fs; struct vnode *devvp; ino_t ino; int mode; struct workhead *wkhd; { struct cg *cgp; struct buf *bp; ufs2_daddr_t cgbno; int error; u_int cg; u_int8_t *inosused; struct cdev *dev; cg = ino_to_cg(fs, ino); if (devvp->v_type == VREG) { /* devvp is a snapshot */ dev = VTOI(devvp)->i_devvp->v_rdev; cgbno = fragstoblks(fs, cgtod(fs, cg)); } else { /* devvp is a normal disk device */ dev = devvp->v_rdev; cgbno = fsbtodb(fs, cgtod(fs, cg)); } if (ino >= fs->fs_ipg * fs->fs_ncg) panic("ffs_freefile: range: dev = %s, ino = %lu, fs = %s", devtoname(dev), (u_long)ino, fs->fs_fsmnt); if ((error = bread(devvp, cgbno, (int)fs->fs_cgsize, NOCRED, &bp))) { brelse(bp); return (error); } cgp = (struct cg *)bp->b_data; if (!cg_chkmagic(cgp)) { brelse(bp); return (0); } bp->b_xflags |= BX_BKGRDWRITE; cgp->cg_old_time = cgp->cg_time = time_second; inosused = cg_inosused(cgp); ino %= fs->fs_ipg; if (isclr(inosused, ino)) { printf("dev = %s, ino = %u, fs = %s\n", devtoname(dev), ino + cg * fs->fs_ipg, fs->fs_fsmnt); if (fs->fs_ronly == 0) panic("ffs_freefile: freeing free inode"); } clrbit(inosused, ino); if (ino < cgp->cg_irotor) cgp->cg_irotor = ino; cgp->cg_cs.cs_nifree++; UFS_LOCK(ump); fs->fs_cstotal.cs_nifree++; fs->fs_cs(fs, cg).cs_nifree++; if ((mode & IFMT) == IFDIR) { cgp->cg_cs.cs_ndir--; fs->fs_cstotal.cs_ndir--; fs->fs_cs(fs, cg).cs_ndir--; } fs->fs_fmod = 1; ACTIVECLEAR(fs, cg); UFS_UNLOCK(ump); if (UFSTOVFS(ump)->mnt_flag & MNT_SOFTDEP && devvp->v_type != VREG) softdep_setup_inofree(UFSTOVFS(ump), bp, ino + cg * fs->fs_ipg, wkhd); bdwrite(bp); return (0); } /* * Check to see if a file is free. */ int ffs_checkfreefile(fs, devvp, ino) struct fs *fs; struct vnode *devvp; ino_t ino; { struct cg *cgp; struct buf *bp; ufs2_daddr_t cgbno; int ret; u_int cg; u_int8_t *inosused; cg = ino_to_cg(fs, ino); if (devvp->v_type == VREG) { /* devvp is a snapshot */ cgbno = fragstoblks(fs, cgtod(fs, cg)); } else { /* devvp is a normal disk device */ cgbno = fsbtodb(fs, cgtod(fs, cg)); } if (ino >= fs->fs_ipg * fs->fs_ncg) return (1); if (bread(devvp, cgbno, (int)fs->fs_cgsize, NOCRED, &bp)) { brelse(bp); return (1); } cgp = (struct cg *)bp->b_data; if (!cg_chkmagic(cgp)) { brelse(bp); return (1); } inosused = cg_inosused(cgp); ino %= fs->fs_ipg; ret = isclr(inosused, ino); brelse(bp); return (ret); } /* * Find a block of the specified size in the specified cylinder group. * * It is a panic if a request is made to find a block if none are * available. */ static ufs1_daddr_t ffs_mapsearch(fs, cgp, bpref, allocsiz) struct fs *fs; struct cg *cgp; ufs2_daddr_t bpref; int allocsiz; { ufs1_daddr_t bno; int start, len, loc, i; int blk, field, subfield, pos; u_int8_t *blksfree; /* * find the fragment by searching through the free block * map for an appropriate bit pattern */ if (bpref) start = dtogd(fs, bpref) / NBBY; else start = cgp->cg_frotor / NBBY; blksfree = cg_blksfree(cgp); len = howmany(fs->fs_fpg, NBBY) - start; loc = scanc((u_int)len, (u_char *)&blksfree[start], fragtbl[fs->fs_frag], (u_char)(1 << (allocsiz - 1 + (fs->fs_frag % NBBY)))); if (loc == 0) { len = start + 1; start = 0; loc = scanc((u_int)len, (u_char *)&blksfree[0], fragtbl[fs->fs_frag], (u_char)(1 << (allocsiz - 1 + (fs->fs_frag % NBBY)))); if (loc == 0) { printf("start = %d, len = %d, fs = %s\n", start, len, fs->fs_fsmnt); panic("ffs_alloccg: map corrupted"); /* NOTREACHED */ } } bno = (start + len - loc) * NBBY; cgp->cg_frotor = bno; /* * found the byte in the map * sift through the bits to find the selected frag */ for (i = bno + NBBY; bno < i; bno += fs->fs_frag) { blk = blkmap(fs, blksfree, bno); blk <<= 1; field = around[allocsiz]; subfield = inside[allocsiz]; for (pos = 0; pos <= fs->fs_frag - allocsiz; pos++) { if ((blk & field) == subfield) return (bno + pos); field <<= 1; subfield <<= 1; } } printf("bno = %lu, fs = %s\n", (u_long)bno, fs->fs_fsmnt); panic("ffs_alloccg: block not in map"); return (-1); } /* * Fserr prints the name of a filesystem with an error diagnostic. * * The form of the error message is: * fs: error message */ static void ffs_fserr(fs, inum, cp) struct fs *fs; ino_t inum; char *cp; { struct thread *td = curthread; /* XXX */ struct proc *p = td->td_proc; log(LOG_ERR, "pid %d (%s), uid %d inumber %d on %s: %s\n", p->p_pid, p->p_comm, td->td_ucred->cr_uid, inum, fs->fs_fsmnt, cp); } /* * This function provides the capability for the fsck program to * update an active filesystem. Fourteen operations are provided: * * adjrefcnt(inode, amt) - adjusts the reference count on the * specified inode by the specified amount. Under normal * operation the count should always go down. Decrementing * the count to zero will cause the inode to be freed. - * adjblkcnt(inode, amt) - adjust the number of blocks used to - * by the specifed amount. + * adjblkcnt(inode, amt) - adjust the number of blocks used by the + * inode by the specified amount. * adjndir, adjbfree, adjifree, adjffree, adjnumclusters(amt) - * adjust the superblock summary. * freedirs(inode, count) - directory inodes [inode..inode + count - 1] * are marked as free. Inodes should never have to be marked * as in use. * freefiles(inode, count) - file inodes [inode..inode + count - 1] * are marked as free. Inodes should never have to be marked * as in use. * freeblks(blockno, size) - blocks [blockno..blockno + size - 1] * are marked as free. Blocks should never have to be marked * as in use. * setflags(flags, set/clear) - the fs_flags field has the specified * flags set (second parameter +1) or cleared (second parameter -1). * setcwd(dirinode) - set the current directory to dirinode in the * filesystem associated with the snapshot. * setdotdot(oldvalue, newvalue) - Verify that the inode number for ".." * in the current directory is oldvalue then change it to newvalue. * unlink(nameptr, oldvalue) - Verify that the inode number associated * with nameptr in the current directory is oldvalue then unlink it. */ static int sysctl_ffs_fsck(SYSCTL_HANDLER_ARGS); SYSCTL_PROC(_vfs_ffs, FFS_ADJ_REFCNT, adjrefcnt, CTLFLAG_WR|CTLTYPE_STRUCT, 0, 0, sysctl_ffs_fsck, "S,fsck", "Adjust Inode Reference Count"); static SYSCTL_NODE(_vfs_ffs, FFS_ADJ_BLKCNT, adjblkcnt, CTLFLAG_WR, sysctl_ffs_fsck, "Adjust Inode Used Blocks Count"); static SYSCTL_NODE(_vfs_ffs, FFS_ADJ_NDIR, adjndir, CTLFLAG_WR, sysctl_ffs_fsck, "Adjust number of directories"); static SYSCTL_NODE(_vfs_ffs, FFS_ADJ_NBFREE, adjnbfree, CTLFLAG_WR, sysctl_ffs_fsck, "Adjust number of free blocks"); static SYSCTL_NODE(_vfs_ffs, FFS_ADJ_NIFREE, adjnifree, CTLFLAG_WR, sysctl_ffs_fsck, "Adjust number of free inodes"); static SYSCTL_NODE(_vfs_ffs, FFS_ADJ_NFFREE, adjnffree, CTLFLAG_WR, sysctl_ffs_fsck, "Adjust number of free frags"); static SYSCTL_NODE(_vfs_ffs, FFS_ADJ_NUMCLUSTERS, adjnumclusters, CTLFLAG_WR, sysctl_ffs_fsck, "Adjust number of free clusters"); static SYSCTL_NODE(_vfs_ffs, FFS_DIR_FREE, freedirs, CTLFLAG_WR, sysctl_ffs_fsck, "Free Range of Directory Inodes"); static SYSCTL_NODE(_vfs_ffs, FFS_FILE_FREE, freefiles, CTLFLAG_WR, sysctl_ffs_fsck, "Free Range of File Inodes"); static SYSCTL_NODE(_vfs_ffs, FFS_BLK_FREE, freeblks, CTLFLAG_WR, sysctl_ffs_fsck, "Free Range of Blocks"); static SYSCTL_NODE(_vfs_ffs, FFS_SET_FLAGS, setflags, CTLFLAG_WR, sysctl_ffs_fsck, "Change Filesystem Flags"); static SYSCTL_NODE(_vfs_ffs, FFS_SET_CWD, setcwd, CTLFLAG_WR, sysctl_ffs_fsck, "Set Current Working Directory"); static SYSCTL_NODE(_vfs_ffs, FFS_SET_DOTDOT, setdotdot, CTLFLAG_WR, sysctl_ffs_fsck, "Change Value of .. Entry"); static SYSCTL_NODE(_vfs_ffs, FFS_UNLINK, unlink, CTLFLAG_WR, sysctl_ffs_fsck, "Unlink a Duplicate Name"); #ifdef DEBUG static int fsckcmds = 0; SYSCTL_INT(_debug, OID_AUTO, fsckcmds, CTLFLAG_RW, &fsckcmds, 0, ""); #endif /* DEBUG */ static int sysctl_ffs_fsck(SYSCTL_HANDLER_ARGS) { struct thread *td = curthread; struct fsck_cmd cmd; struct ufsmount *ump; struct vnode *vp, *vpold, *dvp, *fdvp; struct inode *ip, *dp; struct mount *mp; struct fs *fs; ufs2_daddr_t blkno; long blkcnt, blksize; struct filedesc *fdp; struct file *fp; int vfslocked, filetype, error; if (req->newlen > sizeof cmd) return (EBADRPC); if ((error = SYSCTL_IN(req, &cmd, sizeof cmd)) != 0) return (error); if (cmd.version != FFS_CMD_VERSION) return (ERPCMISMATCH); if ((error = getvnode(curproc->p_fd, cmd.handle, &fp)) != 0) return (error); vp = fp->f_data; if (vp->v_type != VREG && vp->v_type != VDIR) { fdrop(fp, td); return (EINVAL); } vn_start_write(vp, &mp, V_WAIT); if (mp == 0 || strncmp(mp->mnt_stat.f_fstypename, "ufs", MFSNAMELEN)) { vn_finished_write(mp); fdrop(fp, td); return (EINVAL); } if (mp->mnt_flag & MNT_RDONLY) { vn_finished_write(mp); fdrop(fp, td); return (EROFS); } ump = VFSTOUFS(mp); fs = ump->um_fs; filetype = IFREG; switch (oidp->oid_number) { case FFS_SET_FLAGS: #ifdef DEBUG if (fsckcmds) printf("%s: %s flags\n", mp->mnt_stat.f_mntonname, cmd.size > 0 ? "set" : "clear"); #endif /* DEBUG */ if (cmd.size > 0) fs->fs_flags |= (long)cmd.value; else fs->fs_flags &= ~(long)cmd.value; break; case FFS_ADJ_REFCNT: #ifdef DEBUG if (fsckcmds) { printf("%s: adjust inode %jd count by %jd\n", mp->mnt_stat.f_mntonname, (intmax_t)cmd.value, (intmax_t)cmd.size); } #endif /* DEBUG */ if ((error = ffs_vget(mp, (ino_t)cmd.value, LK_EXCLUSIVE, &vp))) break; ip = VTOI(vp); ip->i_nlink += cmd.size; DIP_SET(ip, i_nlink, ip->i_nlink); ip->i_effnlink += cmd.size; ip->i_flag |= IN_CHANGE; if (DOINGSOFTDEP(vp)) softdep_change_linkcnt(ip); vput(vp); break; case FFS_ADJ_BLKCNT: #ifdef DEBUG if (fsckcmds) { printf("%s: adjust inode %jd block count by %jd\n", mp->mnt_stat.f_mntonname, (intmax_t)cmd.value, (intmax_t)cmd.size); } #endif /* DEBUG */ if ((error = ffs_vget(mp, (ino_t)cmd.value, LK_EXCLUSIVE, &vp))) break; ip = VTOI(vp); DIP_SET(ip, i_blocks, DIP(ip, i_blocks) + cmd.size); ip->i_flag |= IN_CHANGE; vput(vp); break; case FFS_DIR_FREE: filetype = IFDIR; /* fall through */ case FFS_FILE_FREE: #ifdef DEBUG if (fsckcmds) { if (cmd.size == 1) printf("%s: free %s inode %d\n", mp->mnt_stat.f_mntonname, filetype == IFDIR ? "directory" : "file", (ino_t)cmd.value); else printf("%s: free %s inodes %d-%d\n", mp->mnt_stat.f_mntonname, filetype == IFDIR ? "directory" : "file", (ino_t)cmd.value, (ino_t)(cmd.value + cmd.size - 1)); } #endif /* DEBUG */ while (cmd.size > 0) { if ((error = ffs_freefile(ump, fs, ump->um_devvp, cmd.value, filetype, NULL))) break; cmd.size -= 1; cmd.value += 1; } break; case FFS_BLK_FREE: #ifdef DEBUG if (fsckcmds) { if (cmd.size == 1) printf("%s: free block %jd\n", mp->mnt_stat.f_mntonname, (intmax_t)cmd.value); else printf("%s: free blocks %jd-%jd\n", mp->mnt_stat.f_mntonname, (intmax_t)cmd.value, (intmax_t)cmd.value + cmd.size - 1); } #endif /* DEBUG */ blkno = cmd.value; blkcnt = cmd.size; blksize = fs->fs_frag - (blkno % fs->fs_frag); while (blkcnt > 0) { if (blksize > blkcnt) blksize = blkcnt; ffs_blkfree(ump, fs, ump->um_devvp, blkno, blksize * fs->fs_fsize, ROOTINO, NULL); blkno += blksize; blkcnt -= blksize; blksize = fs->fs_frag; } break; /* * Adjust superblock summaries. fsck(8) is expected to * submit deltas when necessary. */ case FFS_ADJ_NDIR: #ifdef DEBUG if (fsckcmds) { printf("%s: adjust number of directories by %jd\n", mp->mnt_stat.f_mntonname, (intmax_t)cmd.value); } #endif /* DEBUG */ fs->fs_cstotal.cs_ndir += cmd.value; break; case FFS_ADJ_NBFREE: #ifdef DEBUG if (fsckcmds) { printf("%s: adjust number of free blocks by %+jd\n", mp->mnt_stat.f_mntonname, (intmax_t)cmd.value); } #endif /* DEBUG */ fs->fs_cstotal.cs_nbfree += cmd.value; break; case FFS_ADJ_NIFREE: #ifdef DEBUG if (fsckcmds) { printf("%s: adjust number of free inodes by %+jd\n", mp->mnt_stat.f_mntonname, (intmax_t)cmd.value); } #endif /* DEBUG */ fs->fs_cstotal.cs_nifree += cmd.value; break; case FFS_ADJ_NFFREE: #ifdef DEBUG if (fsckcmds) { printf("%s: adjust number of free frags by %+jd\n", mp->mnt_stat.f_mntonname, (intmax_t)cmd.value); } #endif /* DEBUG */ fs->fs_cstotal.cs_nffree += cmd.value; break; case FFS_ADJ_NUMCLUSTERS: #ifdef DEBUG if (fsckcmds) { printf("%s: adjust number of free clusters by %+jd\n", mp->mnt_stat.f_mntonname, (intmax_t)cmd.value); } #endif /* DEBUG */ fs->fs_cstotal.cs_numclusters += cmd.value; break; case FFS_SET_CWD: #ifdef DEBUG if (fsckcmds) { printf("%s: set current directory to inode %jd\n", mp->mnt_stat.f_mntonname, (intmax_t)cmd.value); } #endif /* DEBUG */ if ((error = ffs_vget(mp, (ino_t)cmd.value, LK_SHARED, &vp))) break; vfslocked = VFS_LOCK_GIANT(vp->v_mount); AUDIT_ARG_VNODE1(vp); if ((error = change_dir(vp, td)) != 0) { vput(vp); VFS_UNLOCK_GIANT(vfslocked); break; } VOP_UNLOCK(vp, 0); VFS_UNLOCK_GIANT(vfslocked); fdp = td->td_proc->p_fd; FILEDESC_XLOCK(fdp); vpold = fdp->fd_cdir; fdp->fd_cdir = vp; FILEDESC_XUNLOCK(fdp); vfslocked = VFS_LOCK_GIANT(vpold->v_mount); vrele(vpold); VFS_UNLOCK_GIANT(vfslocked); break; case FFS_SET_DOTDOT: #ifdef DEBUG if (fsckcmds) { printf("%s: change .. in cwd from %jd to %jd\n", mp->mnt_stat.f_mntonname, (intmax_t)cmd.value, (intmax_t)cmd.size); } #endif /* DEBUG */ /* * First we have to get and lock the parent directory * to which ".." points. */ error = ffs_vget(mp, (ino_t)cmd.value, LK_EXCLUSIVE, &fdvp); if (error) break; /* * Now we get and lock the child directory containing "..". */ FILEDESC_SLOCK(td->td_proc->p_fd); dvp = td->td_proc->p_fd->fd_cdir; FILEDESC_SUNLOCK(td->td_proc->p_fd); if ((error = vget(dvp, LK_EXCLUSIVE, td)) != 0) { vput(fdvp); break; } dp = VTOI(dvp); dp->i_offset = 12; /* XXX mastertemplate.dot_reclen */ error = ufs_dirrewrite(dp, VTOI(fdvp), (ino_t)cmd.size, DT_DIR, 0); cache_purge(fdvp); cache_purge(dvp); vput(dvp); vput(fdvp); break; case FFS_UNLINK: #ifdef DEBUG if (fsckcmds) { char buf[32]; if (copyinstr((char *)(intptr_t)cmd.value, buf,32,NULL)) strncpy(buf, "Name_too_long", 32); printf("%s: unlink %s (inode %jd)\n", mp->mnt_stat.f_mntonname, buf, (intmax_t)cmd.size); } #endif /* DEBUG */ /* * kern_unlinkat will do its own start/finish writes and * they do not nest, so drop ours here. Setting mp == NULL * indicates that vn_finished_write is not needed down below. */ vn_finished_write(mp); mp = NULL; error = kern_unlinkat(td, AT_FDCWD, (char *)(intptr_t)cmd.value, UIO_USERSPACE, (ino_t)cmd.size); break; default: #ifdef DEBUG if (fsckcmds) { printf("Invalid request %d from fsck\n", oidp->oid_number); } #endif /* DEBUG */ error = EINVAL; break; } fdrop(fp, td); vn_finished_write(mp); return (error); } Index: projects/largeSMP/sys =================================================================== --- projects/largeSMP/sys (revision 222782) +++ projects/largeSMP/sys (revision 222783) Property changes on: projects/largeSMP/sys ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys:r222698-222782 Index: projects/largeSMP/tools/regression/bin/sh/expansion/heredoc1.0 =================================================================== --- projects/largeSMP/tools/regression/bin/sh/expansion/heredoc1.0 (nonexistent) +++ projects/largeSMP/tools/regression/bin/sh/expansion/heredoc1.0 (revision 222783) @@ -0,0 +1,25 @@ +# $FreeBSD$ + +f() { return $1; } + +[ `f 42; { cat; } < __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include "pathnames.h" #include "calendar.h" const char *calendarFile = "calendar"; /* default calendar file */ const char *calendarHomes[] = {".calendar", _PATH_INCLUDE}; /* HOME */ const char *calendarNoMail = "nomail"; /* don't sent mail if this file exist */ char path[MAXPATHLEN]; struct fixs neaster, npaskha, ncny, nfullmoon, nnewmoon; struct fixs nmarequinox, nsepequinox, njunsolstice, ndecsolstice; #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) { char *pp, p; FILE *fp; int ch, l; int count, 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; char dbuf[80]; 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 ((fp = opencal()) == NULL) { free(extradata); return; } while (fgets(buf, sizeof(buf), stdin) != NULL) { if ((pp = strchr(buf, '\n')) != NULL) *pp = '\0'; else /* Flush this line */ while ((ch = getchar()) != '\n' && ch != EOF); for (l = strlen(buf); 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 */ 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]); } } event_print_all(fp); closecal(fp); free(extradata); } FILE * opencal(void) { uid_t uid; size_t i; int fd, found, pdes[2]; struct stat sbuf; /* open up calendar file as stdin */ if (!freopen(calendarFile, "r", stdin)) { if (doall) { if (chdir(calendarHomes[0]) != 0) return (NULL); if (stat(calendarNoMail, &sbuf) == 0) return (NULL); if (!freopen(calendarFile, "r", stdin)) 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 && freopen(calendarFile, "r", stdin)) { found = 1; break; } if (!found) errx(1, "can't open calendar file \"%s\": %s (%d)", calendarFile, strerror(errno), errno); } } if (pipe(pdes) < 0) return (NULL); switch (fork()) { case -1: /* error */ (void)close(pdes[0]); (void)close(pdes[1]); return (NULL); case 0: /* child -- stdin already setup, set stdout to pipe input */ if (pdes[1] != STDOUT_FILENO) { (void)dup2(pdes[1], STDOUT_FILENO); (void)close(pdes[1]); } (void)close(pdes[0]); uid = geteuid(); if (setuid(getuid()) < 0) { warnx("first setuid failed"); _exit(1); }; if (setgid(getegid()) < 0) { warnx("setgid failed"); _exit(1); } if (setuid(uid) < 0) { warnx("setuid failed"); _exit(1); } execl(_PATH_CPP, "cpp", "-P", "-traditional", "-nostdinc", /* GCC specific opts */ "-I.", "-I", _PATH_INCLUDE, (char *)NULL); warn(_PATH_CPP); _exit(1); } /* parent -- set stdin to pipe output */ (void)dup2(pdes[0], STDIN_FILENO); (void)close(pdes[0]); (void)close(pdes[1]); /* 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 */ (void)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: ", 12); + 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/largeSMP/usr.bin/calendar =================================================================== --- projects/largeSMP/usr.bin/calendar (revision 222782) +++ projects/largeSMP/usr.bin/calendar (revision 222783) Property changes on: projects/largeSMP/usr.bin/calendar ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/usr.bin/calendar:r222698-222782 Index: projects/largeSMP/usr.bin/csup =================================================================== --- projects/largeSMP/usr.bin/csup (revision 222782) +++ projects/largeSMP/usr.bin/csup (revision 222783) Property changes on: projects/largeSMP/usr.bin/csup ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/usr.bin/csup:r222698-222782 Index: projects/largeSMP/usr.bin/grep/Makefile =================================================================== --- projects/largeSMP/usr.bin/grep/Makefile (revision 222782) +++ projects/largeSMP/usr.bin/grep/Makefile (revision 222783) @@ -1,48 +1,46 @@ # $NetBSD: Makefile,v 1.4 2011/02/16 01:31:33 joerg Exp $ # $FreeBSD$ # $OpenBSD: Makefile,v 1.6 2003/06/25 15:00:04 millert Exp $ .include .if ${MK_BSD_GREP} == "yes" PROG= grep .else PROG= bsdgrep .endif SRCS= fastgrep.c file.c grep.c queue.c util.c .if ${MK_BSD_GREP} == "yes" LINKS= ${BINDIR}/grep ${BINDIR}/egrep \ ${BINDIR}/grep ${BINDIR}/fgrep \ ${BINDIR}/grep ${BINDIR}/zgrep \ ${BINDIR}/grep ${BINDIR}/zegrep \ ${BINDIR}/grep ${BINDIR}/zfgrep MLINKS= grep.1 egrep.1 \ grep.1 fgrep.1 \ grep.1 zgrep.1 \ grep.1 zegrep.1 \ grep.1 zfgrep.1 .endif bsdgrep.1: grep.1 cp ${.ALLSRC} ${.TARGET} -WARNS?= 6 - LDADD= -lz -lbz2 DPADD= ${LIBZ} ${LIBBZ2} .if !defined(WITHOUT_GNU_COMPAT) CFLAGS+= -I/usr/include/gnu LDADD+= -lgnuregex DPADD+= ${LIBGNUREGEX} .endif .if !defined(WITHOUT_NLS) .include "${.CURDIR}/nls/Makefile.inc" .else CFLAGS+= -DWITHOUT_NLS .endif .include Index: projects/largeSMP/usr.bin/iconv/Makefile =================================================================== --- projects/largeSMP/usr.bin/iconv/Makefile (revision 222782) +++ projects/largeSMP/usr.bin/iconv/Makefile (revision 222783) @@ -1,15 +1,13 @@ # $FreeBSD$ # $NetBSD: Makefile,v 1.6 2009/04/14 22:15:21 lukem Exp $ .include PROG= iconv #SRCS= iconv.c MAN= iconv.1 -WARNS?= 6 - LDADD+= -lcrypt DPADD+= ${LIBCRYPT} .include Index: projects/largeSMP/usr.bin/kdump/mksubr =================================================================== --- projects/largeSMP/usr.bin/kdump/mksubr (revision 222782) +++ projects/largeSMP/usr.bin/kdump/mksubr (revision 222783) @@ -1,452 +1,452 @@ #!/bin/sh # # $FreeBSD$ # # Generates kdump_subr.c # mkioctls is a special-purpose script, and works fine as it is # now, so it remains independent. The idea behind how it generates # its list was heavily borrowed here. # # Some functions here are automatically generated. This can mean # the user will see unusual kdump output or errors while building # if the underlying .h files are changed significantly. # # Key: # AUTO: Completely auto-generated with either the "or" or the "switch" # method. # AUTO - Special: Generated automatically, but with some extra commands # that the auto_*_type() functions are inappropriate for. # MANUAL: Manually entered and must therefore be manually updated. set -e LC_ALL=C; export LC_ALL if [ -z "$1" ] then echo "usage: sh $0 include-dir" exit 1 fi include_dir=$1 # # Automatically generates a C function that will print out the # numeric input as a pipe-delimited string of the appropriate # #define keys. ex: # S_IRUSR|S_IWUSR|S_IRGRP|S_IROTH # The XOR is necessary to prevent including the "0"-value in every # line. # auto_or_type () { local name grep file name=$1 grep=$2 file=$3 cat <<_EOF_ /* AUTO */ void $name (int arg) { int or = 0; printf("%#x<", arg); _EOF_ egrep "^#[[:space:]]*define[[:space:]]+"${grep}"[[:space:]]*" \ $include_dir/$file | \ awk '{ for (i = 1; i <= NF; i++) \ if ($i ~ /define/) \ break; \ ++i; \ printf "\tif(!((arg>0)^((%s)>0)))\n\t\tif_print_or(arg, %s, or);\n", $i, $i }' cat <<_EOF_ printf(">"); if (or == 0) (void)printf("%ld", (long)arg); } _EOF_ } # # Automatically generates a C function used when the argument # maps to a single, specific #definition # auto_switch_type () { local name grep file name=$1 grep=$2 file=$3 cat <<_EOF_ /* AUTO */ void $name (int arg) { switch (arg) { _EOF_ egrep "^#[[:space:]]*define[[:space:]]+"${grep}"[[:space:]]*" \ $include_dir/$file | \ awk '{ for (i = 1; i <= NF; i++) \ if ($i ~ /define/) \ break; \ ++i; \ printf "\tcase %s:\n\t\t(void)printf(\"%s\");\n\t\tbreak;\n", $i, $i }' cat <<_EOF_ default: /* Should not reach */ (void)printf("", (long)arg); } } _EOF_ } # # Automatically generates a C function used when the argument # maps to a #definition # auto_if_type () { local name grep file name=$1 grep=$2 file=$3 cat <<_EOF_ /* AUTO */ void $name (int arg) { _EOF_ egrep "^#[[:space:]]*define[[:space:]]+"${grep}"[[:space:]]*" \ $include_dir/$file | \ awk '{ printf "\t"; \ if (NR > 1) \ printf "else " ; \ printf "if (arg == %s) \n\t\tprintf(\"%s\");\n", $2, $2 }' cat <<_EOF_ else /* Should not reach */ (void)printf("", (long)arg); } _EOF_ } # C start cat <<_EOF_ #include #include #include #include #include #include #define _KERNEL #include #undef _KERNEL #include #include #include #include #include #include #include #include #define _KERNEL #include #undef _KERNEL #include #include #include #include #include #include #include #include #include #include "kdump_subr.h" /* * These are simple support macros. print_or utilizes a variable * defined in the calling function to track whether or not it should * print a logical-OR character ('|') before a string. if_print_or * simply handles the necessary "if" statement used in many lines * of this file. */ #define print_or(str,orflag) do { \\ if (orflag) putchar('|'); else orflag = 1; \\ printf (str); } \\ while (0) #define if_print_or(i,flag,orflag) do { \\ if ((i & flag) == flag) \\ print_or(#flag,orflag); } \\ while (0) /* MANUAL */ extern char *signames[]; /* from kdump.c */ void signame (int sig) { if (sig > 0 && sig < NSIG) (void)printf("SIG%s",signames[sig]); else (void)printf("SIG %d", sig); } /* MANUAL */ void semctlname (int cmd) { switch (cmd) { case GETNCNT: (void)printf("GETNCNT"); break; case GETPID: (void)printf("GETPID"); break; case GETVAL: (void)printf("GETVAL"); break; case GETALL: (void)printf("GETALL"); break; case GETZCNT: (void)printf("GETZCNT"); break; case SETVAL: (void)printf("SETVAL"); break; case SETALL: (void)printf("SETALL"); break; case IPC_RMID: (void)printf("IPC_RMID"); break; case IPC_SET: (void)printf("IPC_SET"); break; case IPC_STAT: (void)printf("IPC_STAT"); break; default: /* Should not reach */ (void)printf("", (long)cmd); } } /* MANUAL */ void shmctlname (int cmd) { switch (cmd) { case IPC_RMID: (void)printf("IPC_RMID"); break; case IPC_SET: (void)printf("IPC_SET"); break; case IPC_STAT: (void)printf("IPC_STAT"); break; default: /* Should not reach */ (void)printf("", (long)cmd); } } /* MANUAL */ void semgetname (int flag) { int or = 0; if_print_or(flag, IPC_CREAT, or); if_print_or(flag, IPC_EXCL, or); if_print_or(flag, SEM_R, or); if_print_or(flag, SEM_A, or); if_print_or(flag, (SEM_R>>3), or); if_print_or(flag, (SEM_A>>3), or); if_print_or(flag, (SEM_R>>6), or); if_print_or(flag, (SEM_A>>6), or); } /* * MANUAL * * Only used by SYS_open. Unless O_CREAT is set in flags, the * mode argument is unused (and often bogus and misleading). */ void flagsandmodename (int flags, int mode, int decimal) { flagsname (flags); (void)putchar(','); if ((flags & O_CREAT) == O_CREAT) { modename (mode); } else { if (decimal) { (void)printf("%ld", (long)mode); } else { (void)printf("%#lx", (long)mode); } } } /* * MANUAL * * [g|s]etsockopt's level argument can either be SOL_SOCKET or a value * referring to a line in /etc/protocols . It might be appropriate * to use getprotoent(3) here. */ void sockoptlevelname (int level, int decimal) { if (level == SOL_SOCKET) { (void)printf("SOL_SOCKET"); } else { if (decimal) { (void)printf("%ld", (long)level); } else { (void)printf("%#lx", (long)level); } } } _EOF_ auto_or_type "modename" "S_[A-Z]+[[:space:]]+[0-6]{7}" "sys/stat.h" auto_or_type "flagsname" "O_[A-Z]+[[:space:]]+0x[0-9A-Fa-f]+" "sys/fcntl.h" auto_or_type "accessmodename" "[A-Z]_OK[[:space:]]+0?x?[0-9A-Fa-f]+" "sys/unistd.h" auto_or_type "mmapprotname" "PROT_[A-Z]+[[:space:]]+0x[0-9A-Fa-f]+" "sys/mman.h" auto_or_type "mmapflagsname" "MAP_[A-Z]+[[:space:]]+0x[0-9A-Fa-f]+" "sys/mman.h" auto_or_type "wait4optname" "W[A-Z]+[[:space:]]+[0-9]+" "sys/wait.h" auto_or_type "getfsstatflagsname" "MNT_[A-Z]+[[:space:]]+[1-9][0-9]*" "sys/mount.h" auto_or_type "mountflagsname" "MNT_[A-Z]+[[:space:]]+0x[0-9]+" "sys/mount.h" auto_or_type "rebootoptname" "RB_[A-Z]+[[:space:]]+0x[0-9]+" "sys/reboot.h" auto_or_type "flockname" "LOCK_[A-Z]+[[:space:]]+0x[0-9]+" "sys/fcntl.h" auto_or_type "thrcreateflagsname" "THR_[A-Z]+[[:space:]]+0x[0-9]+" "sys/thr.h" auto_or_type "mlockallname" "MCL_[A-Z]+[[:space:]]+0x[0-9]+" "sys/mman.h" auto_or_type "shmatname" "SHM_[A-Z]+[[:space:]]+[0-9]{6}+" "sys/shm.h" auto_or_type "rforkname" "RF[A-Z]+[[:space:]]+\([0-9]+<<[0-9]+\)" "sys/unistd.h" auto_or_type "nfssvcname" "NFSSVC_[A-Z]+[[:space:]]+0x[0-9]+" "nfsserver/nfs.h" auto_switch_type "whencename" "SEEK_[A-Z]+[[:space:]]+[0-9]+" "sys/unistd.h" auto_switch_type "rlimitname" "RLIMIT_[A-Z]+[[:space:]]+[0-9]+" "sys/resource.h" auto_switch_type "shutdownhowname" "SHUT_[A-Z]+[[:space:]]+[0-9]+" "sys/socket.h" auto_switch_type "prioname" "PRIO_[A-Z]+[[:space:]]+[0-9]" "sys/resource.h" auto_switch_type "madvisebehavname" "_?MADV_[A-Z]+[[:space:]]+[0-9]+" "sys/mman.h" auto_switch_type "msyncflagsname" "MS_[A-Z]+[[:space:]]+0x[0-9]+" "sys/mman.h" auto_switch_type "schedpolicyname" "SCHED_[A-Z]+[[:space:]]+[0-9]+" "sched.h" auto_switch_type "kldunloadfflagsname" "LINKER_UNLOAD_[A-Z]+[[:space:]]+[0-9]+" "sys/linker.h" auto_switch_type "extattrctlname" "EXTATTR_NAMESPACE_[A-Z]+[[:space:]]+0x[0-9]+" "sys/extattr.h" auto_switch_type "kldsymcmdname" "KLDSYM_[A-Z]+[[:space:]]+[0-9]+" "sys/linker.h" auto_switch_type "sendfileflagsname" "SF_[A-Z]+[[:space:]]+[0-9]+" "sys/socket.h" auto_switch_type "acltypename" "ACL_TYPE_[A-Z4_]+[[:space:]]+0x[0-9]+" "sys/acl.h" auto_switch_type "sigprocmaskhowname" "SIG_[A-Z]+[[:space:]]+[0-9]+" "sys/signal.h" auto_switch_type "lio_listioname" "LIO_(NO)?WAIT[[:space:]]+[0-9]+" "aio.h" auto_switch_type "minheritname" "INHERIT_[A-Z]+[[:space:]]+[0-9]+" "sys/mman.h" auto_switch_type "quotactlname" "Q_[A-Z]+[[:space:]]+0x[0-9]+" "ufs/ufs/quota.h" auto_if_type "sockdomainname" "PF_[[:alnum:]]+[[:space:]]+" "sys/socket.h" auto_if_type "sockfamilyname" "AF_[[:alnum:]]+[[:space:]]+" "sys/socket.h" auto_if_type "sockipprotoname" "IPPROTO_[[:alnum:]]+[[:space:]]+" "netinet/in.h" auto_switch_type "sockoptname" "SO_[A-Z]+[[:space:]]+0x[0-9]+" "sys/socket.h" auto_switch_type "socktypename" "SOCK_[A-Z]+[[:space:]]+[1-9]+[0-9]*" "sys/socket.h" -auto_switch_type "ptraceopname" "PT_[[:alnum:]]+[[:space:]]+[0-9]+" "sys/ptrace.h" +auto_switch_type "ptraceopname" "PT_[[:alnum:]_]+[[:space:]]+[0-9]+" "sys/ptrace.h" cat <<_EOF_ /* * AUTO - Special * F_ is used to specify fcntl commands as well as arguments. Both sets are * grouped in fcntl.h, and this awk script grabs the first group. */ void fcntlcmdname (int cmd, int arg, int decimal) { switch (cmd) { _EOF_ egrep "^#[[:space:]]*define[[:space:]]+F_[A-Z]+[[:space:]]+[0-9]+[[:space:]]*" \ $include_dir/sys/fcntl.h | \ awk 'BEGIN { o=0 } { for (i = 1; i <= NF; i++) \ if ($i ~ /define/) \ break; \ ++i; \ if (o <= $(i+1)) \ printf "\tcase %s:\n\t\t(void)printf(\"%s\");\n\t\tbreak;\n", $i, $i; \ else \ exit; \ o = $(i+1) }' cat <<_EOF_ default: /* Should not reach */ (void)printf("", (long)cmd); } (void)putchar(','); if (cmd == F_GETFD || cmd == F_SETFD) { if (arg == FD_CLOEXEC) (void)printf("FD_CLOEXEC"); else if (arg == 0) (void)printf("0"); else { if (decimal) (void)printf("%ld", (long)arg); else (void)printf("%#lx", (long)arg); } } else if (cmd == F_SETFL) { flagsname(arg); } else { if (decimal) (void)printf("%ld", (long)arg); else (void)printf("%#lx", (long)arg); } } /* * AUTO - Special * * The only reason this is not fully automated is due to the * grep -v RTP_PRIO statement. A better egrep line should * make this capable of being a auto_switch_type() function. */ void rtprioname (int func) { switch (func) { _EOF_ egrep "^#[[:space:]]*define[[:space:]]+RTP_[A-Z]+[[:space:]]+0x[0-9]+[[:space:]]*" \ $include_dir/sys/rtprio.h | grep -v RTP_PRIO | \ awk '{ for (i = 1; i <= NF; i++) \ if ($i ~ /define/) \ break; \ ++i; \ printf "\tcase %s:\n\t\t(void)printf(\"%s\");\n\t\tbreak;\n", $i, $i }' cat <<_EOF_ default: /* Should not reach */ (void)printf("", (long)func); } } /* * AUTO - Special * * The send and recv functions have a flags argument which can be * set to 0. There is no corresponding #define. The auto_ functions * detect this as "invalid", which is incorrect here. */ void sendrecvflagsname (int flags) { int or = 0; if (flags == 0) { (void)printf("0"); return; } printf("%#x<", flags); _EOF_ egrep "^#[[:space:]]*define[[:space:]]+MSG_[A-Z]+[[:space:]]+0x[0-9]+[[:space:]]*" $include_dir/sys/socket.h | \ awk '{ for (i = 1; i <= NF; i++) \ if ($i ~ /define/) \ break; \ ++i; \ printf "\tif(!((flags>0)^((%s)>0)))\n\t\tif_print_or(flags, %s, or);\n", $i, $i }' cat <<_EOF_ printf(">"); } _EOF_ Index: projects/largeSMP/usr.bin/procstat =================================================================== --- projects/largeSMP/usr.bin/procstat (revision 222782) +++ projects/largeSMP/usr.bin/procstat (revision 222783) Property changes on: projects/largeSMP/usr.bin/procstat ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/usr.bin/procstat:r222698-222782 Index: projects/largeSMP/usr.bin/rctl/Makefile =================================================================== --- projects/largeSMP/usr.bin/rctl/Makefile (revision 222782) +++ projects/largeSMP/usr.bin/rctl/Makefile (revision 222783) @@ -1,11 +1,9 @@ # $FreeBSD$ PROG= rctl MAN= rctl.8 DPADD= ${LIBUTIL} LDADD= -lutil -WARNS?= 6 - .include Index: projects/largeSMP/usr.bin/su/su.1 =================================================================== --- projects/largeSMP/usr.bin/su/su.1 (revision 222782) +++ projects/largeSMP/usr.bin/su/su.1 (revision 222783) @@ -1,242 +1,248 @@ .\" Copyright (c) 1988, 1990, 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. .\" .\" @(#)su.1 8.2 (Berkeley) 4/18/94 .\" $FreeBSD$ .\" -.Dd July 1, 2008 +.Dd June 6, 2011 .Dt SU 1 .Os .Sh NAME .Nm su .Nd substitute user identity .Sh SYNOPSIS .Nm .Op Fl .Op Fl flms .Op Fl c Ar class .Op Ar login Op Ar args .Sh DESCRIPTION The .Nm utility requests appropriate user credentials via PAM and switches to that user ID (the default user is the superuser). A shell is then executed. .Pp PAM is used to set the policy .Xr su 1 will use. In particular, by default only users in the .Dq Li wheel group can switch to UID 0 .Pq Dq Li root . This group requirement may be changed by modifying the .Dq Li pam_group section of .Pa /etc/pam.d/su . See .Xr pam_group 8 for details on how to modify this setting. .Pp By default, the environment is unmodified with the exception of .Ev USER , .Ev HOME , and .Ev SHELL . .Ev HOME and .Ev SHELL are set to the target login's default values. .Ev USER is set to the target login, unless the target login has a user ID of 0, in which case it is unmodified. The invoked shell is the one belonging to the target login. This is the traditional behavior of .Nm . Resource limits and session priority applicable to the original user's login class (see .Xr login.conf 5 ) are also normally retained unless the target login has a user ID of 0. .Pp The options are as follows: .Bl -tag -width Ds .It Fl f If the invoked shell is .Xr csh 1 , this option prevents it from reading the .Dq Pa .cshrc file. .It Fl l Simulate a full login. The environment is discarded except for .Ev HOME , .Ev SHELL , .Ev PATH , .Ev TERM , and .Ev USER . .Ev HOME and .Ev SHELL are modified as above. .Ev USER is set to the target login. .Ev PATH is set to .Dq Pa /bin:/usr/bin . .Ev TERM is imported from your current environment. Environment variables may be set or overridden from the login class capabilities database according to the class of the target login. The invoked shell is the target login's, and .Nm will change directory to the target login's home directory. Resource limits and session priority are modified to that for the target account's login class. .It Fl (no letter) The same as .Fl l . .It Fl m Leave the environment unmodified. The invoked shell is your login shell, and no directory changes are made. As a security precaution, if the target user's shell is a non-standard shell (as defined by .Xr getusershell 3 ) and the caller's real uid is non-zero, .Nm will fail. .It Fl s Set the MAC label to the user's default label as part of the user credential setup. Setting the MAC label may fail if the MAC label of the invoking process is not sufficient to transition to the user's default MAC label. If the label cannot be set, .Nm will fail. .It Fl c Ar class Use the settings of the specified login class. Only allowed for the super-user. .El .Pp The .Fl l (or .Fl ) and .Fl m options are mutually exclusive; the last one specified overrides any previous ones. .Pp If the optional .Ar args are provided on the command line, they are passed to the login shell of the target login. Note that all command line arguments before the target login name are processed by .Nm itself, everything after the target login name gets passed to the login shell. .Pp By default (unless the prompt is reset by a startup file) the super-user prompt is set to .Dq Sy \&# to remind one of its awesome power. .Sh ENVIRONMENT Environment variables used by .Nm : .Bl -tag -width HOME .It Ev HOME Default home directory of real user ID unless modified as specified above. .It Ev PATH Default search path of real user ID unless modified as specified above. .It Ev TERM Provides terminal type which may be retained for the substituted user ID. .It Ev USER The user ID is always the effective ID (the target user ID) after an .Nm unless the user ID is 0 (root). .El .Sh FILES .Bl -tag -width ".Pa /etc/pam.d/su" -compact .It Pa /etc/pam.d/su PAM configuration for .Nm . .El .Sh EXAMPLES .Bl -tag -width 5n -compact .It Li "su -m man -c catman" -Runs the command -.Li catman -as user -.Li man . +Starts a shell as user +.Li man , +and runs the command +.Li catman . You will be asked for man's password unless your real UID is 0. Note that the .Fl m option is required since user .Dq man does not have a valid shell by default. +In this example, +.Fl c +is passed to the shell of the user +.Dq man , +and is not interpreted as an argument to +.Nm . .It Li "su -m man -c 'catman /usr/share/man /usr/local/man'" Same as above, but the target command consists of more than a single word and hence is quoted for use with the .Fl c option being passed to the shell. (Most shells expect the argument to .Fl c to be a single word). .It Li "su -m -c staff man -c 'catman /usr/share/man /usr/local/man'" Same as above, but the target command is run with the resource limits of the login class .Dq staff . Note: in this example, the first .Fl c option applies to .Nm while the second is an argument to the shell being invoked. .It Li "su -l foo" Simulate a login for user foo. .It Li "su - foo" Same as above. .It Li "su -" Simulate a login for root. .El .Sh SEE ALSO .Xr csh 1 , .Xr sh 1 , .Xr group 5 , .Xr login.conf 5 , .Xr passwd 5 , .Xr environ 7 , .Xr pam_group 8 .Sh HISTORY A .Nm command appeared in .At v1 . Index: projects/largeSMP/usr.sbin/bluetooth/ath3kfw/Makefile =================================================================== --- projects/largeSMP/usr.sbin/bluetooth/ath3kfw/Makefile (revision 222782) +++ projects/largeSMP/usr.sbin/bluetooth/ath3kfw/Makefile (revision 222783) @@ -1,9 +1,8 @@ # $FreeBSD$ PROG= ath3kfw MAN= ath3kfw.8 -WARNS?= 6 DPADD+= ${LIBUSB} LDADD+= -lusb .include Index: projects/largeSMP/usr.sbin/bsdinstall/scripts/netconfig_ipv6 =================================================================== --- projects/largeSMP/usr.sbin/bsdinstall/scripts/netconfig_ipv6 (revision 222782) +++ projects/largeSMP/usr.sbin/bsdinstall/scripts/netconfig_ipv6 (revision 222783) @@ -1,149 +1,148 @@ #!/bin/sh #- # Copyright (c) 2011 Nathan Whitehorn # All rights reserved. # Copyright (c) 2011 The FreeBSD Foundation # All rights reserved. # # Portions of this software were developed by Bjoern Zeeb # 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. # # $FreeBSD$ # # TODO: -# - Add -R /sbin/resolvconf to rtsol once support is in tree. # - Add DHCPv6 support once FreeBSD ships with it. # : ${DIALOG_OK=0} : ${DIALOG_CANCEL=1} : ${DIALOG_HELP=2} : ${DIALOG_EXTRA=3} : ${DIALOG_ITEM_HELP=4} : ${DIALOG_ESC=255} INTERFACE=$1 case "${INTERFACE}" in "") dialog --backtitle 'FreeBSD Installer' --title 'Network Configuration' \ --msgbox 'No interface specified for IPv6 configuration.' 0 0 exit 1 ;; esac AGAIN="" while : ; do MSG="Would you like to try stateless address autoconfiguration (SLAAC)${AGAIN}?" dialog --backtitle 'FreeBSD Installer' --title 'Network Configuration' \ --yesno "${MSG}" 0 0 if [ $? -eq $DIALOG_OK ]; then if [ ! -z $BSDINSTALL_CONFIGCURRENT ]; then dialog --backtitle 'FreeBSD Installer' \ --infobox "Sending Router Solicitation ..." 0 0 ifconfig ${INTERFACE} inet6 -ifdisabled accept_rtadv up rtsol -F $INTERFACE 2>> $BSDINSTALL_LOG if [ $? -ne 0 ]; then dialog --backtitle 'FreeBSD Installer' --msgbox "SLAAC failed." 0 0 AGAIN=" again" continue fi fi echo ifconfig_${INTERFACE}_ipv6=\"inet6 accept_rtadv\" >> $BSDINSTALL_TMPETC/._rc.conf.net exit 0 else break fi done ROUTER6=`netstat -Wrn -f inet6 | awk '/default/ {printf("%s\n", $2);}'` ADDRS=`ifconfig ${INTERFACE} inet6 | \ awk -v dfr="${ROUTER6}" ' BEGIN { n=0; } { if (/inet6/) { if (match($2, "^fe80:")) { next; }; # For the moment ignore all but the first address; it might confuse the user. if (n > 0) { next; }; n++; printf "\"IPv6 Address\" %d 0 \"%s/%s\" %d 16 50 0 0 ", n, $2, $4, n; } } END { if (n == 0) { n++; printf "\"IPv6 Address\" %d 0 \"\" %d 16 50 0 0 ", n, n; } n++; # Nasty trick adding a (hidden, same y) read-only field as a marker # to separate interface address(es) from the default router. printf "\"Default Router\" %d 0 \"%s\" %d 16 50 0 2 ", n, "DefaultRouter", n; printf "\"Default Router\" %d 0 \"%s\" %d 16 50 0 0 ", n, dfr, n; }'` exec 3>&1 IF_CONFIG=$(echo ${ADDRS} | xargs dialog --backtitle 'FreeBSD Installer' \ --title 'Network Configuration' \ --mixedform 'Static IPv6 Network Interface Configuration' 0 0 0 \ 2>&1 1>&3) if [ $? -eq $DIALOG_CANCEL ]; then exit 1; fi exec 3>&- echo ${IF_CONFIG} | tr ' ' '\n' | \ awk -v iface="${INTERFACE}" ' BEGIN { dfr=0; count=0; } { if (/^[[:space:]]+$/) { next; } if (/DefaultRouter/) { dfr=1; next; } if (dfr == 1) { printf("ipv6_defaultrouter=\"%s\"\n", $1); next; } if (count > 0) { # Ignore all but the first IP address for now. next; } count++; if (!match($1, "/")) { sub("$", "/64", $1); } printf("ifconfig_%s_ipv6=\"inet6 %s\"\n", iface, $1); }' >> $BSDINSTALL_TMPETC/._rc.conf.net if [ ! -z $BSDINSTALL_CONFIGCURRENT ]; then . $BSDINSTALL_TMPETC/._rc.conf.net ifconfig ${INTERFACE} `eval echo \\\$ifconfig_${INTERFACE}_ipv6` route delete default route add default ${ipv6_defaultrouter} fi Index: projects/largeSMP/usr.sbin/bsnmpd/modules/snmp_wlan/Makefile =================================================================== --- projects/largeSMP/usr.sbin/bsnmpd/modules/snmp_wlan/Makefile (revision 222782) +++ projects/largeSMP/usr.sbin/bsnmpd/modules/snmp_wlan/Makefile (revision 222783) @@ -1,17 +1,15 @@ # # $FreeBSD$ # MOD= wlan SRCS= wlan_snmp.c wlan_sys.c CFLAGS+= -DSNMPTREE_TYPES -WARNS= 6 - XSYM= begemotWlan BMIBS= BEGEMOT-WIRELESS-MIB.txt MAN= snmp_${MOD}.3 DEFS= ${MOD}_tree.def .include Index: projects/largeSMP/usr.sbin/lastlogin/lastlogin.8 =================================================================== --- projects/largeSMP/usr.sbin/lastlogin/lastlogin.8 (revision 222782) +++ projects/largeSMP/usr.sbin/lastlogin/lastlogin.8 (revision 222783) @@ -1,80 +1,94 @@ .\" $FreeBSD$ .\" $NetBSD: lastlogin.8,v 1.3 1998/02/06 06:20:39 perry Exp $ .\" .\" Copyright (c) 1996 John M. Vinopal .\" 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 for the NetBSD Project .\" by John M. Vinopal. .\" 4. 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. .\" -.Dd January 11, 1996 +.Dd June 6, 2011 .Dt LASTLOGIN 8 .Os .Sh NAME .Nm lastlogin .Nd indicate last login time of users .Sh SYNOPSIS .Nm +.Op Fl f Ar file +.Op Fl rt .Op Ar user ... .Sh DESCRIPTION The .Nm utility will list the last login session of each specified .Ar user , or for all users by default. Each line of output contains the user name, the tty from which the session was conducted, any hostname, and the start time for the session. .Pp If more than one .Ar user is given, the session information for each user is printed in the order given on the command line. -Otherwise, information -for all users is printed, sorted by name. +Otherwise, information for all users is printed. +By default, the entries are sorted by user name. .Pp The .Nm utility differs from .Xr last 1 in that it only prints information regarding the very last login session. The last login database is never turned over or deleted in standard usage. +.Pp +The following options are available: +.Bl -tag -width indent +.It Fl f Ar file +Open last login database +.Ar file +instead of the system-wide database. +.It Fl r +Print the entries in reverse sorted order. +.It Fl t +Sort the elements by last login time, instead of user name. +.El .Sh FILES .Bl -tag -width /var/log/utx.lastlogin -compact .It Pa /var/log/utx.lastlogin last login database .El .Sh SEE ALSO .Xr last 1 , .Xr getutxent 3 , .Xr ac 8 .Sh AUTHORS .An John M. Vinopal wrote this program in January 1996 and contributed it to the .Nx project. Index: projects/largeSMP/usr.sbin/lastlogin/lastlogin.c =================================================================== --- projects/largeSMP/usr.sbin/lastlogin/lastlogin.c (revision 222782) +++ projects/largeSMP/usr.sbin/lastlogin/lastlogin.c (revision 222783) @@ -1,124 +1,152 @@ /* * Copyright (c) 1996 John M. Vinopal * 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 for the NetBSD Project * by John M. Vinopal. * 4. 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. */ #include #ifndef lint __RCSID("$FreeBSD$"); __RCSID("$NetBSD: lastlogin.c,v 1.4 1998/02/03 04:45:35 perry Exp $"); #endif #include #include #include #include #include #include #include int main(int, char **); static void output(struct utmpx *); static void usage(void); +static int utcmp_user(const void *, const void *); +static int order = 1; +static const char *file = NULL; +static int (*utcmp)(const void *, const void *) = utcmp_user; + static int -utcmp(const void *u1, const void *u2) +utcmp_user(const void *u1, const void *u2) { - return (strcmp(((const struct utmpx *)u1)->ut_user, + return (order * strcmp(((const struct utmpx *)u1)->ut_user, ((const struct utmpx *)u2)->ut_user)); } +static int +utcmp_time(const void *u1, const void *u2) +{ + time_t t1, t2; + + t1 = ((const struct utmpx *)u1)->ut_tv.tv_sec; + t2 = ((const struct utmpx *)u2)->ut_tv.tv_sec; + return (t1 < t2 ? order : t1 > t2 ? -order : 0); +} + int main(int argc, char *argv[]) { int ch, i, ulistsize; struct utmpx *u, *ulist; - while ((ch = getopt(argc, argv, "")) != -1) { - usage(); + while ((ch = getopt(argc, argv, "f:rt")) != -1) { + switch (ch) { + case 'f': + file = optarg; + break; + case 'r': + order = -1; + break; + case 't': + utcmp = utcmp_time; + break; + default: + usage(); + } } + argc -= optind; + argv += optind; - /* Process usernames given on the command line. */ - if (argc > 1) { - for (i = 1; i < argc; ++i) { - if (setutxdb(UTXDB_LASTLOGIN, NULL) != 0) - errx(1, "failed to open lastlog database"); + if (argc > 0) { + /* Process usernames given on the command line. */ + for (i = 0; i < argc; i++) { + if (setutxdb(UTXDB_LASTLOGIN, file) != 0) + err(1, "failed to open lastlog database"); if ((u = getutxuser(argv[i])) == NULL) { warnx("user '%s' not found", argv[i]); continue; } output(u); endutxent(); } - } - /* Read all lastlog entries, looking for active ones */ - else { - if (setutxdb(UTXDB_LASTLOGIN, NULL) != 0) - errx(1, "failed to open lastlog database"); + } else { + /* Read all lastlog entries, looking for active ones. */ + if (setutxdb(UTXDB_LASTLOGIN, file) != 0) + err(1, "failed to open lastlog database"); ulist = NULL; ulistsize = 0; while ((u = getutxent()) != NULL) { if (u->ut_type != USER_PROCESS) continue; if ((ulistsize % 16) == 0) { ulist = realloc(ulist, (ulistsize + 16) * sizeof(struct utmpx)); if (ulist == NULL) err(1, "malloc"); } ulist[ulistsize++] = *u; } endutxent(); qsort(ulist, ulistsize, sizeof(struct utmpx), utcmp); for (i = 0; i < ulistsize; i++) output(&ulist[i]); } exit(0); } /* Duplicate the output of last(1) */ static void output(struct utmpx *u) { time_t t = u->ut_tv.tv_sec; printf("%-10s %-8s %-22.22s %s", u->ut_user, u->ut_line, u->ut_host, ctime(&t)); } static void usage(void) { - fprintf(stderr, "usage: lastlogin [user ...]\n"); + fprintf(stderr, "usage: lastlogin [-f file] [-rt] [user ...]\n"); exit(1); } Index: projects/largeSMP/usr.sbin/ndiscvt =================================================================== --- projects/largeSMP/usr.sbin/ndiscvt (revision 222782) +++ projects/largeSMP/usr.sbin/ndiscvt (revision 222783) Property changes on: projects/largeSMP/usr.sbin/ndiscvt ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/usr.sbin/ndiscvt:r222698-222782 Index: projects/largeSMP/usr.sbin/rtadvd/advcap.c =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/advcap.c (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/advcap.c (revision 222783) @@ -1,451 +1,437 @@ /* $FreeBSD$ */ /* $KAME: advcap.c,v 1.11 2003/05/19 09:46:50 keiichi Exp $ */ /* * Copyright (c) 1983 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. */ /* * remcap - routines for dealing with the remote host data base * * derived from termcap */ #include #include #include #include #include #include #include #include #include #include #include "pathnames.h" #ifndef BUFSIZ #define BUFSIZ 1024 #endif #define MAXHOP 32 /* max number of tc= indirections */ #define tgetent agetent #define tnchktc anchktc #define tnamatch anamatch #define tgetnum agetnum #define tgetflag agetflag #define tgetstr agetstr #if 0 #define V_TERMCAP "REMOTE" #define V_TERM "HOST" #endif -char *RM; - /* * termcap - routines for dealing with the terminal capability data base * * BUG: Should use a "last" pointer in tbuf, so that searching * for capabilities alphabetically would not be a n**2/2 * process when large numbers of capabilities are given. * Note: If we add a last pointer now we will screw up the * tc capability. We really should compile termcap. * * Essentially all the work here is scanning and decoding escapes * in string capabilities. We don't use stdio because the editor * doesn't, and because living w/o it is not hard. */ static char *tbuf; static int hopcount; /* detect infinite loops in termcap, init 0 */ -static char *remotefile; +static const char *remotefile; +extern const char *conffile; -extern char *conffile; - int tgetent(char *, char *); -int getent(char *, char *, char *); +int getent(char *, char *, const char *); int tnchktc(void); int tnamatch(char *); static char *tskip(char *); int64_t tgetnum(char *); int tgetflag(char *); char *tgetstr(char *, char **); static char *tdecode(char *, char **); /* * Get an entry for terminal name in buffer bp, * from the termcap file. Parse is very rudimentary; * we just notice escaped newlines. */ int -tgetent(bp, name) - char *bp, *name; +tgetent(char *bp, char *name) { - char *cp; - - remotefile = cp = conffile ? conffile : _PATH_RTADVDCONF; - return (getent(bp, name, cp)); + return (getent(bp, name, conffile)); } int -getent(bp, name, cp) - char *bp, *name, *cp; +getent(char *bp, char *name, const char *cfile) { int c; int i = 0, cnt = 0; char ibuf[BUFSIZ]; + char *cp; int tf; tbuf = bp; tf = 0; /* * TERMCAP can have one of two things in it. It can be the * name of a file to use instead of /etc/termcap. In this * case it better start with a "/". Or it can be an entry to * use so we don't have to read the file. In this case it * has to already have the newlines crunched out. */ - if (cp && *cp) { - tf = open(RM = cp, O_RDONLY); - } + if (cfile && *cfile) + tf = open(cfile, O_RDONLY); + if (tf < 0) { syslog(LOG_INFO, "<%s> open: %s", __func__, strerror(errno)); return (-2); } for (;;) { cp = bp; for (;;) { if (i == cnt) { cnt = read(tf, ibuf, BUFSIZ); if (cnt <= 0) { close(tf); return (0); } i = 0; } c = ibuf[i++]; if (c == '\n') { if (cp > bp && cp[-1] == '\\') { cp--; continue; } break; } if (cp >= bp + BUFSIZ) { write(STDERR_FILENO, "Remcap entry too long\n", 23); break; } else *cp++ = c; } *cp = 0; /* * The real work for the match. */ if (tnamatch(name)) { close(tf); return (tnchktc()); } } } /* * tnchktc: check the last entry, see if it's tc=xxx. If so, * recursively find xxx and append that entry (minus the names) * to take the place of the tc=xxx entry. This allows termcap * entries to say "like an HP2621 but doesn't turn on the labels". * Note that this works because of the left to right scan. */ int -tnchktc() +tnchktc(void) { char *p, *q; char tcname[16]; /* name of similar terminal */ char tcbuf[BUFSIZ]; char *holdtbuf = tbuf; int l; p = tbuf + strlen(tbuf) - 2; /* before the last colon */ while (*--p != ':') if (p < tbuf) { write(STDERR_FILENO, "Bad remcap entry\n", 18); return (0); } p++; /* p now points to beginning of last field */ if (p[0] != 't' || p[1] != 'c') return (1); strlcpy(tcname, p + 3, sizeof tcname); q = tcname; while (*q && *q != ':') q++; *q = 0; if (++hopcount > MAXHOP) { write(STDERR_FILENO, "Infinite tc= loop\n", 18); return (0); } if (getent(tcbuf, tcname, remotefile) != 1) { return (0); } for (q = tcbuf; *q++ != ':'; ) ; l = p - holdtbuf + strlen(q); if (l > BUFSIZ) { write(STDERR_FILENO, "Remcap entry too long\n", 23); q[BUFSIZ - (p-holdtbuf)] = 0; } strcpy(p, q); tbuf = holdtbuf; return (1); } /* * Tnamatch deals with name matching. The first field of the termcap * entry is a sequence of names separated by |'s, so we compare * against each such name. The normal : terminator after the last * name (before the first field) stops us. */ int -tnamatch(np) - char *np; +tnamatch(char *np) { char *Np, *Bp; Bp = tbuf; if (*Bp == '#') return (0); for (;;) { for (Np = np; *Np && *Bp == *Np; Bp++, Np++) continue; if (*Np == 0 && (*Bp == '|' || *Bp == ':' || *Bp == 0)) return (1); while (*Bp && *Bp != ':' && *Bp != '|') Bp++; if (*Bp == 0 || *Bp == ':') return (0); Bp++; } } /* * Skip to the next field. Notice that this is very dumb, not * knowing about \: escapes or any such. If necessary, :'s can be put * into the termcap file in octal. */ static char * -tskip(bp) - char *bp; +tskip(char *bp) { int dquote; dquote = 0; while (*bp) { switch (*bp) { case ':': if (!dquote) goto breakbreak; else bp++; break; case '\\': bp++; if (isdigit(*bp)) { while (isdigit(*bp++)) ; } else bp++; case '"': dquote = (dquote ? 1 : 0); bp++; break; default: bp++; break; } } breakbreak: if (*bp == ':') bp++; return (bp); } /* * Return the (numeric) option id. * Numeric options look like * li#80 * i.e. the option string is separated from the numeric value by * a # character. If the option is not found we return -1. * Note that we handle octal numbers beginning with 0. */ int64_t -tgetnum(id) - char *id; +tgetnum(char *id) { int64_t i; int base; char *bp = tbuf; for (;;) { bp = tskip(bp); if (*bp == 0) return (-1); if (strncmp(bp, id, strlen(id)) != 0) continue; bp += strlen(id); if (*bp == '@') return (-1); if (*bp != '#') continue; bp++; base = 10; if (*bp == '0') base = 8; i = 0; while (isdigit(*bp)) i *= base, i += *bp++ - '0'; return (i); } } /* * Handle a flag option. * Flag options are given "naked", i.e. followed by a : or the end * of the buffer. Return 1 if we find the option, or 0 if it is * not given. */ int -tgetflag(id) - char *id; +tgetflag(char *id) { char *bp = tbuf; for (;;) { bp = tskip(bp); if (!*bp) return (0); if (strncmp(bp, id, strlen(id)) == 0) { bp += strlen(id); if (!*bp || *bp == ':') return (1); else if (*bp == '@') return (0); } } } /* * Get a string valued option. * These are given as * cl=^Z * Much decoding is done on the strings, and the strings are * placed in area, which is a ref parameter which is updated. * No checking on area overflow. */ char * -tgetstr(id, area) - char *id, **area; +tgetstr(char *id, char **area) { char *bp = tbuf; for (;;) { bp = tskip(bp); if (!*bp) return (0); if (strncmp(bp, id, strlen(id)) != 0) continue; bp += strlen(id); if (*bp == '@') return (0); if (*bp != '=') continue; bp++; return (tdecode(bp, area)); } } /* * Tdecode does the grung work to decode the * string capability escapes. */ static char * -tdecode(str, area) - char *str; - char **area; +tdecode(char *str, char **area) { char *cp; int c; - char *dp; + const char *dp; int i; char term; term = ':'; cp = *area; again: if (*str == '"') { term = '"'; str++; } while ((c = *str++) && c != term) { switch (c) { case '^': c = *str++ & 037; break; case '\\': dp = "E\033^^\\\\::n\nr\rt\tb\bf\f\"\""; c = *str++; nextc: if (*dp++ == c) { c = *dp++; break; } dp++; if (*dp) goto nextc; if (isdigit(c)) { c -= '0', i = 2; do c <<= 3, c |= *str++ - '0'; while (--i && isdigit(*str)); } break; } *cp++ = c; } if (c == term && term != ':') { term = ':'; goto again; } *cp++ = 0; str = *area; *area = cp; return (str); } Index: projects/largeSMP/usr.sbin/rtadvd/config.c =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/config.c (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/config.c (revision 222783) @@ -1,1080 +1,1290 @@ /* $FreeBSD$ */ /* $KAME: config.c,v 1.84 2003/08/05 12:34:23 itojun Exp $ */ /* * Copyright (C) 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. */ #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 "rtadvd.h" #include "advcap.h" #include "timer.h" #include "if.h" #include "config.h" +/* label of tcapcode + number + domain name + zero octet */ +static char entbuf[10 + 3 + NI_MAXHOST + 1]; +static char oentbuf[10 + 3 + NI_MAXHOST + 1]; +static char abuf[DNAME_LABELENC_MAXLEN]; + static time_t prefix_timo = (60 * 120); /* 2 hours. * XXX: should be configurable. */ -extern struct rainfo *ralist; static struct rtadvd_timer *prefix_timeout(void *); -static void makeentry(char *, size_t, int, char *); -static int getinet6sysctl(int); +static void makeentry(char *, size_t, int, const char *); +static size_t dname_labelenc(char *, const char *); -void -getconfig(intface) - char *intface; +/* Encode domain name label encoding in RFC 1035 Section 3.1 */ +static size_t +dname_labelenc(char *dst, const char *src) { - int stat, i; - char tbuf[BUFSIZ]; - struct rainfo *tmp; - long val; - int64_t val64; - char buf[BUFSIZ]; - char *bp = buf; - char *addr, *flagstr; - static int forwarding = -1; + char *dst_origin; + char *p; + size_t len; -#define MUSTHAVE(var, cap) \ + dst_origin = dst; + len = strlen(src); + + /* Length fields per 63 octets + '\0' (<= DNAME_LABELENC_MAXLEN) */ + memset(dst, 0, len + len / 64 + 1 + 1); + + syslog(LOG_DEBUG, "<%s> labelenc = %s", __func__, src); + while (src && (len = strlen(src)) != 0) { + /* Put a length field with 63 octet limitation first. */ + p = strchr(src, '.'); + if (p == NULL) + *dst++ = len = MIN(63, len); + else + *dst++ = len = MIN(63, p - src); + /* Copy 63 octets at most. */ + memcpy(dst, src, len); + dst += len; + if (p == NULL) /* the last label */ + break; + src = p + 1; + } + /* Always need a 0-length label at the tail. */ + *dst++ = '\0'; + + syslog(LOG_DEBUG, "<%s> labellen = %td", __func__, dst - dst_origin); + return (dst - dst_origin); +} + +#define MUSTHAVE(var, cap) \ do { \ int64_t t; \ if ((t = agetnum(cap)) < 0) { \ fprintf(stderr, "rtadvd: need %s for interface %s\n", \ cap, intface); \ exit(1); \ } \ var = t; \ } while (0) -#define MAYHAVE(var, cap, def) \ + +#define MAYHAVE(var, cap, def) \ do { \ if ((var = agetnum(cap)) < 0) \ var = def; \ } while (0) +#define ELM_MALLOC(p,error_action) \ + do { \ + p = malloc(sizeof(*p)); \ + if (p == NULL) { \ + syslog(LOG_ERR, "<%s> malloc failed: %s", \ + __func__, strerror(errno)); \ + error_action; \ + } \ + memset(p, 0, sizeof(*p)); \ + } while(0) + +int +rmconfig(int idx) +{ + struct rainfo *rai; + struct prefix *pfx; + struct soliciter *sol; + struct rdnss *rdn; + struct rdnss_addr *rdna; + struct dnssl *dns; + struct rtinfo *rti; + + rai = if_indextorainfo(idx); + if (rai == NULL) { + syslog(LOG_ERR, "<%s>: rainfo not found (idx=%d)", + __func__, idx); + return (-1); + } + + TAILQ_REMOVE(&railist, rai, rai_next); + syslog(LOG_DEBUG, "<%s>: rainfo (idx=%d) removed.", + __func__, idx); + + /* Free all of allocated memories for this entry. */ + rtadvd_remove_timer(rai->rai_timer); + + if (rai->rai_ra_data != NULL) + free(rai->rai_ra_data); + + if (rai->rai_sdl != NULL) + free(rai->rai_sdl); + + while ((pfx = TAILQ_FIRST(&rai->rai_prefix)) != NULL) { + TAILQ_REMOVE(&rai->rai_prefix, pfx, pfx_next); + free(pfx); + } + while ((sol = TAILQ_FIRST(&rai->rai_soliciter)) != NULL) { + TAILQ_REMOVE(&rai->rai_soliciter, sol, sol_next); + free(sol); + } + while ((rdn = TAILQ_FIRST(&rai->rai_rdnss)) != NULL) { + TAILQ_REMOVE(&rai->rai_rdnss, rdn, rd_next); + while ((rdna = TAILQ_FIRST(&rdn->rd_list)) != NULL) { + TAILQ_REMOVE(&rdn->rd_list, rdna, ra_next); + free(rdna); + } + free(rdn); + } + while ((dns = TAILQ_FIRST(&rai->rai_dnssl)) != NULL) { + TAILQ_REMOVE(&rai->rai_dnssl, dns, dn_next); + free(dns); + } + while ((rti = TAILQ_FIRST(&rai->rai_route)) != NULL) { + TAILQ_REMOVE(&rai->rai_route, rti, rti_next); + free(rti); + } + free(rai); + + return (0); +} + +int +getconfig(int idx) +{ + int stat, i; + char tbuf[BUFSIZ]; + struct rainfo *rai; + long val; + int64_t val64; + char buf[BUFSIZ]; + char *bp = buf; + char *addr, *flagstr; + char intface[IFNAMSIZ]; + + if (if_indextoname(idx, intface) == NULL) { + syslog(LOG_ERR, "<%s> invalid index number (%d)", + __func__, idx); + return (-1); + } + if ((stat = agetent(tbuf, intface)) <= 0) { memset(tbuf, 0, sizeof(tbuf)); syslog(LOG_INFO, - "<%s> %s isn't defined in the configuration file" - " or the configuration file doesn't exist." - " Treat it as default", - __func__, intface); + "<%s> %s isn't defined in the configuration file" + " or the configuration file doesn't exist." + " Treat it as default", + __func__, intface); } - tmp = (struct rainfo *)malloc(sizeof(*ralist)); - if (tmp == NULL) { + rai = malloc(sizeof(*rai)); + if (rai == NULL) { syslog(LOG_INFO, "<%s> %s: can't allocate enough memory", __func__, intface); exit(1); } - memset(tmp, 0, sizeof(*tmp)); - tmp->prefix.next = tmp->prefix.prev = &tmp->prefix; + memset(rai, 0, sizeof(*rai)); + TAILQ_INIT(&rai->rai_prefix); #ifdef ROUTEINFO - tmp->route.next = tmp->route.prev = &tmp->route; + TAILQ_INIT(&rai->rai_route); #endif + TAILQ_INIT(&rai->rai_rdnss); + TAILQ_INIT(&rai->rai_dnssl); + TAILQ_INIT(&rai->rai_soliciter); - /* check if we are allowed to forward packets (if not determined) */ - if (forwarding < 0) { - if ((forwarding = getinet6sysctl(IPV6CTL_FORWARDING)) < 0) - exit(1); - } + /* gather on-link prefixes from the network interfaces. */ + if (agetflag("noifprefix")) + rai->rai_advifprefix = 0; + else + rai->rai_advifprefix = 1; /* get interface information */ if (agetflag("nolladdr")) - tmp->advlinkopt = 0; + rai->rai_advlinkopt = 0; else - tmp->advlinkopt = 1; - if (tmp->advlinkopt) { - if ((tmp->sdl = if_nametosdl(intface)) == NULL) { + rai->rai_advlinkopt = 1; + if (rai->rai_advlinkopt) { + if ((rai->rai_sdl = if_nametosdl(intface)) == NULL) { syslog(LOG_ERR, - "<%s> can't get information of %s", - __func__, intface); - exit(1); + "<%s> can't get information of %s", + __func__, intface); + return (-1); } - tmp->ifindex = tmp->sdl->sdl_index; + rai->rai_ifindex = rai->rai_sdl->sdl_index; } else - tmp->ifindex = if_nametoindex(intface); - strncpy(tmp->ifname, intface, sizeof(tmp->ifname)); - if ((tmp->phymtu = if_getmtu(intface)) == 0) { - tmp->phymtu = IPV6_MMTU; + rai->rai_ifindex = if_nametoindex(intface); + strncpy(rai->rai_ifname, intface, sizeof(rai->rai_ifname)); + syslog(LOG_DEBUG, + "<%s> ifindex = %d on %s", __func__, rai->rai_ifindex, + rai->rai_ifname); + + if ((rai->rai_phymtu = if_getmtu(intface)) == 0) { + rai->rai_phymtu = IPV6_MMTU; syslog(LOG_WARNING, - "<%s> can't get interface mtu of %s. Treat as %d", - __func__, intface, IPV6_MMTU); + "<%s> can't get interface mtu of %s. Treat as %d", + __func__, intface, IPV6_MMTU); } /* * set router configuration variables. */ MAYHAVE(val, "maxinterval", DEF_MAXRTRADVINTERVAL); if (val < MIN_MAXINTERVAL || val > MAX_MAXINTERVAL) { syslog(LOG_ERR, - "<%s> maxinterval (%ld) on %s is invalid " - "(must be between %u and %u)", __func__, val, - intface, MIN_MAXINTERVAL, MAX_MAXINTERVAL); - exit(1); + "<%s> maxinterval (%ld) on %s is invalid " + "(must be between %u and %u)", __func__, val, + intface, MIN_MAXINTERVAL, MAX_MAXINTERVAL); + return (-1); } - tmp->maxinterval = (u_int)val; - MAYHAVE(val, "mininterval", tmp->maxinterval/3); - if (val < MIN_MININTERVAL || val > (tmp->maxinterval * 3) / 4) { + rai->rai_maxinterval = (u_int)val; + + MAYHAVE(val, "mininterval", rai->rai_maxinterval/3); + if ((u_int)val < MIN_MININTERVAL || + (u_int)val > (rai->rai_maxinterval * 3) / 4) { syslog(LOG_ERR, - "<%s> mininterval (%ld) on %s is invalid " - "(must be between %d and %d)", - __func__, val, intface, MIN_MININTERVAL, - (tmp->maxinterval * 3) / 4); - exit(1); + "<%s> mininterval (%ld) on %s is invalid " + "(must be between %d and %d)", + __func__, val, intface, MIN_MININTERVAL, + (rai->rai_maxinterval * 3) / 4); + return (-1); } - tmp->mininterval = (u_int)val; + rai->rai_mininterval = (u_int)val; MAYHAVE(val, "chlim", DEF_ADVCURHOPLIMIT); - tmp->hoplimit = val & 0xff; + rai->rai_hoplimit = val & 0xff; if ((flagstr = (char *)agetstr("raflags", &bp))) { val = 0; if (strchr(flagstr, 'm')) val |= ND_RA_FLAG_MANAGED; if (strchr(flagstr, 'o')) val |= ND_RA_FLAG_OTHER; if (strchr(flagstr, 'h')) val |= ND_RA_FLAG_RTPREF_HIGH; if (strchr(flagstr, 'l')) { if ((val & ND_RA_FLAG_RTPREF_HIGH)) { syslog(LOG_ERR, "<%s> the \'h\' and \'l\'" " router flags are exclusive", __func__); - exit(1); + return (-1); } val |= ND_RA_FLAG_RTPREF_LOW; } - } else { + } else MAYHAVE(val, "raflags", 0); - } - tmp->managedflg = val & ND_RA_FLAG_MANAGED; - tmp->otherflg = val & ND_RA_FLAG_OTHER; + + rai->rai_managedflg = val & ND_RA_FLAG_MANAGED; + rai->rai_otherflg = val & ND_RA_FLAG_OTHER; #ifndef ND_RA_FLAG_RTPREF_MASK #define ND_RA_FLAG_RTPREF_MASK 0x18 /* 00011000 */ #define ND_RA_FLAG_RTPREF_RSV 0x10 /* 00010000 */ #endif - tmp->rtpref = val & ND_RA_FLAG_RTPREF_MASK; - if (tmp->rtpref == ND_RA_FLAG_RTPREF_RSV) { + rai->rai_rtpref = val & ND_RA_FLAG_RTPREF_MASK; + if (rai->rai_rtpref == ND_RA_FLAG_RTPREF_RSV) { syslog(LOG_ERR, "<%s> invalid router preference (%02x) on %s", - __func__, tmp->rtpref, intface); - exit(1); + __func__, rai->rai_rtpref, intface); + return (-1); } - MAYHAVE(val, "rltime", tmp->maxinterval * 3); - if (val && (val < tmp->maxinterval || val > MAXROUTERLIFETIME)) { + MAYHAVE(val, "rltime", rai->rai_maxinterval * 3); + if ((u_int)val && ((u_int)val < rai->rai_maxinterval || + (u_int)val > MAXROUTERLIFETIME)) { syslog(LOG_ERR, - "<%s> router lifetime (%ld) on %s is invalid " - "(must be 0 or between %d and %d)", - __func__, val, intface, - tmp->maxinterval, - MAXROUTERLIFETIME); - exit(1); + "<%s> router lifetime (%ld) on %s is invalid " + "(must be 0 or between %d and %d)", + __func__, val, intface, rai->rai_maxinterval, + MAXROUTERLIFETIME); + return (-1); } - /* - * Basically, hosts MUST NOT send Router Advertisement messages at any - * time (RFC 2461, Section 6.2.3). However, it would sometimes be - * useful to allow hosts to advertise some parameters such as prefix - * information and link MTU. Thus, we allow hosts to invoke rtadvd - * only when router lifetime (on every advertising interface) is - * explicitly set zero. (see also the above section) - */ - if (val && forwarding == 0) { - syslog(LOG_ERR, - "<%s> non zero router lifetime is specified for %s, " - "which must not be allowed for hosts. you must " - "change router lifetime or enable IPv6 forwarding.", - __func__, intface); - exit(1); - } - tmp->lifetime = val & 0xffff; + rai->rai_lifetime = val & 0xffff; MAYHAVE(val, "rtime", DEF_ADVREACHABLETIME); if (val < 0 || val > MAXREACHABLETIME) { syslog(LOG_ERR, - "<%s> reachable time (%ld) on %s is invalid " - "(must be no greater than %d)", - __func__, val, intface, MAXREACHABLETIME); - exit(1); + "<%s> reachable time (%ld) on %s is invalid " + "(must be no greater than %d)", + __func__, val, intface, MAXREACHABLETIME); + return (-1); } - tmp->reachabletime = (u_int32_t)val; + rai->rai_reachabletime = (u_int32_t)val; MAYHAVE(val64, "retrans", DEF_ADVRETRANSTIMER); if (val64 < 0 || val64 > 0xffffffff) { syslog(LOG_ERR, "<%s> retrans time (%lld) on %s out of range", - __func__, (long long)val64, intface); - exit(1); + __func__, (long long)val64, intface); + return (-1); } - tmp->retranstimer = (u_int32_t)val64; + rai->rai_retranstimer = (u_int32_t)val64; if (agetnum("hapref") != -1 || agetnum("hatime") != -1) { syslog(LOG_ERR, - "<%s> mobile-ip6 configuration not supported", - __func__); - exit(1); + "<%s> mobile-ip6 configuration not supported", + __func__); + return (-1); } /* prefix information */ /* * This is an implementation specific parameter to consider * link propagation delays and poorly synchronized clocks when * checking consistency of advertised lifetimes. */ MAYHAVE(val, "clockskew", 0); - tmp->clockskew = val; + rai->rai_clockskew = val; - tmp->pfxs = 0; + rai->rai_pfxs = 0; for (i = -1; i < MAXPREFIX; i++) { struct prefix *pfx; - char entbuf[256]; makeentry(entbuf, sizeof(entbuf), i, "addr"); addr = (char *)agetstr(entbuf, &bp); if (addr == NULL) continue; /* allocate memory to store prefix information */ - if ((pfx = malloc(sizeof(struct prefix))) == NULL) { - syslog(LOG_ERR, - "<%s> can't allocate enough memory", - __func__); - exit(1); - } - memset(pfx, 0, sizeof(*pfx)); + ELM_MALLOC(pfx, exit(1)); /* link into chain */ - insque(pfx, &tmp->prefix); - tmp->pfxs++; - pfx->rainfo = tmp; + TAILQ_INSERT_TAIL(&rai->rai_prefix, pfx, pfx_next); + rai->rai_pfxs++; + pfx->pfx_origin = PREFIX_FROM_CONFIG; - pfx->origin = PREFIX_FROM_CONFIG; - - if (inet_pton(AF_INET6, addr, &pfx->prefix) != 1) { + if (inet_pton(AF_INET6, addr, &pfx->pfx_prefix) != 1) { syslog(LOG_ERR, - "<%s> inet_pton failed for %s", - __func__, addr); - exit(1); + "<%s> inet_pton failed for %s", + __func__, addr); + return (-1); } - if (IN6_IS_ADDR_MULTICAST(&pfx->prefix)) { + if (IN6_IS_ADDR_MULTICAST(&pfx->pfx_prefix)) { syslog(LOG_ERR, - "<%s> multicast prefix (%s) must " - "not be advertised on %s", - __func__, addr, intface); - exit(1); + "<%s> multicast prefix (%s) must " + "not be advertised on %s", + __func__, addr, intface); + return (-1); } - if (IN6_IS_ADDR_LINKLOCAL(&pfx->prefix)) + if (IN6_IS_ADDR_LINKLOCAL(&pfx->pfx_prefix)) syslog(LOG_NOTICE, - "<%s> link-local prefix (%s) will be" - " advertised on %s", - __func__, addr, intface); + "<%s> link-local prefix (%s) will be" + " advertised on %s", + __func__, addr, intface); makeentry(entbuf, sizeof(entbuf), i, "prefixlen"); MAYHAVE(val, entbuf, 64); if (val < 0 || val > 128) { syslog(LOG_ERR, "<%s> prefixlen (%ld) for %s " - "on %s out of range", - __func__, val, addr, intface); - exit(1); + "on %s out of range", + __func__, val, addr, intface); + return (-1); } - pfx->prefixlen = (int)val; + pfx->pfx_prefixlen = (int)val; makeentry(entbuf, sizeof(entbuf), i, "pinfoflags"); if ((flagstr = (char *)agetstr(entbuf, &bp))) { val = 0; if (strchr(flagstr, 'l')) val |= ND_OPT_PI_FLAG_ONLINK; if (strchr(flagstr, 'a')) val |= ND_OPT_PI_FLAG_AUTO; } else { MAYHAVE(val, entbuf, (ND_OPT_PI_FLAG_ONLINK|ND_OPT_PI_FLAG_AUTO)); } - pfx->onlinkflg = val & ND_OPT_PI_FLAG_ONLINK; - pfx->autoconfflg = val & ND_OPT_PI_FLAG_AUTO; + pfx->pfx_onlinkflg = val & ND_OPT_PI_FLAG_ONLINK; + pfx->pfx_autoconfflg = val & ND_OPT_PI_FLAG_AUTO; makeentry(entbuf, sizeof(entbuf), i, "vltime"); MAYHAVE(val64, entbuf, DEF_ADVVALIDLIFETIME); if (val64 < 0 || val64 > 0xffffffff) { syslog(LOG_ERR, "<%s> vltime (%lld) for " "%s/%d on %s is out of range", __func__, (long long)val64, - addr, pfx->prefixlen, intface); - exit(1); + addr, pfx->pfx_prefixlen, intface); + return (-1); } - pfx->validlifetime = (u_int32_t)val64; + pfx->pfx_validlifetime = (u_int32_t)val64; makeentry(entbuf, sizeof(entbuf), i, "vltimedecr"); if (agetflag(entbuf)) { struct timeval now; gettimeofday(&now, 0); - pfx->vltimeexpire = - now.tv_sec + pfx->validlifetime; + pfx->pfx_vltimeexpire = + now.tv_sec + pfx->pfx_validlifetime; } makeentry(entbuf, sizeof(entbuf), i, "pltime"); MAYHAVE(val64, entbuf, DEF_ADVPREFERREDLIFETIME); if (val64 < 0 || val64 > 0xffffffff) { syslog(LOG_ERR, "<%s> pltime (%lld) for %s/%d on %s " "is out of range", __func__, (long long)val64, - addr, pfx->prefixlen, intface); - exit(1); + addr, pfx->pfx_prefixlen, intface); + return (-1); } - pfx->preflifetime = (u_int32_t)val64; + pfx->pfx_preflifetime = (u_int32_t)val64; makeentry(entbuf, sizeof(entbuf), i, "pltimedecr"); if (agetflag(entbuf)) { struct timeval now; gettimeofday(&now, 0); - pfx->pltimeexpire = - now.tv_sec + pfx->preflifetime; + pfx->pfx_pltimeexpire = + now.tv_sec + pfx->pfx_preflifetime; } } - if (tmp->pfxs == 0) - get_prefix(tmp); + if (rai->rai_advifprefix && rai->rai_pfxs == 0) + get_prefix(rai); MAYHAVE(val, "mtu", 0); - if (val < 0 || val > 0xffffffff) { + if (val < 0 || (u_int)val > 0xffffffff) { syslog(LOG_ERR, - "<%s> mtu (%ld) on %s out of range", - __func__, val, intface); - exit(1); + "<%s> mtu (%ld) on %s out of range", + __func__, val, intface); + return (-1); } - tmp->linkmtu = (u_int32_t)val; - if (tmp->linkmtu == 0) { + rai->rai_linkmtu = (u_int32_t)val; + if (rai->rai_linkmtu == 0) { char *mtustr; if ((mtustr = (char *)agetstr("mtu", &bp)) && strcmp(mtustr, "auto") == 0) - tmp->linkmtu = tmp->phymtu; + rai->rai_linkmtu = rai->rai_phymtu; } - else if (tmp->linkmtu < IPV6_MMTU || tmp->linkmtu > tmp->phymtu) { + else if (rai->rai_linkmtu < IPV6_MMTU || + rai->rai_linkmtu > rai->rai_phymtu) { syslog(LOG_ERR, - "<%s> advertised link mtu (%lu) on %s is invalid (must " - "be between least MTU (%d) and physical link MTU (%d)", - __func__, (unsigned long)tmp->linkmtu, intface, - IPV6_MMTU, tmp->phymtu); - exit(1); + "<%s> advertised link mtu (%lu) on %s is invalid (must " + "be between least MTU (%d) and physical link MTU (%d)", + __func__, (unsigned long)rai->rai_linkmtu, intface, + IPV6_MMTU, rai->rai_phymtu); + return (-1); } #ifdef SIOCSIFINFO_IN6 { struct in6_ndireq ndi; int s; if ((s = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) { syslog(LOG_ERR, "<%s> socket: %s", __func__, - strerror(errno)); + strerror(errno)); exit(1); } memset(&ndi, 0, sizeof(ndi)); strncpy(ndi.ifname, intface, IFNAMSIZ); - if (ioctl(s, SIOCGIFINFO_IN6, (caddr_t)&ndi) < 0) { + if (ioctl(s, SIOCGIFINFO_IN6, (caddr_t)&ndi) < 0) syslog(LOG_INFO, "<%s> ioctl:SIOCGIFINFO_IN6 at %s: %s", - __func__, intface, strerror(errno)); - } + __func__, intface, strerror(errno)); /* reflect the RA info to the host variables in kernel */ - ndi.ndi.chlim = tmp->hoplimit; - ndi.ndi.retrans = tmp->retranstimer; - ndi.ndi.basereachable = tmp->reachabletime; - if (ioctl(s, SIOCSIFINFO_IN6, (caddr_t)&ndi) < 0) { + ndi.ndi.chlim = rai->rai_hoplimit; + ndi.ndi.retrans = rai->rai_retranstimer; + ndi.ndi.basereachable = rai->rai_reachabletime; + if (ioctl(s, SIOCSIFINFO_IN6, (caddr_t)&ndi) < 0) syslog(LOG_INFO, "<%s> ioctl:SIOCSIFINFO_IN6 at %s: %s", - __func__, intface, strerror(errno)); - } + __func__, intface, strerror(errno)); + close(s); } #endif /* route information */ #ifdef ROUTEINFO - tmp->routes = 0; + rai->rai_routes = 0; for (i = -1; i < MAXROUTE; i++) { struct rtinfo *rti; - char entbuf[256], oentbuf[256]; makeentry(entbuf, sizeof(entbuf), i, "rtprefix"); addr = (char *)agetstr(entbuf, &bp); if (addr == NULL) { makeentry(oentbuf, sizeof(oentbuf), i, "rtrprefix"); addr = (char *)agetstr(oentbuf, &bp); - if (addr) { + if (addr) fprintf(stderr, "%s was obsoleted. Use %s.\n", - oentbuf, entbuf); - } + oentbuf, entbuf); } if (addr == NULL) continue; /* allocate memory to store prefix information */ - if ((rti = malloc(sizeof(struct rtinfo))) == NULL) { - syslog(LOG_ERR, - "<%s> can't allocate enough memory", - __func__); - exit(1); - } - memset(rti, 0, sizeof(*rti)); + ELM_MALLOC(rti, exit(1)); /* link into chain */ - insque(rti, &tmp->route); - tmp->routes++; + TAILQ_INSERT_TAIL(&rai->rai_route, rti, rti_next); + rai->rai_routes++; - if (inet_pton(AF_INET6, addr, &rti->prefix) != 1) { + if (inet_pton(AF_INET6, addr, &rti->rti_prefix) != 1) { syslog(LOG_ERR, "<%s> inet_pton failed for %s", - __func__, addr); - exit(1); + __func__, addr); + return (-1); } #if 0 /* * XXX: currently there's no restriction in route information * prefix according to * draft-ietf-ipngwg-router-selection-00.txt. * However, I think the similar restriction be necessary. */ MAYHAVE(val64, entbuf, DEF_ADVVALIDLIFETIME); if (IN6_IS_ADDR_MULTICAST(&rti->prefix)) { syslog(LOG_ERR, - "<%s> multicast route (%s) must " - "not be advertised on %s", - __func__, addr, intface); - exit(1); + "<%s> multicast route (%s) must " + "not be advertised on %s", + __func__, addr, intface); + return (-1); } if (IN6_IS_ADDR_LINKLOCAL(&rti->prefix)) { syslog(LOG_NOTICE, - "<%s> link-local route (%s) will " - "be advertised on %s", - __func__, addr, intface); - exit(1); + "<%s> link-local route (%s) will " + "be advertised on %s", + __func__, addr, intface); + return (-1); } #endif makeentry(entbuf, sizeof(entbuf), i, "rtplen"); /* XXX: 256 is a magic number for compatibility check. */ MAYHAVE(val, entbuf, 256); if (val == 256) { makeentry(oentbuf, sizeof(oentbuf), i, "rtrplen"); MAYHAVE(val, oentbuf, 256); - if (val != 256) { + if (val != 256) fprintf(stderr, "%s was obsoleted. Use %s.\n", - oentbuf, entbuf); - } else + oentbuf, entbuf); + else val = 64; } if (val < 0 || val > 128) { syslog(LOG_ERR, "<%s> prefixlen (%ld) for %s on %s " - "out of range", - __func__, val, addr, intface); - exit(1); + "out of range", + __func__, val, addr, intface); + return (-1); } - rti->prefixlen = (int)val; + rti->rti_prefixlen = (int)val; makeentry(entbuf, sizeof(entbuf), i, "rtflags"); if ((flagstr = (char *)agetstr(entbuf, &bp))) { val = 0; if (strchr(flagstr, 'h')) val |= ND_RA_FLAG_RTPREF_HIGH; if (strchr(flagstr, 'l')) { if ((val & ND_RA_FLAG_RTPREF_HIGH)) { syslog(LOG_ERR, "<%s> the \'h\' and \'l\' route" " preferences are exclusive", __func__); exit(1); } val |= ND_RA_FLAG_RTPREF_LOW; } } else MAYHAVE(val, entbuf, 256); /* XXX */ if (val == 256) { makeentry(oentbuf, sizeof(oentbuf), i, "rtrflags"); MAYHAVE(val, oentbuf, 256); if (val != 256) { fprintf(stderr, "%s was obsoleted. Use %s.\n", - oentbuf, entbuf); + oentbuf, entbuf); } else val = 0; } - rti->rtpref = val & ND_RA_FLAG_RTPREF_MASK; - if (rti->rtpref == ND_RA_FLAG_RTPREF_RSV) { + rti->rti_rtpref = val & ND_RA_FLAG_RTPREF_MASK; + if (rti->rti_rtpref == ND_RA_FLAG_RTPREF_RSV) { syslog(LOG_ERR, "<%s> invalid route preference (%02x) " - "for %s/%d on %s", - __func__, rti->rtpref, addr, - rti->prefixlen, intface); - exit(1); + "for %s/%d on %s", + __func__, rti->rti_rtpref, addr, + rti->rti_prefixlen, intface); + return (-1); } /* * Since the spec does not a default value, we should make * this entry mandatory. However, FreeBSD 4.4 has shipped * with this field being optional, we use the router lifetime * as an ad-hoc default value with a warning message. */ makeentry(entbuf, sizeof(entbuf), i, "rtltime"); MAYHAVE(val64, entbuf, -1); if (val64 == -1) { makeentry(oentbuf, sizeof(oentbuf), i, "rtrltime"); MAYHAVE(val64, oentbuf, -1); - if (val64 != -1) { + if (val64 != -1) fprintf(stderr, "%s was obsoleted. Use %s.\n", - oentbuf, entbuf); - } else { + oentbuf, entbuf); + else { fprintf(stderr, "%s should be specified " - "for interface %s.\n", - entbuf, intface); - val64 = tmp->lifetime; + "for interface %s.\n", entbuf, intface); + val64 = rai->rai_lifetime; } } if (val64 < 0 || val64 > 0xffffffff) { syslog(LOG_ERR, "<%s> route lifetime (%lld) for " "%s/%d on %s out of range", __func__, - (long long)val64, addr, rti->prefixlen, intface); - exit(1); + (long long)val64, addr, rti->rti_prefixlen, intface); + return (-1); } - rti->ltime = (u_int32_t)val64; + rti->rti_ltime = (u_int32_t)val64; } #endif + /* DNS server and DNS search list information */ + for (i = -1; i < MAXRDNSSENT ; i++) { + struct rdnss *rdn; + struct rdnss_addr *rdna; + char *ap; + int c; - /* okey */ - tmp->next = ralist; - ralist = tmp; + makeentry(entbuf, sizeof(entbuf), i, "rdnss"); + addr = (char *)agetstr(entbuf, &bp); + if (addr == NULL) + break; + ELM_MALLOC(rdn, exit(1)); + TAILQ_INIT(&rdn->rd_list); + + for (ap = addr; ap - addr < (ssize_t)strlen(addr); ap += c+1) { + c = strcspn(ap, ","); + strncpy(abuf, ap, c); + abuf[c] = '\0'; + ELM_MALLOC(rdna, exit(1)); + if (inet_pton(AF_INET6, abuf, &rdna->ra_dns) != 1) { + syslog(LOG_ERR, "<%s> inet_pton failed for %s", + __func__, abuf); + free(rdna); + return (-1); + } + TAILQ_INSERT_TAIL(&rdn->rd_list, rdna, ra_next); + } + + makeentry(entbuf, sizeof(entbuf), i, "rdnssltime"); + MAYHAVE(val, entbuf, (rai->rai_maxinterval * 3 / 2)); + if ((u_int)val < rai->rai_maxinterval || + (u_int)val > rai->rai_maxinterval * 2) { + syslog(LOG_ERR, "%s (%ld) on %s is invalid " + "(must be between %d and %d)", + entbuf, val, intface, rai->rai_maxinterval, + rai->rai_maxinterval * 2); + return (-1); + } + rdn->rd_ltime = val; + + /* link into chain */ + TAILQ_INSERT_TAIL(&rai->rai_rdnss, rdn, rd_next); + } + + for (i = -1; i < MAXDNSSLENT ; i++) { + struct dnssl *dns; + struct dnssl_addr *dnsa; + char *ap; + int c; + + makeentry(entbuf, sizeof(entbuf), i, "dnssl"); + addr = (char *)agetstr(entbuf, &bp); + if (addr == NULL) + break; + + ELM_MALLOC(dns, exit(1)); + + TAILQ_INIT(&dns->dn_list); + + for (ap = addr; ap - addr < (ssize_t)strlen(addr); ap += c+1) { + c = strcspn(ap, ","); + strncpy(abuf, ap, c); + abuf[c] = '\0'; + ELM_MALLOC(dnsa, exit(1)); + dnsa->da_len = dname_labelenc(dnsa->da_dom, abuf); + syslog(LOG_DEBUG, "<%s>: dnsa->da_len = %d", __func__, + dnsa->da_len); + TAILQ_INSERT_TAIL(&dns->dn_list, dnsa, da_next); + } + + makeentry(entbuf, sizeof(entbuf), i, "dnsslltime"); + MAYHAVE(val, entbuf, (rai->rai_maxinterval * 3 / 2)); + if ((u_int)val < rai->rai_maxinterval || + (u_int)val > rai->rai_maxinterval * 2) { + syslog(LOG_ERR, "%s (%ld) on %s is invalid " + "(must be between %d and %d)", + entbuf, val, intface, rai->rai_maxinterval, + rai->rai_maxinterval * 2); + return (-1); + } + dns->dn_ltime = val; + + /* link into chain */ + TAILQ_INSERT_TAIL(&rai->rai_dnssl, dns, dn_next); + } /* construct the sending packet */ - make_packet(tmp); + make_packet(rai); + TAILQ_INSERT_TAIL(&railist, rai, rai_next); /* set timer */ - tmp->timer = rtadvd_add_timer(ra_timeout, ra_timer_update, - tmp, tmp); - ra_timer_update((void *)tmp, &tmp->timer->tm); - rtadvd_set_timer(&tmp->timer->tm, tmp->timer); + rai->rai_timer = rtadvd_add_timer(ra_timeout, ra_timer_update, + rai, rai); + ra_timer_update((void *)rai, &rai->rai_timer->rat_tm); + rtadvd_set_timer(&rai->rai_timer->rat_tm, rai->rai_timer); + + return (0); } void get_prefix(struct rainfo *rai) { struct ifaddrs *ifap, *ifa; - struct prefix *pp; + struct prefix *pfx; struct in6_addr *a; u_char *p, *ep, *m, *lim; u_char ntopbuf[INET6_ADDRSTRLEN]; if (getifaddrs(&ifap) < 0) { syslog(LOG_ERR, - "<%s> can't get interface addresses", - __func__); + "<%s> can't get interface addresses", + __func__); exit(1); } for (ifa = ifap; ifa; ifa = ifa->ifa_next) { int plen; - if (strcmp(ifa->ifa_name, rai->ifname) != 0) + if (strcmp(ifa->ifa_name, rai->rai_ifname) != 0) continue; if (ifa->ifa_addr->sa_family != AF_INET6) continue; a = &((struct sockaddr_in6 *)ifa->ifa_addr)->sin6_addr; if (IN6_IS_ADDR_LINKLOCAL(a)) continue; /* get prefix length */ m = (u_char *)&((struct sockaddr_in6 *)ifa->ifa_netmask)->sin6_addr; lim = (u_char *)(ifa->ifa_netmask) + ifa->ifa_netmask->sa_len; plen = prefixlen(m, lim); if (plen <= 0 || plen > 128) { syslog(LOG_ERR, "<%s> failed to get prefixlen " - "or prefix is invalid", - __func__); + "or prefix is invalid", + __func__); exit(1); } if (plen == 128) /* XXX */ continue; if (find_prefix(rai, a, plen)) { /* ignore a duplicated prefix. */ continue; } /* allocate memory to store prefix info. */ - if ((pp = malloc(sizeof(*pp))) == NULL) { - syslog(LOG_ERR, - "<%s> can't get allocate buffer for prefix", - __func__); - exit(1); - } - memset(pp, 0, sizeof(*pp)); + ELM_MALLOC(pfx, exit(1)); /* set prefix, sweep bits outside of prefixlen */ - pp->prefixlen = plen; - memcpy(&pp->prefix, a, sizeof(*a)); - p = (u_char *)&pp->prefix; - ep = (u_char *)(&pp->prefix + 1); + pfx->pfx_prefixlen = plen; + memcpy(&pfx->pfx_prefix, a, sizeof(*a)); + p = (u_char *)&pfx->pfx_prefix; + ep = (u_char *)(&pfx->pfx_prefix + 1); while (m < lim && p < ep) *p++ &= *m++; while (p < ep) *p++ = 0x00; - if (!inet_ntop(AF_INET6, &pp->prefix, ntopbuf, + if (!inet_ntop(AF_INET6, &pfx->pfx_prefix, ntopbuf, sizeof(ntopbuf))) { syslog(LOG_ERR, "<%s> inet_ntop failed", __func__); exit(1); } syslog(LOG_DEBUG, - "<%s> add %s/%d to prefix list on %s", - __func__, ntopbuf, pp->prefixlen, rai->ifname); + "<%s> add %s/%d to prefix list on %s", + __func__, ntopbuf, pfx->pfx_prefixlen, rai->rai_ifname); /* set other fields with protocol defaults */ - pp->validlifetime = DEF_ADVVALIDLIFETIME; - pp->preflifetime = DEF_ADVPREFERREDLIFETIME; - pp->onlinkflg = 1; - pp->autoconfflg = 1; - pp->origin = PREFIX_FROM_KERNEL; - pp->rainfo = rai; + pfx->pfx_validlifetime = DEF_ADVVALIDLIFETIME; + pfx->pfx_preflifetime = DEF_ADVPREFERREDLIFETIME; + pfx->pfx_onlinkflg = 1; + pfx->pfx_autoconfflg = 1; + pfx->pfx_origin = PREFIX_FROM_KERNEL; + pfx->pfx_rainfo = rai; /* link into chain */ - insque(pp, &rai->prefix); + TAILQ_INSERT_TAIL(&rai->rai_prefix, pfx, pfx_next); /* counter increment */ - rai->pfxs++; + rai->rai_pfxs++; } freeifaddrs(ifap); } static void -makeentry(buf, len, id, string) - char *buf; - size_t len; - int id; - char *string; +makeentry(char *buf, size_t len, int id, const char *string) { if (id < 0) strlcpy(buf, string, len); else snprintf(buf, len, "%s%d", string, id); } /* * Add a prefix to the list of specified interface and reconstruct * the outgoing packet. * The prefix must not be in the list. * XXX: other parameters of the prefix (e.g. lifetime) should be * able to be specified. */ static void add_prefix(struct rainfo *rai, struct in6_prefixreq *ipr) { - struct prefix *prefix; + struct prefix *pfx; u_char ntopbuf[INET6_ADDRSTRLEN]; - if ((prefix = malloc(sizeof(*prefix))) == NULL) { - syslog(LOG_ERR, "<%s> memory allocation failed", - __func__); - return; /* XXX: error or exit? */ - } - memset(prefix, 0, sizeof(*prefix)); - prefix->prefix = ipr->ipr_prefix.sin6_addr; - prefix->prefixlen = ipr->ipr_plen; - prefix->validlifetime = ipr->ipr_vltime; - prefix->preflifetime = ipr->ipr_pltime; - prefix->onlinkflg = ipr->ipr_raf_onlink; - prefix->autoconfflg = ipr->ipr_raf_auto; - prefix->origin = PREFIX_FROM_DYNAMIC; + ELM_MALLOC(pfx, return); + pfx->pfx_prefix = ipr->ipr_prefix.sin6_addr; + pfx->pfx_prefixlen = ipr->ipr_plen; + pfx->pfx_validlifetime = ipr->ipr_vltime; + pfx->pfx_preflifetime = ipr->ipr_pltime; + pfx->pfx_onlinkflg = ipr->ipr_raf_onlink; + pfx->pfx_autoconfflg = ipr->ipr_raf_auto; + pfx->pfx_origin = PREFIX_FROM_DYNAMIC; - insque(prefix, &rai->prefix); - prefix->rainfo = rai; + TAILQ_INSERT_TAIL(&rai->rai_prefix, pfx, pfx_next); + pfx->pfx_rainfo = rai; syslog(LOG_DEBUG, "<%s> new prefix %s/%d was added on %s", - __func__, inet_ntop(AF_INET6, &ipr->ipr_prefix.sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - ipr->ipr_plen, rai->ifname); + __func__, + inet_ntop(AF_INET6, &ipr->ipr_prefix.sin6_addr, ntopbuf, + sizeof(ntopbuf)), ipr->ipr_plen, rai->rai_ifname); - /* free the previous packet */ - free(rai->ra_data); - rai->ra_data = NULL; - /* reconstruct the packet */ - rai->pfxs++; + rai->rai_pfxs++; make_packet(rai); } /* * Delete a prefix to the list of specified interface and reconstruct * the outgoing packet. * The prefix must be in the list. */ void -delete_prefix(struct prefix *prefix) +delete_prefix(struct prefix *pfx) { u_char ntopbuf[INET6_ADDRSTRLEN]; - struct rainfo *rai = prefix->rainfo; + struct rainfo *rai; - remque(prefix); + rai = pfx->pfx_rainfo; + TAILQ_REMOVE(&rai->rai_prefix, pfx, pfx_next); syslog(LOG_DEBUG, "<%s> prefix %s/%d was deleted on %s", - __func__, inet_ntop(AF_INET6, &prefix->prefix, - ntopbuf, INET6_ADDRSTRLEN), - prefix->prefixlen, rai->ifname); - if (prefix->timer) - rtadvd_remove_timer(&prefix->timer); - free(prefix); - rai->pfxs--; + __func__, + inet_ntop(AF_INET6, &pfx->pfx_prefix, ntopbuf, + sizeof(ntopbuf)), pfx->pfx_prefixlen, rai->rai_ifname); + if (pfx->pfx_timer) + rtadvd_remove_timer(pfx->pfx_timer); + free(pfx); + rai->rai_pfxs--; + make_packet(rai); } void -invalidate_prefix(struct prefix *prefix) +invalidate_prefix(struct prefix *pfx) { u_char ntopbuf[INET6_ADDRSTRLEN]; struct timeval timo; - struct rainfo *rai = prefix->rainfo; + struct rainfo *rai; - if (prefix->timer) { /* sanity check */ + rai = pfx->pfx_rainfo; + if (pfx->pfx_timer) { /* sanity check */ syslog(LOG_ERR, "<%s> assumption failure: timer already exists", __func__); exit(1); } syslog(LOG_DEBUG, "<%s> prefix %s/%d was invalidated on %s, " "will expire in %ld seconds", __func__, - inet_ntop(AF_INET6, &prefix->prefix, ntopbuf, INET6_ADDRSTRLEN), - prefix->prefixlen, rai->ifname, (long)prefix_timo); + inet_ntop(AF_INET6, &pfx->pfx_prefix, ntopbuf, sizeof(ntopbuf)), + pfx->pfx_prefixlen, rai->rai_ifname, (long)prefix_timo); /* set the expiration timer */ - prefix->timer = rtadvd_add_timer(prefix_timeout, NULL, prefix, NULL); - if (prefix->timer == NULL) { + pfx->pfx_timer = rtadvd_add_timer(prefix_timeout, NULL, pfx, NULL); + if (pfx->pfx_timer == NULL) { syslog(LOG_ERR, "<%s> failed to add a timer for a prefix. " "remove the prefix", __func__); - delete_prefix(prefix); + delete_prefix(pfx); } timo.tv_sec = prefix_timo; timo.tv_usec = 0; - rtadvd_set_timer(&timo, prefix->timer); + rtadvd_set_timer(&timo, pfx->pfx_timer); } static struct rtadvd_timer * prefix_timeout(void *arg) { - struct prefix *prefix = (struct prefix *)arg; - - delete_prefix(prefix); - return(NULL); + delete_prefix((struct prefix *)arg); + + return (NULL); } void -update_prefix(struct prefix * prefix) +update_prefix(struct prefix *pfx) { u_char ntopbuf[INET6_ADDRSTRLEN]; - struct rainfo *rai = prefix->rainfo; + struct rainfo *rai; - if (prefix->timer == NULL) { /* sanity check */ + rai = pfx->pfx_rainfo; + if (pfx->pfx_timer == NULL) { /* sanity check */ syslog(LOG_ERR, "<%s> assumption failure: timer does not exist", __func__); exit(1); } syslog(LOG_DEBUG, "<%s> prefix %s/%d was re-enabled on %s", - __func__, inet_ntop(AF_INET6, &prefix->prefix, ntopbuf, - INET6_ADDRSTRLEN), prefix->prefixlen, rai->ifname); + __func__, inet_ntop(AF_INET6, &pfx->pfx_prefix, ntopbuf, + sizeof(ntopbuf)), pfx->pfx_prefixlen, rai->rai_ifname); /* stop the expiration timer */ - rtadvd_remove_timer(&prefix->timer); + rtadvd_remove_timer(pfx->pfx_timer); + pfx->pfx_timer = NULL; } /* * Try to get an in6_prefixreq contents for a prefix which matches * ipr->ipr_prefix and ipr->ipr_plen and belongs to * the interface whose name is ipr->ipr_name[]. */ static int init_prefix(struct in6_prefixreq *ipr) { #if 0 int s; if ((s = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) { syslog(LOG_ERR, "<%s> socket: %s", __func__, - strerror(errno)); + strerror(errno)); exit(1); } if (ioctl(s, SIOCGIFPREFIX_IN6, (caddr_t)ipr) < 0) { syslog(LOG_INFO, "<%s> ioctl:SIOCGIFPREFIX %s", __func__, - strerror(errno)); + strerror(errno)); ipr->ipr_vltime = DEF_ADVVALIDLIFETIME; ipr->ipr_pltime = DEF_ADVPREFERREDLIFETIME; ipr->ipr_raf_onlink = 1; ipr->ipr_raf_auto = 1; /* omit other field initialization */ } else if (ipr->ipr_origin < PR_ORIG_RR) { u_char ntopbuf[INET6_ADDRSTRLEN]; syslog(LOG_WARNING, "<%s> Added prefix(%s)'s origin %d is" - "lower than PR_ORIG_RR(router renumbering)." - "This should not happen if I am router", __func__, - inet_ntop(AF_INET6, &ipr->ipr_prefix.sin6_addr, ntopbuf, - sizeof(ntopbuf)), ipr->ipr_origin); + "lower than PR_ORIG_RR(router renumbering)." + "This should not happen if I am router", __func__, + inet_ntop(AF_INET6, &ipr->ipr_prefix.sin6_addr, ntopbuf, + sizeof(ntopbuf)), ipr->ipr_origin); close(s); - return 1; + return (1); } close(s); - return 0; + return (0); #else ipr->ipr_vltime = DEF_ADVVALIDLIFETIME; ipr->ipr_pltime = DEF_ADVPREFERREDLIFETIME; ipr->ipr_raf_onlink = 1; ipr->ipr_raf_auto = 1; - return 0; + return (0); #endif } void make_prefix(struct rainfo *rai, int ifindex, struct in6_addr *addr, int plen) { struct in6_prefixreq ipr; memset(&ipr, 0, sizeof(ipr)); if (if_indextoname(ifindex, ipr.ipr_name) == NULL) { syslog(LOG_ERR, "<%s> Prefix added interface No.%d doesn't" - "exist. This should not happen! %s", __func__, - ifindex, strerror(errno)); + "exist. This should not happen! %s", __func__, + ifindex, strerror(errno)); exit(1); } ipr.ipr_prefix.sin6_len = sizeof(ipr.ipr_prefix); ipr.ipr_prefix.sin6_family = AF_INET6; ipr.ipr_prefix.sin6_addr = *addr; ipr.ipr_plen = plen; if (init_prefix(&ipr)) return; /* init failed by some error */ add_prefix(rai, &ipr); } void -make_packet(struct rainfo *rainfo) +make_packet(struct rainfo *rai) { size_t packlen, lladdroptlen = 0; char *buf; struct nd_router_advert *ra; struct nd_opt_prefix_info *ndopt_pi; struct nd_opt_mtu *ndopt_mtu; #ifdef ROUTEINFO struct nd_opt_route_info *ndopt_rti; struct rtinfo *rti; #endif + struct nd_opt_rdnss *ndopt_rdnss; + struct rdnss *rdn; + struct nd_opt_dnssl *ndopt_dnssl; + struct dnssl *dns; + size_t len; struct prefix *pfx; /* calculate total length */ packlen = sizeof(struct nd_router_advert); - if (rainfo->advlinkopt) { - if ((lladdroptlen = lladdropt_length(rainfo->sdl)) == 0) { + if (rai->rai_advlinkopt) { + if ((lladdroptlen = lladdropt_length(rai->rai_sdl)) == 0) { syslog(LOG_INFO, - "<%s> link-layer address option has" - " null length on %s. Treat as not included.", - __func__, rainfo->ifname); - rainfo->advlinkopt = 0; + "<%s> link-layer address option has" + " null length on %s. Treat as not included.", + __func__, rai->rai_ifname); + rai->rai_advlinkopt = 0; } packlen += lladdroptlen; } - if (rainfo->pfxs) - packlen += sizeof(struct nd_opt_prefix_info) * rainfo->pfxs; - if (rainfo->linkmtu) + if (rai->rai_pfxs) + packlen += sizeof(struct nd_opt_prefix_info) * rai->rai_pfxs; + if (rai->rai_linkmtu) packlen += sizeof(struct nd_opt_mtu); #ifdef ROUTEINFO - for (rti = rainfo->route.next; rti != &rainfo->route; rti = rti->next) - packlen += sizeof(struct nd_opt_route_info) + - ((rti->prefixlen + 0x3f) >> 6) * 8; + TAILQ_FOREACH(rti, &rai->rai_route, rti_next) + packlen += sizeof(struct nd_opt_route_info) + + ((rti->rti_prefixlen + 0x3f) >> 6) * 8; #endif + TAILQ_FOREACH(rdn, &rai->rai_rdnss, rd_next) { + struct rdnss_addr *rdna; + packlen += sizeof(struct nd_opt_rdnss); + TAILQ_FOREACH(rdna, &rdn->rd_list, ra_next) + packlen += sizeof(rdna->ra_dns); + } + TAILQ_FOREACH(dns, &rai->rai_dnssl, dn_next) { + struct dnssl_addr *dnsa; + + packlen += sizeof(struct nd_opt_dnssl); + len = 0; + TAILQ_FOREACH(dnsa, &dns->dn_list, da_next) + len += dnsa->da_len; + + /* A zero octet and 8 octet boundary */ + len++; + len += (len % 8) ? 8 - len % 8 : 0; + + packlen += len; + } /* allocate memory for the packet */ if ((buf = malloc(packlen)) == NULL) { syslog(LOG_ERR, - "<%s> can't get enough memory for an RA packet", - __func__); + "<%s> can't get enough memory for an RA packet", + __func__); exit(1); } - if (rainfo->ra_data) { - /* free the previous packet */ - free(rainfo->ra_data); - rainfo->ra_data = NULL; - } - rainfo->ra_data = buf; + memset(buf, 0, packlen); + if (rai->rai_ra_data) /* Free old data if any. */ + free(rai->rai_ra_data); + rai->rai_ra_data = buf; /* XXX: what if packlen > 576? */ - rainfo->ra_datalen = packlen; + rai->rai_ra_datalen = packlen; /* * construct the packet */ ra = (struct nd_router_advert *)buf; ra->nd_ra_type = ND_ROUTER_ADVERT; ra->nd_ra_code = 0; ra->nd_ra_cksum = 0; - ra->nd_ra_curhoplimit = (u_int8_t)(0xff & rainfo->hoplimit); + ra->nd_ra_curhoplimit = (u_int8_t)(0xff & rai->rai_hoplimit); ra->nd_ra_flags_reserved = 0; /* just in case */ /* * XXX: the router preference field, which is a 2-bit field, should be * initialized before other fields. */ - ra->nd_ra_flags_reserved = 0xff & rainfo->rtpref; + ra->nd_ra_flags_reserved = 0xff & rai->rai_rtpref; ra->nd_ra_flags_reserved |= - rainfo->managedflg ? ND_RA_FLAG_MANAGED : 0; + rai->rai_managedflg ? ND_RA_FLAG_MANAGED : 0; ra->nd_ra_flags_reserved |= - rainfo->otherflg ? ND_RA_FLAG_OTHER : 0; - ra->nd_ra_router_lifetime = htons(rainfo->lifetime); - ra->nd_ra_reachable = htonl(rainfo->reachabletime); - ra->nd_ra_retransmit = htonl(rainfo->retranstimer); + rai->rai_otherflg ? ND_RA_FLAG_OTHER : 0; + ra->nd_ra_router_lifetime = htons(rai->rai_lifetime); + ra->nd_ra_reachable = htonl(rai->rai_reachabletime); + ra->nd_ra_retransmit = htonl(rai->rai_retranstimer); buf += sizeof(*ra); - if (rainfo->advlinkopt) { - lladdropt_fill(rainfo->sdl, (struct nd_opt_hdr *)buf); + if (rai->rai_advlinkopt) { + lladdropt_fill(rai->rai_sdl, (struct nd_opt_hdr *)buf); buf += lladdroptlen; } - if (rainfo->linkmtu) { + if (rai->rai_linkmtu) { ndopt_mtu = (struct nd_opt_mtu *)buf; ndopt_mtu->nd_opt_mtu_type = ND_OPT_MTU; ndopt_mtu->nd_opt_mtu_len = 1; ndopt_mtu->nd_opt_mtu_reserved = 0; - ndopt_mtu->nd_opt_mtu_mtu = htonl(rainfo->linkmtu); + ndopt_mtu->nd_opt_mtu_mtu = htonl(rai->rai_linkmtu); buf += sizeof(struct nd_opt_mtu); } - for (pfx = rainfo->prefix.next; - pfx != &rainfo->prefix; pfx = pfx->next) { + TAILQ_FOREACH(pfx, &rai->rai_prefix, pfx_next) { u_int32_t vltime, pltime; struct timeval now; ndopt_pi = (struct nd_opt_prefix_info *)buf; ndopt_pi->nd_opt_pi_type = ND_OPT_PREFIX_INFORMATION; ndopt_pi->nd_opt_pi_len = 4; - ndopt_pi->nd_opt_pi_prefix_len = pfx->prefixlen; + ndopt_pi->nd_opt_pi_prefix_len = pfx->pfx_prefixlen; ndopt_pi->nd_opt_pi_flags_reserved = 0; - if (pfx->onlinkflg) + if (pfx->pfx_onlinkflg) ndopt_pi->nd_opt_pi_flags_reserved |= ND_OPT_PI_FLAG_ONLINK; - if (pfx->autoconfflg) + if (pfx->pfx_autoconfflg) ndopt_pi->nd_opt_pi_flags_reserved |= ND_OPT_PI_FLAG_AUTO; - if (pfx->timer) + if (pfx->pfx_timer) vltime = 0; else { - if (pfx->vltimeexpire || pfx->pltimeexpire) + if (pfx->pfx_vltimeexpire || pfx->pfx_pltimeexpire) gettimeofday(&now, NULL); - if (pfx->vltimeexpire == 0) - vltime = pfx->validlifetime; + if (pfx->pfx_vltimeexpire == 0) + vltime = pfx->pfx_validlifetime; else - vltime = (pfx->vltimeexpire > now.tv_sec) ? - pfx->vltimeexpire - now.tv_sec : 0; + vltime = (pfx->pfx_vltimeexpire > now.tv_sec) ? + pfx->pfx_vltimeexpire - now.tv_sec : 0; } - if (pfx->timer) + if (pfx->pfx_timer) pltime = 0; else { - if (pfx->pltimeexpire == 0) - pltime = pfx->preflifetime; + if (pfx->pfx_pltimeexpire == 0) + pltime = pfx->pfx_preflifetime; else - pltime = (pfx->pltimeexpire > now.tv_sec) ? - pfx->pltimeexpire - now.tv_sec : 0; + pltime = (pfx->pfx_pltimeexpire > now.tv_sec) ? + pfx->pfx_pltimeexpire - now.tv_sec : 0; } if (vltime < pltime) { /* * this can happen if vltime is decrement but pltime * is not. */ pltime = vltime; } ndopt_pi->nd_opt_pi_valid_time = htonl(vltime); ndopt_pi->nd_opt_pi_preferred_time = htonl(pltime); ndopt_pi->nd_opt_pi_reserved2 = 0; - ndopt_pi->nd_opt_pi_prefix = pfx->prefix; + ndopt_pi->nd_opt_pi_prefix = pfx->pfx_prefix; buf += sizeof(struct nd_opt_prefix_info); } #ifdef ROUTEINFO - for (rti = rainfo->route.next; rti != &rainfo->route; rti = rti->next) { - u_int8_t psize = (rti->prefixlen + 0x3f) >> 6; + TAILQ_FOREACH(rti, &rai->rai_route, rti_next) { + u_int8_t psize = (rti->rti_prefixlen + 0x3f) >> 6; ndopt_rti = (struct nd_opt_route_info *)buf; ndopt_rti->nd_opt_rti_type = ND_OPT_ROUTE_INFO; ndopt_rti->nd_opt_rti_len = 1 + psize; - ndopt_rti->nd_opt_rti_prefixlen = rti->prefixlen; - ndopt_rti->nd_opt_rti_flags = 0xff & rti->rtpref; - ndopt_rti->nd_opt_rti_lifetime = htonl(rti->ltime); - memcpy(ndopt_rti + 1, &rti->prefix, psize * 8); + ndopt_rti->nd_opt_rti_prefixlen = rti->rti_prefixlen; + ndopt_rti->nd_opt_rti_flags = 0xff & rti->rti_rtpref; + ndopt_rti->nd_opt_rti_lifetime = htonl(rti->rti_ltime); + memcpy(ndopt_rti + 1, &rti->rti_prefix, psize * 8); buf += sizeof(struct nd_opt_route_info) + psize * 8; } #endif + TAILQ_FOREACH(rdn, &rai->rai_rdnss, rd_next) { + struct rdnss_addr *rdna; - return; -} + ndopt_rdnss = (struct nd_opt_rdnss *)buf; + ndopt_rdnss->nd_opt_rdnss_type = ND_OPT_RDNSS; + ndopt_rdnss->nd_opt_rdnss_len = 0; + ndopt_rdnss->nd_opt_rdnss_reserved = 0; + ndopt_rdnss->nd_opt_rdnss_lifetime = htonl(rdn->rd_ltime); + buf += sizeof(struct nd_opt_rdnss); -static int -getinet6sysctl(int code) -{ - int mib[] = { CTL_NET, PF_INET6, IPPROTO_IPV6, 0 }; - int value; - size_t size; + TAILQ_FOREACH(rdna, &rdn->rd_list, ra_next) { + memcpy(buf, &rdna->ra_dns, sizeof(rdna->ra_dns)); + buf += sizeof(rdna->ra_dns); + } + /* Length field should be in 8 octets */ + ndopt_rdnss->nd_opt_rdnss_len = (buf - (char *)ndopt_rdnss) / 8; - mib[3] = code; - size = sizeof(value); - if (sysctl(mib, sizeof(mib)/sizeof(mib[0]), &value, &size, NULL, 0) - < 0) { - syslog(LOG_ERR, "<%s>: failed to get ip6 sysctl(%d): %s", - __func__, code, - strerror(errno)); - return(-1); + syslog(LOG_DEBUG, "<%s>: nd_opt_dnss_len = %d", __func__, + ndopt_rdnss->nd_opt_rdnss_len); } - else - return(value); + TAILQ_FOREACH(dns, &rai->rai_dnssl, dn_next) { + struct dnssl_addr *dnsa; + + ndopt_dnssl = (struct nd_opt_dnssl *)buf; + ndopt_dnssl->nd_opt_dnssl_type = ND_OPT_DNSSL; + ndopt_dnssl->nd_opt_dnssl_len = 0; + ndopt_dnssl->nd_opt_dnssl_reserved = 0; + ndopt_dnssl->nd_opt_dnssl_lifetime = htonl(dns->dn_ltime); + buf += sizeof(*ndopt_dnssl); + + TAILQ_FOREACH(dnsa, &dns->dn_list, da_next) { + memcpy(buf, dnsa->da_dom, dnsa->da_len); + buf += dnsa->da_len; + } + + /* A zero octet after encoded DNS server list. */ + *buf++ = '\0'; + + /* Padding to next 8 octets boundary */ + len = buf - (char *)ndopt_dnssl; + len += (len % 8) ? 8 - len % 8 : 0; + + /* Length field must be in 8 octets */ + ndopt_dnssl->nd_opt_dnssl_len = len / 8; + + syslog(LOG_DEBUG, "<%s>: nd_opt_dnssl_len = %d", __func__, + ndopt_dnssl->nd_opt_dnssl_len); + } + return; } Index: projects/largeSMP/usr.sbin/rtadvd/config.h =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/config.h (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/config.h (revision 222783) @@ -1,47 +1,50 @@ /* $FreeBSD$ */ /* $KAME: config.h,v 1.8 2003/06/17 08:26:22 itojun Exp $ */ /* * Copyright (C) 1995, 1996, 1997, 1998, and 1999 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. */ -extern void getconfig(char *); +extern int getconfig(int); +extern int rmconfig(int); extern void delete_prefix(struct prefix *); extern void invalidate_prefix(struct prefix *); extern void update_prefix(struct prefix *); extern void make_prefix(struct rainfo *, int, struct in6_addr *, int); extern void make_packet(struct rainfo *); extern void get_prefix(struct rainfo *); /* * it is highly unlikely to have 100 prefix information options, * so it should be okay to limit it */ #define MAXPREFIX 100 #define MAXROUTE 100 +#define MAXRDNSSENT 100 +#define MAXDNSSLENT 100 Index: projects/largeSMP/usr.sbin/rtadvd/dump.c =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/dump.c (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/dump.c (revision 222783) @@ -1,252 +1,321 @@ /* $FreeBSD$ */ /* $KAME: dump.c,v 1.32 2003/05/19 09:46:50 keiichi Exp $ */ /* * Copyright (C) 2000 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. */ #include #include #include #include #include #include #include /* XXX: the following two are non-standard include files */ #include #include #include +#include #include #include #include #include #include #include #include "rtadvd.h" #include "timer.h" #include "if.h" #include "dump.h" static FILE *fp; extern struct rainfo *ralist; static char *ether_str(struct sockaddr_dl *); static void if_dump(void); +static size_t dname_labeldec(char *, size_t, const char *); -static char *rtpref_str[] = { +static const char *rtpref_str[] = { "medium", /* 00 */ "high", /* 01 */ "rsv", /* 10 */ "low" /* 11 */ }; static char * -ether_str(sdl) - struct sockaddr_dl *sdl; +ether_str(struct sockaddr_dl *sdl) { static char hbuf[32]; u_char *cp; if (sdl->sdl_alen && sdl->sdl_alen > 5) { cp = (u_char *)LLADDR(sdl); snprintf(hbuf, sizeof(hbuf), "%x:%x:%x:%x:%x:%x", cp[0], cp[1], cp[2], cp[3], cp[4], cp[5]); } else snprintf(hbuf, sizeof(hbuf), "NONE"); - return(hbuf); + return (hbuf); } static void -if_dump() +if_dump(void) { struct rainfo *rai; struct prefix *pfx; #ifdef ROUTEINFO struct rtinfo *rti; #endif + struct rdnss *rdn; + struct dnssl *dns; char prefixbuf[INET6_ADDRSTRLEN]; - int first; struct timeval now; gettimeofday(&now, NULL); /* XXX: unused in most cases */ - for (rai = ralist; rai; rai = rai->next) { - fprintf(fp, "%s:\n", rai->ifname); + TAILQ_FOREACH(rai, &railist, rai_next) { + fprintf(fp, "%s:\n", rai->rai_ifname); fprintf(fp, " Status: %s\n", - (iflist[rai->ifindex]->ifm_flags & IFF_UP) ? "UP" : - "DOWN"); + (iflist[rai->rai_ifindex]->ifm_flags & IFF_UP) ? "UP" : + "DOWN"); /* control information */ - if (rai->lastsent.tv_sec) { + if (rai->rai_lastsent.tv_sec) { /* note that ctime() appends CR by itself */ fprintf(fp, " Last RA sent: %s", - ctime((time_t *)&rai->lastsent.tv_sec)); + ctime((time_t *)&rai->rai_lastsent.tv_sec)); } - if (rai->timer) { + if (rai->rai_timer) fprintf(fp, " Next RA will be sent: %s", - ctime((time_t *)&rai->timer->tm.tv_sec)); - } + ctime((time_t *)&rai->rai_timer->rat_tm.tv_sec)); else fprintf(fp, " RA timer is stopped"); fprintf(fp, " waits: %d, initcount: %d\n", - rai->waiting, rai->initcounter); + rai->rai_waiting, rai->rai_initcounter); /* statistics */ fprintf(fp, " statistics: RA(out/in/inconsistent): " "%llu/%llu/%llu, ", - (unsigned long long)rai->raoutput, - (unsigned long long)rai->rainput, - (unsigned long long)rai->rainconsistent); + (unsigned long long)rai->rai_raoutput, + (unsigned long long)rai->rai_rainput, + (unsigned long long)rai->rai_rainconsistent); fprintf(fp, "RS(input): %llu\n", - (unsigned long long)rai->rsinput); + (unsigned long long)rai->rai_rsinput); /* interface information */ - if (rai->advlinkopt) + if (rai->rai_advlinkopt) fprintf(fp, " Link-layer address: %s\n", - ether_str(rai->sdl)); - fprintf(fp, " MTU: %d\n", rai->phymtu); + ether_str(rai->rai_sdl)); + fprintf(fp, " MTU: %d\n", rai->rai_phymtu); /* Router configuration variables */ fprintf(fp, " DefaultLifetime: %d, MaxAdvInterval: %d, " - "MinAdvInterval: %d\n", rai->lifetime, rai->maxinterval, - rai->mininterval); - fprintf(fp, " Flags: %s%s%s, ", - rai->managedflg ? "M" : "", rai->otherflg ? "O" : "", ""); + "MinAdvInterval: %d\n", rai->rai_lifetime, + rai->rai_maxinterval, rai->rai_mininterval); + fprintf(fp, " Flags: "); + if (rai->rai_managedflg || rai->rai_otherflg) { + fprintf(fp, "%s", rai->rai_managedflg ? "M" : ""); + fprintf(fp, "%s", rai->rai_otherflg ? "O" : ""); + } else + fprintf(fp, ""); + fprintf(fp, ", "); fprintf(fp, "Preference: %s, ", - rtpref_str[(rai->rtpref >> 3) & 0xff]); - fprintf(fp, "MTU: %d\n", rai->linkmtu); + rtpref_str[(rai->rai_rtpref >> 3) & 0xff]); + fprintf(fp, "MTU: %d\n", rai->rai_linkmtu); fprintf(fp, " ReachableTime: %d, RetransTimer: %d, " - "CurHopLimit: %d\n", rai->reachabletime, - rai->retranstimer, rai->hoplimit); - if (rai->clockskew) + "CurHopLimit: %d\n", rai->rai_reachabletime, + rai->rai_retranstimer, rai->rai_hoplimit); + if (rai->rai_clockskew) fprintf(fp, " Clock skew: %ldsec\n", - rai->clockskew); - for (first = 1, pfx = rai->prefix.next; pfx != &rai->prefix; - pfx = pfx->next) { - if (first) { + rai->rai_clockskew); + TAILQ_FOREACH(pfx, &rai->rai_prefix, pfx_next) { + if (pfx == TAILQ_FIRST(&rai->rai_prefix)) fprintf(fp, " Prefixes:\n"); - first = 0; - } fprintf(fp, " %s/%d(", - inet_ntop(AF_INET6, &pfx->prefix, prefixbuf, - sizeof(prefixbuf)), pfx->prefixlen); - switch (pfx->origin) { + inet_ntop(AF_INET6, &pfx->pfx_prefix, prefixbuf, + sizeof(prefixbuf)), pfx->pfx_prefixlen); + switch (pfx->pfx_origin) { case PREFIX_FROM_KERNEL: fprintf(fp, "KERNEL, "); break; case PREFIX_FROM_CONFIG: fprintf(fp, "CONFIG, "); break; case PREFIX_FROM_DYNAMIC: fprintf(fp, "DYNAMIC, "); break; } - if (pfx->validlifetime == ND6_INFINITE_LIFETIME) + if (pfx->pfx_validlifetime == ND6_INFINITE_LIFETIME) fprintf(fp, "vltime: infinity"); else fprintf(fp, "vltime: %ld", - (long)pfx->validlifetime); - if (pfx->vltimeexpire != 0) - fprintf(fp, "(decr,expire %ld), ", (long) - pfx->vltimeexpire > now.tv_sec ? - pfx->vltimeexpire - now.tv_sec : 0); + (long)pfx->pfx_validlifetime); + if (pfx->pfx_vltimeexpire != 0) + fprintf(fp, "(decr,expire %ld), ", + (long)pfx->pfx_vltimeexpire > now.tv_sec ? + (long)pfx->pfx_vltimeexpire - now.tv_sec : + 0); else fprintf(fp, ", "); - if (pfx->preflifetime == ND6_INFINITE_LIFETIME) + if (pfx->pfx_preflifetime == ND6_INFINITE_LIFETIME) fprintf(fp, "pltime: infinity"); else fprintf(fp, "pltime: %ld", - (long)pfx->preflifetime); - if (pfx->pltimeexpire != 0) - fprintf(fp, "(decr,expire %ld), ", (long) - pfx->pltimeexpire > now.tv_sec ? - pfx->pltimeexpire - now.tv_sec : 0); + (long)pfx->pfx_preflifetime); + if (pfx->pfx_pltimeexpire != 0) + fprintf(fp, "(decr,expire %ld), ", + (long)pfx->pfx_pltimeexpire > now.tv_sec ? + (long)pfx->pfx_pltimeexpire - now.tv_sec : + 0); else fprintf(fp, ", "); - fprintf(fp, "flags: %s%s%s", - pfx->onlinkflg ? "L" : "", - pfx->autoconfflg ? "A" : "", - ""); - if (pfx->timer) { + fprintf(fp, "flags: "); + if (pfx->pfx_onlinkflg || pfx->pfx_autoconfflg) { + fprintf(fp, "%s", + pfx->pfx_onlinkflg ? "L" : ""); + fprintf(fp, "%s", + pfx->pfx_autoconfflg ? "A" : ""); + } else + fprintf(fp, ""); + if (pfx->pfx_timer) { struct timeval *rest; - rest = rtadvd_timer_rest(pfx->timer); + rest = rtadvd_timer_rest(pfx->pfx_timer); if (rest) { /* XXX: what if not? */ fprintf(fp, ", expire in: %ld", (long)rest->tv_sec); } } fprintf(fp, ")\n"); } #ifdef ROUTEINFO - for (first = 1, rti = rai->route.next; rti != &rai->route; - rti = rti->next) { - if (first) { + TAILQ_FOREACH(rti, &rai->rai_route, rti_next) { + if (rti == TAILQ_FIRST(&rai->rai_route)) fprintf(fp, " Route Information:\n"); - first = 0; - } fprintf(fp, " %s/%d (", - inet_ntop(AF_INET6, &rti->prefix, - prefixbuf, sizeof(prefixbuf)), - rti->prefixlen); + inet_ntop(AF_INET6, &rti->rti_prefix, + prefixbuf, sizeof(prefixbuf)), + rti->rti_prefixlen); fprintf(fp, "preference: %s, ", - rtpref_str[0xff & (rti->rtpref >> 3)]); - if (rti->ltime == ND6_INFINITE_LIFETIME) + rtpref_str[0xff & (rti->rti_rtpref >> 3)]); + if (rti->rti_ltime == ND6_INFINITE_LIFETIME) fprintf(fp, "lifetime: infinity"); else - fprintf(fp, "lifetime: %ld", (long)rti->ltime); + fprintf(fp, "lifetime: %ld", + (long)rti->rti_ltime); fprintf(fp, ")\n"); } #endif + TAILQ_FOREACH(rdn, &rai->rai_rdnss, rd_next) { + struct rdnss_addr *rdna; + + if (rdn == TAILQ_FIRST(&rai->rai_rdnss)) + fprintf(fp, " Recursive DNS servers:\n" + " Lifetime\tServers\n"); + + fprintf(fp, " %8u\t", rdn->rd_ltime); + TAILQ_FOREACH(rdna, &rdn->rd_list, ra_next) { + inet_ntop(AF_INET6, &rdna->ra_dns, + prefixbuf, sizeof(prefixbuf)); + + if (rdna != TAILQ_FIRST(&rdn->rd_list)) + fprintf(fp, " \t"); + fprintf(fp, "%s\n", prefixbuf); + } + fprintf(fp, "\n"); + } + + TAILQ_FOREACH(dns, &rai->rai_dnssl, dn_next) { + struct dnssl_addr *dnsa; + char buf[NI_MAXHOST]; + + if (dns == TAILQ_FIRST(&rai->rai_dnssl)) + fprintf(fp, " DNS search list:\n" + " Lifetime\tDomains\n"); + + fprintf(fp, " %8u\t", dns->dn_ltime); + TAILQ_FOREACH(dnsa, &dns->dn_list, da_next) { + dname_labeldec(buf, sizeof(buf), dnsa->da_dom); + if (dnsa != TAILQ_FIRST(&dns->dn_list)) + fprintf(fp, " \t"); + fprintf(fp, "%s(%d)\n", buf, dnsa->da_len); + } + fprintf(fp, "\n"); + } } } void -rtadvd_dump_file(dumpfile) - char *dumpfile; +rtadvd_dump_file(const char *dumpfile) { syslog(LOG_DEBUG, "<%s> dump current status to %s", __func__, dumpfile); if ((fp = fopen(dumpfile, "w")) == NULL) { syslog(LOG_WARNING, "<%s> open a dump file(%s)", __func__, dumpfile); return; } if_dump(); fclose(fp); +} + +/* Decode domain name label encoding in RFC 1035 Section 3.1 */ +static size_t +dname_labeldec(char *dst, size_t dlen, const char *src) +{ + size_t len; + const char *src_origin; + const char *src_last; + const char *dst_origin; + + src_origin = src; + src_last = strchr(src, '\0'); + dst_origin = dst; + memset(dst, '\0', dlen); + while (src && (len = (uint8_t)(*src++) & 0x3f) && + (src + len) <= src_last) { + if (dst != dst_origin) + *dst++ = '.'; + syslog(LOG_DEBUG, "<%s> labellen = %zd", __func__, len); + memcpy(dst, src, len); + src += len; + dst += len; + } + *dst = '\0'; + + return (src - src_origin); } Index: projects/largeSMP/usr.sbin/rtadvd/dump.h =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/dump.h (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/dump.h (revision 222783) @@ -1,33 +1,33 @@ /* $FreeBSD$ */ /* $KAME: dump.h,v 1.1 2000/05/23 11:31:26 itojun Exp $ */ /* * Copyright (C) 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. */ -extern void rtadvd_dump_file(char *); +extern void rtadvd_dump_file(const char *); Index: projects/largeSMP/usr.sbin/rtadvd/if.c =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/if.c (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/if.c (revision 222783) @@ -1,580 +1,591 @@ /* $FreeBSD$ */ /* $KAME: if.c,v 1.17 2001/01/21 15:27:30 itojun Exp $ */ /* * 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. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include +#include #include #include #include #include "rtadvd.h" #include "if.h" -#define ROUNDUP(a, size) \ +#define ROUNDUP(a, size) \ (((a) & ((size)-1)) ? (1 + ((a) | ((size)-1))) : (a)) -#define NEXT_SA(ap) (ap) = (struct sockaddr *) \ - ((caddr_t)(ap) + ((ap)->sa_len ? ROUNDUP((ap)->sa_len,\ - sizeof(u_long)) :\ - sizeof(u_long))) +#define NEXT_SA(ap) \ + (ap) = (struct sockaddr *)((caddr_t)(ap) + \ + ((ap)->sa_len ? ROUNDUP((ap)->sa_len, sizeof(u_long)) : \ + sizeof(u_long))) struct if_msghdr **iflist; int iflist_init_ok; size_t ifblock_size; char *ifblock; -static void get_iflist(char **buf, size_t *size); -static void parse_iflist(struct if_msghdr ***ifmlist_p, char *buf, - size_t bufsize); +static void get_iflist(char **buf, size_t *size); +static void parse_iflist(struct if_msghdr ***ifmlist_p, + char *buf, size_t bufsize); static void get_rtaddrs(int addrs, struct sockaddr *sa, struct sockaddr **rti_info) { int i; - + for (i = 0; i < RTAX_MAX; i++) { if (addrs & (1 << i)) { rti_info[i] = sa; NEXT_SA(sa); } else rti_info[i] = NULL; } } struct sockaddr_dl * if_nametosdl(char *name) { int mib[6] = {CTL_NET, AF_ROUTE, 0, 0, NET_RT_IFLIST, 0}; char *buf, *next, *lim; size_t len; struct if_msghdr *ifm; struct sockaddr *sa, *rti_info[RTAX_MAX]; struct sockaddr_dl *sdl = NULL, *ret_sdl; if (sysctl(mib, 6, NULL, &len, NULL, 0) < 0) - return(NULL); + return (NULL); if ((buf = malloc(len)) == NULL) - return(NULL); + return (NULL); if (sysctl(mib, 6, buf, &len, NULL, 0) < 0) { free(buf); - return(NULL); + return (NULL); } lim = buf + len; for (next = buf; next < lim; next += ifm->ifm_msglen) { ifm = (struct if_msghdr *)next; if (ifm->ifm_type == RTM_IFINFO) { sa = (struct sockaddr *)(ifm + 1); get_rtaddrs(ifm->ifm_addrs, sa, rti_info); if ((sa = rti_info[RTAX_IFP]) != NULL) { if (sa->sa_family == AF_LINK) { sdl = (struct sockaddr_dl *)sa; if (strlen(name) != sdl->sdl_nlen) continue; /* not same len */ if (strncmp(&sdl->sdl_data[0], name, sdl->sdl_nlen) == 0) { break; } } } } } if (next == lim) { /* search failed */ free(buf); - return(NULL); + return (NULL); } if ((ret_sdl = malloc(sdl->sdl_len)) == NULL) goto end; memcpy((caddr_t)ret_sdl, (caddr_t)sdl, sdl->sdl_len); end: free(buf); - return(ret_sdl); + return (ret_sdl); } int if_getmtu(char *name) { struct ifaddrs *ifap, *ifa; struct if_data *ifd; u_long mtu = 0; if (getifaddrs(&ifap) < 0) - return(0); + return (0); for (ifa = ifap; ifa; ifa = ifa->ifa_next) { if (strcmp(ifa->ifa_name, name) == 0) { ifd = ifa->ifa_data; if (ifd) mtu = ifd->ifi_mtu; break; } } freeifaddrs(ifap); #ifdef SIOCGIFMTU /* XXX: this ifdef may not be necessary */ if (mtu == 0) { struct ifreq ifr; int s; if ((s = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) - return(0); + return (0); ifr.ifr_addr.sa_family = AF_INET6; strncpy(ifr.ifr_name, name, sizeof(ifr.ifr_name)); if (ioctl(s, SIOCGIFMTU, (caddr_t)&ifr) < 0) { close(s); - return(0); + return (0); } close(s); mtu = ifr.ifr_mtu; } #endif - return(mtu); + return (mtu); } /* give interface index and its old flags, then new flags returned */ int if_getflags(int ifindex, int oifflags) { struct ifreq ifr; int s; if ((s = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) { syslog(LOG_ERR, "<%s> socket: %s", __func__, - strerror(errno)); + strerror(errno)); return (oifflags & ~IFF_UP); } if_indextoname(ifindex, ifr.ifr_name); if (ioctl(s, SIOCGIFFLAGS, (caddr_t)&ifr) < 0) { syslog(LOG_ERR, "<%s> ioctl:SIOCGIFFLAGS: failed for %s", - __func__, ifr.ifr_name); + __func__, ifr.ifr_name); close(s); return (oifflags & ~IFF_UP); } close(s); return (ifr.ifr_flags); } #define ROUNDUP8(a) (1 + (((a) - 1) | 7)) int lladdropt_length(struct sockaddr_dl *sdl) { switch (sdl->sdl_type) { case IFT_ETHER: - return(ROUNDUP8(ETHER_ADDR_LEN + 2)); + return (ROUNDUP8(ETHER_ADDR_LEN + 2)); default: - return(0); + return (0); } } void lladdropt_fill(struct sockaddr_dl *sdl, struct nd_opt_hdr *ndopt) { char *addr; ndopt->nd_opt_type = ND_OPT_SOURCE_LINKADDR; /* fixed */ switch (sdl->sdl_type) { case IFT_ETHER: ndopt->nd_opt_len = (ROUNDUP8(ETHER_ADDR_LEN + 2)) >> 3; addr = (char *)(ndopt + 1); memcpy(addr, LLADDR(sdl), ETHER_ADDR_LEN); break; default: syslog(LOG_ERR, "<%s> unsupported link type(%d)", __func__, sdl->sdl_type); exit(1); } return; } int -rtbuf_len() +rtbuf_len(void) { size_t len; - int mib[6] = {CTL_NET, AF_ROUTE, 0, AF_INET6, NET_RT_DUMP, 0}; if (sysctl(mib, 6, NULL, &len, NULL, 0) < 0) - return(-1); + return (-1); - return(len); + return (len); } #define FILTER_MATCH(type, filter) ((0x1 << type) & filter) #define SIN6(s) ((struct sockaddr_in6 *)(s)) #define SDL(s) ((struct sockaddr_dl *)(s)) char * get_next_msg(char *buf, char *lim, int ifindex, size_t *lenp, int filter) { struct rt_msghdr *rtm; struct ifa_msghdr *ifam; struct sockaddr *sa, *dst, *gw, *ifa, *rti_info[RTAX_MAX]; *lenp = 0; for (rtm = (struct rt_msghdr *)buf; rtm < (struct rt_msghdr *)lim; rtm = (struct rt_msghdr *)(((char *)rtm) + rtm->rtm_msglen)) { /* just for safety */ if (!rtm->rtm_msglen) { syslog(LOG_WARNING, "<%s> rtm_msglen is 0 " - "(buf=%p lim=%p rtm=%p)", __func__, - buf, lim, rtm); + "(buf=%p lim=%p rtm=%p)", __func__, + buf, lim, rtm); break; } - if (FILTER_MATCH(rtm->rtm_type, filter) == 0) { + if (((struct rt_msghdr *)buf)->rtm_version != RTM_VERSION) { + syslog(LOG_WARNING, + "<%s> routing message version mismatch " + "(buf=%p lim=%p rtm=%p)", __func__, + buf, lim, rtm); continue; } + if (FILTER_MATCH(rtm->rtm_type, filter) == 0) + continue; + switch (rtm->rtm_type) { case RTM_GET: case RTM_ADD: case RTM_DELETE: /* address related checks */ sa = (struct sockaddr *)(rtm + 1); get_rtaddrs(rtm->rtm_addrs, sa, rti_info); if ((dst = rti_info[RTAX_DST]) == NULL || dst->sa_family != AF_INET6) continue; if (IN6_IS_ADDR_LINKLOCAL(&SIN6(dst)->sin6_addr) || IN6_IS_ADDR_MULTICAST(&SIN6(dst)->sin6_addr)) continue; if ((gw = rti_info[RTAX_GATEWAY]) == NULL || gw->sa_family != AF_LINK) continue; if (ifindex && SDL(gw)->sdl_index != ifindex) continue; if (rti_info[RTAX_NETMASK] == NULL) continue; /* found */ *lenp = rtm->rtm_msglen; return (char *)rtm; /* NOTREACHED */ case RTM_NEWADDR: case RTM_DELADDR: ifam = (struct ifa_msghdr *)rtm; /* address related checks */ sa = (struct sockaddr *)(ifam + 1); get_rtaddrs(ifam->ifam_addrs, sa, rti_info); if ((ifa = rti_info[RTAX_IFA]) == NULL || (ifa->sa_family != AF_INET && ifa->sa_family != AF_INET6)) continue; if (ifa->sa_family == AF_INET6 && (IN6_IS_ADDR_LINKLOCAL(&SIN6(ifa)->sin6_addr) || IN6_IS_ADDR_MULTICAST(&SIN6(ifa)->sin6_addr))) continue; if (ifindex && ifam->ifam_index != ifindex) continue; /* found */ *lenp = ifam->ifam_msglen; return (char *)rtm; /* NOTREACHED */ case RTM_IFINFO: + case RTM_IFANNOUNCE: /* found */ *lenp = rtm->rtm_msglen; return (char *)rtm; /* NOTREACHED */ } } - return (char *)rtm; + return ((char *)rtm); } #undef FILTER_MATCH struct in6_addr * get_addr(char *buf) { struct rt_msghdr *rtm = (struct rt_msghdr *)buf; struct sockaddr *sa, *rti_info[RTAX_MAX]; sa = (struct sockaddr *)(rtm + 1); get_rtaddrs(rtm->rtm_addrs, sa, rti_info); - return(&SIN6(rti_info[RTAX_DST])->sin6_addr); + return (&SIN6(rti_info[RTAX_DST])->sin6_addr); } int get_rtm_ifindex(char *buf) { struct rt_msghdr *rtm = (struct rt_msghdr *)buf; struct sockaddr *sa, *rti_info[RTAX_MAX]; sa = (struct sockaddr *)(rtm + 1); get_rtaddrs(rtm->rtm_addrs, sa, rti_info); - return(((struct sockaddr_dl *)rti_info[RTAX_GATEWAY])->sdl_index); + return (((struct sockaddr_dl *)rti_info[RTAX_GATEWAY])->sdl_index); } int get_ifm_ifindex(char *buf) { struct if_msghdr *ifm = (struct if_msghdr *)buf; return ((int)ifm->ifm_index); } int get_ifam_ifindex(char *buf) { struct ifa_msghdr *ifam = (struct ifa_msghdr *)buf; return ((int)ifam->ifam_index); } int get_ifm_flags(char *buf) { struct if_msghdr *ifm = (struct if_msghdr *)buf; return (ifm->ifm_flags); } int get_prefixlen(char *buf) { struct rt_msghdr *rtm = (struct rt_msghdr *)buf; struct sockaddr *sa, *rti_info[RTAX_MAX]; u_char *p, *lim; - + sa = (struct sockaddr *)(rtm + 1); get_rtaddrs(rtm->rtm_addrs, sa, rti_info); sa = rti_info[RTAX_NETMASK]; p = (u_char *)(&SIN6(sa)->sin6_addr); lim = (u_char *)sa + sa->sa_len; return prefixlen(p, lim); } int prefixlen(u_char *p, u_char *lim) { int masklen; for (masklen = 0; p < lim; p++) { switch (*p) { case 0xff: masklen += 8; break; case 0xfe: masklen += 7; break; case 0xfc: masklen += 6; break; case 0xf8: masklen += 5; break; case 0xf0: masklen += 4; break; case 0xe0: masklen += 3; break; case 0xc0: masklen += 2; break; case 0x80: masklen += 1; break; case 0x00: break; default: - return(-1); + return (-1); } } - return(masklen); + return (masklen); } int rtmsg_type(char *buf) { struct rt_msghdr *rtm = (struct rt_msghdr *)buf; - return(rtm->rtm_type); + return (rtm->rtm_type); } int rtmsg_len(char *buf) { struct rt_msghdr *rtm = (struct rt_msghdr *)buf; - return(rtm->rtm_msglen); + return (rtm->rtm_msglen); } int ifmsg_len(char *buf) { struct if_msghdr *ifm = (struct if_msghdr *)buf; - return(ifm->ifm_msglen); + return (ifm->ifm_msglen); } /* * alloc buffer and get if_msghdrs block from kernel, * and put them into the buffer */ static void get_iflist(char **buf, size_t *size) { int mib[6]; mib[0] = CTL_NET; mib[1] = PF_ROUTE; mib[2] = 0; mib[3] = AF_INET6; mib[4] = NET_RT_IFLIST; mib[5] = 0; if (sysctl(mib, 6, NULL, size, NULL, 0) < 0) { syslog(LOG_ERR, "<%s> sysctl: iflist size get failed", - __func__); + __func__); exit(1); } if ((*buf = malloc(*size)) == NULL) { syslog(LOG_ERR, "<%s> malloc failed", __func__); exit(1); } if (sysctl(mib, 6, *buf, size, NULL, 0) < 0) { syslog(LOG_ERR, "<%s> sysctl: iflist get failed", - __func__); + __func__); exit(1); } return; } /* * alloc buffer and parse if_msghdrs block passed as arg, * and init the buffer as list of pointers ot each of the if_msghdr. */ static void parse_iflist(struct if_msghdr ***ifmlist_p, char *buf, size_t bufsize) { int iflentry_size, malloc_size; struct if_msghdr *ifm; struct ifa_msghdr *ifam; char *lim; /* * Estimate least size of an iflist entry, to be obtained from kernel. * Should add sizeof(sockaddr) ?? */ iflentry_size = sizeof(struct if_msghdr); /* roughly estimate max list size of pointers to each if_msghdr */ malloc_size = (bufsize/iflentry_size) * sizeof(size_t); if ((*ifmlist_p = (struct if_msghdr **)malloc(malloc_size)) == NULL) { syslog(LOG_ERR, "<%s> malloc failed", __func__); exit(1); } lim = buf + bufsize; for (ifm = (struct if_msghdr *)buf; ifm < (struct if_msghdr *)lim;) { if (ifm->ifm_msglen == 0) { syslog(LOG_WARNING, "<%s> ifm_msglen is 0 " - "(buf=%p lim=%p ifm=%p)", __func__, - buf, lim, ifm); + "(buf=%p lim=%p ifm=%p)", __func__, + buf, lim, ifm); return; } if (ifm->ifm_type == RTM_IFINFO) { (*ifmlist_p)[ifm->ifm_index] = ifm; } else { syslog(LOG_ERR, "out of sync parsing NET_RT_IFLIST\n" - "expected %d, got %d\n msglen = %d\n" - "buf:%p, ifm:%p, lim:%p\n", - RTM_IFINFO, ifm->ifm_type, ifm->ifm_msglen, - buf, ifm, lim); + "expected %d, got %d\n msglen = %d\n" + "buf:%p, ifm:%p, lim:%p\n", + RTM_IFINFO, ifm->ifm_type, ifm->ifm_msglen, + buf, ifm, lim); exit (1); } for (ifam = (struct ifa_msghdr *) ((char *)ifm + ifm->ifm_msglen); ifam < (struct ifa_msghdr *)lim; ifam = (struct ifa_msghdr *) ((char *)ifam + ifam->ifam_msglen)) { /* just for safety */ if (!ifam->ifam_msglen) { syslog(LOG_WARNING, "<%s> ifa_msglen is 0 " - "(buf=%p lim=%p ifam=%p)", __func__, - buf, lim, ifam); + "(buf=%p lim=%p ifam=%p)", __func__, + buf, lim, ifam); return; } if (ifam->ifam_type != RTM_NEWADDR) break; } ifm = (struct if_msghdr *)ifam; } } void -init_iflist() +init_iflist(void) { + syslog(LOG_DEBUG, + "<%s> generate iflist.", __func__); + if (ifblock) { free(ifblock); ifblock_size = 0; } if (iflist) free(iflist); /* get iflist block from kernel */ get_iflist(&ifblock, &ifblock_size); /* make list of pointers to each if_msghdr */ parse_iflist(&iflist, ifblock, ifblock_size); } Index: projects/largeSMP/usr.sbin/rtadvd/if.h =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/if.h (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/if.h (revision 222783) @@ -1,58 +1,58 @@ /* $FreeBSD$ */ /* $KAME: if.h,v 1.10 2003/02/24 11:29:10 ono Exp $ */ /* * Copyright (C) 1995, 1996, 1997, 1998, and 1999 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. */ #define RTADV_TYPE2BITMASK(type) (0x1 << type) extern struct if_msghdr **iflist; extern size_t ifblock_size; extern char *ifblock; struct nd_opt_hdr; struct sockaddr_dl *if_nametosdl(char *); int if_getmtu(char *); int if_getflags(int, int); int lladdropt_length(struct sockaddr_dl *); void lladdropt_fill(struct sockaddr_dl *, struct nd_opt_hdr *); int rtbuf_len(void); char *get_next_msg(char *, char *, int, size_t *, int); struct in6_addr *get_addr(char *); int get_rtm_ifindex(char *); int get_ifm_ifindex(char *); int get_ifam_ifindex(char *); int get_ifm_flags(char *); int get_prefixlen(char *); int prefixlen(u_char *, u_char *); int rtmsg_type(char *); int ifmsg_type(char *); int rtmsg_len(char *); int ifmsg_len(char *); void init_iflist(void); Index: projects/largeSMP/usr.sbin/rtadvd/pathnames.h =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/pathnames.h (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/pathnames.h (revision 222783) @@ -1,4 +1,7 @@ /* $KAME: pathnames.h,v 1.2 2000/05/16 13:34:13 itojun Exp $ */ /* $FreeBSD$ */ -#define _PATH_RTADVDCONF "/etc/rtadvd.conf" +#define _PATH_RTADVDCONF "/etc/rtadvd.conf" +#define _PATH_RTADVDDUMP "/var/run/rtadvd.dump" +#define _PATH_RTADVDPID "/var/run/rtadvd.pid" + Index: projects/largeSMP/usr.sbin/rtadvd/rrenum.c =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/rrenum.c (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/rrenum.c (revision 222783) @@ -1,486 +1,490 @@ /* $FreeBSD$ */ /* $KAME: rrenum.c,v 1.12 2002/06/10 19:59:47 itojun Exp $ */ /* * 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. */ #include #include #include #include #include #include #include #include #include #include #include #include #include +#include #include #include #include #include "rtadvd.h" #include "rrenum.h" #include "if.h" #define RR_ISSET_SEGNUM(segnum_bits, segnum) \ ((((segnum_bits)[(segnum) >> 5]) & (1 << ((segnum) & 31))) != 0) #define RR_SET_SEGNUM(segnum_bits, segnum) \ (((segnum_bits)[(segnum) >> 5]) |= (1 << ((segnum) & 31))) struct rr_operation { u_long rro_seqnum; u_long rro_segnum_bits[8]; }; static struct rr_operation rro; static int rr_rcvifindex; static int rrcmd2pco[RPM_PCO_MAX] = { 0, SIOCAIFPREFIX_IN6, SIOCCIFPREFIX_IN6, SIOCSGIFPREFIX_IN6 }; static int s = -1; /* * Check validity of a Prefix Control Operation(PCO). - * Return 0 on success, 1 on failure. + * return 0 on success, 1 on failure. */ static int rr_pco_check(int len, struct rr_pco_match *rpm) { struct rr_pco_use *rpu, *rpulim; int checklen; /* rpm->rpm_len must be (4N * 3) as router-renum-05.txt */ if ((rpm->rpm_len - 3) < 0 || /* must be at least 3 */ (rpm->rpm_len - 3) & 0x3) { /* must be multiple of 4 */ syslog(LOG_WARNING, "<%s> rpm_len %d is not 4N * 3", - __func__, rpm->rpm_len); - return 1; + __func__, rpm->rpm_len); + return (1); } /* rpm->rpm_code must be valid value */ switch (rpm->rpm_code) { case RPM_PCO_ADD: case RPM_PCO_CHANGE: case RPM_PCO_SETGLOBAL: break; default: syslog(LOG_WARNING, "<%s> unknown rpm_code %d", __func__, - rpm->rpm_code); - return 1; + rpm->rpm_code); + return (1); } /* rpm->rpm_matchlen must be 0 to 128 inclusive */ if (rpm->rpm_matchlen > 128) { syslog(LOG_WARNING, "<%s> rpm_matchlen %d is over 128", - __func__, rpm->rpm_matchlen); - return 1; + __func__, rpm->rpm_matchlen); + return (1); } /* * rpu->rpu_uselen, rpu->rpu_keeplen, and sum of them must be * between 0 and 128 inclusive */ for (rpu = (struct rr_pco_use *)(rpm + 1), rpulim = (struct rr_pco_use *)((char *)rpm + len); rpu < rpulim; rpu += 1) { checklen = rpu->rpu_uselen; checklen += rpu->rpu_keeplen; /* * omit these check, because either of rpu_uselen * and rpu_keeplen is unsigned char * (128 > rpu_uselen > 0) * (128 > rpu_keeplen > 0) * (rpu_uselen + rpu_keeplen > 0) */ if (checklen > 128) { syslog(LOG_WARNING, "<%s> sum of rpu_uselen %d and" - " rpu_keeplen %d is %d(over 128)", - __func__, rpu->rpu_uselen, - rpu->rpu_keeplen, - rpu->rpu_uselen + rpu->rpu_keeplen); - return 1; + " rpu_keeplen %d is %d(over 128)", + __func__, rpu->rpu_uselen, rpu->rpu_keeplen, + rpu->rpu_uselen + rpu->rpu_keeplen); + return (1); } } - return 0; + return (0); } static void do_use_prefix(int len, struct rr_pco_match *rpm, - struct in6_rrenumreq *irr, int ifindex) + struct in6_rrenumreq *irr, int ifindex) { struct rr_pco_use *rpu, *rpulim; struct rainfo *rai; - struct prefix *pp; + struct prefix *pfx; rpu = (struct rr_pco_use *)(rpm + 1); rpulim = (struct rr_pco_use *)((char *)rpm + len); if (rpu == rpulim) { /* no use prefix */ if (rpm->rpm_code == RPM_PCO_ADD) return; irr->irr_u_uselen = 0; irr->irr_u_keeplen = 0; irr->irr_raf_mask_onlink = 0; irr->irr_raf_mask_auto = 0; irr->irr_vltime = 0; irr->irr_pltime = 0; memset(&irr->irr_flags, 0, sizeof(irr->irr_flags)); irr->irr_useprefix.sin6_len = 0; /* let it mean, no addition */ irr->irr_useprefix.sin6_family = 0; irr->irr_useprefix.sin6_addr = in6addr_any; if (ioctl(s, rrcmd2pco[rpm->rpm_code], (caddr_t)irr) < 0 && errno != EADDRNOTAVAIL) syslog(LOG_ERR, "<%s> ioctl: %s", __func__, - strerror(errno)); + strerror(errno)); return; } for (rpu = (struct rr_pco_use *)(rpm + 1), rpulim = (struct rr_pco_use *)((char *)rpm + len); rpu < rpulim; rpu += 1) { /* init in6_rrenumreq fields */ irr->irr_u_uselen = rpu->rpu_uselen; irr->irr_u_keeplen = rpu->rpu_keeplen; irr->irr_raf_mask_onlink = - !!(rpu->rpu_ramask & ICMP6_RR_PCOUSE_RAFLAGS_ONLINK); + !!(rpu->rpu_ramask & ICMP6_RR_PCOUSE_RAFLAGS_ONLINK); irr->irr_raf_mask_auto = - !!(rpu->rpu_ramask & ICMP6_RR_PCOUSE_RAFLAGS_AUTO); + !!(rpu->rpu_ramask & ICMP6_RR_PCOUSE_RAFLAGS_AUTO); irr->irr_vltime = ntohl(rpu->rpu_vltime); irr->irr_pltime = ntohl(rpu->rpu_pltime); irr->irr_raf_onlink = - (rpu->rpu_raflags & ICMP6_RR_PCOUSE_RAFLAGS_ONLINK) == 0 ? 0 : 1; + (rpu->rpu_raflags & ICMP6_RR_PCOUSE_RAFLAGS_ONLINK) == 0 ? + 0 : 1; irr->irr_raf_auto = - (rpu->rpu_raflags & ICMP6_RR_PCOUSE_RAFLAGS_AUTO) == 0 ? 0 : 1; + (rpu->rpu_raflags & ICMP6_RR_PCOUSE_RAFLAGS_AUTO) == 0 ? + 0 : 1; irr->irr_rrf_decrvalid = - (rpu->rpu_flags & ICMP6_RR_PCOUSE_FLAGS_DECRVLTIME) == 0 ? 0 : 1; + (rpu->rpu_flags & ICMP6_RR_PCOUSE_FLAGS_DECRVLTIME) == 0 ? + 0 : 1; irr->irr_rrf_decrprefd = - (rpu->rpu_flags & ICMP6_RR_PCOUSE_FLAGS_DECRPLTIME) == 0 ? 0 : 1; + (rpu->rpu_flags & ICMP6_RR_PCOUSE_FLAGS_DECRPLTIME) == 0 ? + 0 : 1; irr->irr_useprefix.sin6_len = sizeof(irr->irr_useprefix); irr->irr_useprefix.sin6_family = AF_INET6; irr->irr_useprefix.sin6_addr = rpu->rpu_prefix; if (ioctl(s, rrcmd2pco[rpm->rpm_code], (caddr_t)irr) < 0 && errno != EADDRNOTAVAIL) syslog(LOG_ERR, "<%s> ioctl: %s", __func__, - strerror(errno)); + strerror(errno)); /* very adhoc: should be rewritten */ if (rpm->rpm_code == RPM_PCO_CHANGE && IN6_ARE_ADDR_EQUAL(&rpm->rpm_prefix, &rpu->rpu_prefix) && rpm->rpm_matchlen == rpu->rpu_uselen && rpu->rpu_uselen == rpu->rpu_keeplen) { if ((rai = if_indextorainfo(ifindex)) == NULL) continue; /* non-advertising IF */ - for (pp = rai->prefix.next; pp != &rai->prefix; - pp = pp->next) { + TAILQ_FOREACH(pfx, &rai->rai_prefix, pfx_next) { struct timeval now; - if (prefix_match(&pp->prefix, pp->prefixlen, - &rpm->rpm_prefix, - rpm->rpm_matchlen)) { + if (prefix_match(&pfx->pfx_prefix, + pfx->pfx_prefixlen, &rpm->rpm_prefix, + rpm->rpm_matchlen)) { /* change parameters */ - pp->validlifetime = ntohl(rpu->rpu_vltime); - pp->preflifetime = ntohl(rpu->rpu_pltime); + pfx->pfx_validlifetime = + ntohl(rpu->rpu_vltime); + pfx->pfx_preflifetime = + ntohl(rpu->rpu_pltime); if (irr->irr_rrf_decrvalid) { gettimeofday(&now, 0); - pp->vltimeexpire = - now.tv_sec + pp->validlifetime; + pfx->pfx_vltimeexpire = + now.tv_sec + + pfx->pfx_validlifetime; } else - pp->vltimeexpire = 0; + pfx->pfx_vltimeexpire = 0; if (irr->irr_rrf_decrprefd) { gettimeofday(&now, 0); - pp->pltimeexpire = - now.tv_sec + pp->preflifetime; + pfx->pfx_pltimeexpire = + now.tv_sec + + pfx->pfx_preflifetime; } else - pp->pltimeexpire = 0; + pfx->pfx_pltimeexpire = 0; } } } } } /* * process a Prefix Control Operation(PCO). * return 0 on success, 1 on failure */ static int do_pco(struct icmp6_router_renum *rr, int len, struct rr_pco_match *rpm) { int ifindex = 0; struct in6_rrenumreq irr; if ((rr_pco_check(len, rpm) != 0)) - return 1; + return (1); if (s == -1 && (s = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) { syslog(LOG_ERR, "<%s> socket: %s", __func__, - strerror(errno)); + strerror(errno)); exit(1); } memset(&irr, 0, sizeof(irr)); irr.irr_origin = PR_ORIG_RR; irr.irr_m_len = rpm->rpm_matchlen; irr.irr_m_minlen = rpm->rpm_minlen; irr.irr_m_maxlen = rpm->rpm_maxlen; irr.irr_matchprefix.sin6_len = sizeof(irr.irr_matchprefix); irr.irr_matchprefix.sin6_family = AF_INET6; irr.irr_matchprefix.sin6_addr = rpm->rpm_prefix; while (if_indextoname(++ifindex, irr.irr_name)) { /* - * if ICMP6_RR_FLAGS_FORCEAPPLY(A flag) is 0 and IFF_UP is off, - * the interface is not applied + * if ICMP6_RR_FLAGS_FORCEAPPLY(A flag) is 0 and + * IFF_UP is off, the interface is not applied */ if ((rr->rr_flags & ICMP6_RR_FLAGS_FORCEAPPLY) == 0 && (iflist[ifindex]->ifm_flags & IFF_UP) == 0) continue; /* TODO: interface scope check */ do_use_prefix(len, rpm, &irr, ifindex); } if (errno == ENXIO) - return 0; + return (0); else if (errno) { syslog(LOG_ERR, "<%s> if_indextoname: %s", __func__, - strerror(errno)); - return 1; + strerror(errno)); + return (1); } - return 0; + return (0); } /* * call do_pco() for each Prefix Control Operations(PCOs) in a received * Router Renumbering Command packet. * return 0 on success, 1 on failure */ static int do_rr(int len, struct icmp6_router_renum *rr) { struct rr_pco_match *rpm; char *cp, *lim; lim = (char *)rr + len; cp = (char *)(rr + 1); len -= sizeof(struct icmp6_router_renum); /* get iflist block from kernel again, to get up-to-date information */ init_iflist(); while (cp < lim) { int rpmlen; rpm = (struct rr_pco_match *)cp; - if (len < sizeof(struct rr_pco_match)) { + if ((size_t)len < sizeof(struct rr_pco_match)) { tooshort: syslog(LOG_ERR, "<%s> pkt too short. left len = %d. " - "gabage at end of pkt?", __func__, len); - return 1; + "gabage at end of pkt?", __func__, len); + return (1); } rpmlen = rpm->rpm_len << 3; if (len < rpmlen) goto tooshort; if (do_pco(rr, rpmlen, rpm)) { syslog(LOG_WARNING, "<%s> invalid PCO", __func__); goto next; } next: cp += rpmlen; len -= rpmlen; } - return 0; + return (0); } /* * check validity of a router renumbering command packet * return 0 on success, 1 on failure */ static int rr_command_check(int len, struct icmp6_router_renum *rr, struct in6_addr *from, - struct in6_addr *dst) + struct in6_addr *dst) { u_char ntopbuf[INET6_ADDRSTRLEN]; /* omit rr minimal length check. hope kernel have done it. */ /* rr_command length check */ - if (len < (sizeof(struct icmp6_router_renum) + - sizeof(struct rr_pco_match))) { + if ((size_t)len < (sizeof(struct icmp6_router_renum) + + sizeof(struct rr_pco_match))) { syslog(LOG_ERR, "<%s> rr_command len %d is too short", - __func__, len); - return 1; + __func__, len); + return (1); } /* destination check. only for multicast. omit unicast check. */ if (IN6_IS_ADDR_MULTICAST(dst) && !IN6_IS_ADDR_MC_LINKLOCAL(dst) && !IN6_IS_ADDR_MC_SITELOCAL(dst)) { syslog(LOG_ERR, "<%s> dst mcast addr %s is illegal", - __func__, - inet_ntop(AF_INET6, dst, ntopbuf, INET6_ADDRSTRLEN)); - return 1; + __func__, + inet_ntop(AF_INET6, dst, ntopbuf, sizeof(ntopbuf))); + return (1); } /* seqnum and segnum check */ if (rro.rro_seqnum > rr->rr_seqnum) { syslog(LOG_WARNING, - "<%s> rcvd old seqnum %d from %s", - __func__, (u_int32_t)ntohl(rr->rr_seqnum), - inet_ntop(AF_INET6, from, ntopbuf, INET6_ADDRSTRLEN)); - return 1; + "<%s> rcvd old seqnum %d from %s", + __func__, (u_int32_t)ntohl(rr->rr_seqnum), + inet_ntop(AF_INET6, from, ntopbuf, sizeof(ntopbuf))); + return (1); } if (rro.rro_seqnum == rr->rr_seqnum && (rr->rr_flags & ICMP6_RR_FLAGS_TEST) == 0 && RR_ISSET_SEGNUM(rro.rro_segnum_bits, rr->rr_segnum)) { if ((rr->rr_flags & ICMP6_RR_FLAGS_REQRESULT) != 0) syslog(LOG_WARNING, - "<%s> rcvd duped segnum %d from %s", - __func__, rr->rr_segnum, - inet_ntop(AF_INET6, from, ntopbuf, - INET6_ADDRSTRLEN)); - return 0; + "<%s> rcvd duped segnum %d from %s", + __func__, rr->rr_segnum, inet_ntop(AF_INET6, from, + ntopbuf, sizeof(ntopbuf))); + return (0); } /* update seqnum */ if (rro.rro_seqnum != rr->rr_seqnum) { /* then must be "<" */ /* init rro_segnum_bits */ memset(rro.rro_segnum_bits, 0, - sizeof(rro.rro_segnum_bits)); + sizeof(rro.rro_segnum_bits)); } rro.rro_seqnum = rr->rr_seqnum; - return 0; + return (0); } static void rr_command_input(int len, struct icmp6_router_renum *rr, - struct in6_addr *from, struct in6_addr *dst) + struct in6_addr *from, struct in6_addr *dst) { /* rr_command validity check */ if (rr_command_check(len, rr, from, dst)) goto failed; if ((rr->rr_flags & (ICMP6_RR_FLAGS_TEST|ICMP6_RR_FLAGS_REQRESULT)) == ICMP6_RR_FLAGS_TEST) return; /* do router renumbering */ - if (do_rr(len, rr)) { + if (do_rr(len, rr)) goto failed; - } /* update segnum */ RR_SET_SEGNUM(rro.rro_segnum_bits, rr->rr_segnum); return; failed: syslog(LOG_ERR, "<%s> received RR was invalid", __func__); return; } void rr_input(int len, struct icmp6_router_renum *rr, struct in6_pktinfo *pi, - struct sockaddr_in6 *from, struct in6_addr *dst) + struct sockaddr_in6 *from, struct in6_addr *dst) { u_char ntopbuf[2][INET6_ADDRSTRLEN], ifnamebuf[IFNAMSIZ]; syslog(LOG_DEBUG, - "<%s> RR received from %s to %s on %s", - __func__, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf[0], INET6_ADDRSTRLEN), - inet_ntop(AF_INET6, &dst, ntopbuf[1], INET6_ADDRSTRLEN), - if_indextoname(pi->ipi6_ifindex, ifnamebuf)); + "<%s> RR received from %s to %s on %s", + __func__, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf[0] ,sizeof(ntopbuf[0])), + inet_ntop(AF_INET6, &dst, ntopbuf[1], sizeof(ntopbuf[1])), + if_indextoname(pi->ipi6_ifindex, ifnamebuf)); /* packet validation based on Section 4.1 of RFC2894 */ - if (len < sizeof(struct icmp6_router_renum)) { + if ((size_t)len < sizeof(struct icmp6_router_renum)) { syslog(LOG_NOTICE, - "<%s>: RR short message (size %d) from %s to %s on %s", - __func__, len, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf[0], INET6_ADDRSTRLEN), - inet_ntop(AF_INET6, &dst, ntopbuf[1], INET6_ADDRSTRLEN), - if_indextoname(pi->ipi6_ifindex, ifnamebuf)); + "<%s>: RR short message (size %d) from %s to %s on %s", + __func__, len, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf[0], + sizeof(ntopbuf[0])), + inet_ntop(AF_INET6, &dst, ntopbuf[1], sizeof(ntopbuf[1])), + if_indextoname(pi->ipi6_ifindex, ifnamebuf)); return; } /* * If the IPv6 destination address is neither an All Routers multicast * address [AARCH] nor one of the receiving router's unicast addresses, * the message MUST be discarded and SHOULD be logged to network * management. * We rely on the kernel input routine for unicast addresses, and thus * check multicast destinations only. */ - if (IN6_IS_ADDR_MULTICAST(&pi->ipi6_addr) && - !IN6_ARE_ADDR_EQUAL(&in6a_site_allrouters, &pi->ipi6_addr)) { + if (IN6_IS_ADDR_MULTICAST(&pi->ipi6_addr) && !IN6_ARE_ADDR_EQUAL( + &sin6_sitelocal_allrouters.sin6_addr, &pi->ipi6_addr)) { syslog(LOG_NOTICE, - "<%s>: RR message with invalid destination (%s) " - "from %s on %s", - __func__, - inet_ntop(AF_INET6, &dst, ntopbuf[0], INET6_ADDRSTRLEN), - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf[1], INET6_ADDRSTRLEN), - if_indextoname(pi->ipi6_ifindex, ifnamebuf)); + "<%s>: RR message with invalid destination (%s) " + "from %s on %s", + __func__, + inet_ntop(AF_INET6, &dst, ntopbuf[0], sizeof(ntopbuf[0])), + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf[1], + sizeof(ntopbuf[1])), + if_indextoname(pi->ipi6_ifindex, ifnamebuf)); return; } rr_rcvifindex = pi->ipi6_ifindex; switch (rr->rr_code) { case ICMP6_ROUTER_RENUMBERING_COMMAND: rr_command_input(len, rr, &from->sin6_addr, dst); /* TODO: send reply msg */ break; case ICMP6_ROUTER_RENUMBERING_RESULT: /* RESULT will be processed by rrenumd */ break; case ICMP6_ROUTER_RENUMBERING_SEQNUM_RESET: /* TODO: sequence number reset */ break; default: syslog(LOG_ERR, "<%s> received unknown code %d", - __func__, rr->rr_code); + __func__, rr->rr_code); break; } return; } Index: projects/largeSMP/usr.sbin/rtadvd/rrenum.h =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/rrenum.h (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/rrenum.h (revision 222783) @@ -1,34 +1,34 @@ /* $FreeBSD$ */ /* $KAME: rrenum.h,v 1.3 2001/01/21 15:37:14 itojun Exp $ */ /* * Copyright (C) 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. */ void rr_input(int, struct icmp6_router_renum *, struct in6_pktinfo *, struct sockaddr_in6 *, struct in6_addr *); Index: projects/largeSMP/usr.sbin/rtadvd/rtadvd.8 =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/rtadvd.8 (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/rtadvd.8 (revision 222783) @@ -1,209 +1,238 @@ .\" $KAME: rtadvd.8,v 1.24 2002/05/31 16:16:08 jinmei Exp $ .\" .\" 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. .\" .\" $FreeBSD$ .\" .Dd December 22, 2010 .Dt RTADVD 8 .Os .Sh NAME .Nm rtadvd .Nd router advertisement daemon .Sh SYNOPSIS .Nm -.Op Fl dDfMRs +.Op Fl dDfRs .Op Fl c Ar configfile .Op Fl F Ar dumpfile +.Op Fl M Ar ifname .Op Fl p Ar pidfile .Ar interface ... .Sh DESCRIPTION .Nm sends router advertisement packets to the specified .Ar interfaces . .Pp The program will daemonize itself on invocation. It will then send router advertisement packets periodically, as well as in response to router solicitation messages sent by end hosts. .Pp Router advertisements can be configured on a per-interface basis, as described in .Xr rtadvd.conf 5 . .Pp If there is no configuration file entry for an interface, or if the configuration file does not exist altogether, .Nm sets all the parameters to their default values. In particular, .Nm reads all the interface routes from the routing table and advertises them as on-link prefixes. .Pp .Nm also watches the routing table. If an interface direct route is added on an advertising interface and no static prefixes are specified by the configuration file, .Nm adds the corresponding prefix to its advertising list. .Pp Similarly, when an interface direct route is deleted, .Nm will start advertising the prefixes with zero valid and preferred lifetimes to help the receiving hosts switch to a new prefix when renumbering. Note, however, that the zero valid lifetime cannot invalidate the autoconfigured addresses at a receiving host immediately. According to the specification, the host will retain the address for a certain period, which will typically be two hours. The zero lifetimes rather intend to make the address deprecated, indicating that a new non-deprecated address should be used as the source address of a new connection. This behavior will last for two hours. Then .Nm will completely remove the prefix from the advertising list, and succeeding advertisements will not contain the prefix information. .Pp Moreover, if the status of an advertising interface changes, .Nm will start or stop sending router advertisements according to the latest status. .Pp The .Fl s option may be used to disable this behavior; .Nm will not watch the routing table and the whole functionality described above will be suppressed. .Pp Basically, hosts MUST NOT send Router Advertisement messages at any -time (RFC 2461, Section 6.2.3). +time (RFC 4861, Section 6.2.3). However, it would sometimes be useful to allow hosts to advertise some parameters such as prefix information and link MTU. Thus, .Nm can be invoked if router lifetime is explicitly set zero on every advertising interface. .Pp The command line options are: .Bl -tag -width indent .\" .It Fl c Specify an alternate location, .Ar configfile , for the configuration file. By default, .Pa /etc/rtadvd.conf is used. .It Fl d Print debugging information. .It Fl D Even more debugging information is printed. .It Fl f Foreground mode (useful when debugging). Log messages will be dumped to stderr when this option is specified. .It Fl F Specify an alternative file in which to dump internal states when .Nm receives signal .Dv SIGUSR1 . The default is .Pa /var/run/rtadvd.dump . .It Fl M Specify an interface to join the all-routers site-local multicast group. By default, .Nm tries to join the first advertising interface appearing on the command line. This option has meaning only with the .Fl R option, which enables routing renumbering protocol support. .It Fl p Specify an alternative file in which to store the process ID. The default is .Pa /var/run/rtadvd.pid. .It Fl R Accept router renumbering requests. If you enable it, certain IPsec setup is suggested for security reasons. This option is currently disabled, and is ignored by .Nm with a warning message. .It Fl s Do not add or delete prefixes dynamically. Only statically configured prefixes, if any, will be advertised. .El .Pp Upon receipt of signal .Dv SIGUSR1 , .Nm will dump the current internal state into .Pa /var/run/rtadvd.dump or the file specified with option .Fl F . .Pp Use .Dv SIGTERM to kill .Nm gracefully. In this case, .Nm will transmit router advertisement with router lifetime 0 to all the interfaces -.Pq in accordance with RFC2461 6.2.5 . +.Pq in accordance with RFC 4861 6.2.5 . .Sh FILES .Bl -tag -width Pa -compact .It Pa /etc/rtadvd.conf The default configuration file. .It Pa /var/run/rtadvd.pid The default process ID file. .It Pa /var/run/rtadvd.dump The default file in which .Nm dumps its internal state. .El .Sh EXIT STATUS .Ex -std .Sh SEE ALSO .Xr rtadvd.conf 5 , .Xr rtsol 8 +.Rs +.%A Thomas Narten +.%A Erik Nordmark +.%A W. A. Simpson +.%A Hesham Soliman +.%T Neighbor Discovery for IP version 6 (IPv6) +.%R RFC 4861 +.Re +.Rs +.%A Thomas Narten +.%A Erik Nordmark +.%A W. A. Simpson +.%T Neighbor Discovery for IP version 6 (IPv6) +.%R RFC 2461 (obsoleted by RFC 4861) +.Re +.Rs +.%A Richard Draves +.%T Default Router Preferences and More-Specific Routes +.%R draft-ietf-ipngwg-router-selection-xx.txt +.Re +.Rs +.%A J. Jeong +.%A S. Park +.%A L. Beloeil +.%A S. Madanapalli +.%T IPv6 Router Advertisement Options for DNS Configuration +.%R RFC 6106 +.Re .Sh HISTORY The .Nm command first appeared in the WIDE Hydrangea IPv6 protocol stack kit. .Sh BUGS There used to be some text that recommended users not to let .Nm advertise Router Advertisement messages on an upstream link to avoid undesirable .Xr icmp6 4 redirect messages. However, based on the later discussion in the IETF ipng working group, all routers should rather advertise the messages regardless of the network topology, in order to ensure reachability. Index: projects/largeSMP/usr.sbin/rtadvd/rtadvd.c =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/rtadvd.c (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/rtadvd.c (revision 222783) @@ -1,1702 +1,1872 @@ /* $FreeBSD$ */ /* $KAME: rtadvd.c,v 1.82 2003/08/05 12:34:23 itojun Exp $ */ /* * 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. */ #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 HAVE_POLL_H #include #endif #include "rtadvd.h" #include "rrenum.h" #include "advcap.h" #include "timer.h" #include "if.h" #include "config.h" #include "dump.h" +#include "pathnames.h" struct msghdr rcvmhdr; static u_char *rcvcmsgbuf; static size_t rcvcmsgbuflen; static u_char *sndcmsgbuf = NULL; static size_t sndcmsgbuflen; volatile sig_atomic_t do_dump; volatile sig_atomic_t do_die; struct msghdr sndmhdr; struct iovec rcviov[2]; struct iovec sndiov[2]; struct sockaddr_in6 rcvfrom; -struct sockaddr_in6 sin6_allnodes = {sizeof(sin6_allnodes), AF_INET6}; -struct in6_addr in6a_site_allrouters; -static char *dumpfilename = "/var/run/rtadvd.dump"; -static char *pidfilename = "/var/run/rtadvd.pid"; +static const char *dumpfilename = _PATH_RTADVDDUMP; +static const char *pidfilename = _PATH_RTADVDPID; +const char *conffile = _PATH_RTADVDCONF; static struct pidfh *pfh; static char *mcastif; int sock; int rtsock = -1; int accept_rr = 0; int dflag = 0, sflag = 0; +static int ifl_len; +static char **ifl_names; -u_char *conffile = NULL; +struct railist_head_t railist = + TAILQ_HEAD_INITIALIZER(railist); -struct rainfo *ralist = NULL; struct nd_optlist { - struct nd_optlist *next; - struct nd_opt_hdr *opt; + TAILQ_ENTRY(nd_optlist) nol_next; + struct nd_opt_hdr *nol_opt; }; -union nd_opts { - struct nd_opt_hdr *nd_opt_array[9]; +union nd_opt { + struct nd_opt_hdr *opt_array[9]; struct { struct nd_opt_hdr *zero; struct nd_opt_hdr *src_lladdr; struct nd_opt_hdr *tgt_lladdr; struct nd_opt_prefix_info *pi; struct nd_opt_rd_hdr *rh; struct nd_opt_mtu *mtu; - struct nd_optlist *list; + TAILQ_HEAD(, nd_optlist) opt_list; } 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 -#define nd_opts_rh nd_opt_each.rh -#define nd_opts_mtu nd_opt_each.mtu -#define nd_opts_list nd_opt_each.list +#define opt_src_lladdr nd_opt_each.src_lladdr +#define opt_tgt_lladdr nd_opt_each.tgt_lladdr +#define opt_pi nd_opt_each.pi +#define opt_rh nd_opt_each.rh +#define opt_mtu nd_opt_each.mtu +#define opt_list nd_opt_each.opt_list -#define NDOPT_FLAG_SRCLINKADDR 0x1 -#define NDOPT_FLAG_TGTLINKADDR 0x2 -#define NDOPT_FLAG_PREFIXINFO 0x4 -#define NDOPT_FLAG_RDHDR 0x8 -#define NDOPT_FLAG_MTU 0x10 +#define NDOPT_FLAG_SRCLINKADDR (1 << 0) +#define NDOPT_FLAG_TGTLINKADDR (1 << 1) +#define NDOPT_FLAG_PREFIXINFO (1 << 2) +#define NDOPT_FLAG_RDHDR (1 << 3) +#define NDOPT_FLAG_MTU (1 << 4) +#define NDOPT_FLAG_RDNSS (1 << 5) +#define NDOPT_FLAG_DNSSL (1 << 6) u_int32_t ndopt_flags[] = { - 0, NDOPT_FLAG_SRCLINKADDR, NDOPT_FLAG_TGTLINKADDR, - NDOPT_FLAG_PREFIXINFO, NDOPT_FLAG_RDHDR, NDOPT_FLAG_MTU, + [ND_OPT_SOURCE_LINKADDR] = NDOPT_FLAG_SRCLINKADDR, + [ND_OPT_TARGET_LINKADDR] = NDOPT_FLAG_TGTLINKADDR, + [ND_OPT_PREFIX_INFORMATION] = NDOPT_FLAG_PREFIXINFO, + [ND_OPT_REDIRECTED_HEADER] = NDOPT_FLAG_RDHDR, + [ND_OPT_MTU] = NDOPT_FLAG_MTU, + [ND_OPT_RDNSS] = NDOPT_FLAG_RDNSS, + [ND_OPT_DNSSL] = NDOPT_FLAG_DNSSL, }; -int main(int, char *[]); -static void set_die(int); -static void die(void); -static void sock_open(void); -static void rtsock_open(void); -static void rtadvd_input(void); -static void rs_input(int, struct nd_router_solicit *, - struct in6_pktinfo *, struct sockaddr_in6 *); -static void ra_input(int, struct nd_router_advert *, - struct in6_pktinfo *, struct sockaddr_in6 *); -static int prefix_check(struct nd_opt_prefix_info *, struct rainfo *, - struct sockaddr_in6 *); -static int nd6_options(struct nd_opt_hdr *, int, - union nd_opts *, u_int32_t); -static void free_ndopts(union nd_opts *); -static void ra_output(struct rainfo *); -static void rtmsg_input(void); -static void rtadvd_set_dump_file(int); -static void set_short_delay(struct rainfo *); +struct sockaddr_in6 sin6_linklocal_allnodes = { + .sin6_len = sizeof(sin6_linklocal_allnodes), + .sin6_family = AF_INET6, + .sin6_addr = IN6ADDR_LINKLOCAL_ALLNODES_INIT, +}; +struct sockaddr_in6 sin6_linklocal_allrouters = { + .sin6_len = sizeof(sin6_linklocal_allrouters), + .sin6_family = AF_INET6, + .sin6_addr = IN6ADDR_LINKLOCAL_ALLROUTERS_INIT, +}; + +struct sockaddr_in6 sin6_sitelocal_allrouters = { + .sin6_len = sizeof(sin6_sitelocal_allrouters), + .sin6_family = AF_INET6, + .sin6_addr = IN6ADDR_SITELOCAL_ALLROUTERS_INIT, +}; + +static void set_die(int); +static void die(void); +static void sock_open(void); +static void rtsock_open(void); +static void rtadvd_input(void); +static void rs_input(int, struct nd_router_solicit *, + struct in6_pktinfo *, struct sockaddr_in6 *); +static void ra_input(int, struct nd_router_advert *, + struct in6_pktinfo *, struct sockaddr_in6 *); +static int prefix_check(struct nd_opt_prefix_info *, struct rainfo *, + struct sockaddr_in6 *); +static int nd6_options(struct nd_opt_hdr *, int, + union nd_opt *, u_int32_t); +static void free_ndopts(union nd_opt *); +static void ra_output(struct rainfo *); +static void rtmsg_input(void); +static void rtadvd_set_dump_file(int); +static void set_short_delay(struct rainfo *); +static int ifl_lookup(char *, char **, int); +static int check_accept_rtadv(int); +static int getinet6sysctl(int); + int -main(argc, argv) - int argc; - char *argv[]; +main(int argc, char *argv[]) { #ifdef HAVE_POLL_H struct pollfd set[2]; #else fd_set *fdsetp, *selectfdp; int fdmasks; int maxfd = 0; #endif struct timeval *timeout; int i, ch; int fflag = 0, logopt; pid_t pid, otherpid; + int error; /* get command line options and arguments */ - while ((ch = getopt(argc, argv, "c:dDF:fMp:Rs")) != -1) { + while ((ch = getopt(argc, argv, "c:dDfF:M:p:Rs")) != -1) { switch (ch) { case 'c': conffile = optarg; break; case 'd': - dflag = 1; + dflag++; break; case 'D': - dflag = 2; + dflag += 2; break; case 'f': fflag = 1; break; case 'M': mcastif = optarg; break; case 'R': fprintf(stderr, "rtadvd: " "the -R option is currently ignored.\n"); /* accept_rr = 1; */ /* run anyway... */ break; case 's': sflag = 1; break; case 'p': pidfilename = optarg; break; case 'F': dumpfilename = optarg; break; } } argc -= optind; argv += optind; if (argc == 0) { fprintf(stderr, - "usage: rtadvd [-dDfMRs] [-c conffile] " - "[-F dumpfile] [-p pidfile] interfaces...\n"); + "usage: rtadvd [-dDfRs] [-c conffile] " + "[-F dumpfile] [-M ifname] " + "[-p pidfile] interfaces...\n"); exit(1); } logopt = LOG_NDELAY | LOG_PID; if (fflag) logopt |= LOG_PERROR; openlog("rtadvd", logopt, LOG_DAEMON); /* set log level */ - if (dflag == 0) - (void)setlogmask(LOG_UPTO(LOG_ERR)); - if (dflag == 1) + if (dflag > 1) + (void)setlogmask(LOG_UPTO(LOG_DEBUG)); + else if (dflag > 0) (void)setlogmask(LOG_UPTO(LOG_INFO)); + else + (void)setlogmask(LOG_UPTO(LOG_ERR)); /* timer initialization */ rtadvd_timer_init(); #ifndef HAVE_ARC4RANDOM /* random value initialization */ #ifdef __FreeBSD__ srandomdev(); #else srandom((u_long)time(NULL)); #endif #endif - /* get iflist block from kernel */ init_iflist(); + ifl_names = argv; + ifl_len = argc; - while (argc--) - getconfig(*argv++); + for (i = 0; i < ifl_len; i++) { + int idx; - if (inet_pton(AF_INET6, ALLNODES, &sin6_allnodes.sin6_addr) != 1) { - fprintf(stderr, "fatal: inet_pton failed\n"); - exit(1); + idx = if_nametoindex(ifl_names[i]); + if (idx == 0) { + syslog(LOG_INFO, + "<%s> interface %s not found." + "Ignored at this moment.", __func__, ifl_names[i]); + continue; + } + error = getconfig(idx); + if (error) + syslog(LOG_INFO, + "<%s> invalid configuration for %s." + "Ignored at this moment.", __func__, ifl_names[i]); } pfh = pidfile_open(pidfilename, 0600, &otherpid); if (pfh == NULL) { if (errno == EEXIST) errx(1, "%s already running, pid: %d", getprogname(), otherpid); syslog(LOG_ERR, "<%s> failed to open the pid log file, run anyway.", __func__); } if (!fflag) daemon(1, 0); sock_open(); /* record the current PID */ pid = getpid(); pidfile_write(pfh); #ifdef HAVE_POLL_H set[0].fd = sock; set[0].events = POLLIN; if (sflag == 0) { rtsock_open(); set[1].fd = rtsock; set[1].events = POLLIN; } else set[1].fd = -1; #else maxfd = sock; if (sflag == 0) { rtsock_open(); if (rtsock > sock) maxfd = rtsock; } else rtsock = -1; fdmasks = howmany(maxfd + 1, NFDBITS) * sizeof(fd_mask); if ((fdsetp = malloc(fdmasks)) == NULL) { err(1, "malloc"); /*NOTREACHED*/ } if ((selectfdp = malloc(fdmasks)) == NULL) { err(1, "malloc"); /*NOTREACHED*/ } memset(fdsetp, 0, fdmasks); FD_SET(sock, fdsetp); if (rtsock >= 0) FD_SET(rtsock, fdsetp); #endif - signal(SIGTERM, set_die); signal(SIGUSR1, rtadvd_set_dump_file); while (1) { #ifndef HAVE_POLL_H memcpy(selectfdp, fdsetp, fdmasks); /* reinitialize */ #endif - if (do_dump) { /* SIGUSR1 */ do_dump = 0; rtadvd_dump_file(dumpfilename); } if (do_die) { die(); /*NOTREACHED*/ } /* timer expiration check and reset the timer */ timeout = rtadvd_check_timer(); if (timeout != NULL) { syslog(LOG_DEBUG, "<%s> set timer to %ld:%ld. waiting for " "inputs or timeout", __func__, (long int)timeout->tv_sec, (long int)timeout->tv_usec); } else { syslog(LOG_DEBUG, "<%s> there's no timer. waiting for inputs", __func__); } - #ifdef HAVE_POLL_H if ((i = poll(set, 2, timeout ? (timeout->tv_sec * 1000 + timeout->tv_usec / 1000) : INFTIM)) < 0) #else if ((i = select(maxfd + 1, selectfdp, NULL, NULL, timeout)) < 0) #endif { /* EINTR would occur upon SIGUSR1 for status dump */ if (errno != EINTR) syslog(LOG_ERR, "<%s> select: %s", __func__, strerror(errno)); continue; } if (i == 0) /* timeout */ continue; #ifdef HAVE_POLL_H if (rtsock != -1 && set[1].revents & POLLIN) #else if (rtsock != -1 && FD_ISSET(rtsock, selectfdp)) #endif rtmsg_input(); #ifdef HAVE_POLL_H if (set[0].revents & POLLIN) #else if (FD_ISSET(sock, selectfdp)) #endif rtadvd_input(); } exit(0); /* NOTREACHED */ } +static int +ifl_lookup(char *ifn, char **names, int len) +{ + while (len--) + if (strncmp(names[len], ifn, IFNAMSIZ) == 0) + return (0); + return (-1); +} + static void -rtadvd_set_dump_file(sig) - int sig; +rtadvd_set_dump_file(int sig __unused) { + do_dump = 1; } static void -set_die(sig) - int sig; +set_die(int sig __unused) { + do_die = 1; } static void -die() +die(void) { - struct rainfo *ra; + struct rainfo *rai; + struct rdnss *rdn; + struct dnssl *dns; int i; const int retrans = MAX_FINAL_RTR_ADVERTISEMENTS; - if (dflag > 1) { - syslog(LOG_DEBUG, "<%s> cease to be an advertising router\n", - __func__); - } + syslog(LOG_DEBUG, "<%s> cease to be an advertising router\n", + __func__); - for (ra = ralist; ra; ra = ra->next) { - ra->lifetime = 0; - make_packet(ra); + TAILQ_FOREACH(rai, &railist, rai_next) { + rai->rai_lifetime = 0; + TAILQ_FOREACH(rdn, &rai->rai_rdnss, rd_next) + rdn->rd_ltime = 0; + TAILQ_FOREACH(dns, &rai->rai_dnssl, dn_next) + dns->dn_ltime = 0; + make_packet(rai); } for (i = 0; i < retrans; i++) { - for (ra = ralist; ra; ra = ra->next) - ra_output(ra); + TAILQ_FOREACH(rai, &railist, rai_next) + ra_output(rai); sleep(MIN_DELAY_BETWEEN_RAS); } pidfile_remove(pfh); + exit(0); - /*NOTREACHED*/ } static void -rtmsg_input() +rtmsg_input(void) { int n, type, ifindex = 0, plen; size_t len; char msg[2048], *next, *lim; - u_char ifname[IF_NAMESIZE]; - struct prefix *prefix; + u_char ifname[IFNAMSIZ]; + struct if_announcemsghdr *ifan; + struct prefix *pfx; struct rainfo *rai; struct in6_addr *addr; char addrbuf[INET6_ADDRSTRLEN]; int prefixchange = 0; + int error; n = read(rtsock, msg, sizeof(msg)); - if (dflag > 1) { - syslog(LOG_DEBUG, "<%s> received a routing message " - "(type = %d, len = %d)", __func__, rtmsg_type(msg), n); - } + syslog(LOG_DEBUG, "<%s> received a routing message " + "(type = %d, len = %d)", __func__, rtmsg_type(msg), n); + if (n > rtmsg_len(msg)) { /* - * This usually won't happen for messages received on + * This usually won't happen for messages received on * a routing socket. */ - if (dflag > 1) - syslog(LOG_DEBUG, - "<%s> received data length is larger than " - "1st routing message len. multiple messages? " - "read %d bytes, but 1st msg len = %d", - __func__, n, rtmsg_len(msg)); + syslog(LOG_DEBUG, + "<%s> received data length is larger than " + "1st routing message len. multiple messages? " + "read %d bytes, but 1st msg len = %d", + __func__, n, rtmsg_len(msg)); #if 0 /* adjust length */ n = rtmsg_len(msg); #endif } lim = msg + n; for (next = msg; next < lim; next += len) { int oldifflags; next = get_next_msg(next, lim, 0, &len, - RTADV_TYPE2BITMASK(RTM_ADD) | - RTADV_TYPE2BITMASK(RTM_DELETE) | - RTADV_TYPE2BITMASK(RTM_NEWADDR) | - RTADV_TYPE2BITMASK(RTM_DELADDR) | - RTADV_TYPE2BITMASK(RTM_IFINFO)); + RTADV_TYPE2BITMASK(RTM_ADD) | + RTADV_TYPE2BITMASK(RTM_DELETE) | + RTADV_TYPE2BITMASK(RTM_NEWADDR) | + RTADV_TYPE2BITMASK(RTM_DELADDR) | + RTADV_TYPE2BITMASK(RTM_IFINFO) | + RTADV_TYPE2BITMASK(RTM_IFANNOUNCE)); if (len == 0) break; type = rtmsg_type(next); switch (type) { case RTM_ADD: case RTM_DELETE: ifindex = get_rtm_ifindex(next); break; case RTM_NEWADDR: case RTM_DELADDR: ifindex = get_ifam_ifindex(next); break; case RTM_IFINFO: ifindex = get_ifm_ifindex(next); break; - default: - /* should not reach here */ - if (dflag > 1) { + case RTM_IFANNOUNCE: + ifan = (struct if_announcemsghdr *)next; + switch (ifan->ifan_what) { + case IFAN_ARRIVAL: + case IFAN_DEPARTURE: + break; + default: syslog(LOG_DEBUG, - "<%s:%d> unknown rtmsg %d on %s", - __func__, __LINE__, type, - if_indextoname(ifindex, ifname)); + "<%s:%d> unknown ifan msg (ifan_what=%d)", + __func__, __LINE__, ifan->ifan_what); + continue; } + + syslog(LOG_INFO, "<%s>: if_announcemsg (idx=%d:%d)", + __func__, ifan->ifan_index, ifan->ifan_what); + init_iflist(); + error = ifl_lookup(ifan->ifan_name, + ifl_names, ifl_len); + if (error) { + syslog(LOG_INFO, "<%s>: not a target " + "interface (idx=%d)", __func__, + ifan->ifan_index); + continue; + } + + switch (ifan->ifan_what) { + case IFAN_ARRIVAL: + error = getconfig(ifan->ifan_index); + if (error) + syslog(LOG_ERR, + "<%s>: getconfig failed (idx=%d)" + " Ignored.", __func__, + ifan->ifan_index); + break; + case IFAN_DEPARTURE: + error = rmconfig(ifan->ifan_index); + if (error) + syslog(LOG_ERR, + "<%s>: rmconfig failed (idx=%d)" + " Ignored.", __func__, + ifan->ifan_index); + break; + } continue; + default: + /* should not reach here */ + syslog(LOG_DEBUG, + "<%s:%d> unknown rtmsg %d on %s", + __func__, __LINE__, type, + if_indextoname(ifindex, ifname)); + continue; } if ((rai = if_indextorainfo(ifindex)) == NULL) { - if (dflag > 1) { - syslog(LOG_DEBUG, - "<%s> route changed on " - "non advertising interface(%s)", - __func__, - if_indextoname(ifindex, ifname)); - } + syslog(LOG_DEBUG, + "<%s> route changed on " + "non advertising interface(%s)", + __func__, + if_indextoname(ifindex, ifname)); continue; } oldifflags = iflist[ifindex]->ifm_flags; switch (type) { case RTM_ADD: /* init ifflags because it may have changed */ iflist[ifindex]->ifm_flags = if_getflags(ifindex, iflist[ifindex]->ifm_flags); if (sflag) break; /* we aren't interested in prefixes */ addr = get_addr(msg); plen = get_prefixlen(msg); /* sanity check for plen */ /* as RFC2373, prefixlen is at least 4 */ if (plen < 4 || plen > 127) { syslog(LOG_INFO, "<%s> new interface route's" "plen %d is invalid for a prefix", __func__, plen); break; } - prefix = find_prefix(rai, addr, plen); - if (prefix) { - if (prefix->timer) { + pfx = find_prefix(rai, addr, plen); + if (pfx) { + if (pfx->pfx_timer) { /* * If the prefix has been invalidated, * make it available again. */ - update_prefix(prefix); + update_prefix(pfx); prefixchange = 1; - } else if (dflag > 1) { + } else syslog(LOG_DEBUG, "<%s> new prefix(%s/%d) " "added on %s, " "but it was already in list", __func__, inet_ntop(AF_INET6, addr, - (char *)addrbuf, INET6_ADDRSTRLEN), - plen, rai->ifname); - } + (char *)addrbuf, + sizeof(addrbuf)), + plen, rai->rai_ifname); break; } make_prefix(rai, ifindex, addr, plen); prefixchange = 1; break; case RTM_DELETE: /* init ifflags because it may have changed */ iflist[ifindex]->ifm_flags = if_getflags(ifindex, iflist[ifindex]->ifm_flags); if (sflag) break; addr = get_addr(msg); plen = get_prefixlen(msg); /* sanity check for plen */ /* as RFC2373, prefixlen is at least 4 */ if (plen < 4 || plen > 127) { syslog(LOG_INFO, "<%s> deleted interface route's " "plen %d is invalid for a prefix", __func__, plen); break; } - prefix = find_prefix(rai, addr, plen); - if (prefix == NULL) { - if (dflag > 1) { - syslog(LOG_DEBUG, - "<%s> prefix(%s/%d) was " - "deleted on %s, " - "but it was not in list", - __func__, - inet_ntop(AF_INET6, addr, - (char *)addrbuf, INET6_ADDRSTRLEN), - plen, rai->ifname); - } + pfx = find_prefix(rai, addr, plen); + if (pfx == NULL) { + syslog(LOG_DEBUG, + "<%s> prefix(%s/%d) was deleted on %s, " + "but it was not in list", + __func__, inet_ntop(AF_INET6, addr, + (char *)addrbuf, sizeof(addrbuf)), + plen, rai->rai_ifname); break; } - invalidate_prefix(prefix); + invalidate_prefix(pfx); prefixchange = 1; break; case RTM_NEWADDR: case RTM_DELADDR: /* init ifflags because it may have changed */ iflist[ifindex]->ifm_flags = if_getflags(ifindex, iflist[ifindex]->ifm_flags); break; case RTM_IFINFO: iflist[ifindex]->ifm_flags = get_ifm_flags(next); break; default: /* should not reach here */ - if (dflag > 1) { - syslog(LOG_DEBUG, - "<%s:%d> unknown rtmsg %d on %s", - __func__, __LINE__, type, - if_indextoname(ifindex, ifname)); - } + syslog(LOG_DEBUG, + "<%s:%d> unknown rtmsg %d on %s", + __func__, __LINE__, type, + if_indextoname(ifindex, ifname)); return; } /* check if an interface flag is changed */ if ((oldifflags & IFF_UP) && /* UP to DOWN */ !(iflist[ifindex]->ifm_flags & IFF_UP)) { syslog(LOG_INFO, "<%s> interface %s becomes down. stop timer.", - __func__, rai->ifname); - rtadvd_remove_timer(&rai->timer); + __func__, rai->rai_ifname); + rtadvd_remove_timer(rai->rai_timer); + rai->rai_timer = NULL; } else if (!(oldifflags & IFF_UP) && /* DOWN to UP */ - (iflist[ifindex]->ifm_flags & IFF_UP)) { + (iflist[ifindex]->ifm_flags & IFF_UP)) { syslog(LOG_INFO, "<%s> interface %s becomes up. restart timer.", - __func__, rai->ifname); + __func__, rai->rai_ifname); - rai->initcounter = 0; /* reset the counter */ - rai->waiting = 0; /* XXX */ - rai->timer = rtadvd_add_timer(ra_timeout, + rai->rai_initcounter = 0; /* reset the counter */ + rai->rai_waiting = 0; /* XXX */ + rai->rai_timer = rtadvd_add_timer(ra_timeout, ra_timer_update, rai, rai); - ra_timer_update((void *)rai, &rai->timer->tm); - rtadvd_set_timer(&rai->timer->tm, rai->timer); + ra_timer_update(rai, &rai->rai_timer->rat_tm); + rtadvd_set_timer(&rai->rai_timer->rat_tm, + rai->rai_timer); } else if (prefixchange && (iflist[ifindex]->ifm_flags & IFF_UP)) { /* * An advertised prefix has been added or invalidated. * Will notice the change in a short delay. */ - rai->initcounter = 0; + rai->rai_initcounter = 0; set_short_delay(rai); } } return; } void -rtadvd_input() +rtadvd_input(void) { - int i; + ssize_t i; int *hlimp = NULL; #ifdef OLDRAWSOCKET struct ip6_hdr *ip; -#endif +#endif struct icmp6_hdr *icp; int ifindex = 0; struct cmsghdr *cm; struct in6_pktinfo *pi = NULL; u_char ntopbuf[INET6_ADDRSTRLEN], ifnamebuf[IFNAMSIZ]; struct in6_addr dst = in6addr_any; /* * Get message. We reset msg_controllen since the field could * be modified if we had received a message before setting * receive options. */ rcvmhdr.msg_controllen = rcvcmsgbuflen; if ((i = recvmsg(sock, &rcvmhdr, 0)) < 0) return; /* extract optional information via Advanced API */ for (cm = (struct cmsghdr *)CMSG_FIRSTHDR(&rcvmhdr); cm; cm = (struct cmsghdr *)CMSG_NXTHDR(&rcvmhdr, cm)) { if (cm->cmsg_level == IPPROTO_IPV6 && cm->cmsg_type == IPV6_PKTINFO && cm->cmsg_len == CMSG_LEN(sizeof(struct in6_pktinfo))) { pi = (struct in6_pktinfo *)(CMSG_DATA(cm)); ifindex = pi->ipi6_ifindex; dst = pi->ipi6_addr; } if (cm->cmsg_level == IPPROTO_IPV6 && cm->cmsg_type == IPV6_HOPLIMIT && cm->cmsg_len == CMSG_LEN(sizeof(int))) hlimp = (int *)CMSG_DATA(cm); } if (ifindex == 0) { syslog(LOG_ERR, - "<%s> failed to get receiving interface", - __func__); + "<%s> failed to get receiving interface", + __func__); return; } if (hlimp == NULL) { syslog(LOG_ERR, - "<%s> failed to get receiving hop limit", - __func__); + "<%s> failed to get receiving hop limit", + __func__); return; } /* * If we happen to receive data on an interface which is now gone * or down, just discard the data. */ if (iflist[pi->ipi6_ifindex] == NULL || (iflist[pi->ipi6_ifindex]->ifm_flags & IFF_UP) == 0) { syslog(LOG_INFO, - "<%s> received data on a disabled interface (%s)", - __func__, - (iflist[pi->ipi6_ifindex] == NULL) ? "[gone]" : - if_indextoname(pi->ipi6_ifindex, ifnamebuf)); + "<%s> received data on a disabled interface (%s)", + __func__, + (iflist[pi->ipi6_ifindex] == NULL) ? "[gone]" : + if_indextoname(pi->ipi6_ifindex, ifnamebuf)); return; } #ifdef OLDRAWSOCKET - if (i < sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr)) { + if ((size_t)i < sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr)) { syslog(LOG_ERR, - "<%s> packet size(%d) is too short", - __func__, i); + "<%s> packet size(%d) is too short", + __func__, i); return; } ip = (struct ip6_hdr *)rcvmhdr.msg_iov[0].iov_base; icp = (struct icmp6_hdr *)(ip + 1); /* XXX: ext. hdr? */ #else - if (i < sizeof(struct icmp6_hdr)) { + if ((size_t)i < sizeof(struct icmp6_hdr)) { syslog(LOG_ERR, - "<%s> packet size(%d) is too short", - __func__, i); + "<%s> packet size(%zd) is too short", + __func__, i); return; } icp = (struct icmp6_hdr *)rcvmhdr.msg_iov[0].iov_base; #endif switch (icp->icmp6_type) { case ND_ROUTER_SOLICIT: /* - * Message verification - RFC-2461 6.1.1 + * Message verification - RFC 4861 6.1.1 * XXX: these checks must be done in the kernel as well, * but we can't completely rely on them. */ if (*hlimp != 255) { syslog(LOG_NOTICE, "<%s> RS with invalid hop limit(%d) " "received from %s on %s", __func__, *hlimp, inet_ntop(AF_INET6, &rcvfrom.sin6_addr, ntopbuf, - INET6_ADDRSTRLEN), + sizeof(ntopbuf)), if_indextoname(pi->ipi6_ifindex, ifnamebuf)); return; } if (icp->icmp6_code) { syslog(LOG_NOTICE, "<%s> RS with invalid ICMP6 code(%d) " "received from %s on %s", __func__, icp->icmp6_code, inet_ntop(AF_INET6, &rcvfrom.sin6_addr, ntopbuf, - INET6_ADDRSTRLEN), + sizeof(ntopbuf)), if_indextoname(pi->ipi6_ifindex, ifnamebuf)); return; } - if (i < sizeof(struct nd_router_solicit)) { + if ((size_t)i < sizeof(struct nd_router_solicit)) { syslog(LOG_NOTICE, "<%s> RS from %s on %s does not have enough " - "length (len = %d)", + "length (len = %zd)", __func__, inet_ntop(AF_INET6, &rcvfrom.sin6_addr, ntopbuf, - INET6_ADDRSTRLEN), + sizeof(ntopbuf)), if_indextoname(pi->ipi6_ifindex, ifnamebuf), i); return; } rs_input(i, (struct nd_router_solicit *)icp, pi, &rcvfrom); break; case ND_ROUTER_ADVERT: /* - * Message verification - RFC-2461 6.1.2 - * XXX: there's a same dilemma as above... + * Message verification - RFC 4861 6.1.2 + * XXX: there's the same dilemma as above... */ + if (!IN6_IS_ADDR_LINKLOCAL(&rcvfrom.sin6_addr)) { + syslog(LOG_NOTICE, + "<%s> RA witn non-linklocal source address " + "received from %s on %s", + __func__, inet_ntop(AF_INET6, &rcvfrom.sin6_addr, + ntopbuf, sizeof(ntopbuf)), + if_indextoname(pi->ipi6_ifindex, ifnamebuf)); + return; + } if (*hlimp != 255) { syslog(LOG_NOTICE, "<%s> RA with invalid hop limit(%d) " "received from %s on %s", __func__, *hlimp, inet_ntop(AF_INET6, &rcvfrom.sin6_addr, ntopbuf, - INET6_ADDRSTRLEN), + sizeof(ntopbuf)), if_indextoname(pi->ipi6_ifindex, ifnamebuf)); return; } if (icp->icmp6_code) { syslog(LOG_NOTICE, "<%s> RA with invalid ICMP6 code(%d) " "received from %s on %s", __func__, icp->icmp6_code, inet_ntop(AF_INET6, &rcvfrom.sin6_addr, ntopbuf, - INET6_ADDRSTRLEN), + sizeof(ntopbuf)), if_indextoname(pi->ipi6_ifindex, ifnamebuf)); return; } - if (i < sizeof(struct nd_router_advert)) { + if ((size_t)i < sizeof(struct nd_router_advert)) { syslog(LOG_NOTICE, "<%s> RA from %s on %s does not have enough " - "length (len = %d)", + "length (len = %zd)", __func__, inet_ntop(AF_INET6, &rcvfrom.sin6_addr, ntopbuf, - INET6_ADDRSTRLEN), + sizeof(ntopbuf)), if_indextoname(pi->ipi6_ifindex, ifnamebuf), i); return; } ra_input(i, (struct nd_router_advert *)icp, pi, &rcvfrom); break; case ICMP6_ROUTER_RENUMBERING: if (accept_rr == 0) { syslog(LOG_ERR, "<%s> received a router renumbering " "message, but not allowed to be accepted", __func__); break; } rr_input(i, (struct icmp6_router_renum *)icp, pi, &rcvfrom, - &dst); + &dst); break; default: /* * Note that this case is POSSIBLE, especially just * after invocation of the daemon. This is because we * could receive message after opening the socket and * before setting ICMP6 type filter(see sock_open()). */ syslog(LOG_ERR, "<%s> invalid icmp type(%d)", __func__, icp->icmp6_type); return; } return; } static void rs_input(int len, struct nd_router_solicit *rs, struct in6_pktinfo *pi, struct sockaddr_in6 *from) { - u_char ntopbuf[INET6_ADDRSTRLEN], ifnamebuf[IFNAMSIZ]; - union nd_opts ndopts; - struct rainfo *ra; + u_char ntopbuf[INET6_ADDRSTRLEN]; + u_char ifnamebuf[IFNAMSIZ]; + union nd_opt ndopts; + struct rainfo *rai; struct soliciter *sol; syslog(LOG_DEBUG, - "<%s> RS received from %s on %s", - __func__, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - if_indextoname(pi->ipi6_ifindex, ifnamebuf)); + "<%s> RS received from %s on %s", + __func__, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, sizeof(ntopbuf)), + if_indextoname(pi->ipi6_ifindex, ifnamebuf)); /* ND option check */ memset(&ndopts, 0, sizeof(ndopts)); if (nd6_options((struct nd_opt_hdr *)(rs + 1), len - sizeof(struct nd_router_solicit), &ndopts, NDOPT_FLAG_SRCLINKADDR)) { syslog(LOG_INFO, - "<%s> ND option check failed for an RS from %s on %s", - __func__, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - if_indextoname(pi->ipi6_ifindex, ifnamebuf)); + "<%s> ND option check failed for an RS from %s on %s", + __func__, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), + if_indextoname(pi->ipi6_ifindex, ifnamebuf)); return; } /* * If the IP source address is the unspecified address, there * must be no source link-layer address option in the message. - * (RFC-2461 6.1.1) + * (RFC 4861 6.1.1) */ if (IN6_IS_ADDR_UNSPECIFIED(&from->sin6_addr) && - ndopts.nd_opts_src_lladdr) { + ndopts.opt_src_lladdr) { syslog(LOG_INFO, - "<%s> RS from unspecified src on %s has a link-layer" - " address option", - __func__, - if_indextoname(pi->ipi6_ifindex, ifnamebuf)); + "<%s> RS from unspecified src on %s has a link-layer" + " address option", + __func__, if_indextoname(pi->ipi6_ifindex, ifnamebuf)); goto done; } - ra = ralist; - while (ra != NULL) { - if (pi->ipi6_ifindex == ra->ifindex) + TAILQ_FOREACH(rai, &railist, rai_next) + if (pi->ipi6_ifindex == (unsigned int)rai->rai_ifindex) break; - ra = ra->next; - } - if (ra == NULL) { + + if (rai == NULL) { syslog(LOG_INFO, "<%s> RS received on non advertising interface(%s)", __func__, if_indextoname(pi->ipi6_ifindex, ifnamebuf)); goto done; } - ra->rsinput++; /* increment statistics */ + rai->rai_rsinput++; /* increment statistics */ /* * Decide whether to send RA according to the rate-limit * consideration. */ /* record sockaddr waiting for RA, if possible */ sol = (struct soliciter *)malloc(sizeof(*sol)); if (sol) { - sol->addr = *from; - /* XXX RFC2553 need clarification on flowinfo */ - sol->addr.sin6_flowinfo = 0; - sol->next = ra->soliciter; - ra->soliciter = sol; + sol->sol_addr = *from; + /* XXX RFC 2553 need clarification on flowinfo */ + sol->sol_addr.sin6_flowinfo = 0; + TAILQ_INSERT_TAIL(&rai->rai_soliciter, sol, sol_next); } /* * If there is already a waiting RS packet, don't * update the timer. */ - if (ra->waiting++) + if (rai->rai_waiting++) goto done; - set_short_delay(ra); + set_short_delay(rai); done: free_ndopts(&ndopts); return; } static void -set_short_delay(rai) - struct rainfo *rai; +set_short_delay(struct rainfo *rai) { long delay; /* must not be greater than 1000000 */ struct timeval interval, now, min_delay, tm_tmp, *rest; /* * Compute a random delay. If the computed value * corresponds to a time later than the time the next * multicast RA is scheduled to be sent, ignore the random * delay and send the advertisement at the - * already-scheduled time. RFC-2461 6.2.6 + * already-scheduled time. RFC 4861 6.2.6 */ #ifdef HAVE_ARC4RANDOM delay = arc4random_uniform(MAX_RA_DELAY_TIME); #else delay = random() % MAX_RA_DELAY_TIME; #endif interval.tv_sec = 0; interval.tv_usec = delay; - rest = rtadvd_timer_rest(rai->timer); - if (TIMEVAL_LT(*rest, interval)) { + rest = rtadvd_timer_rest(rai->rai_timer); + if (TIMEVAL_LT(rest, &interval)) { syslog(LOG_DEBUG, "<%s> random delay is larger than " "the rest of the current timer", __func__); interval = *rest; } /* * If we sent a multicast Router Advertisement within * the last MIN_DELAY_BETWEEN_RAS seconds, schedule * the advertisement to be sent at a time corresponding to * MIN_DELAY_BETWEEN_RAS plus the random value after the * previous advertisement was sent. */ gettimeofday(&now, NULL); - TIMEVAL_SUB(&now, &rai->lastsent, &tm_tmp); + TIMEVAL_SUB(&now, &rai->rai_lastsent, &tm_tmp); min_delay.tv_sec = MIN_DELAY_BETWEEN_RAS; min_delay.tv_usec = 0; - if (TIMEVAL_LT(tm_tmp, min_delay)) { + if (TIMEVAL_LT(&tm_tmp, &min_delay)) { TIMEVAL_SUB(&min_delay, &tm_tmp, &min_delay); TIMEVAL_ADD(&min_delay, &interval, &interval); } - rtadvd_set_timer(&interval, rai->timer); + rtadvd_set_timer(&interval, rai->rai_timer); } +static int +check_accept_rtadv(int idx) +{ + struct in6_ndireq nd; + u_char ifname[IFNAMSIZ]; + int s6; + int error; + + if ((s6 = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) { + syslog(LOG_ERR, + "<%s> open socket failed for idx=%d.", + __func__, idx); + return (0); + } + if ((if_indextoname(idx, ifname)) == NULL) { + syslog(LOG_ERR, + "<%s> ifindex->ifname failed (idx=%d).", + __func__, idx); + close(s6); + return (0); + } + memset(&nd, 0, sizeof(nd)); + strncpy(nd.ifname, ifname, sizeof(nd.ifname)); + error = ioctl(s6, SIOCGIFINFO_IN6, &nd); + if (error) { + syslog(LOG_ERR, + "<%s> ioctl(SIOCGIFINFO_IN6) failed for idx=%d.", + __func__, idx); + nd.ndi.flags = 0; + } + close(s6); + + return (nd.ndi.flags & ND6_IFF_ACCEPT_RTADV); +} + +static int +getinet6sysctl(int code) +{ + int mib[] = { CTL_NET, PF_INET6, IPPROTO_IPV6, 0 }; + int value; + size_t size; + + mib[3] = code; + size = sizeof(value); + if (sysctl(mib, sizeof(mib)/sizeof(mib[0]), &value, &size, NULL, 0) + < 0) { + syslog(LOG_ERR, "<%s>: failed to get ip6 sysctl(%d): %s", + __func__, code, + strerror(errno)); + return (-1); + } + else + return (value); +} + static void -ra_input(int len, struct nd_router_advert *ra, +ra_input(int len, struct nd_router_advert *nra, struct in6_pktinfo *pi, struct sockaddr_in6 *from) { struct rainfo *rai; - u_char ntopbuf[INET6_ADDRSTRLEN], ifnamebuf[IFNAMSIZ]; - union nd_opts ndopts; - char *on_off[] = {"OFF", "ON"}; + u_char ntopbuf[INET6_ADDRSTRLEN]; + u_char ifnamebuf[IFNAMSIZ]; + union nd_opt ndopts; + const char *on_off[] = {"OFF", "ON"}; u_int32_t reachabletime, retranstimer, mtu; int inconsistent = 0; + int error; - syslog(LOG_DEBUG, - "<%s> RA received from %s on %s", - __func__, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - if_indextoname(pi->ipi6_ifindex, ifnamebuf)); - + syslog(LOG_DEBUG, "<%s> RA received from %s on %s", __func__, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, sizeof(ntopbuf)), + if_indextoname(pi->ipi6_ifindex, ifnamebuf)); + + if (!check_accept_rtadv(pi->ipi6_ifindex)) { + syslog(LOG_INFO, + "<%s> An RA from %s on %s ignored (no ACCEPT_RTADV flag).", + __func__, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), if_indextoname(pi->ipi6_ifindex, + ifnamebuf)); + return; + } + /* ND option check */ memset(&ndopts, 0, sizeof(ndopts)); - if (nd6_options((struct nd_opt_hdr *)(ra + 1), - len - sizeof(struct nd_router_advert), - &ndopts, NDOPT_FLAG_SRCLINKADDR | - NDOPT_FLAG_PREFIXINFO | NDOPT_FLAG_MTU)) { + error = nd6_options((struct nd_opt_hdr *)(nra + 1), + len - sizeof(struct nd_router_advert), &ndopts, + NDOPT_FLAG_SRCLINKADDR | NDOPT_FLAG_PREFIXINFO | NDOPT_FLAG_MTU | + NDOPT_FLAG_RDNSS | NDOPT_FLAG_DNSSL); + if (error) { syslog(LOG_INFO, - "<%s> ND option check failed for an RA from %s on %s", - __func__, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - if_indextoname(pi->ipi6_ifindex, ifnamebuf)); + "<%s> ND option check failed for an RA from %s on %s", + __func__, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), if_indextoname(pi->ipi6_ifindex, + ifnamebuf)); return; } /* - * RA consistency check according to RFC-2461 6.2.7 + * RA consistency check according to RFC 4861 6.2.7 */ - if ((rai = if_indextorainfo(pi->ipi6_ifindex)) == 0) { + rai = if_indextorainfo(pi->ipi6_ifindex); + if (rai == NULL) { syslog(LOG_INFO, - "<%s> received RA from %s on non-advertising" - " interface(%s)", - __func__, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - if_indextoname(pi->ipi6_ifindex, ifnamebuf)); + "<%s> received RA from %s on non-advertising" + " interface(%s)", + __func__, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), if_indextoname(pi->ipi6_ifindex, + ifnamebuf)); goto done; } - rai->rainput++; /* increment statistics */ - + rai->rai_rainput++; /* increment statistics */ + /* Cur Hop Limit value */ - if (ra->nd_ra_curhoplimit && rai->hoplimit && - ra->nd_ra_curhoplimit != rai->hoplimit) { + if (nra->nd_ra_curhoplimit && rai->rai_hoplimit && + nra->nd_ra_curhoplimit != rai->rai_hoplimit) { syslog(LOG_INFO, - "<%s> CurHopLimit inconsistent on %s:" - " %d from %s, %d from us", - __func__, - rai->ifname, - ra->nd_ra_curhoplimit, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - rai->hoplimit); + "<%s> CurHopLimit inconsistent on %s:" + " %d from %s, %d from us", + __func__, rai->rai_ifname, nra->nd_ra_curhoplimit, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), rai->rai_hoplimit); inconsistent++; } /* M flag */ - if ((ra->nd_ra_flags_reserved & ND_RA_FLAG_MANAGED) != - rai->managedflg) { + if ((nra->nd_ra_flags_reserved & ND_RA_FLAG_MANAGED) != + rai->rai_managedflg) { syslog(LOG_INFO, - "<%s> M flag inconsistent on %s:" - " %s from %s, %s from us", - __func__, - rai->ifname, - on_off[!rai->managedflg], - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - on_off[rai->managedflg]); + "<%s> M flag inconsistent on %s:" + " %s from %s, %s from us", + __func__, rai->rai_ifname, on_off[!rai->rai_managedflg], + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), on_off[rai->rai_managedflg]); inconsistent++; } /* O flag */ - if ((ra->nd_ra_flags_reserved & ND_RA_FLAG_OTHER) != - rai->otherflg) { + if ((nra->nd_ra_flags_reserved & ND_RA_FLAG_OTHER) != + rai->rai_otherflg) { syslog(LOG_INFO, - "<%s> O flag inconsistent on %s:" - " %s from %s, %s from us", - __func__, - rai->ifname, - on_off[!rai->otherflg], - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - on_off[rai->otherflg]); + "<%s> O flag inconsistent on %s:" + " %s from %s, %s from us", + __func__, rai->rai_ifname, on_off[!rai->rai_otherflg], + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), on_off[rai->rai_otherflg]); inconsistent++; } /* Reachable Time */ - reachabletime = ntohl(ra->nd_ra_reachable); - if (reachabletime && rai->reachabletime && - reachabletime != rai->reachabletime) { + reachabletime = ntohl(nra->nd_ra_reachable); + if (reachabletime && rai->rai_reachabletime && + reachabletime != rai->rai_reachabletime) { syslog(LOG_INFO, - "<%s> ReachableTime inconsistent on %s:" - " %d from %s, %d from us", - __func__, - rai->ifname, - reachabletime, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - rai->reachabletime); + "<%s> ReachableTime inconsistent on %s:" + " %d from %s, %d from us", + __func__, rai->rai_ifname, reachabletime, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), rai->rai_reachabletime); inconsistent++; } /* Retrans Timer */ - retranstimer = ntohl(ra->nd_ra_retransmit); - if (retranstimer && rai->retranstimer && - retranstimer != rai->retranstimer) { + retranstimer = ntohl(nra->nd_ra_retransmit); + if (retranstimer && rai->rai_retranstimer && + retranstimer != rai->rai_retranstimer) { syslog(LOG_INFO, - "<%s> RetranceTimer inconsistent on %s:" - " %d from %s, %d from us", - __func__, - rai->ifname, - retranstimer, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - rai->retranstimer); + "<%s> RetranceTimer inconsistent on %s:" + " %d from %s, %d from us", + __func__, rai->rai_ifname, retranstimer, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), rai->rai_retranstimer); inconsistent++; } /* Values in the MTU options */ - if (ndopts.nd_opts_mtu) { - mtu = ntohl(ndopts.nd_opts_mtu->nd_opt_mtu_mtu); - if (mtu && rai->linkmtu && mtu != rai->linkmtu) { + if (ndopts.opt_mtu) { + mtu = ntohl(ndopts.opt_mtu->nd_opt_mtu_mtu); + if (mtu && rai->rai_linkmtu && mtu != rai->rai_linkmtu) { syslog(LOG_INFO, - "<%s> MTU option value inconsistent on %s:" - " %d from %s, %d from us", - __func__, - rai->ifname, mtu, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - rai->linkmtu); + "<%s> MTU option value inconsistent on %s:" + " %d from %s, %d from us", + __func__, rai->rai_ifname, mtu, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), rai->rai_linkmtu); inconsistent++; } } /* Preferred and Valid Lifetimes for prefixes */ { - struct nd_optlist *optp = ndopts.nd_opts_list; + struct nd_optlist *nol; - if (ndopts.nd_opts_pi) { - if (prefix_check(ndopts.nd_opts_pi, rai, from)) + if (ndopts.opt_pi) + if (prefix_check(ndopts.opt_pi, rai, from)) inconsistent++; - } - while (optp) { - if (prefix_check((struct nd_opt_prefix_info *)optp->opt, - rai, from)) + + TAILQ_FOREACH(nol, &ndopts.opt_list, nol_next) + if (prefix_check((struct nd_opt_prefix_info *)nol->nol_opt, + rai, from)) inconsistent++; - optp = optp->next; - } } if (inconsistent) - rai->rainconsistent++; - + rai->rai_rainconsistent++; + done: free_ndopts(&ndopts); return; } /* return a non-zero value if the received prefix is inconsitent with ours */ static int prefix_check(struct nd_opt_prefix_info *pinfo, - struct rainfo *rai, struct sockaddr_in6 *from) + struct rainfo *rai, struct sockaddr_in6 *from) { u_int32_t preferred_time, valid_time; - struct prefix *pp; + struct prefix *pfx; int inconsistent = 0; - u_char ntopbuf[INET6_ADDRSTRLEN], prefixbuf[INET6_ADDRSTRLEN]; + u_char ntopbuf[INET6_ADDRSTRLEN]; + u_char prefixbuf[INET6_ADDRSTRLEN]; struct timeval now; #if 0 /* impossible */ if (pinfo->nd_opt_pi_type != ND_OPT_PREFIX_INFORMATION) - return(0); + return (0); #endif /* * log if the adveritsed prefix has link-local scope(sanity check?) */ - if (IN6_IS_ADDR_LINKLOCAL(&pinfo->nd_opt_pi_prefix)) { + if (IN6_IS_ADDR_LINKLOCAL(&pinfo->nd_opt_pi_prefix)) syslog(LOG_INFO, - "<%s> link-local prefix %s/%d is advertised " - "from %s on %s", - __func__, - inet_ntop(AF_INET6, &pinfo->nd_opt_pi_prefix, - prefixbuf, INET6_ADDRSTRLEN), - pinfo->nd_opt_pi_prefix_len, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - rai->ifname); - } + "<%s> link-local prefix %s/%d is advertised " + "from %s on %s", + __func__, + inet_ntop(AF_INET6, &pinfo->nd_opt_pi_prefix, prefixbuf, + sizeof(prefixbuf)), + pinfo->nd_opt_pi_prefix_len, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), rai->rai_ifname); - if ((pp = find_prefix(rai, &pinfo->nd_opt_pi_prefix, - pinfo->nd_opt_pi_prefix_len)) == NULL) { + if ((pfx = find_prefix(rai, &pinfo->nd_opt_pi_prefix, + pinfo->nd_opt_pi_prefix_len)) == NULL) { syslog(LOG_INFO, - "<%s> prefix %s/%d from %s on %s is not in our list", - __func__, - inet_ntop(AF_INET6, &pinfo->nd_opt_pi_prefix, - prefixbuf, INET6_ADDRSTRLEN), - pinfo->nd_opt_pi_prefix_len, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - rai->ifname); - return(0); + "<%s> prefix %s/%d from %s on %s is not in our list", + __func__, + inet_ntop(AF_INET6, &pinfo->nd_opt_pi_prefix, prefixbuf, + sizeof(prefixbuf)), + pinfo->nd_opt_pi_prefix_len, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), rai->rai_ifname); + return (0); } preferred_time = ntohl(pinfo->nd_opt_pi_preferred_time); - if (pp->pltimeexpire) { + if (pfx->pfx_pltimeexpire) { /* * The lifetime is decremented in real time, so we should * compare the expiration time. * (RFC 2461 Section 6.2.7.) * XXX: can we really expect that all routers on the link * have synchronized clocks? */ gettimeofday(&now, NULL); preferred_time += now.tv_sec; - if (!pp->timer && rai->clockskew && - abs(preferred_time - pp->pltimeexpire) > rai->clockskew) { + if (!pfx->pfx_timer && rai->rai_clockskew && + abs(preferred_time - pfx->pfx_pltimeexpire) > rai->rai_clockskew) { syslog(LOG_INFO, - "<%s> preferred lifetime for %s/%d" - " (decr. in real time) inconsistent on %s:" - " %d from %s, %ld from us", - __func__, - inet_ntop(AF_INET6, &pinfo->nd_opt_pi_prefix, - prefixbuf, INET6_ADDRSTRLEN), - pinfo->nd_opt_pi_prefix_len, - rai->ifname, preferred_time, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - pp->pltimeexpire); + "<%s> preferred lifetime for %s/%d" + " (decr. in real time) inconsistent on %s:" + " %d from %s, %ld from us", + __func__, + inet_ntop(AF_INET6, &pinfo->nd_opt_pi_prefix, prefixbuf, + sizeof(prefixbuf)), + pinfo->nd_opt_pi_prefix_len, + rai->rai_ifname, preferred_time, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), pfx->pfx_pltimeexpire); inconsistent++; } - } else if (!pp->timer && preferred_time != pp->preflifetime) { + } else if (!pfx->pfx_timer && preferred_time != pfx->pfx_preflifetime) syslog(LOG_INFO, - "<%s> preferred lifetime for %s/%d" - " inconsistent on %s:" - " %d from %s, %d from us", - __func__, - inet_ntop(AF_INET6, &pinfo->nd_opt_pi_prefix, - prefixbuf, INET6_ADDRSTRLEN), - pinfo->nd_opt_pi_prefix_len, - rai->ifname, preferred_time, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - pp->preflifetime); - } + "<%s> preferred lifetime for %s/%d" + " inconsistent on %s:" + " %d from %s, %d from us", + __func__, + inet_ntop(AF_INET6, &pinfo->nd_opt_pi_prefix, prefixbuf, + sizeof(prefixbuf)), + pinfo->nd_opt_pi_prefix_len, + rai->rai_ifname, preferred_time, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), pfx->pfx_preflifetime); valid_time = ntohl(pinfo->nd_opt_pi_valid_time); - if (pp->vltimeexpire) { + if (pfx->pfx_vltimeexpire) { gettimeofday(&now, NULL); valid_time += now.tv_sec; - if (!pp->timer && rai->clockskew && - abs(valid_time - pp->vltimeexpire) > rai->clockskew) { + if (!pfx->pfx_timer && rai->rai_clockskew && + abs(valid_time - pfx->pfx_vltimeexpire) > rai->rai_clockskew) { syslog(LOG_INFO, - "<%s> valid lifetime for %s/%d" - " (decr. in real time) inconsistent on %s:" - " %d from %s, %ld from us", - __func__, - inet_ntop(AF_INET6, &pinfo->nd_opt_pi_prefix, - prefixbuf, INET6_ADDRSTRLEN), - pinfo->nd_opt_pi_prefix_len, - rai->ifname, preferred_time, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - pp->vltimeexpire); + "<%s> valid lifetime for %s/%d" + " (decr. in real time) inconsistent on %s:" + " %d from %s, %ld from us", + __func__, + inet_ntop(AF_INET6, &pinfo->nd_opt_pi_prefix, prefixbuf, + sizeof(prefixbuf)), + pinfo->nd_opt_pi_prefix_len, + rai->rai_ifname, preferred_time, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), pfx->pfx_vltimeexpire); inconsistent++; } - } else if (!pp->timer && valid_time != pp->validlifetime) { + } else if (!pfx->pfx_timer && valid_time != pfx->pfx_validlifetime) { syslog(LOG_INFO, - "<%s> valid lifetime for %s/%d" - " inconsistent on %s:" - " %d from %s, %d from us", - __func__, - inet_ntop(AF_INET6, &pinfo->nd_opt_pi_prefix, - prefixbuf, INET6_ADDRSTRLEN), - pinfo->nd_opt_pi_prefix_len, - rai->ifname, valid_time, - inet_ntop(AF_INET6, &from->sin6_addr, - ntopbuf, INET6_ADDRSTRLEN), - pp->validlifetime); + "<%s> valid lifetime for %s/%d" + " inconsistent on %s:" + " %d from %s, %d from us", + __func__, + inet_ntop(AF_INET6, &pinfo->nd_opt_pi_prefix, prefixbuf, + sizeof(prefixbuf)), + pinfo->nd_opt_pi_prefix_len, + rai->rai_ifname, valid_time, + inet_ntop(AF_INET6, &from->sin6_addr, ntopbuf, + sizeof(ntopbuf)), pfx->pfx_validlifetime); inconsistent++; } - return(inconsistent); + return (inconsistent); } struct prefix * find_prefix(struct rainfo *rai, struct in6_addr *prefix, int plen) { - struct prefix *pp; + struct prefix *pfx; int bytelen, bitlen; u_char bitmask; - for (pp = rai->prefix.next; pp != &rai->prefix; pp = pp->next) { - if (plen != pp->prefixlen) + TAILQ_FOREACH(pfx, &rai->rai_prefix, pfx_next) { + if (plen != pfx->pfx_prefixlen) continue; + bytelen = plen / 8; bitlen = plen % 8; bitmask = 0xff << (8 - bitlen); - if (memcmp((void *)prefix, (void *)&pp->prefix, bytelen)) + + if (memcmp((void *)prefix, (void *)&pfx->pfx_prefix, bytelen)) continue; + if (bitlen == 0 || - ((prefix->s6_addr[bytelen] & bitmask) == - (pp->prefix.s6_addr[bytelen] & bitmask))) { - return(pp); + ((prefix->s6_addr[bytelen] & bitmask) == + (pfx->pfx_prefix.s6_addr[bytelen] & bitmask))) { + return (pfx); } } - return(NULL); + return (NULL); } /* check if p0/plen0 matches p1/plen1; return 1 if matches, otherwise 0. */ int prefix_match(struct in6_addr *p0, int plen0, - struct in6_addr *p1, int plen1) + struct in6_addr *p1, int plen1) { int bytelen, bitlen; u_char bitmask; if (plen0 < plen1) - return(0); + return (0); + bytelen = plen1 / 8; bitlen = plen1 % 8; bitmask = 0xff << (8 - bitlen); + if (memcmp((void *)p0, (void *)p1, bytelen)) - return(0); + return (0); + if (bitlen == 0 || ((p0->s6_addr[bytelen] & bitmask) == (p1->s6_addr[bytelen] & bitmask))) { - return(1); + return (1); } - return(0); + return (0); } static int nd6_options(struct nd_opt_hdr *hdr, int limit, - union nd_opts *ndopts, u_int32_t optflags) + union nd_opt *ndopts, u_int32_t optflags) { int optlen = 0; for (; limit > 0; limit -= optlen) { - if (limit < sizeof(struct nd_opt_hdr)) { + if ((size_t)limit < sizeof(struct nd_opt_hdr)) { syslog(LOG_INFO, "<%s> short option header", __func__); goto bad; } hdr = (struct nd_opt_hdr *)((caddr_t)hdr + optlen); if (hdr->nd_opt_len == 0) { syslog(LOG_INFO, "<%s> bad ND option length(0) (type = %d)", __func__, hdr->nd_opt_type); goto bad; } optlen = hdr->nd_opt_len << 3; if (optlen > limit) { syslog(LOG_INFO, "<%s> short option", __func__); goto bad; } - if (hdr->nd_opt_type > ND_OPT_MTU) { + if (hdr->nd_opt_type > ND_OPT_MTU && + hdr->nd_opt_type != ND_OPT_RDNSS && + hdr->nd_opt_type != ND_OPT_DNSSL) { syslog(LOG_INFO, "<%s> unknown ND option(type %d)", __func__, hdr->nd_opt_type); continue; } if ((ndopt_flags[hdr->nd_opt_type] & optflags) == 0) { syslog(LOG_INFO, "<%s> unexpected ND option(type %d)", __func__, hdr->nd_opt_type); continue; } /* * Option length check. Do it here for all fixed-length * options. */ - if ((hdr->nd_opt_type == ND_OPT_MTU && - (optlen != sizeof(struct nd_opt_mtu))) || - ((hdr->nd_opt_type == ND_OPT_PREFIX_INFORMATION && - optlen != sizeof(struct nd_opt_prefix_info)))) { + switch (hdr->nd_opt_type) { + case ND_OPT_MTU: + if (optlen == sizeof(struct nd_opt_mtu)) + break; + goto skip; + case ND_OPT_RDNSS: + if (optlen >= 24 && + (optlen - sizeof(struct nd_opt_rdnss)) % 16 == 0) + break; + goto skip; + case ND_OPT_DNSSL: + if (optlen >= 16 && + (optlen - sizeof(struct nd_opt_dnssl)) % 8 == 0) + break; + goto skip; + case ND_OPT_PREFIX_INFORMATION: + if (optlen == sizeof(struct nd_opt_prefix_info)) + break; +skip: syslog(LOG_INFO, "<%s> invalid option length", __func__); continue; } switch (hdr->nd_opt_type) { case ND_OPT_TARGET_LINKADDR: case ND_OPT_REDIRECTED_HEADER: + case ND_OPT_RDNSS: + case ND_OPT_DNSSL: break; /* we don't care about these options */ case ND_OPT_SOURCE_LINKADDR: case ND_OPT_MTU: - if (ndopts->nd_opt_array[hdr->nd_opt_type]) { + if (ndopts->opt_array[hdr->nd_opt_type]) { syslog(LOG_INFO, "<%s> duplicated ND option (type = %d)", __func__, hdr->nd_opt_type); } - ndopts->nd_opt_array[hdr->nd_opt_type] = hdr; + ndopts->opt_array[hdr->nd_opt_type] = hdr; break; case ND_OPT_PREFIX_INFORMATION: { - struct nd_optlist *pfxlist; + struct nd_optlist *nol; - if (ndopts->nd_opts_pi == 0) { - ndopts->nd_opts_pi = + if (ndopts->opt_pi == 0) { + ndopts->opt_pi = (struct nd_opt_prefix_info *)hdr; continue; } - if ((pfxlist = malloc(sizeof(*pfxlist))) == NULL) { + nol = malloc(sizeof(*nol)); + if (nol == NULL) { syslog(LOG_ERR, "<%s> can't allocate memory", __func__); goto bad; } - pfxlist->next = ndopts->nd_opts_list; - pfxlist->opt = hdr; - ndopts->nd_opts_list = pfxlist; + nol->nol_opt = hdr; + TAILQ_INSERT_TAIL(&(ndopts->opt_list), nol, nol_next); break; } default: /* impossible */ break; } } - return(0); + return (0); bad: free_ndopts(ndopts); - return(-1); + return (-1); } static void -free_ndopts(union nd_opts *ndopts) +free_ndopts(union nd_opt *ndopts) { - struct nd_optlist *opt = ndopts->nd_opts_list, *next; + struct nd_optlist *nol; - while (opt) { - next = opt->next; - free(opt); - opt = next; + while ((nol = TAILQ_FIRST(&ndopts->opt_list)) != NULL) { + TAILQ_REMOVE(&ndopts->opt_list, nol, nol_next); + free(nol); } } void -sock_open() +sock_open(void) { struct icmp6_filter filt; struct ipv6_mreq mreq; - struct rainfo *ra = ralist; + struct rainfo *rai; int on; /* XXX: should be max MTU attached to the node */ static u_char answer[1500]; rcvcmsgbuflen = CMSG_SPACE(sizeof(struct in6_pktinfo)) + - CMSG_SPACE(sizeof(int)); + CMSG_SPACE(sizeof(int)); rcvcmsgbuf = (u_char *)malloc(rcvcmsgbuflen); if (rcvcmsgbuf == NULL) { syslog(LOG_ERR, "<%s> not enough core", __func__); exit(1); } - sndcmsgbuflen = CMSG_SPACE(sizeof(struct in6_pktinfo)) + - CMSG_SPACE(sizeof(int)); + sndcmsgbuflen = CMSG_SPACE(sizeof(struct in6_pktinfo)) + + CMSG_SPACE(sizeof(int)); sndcmsgbuf = (u_char *)malloc(sndcmsgbuflen); if (sndcmsgbuf == NULL) { syslog(LOG_ERR, "<%s> not enough core", __func__); exit(1); } if ((sock = socket(AF_INET6, SOCK_RAW, IPPROTO_ICMPV6)) < 0) { - syslog(LOG_ERR, "<%s> socket: %s", __func__, - strerror(errno)); + syslog(LOG_ERR, "<%s> socket: %s", __func__, strerror(errno)); exit(1); } - /* specify to tell receiving interface */ on = 1; -#ifdef IPV6_RECVPKTINFO if (setsockopt(sock, IPPROTO_IPV6, IPV6_RECVPKTINFO, &on, - sizeof(on)) < 0) { - syslog(LOG_ERR, "<%s> IPV6_RECVPKTINFO: %s", - __func__, strerror(errno)); + sizeof(on)) < 0) { + syslog(LOG_ERR, "<%s> IPV6_RECVPKTINFO: %s", __func__, + strerror(errno)); exit(1); } -#else /* old adv. API */ - if (setsockopt(sock, IPPROTO_IPV6, IPV6_PKTINFO, &on, - sizeof(on)) < 0) { - syslog(LOG_ERR, "<%s> IPV6_PKTINFO: %s", - __func__, strerror(errno)); - exit(1); - } -#endif - on = 1; /* specify to tell value of hoplimit field of received IP6 hdr */ -#ifdef IPV6_RECVHOPLIMIT if (setsockopt(sock, IPPROTO_IPV6, IPV6_RECVHOPLIMIT, &on, - sizeof(on)) < 0) { - syslog(LOG_ERR, "<%s> IPV6_RECVHOPLIMIT: %s", - __func__, strerror(errno)); + sizeof(on)) < 0) { + syslog(LOG_ERR, "<%s> IPV6_RECVHOPLIMIT: %s", __func__, + strerror(errno)); exit(1); } -#else /* old adv. API */ - if (setsockopt(sock, IPPROTO_IPV6, IPV6_HOPLIMIT, &on, - sizeof(on)) < 0) { - syslog(LOG_ERR, "<%s> IPV6_HOPLIMIT: %s", - __func__, strerror(errno)); - exit(1); - } -#endif - ICMP6_FILTER_SETBLOCKALL(&filt); ICMP6_FILTER_SETPASS(ND_ROUTER_SOLICIT, &filt); ICMP6_FILTER_SETPASS(ND_ROUTER_ADVERT, &filt); if (accept_rr) ICMP6_FILTER_SETPASS(ICMP6_ROUTER_RENUMBERING, &filt); + if (setsockopt(sock, IPPROTO_ICMPV6, ICMP6_FILTER, &filt, - sizeof(filt)) < 0) { + sizeof(filt)) < 0) { syslog(LOG_ERR, "<%s> IICMP6_FILTER: %s", - __func__, strerror(errno)); + __func__, strerror(errno)); exit(1); } /* * join all routers multicast address on each advertising interface. */ - if (inet_pton(AF_INET6, ALLROUTERS_LINK, - &mreq.ipv6mr_multiaddr.s6_addr) - != 1) { - syslog(LOG_ERR, "<%s> inet_pton failed(library bug?)", - __func__); - exit(1); - } - while (ra) { - mreq.ipv6mr_interface = ra->ifindex; + memcpy(&mreq.ipv6mr_multiaddr.s6_addr, + &sin6_linklocal_allrouters.sin6_addr, + sizeof(mreq.ipv6mr_multiaddr.s6_addr)); + TAILQ_FOREACH(rai, &railist, rai_next) { + mreq.ipv6mr_interface = rai->rai_ifindex; if (setsockopt(sock, IPPROTO_IPV6, IPV6_JOIN_GROUP, &mreq, - sizeof(mreq)) < 0) { + sizeof(mreq)) < 0) { syslog(LOG_ERR, "<%s> IPV6_JOIN_GROUP(link) on %s: %s", - __func__, ra->ifname, strerror(errno)); + __func__, rai->rai_ifname, strerror(errno)); exit(1); } - ra = ra->next; } /* * When attending router renumbering, join all-routers site-local - * multicast group. + * multicast group. */ if (accept_rr) { - if (inet_pton(AF_INET6, ALLROUTERS_SITE, - &in6a_site_allrouters) != 1) { - syslog(LOG_ERR, "<%s> inet_pton failed(library bug?)", - __func__); - exit(1); - } - mreq.ipv6mr_multiaddr = in6a_site_allrouters; + memcpy(&mreq.ipv6mr_multiaddr.s6_addr, + &sin6_sitelocal_allrouters.sin6_addr, + sizeof(mreq.ipv6mr_multiaddr.s6_addr)); if (mcastif) { if ((mreq.ipv6mr_interface = if_nametoindex(mcastif)) == 0) { syslog(LOG_ERR, - "<%s> invalid interface: %s", - __func__, mcastif); + "<%s> invalid interface: %s", + __func__, mcastif); exit(1); } } else - mreq.ipv6mr_interface = ralist->ifindex; + mreq.ipv6mr_interface = + TAILQ_FIRST(&railist)->rai_ifindex; if (setsockopt(sock, IPPROTO_IPV6, IPV6_JOIN_GROUP, - &mreq, sizeof(mreq)) < 0) { + &mreq, sizeof(mreq)) < 0) { syslog(LOG_ERR, - "<%s> IPV6_JOIN_GROUP(site) on %s: %s", - __func__, - mcastif ? mcastif : ralist->ifname, - strerror(errno)); + "<%s> IPV6_JOIN_GROUP(site) on %s: %s", __func__, + mcastif ? mcastif : + TAILQ_FIRST(&railist)->rai_ifname, + strerror(errno)); exit(1); } } - + /* initialize msghdr for receiving packets */ rcviov[0].iov_base = (caddr_t)answer; rcviov[0].iov_len = sizeof(answer); rcvmhdr.msg_name = (caddr_t)&rcvfrom; rcvmhdr.msg_namelen = sizeof(rcvfrom); rcvmhdr.msg_iov = rcviov; rcvmhdr.msg_iovlen = 1; rcvmhdr.msg_control = (caddr_t) rcvcmsgbuf; rcvmhdr.msg_controllen = rcvcmsgbuflen; /* initialize msghdr for sending packets */ sndmhdr.msg_namelen = sizeof(struct sockaddr_in6); sndmhdr.msg_iov = sndiov; sndmhdr.msg_iovlen = 1; sndmhdr.msg_control = (caddr_t)sndcmsgbuf; sndmhdr.msg_controllen = sndcmsgbuflen; - + return; } /* open a routing socket to watch the routing table */ static void -rtsock_open() +rtsock_open(void) { if ((rtsock = socket(PF_ROUTE, SOCK_RAW, 0)) < 0) { syslog(LOG_ERR, - "<%s> socket: %s", __func__, strerror(errno)); + "<%s> socket: %s", __func__, strerror(errno)); exit(1); } } struct rainfo * if_indextorainfo(int idx) { - struct rainfo *rai = ralist; + struct rainfo *rai; - for (rai = ralist; rai; rai = rai->next) { - if (rai->ifindex == idx) - return(rai); + TAILQ_FOREACH(rai, &railist, rai_next) { + syslog(LOG_DEBUG, "<%s> rai->rai_ifindex %d == idx %d?", + __func__, rai->rai_ifindex, idx); + if (rai->rai_ifindex == idx) + return (rai); } - return(NULL); /* search failed */ + return (NULL); /* search failed */ } static void -ra_output(rainfo) -struct rainfo *rainfo; +ra_output(struct rainfo *rai) { int i; struct cmsghdr *cm; struct in6_pktinfo *pi; - struct soliciter *sol, *nextsol; + struct soliciter *sol; - if ((iflist[rainfo->ifindex]->ifm_flags & IFF_UP) == 0) { + if ((iflist[rai->rai_ifindex]->ifm_flags & IFF_UP) == 0) { syslog(LOG_DEBUG, "<%s> %s is not up, skip sending RA", - __func__, rainfo->ifname); + __func__, rai->rai_ifname); return; } - make_packet(rainfo); /* XXX: inefficient */ + /* + * Check lifetime, ACCEPT_RTADV flag, and ip6.forwarding. + * + * (lifetime == 0) = output + * (lifetime != 0 && (ACCEPT_RTADV || !ip6.forwarding) = no output + * + * Basically, hosts MUST NOT send Router Advertisement + * messages at any time (RFC 4861, Section 6.2.3). However, it + * would sometimes be useful to allow hosts to advertise some + * parameters such as prefix information and link MTU. Thus, + * we allow hosts to invoke rtadvd only when router lifetime + * (on every advertising interface) is explicitly set + * zero. (see also the above section) + */ + syslog(LOG_DEBUG, + "<%s> check lifetime=%d, ACCEPT_RTADV=%d, ip6.forwarding=%d on %s", + __func__, rai->rai_lifetime, check_accept_rtadv(rai->rai_ifindex), + getinet6sysctl(IPV6CTL_FORWARDING), rai->rai_ifname); + if (rai->rai_lifetime != 0) { + if (check_accept_rtadv(rai->rai_ifindex)) { + syslog(LOG_INFO, + "<%s> non-zero lifetime RA " + "on RA receiving interface %s." + " Ignored.", __func__, rai->rai_ifname); + return; + } + if (getinet6sysctl(IPV6CTL_FORWARDING) == 0) { + syslog(LOG_INFO, + "<%s> non-zero lifetime RA " + "but net.inet6.ip6.forwarding=0. " + "Ignored.", __func__); + return; + } + } - sndmhdr.msg_name = (caddr_t)&sin6_allnodes; - sndmhdr.msg_iov[0].iov_base = (caddr_t)rainfo->ra_data; - sndmhdr.msg_iov[0].iov_len = rainfo->ra_datalen; + make_packet(rai); /* XXX: inefficient */ + sndmhdr.msg_name = (caddr_t)&sin6_linklocal_allnodes; + sndmhdr.msg_iov[0].iov_base = (caddr_t)rai->rai_ra_data; + sndmhdr.msg_iov[0].iov_len = rai->rai_ra_datalen; + cm = CMSG_FIRSTHDR(&sndmhdr); /* specify the outgoing interface */ cm->cmsg_level = IPPROTO_IPV6; cm->cmsg_type = IPV6_PKTINFO; cm->cmsg_len = CMSG_LEN(sizeof(struct in6_pktinfo)); pi = (struct in6_pktinfo *)CMSG_DATA(cm); memset(&pi->ipi6_addr, 0, sizeof(pi->ipi6_addr)); /*XXX*/ - pi->ipi6_ifindex = rainfo->ifindex; + pi->ipi6_ifindex = rai->rai_ifindex; /* specify the hop limit of the packet */ { int hoplimit = 255; cm = CMSG_NXTHDR(&sndmhdr, cm); cm->cmsg_level = IPPROTO_IPV6; cm->cmsg_type = IPV6_HOPLIMIT; cm->cmsg_len = CMSG_LEN(sizeof(int)); memcpy(CMSG_DATA(cm), &hoplimit, sizeof(int)); } syslog(LOG_DEBUG, - "<%s> send RA on %s, # of waitings = %d", - __func__, rainfo->ifname, rainfo->waiting); + "<%s> send RA on %s, # of waitings = %d", + __func__, rai->rai_ifname, rai->rai_waiting); i = sendmsg(sock, &sndmhdr, 0); - if (i < 0 || i != rainfo->ra_datalen) { + if (i < 0 || (size_t)i != rai->rai_ra_datalen) { if (i < 0) { syslog(LOG_ERR, "<%s> sendmsg on %s: %s", - __func__, rainfo->ifname, - strerror(errno)); + __func__, rai->rai_ifname, + strerror(errno)); } } /* update counter */ - if (rainfo->initcounter < MAX_INITIAL_RTR_ADVERTISEMENTS) - rainfo->initcounter++; - rainfo->raoutput++; + if (rai->rai_initcounter < MAX_INITIAL_RTR_ADVERTISEMENTS) + rai->rai_initcounter++; + rai->rai_raoutput++; /* * unicast advertisements * XXX commented out. reason: though spec does not forbit it, unicast * advert does not really help */ - for (sol = rainfo->soliciter; sol; sol = nextsol) { - nextsol = sol->next; - - sol->next = NULL; + while ((sol = TAILQ_FIRST(&rai->rai_soliciter)) != NULL) { + TAILQ_REMOVE(&rai->rai_soliciter, sol, sol_next); free(sol); } - rainfo->soliciter = NULL; /* update timestamp */ - gettimeofday(&rainfo->lastsent, NULL); + gettimeofday(&rai->rai_lastsent, NULL); /* reset waiting conter */ - rainfo->waiting = 0; + rai->rai_waiting = 0; } /* process RA timer */ struct rtadvd_timer * -ra_timeout(void *data) +ra_timeout(void *arg) { - struct rainfo *rai = (struct rainfo *)data; + struct rainfo *rai; #ifdef notyet /* if necessary, reconstruct the packet. */ #endif + rai = (struct rainfo *)arg; + syslog(LOG_DEBUG, "<%s> RA timer on %s is expired", + __func__, rai->rai_ifname); - syslog(LOG_DEBUG, - "<%s> RA timer on %s is expired", - __func__, rai->ifname); - ra_output(rai); - return(rai->timer); + return (rai->rai_timer); } /* update RA timer */ void -ra_timer_update(void *data, struct timeval *tm) +ra_timer_update(void *arg, struct timeval *tm) { - struct rainfo *rai = (struct rainfo *)data; long interval; + struct rainfo *rai; + rai = (struct rainfo *)arg; /* * Whenever a multicast advertisement is sent from an interface, * the timer is reset to a uniformly-distributed random value * between the interface's configured MinRtrAdvInterval and * MaxRtrAdvInterval (RFC2461 6.2.4). */ - interval = rai->mininterval; + interval = rai->rai_mininterval; #ifdef HAVE_ARC4RANDOM - interval += arc4random_uniform(rai->maxinterval - rai->mininterval); + interval += arc4random_uniform(rai->rai_maxinterval - + rai->rai_mininterval); #else - interval += random() % (rai->maxinterval - rai->mininterval); + interval += random() % (rai->rai_maxinterval - + rai->rai_mininterval); #endif /* * For the first few advertisements (up to * MAX_INITIAL_RTR_ADVERTISEMENTS), if the randomly chosen interval * is greater than MAX_INITIAL_RTR_ADVERT_INTERVAL, the timer * SHOULD be set to MAX_INITIAL_RTR_ADVERT_INTERVAL instead. - * (RFC-2461 6.2.4) + * (RFC 4861 6.2.4) */ - if (rai->initcounter < MAX_INITIAL_RTR_ADVERTISEMENTS && + if (rai->rai_initcounter < MAX_INITIAL_RTR_ADVERTISEMENTS && interval > MAX_INITIAL_RTR_ADVERT_INTERVAL) interval = MAX_INITIAL_RTR_ADVERT_INTERVAL; tm->tv_sec = interval; tm->tv_usec = 0; syslog(LOG_DEBUG, - "<%s> RA timer on %s is set to %ld:%ld", - __func__, rai->ifname, - (long int)tm->tv_sec, (long int)tm->tv_usec); + "<%s> RA timer on %s is set to %ld:%ld", + __func__, rai->rai_ifname, + (long int)tm->tv_sec, (long int)tm->tv_usec); return; } Index: projects/largeSMP/usr.sbin/rtadvd/rtadvd.conf =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/rtadvd.conf (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/rtadvd.conf (revision 222783) @@ -1,21 +1,22 @@ # $FreeBSD$ # $KAME: rtadvd.conf,v 1.13 2003/06/25 03:45:21 itojun Exp $ # # Note: All of the following parameters have default values defined # in specifications, and hence you usually do not have to set them # by hand unless you need special non-default values. # # You even do not need to create the configuration file. rtadvd # would usually work well without a configuration file. # See also: rtadvd(8) # per-interface definitions. # Mainly IPv6 prefixes are configured in this part. However, rtadvd # automatically learns appropriate prefixes from the kernel's routing # table, and advertises the prefixes, so you don't have to configure # this part, either. # If you don't want the automatic advertisement, (uncomment and) configure # this part by hand, and then invoke rtadvd with the -s option. #ef0:\ -# :addr="3ffe:501:ffff:1000::":prefixlen#64: +# :addr="2001:db8:ffff:1000::":prefixlen#64:\ +# :rdnss="2001:db8:ffff:1000::1":dnssl="example.com": Index: projects/largeSMP/usr.sbin/rtadvd/rtadvd.conf.5 =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/rtadvd.conf.5 (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/rtadvd.conf.5 (revision 222783) @@ -1,427 +1,530 @@ .\" $KAME: rtadvd.conf.5,v 1.50 2005/01/14 05:30:59 jinmei Exp $ .\" .\" 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. .\" .\" $FreeBSD$ .\" -.Dd May 17, 1998 +.Dd June 4, 2011 .Dt RTADVD.CONF 5 .Os .Sh NAME .Nm rtadvd.conf .Nd config file for router advertisement daemon .Sh DESCRIPTION This file describes how the router advertisement packets must be constructed for each of the interfaces. .Pp As described in .Xr rtadvd 8 , you do not have to set this configuration file up at all, unless you need some special configurations. You may even omit the file as a whole. In such cases, the .Nm rtadvd daemon will automatically configure itself using default values specified in the specification. .Pp It obeys the famous .Xr termcap 5 file format. Each line in the file describes a network interface. Fields are separated by a colon .Pq Sq \&: , and each field contains one capability description. Lines may be concatenated by the .Sq \e character. The comment marker is the .Sq \&# character. .Sh CAPABILITIES Capabilities describe the value to be filled into ICMPv6 router advertisement messages and to control .Xr rtadvd 8 behavior. Therefore, you are encouraged to read IETF neighbor discovery documents if you would like to modify the sample configuration file. .Pp Note that almost all items have default values. If you omit an item, the default value of the item will be used. .Pp There are two items which control the interval of sending router advertisements. These items can be omitted, then .Nm rtadvd will use the default values. .Bl -tag -width indent .It Cm \&maxinterval (num) The maximum time allowed between sending unsolicited multicast router advertisements .Pq unit: seconds . The default value is 600. Its value must be no less than 4 seconds and no greater than 1800 seconds. .It Cm \&mininterval (num) The minimum time allowed between sending unsolicited multicast router advertisements .Pq unit: seconds . The default value is the one third of value of .Cm maxinterval . Its value must be no less than 3 seconds and no greater than .75 * the value of .Cm maxinterval . .El .Pp The following items are for ICMPv6 router advertisement message header. These items can be omitted, then .Nm rtadvd will use the default values. .Bl -tag -width indent .It Cm \&chlim (num) The value for Cur Hop Limit field. The default value is 64. .It Cm \&raflags (str or num) A 8-bit flags field in router advertisement message header. This field can be specified either as a case-sensitive string or as an integer. A string consists of characters each of which corresponds to a particular flag bit(s). An integer should be the logical OR of all enabled bits. Bit 7 .Po .Li 'm' or 0x80 .Pc means Managed address configuration flag bit, and Bit 6 .Po .Li 'o' or 0x40 .Pc means Other stateful configuration flag bit. Bit 4 .Po .Li 0x10 .Pc and Bit 3 .Po .Li 0x08 .Pc are used to encode router preference. Bits 01 .Po or 'h' .Pc means high, 00 means medium, and 11 .Po or 'l' .Pc means low. Bits 10 is reserved, and must not be specified. There is no character to specify the medium preference explicitly. The default value of the entire flag is 0 .Po or a null string, .Pc which means no additional configuration methods, and the medium router preference. .It Cm \&rltime (num) Router lifetime field .Pq unit: seconds . The value must be either zero or between the value of .Cm maxinterval and 9000. When .Nm rtadvd runs on a host, this value must explicitly set 0 on all the advertising interfaces as described in .Xr rtadvd 8 . The default value is 1800. .It Cm \&rtime (num) Reachable time field .Pq unit: milliseconds . The default value is 0, which means unspecified by this router. .It Cm \&retrans (num) Retrans Timer field .Pq unit: milliseconds . The default value is 0, which means unspecified by this router. .El .Pp The following items are for ICMPv6 prefix information option, which will be attached to router advertisement header. These items can be omitted, then .Nm rtadvd will automatically get appropriate prefixes from the kernel's routing table, and advertise the prefixes with the default parameters. Keywords other than .Cm clockskew +and +.Cm noifprefix can be augmented with a number, like .Dq Li prefix2 , to specify multiple prefixes. .Bl -tag -width indent +.It Cm \&noifprefix +(bool) Specifies no prefix on the network interfaces will be advertised. +By default +.Nm rtadvd +automatically gathers on-link prefixes from all of the network interfaces +and advertise them. +The +.Cm noifprefix +disables that behavior. +If this is specified and no +.Cm addr +keyword is specified, no prefix information option will be included in the +message. .It Cm \&clockskew (num) Time skew to adjust link propagation delays and clock skews between routers on the link .Pq unit: seconds . This value is used in consistency check for locally-configured and advertised prefix lifetimes, and has its meaning when the local router configures a prefix on the link with a lifetime that decrements in real time. If the value is 0, it means the consistency check will be skipped for such prefixes. The default value is 0. .It Cm \&prefixlen (num) Prefix length field. The default value is 64. .It Cm \&pinfoflags (str or num) A 8-bit flags field in prefix information option. This field can be specified either as a case-sensitive string or as an integer. A string consists of characters each of which corresponds to a particular flag bit(s). An integer should be the logical OR of all enabled bits. Bit 7 .Po .Li 'l' or 0x80 .Pc means On-link flag bit, and Bit 6 .Po .Li 'a' or 0x40 .Pc means Autonomous address-configuration flag bit. The default value is "la" or 0xc0, i.e., both bits are set. .It Cm \&addr (str) The address filled into Prefix field. Since .Dq \&: is used for .Xr termcap 5 file format as well as IPv6 numeric address, the field MUST be quoted by doublequote character. .It Cm \&vltime (num) Valid lifetime field .Pq unit: seconds . The default value is 2592000 (30 days). .It Cm \&vltimedecr (bool) This item means the advertised valid lifetime will decrement in real time, which is disabled by default. .It Cm \&pltime (num) Preferred lifetime field .Pq unit: seconds . The default value is 604800 (7 days). .It Cm \&pltimedecr (bool) This item means the advertised preferred lifetime will decrement in real time, which is disabled by default. .El .Pp The following item is for ICMPv6 MTU option, which will be attached to router advertisement header. This item can be omitted, then .Nm rtadvd will use the default value. .Bl -tag -width indent .It Cm \&mtu (num or str) MTU (maximum transmission unit) field. If 0 is specified, it means that the option will not be included. The default value is 0. If the special string .Dq auto is specified for this item, MTU option will be included and its value will be set to the interface MTU automatically. .El .Pp The following item controls ICMPv6 source link-layer address option, which will be attached to router advertisement header. As noted above, you can just omit the item, then .Nm rtadvd will use the default value. .Bl -tag -width indent .It Cm \&nolladdr (bool) By default .Po if .Cm \&nolladdr is not specified .Pc , .Xr rtadvd 8 will try to get link-layer address for the interface from the kernel, and attach that in source link-layer address option. If this capability exists, .Xr rtadvd 8 will not attach source link-layer address option to router advertisement packets. .El .Pp The following item controls ICMPv6 home agent information option, which was defined with mobile IPv6 support. It will be attached to router advertisement header just like other options do. .Bl -tag -width indent .It Cm \&hapref (num) Specifies home agent preference. If set to non-zero, .Cm \&hatime must be present as well. .It Cm \&hatime (num) Specifies home agent lifetime. .El .Pp When mobile IPv6 support is turned on for .Xr rtadvd 8 , advertisement interval option will be attached to router advertisement packet, by configuring .Cm \&maxinterval explicitly. .Pp The following items are for ICMPv6 route information option, which will be attached to router advertisement header. These items are optional. Each items can be augmented with number, like .Dq Li rtplen2 , to specify multiple routes. .Bl -tag -width indent .It Cm \&rtprefix (str) The prefix filled into the Prefix field of route information option. Since .Dq \&: is used for .Xr termcap 5 file format as well as IPv6 numeric address, the field MUST be quoted by doublequote character. .It Cm \&rtplen (num) Prefix length field in route information option. The default value is 64. .It Cm \&rtflags (str or num) A 8-bit flags field in route information option. Currently only the preference values are defined. The notation is same as that of the raflags field. Bit 4 .Po .Li 0x10 .Pc and Bit 3 .Po .Li 0x08 .Pc are used to encode the route preference for the route. The default value is 0x00, i.e., medium preference. .It Cm \&rtltime (num) route lifetime field in route information option. .Pq unit: seconds . Since the specification does not define the default value of this item, the value for this item should be specified by hand. However, .Nm rtadvd allows this item to be unspecified, and uses the router lifetime as the default value in such a case, just for compatibility with an old version of the program. .El .Pp In the above list, each keyword beginning with .Dq Li rt could be replaced with the one beginning with .Dq Li rtr for backward compatibility reason. For example, .Cm rtrplen is accepted instead of .Cm rtplen . However, keywords that start with .Dq Li rtr have basically been obsoleted, and should not be used any more. .Pp +The following items are for ICMPv6 Recursive DNS Server Option and +DNS Search List Option +.Pq RFC 6106 , +which will be attached to router advertisement header. +These items are optional. +.Bl -tag -width indent +.It Cm \&rdnss +(str) The IPv6 address of one or more recursive DNS servers. +The argument must be inside double quotes. +Multiple DNS servers can be specified in a comma-separated string. +If different lifetimes are needed for different servers, +separate entries can be given by using +.Cm rdnss , +.Cm rdnss0 , +.Cm rdnss1 , +.Cm rdnss2 ... +options with corresponding +.Cm rdnssltime , +.Cm rdnssltime0 , +.Cm rdnssltime1 , +.Cm rdnssltime2 ... +entries. +Note that the maximum number of servers depends on the receiver side. +See also +.Xr resolver 5 +manual page for resolver implementation in +.Fx . +.It Cm \&rdnssltime +The lifetime of the +.Cm rdnss +DNS server entries. +The default value is 3/2 of the interval time. +.It Cm \&dnssl +(str) One or more domain names in a comma-separated string. +These domain names will be used when making DNS queries on a +non-fully-qualified domain name. +If different lifetimes are needed for different domains, separate entries +can be given by using +.Cm dnssl , +.Cm dnssl0 , +.Cm dnssl1 , +.Cm dnssl2 ... +options with corresponding +.Cm dnsslltime , +.Cm dnsslltime0 , +.Cm dnsslltime1 , +.Cm dnsslltime2 ... +entries. +Note that the maximum number of names depends on the receiver side. +See also +.Xr resolver 5 +manual page for resolver implementation in +.Fx . +.It Cm \&dnsslltime +The lifetime of the +.Cm dnssl +DNS search list entries. +The default value is 3/2 of the interval time. +.El +.Pp You can also refer one line from another by using .Cm tc capability. See .Xr termcap 5 for details on the capability. .Sh EXAMPLES As presented above, all of the advertised parameters have default values defined in specifications, and hence you usually do not have to set them by hand, unless you need special non-default values. It can cause interoperability problem if you use an ill-configured parameter. .Pp To override a configuration parameter, you can specify the parameter alone. With the following configuration, .Xr rtadvd 8 overrides the router lifetime parameter for the .Li ne0 interface. .Bd -literal -offset ne0:\\ :rltime#0: .Ed .Pp The following example manually configures prefixes advertised from the .Li ef0 interface. The configuration must be used with the .Fl s option to .Xr rtadvd 8 . .Bd -literal -offset ef0:\\ - :addr="3ffe:501:ffff:1000::":prefixlen#64: + :addr="2001:db8:ffff:1000::":prefixlen#64: .Ed .Pp +The following example configures the +.Li wlan0 +interface and adds two DNS servers and a DNS domain search options +using the default option lifetime values. +.Bd -literal -offset +wlan0:\\ + :addr="2001:db8:ffff:1000::":prefixlen#64:\\ + :rdnss="2001:db8:ffff::10,2001:db8:ffff::2:43":\\ + :dnssl="example.com": +.Ed +.Pp The following example presents the default values in an explicit manner. The configuration is provided just for reference purposes; YOU DO NOT NEED TO HAVE IT AT ALL. .Bd -literal -offset default:\\ :chlim#64:raflags#0:rltime#1800:rtime#0:retrans#0:\\ :pinfoflags="la":vltime#2592000:pltime#604800:mtu#0: ef0:\\ - :addr="3ffe:501:ffff:1000::":prefixlen#64:tc=default: + :addr="2001:db8:ffff:1000::":prefixlen#64:tc=default: .Ed .Sh SEE ALSO .Xr termcap 5 , +.Xr resolver 5 , .Xr rtadvd 8 , .Xr rtsol 8 .Rs .%A Thomas Narten .%A Erik Nordmark .%A W. A. Simpson +.%A Hesham Soliman .%T Neighbor Discovery for IP version 6 (IPv6) -.%R RFC 2461 +.%R RFC 4861 .Re .Rs +.%A Thomas Narten +.%A Erik Nordmark +.%A W. A. Simpson +.%T Neighbor Discovery for IP version 6 (IPv6) +.%R RFC 2461 (obsoleted by RFC 4861) +.Re +.Rs .%A Richard Draves .%T Default Router Preferences and More-Specific Routes .%R draft-ietf-ipngwg-router-selection-xx.txt +.Re +.Rs +.%A J. Jeong +.%A S. Park +.%A L. Beloeil +.%A S. Madanapalli +.%T IPv6 Router Advertisement Options for DNS Configuration +.%R RFC 6106 .Re .Sh HISTORY The .Xr rtadvd 8 and the configuration file .Nm first appeared in WIDE Hydrangea IPv6 protocol stack kit. .\" .Sh BUGS .\" (to be written) Index: projects/largeSMP/usr.sbin/rtadvd/rtadvd.h =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/rtadvd.h (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/rtadvd.h (revision 222783) @@ -1,161 +1,241 @@ /* $FreeBSD$ */ /* $KAME: rtadvd.h,v 1.26 2003/08/05 12:34:23 itojun Exp $ */ /* * Copyright (C) 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. */ -#define ALLNODES "ff02::1" -#define ALLROUTERS_LINK "ff02::2" -#define ALLROUTERS_SITE "ff05::2" -#define ANY "::" -#define RTSOLLEN 8 +#define IN6ADDR_LINKLOCAL_ALLNODES_INIT \ + {{{ 0xff, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, \ + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01 }}} +#define IN6ADDR_LINKLOCAL_ALLROUTERS_INIT \ + {{{ 0xff, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, \ + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02 }}} + +#define IN6ADDR_SITELOCAL_ALLROUTERS_INIT \ + {{{ 0xff, 0x05, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, \ + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02 }}} + +extern struct sockaddr_in6 sin6_linklocal_allnodes; +extern struct sockaddr_in6 sin6_linklocal_allrouters; +extern struct sockaddr_in6 sin6_sitelocal_allrouters; + +/* + * RFC 3542 API deprecates IPV6_PKTINFO in favor of + * IPV6_RECVPKTINFO + */ +#ifndef IPV6_RECVPKTINFO +#ifdef IPV6_PKTINFO +#define IPV6_RECVPKTINFO IPV6_PKTINFO +#endif +#endif + +/* + * RFC 3542 API deprecates IPV6_HOPLIMIT in favor of + * IPV6_RECVHOPLIMIT + */ +#ifndef IPV6_RECVHOPLIMIT +#ifdef IPV6_HOPLIMIT +#define IPV6_RECVHOPLIMIT IPV6_HOPLIMIT +#endif +#endif + /* protocol constants and default values */ #define DEF_MAXRTRADVINTERVAL 600 #define DEF_ADVLINKMTU 0 #define DEF_ADVREACHABLETIME 0 #define DEF_ADVRETRANSTIMER 0 #define DEF_ADVCURHOPLIMIT 64 #define DEF_ADVVALIDLIFETIME 2592000 #define DEF_ADVPREFERREDLIFETIME 604800 #define MAXROUTERLIFETIME 9000 #define MIN_MAXINTERVAL 4 #define MAX_MAXINTERVAL 1800 #define MIN_MININTERVAL 3 #define MAXREACHABLETIME 3600000 #define MAX_INITIAL_RTR_ADVERT_INTERVAL 16 #define MAX_INITIAL_RTR_ADVERTISEMENTS 3 #define MAX_FINAL_RTR_ADVERTISEMENTS 3 #define MIN_DELAY_BETWEEN_RAS 3 #define MAX_RA_DELAY_TIME 500000 /* usec */ #define PREFIX_FROM_KERNEL 1 #define PREFIX_FROM_CONFIG 2 #define PREFIX_FROM_DYNAMIC 3 struct prefix { - struct prefix *next; /* forward link */ - struct prefix *prev; /* previous link */ + TAILQ_ENTRY(prefix) pfx_next; - struct rainfo *rainfo; /* back pointer to the interface */ + struct rainfo *pfx_rainfo; /* back pointer to the interface */ + /* + * Expiration timer. This is used when a prefix derived from + * the kernel is deleted. + */ + struct rtadvd_timer *pfx_timer; - struct rtadvd_timer *timer; /* expiration timer. used when a prefix - * derived from the kernel is deleted. - */ + u_int32_t pfx_validlifetime; /* AdvValidLifetime */ + long pfx_vltimeexpire; /* Expiration of vltime */ + u_int32_t pfx_preflifetime; /* AdvPreferredLifetime */ + long pfx_pltimeexpire; /* Expiration of pltime */ + u_int pfx_onlinkflg; /* bool: AdvOnLinkFlag */ + u_int pfx_autoconfflg; /* bool: AdvAutonomousFlag */ + int pfx_prefixlen; + int pfx_origin; /* From kernel or config */ - u_int32_t validlifetime; /* AdvValidLifetime */ - long vltimeexpire; /* expiration of vltime; decrement case only */ - u_int32_t preflifetime; /* AdvPreferredLifetime */ - long pltimeexpire; /* expiration of pltime; decrement case only */ - u_int onlinkflg; /* bool: AdvOnLinkFlag */ - u_int autoconfflg; /* bool: AdvAutonomousFlag */ - int prefixlen; - int origin; /* from kernel or config */ - struct in6_addr prefix; + struct in6_addr pfx_prefix; }; #ifdef ROUTEINFO struct rtinfo { - struct rtinfo *prev; /* previous link */ - struct rtinfo *next; /* forward link */ + TAILQ_ENTRY(rtinfo) rti_next; - u_int32_t ltime; /* route lifetime */ - u_int rtpref; /* route preference */ - int prefixlen; - struct in6_addr prefix; + u_int32_t rti_ltime; /* route lifetime */ + u_int rti_rtpref; /* route preference */ + int rti_prefixlen; + struct in6_addr rti_prefix; }; #endif +struct rdnss_addr { + TAILQ_ENTRY(rdnss_addr) ra_next; + + struct in6_addr ra_dns; /* DNS server entry */ +}; + +struct rdnss { + TAILQ_ENTRY(rdnss) rd_next; + + TAILQ_HEAD(, rdnss_addr) rd_list; /* list of DNS servers */ + int rd_cnt; /* number of DNS servers */ + u_int32_t rd_ltime; /* number of seconds valid */ +}; + +/* + * The maximum length of a domain name in a DNS search list is calculated + * by a domain name + length fields per 63 octets + a zero octet at + * the tail and adding 8 octet boundary padding. + */ +#define _DNAME_LABELENC_MAXLEN \ + (NI_MAXHOST + (NI_MAXHOST / 64 + 1) + 1) + +#define DNAME_LABELENC_MAXLEN \ + (_DNAME_LABELENC_MAXLEN + 8 - _DNAME_LABELENC_MAXLEN % 8) + +struct dnssl_addr { + TAILQ_ENTRY(dnssl_addr) da_next; + + int da_len; /* length of entry */ + char da_dom[DNAME_LABELENC_MAXLEN]; /* search domain name entry */ +}; + +struct dnssl { + TAILQ_ENTRY(dnssl) dn_next; + + TAILQ_HEAD(, dnssl_addr) dn_list; /* list of search domains */ + u_int32_t dn_ltime; /* number of seconds valid */ +}; + struct soliciter { - struct soliciter *next; - struct sockaddr_in6 addr; + TAILQ_ENTRY(soliciter) sol_next; + + struct sockaddr_in6 sol_addr; }; struct rainfo { /* pointer for list */ - struct rainfo *next; + TAILQ_ENTRY(rainfo) rai_next; /* timer related parameters */ - struct rtadvd_timer *timer; - int initcounter; /* counter for the first few advertisements */ - struct timeval lastsent; /* timestamp when the latest RA was sent */ - int waiting; /* number of RS waiting for RA */ + struct rtadvd_timer *rai_timer; + /* counter for the first few advertisements */ + int rai_initcounter; + /* timestamp when the latest RA was sent */ + struct timeval rai_lastsent; + /* number of RS waiting for RA */ + int rai_waiting; /* interface information */ - int ifindex; - int advlinkopt; /* bool: whether include link-layer addr opt */ - struct sockaddr_dl *sdl; - char ifname[16]; - int phymtu; /* mtu of the physical interface */ + int rai_ifindex; + int rai_advlinkopt; /* bool: whether include link-layer addr opt */ + int rai_advifprefix; /* bool: gather IF prefixes? */ + struct sockaddr_dl *rai_sdl; + char rai_ifname[IFNAMSIZ]; + u_int32_t rai_phymtu; /* mtu of the physical interface */ /* Router configuration variables */ - u_short lifetime; /* AdvDefaultLifetime */ - u_int maxinterval; /* MaxRtrAdvInterval */ - u_int mininterval; /* MinRtrAdvInterval */ - int managedflg; /* AdvManagedFlag */ - int otherflg; /* AdvOtherConfigFlag */ + u_short rai_lifetime; /* AdvDefaultLifetime */ + u_int rai_maxinterval; /* MaxRtrAdvInterval */ + u_int rai_mininterval; /* MinRtrAdvInterval */ + int rai_managedflg; /* AdvManagedFlag */ + int rai_otherflg; /* AdvOtherConfigFlag */ - int rtpref; /* router preference */ - u_int32_t linkmtu; /* AdvLinkMTU */ - u_int32_t reachabletime; /* AdvReachableTime */ - u_int32_t retranstimer; /* AdvRetransTimer */ - u_int hoplimit; /* AdvCurHopLimit */ - struct prefix prefix; /* AdvPrefixList(link head) */ - int pfxs; /* number of prefixes */ - long clockskew; /* used for consisitency check of lifetimes */ + int rai_rtpref; /* router preference */ + u_int32_t rai_linkmtu; /* AdvLinkMTU */ + u_int32_t rai_reachabletime; /* AdvReachableTime */ + u_int32_t rai_retranstimer; /* AdvRetransTimer */ + u_int rai_hoplimit; /* AdvCurHopLimit */ + TAILQ_HEAD(, prefix) rai_prefix;/* AdvPrefixList(link head) */ + int rai_pfxs; /* number of prefixes */ + + long rai_clockskew; /* used for consisitency check of lifetimes */ + + TAILQ_HEAD(, rdnss) rai_rdnss; /* DNS server list */ + TAILQ_HEAD(, dnssl) rai_dnssl; /* search domain list */ #ifdef ROUTEINFO - struct rtinfo route; /* route information option (link head) */ - int routes; /* number of route information options */ + TAILQ_HEAD(, rtinfo) rai_route; /* route information option (link head) */ + int rai_routes; /* number of route information options */ #endif - /* actual RA packet data and its length */ - size_t ra_datalen; - u_char *ra_data; + size_t rai_ra_datalen; + u_char *rai_ra_data; /* statistics */ - u_quad_t raoutput; /* number of RAs sent */ - u_quad_t rainput; /* number of RAs received */ - u_quad_t rainconsistent; /* number of RAs inconsistent with ours */ - u_quad_t rsinput; /* number of RSs received */ + u_quad_t rai_raoutput; /* # of RAs sent */ + u_quad_t rai_rainput; /* # of RAs received */ + u_quad_t rai_rainconsistent; /* # of RAs inconsistent with ours */ + u_quad_t rai_rsinput; /* # of RSs received */ /* info about soliciter */ - struct soliciter *soliciter; /* recent solication source */ + TAILQ_HEAD(, soliciter) rai_soliciter; /* recent solication source */ }; -struct rtadvd_timer *ra_timeout(void *); -void ra_timer_update(void *, struct timeval *); +/* Interface list including RA information */ +extern TAILQ_HEAD(railist_head_t, rainfo) railist; -int prefix_match(struct in6_addr *, int, struct in6_addr *, int); -struct rainfo *if_indextorainfo(int); -struct prefix *find_prefix(struct rainfo *, struct in6_addr *, int); +struct rtadvd_timer *ra_timeout(void *); +void ra_timer_update(void *, struct timeval *); -extern struct in6_addr in6a_site_allrouters; +int prefix_match(struct in6_addr *, int, + struct in6_addr *, int); +struct rainfo *if_indextorainfo(int); +struct prefix *find_prefix(struct rainfo *, + struct in6_addr *, int); Index: projects/largeSMP/usr.sbin/rtadvd/timer.c =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/timer.c (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/timer.c (revision 222783) @@ -1,209 +1,205 @@ /* $FreeBSD$ */ /* $KAME: timer.c,v 1.9 2002/06/10 19:59:47 itojun Exp $ */ /* * Copyright (C) 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. */ #include +#include #include #include #include #include #include #include "timer.h" -static struct rtadvd_timer timer_head; - #define MILLION 1000000 -#define TIMEVAL_EQUAL(t1,t2) ((t1)->tv_sec == (t2)->tv_sec &&\ - (t1)->tv_usec == (t2)->tv_usec) -static struct timeval tm_max = {0x7fffffff, 0x7fffffff}; +struct rtadvd_timer_head_t ra_timer = + TAILQ_HEAD_INITIALIZER(ra_timer); +static struct timeval tm_limit = {0x7fffffff, 0x7fffffff}; +static struct timeval tm_max; void -rtadvd_timer_init() +rtadvd_timer_init(void) { - memset(&timer_head, 0, sizeof(timer_head)); - timer_head.next = timer_head.prev = &timer_head; - timer_head.tm = tm_max; + tm_max = tm_limit; + TAILQ_INIT(&ra_timer); } struct rtadvd_timer * rtadvd_add_timer(struct rtadvd_timer *(*timeout)(void *), void (*update)(void *, struct timeval *), void *timeodata, void *updatedata) { - struct rtadvd_timer *newtimer; + struct rtadvd_timer *rat; - if ((newtimer = malloc(sizeof(*newtimer))) == NULL) { + if (timeout == NULL) { syslog(LOG_ERR, - "<%s> can't allocate memory", __func__); + "<%s> timeout function unspecified", __func__); exit(1); } - memset(newtimer, 0, sizeof(*newtimer)); - - if (timeout == NULL) { + rat = malloc(sizeof(*rat)); + if (rat == NULL) { syslog(LOG_ERR, - "<%s> timeout function unspecified", __func__); + "<%s> can't allocate memory", __func__); exit(1); } - newtimer->expire = timeout; - newtimer->update = update; - newtimer->expire_data = timeodata; - newtimer->update_data = updatedata; - newtimer->tm = tm_max; + memset(rat, 0, sizeof(*rat)); + rat->rat_expire = timeout; + rat->rat_update = update; + rat->rat_expire_data = timeodata; + rat->rat_update_data = updatedata; + rat->rat_tm = tm_max; + /* link into chain */ - insque(newtimer, &timer_head); + TAILQ_INSERT_TAIL(&ra_timer, rat, rat_next); - return(newtimer); + return (rat); } void -rtadvd_remove_timer(struct rtadvd_timer **timer) +rtadvd_remove_timer(struct rtadvd_timer *rat) { - remque(*timer); - free(*timer); - *timer = NULL; + + if (rat == NULL) + return; + + TAILQ_REMOVE(&ra_timer, rat, rat_next); + free(rat); } void -rtadvd_set_timer(struct timeval *tm, struct rtadvd_timer *timer) +rtadvd_set_timer(struct timeval *tm, struct rtadvd_timer *rat) { struct timeval now; /* reset the timer */ gettimeofday(&now, NULL); + TIMEVAL_ADD(&now, tm, &rat->rat_tm); - TIMEVAL_ADD(&now, tm, &timer->tm); - /* update the next expiration time */ - if (TIMEVAL_LT(timer->tm, timer_head.tm)) - timer_head.tm = timer->tm; + if (TIMEVAL_LT(&rat->rat_tm, &tm_max)) + tm_max = rat->rat_tm; return; } /* * Check expiration for each timer. If a timer expires, * call the expire function for the timer and update the timer. * Return the next interval for select() call. */ struct timeval * -rtadvd_check_timer() +rtadvd_check_timer(void) { static struct timeval returnval; struct timeval now; - struct rtadvd_timer *tm = timer_head.next, *tm_next; + struct rtadvd_timer *rat; gettimeofday(&now, NULL); - - timer_head.tm = tm_max; - - for (tm = timer_head.next; tm != &timer_head; tm = tm_next) { - tm_next = tm->next; - - if (TIMEVAL_LEQ(tm->tm, now)) { - if (((*tm->expire)(tm->expire_data) == NULL)) + tm_max = tm_limit; + TAILQ_FOREACH(rat, &ra_timer, rat_next) { + if (TIMEVAL_LEQ(&rat->rat_tm, &now)) { + if (((*rat->rat_expire)(rat->rat_expire_data) == NULL)) continue; /* the timer was removed */ - if (tm->update) - (*tm->update)(tm->update_data, &tm->tm); - TIMEVAL_ADD(&tm->tm, &now, &tm->tm); + if (rat->rat_update) + (*rat->rat_update)(rat->rat_update_data, &rat->rat_tm); + TIMEVAL_ADD(&rat->rat_tm, &now, &rat->rat_tm); } - - if (TIMEVAL_LT(tm->tm, timer_head.tm)) - timer_head.tm = tm->tm; + if (TIMEVAL_LT(&rat->rat_tm, &tm_max)) + tm_max = rat->rat_tm; } - - if (TIMEVAL_EQUAL(&tm_max, &timer_head.tm)) { + if (TIMEVAL_EQUAL(&tm_max, &tm_limit)) { /* no need to timeout */ - return(NULL); - } else if (TIMEVAL_LT(timer_head.tm, now)) { + return (NULL); + } else if (TIMEVAL_LT(&tm_max, &now)) { /* this may occur when the interval is too small */ returnval.tv_sec = returnval.tv_usec = 0; } else - TIMEVAL_SUB(&timer_head.tm, &now, &returnval); - return(&returnval); + TIMEVAL_SUB(&tm_max, &now, &returnval); + return (&returnval); } struct timeval * -rtadvd_timer_rest(struct rtadvd_timer *timer) +rtadvd_timer_rest(struct rtadvd_timer *rat) { static struct timeval returnval, now; gettimeofday(&now, NULL); - if (TIMEVAL_LEQ(timer->tm, now)) { + if (TIMEVAL_LEQ(&rat->rat_tm, &now)) { syslog(LOG_DEBUG, - "<%s> a timer must be expired, but not yet", - __func__); + "<%s> a timer must be expired, but not yet", + __func__); returnval.tv_sec = returnval.tv_usec = 0; } else - TIMEVAL_SUB(&timer->tm, &now, &returnval); + TIMEVAL_SUB(&rat->rat_tm, &now, &returnval); - return(&returnval); + return (&returnval); } /* result = a + b */ void TIMEVAL_ADD(struct timeval *a, struct timeval *b, struct timeval *result) { long l; if ((l = a->tv_usec + b->tv_usec) < MILLION) { result->tv_usec = l; result->tv_sec = a->tv_sec + b->tv_sec; } else { result->tv_usec = l - MILLION; result->tv_sec = a->tv_sec + b->tv_sec + 1; } } /* * result = a - b * XXX: this function assumes that a >= b. */ void TIMEVAL_SUB(struct timeval *a, struct timeval *b, struct timeval *result) { long l; if ((l = a->tv_usec - b->tv_usec) >= 0) { result->tv_usec = l; result->tv_sec = a->tv_sec - b->tv_sec; } else { result->tv_usec = MILLION + l; result->tv_sec = a->tv_sec - b->tv_sec - 1; } } Index: projects/largeSMP/usr.sbin/rtadvd/timer.h =================================================================== --- projects/largeSMP/usr.sbin/rtadvd/timer.h (revision 222782) +++ projects/largeSMP/usr.sbin/rtadvd/timer.h (revision 222783) @@ -1,65 +1,72 @@ /* $FreeBSD$ */ /* $KAME: timer.h,v 1.5 2002/05/31 13:30:38 jinmei Exp $ */ /* * Copyright (C) 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. */ /* a < b */ -#define TIMEVAL_LT(a, b) (((a).tv_sec < (b).tv_sec) ||\ - (((a).tv_sec == (b).tv_sec) && \ - ((a).tv_usec < (b).tv_usec))) +#define TIMEVAL_LT(a, b) \ + (((a)->tv_sec < (b)->tv_sec) || \ + (((a)->tv_sec == (b)->tv_sec) && \ + ((a)->tv_usec < (b)->tv_usec))) /* a <= b */ -#define TIMEVAL_LEQ(a, b) (((a).tv_sec < (b).tv_sec) ||\ - (((a).tv_sec == (b).tv_sec) &&\ - ((a).tv_usec <= (b).tv_usec))) +#define TIMEVAL_LEQ(a, b) \ + (((a)->tv_sec < (b)->tv_sec) || \ + (((a)->tv_sec == (b)->tv_sec) && \ + ((a)->tv_usec <= (b)->tv_usec))) +#define TIMEVAL_EQUAL(a,b) \ + (((a)->tv_sec == (b)->tv_sec) && \ + ((a)->tv_usec == (b)->tv_usec)) + +extern TAILQ_HEAD(rtadvd_timer_head_t, rtadvd_timer) ra_timer; struct rtadvd_timer { - struct rtadvd_timer *next; - struct rtadvd_timer *prev; - struct rainfo *rai; - struct timeval tm; + TAILQ_ENTRY(rtadvd_timer) rat_next; - struct rtadvd_timer *(*expire)(void *); /* expiration function */ - void *expire_data; - void (*update)(void *, struct timeval *); /* update function */ - void *update_data; + struct rainfo *rat_rai; + struct timeval rat_tm; + struct rtadvd_timer *(*rat_expire)(void *); + void *rat_expire_data; + void (*rat_update)(void *, struct timeval *); + void *rat_update_data; }; -void rtadvd_timer_init(void); -struct rtadvd_timer *rtadvd_add_timer(struct rtadvd_timer *(*)(void *), - void (*)(void *, struct timeval *), void *, void *); -void rtadvd_set_timer(struct timeval *, struct rtadvd_timer *); -void rtadvd_remove_timer(struct rtadvd_timer **); -struct timeval * rtadvd_check_timer(void); -struct timeval * rtadvd_timer_rest(struct rtadvd_timer *); -void TIMEVAL_ADD(struct timeval *, struct timeval *, - struct timeval *); -void TIMEVAL_SUB(struct timeval *, struct timeval *, - struct timeval *); +void rtadvd_timer_init(void); +struct rtadvd_timer *rtadvd_add_timer(struct rtadvd_timer *(*)(void *), + void (*)(void *, struct timeval *), void *, void *); +void rtadvd_set_timer(struct timeval *, + struct rtadvd_timer *); +void rtadvd_remove_timer(struct rtadvd_timer *); +struct timeval *rtadvd_check_timer(void); +struct timeval *rtadvd_timer_rest(struct rtadvd_timer *); +void TIMEVAL_ADD(struct timeval *, struct timeval *, + struct timeval *); +void TIMEVAL_SUB(struct timeval *, struct timeval *, + struct timeval *); Index: projects/largeSMP/usr.sbin/rtsold/Makefile =================================================================== --- projects/largeSMP/usr.sbin/rtsold/Makefile (revision 222782) +++ projects/largeSMP/usr.sbin/rtsold/Makefile (revision 222783) @@ -1,27 +1,26 @@ # Copyright (c) 1996 WIDE Project. All rights reserved. # # Redistribution and use in source and binary forms, with or without # modifications, are permitted provided that the above copyright notice # and this paragraph are duplicated in all such forms and that any # documentation, advertising materials, and other materials related to # such distribution and use acknowledge that the software was developed # by the WIDE Project, Japan. The name of the Project may not be used to # endorse or promote products derived from this software without # specific prior written permission. THIS SOFTWARE IS PROVIDED ``AS IS'' # AND WITHOUT ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, WITHOUT # LIMITATION, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR # A PARTICULAR PURPOSE. # # $FreeBSD$ PROG= rtsold MAN= rtsold.8 MLINKS= rtsold.8 rtsol.8 SRCS= rtsold.c rtsol.c if.c probe.c dump.c rtsock.c -WARNS?= 3 CFLAGS+= -DHAVE_ARC4RANDOM -DHAVE_POLL_H DPADD= ${LIBKVM} LDADD= -lkvm .include Index: projects/largeSMP/usr.sbin/rtsold/dump.c =================================================================== --- projects/largeSMP/usr.sbin/rtsold/dump.c (revision 222782) +++ projects/largeSMP/usr.sbin/rtsold/dump.c (revision 222783) @@ -1,149 +1,151 @@ /* $KAME: dump.c,v 1.13 2003/10/05 00:09:36 itojun Exp $ */ /* * Copyright (C) 1999 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. * * $FreeBSD$ */ #include #include #include +#include #include #include #include #include #include #include #include #include #include "rtsold.h" static FILE *fp; extern struct ifinfo *iflist; static void dump_interface_status(void); static const char *sec2str(time_t); static const char * const ifstatstr[] = {"IDLE", "DELAY", "PROBE", "DOWN", "TENTATIVE"}; static void dump_interface_status(void) { - struct ifinfo *ifinfo; + struct ifinfo *ifi; struct timeval now; gettimeofday(&now, NULL); - for (ifinfo = iflist; ifinfo; ifinfo = ifinfo->next) { - fprintf(fp, "Interface %s\n", ifinfo->ifname); + TAILQ_FOREACH(ifi, &ifinfo_head, ifi_next) { + fprintf(fp, "Interface %s\n", ifi->ifname); fprintf(fp, " probe interval: "); - if (ifinfo->probeinterval) { - fprintf(fp, "%d\n", ifinfo->probeinterval); - fprintf(fp, " probe timer: %d\n", ifinfo->probetimer); + if (ifi->probeinterval) { + fprintf(fp, "%d\n", ifi->probeinterval); + fprintf(fp, " probe timer: %d\n", ifi->probetimer); } else { fprintf(fp, "infinity\n"); fprintf(fp, " no probe timer\n"); } fprintf(fp, " interface status: %s\n", - ifinfo->active > 0 ? "active" : "inactive"); + ifi->active > 0 ? "active" : "inactive"); fprintf(fp, " other config: %s\n", - ifinfo->otherconfig ? "on" : "off"); - fprintf(fp, " rtsold status: %s\n", ifstatstr[ifinfo->state]); + ifi->otherconfig ? "on" : "off"); + fprintf(fp, " rtsold status: %s\n", ifstatstr[ifi->state]); fprintf(fp, " carrier detection: %s\n", - ifinfo->mediareqok ? "available" : "unavailable"); + ifi->mediareqok ? "available" : "unavailable"); fprintf(fp, " probes: %d, dadcount = %d\n", - ifinfo->probes, ifinfo->dadcount); - if (ifinfo->timer.tv_sec == tm_max.tv_sec && - ifinfo->timer.tv_usec == tm_max.tv_usec) + ifi->probes, ifi->dadcount); + if (ifi->timer.tv_sec == tm_max.tv_sec && + ifi->timer.tv_usec == tm_max.tv_usec) fprintf(fp, " no timer\n"); else { fprintf(fp, " timer: interval=%d:%d, expire=%s\n", - (int)ifinfo->timer.tv_sec, - (int)ifinfo->timer.tv_usec, - (ifinfo->expire.tv_sec < now.tv_sec) ? "expired" - : sec2str(ifinfo->expire.tv_sec - now.tv_sec)); + (int)ifi->timer.tv_sec, + (int)ifi->timer.tv_usec, + (ifi->expire.tv_sec < now.tv_sec) ? "expired" + : sec2str(ifi->expire.tv_sec - now.tv_sec)); } - fprintf(fp, " number of valid RAs: %d\n", ifinfo->racnt); + fprintf(fp, " number of valid RAs: %d\n", ifi->racnt); } } void rtsold_dump_file(const char *dumpfile) { if ((fp = fopen(dumpfile, "w")) == NULL) { warnmsg(LOG_WARNING, __func__, "open a dump file(%s): %s", dumpfile, strerror(errno)); return; } dump_interface_status(); fclose(fp); } static const char * sec2str(time_t total) { static char result[256]; int days, hours, mins, secs; int first = 1; char *p = result; char *ep = &result[sizeof(result)]; int n; days = total / 3600 / 24; hours = (total / 3600) % 24; mins = (total / 60) % 60; secs = total % 60; if (days) { first = 0; n = snprintf(p, ep - p, "%dd", days); if (n < 0 || n >= ep - p) return "?"; p += n; } if (!first || hours) { first = 0; n = snprintf(p, ep - p, "%dh", hours); if (n < 0 || n >= ep - p) return "?"; p += n; } if (!first || mins) { first = 0; n = snprintf(p, ep - p, "%dm", mins); if (n < 0 || n >= ep - p) return "?"; p += n; } snprintf(p, ep - p, "%ds", secs); - return(result); + + return (result); } Index: projects/largeSMP/usr.sbin/rtsold/if.c =================================================================== --- projects/largeSMP/usr.sbin/rtsold/if.c (revision 222782) +++ projects/largeSMP/usr.sbin/rtsold/if.c (revision 222783) @@ -1,438 +1,436 @@ /* $KAME: if.c,v 1.27 2003/10/05 00:09:36 itojun Exp $ */ /* * 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. * * $FreeBSD$ */ #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 "rtsold.h" extern int rssock; static int ifsock; static int get_llflag(const char *); static void get_rtaddrs(int, struct sockaddr *, struct sockaddr **); int ifinit(void) { ifsock = rssock; return(0); } int interface_up(char *name) { struct ifreq ifr; struct in6_ndireq nd; int llflag; int s; memset(&ifr, 0, sizeof(ifr)); strncpy(ifr.ifr_name, name, sizeof(ifr.ifr_name)); memset(&nd, 0, sizeof(nd)); strlcpy(nd.ifname, name, sizeof(nd.ifname)); if (ioctl(ifsock, SIOCGIFFLAGS, (caddr_t)&ifr) < 0) { warnmsg(LOG_WARNING, __func__, "ioctl(SIOCGIFFLAGS): %s", strerror(errno)); - return(-1); + return (-1); } if (!(ifr.ifr_flags & IFF_UP)) { ifr.ifr_flags |= IFF_UP; if (ioctl(ifsock, SIOCSIFFLAGS, (caddr_t)&ifr) < 0) warnmsg(LOG_ERR, __func__, "ioctl(SIOCSIFFLAGS): %s", strerror(errno)); - return(-1); + return (-1); } if ((s = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) { warnmsg(LOG_WARNING, __func__, "socket(AF_INET6, SOCK_DGRAM): %s", strerror(errno)); - return(-1); + return (-1); } if (ioctl(s, SIOCGIFINFO_IN6, (caddr_t)&nd) < 0) { warnmsg(LOG_WARNING, __func__, "ioctl(SIOCGIFINFO_IN6): %s", strerror(errno)); close(s); - return(-1); + return (-1); } warnmsg(LOG_DEBUG, __func__, "checking if %s is ready...", name); if (nd.ndi.flags & ND6_IFF_IFDISABLED) { if (Fflag) { nd.ndi.flags &= ~ND6_IFF_IFDISABLED; if (ioctl(s, SIOCSIFINFO_IN6, (caddr_t)&nd)) { warnmsg(LOG_WARNING, __func__, "ioctl(SIOCSIFINFO_IN6): %s", strerror(errno)); close(s); - return(-1); + return (-1); } } else { warnmsg(LOG_WARNING, __func__, "%s is disabled.", name); close(s); - return(-1); + return (-1); } } if (!(nd.ndi.flags & ND6_IFF_ACCEPT_RTADV)) { if (Fflag) { nd.ndi.flags |= ND6_IFF_ACCEPT_RTADV; if (ioctl(s, SIOCSIFINFO_IN6, (caddr_t)&nd)) { warnmsg(LOG_WARNING, __func__, "ioctl(SIOCSIFINFO_IN6): %s", strerror(errno)); close(s); - return(-1); + return (-1); } } else { warnmsg(LOG_WARNING, __func__, "%s does not accept Router Advertisement.", name); close(s); - return(-1); + return (-1); } } close(s); llflag = get_llflag(name); if (llflag < 0) { warnmsg(LOG_WARNING, __func__, "get_llflag() failed, anyway I'll try"); - return 0; + return (0); } if (!(llflag & IN6_IFF_NOTREADY)) { warnmsg(LOG_DEBUG, __func__, "%s is ready", name); - return(0); + return (0); } else { if (llflag & IN6_IFF_TENTATIVE) { warnmsg(LOG_DEBUG, __func__, "%s is tentative", name); - return IFS_TENTATIVE; + return (IFS_TENTATIVE); } if (llflag & IN6_IFF_DUPLICATED) warnmsg(LOG_DEBUG, __func__, "%s is duplicated", name); - return -1; + return (-1); } } int interface_status(struct ifinfo *ifinfo) { char *ifname = ifinfo->ifname; struct ifreq ifr; struct ifmediareq ifmr; /* get interface flags */ memset(&ifr, 0, sizeof(ifr)); strncpy(ifr.ifr_name, ifname, sizeof(ifr.ifr_name)); if (ioctl(ifsock, SIOCGIFFLAGS, &ifr) < 0) { warnmsg(LOG_ERR, __func__, "ioctl(SIOCGIFFLAGS) on %s: %s", ifname, strerror(errno)); - return(-1); + return (-1); } /* * if one of UP and RUNNING flags is dropped, * the interface is not active. */ - if ((ifr.ifr_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING)) { + if ((ifr.ifr_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING)) goto inactive; - } - /* Next, check carrier on the interface, if possible */ if (!ifinfo->mediareqok) goto active; memset(&ifmr, 0, sizeof(ifmr)); strncpy(ifmr.ifm_name, ifname, sizeof(ifmr.ifm_name)); if (ioctl(ifsock, SIOCGIFMEDIA, (caddr_t)&ifmr) < 0) { if (errno != EINVAL) { warnmsg(LOG_DEBUG, __func__, "ioctl(SIOCGIFMEDIA) on %s: %s", ifname, strerror(errno)); return(-1); } /* * EINVAL simply means that the interface does not support * the SIOCGIFMEDIA ioctl. We regard it alive. */ ifinfo->mediareqok = 0; goto active; } if (ifmr.ifm_status & IFM_AVALID) { switch (ifmr.ifm_active & IFM_NMASK) { case IFM_ETHER: case IFM_IEEE80211: if (ifmr.ifm_status & IFM_ACTIVE) goto active; else goto inactive; break; default: goto inactive; } } inactive: - return(0); + return (0); active: - return(1); + return (1); } #define ROUNDUP(a, size) \ (((a) & ((size)-1)) ? (1 + ((a) | ((size)-1))) : (a)) #define NEXT_SA(ap) (ap) = (struct sockaddr *) \ ((caddr_t)(ap) + ((ap)->sa_len ? ROUNDUP((ap)->sa_len,\ sizeof(u_long)) : sizeof(u_long))) #define ROUNDUP8(a) (1 + (((a) - 1) | 7)) int lladdropt_length(struct sockaddr_dl *sdl) { switch (sdl->sdl_type) { case IFT_ETHER: #ifdef IFT_IEEE80211 case IFT_IEEE80211: #endif - return(ROUNDUP8(ETHER_ADDR_LEN + 2)); + return (ROUNDUP8(ETHER_ADDR_LEN + 2)); default: - return(0); + return (0); } } void lladdropt_fill(struct sockaddr_dl *sdl, struct nd_opt_hdr *ndopt) { char *addr; ndopt->nd_opt_type = ND_OPT_SOURCE_LINKADDR; /* fixed */ switch (sdl->sdl_type) { case IFT_ETHER: #ifdef IFT_IEEE80211 case IFT_IEEE80211: #endif ndopt->nd_opt_len = (ROUNDUP8(ETHER_ADDR_LEN + 2)) >> 3; addr = (char *)(ndopt + 1); memcpy(addr, LLADDR(sdl), ETHER_ADDR_LEN); break; default: warnmsg(LOG_ERR, __func__, "unsupported link type(%d)", sdl->sdl_type); exit(1); } return; } struct sockaddr_dl * if_nametosdl(char *name) { int mib[6] = {CTL_NET, AF_ROUTE, 0, 0, NET_RT_IFLIST, 0}; char *buf, *next, *lim; size_t len; struct if_msghdr *ifm; struct sockaddr *sa, *rti_info[RTAX_MAX]; struct sockaddr_dl *sdl = NULL, *ret_sdl; if (sysctl(mib, 6, NULL, &len, NULL, 0) < 0) return(NULL); if ((buf = malloc(len)) == NULL) return(NULL); if (sysctl(mib, 6, buf, &len, NULL, 0) < 0) { free(buf); - return(NULL); + return (NULL); } lim = buf + len; for (next = buf; next < lim; next += ifm->ifm_msglen) { ifm = (struct if_msghdr *)next; if (ifm->ifm_type == RTM_IFINFO) { sa = (struct sockaddr *)(ifm + 1); get_rtaddrs(ifm->ifm_addrs, sa, rti_info); if ((sa = rti_info[RTAX_IFP]) != NULL) { if (sa->sa_family == AF_LINK) { sdl = (struct sockaddr_dl *)sa; if (strlen(name) != sdl->sdl_nlen) continue; /* not same len */ if (strncmp(&sdl->sdl_data[0], name, sdl->sdl_nlen) == 0) { break; } } } } } if (next == lim) { /* search failed */ free(buf); - return(NULL); + return (NULL); } if ((ret_sdl = malloc(sdl->sdl_len)) == NULL) { free(buf); - return(NULL); + return (NULL); } memcpy((caddr_t)ret_sdl, (caddr_t)sdl, sdl->sdl_len); free(buf); - return(ret_sdl); + return (ret_sdl); } int getinet6sysctl(int code) { int mib[] = { CTL_NET, PF_INET6, IPPROTO_IPV6, 0 }; int value; size_t size; mib[3] = code; size = sizeof(value); if (sysctl(mib, sizeof(mib)/sizeof(mib[0]), &value, &size, NULL, 0) < 0) - return -1; + return (-1); else - return value; + return (value); } int setinet6sysctl(int code, int newval) { int mib[] = { CTL_NET, PF_INET6, IPPROTO_IPV6, 0 }; int value; size_t size; mib[3] = code; size = sizeof(value); if (sysctl(mib, sizeof(mib)/sizeof(mib[0]), &value, &size, &newval, sizeof(newval)) < 0) - return -1; + return (-1); else - return value; + return (value); } /*------------------------------------------------------------*/ /* get ia6_flags for link-local addr on if. returns -1 on error. */ static int get_llflag(const char *name) { struct ifaddrs *ifap, *ifa; struct in6_ifreq ifr6; struct sockaddr_in6 *sin6; int s; if ((s = socket(PF_INET6, SOCK_DGRAM, 0)) < 0) { warnmsg(LOG_ERR, __func__, "socket(SOCK_DGRAM): %s", strerror(errno)); exit(1); } if (getifaddrs(&ifap) != 0) { warnmsg(LOG_ERR, __func__, "getifaddrs: %s", strerror(errno)); exit(1); } for (ifa = ifap; ifa; ifa = ifa->ifa_next) { if (strlen(ifa->ifa_name) != strlen(name) || strncmp(ifa->ifa_name, name, strlen(name)) != 0) continue; if (ifa->ifa_addr->sa_family != AF_INET6) continue; sin6 = (struct sockaddr_in6 *)ifa->ifa_addr; if (!IN6_IS_ADDR_LINKLOCAL(&sin6->sin6_addr)) continue; memset(&ifr6, 0, sizeof(ifr6)); strncpy(ifr6.ifr_name, name, sizeof(ifr6.ifr_name)); memcpy(&ifr6.ifr_ifru.ifru_addr, sin6, sin6->sin6_len); if (ioctl(s, SIOCGIFAFLAG_IN6, &ifr6) < 0) { warnmsg(LOG_ERR, __func__, "ioctl(SIOCGIFAFLAG_IN6): %s", strerror(errno)); exit(1); } freeifaddrs(ifap); close(s); - return ifr6.ifr_ifru.ifru_flags6; + return (ifr6.ifr_ifru.ifru_flags6); } freeifaddrs(ifap); close(s); - return -1; + return (-1); } static void get_rtaddrs(int addrs, struct sockaddr *sa, struct sockaddr **rti_info) { int i; for (i = 0; i < RTAX_MAX; i++) { if (addrs & (1 << i)) { rti_info[i] = sa; NEXT_SA(sa); } else rti_info[i] = NULL; } } Index: projects/largeSMP/usr.sbin/rtsold/probe.c =================================================================== --- projects/largeSMP/usr.sbin/rtsold/probe.c (revision 222782) +++ projects/largeSMP/usr.sbin/rtsold/probe.c (revision 222783) @@ -1,184 +1,185 @@ /* $KAME: probe.c,v 1.17 2003/10/05 00:09:36 itojun Exp $ */ /* * Copyright (C) 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. * * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "rtsold.h" static struct msghdr sndmhdr; static struct iovec sndiov[2]; static int probesock; static void sendprobe(struct in6_addr *, struct ifinfo *); int probe_init(void) { int scmsglen = CMSG_SPACE(sizeof(struct in6_pktinfo)) + CMSG_SPACE(sizeof(int)); static u_char *sndcmsgbuf = NULL; if (sndcmsgbuf == NULL && (sndcmsgbuf = (u_char *)malloc(scmsglen)) == NULL) { warnmsg(LOG_ERR, __func__, "malloc failed"); - return(-1); + return (-1); } if ((probesock = socket(AF_INET6, SOCK_RAW, IPPROTO_NONE)) < 0) { warnmsg(LOG_ERR, __func__, "socket: %s", strerror(errno)); - return(-1); + return (-1); } /* make the socket send-only */ if (shutdown(probesock, 0)) { warnmsg(LOG_ERR, __func__, "shutdown: %s", strerror(errno)); - return(-1); + return (-1); } /* initialize msghdr for sending packets */ sndmhdr.msg_namelen = sizeof(struct sockaddr_in6); sndmhdr.msg_iov = sndiov; sndmhdr.msg_iovlen = 1; sndmhdr.msg_control = (caddr_t)sndcmsgbuf; sndmhdr.msg_controllen = scmsglen; - return(0); + + return (0); } /* * Probe if each router in the default router list is still alive. */ void defrouter_probe(struct ifinfo *ifinfo) { u_char ntopbuf[INET6_ADDRSTRLEN]; struct in6_drlist dr; int s, i; int ifindex = ifinfo->sdl->sdl_index; if ((s = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) { warnmsg(LOG_ERR, __func__, "socket: %s", strerror(errno)); return; } memset(&dr, 0, sizeof(dr)); strlcpy(dr.ifname, "lo0", sizeof dr.ifname); /* dummy interface */ if (ioctl(s, SIOCGDRLST_IN6, (caddr_t)&dr) < 0) { warnmsg(LOG_ERR, __func__, "ioctl(SIOCGDRLST_IN6): %s", strerror(errno)); goto closeandend; } for (i = 0; i < DRLSTSIZ && dr.defrouter[i].if_index; i++) { if (ifindex && dr.defrouter[i].if_index == ifindex) { /* sanity check */ if (!IN6_IS_ADDR_LINKLOCAL(&dr.defrouter[i].rtaddr)) { warnmsg(LOG_ERR, __func__, "default router list contains a " "non-link-local address(%s)", inet_ntop(AF_INET6, &dr.defrouter[i].rtaddr, ntopbuf, INET6_ADDRSTRLEN)); continue; /* ignore the address */ } sendprobe(&dr.defrouter[i].rtaddr, ifinfo); } } closeandend: close(s); } static void sendprobe(struct in6_addr *addr, struct ifinfo *ifinfo) { u_char ntopbuf[INET6_ADDRSTRLEN], ifnamebuf[IFNAMSIZ]; struct sockaddr_in6 sa6_probe; struct in6_pktinfo *pi; struct cmsghdr *cm; u_int32_t ifindex = ifinfo->sdl->sdl_index; int hoplimit = 1; memset(&sa6_probe, 0, sizeof(sa6_probe)); sa6_probe.sin6_family = AF_INET6; sa6_probe.sin6_len = sizeof(sa6_probe); sa6_probe.sin6_addr = *addr; sa6_probe.sin6_scope_id = ifinfo->linkid; sndmhdr.msg_name = (caddr_t)&sa6_probe; sndmhdr.msg_iov[0].iov_base = NULL; sndmhdr.msg_iov[0].iov_len = 0; cm = CMSG_FIRSTHDR(&sndmhdr); /* specify the outgoing interface */ cm->cmsg_level = IPPROTO_IPV6; cm->cmsg_type = IPV6_PKTINFO; cm->cmsg_len = CMSG_LEN(sizeof(struct in6_pktinfo)); pi = (struct in6_pktinfo *)CMSG_DATA(cm); memset(&pi->ipi6_addr, 0, sizeof(pi->ipi6_addr)); /*XXX*/ pi->ipi6_ifindex = ifindex; /* specify the hop limit of the packet for safety */ cm = CMSG_NXTHDR(&sndmhdr, cm); cm->cmsg_level = IPPROTO_IPV6; cm->cmsg_type = IPV6_HOPLIMIT; cm->cmsg_len = CMSG_LEN(sizeof(int)); memcpy(CMSG_DATA(cm), &hoplimit, sizeof(int)); warnmsg(LOG_DEBUG, __func__, "probe a router %s on %s", inet_ntop(AF_INET6, addr, ntopbuf, INET6_ADDRSTRLEN), if_indextoname(ifindex, ifnamebuf)); if (sendmsg(probesock, &sndmhdr, 0)) warnmsg(LOG_ERR, __func__, "sendmsg on %s: %s", if_indextoname(ifindex, ifnamebuf), strerror(errno)); } Index: projects/largeSMP/usr.sbin/rtsold/rtsock.c =================================================================== --- projects/largeSMP/usr.sbin/rtsold/rtsock.c (revision 222782) +++ projects/largeSMP/usr.sbin/rtsold/rtsock.c (revision 222783) @@ -1,175 +1,175 @@ /* $KAME: rtsock.c,v 1.3 2000/10/10 08:46:45 itojun Exp $ */ /* $FreeBSD$ */ /* * Copyright (C) 2000 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. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "rtsold.h" #define ROUNDUP(a, size) \ (((a) & ((size)-1)) ? (1 + ((a) | ((size)-1))) : (a)) #define NEXT_SA(ap) (ap) = (struct sockaddr *) \ ((caddr_t)(ap) + \ ((ap)->sa_len ? ROUNDUP((ap)->sa_len, sizeof(u_long)) \ : sizeof(u_long))) #ifdef RTM_IFANNOUNCE /*NetBSD 1.5 or later*/ static int rtsock_input_ifannounce(int, struct rt_msghdr *, char *); #endif static struct { u_char type; size_t minlen; int (*func)(int, struct rt_msghdr *, char *); } rtsock_dispatch[] = { #ifdef RTM_IFANNOUNCE /*NetBSD 1.5 or later*/ { RTM_IFANNOUNCE, sizeof(struct if_announcemsghdr), rtsock_input_ifannounce }, #endif { 0, 0, NULL }, }; int rtsock_open(void) { - return socket(PF_ROUTE, SOCK_RAW, 0); + return (socket(PF_ROUTE, SOCK_RAW, 0)); } int rtsock_input(int s) { ssize_t n; char msg[2048]; char *lim, *next; struct rt_msghdr *rtm; int idx; ssize_t len; int ret = 0; const ssize_t lenlim = offsetof(struct rt_msghdr, rtm_msglen) + sizeof(rtm->rtm_msglen); n = read(s, msg, sizeof(msg)); lim = msg + n; for (next = msg; next < lim; next += len) { rtm = (struct rt_msghdr *)next; if (lim - next < lenlim) break; len = rtm->rtm_msglen; if (len < lenlim) break; if (dflag > 1) { warnmsg(LOG_INFO, __func__, "rtmsg type %d, len=%lu", rtm->rtm_type, (u_long)len); } for (idx = 0; rtsock_dispatch[idx].func; idx++) { if (rtm->rtm_type != rtsock_dispatch[idx].type) continue; if (rtm->rtm_msglen < rtsock_dispatch[idx].minlen) { warnmsg(LOG_INFO, __func__, "rtmsg type %d too short!", rtm->rtm_type); continue; } ret = (*rtsock_dispatch[idx].func)(s, rtm, lim); break; } } - return ret; + return (ret); } #ifdef RTM_IFANNOUNCE /*NetBSD 1.5 or later*/ static int rtsock_input_ifannounce(int s __unused, struct rt_msghdr *rtm, char *lim) { struct if_announcemsghdr *ifan; struct ifinfo *ifinfo; ifan = (struct if_announcemsghdr *)rtm; if ((char *)(ifan + 1) > lim) - return -1; + return (-1); switch (ifan->ifan_what) { case IFAN_ARRIVAL: /* * XXX for NetBSD 1.5, interface index will monotonically be * increased as new pcmcia card gets inserted. * we may be able to do a name-based interface match, * and call ifreconfig() to enable the interface again. */ warnmsg(LOG_INFO, __func__, "interface %s inserted", ifan->ifan_name); break; case IFAN_DEPARTURE: warnmsg(LOG_WARNING, __func__, "interface %s removed", ifan->ifan_name); ifinfo = find_ifinfo(ifan->ifan_index); if (ifinfo) { if (dflag > 1) { warnmsg(LOG_INFO, __func__, "bring interface %s to DOWN state", ifan->ifan_name); } ifinfo->state = IFS_DOWN; } break; } - return 0; + return (0); } #endif Index: projects/largeSMP/usr.sbin/rtsold/rtsol.c =================================================================== --- projects/largeSMP/usr.sbin/rtsold/rtsol.c (revision 222782) +++ projects/largeSMP/usr.sbin/rtsold/rtsol.c (revision 222783) @@ -1,473 +1,826 @@ /* $KAME: rtsol.c,v 1.27 2003/10/05 00:09:36 itojun Exp $ */ /* * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. + * Copyright (C) 2011 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. * 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. * * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include +#define __BSD_VISIBLE 1 /* IN6ADDR_LINKLOCAL_ALLROUTERS_INIT */ #include +#undef __BSD_VISIBLE #include #include #include #include +#include #include #include #include #include #include #include #include #include #include #include "rtsold.h" -#define ALLROUTER "ff02::2" - static struct msghdr rcvmhdr; static struct msghdr sndmhdr; static struct iovec rcviov[2]; static struct iovec sndiov[2]; static struct sockaddr_in6 from; static int rcvcmsglen; int rssock; +struct ifinfo_head_t ifinfo_head = + TAILQ_HEAD_INITIALIZER(ifinfo_head); -static struct sockaddr_in6 sin6_allrouters = { +static const struct sockaddr_in6 sin6_allrouters = { .sin6_len = sizeof(sin6_allrouters), .sin6_family = AF_INET6, + .sin6_addr = IN6ADDR_LINKLOCAL_ALLROUTERS_INIT, }; -static void call_script(char *, char *); +static void call_script(const int, const char *const *, void *); +static size_t dname_labeldec(char *, size_t, const char *); static int safefile(const char *); +#define _ARGS_OTHER otherconf_script, ifi->ifname +#define _ARGS_RESADD resolvconf_script, "-a", ifi->ifname +#define _ARGS_RESDEL resolvconf_script, "-d", ifi->ifname + +#define CALL_SCRIPT(name, sm_head) \ + do { \ + const char *const sarg[] = { _ARGS_##name, NULL }; \ + call_script(sizeof(sarg), sarg, sm_head); \ + } while(0) + +#define ELM_MALLOC(p,error_action) \ + do { \ + p = malloc(sizeof(*p)); \ + if (p == NULL) { \ + warnmsg(LOG_ERR, __func__, "malloc failed: %s", \ + strerror(errno)); \ + error_action; \ + } \ + memset(p, 0, sizeof(*p)); \ + } while(0) + int sockopen(void) { static u_char *rcvcmsgbuf = NULL, *sndcmsgbuf = NULL; int sndcmsglen, on; static u_char answer[1500]; struct icmp6_filter filt; sndcmsglen = rcvcmsglen = CMSG_SPACE(sizeof(struct in6_pktinfo)) + CMSG_SPACE(sizeof(int)); if (rcvcmsgbuf == NULL && (rcvcmsgbuf = malloc(rcvcmsglen)) == NULL) { warnmsg(LOG_ERR, __func__, "malloc for receive msghdr failed"); - return(-1); + return (-1); } if (sndcmsgbuf == NULL && (sndcmsgbuf = malloc(sndcmsglen)) == NULL) { warnmsg(LOG_ERR, __func__, "malloc for send msghdr failed"); - return(-1); + return (-1); } - memset(&sin6_allrouters, 0, sizeof(struct sockaddr_in6)); - sin6_allrouters.sin6_family = AF_INET6; - sin6_allrouters.sin6_len = sizeof(sin6_allrouters); - if (inet_pton(AF_INET6, ALLROUTER, - &sin6_allrouters.sin6_addr.s6_addr) != 1) { - warnmsg(LOG_ERR, __func__, "inet_pton failed for %s", - ALLROUTER); - return(-1); - } - if ((rssock = socket(AF_INET6, SOCK_RAW, IPPROTO_ICMPV6)) < 0) { warnmsg(LOG_ERR, __func__, "socket: %s", strerror(errno)); - return(-1); + return (-1); } /* specify to tell receiving interface */ on = 1; -#ifdef IPV6_RECVPKTINFO if (setsockopt(rssock, IPPROTO_IPV6, IPV6_RECVPKTINFO, &on, sizeof(on)) < 0) { warnmsg(LOG_ERR, __func__, "IPV6_RECVPKTINFO: %s", strerror(errno)); exit(1); } -#else /* old adv. API */ - if (setsockopt(rssock, IPPROTO_IPV6, IPV6_PKTINFO, &on, - sizeof(on)) < 0) { - warnmsg(LOG_ERR, __func__, "IPV6_PKTINFO: %s", - strerror(errno)); - exit(1); - } -#endif - on = 1; /* specify to tell value of hoplimit field of received IP6 hdr */ -#ifdef IPV6_RECVHOPLIMIT + on = 1; if (setsockopt(rssock, IPPROTO_IPV6, IPV6_RECVHOPLIMIT, &on, sizeof(on)) < 0) { warnmsg(LOG_ERR, __func__, "IPV6_RECVHOPLIMIT: %s", strerror(errno)); exit(1); } -#else /* old adv. API */ - if (setsockopt(rssock, IPPROTO_IPV6, IPV6_HOPLIMIT, &on, - sizeof(on)) < 0) { - warnmsg(LOG_ERR, __func__, "IPV6_HOPLIMIT: %s", - strerror(errno)); - exit(1); - } -#endif /* specfiy to accept only router advertisements on the socket */ ICMP6_FILTER_SETBLOCKALL(&filt); ICMP6_FILTER_SETPASS(ND_ROUTER_ADVERT, &filt); if (setsockopt(rssock, IPPROTO_ICMPV6, ICMP6_FILTER, &filt, sizeof(filt)) == -1) { warnmsg(LOG_ERR, __func__, "setsockopt(ICMP6_FILTER): %s", strerror(errno)); return(-1); } /* initialize msghdr for receiving packets */ rcviov[0].iov_base = (caddr_t)answer; rcviov[0].iov_len = sizeof(answer); rcvmhdr.msg_name = (caddr_t)&from; rcvmhdr.msg_iov = rcviov; rcvmhdr.msg_iovlen = 1; rcvmhdr.msg_control = (caddr_t) rcvcmsgbuf; /* initialize msghdr for sending packets */ sndmhdr.msg_namelen = sizeof(struct sockaddr_in6); sndmhdr.msg_iov = sndiov; sndmhdr.msg_iovlen = 1; sndmhdr.msg_control = (caddr_t)sndcmsgbuf; sndmhdr.msg_controllen = sndcmsglen; - return(rssock); + return (rssock); } void -sendpacket(struct ifinfo *ifinfo) +sendpacket(struct ifinfo *ifi) { struct in6_pktinfo *pi; struct cmsghdr *cm; int hoplimit = 255; ssize_t i; struct sockaddr_in6 dst; dst = sin6_allrouters; - dst.sin6_scope_id = ifinfo->linkid; + dst.sin6_scope_id = ifi->linkid; sndmhdr.msg_name = (caddr_t)&dst; - sndmhdr.msg_iov[0].iov_base = (caddr_t)ifinfo->rs_data; - sndmhdr.msg_iov[0].iov_len = ifinfo->rs_datalen; + sndmhdr.msg_iov[0].iov_base = (caddr_t)ifi->rs_data; + sndmhdr.msg_iov[0].iov_len = ifi->rs_datalen; cm = CMSG_FIRSTHDR(&sndmhdr); /* specify the outgoing interface */ cm->cmsg_level = IPPROTO_IPV6; cm->cmsg_type = IPV6_PKTINFO; cm->cmsg_len = CMSG_LEN(sizeof(struct in6_pktinfo)); pi = (struct in6_pktinfo *)CMSG_DATA(cm); memset(&pi->ipi6_addr, 0, sizeof(pi->ipi6_addr)); /*XXX*/ - pi->ipi6_ifindex = ifinfo->sdl->sdl_index; + pi->ipi6_ifindex = ifi->sdl->sdl_index; /* specify the hop limit of the packet */ cm = CMSG_NXTHDR(&sndmhdr, cm); cm->cmsg_level = IPPROTO_IPV6; cm->cmsg_type = IPV6_HOPLIMIT; cm->cmsg_len = CMSG_LEN(sizeof(int)); memcpy(CMSG_DATA(cm), &hoplimit, sizeof(int)); warnmsg(LOG_DEBUG, __func__, "send RS on %s, whose state is %d", - ifinfo->ifname, ifinfo->state); + ifi->ifname, ifi->state); i = sendmsg(rssock, &sndmhdr, 0); - if (i < 0 || (size_t)i != ifinfo->rs_datalen) { + if (i < 0 || (size_t)i != ifi->rs_datalen) { /* * ENETDOWN is not so serious, especially when using several * network cards on a mobile node. We ignore it. */ if (errno != ENETDOWN || dflag > 0) warnmsg(LOG_ERR, __func__, "sendmsg on %s: %s", - ifinfo->ifname, strerror(errno)); + ifi->ifname, strerror(errno)); } /* update counter */ - ifinfo->probes++; + ifi->probes++; } void rtsol_input(int s) { u_char ntopbuf[INET6_ADDRSTRLEN], ifnamebuf[IFNAMSIZ]; - int ifindex = 0, *hlimp = NULL; - ssize_t i; + int l, ifindex = 0, *hlimp = NULL; + ssize_t msglen; struct in6_pktinfo *pi = NULL; struct ifinfo *ifi = NULL; + struct ra_opt *rao = NULL; struct icmp6_hdr *icp; struct nd_router_advert *nd_ra; struct cmsghdr *cm; + char *raoptp; + char *p; + struct in6_addr *addr; + struct nd_opt_hdr *ndo; + struct nd_opt_rdnss *rdnss; + struct nd_opt_dnssl *dnssl; + size_t len; + char nsbuf[INET6_ADDRSTRLEN + 1 + IFNAMSIZ + 1]; + char dname[NI_MAXHOST]; + struct timeval now; + struct timeval lifetime; /* get message. namelen and controllen must always be initialized. */ rcvmhdr.msg_namelen = sizeof(from); rcvmhdr.msg_controllen = rcvcmsglen; - if ((i = recvmsg(s, &rcvmhdr, 0)) < 0) { + if ((msglen = recvmsg(s, &rcvmhdr, 0)) < 0) { warnmsg(LOG_ERR, __func__, "recvmsg: %s", strerror(errno)); return; } /* extract optional information via Advanced API */ for (cm = (struct cmsghdr *)CMSG_FIRSTHDR(&rcvmhdr); cm; cm = (struct cmsghdr *)CMSG_NXTHDR(&rcvmhdr, cm)) { if (cm->cmsg_level == IPPROTO_IPV6 && cm->cmsg_type == IPV6_PKTINFO && cm->cmsg_len == CMSG_LEN(sizeof(struct in6_pktinfo))) { pi = (struct in6_pktinfo *)(CMSG_DATA(cm)); ifindex = pi->ipi6_ifindex; } if (cm->cmsg_level == IPPROTO_IPV6 && cm->cmsg_type == IPV6_HOPLIMIT && cm->cmsg_len == CMSG_LEN(sizeof(int))) hlimp = (int *)CMSG_DATA(cm); } if (ifindex == 0) { warnmsg(LOG_ERR, __func__, "failed to get receiving interface"); return; } if (hlimp == NULL) { warnmsg(LOG_ERR, __func__, "failed to get receiving hop limit"); return; } - if ((size_t)i < sizeof(struct nd_router_advert)) { + if ((size_t)msglen < sizeof(struct nd_router_advert)) { warnmsg(LOG_INFO, __func__, - "packet size(%zd) is too short", i); + "packet size(%zd) is too short", msglen); return; } icp = (struct icmp6_hdr *)rcvmhdr.msg_iov[0].iov_base; if (icp->icmp6_type != ND_ROUTER_ADVERT) { /* * this should not happen because we configured a filter * that only passes RAs on the receiving socket. */ warnmsg(LOG_ERR, __func__, "invalid icmp type(%d) from %s on %s", icp->icmp6_type, inet_ntop(AF_INET6, &from.sin6_addr, ntopbuf, INET6_ADDRSTRLEN), if_indextoname(pi->ipi6_ifindex, ifnamebuf)); return; } if (icp->icmp6_code != 0) { warnmsg(LOG_INFO, __func__, "invalid icmp code(%d) from %s on %s", icp->icmp6_code, inet_ntop(AF_INET6, &from.sin6_addr, ntopbuf, INET6_ADDRSTRLEN), if_indextoname(pi->ipi6_ifindex, ifnamebuf)); return; } if (*hlimp != 255) { warnmsg(LOG_INFO, __func__, "invalid RA with hop limit(%d) from %s on %s", *hlimp, inet_ntop(AF_INET6, &from.sin6_addr, ntopbuf, INET6_ADDRSTRLEN), if_indextoname(pi->ipi6_ifindex, ifnamebuf)); return; } if (pi && !IN6_IS_ADDR_LINKLOCAL(&from.sin6_addr)) { warnmsg(LOG_INFO, __func__, "invalid RA with non link-local source from %s on %s", inet_ntop(AF_INET6, &from.sin6_addr, ntopbuf, INET6_ADDRSTRLEN), if_indextoname(pi->ipi6_ifindex, ifnamebuf)); return; } /* xxx: more validation? */ if ((ifi = find_ifinfo(pi->ipi6_ifindex)) == NULL) { warnmsg(LOG_INFO, __func__, "received RA from %s on an unexpected IF(%s)", inet_ntop(AF_INET6, &from.sin6_addr, ntopbuf, INET6_ADDRSTRLEN), if_indextoname(pi->ipi6_ifindex, ifnamebuf)); return; } warnmsg(LOG_DEBUG, __func__, "received RA from %s on %s, state is %d", inet_ntop(AF_INET6, &from.sin6_addr, ntopbuf, INET6_ADDRSTRLEN), ifi->ifname, ifi->state); nd_ra = (struct nd_router_advert *)icp; /* * Process the "O bit." * If the value of OtherConfigFlag changes from FALSE to TRUE, the * host should invoke the stateful autoconfiguration protocol, * requesting information. * [RFC 2462 Section 5.5.3] */ if (((nd_ra->nd_ra_flags_reserved) & ND_RA_FLAG_OTHER) && !ifi->otherconfig) { warnmsg(LOG_DEBUG, __func__, "OtherConfigFlag on %s is turned on", ifi->ifname); ifi->otherconfig = 1; - call_script(otherconf_script, ifi->ifname); + CALL_SCRIPT(OTHER, NULL); } + /* Initialize ra_opt per-interface structure. */ + gettimeofday(&now, NULL); + if (!TAILQ_EMPTY(&ifi->ifi_ra_opt)) + while ((rao = TAILQ_FIRST(&ifi->ifi_ra_opt)) != NULL) { + if (rao->rao_msg != NULL) + free(rao->rao_msg); + TAILQ_REMOVE(&ifi->ifi_ra_opt, rao, rao_next); + free(rao); + } + else + TAILQ_INIT(&ifi->ifi_ra_opt); + +#define RA_OPT_NEXT_HDR(x) (struct nd_opt_hdr *)((char *)x + \ + (((struct nd_opt_hdr *)x)->nd_opt_len * 8)) + + /* Process RA options. */ + warnmsg(LOG_DEBUG, __func__, "Processing RA"); + raoptp = (char *)icp + sizeof(struct nd_router_advert); + while (raoptp < (char *)icp + msglen) { + ndo = (struct nd_opt_hdr *)raoptp; + warnmsg(LOG_DEBUG, __func__, "ndo = %p", raoptp); + warnmsg(LOG_DEBUG, __func__, "ndo->nd_opt_type = %d", + ndo->nd_opt_type); + warnmsg(LOG_DEBUG, __func__, "ndo->nd_opt_len = %d", + ndo->nd_opt_len); + + switch (ndo->nd_opt_type) { + case ND_OPT_RDNSS: + rdnss = (struct nd_opt_rdnss *)raoptp; + + /* Optlen sanity check (Section 5.3.1 in RFC 6106) */ + if (rdnss->nd_opt_rdnss_len < 3) { + warnmsg(LOG_INFO, __func__, + "too short RDNSS option" + "in RA from %s was ignored.", + inet_ntop(AF_INET6, &from.sin6_addr, + ntopbuf, INET6_ADDRSTRLEN)); + break; + } + + addr = (struct in6_addr *)(raoptp + sizeof(*rdnss)); + while ((char *)addr < (char *)RA_OPT_NEXT_HDR(raoptp)) { + if (inet_ntop(AF_INET6, addr, ntopbuf, + INET6_ADDRSTRLEN) == NULL) { + warnmsg(LOG_INFO, __func__, + "an invalid address in RDNSS option" + " in RA from %s was ignored.", + inet_ntop(AF_INET6, &from.sin6_addr, + ntopbuf, INET6_ADDRSTRLEN)); + addr++; + continue; + } + if (IN6_IS_ADDR_LINKLOCAL(addr)) + /* XXX: % has to be escaped here */ + l = snprintf(nsbuf, sizeof(nsbuf), + "%s%c%s", ntopbuf, + SCOPE_DELIMITER, + ifi->ifname); + else + l = snprintf(nsbuf, sizeof(nsbuf), + "%s", ntopbuf); + if (l < 0 || (size_t)l >= sizeof(nsbuf)) { + warnmsg(LOG_ERR, __func__, + "address copying error in " + "RDNSS option: %d.", l); + addr++; + continue; + } + warnmsg(LOG_DEBUG, __func__, "nsbuf = %s", + nsbuf); + + ELM_MALLOC(rao, break); + rao->rao_type = ndo->nd_opt_type; + rao->rao_len = strlen(nsbuf); + rao->rao_msg = strdup(nsbuf); + if (rao->rao_msg == NULL) { + warnmsg(LOG_ERR, __func__, + "strdup failed: %s", + strerror(errno)); + free(rao); + addr++; + continue; + } + /* Set expiration timer */ + memset(&rao->rao_expire, 0, sizeof(rao->rao_expire)); + memset(&lifetime, 0, sizeof(lifetime)); + lifetime.tv_sec = ntohl(rdnss->nd_opt_rdnss_lifetime); + timeradd(&now, &lifetime, &rao->rao_expire); + + TAILQ_INSERT_TAIL(&ifi->ifi_ra_opt, rao, rao_next); + addr++; + } + break; + case ND_OPT_DNSSL: + dnssl = (struct nd_opt_dnssl *)raoptp; + + /* Optlen sanity check (Section 5.3.1 in RFC 6106) */ + if (dnssl->nd_opt_dnssl_len < 2) { + warnmsg(LOG_INFO, __func__, + "too short DNSSL option" + "in RA from %s was ignored.", + inet_ntop(AF_INET6, &from.sin6_addr, + ntopbuf, INET6_ADDRSTRLEN)); + break; + } + + /* + * Ensure NUL-termination in DNSSL in case of + * malformed field. + */ + p = (char *)RA_OPT_NEXT_HDR(raoptp); + *(p - 1) = '\0'; + + p = raoptp + sizeof(*dnssl); + while (1 < (len = dname_labeldec(dname, sizeof(dname), + p))) { + /* length == 1 means empty string */ + warnmsg(LOG_DEBUG, __func__, "dname = %s", + dname); + + ELM_MALLOC(rao, break); + rao->rao_type = ndo->nd_opt_type; + rao->rao_len = strlen(dname); + rao->rao_msg = strdup(dname); + if (rao->rao_msg == NULL) { + warnmsg(LOG_ERR, __func__, + "strdup failed: %s", + strerror(errno)); + free(rao); + break; + } + /* Set expiration timer */ + memset(&rao->rao_expire, 0, sizeof(rao->rao_expire)); + memset(&lifetime, 0, sizeof(lifetime)); + lifetime.tv_sec = ntohl(dnssl->nd_opt_dnssl_lifetime); + timeradd(&now, &lifetime, &rao->rao_expire); + + TAILQ_INSERT_TAIL(&ifi->ifi_ra_opt, rao, rao_next); + p += len; + } + break; + default: + /* nothing to do for other options */ + break; + } + raoptp = (char *)RA_OPT_NEXT_HDR(raoptp); + } + ra_opt_handler(ifi); ifi->racnt++; switch (ifi->state) { case IFS_IDLE: /* should be ignored */ case IFS_DELAY: /* right? */ break; case IFS_PROBE: ifi->state = IFS_IDLE; ifi->probes = 0; rtsol_timer_update(ifi); break; } } +static char resstr_ns_prefix[] = "nameserver "; +static char resstr_sh_prefix[] = "search "; +static char resstr_nl[] = "\n"; +static char resstr_sp[] = " "; + +int +ra_opt_handler(struct ifinfo *ifi) +{ + struct ra_opt *rao; + struct script_msg *smp1, *smp2, *smp3; + struct timeval now; + TAILQ_HEAD(, script_msg) sm_rdnss_head = + TAILQ_HEAD_INITIALIZER(sm_rdnss_head); + TAILQ_HEAD(, script_msg) sm_dnssl_head = + TAILQ_HEAD_INITIALIZER(sm_dnssl_head); + int dcount, dlen; + + dcount = 0; + dlen = strlen(resstr_sh_prefix) + strlen(resstr_nl); + gettimeofday(&now, NULL); + TAILQ_FOREACH(rao, &ifi->ifi_ra_opt, rao_next) { + switch (rao->rao_type) { + case ND_OPT_RDNSS: + if (timercmp(&now, &rao->rao_expire, >)) { + warnmsg(LOG_INFO, __func__, + "expired rdnss entry: %s", + (char *)rao->rao_msg); + break; + } + ELM_MALLOC(smp1, continue); + ELM_MALLOC(smp2, goto free1); + ELM_MALLOC(smp3, goto free2); + smp1->sm_msg = resstr_ns_prefix; + TAILQ_INSERT_TAIL(&sm_rdnss_head, smp1, sm_next); + smp2->sm_msg = rao->rao_msg; + TAILQ_INSERT_TAIL(&sm_rdnss_head, smp2, sm_next); + smp3->sm_msg = resstr_nl; + TAILQ_INSERT_TAIL(&sm_rdnss_head, smp3, sm_next); + + break; + case ND_OPT_DNSSL: + if (timercmp(&now, &rao->rao_expire, >)) { + warnmsg(LOG_INFO, __func__, + "expired dnssl entry: %s", + (char *)rao->rao_msg); + break; + } + dcount++; + /* Check resolv.conf(5) restrictions. */ + if (dcount > 6) { + warnmsg(LOG_INFO, __func__, + "dnssl entry exceeding maximum count (%d>6)" + ": %s", dcount, (char *)rao->rao_msg); + break; + } + if (256 < dlen + strlen(rao->rao_msg) + + strlen(resstr_sp)) { + warnmsg(LOG_INFO, __func__, + "dnssl entry exceeding maximum length " + "(>256): %s", (char *)rao->rao_msg); + break; + } + ELM_MALLOC(smp1, continue); + ELM_MALLOC(smp2, goto free1); + if (TAILQ_EMPTY(&sm_dnssl_head)) { + ELM_MALLOC(smp3, goto free2); + smp3->sm_msg = resstr_sh_prefix; + TAILQ_INSERT_TAIL(&sm_dnssl_head, smp3, + sm_next); + } + smp1->sm_msg = rao->rao_msg; + TAILQ_INSERT_TAIL(&sm_dnssl_head, smp1, sm_next); + smp2->sm_msg = resstr_sp; + TAILQ_INSERT_TAIL(&sm_dnssl_head, smp2, sm_next); + dlen += strlen(rao->rao_msg) + strlen(resstr_sp); + break; + default: + break; + } + continue; +free2: + free(smp2); +free1: + free(smp1); + } + /* Add \n for DNSSL list. */ + if (!TAILQ_EMPTY(&sm_dnssl_head)) { + ELM_MALLOC(smp1, goto ra_opt_handler_freeit); + smp1->sm_msg = resstr_nl; + TAILQ_INSERT_TAIL(&sm_dnssl_head, smp1, sm_next); + } + TAILQ_CONCAT(&sm_rdnss_head, &sm_dnssl_head, sm_next); + + if (!TAILQ_EMPTY(&sm_rdnss_head)) + CALL_SCRIPT(RESADD, &sm_rdnss_head); + else + CALL_SCRIPT(RESDEL, NULL); + +ra_opt_handler_freeit: + /* Clear script message queue. */ + if (!TAILQ_EMPTY(&sm_rdnss_head)) { + while ((smp1 = TAILQ_FIRST(&sm_rdnss_head)) != NULL) { + TAILQ_REMOVE(&sm_rdnss_head, smp1, sm_next); + free(smp1); + } + } + return (0); +} + static void -call_script(char *scriptpath, char *ifname) +call_script(const int argc, const char *const argv[], void *head) { + const char *scriptpath; + int fd[2]; + int error; pid_t pid, wpid; + TAILQ_HEAD(, script_msg) *sm_head; - if (scriptpath == NULL) + if ((scriptpath = argv[0]) == NULL) return; + fd[0] = fd[1] = -1; + sm_head = head; + if (sm_head != NULL && !TAILQ_EMPTY(sm_head)) { + error = pipe(fd); + if (error) { + warnmsg(LOG_ERR, __func__, + "failed to create a pipe: %s", strerror(errno)); + return; + } + } + /* launch the script */ pid = fork(); if (pid < 0) { warnmsg(LOG_ERR, __func__, "failed to fork: %s", strerror(errno)); return; - } else if (pid) { + } else if (pid) { /* parent */ int wstatus; + if (fd[0] != -1) { /* Send message to the child if any. */ + ssize_t len; + struct script_msg *smp; + + close(fd[0]); + TAILQ_FOREACH(smp, sm_head, sm_next) { + len = strlen(smp->sm_msg); + warnmsg(LOG_DEBUG, __func__, + "write to child = %s(%zd)", + smp->sm_msg, len); + if (write(fd[1], smp->sm_msg, len) != len) { + warnmsg(LOG_ERR, __func__, + "write to child failed: %s", + strerror(errno)); + break; + } + } + close(fd[1]); + } do { wpid = wait(&wstatus); } while (wpid != pid && wpid > 0); if (wpid < 0) warnmsg(LOG_ERR, __func__, "wait: %s", strerror(errno)); - else { + else warnmsg(LOG_DEBUG, __func__, "script \"%s\" terminated", scriptpath); - } - } else { - char *argv[3]; - int fd; + } else { /* child */ + int nullfd; + char **_argv; - argv[0] = scriptpath; - argv[1] = ifname; - argv[2] = NULL; - if (safefile(scriptpath)) { warnmsg(LOG_ERR, __func__, "script \"%s\" cannot be executed safely", scriptpath); exit(1); } - - if ((fd = open("/dev/null", O_RDWR)) != -1) { - dup2(fd, STDIN_FILENO); - dup2(fd, STDOUT_FILENO); - dup2(fd, STDERR_FILENO); - if (fd > STDERR_FILENO) - close(fd); + nullfd = open("/dev/null", O_RDWR); + if (nullfd < 0) { + warnmsg(LOG_ERR, __func__, + "open /dev/null: %s", strerror(errno)); + exit(1); } + if (fd[0] != -1) { /* Receive message from STDIN if any. */ + close(fd[1]); + if (fd[0] != STDIN_FILENO) { + /* Connect a pipe read-end to child's STDIN. */ + if (dup2(fd[0], STDIN_FILENO) != STDIN_FILENO) { + warnmsg(LOG_ERR, __func__, + "dup2 STDIN: %s", strerror(errno)); + exit(1); + } + close(fd[0]); + } + } else + dup2(nullfd, STDIN_FILENO); + + dup2(nullfd, STDOUT_FILENO); + dup2(nullfd, STDERR_FILENO); + if (nullfd > STDERR_FILENO) + close(nullfd); - execv(scriptpath, argv); - + _argv = malloc(sizeof(*_argv) * argc); + if (_argv == NULL) { + warnmsg(LOG_ERR, __func__, + "malloc: %s", strerror(errno)); + exit(1); + } + memcpy(_argv, argv, (size_t)argc); + execv(scriptpath, (char *const *)_argv); warnmsg(LOG_ERR, __func__, "child: exec failed: %s", strerror(errno)); - exit(0); + exit(1); } return; } static int safefile(const char *path) { struct stat s; uid_t myuid; /* no setuid */ if (getuid() != geteuid()) { warnmsg(LOG_NOTICE, __func__, "setuid'ed execution not allowed\n"); return (-1); } if (lstat(path, &s) != 0) { warnmsg(LOG_NOTICE, __func__, "lstat failed: %s", strerror(errno)); return (-1); } /* the file must be owned by the running uid */ myuid = getuid(); if (s.st_uid != myuid) { warnmsg(LOG_NOTICE, __func__, "%s has invalid owner uid\n", path); return (-1); } switch (s.st_mode & S_IFMT) { case S_IFREG: break; default: warnmsg(LOG_NOTICE, __func__, "%s is an invalid file type 0x%o\n", path, (s.st_mode & S_IFMT)); return (-1); } return (0); +} + +/* Decode domain name label encoding in RFC 1035 Section 3.1 */ +static size_t +dname_labeldec(char *dst, size_t dlen, const char *src) +{ + size_t len; + const char *src_origin; + const char *src_last; + const char *dst_origin; + + src_origin = src; + src_last = strchr(src, '\0'); + dst_origin = dst; + memset(dst, '\0', dlen); + while (src && (len = (uint8_t)(*src++) & 0x3f) && + (src + len) <= src_last) { + if (dst != dst_origin) + *dst++ = '.'; + warnmsg(LOG_DEBUG, __func__, "labellen = %zd", len); + memcpy(dst, src, len); + src += len; + dst += len; + } + *dst = '\0'; + + /* + * XXX validate that domain name only contains valid characters + * for two reasons: 1) correctness, 2) we do not want to pass + * possible malicious, unescaped characters like `` to a script + * or program that could be exploited that way. + */ + + return (src - src_origin); } Index: projects/largeSMP/usr.sbin/rtsold/rtsold.8 =================================================================== --- projects/largeSMP/usr.sbin/rtsold/rtsold.8 (revision 222782) +++ projects/largeSMP/usr.sbin/rtsold/rtsold.8 (revision 222783) @@ -1,272 +1,297 @@ .\" $KAME: rtsold.8,v 1.20 2003/04/11 12:46:12 jinmei Exp $ .\" .\" 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. .\" .\" $FreeBSD$ .\" -.Dd September 2, 2009 +.Dd May 28, 2011 .Dt RTSOLD 8 .Os .\" .Sh NAME .Nm rtsold , rtsol .Nd router solicitation daemon .\" .Sh SYNOPSIS .Nm .Op Fl dDfFm1 .Op Fl O Ar script-name +.Op Fl P Ar pidfile +.Op Fl R Ar script-name .Ar interface ... .Nm .Op Fl dDfFm1 .Op Fl O Ar script-name +.Op Fl P Ar pidfile +.Op Fl R Ar script-name .Fl a .Nm rtsol -.Op Fl dDF +.Op Fl dD .Op Fl O Ar script-name +.Op Fl R Ar script-name .Ar interface ... .Nm rtsol .Op Fl dD .Op Fl O Ar script-name +.Op Fl R Ar script-name .Fl a .\" .Sh DESCRIPTION .Nm is the daemon program to send ICMPv6 Router Solicitation messages on the specified interfaces. If a node (re)attaches to a link, .Nm sends some Router Solicitations on the link destined to the link-local scope all-routers multicast address to discover new routers and to get non link-local addresses. .Pp .Nm should be used on IPv6 hosts .Pq non-router nodes only. .Pp If you invoke the program as .Nm rtsol , it will transmit probes from the specified .Ar interface , without becoming a daemon. In other words, .Nm rtsol behaves as .Do .Nm .Fl f1 .Ar interfaces .Dc . .Pp Specifically, .Nm sends at most 3 Router Solicitations on an interface after one of the following events: .Pp .Bl -bullet -compact .It Just after invocation of .Nm daemon. .It The interface is up after a temporary interface failure. .Nm detects such failures by periodically probing to see if the status of the interface is active or not. Note that some network cards and drivers do not allow the extraction of link state. In such cases, .Nm cannot detect the change of the interface status. .It Every 60 seconds if the .Fl m option is specified and the .Nm daemon cannot get the interface status. This feature does not conform to the IPv6 neighbor discovery specification, but is provided for mobile stations. The default interval for router advertisements, which is on the order of 10 minutes, is slightly long for mobile stations. This feature is provided for such stations so that they can find new routers as soon as possible when they attach to another link. .El .Lp Once .Nm has sent a Router Solicitation, and has received a valid Router Advertisement, it refrains from sending additional solicitations on that interface, until the next time one of the above events occurs. .Lp When sending a Router Solicitation on an interface, .Nm includes a Source Link-layer address option if the interface has a link-layer address. .Lp .Nm manages a per-interface parameter to detect if a separate protocol is needed for configuration parameters other than host's addresses. At the invocation time, the flag is FALSE, and becomes TRUE when the daemon receives a router advertisement with the OtherConfig flag being set. A script file can be specified to deal with the case .Pq see below . When .Nm start resending router solicitation messages by one of the conditions events, the daemon resets the parameter because the event may indicate a change on the attached link. .Pp Upon receipt of signal .Dv SIGUSR1 , .Nm will dump the current internal state into .Pa /var/run/rtsold.dump . .\" .Pp The options are as follows: .Bl -tag -width indent .It Fl a Autoprobe outgoing interface. .Nm will try to find a non-loopback, non-point-to-point, IPv6-capable interface. If .Nm finds multiple interfaces, .Nm will exit with error. .\" .It Fl d Enable debugging. .It Fl D Enable more debugging including the printing of internal timer information. .It Fl f Prevent .Nm from becoming a daemon (foreground mode). Warning messages are generated to standard error instead of .Xr syslog 3 . .It Fl F Explicitly configure the kernel to accept Router Advertisements and disable IPv6 forwarding. These settings are required for proper .Nm operation. Without this option, the current settings will be obeyed; if they are incompatible with proper operation, warning messages will be generated, but Router Solicitations will still be sent. The settings may be changed manually with .Xr sysctl 8 and .Xr ifconfig 8 . .It Fl m Enable mobility support. If this option is specified, .Nm sends probing packets to default routers that have advertised Router Advertisements when the node (re)attaches to an interface. Moreover, if the option is specified, .Nm periodically sends Router Solicitation on an interface that does not support .Dv SIOCGIFMEDIA ioctl. .It Fl 1 Perform only one probe. Transmit Router Solicitation packets until at least one valid Router Advertisement packet has arrived on each .Ar interface , then exit. .It Fl O Ar script-name Specifies a supplement script file to handle the Other Configuration flag of the router advertisement. When the flag changes from FALSE to TRUE, .Nm will invoke .Ar script-name with a single argument of the receiving interface name, expecting the script will then start a protocol for the other configuration. .Ar script-name must be the absolute path from root to the script file, be a regular file, and be created by the same owner who runs .Nm . +.It Fl P Ar pidfile +Writes the process ID of +.Nm +to +.Pa pidfile +instead of the default PID file +.Pa /var/run/rtsold.pid . +.It Fl R Ar script-name +Specifies a script to run when router advertisment options +.Dv RDNSS Pq Recursive DNS Server +or +.Dv DNSSL Pq DNS Search List +are encountered. +The information of DNS servers and DNS search domains will be sent to +standard input of this script. +The +.Xr resolvconf 8 +script is used by default. .El .Sh FILES .Bl -tag -width /var/run/rtsold.dump -compact .It Pa /var/run/rtsold.pid the pid of the currently running .Nm . .It Pa /var/run/rtsold.dump dumps internal state on. .El .\" .Sh EXIT STATUS .Ex -std .\" .Sh SEE ALSO +.Xr resolvconf 8 , .Xr rtadvd 8 , .Xr sysctl 8 .\" .Sh HISTORY The .Nm command is based on the .Nm rtsol command, which first appeared in WIDE/KAME IPv6 protocol stack kit. .Nm rtsol is now integrated into .Xr rtsold 8 . .\" .Sh BUGS In some operating systems, when a PCMCIA network card is removed and reinserted, the corresponding interface index is changed. However, .Nm assumes such changes will not occur, and always uses the index that it got at invocation. As a result, .Nm may not work if you reinsert a network card. In such a case, .Nm should be killed and restarted. .Pp The IPv6 autoconfiguration specification assumes a single-interface host. You may see kernel error messages if you try to autoconfigure a host with multiple interfaces. Also, it seems contradictory for .Nm to accept multiple .Ar interface arguments. Index: projects/largeSMP/usr.sbin/rtsold/rtsold.c =================================================================== --- projects/largeSMP/usr.sbin/rtsold/rtsold.c (revision 222782) +++ projects/largeSMP/usr.sbin/rtsold/rtsold.c (revision 222783) @@ -1,887 +1,919 @@ /* $KAME: rtsold.c,v 1.67 2003/05/17 18:16:15 itojun Exp $ */ /* * 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. * * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef HAVE_POLL_H #include #endif #include "rtsold.h" +#define RTSOL_DUMPFILE "/var/run/rtsold.dump"; +#define RTSOL_PIDFILE "/var/run/rtsold.pid"; + struct ifinfo *iflist; struct timeval tm_max = {0x7fffffff, 0x7fffffff}; static int log_upto = 999; static int fflag = 0; int Fflag = 0; /* force setting sysctl parameters */ int aflag = 0; int dflag = 0; -char *otherconf_script; +const char *otherconf_script; +const char *resolvconf_script = "/sbin/resolvconf"; /* protocol constants */ #define MAX_RTR_SOLICITATION_DELAY 1 /* second */ #define RTR_SOLICITATION_INTERVAL 4 /* seconds */ #define MAX_RTR_SOLICITATIONS 3 /* times */ /* * implementation dependent constants in seconds * XXX: should be configurable */ #define PROBE_INTERVAL 60 -int main(int, char **); - /* static variables and functions */ static int mobile_node = 0; +static const char *pidfilename = RTSOL_PIDFILE; + #ifndef SMALL static int do_dump; -static const char *dumpfilename = "/var/run/rtsold.dump"; /* XXX: should be configurable */ +static const char *dumpfilename = RTSOL_DUMPFILE; #endif -#if 1 -static const char *pidfilename = "/var/run/rtsold.pid"; /* should be configurable */ -#endif #if 0 static int ifreconfig(char *); #endif + static int make_packet(struct ifinfo *); static struct timeval *rtsol_check_timer(void); #ifndef SMALL static void rtsold_set_dump_file(int); #endif -static void usage(char *); +static void usage(void); int main(int argc, char **argv) { int s, ch, once = 0; struct timeval *timeout; - char *argv0; const char *opts; #ifdef HAVE_POLL_H struct pollfd set[2]; #else fd_set *fdsetp, *selectfdp; int fdmasks; int maxfd; #endif int rtsock; - /* - * Initialization - */ - argv0 = argv[0]; - - /* get option */ - if (argv0 && argv0[strlen(argv0) - 1] != 'd') { - fflag = 1; - once = 1; - opts = "adDFO:"; - } else - opts = "adDfFm1O:"; - +#ifndef SMALL + /* rtsold */ + opts = "adDfFm1O:P:R:"; +#else + /* rtsol */ + opts = "adDFO:P:R:"; + fflag = 1; + once = 1; +#endif while ((ch = getopt(argc, argv, opts)) != -1) { switch (ch) { case 'a': aflag = 1; break; case 'd': dflag = 1; break; case 'D': dflag = 2; break; case 'f': fflag = 1; break; case 'F': Fflag = 1; break; case 'm': mobile_node = 1; break; case '1': once = 1; break; case 'O': otherconf_script = optarg; break; + case 'P': + pidfilename = optarg; + break; + case 'R': + resolvconf_script = optarg; + break; default: - usage(argv0); - /*NOTREACHED*/ + usage(); + exit(1); } } argc -= optind; argv += optind; if ((!aflag && argc == 0) || (aflag && argc != 0)) { - usage(argv0); - /*NOTREACHED*/ + usage(); + exit(1); } /* set log level */ if (dflag == 0) log_upto = LOG_NOTICE; if (!fflag) { char *ident; - ident = strrchr(argv0, '/'); + ident = strrchr(argv[0], '/'); if (!ident) - ident = argv0; + ident = argv[0]; else ident++; openlog(ident, LOG_NDELAY|LOG_PID, LOG_DAEMON); if (log_upto >= 0) setlogmask(LOG_UPTO(log_upto)); } if (otherconf_script && *otherconf_script != '/') { errx(1, "configuration script (%s) must be an absolute path", otherconf_script); } - + if (resolvconf_script && *resolvconf_script != '/') { + errx(1, "configuration script (%s) must be an absolute path", + resolvconf_script); + } + if (pidfilename && *pidfilename != '/') { + errx(1, "pid filename (%s) must be an absolute path", + pidfilename); + } #ifndef HAVE_ARC4RANDOM /* random value initialization */ srandom((u_long)time(NULL)); #endif if (Fflag) { setinet6sysctl(IPV6CTL_FORWARDING, 0); } else { /* warn if forwarding is up */ if (getinet6sysctl(IPV6CTL_FORWARDING)) warnx("kernel is configured as a router, not a host"); } #ifndef SMALL /* initialization to dump internal status to a file */ signal(SIGUSR1, rtsold_set_dump_file); #endif if (!fflag) daemon(0, 0); /* act as a daemon */ /* * Open a socket for sending RS and receiving RA. * This should be done before calling ifinit(), since the function * uses the socket. */ if ((s = sockopen()) < 0) { warnmsg(LOG_ERR, __func__, "failed to open a socket"); exit(1); - /*NOTREACHED*/ } #ifdef HAVE_POLL_H set[0].fd = s; set[0].events = POLLIN; #else maxfd = s; #endif #ifdef HAVE_POLL_H set[1].fd = -1; #endif if ((rtsock = rtsock_open()) < 0) { warnmsg(LOG_ERR, __func__, "failed to open a socket"); exit(1); - /*NOTREACHED*/ } #ifdef HAVE_POLL_H set[1].fd = rtsock; set[1].events = POLLIN; #else if (rtsock > maxfd) maxfd = rtsock; #endif #ifndef HAVE_POLL_H fdmasks = howmany(maxfd + 1, NFDBITS) * sizeof(fd_mask); if ((fdsetp = malloc(fdmasks)) == NULL) { - err(1, "malloc"); - /*NOTREACHED*/ + warnmsg(LOG_ERR, __func__, "malloc"); + exit(1); } if ((selectfdp = malloc(fdmasks)) == NULL) { - err(1, "malloc"); - /*NOTREACHED*/ + warnmsg(LOG_ERR, __func__, "malloc"); + exit(1); } #endif /* configuration per interface */ if (ifinit()) { warnmsg(LOG_ERR, __func__, "failed to initialize interfaces"); exit(1); - /*NOTREACHED*/ } if (aflag) argv = autoifprobe(); while (argv && *argv) { if (ifconfig(*argv)) { warnmsg(LOG_ERR, __func__, "failed to initialize %s", *argv); exit(1); - /*NOTREACHED*/ } argv++; } /* setup for probing default routers */ if (probe_init()) { warnmsg(LOG_ERR, __func__, "failed to setup for probing routers"); exit(1); /*NOTREACHED*/ } -#if 1 /* dump the current pid */ if (!once) { pid_t pid = getpid(); FILE *fp; if ((fp = fopen(pidfilename, "w")) == NULL) warnmsg(LOG_ERR, __func__, "failed to open a pid log file(%s): %s", pidfilename, strerror(errno)); else { fprintf(fp, "%d\n", pid); fclose(fp); } } -#endif - #ifndef HAVE_POLL_H memset(fdsetp, 0, fdmasks); FD_SET(s, fdsetp); FD_SET(rtsock, fdsetp); #endif while (1) { /* main loop */ int e; #ifndef HAVE_POLL_H memcpy(selectfdp, fdsetp, fdmasks); #endif #ifndef SMALL if (do_dump) { /* SIGUSR1 */ do_dump = 0; rtsold_dump_file(dumpfilename); } #endif timeout = rtsol_check_timer(); if (once) { struct ifinfo *ifi; /* if we have no timeout, we are done (or failed) */ if (timeout == NULL) break; /* if all interfaces have got RA packet, we are done */ - for (ifi = iflist; ifi; ifi = ifi->next) { + TAILQ_FOREACH(ifi, &ifinfo_head, ifi_next) { if (ifi->state != IFS_DOWN && ifi->racnt == 0) break; } if (ifi == NULL) break; } #ifdef HAVE_POLL_H e = poll(set, 2, timeout ? (timeout->tv_sec * 1000 + timeout->tv_usec / 1000) : INFTIM); #else e = select(maxfd + 1, selectfdp, NULL, NULL, timeout); #endif if (e < 1) { if (e < 0 && errno != EINTR) { warnmsg(LOG_ERR, __func__, "select: %s", strerror(errno)); } continue; } /* packet reception */ #ifdef HAVE_POLL_H if (set[1].revents & POLLIN) #else if (FD_ISSET(rtsock, selectfdp)) #endif rtsock_input(rtsock); #ifdef HAVE_POLL_H if (set[0].revents & POLLIN) #else if (FD_ISSET(s, selectfdp)) #endif rtsol_input(s); } /* NOTREACHED */ - return 0; + return (0); } int ifconfig(char *ifname) { - struct ifinfo *ifinfo; + struct ifinfo *ifi; struct sockaddr_dl *sdl; int flags; if ((sdl = if_nametosdl(ifname)) == NULL) { warnmsg(LOG_ERR, __func__, "failed to get link layer information for %s", ifname); - return(-1); + return (-1); } if (find_ifinfo(sdl->sdl_index)) { warnmsg(LOG_ERR, __func__, "interface %s was already configured", ifname); free(sdl); - return(-1); + return (-1); } - if ((ifinfo = malloc(sizeof(*ifinfo))) == NULL) { + if ((ifi = malloc(sizeof(*ifi))) == NULL) { warnmsg(LOG_ERR, __func__, "memory allocation failed"); free(sdl); - return(-1); + return (-1); } - memset(ifinfo, 0, sizeof(*ifinfo)); - ifinfo->sdl = sdl; + memset(ifi, 0, sizeof(*ifi)); + ifi->sdl = sdl; - strlcpy(ifinfo->ifname, ifname, sizeof(ifinfo->ifname)); + strlcpy(ifi->ifname, ifname, sizeof(ifi->ifname)); /* construct a router solicitation message */ - if (make_packet(ifinfo)) + if (make_packet(ifi)) goto bad; /* set link ID of this interface. */ #ifdef HAVE_SCOPELIB - if (inet_zoneid(AF_INET6, 2, ifname, &ifinfo->linkid)) + if (inet_zoneid(AF_INET6, 2, ifname, &ifi->linkid)) goto bad; #else /* XXX: assume interface IDs as link IDs */ - ifinfo->linkid = ifinfo->sdl->sdl_index; + ifi->linkid = ifi->sdl->sdl_index; #endif /* * check if the interface is available. * also check if SIOCGIFMEDIA ioctl is OK on the interface. */ - ifinfo->mediareqok = 1; - ifinfo->active = interface_status(ifinfo); - if (!ifinfo->mediareqok) { + ifi->mediareqok = 1; + ifi->active = interface_status(ifi); + if (!ifi->mediareqok) { /* * probe routers periodically even if the link status * does not change. */ - ifinfo->probeinterval = PROBE_INTERVAL; + ifi->probeinterval = PROBE_INTERVAL; } /* activate interface: interface_up returns 0 on success */ - flags = interface_up(ifinfo->ifname); + flags = interface_up(ifi->ifname); if (flags == 0) - ifinfo->state = IFS_DELAY; + ifi->state = IFS_DELAY; else if (flags == IFS_TENTATIVE) - ifinfo->state = IFS_TENTATIVE; + ifi->state = IFS_TENTATIVE; else - ifinfo->state = IFS_DOWN; + ifi->state = IFS_DOWN; - rtsol_timer_update(ifinfo); + rtsol_timer_update(ifi); - /* link into chain */ - if (iflist) - ifinfo->next = iflist; - iflist = ifinfo; + TAILQ_INSERT_TAIL(&ifinfo_head, ifi, ifi_next); + return (0); - return(0); - bad: - free(ifinfo->sdl); - free(ifinfo); - return(-1); + free(ifi->sdl); + free(ifi); + return (-1); } void iflist_init(void) { - struct ifinfo *ifi, *next; + struct ifinfo *ifi; - for (ifi = iflist; ifi; ifi = next) { - next = ifi->next; - if (ifi->sdl) + while ((ifi = TAILQ_FIRST(&ifinfo_head)) != NULL) { + TAILQ_REMOVE(&ifinfo_head, ifi, ifi_next); + if (ifi->sdl != NULL) free(ifi->sdl); - if (ifi->rs_data) + if (ifi->rs_data != NULL) free(ifi->rs_data); free(ifi); - iflist = NULL; } } #if 0 static int ifreconfig(char *ifname) { struct ifinfo *ifi, *prev; int rv; prev = NULL; - for (ifi = iflist; ifi; ifi = ifi->next) { + TAILQ_FOREACH(ifi, &ifinfo_head, ifi_next) { if (strncmp(ifi->ifname, ifname, sizeof(ifi->ifname)) == 0) break; prev = ifi; } prev->next = ifi->next; rv = ifconfig(ifname); /* reclaim it after ifconfig() in case ifname is pointer inside ifi */ if (ifi->rs_data) free(ifi->rs_data); free(ifi->sdl); free(ifi); - return rv; + + return (rv); } #endif struct ifinfo * find_ifinfo(int ifindex) { struct ifinfo *ifi; - for (ifi = iflist; ifi; ifi = ifi->next) + TAILQ_FOREACH(ifi, &ifinfo_head, ifi_next) { if (ifi->sdl->sdl_index == ifindex) - return(ifi); - return(NULL); + return (ifi); + } + return (NULL); } static int -make_packet(struct ifinfo *ifinfo) +make_packet(struct ifinfo *ifi) { size_t packlen = sizeof(struct nd_router_solicit), lladdroptlen = 0; struct nd_router_solicit *rs; char *buf; - if ((lladdroptlen = lladdropt_length(ifinfo->sdl)) == 0) { + if ((lladdroptlen = lladdropt_length(ifi->sdl)) == 0) { warnmsg(LOG_INFO, __func__, "link-layer address option has null length" - " on %s. Treat as not included.", ifinfo->ifname); + " on %s. Treat as not included.", ifi->ifname); } packlen += lladdroptlen; - ifinfo->rs_datalen = packlen; + ifi->rs_datalen = packlen; /* allocate buffer */ if ((buf = malloc(packlen)) == NULL) { warnmsg(LOG_ERR, __func__, - "memory allocation failed for %s", ifinfo->ifname); - return(-1); + "memory allocation failed for %s", ifi->ifname); + return (-1); } - ifinfo->rs_data = buf; + ifi->rs_data = buf; /* fill in the message */ rs = (struct nd_router_solicit *)buf; rs->nd_rs_type = ND_ROUTER_SOLICIT; rs->nd_rs_code = 0; rs->nd_rs_cksum = 0; rs->nd_rs_reserved = 0; buf += sizeof(*rs); /* fill in source link-layer address option */ if (lladdroptlen) - lladdropt_fill(ifinfo->sdl, (struct nd_opt_hdr *)buf); + lladdropt_fill(ifi->sdl, (struct nd_opt_hdr *)buf); - return(0); + return (0); } static struct timeval * rtsol_check_timer(void) { static struct timeval returnval; struct timeval now, rtsol_timer; - struct ifinfo *ifinfo; + struct ifinfo *ifi; + struct ra_opt *rao; int flags; gettimeofday(&now, NULL); rtsol_timer = tm_max; - for (ifinfo = iflist; ifinfo; ifinfo = ifinfo->next) { - if (timercmp(&ifinfo->expire, &now, <=)) { + TAILQ_FOREACH(ifi, &ifinfo_head, ifi_next) { + if (timercmp(&ifi->expire, &now, <=)) { if (dflag > 1) warnmsg(LOG_DEBUG, __func__, "timer expiration on %s, " - "state = %d", ifinfo->ifname, - ifinfo->state); + "state = %d", ifi->ifname, + ifi->state); - switch (ifinfo->state) { + /* Remove all RA options. */ + while ((rao = TAILQ_FIRST(&ifi->ifi_ra_opt)) != NULL) { + if (rao->rao_msg != NULL) + free(rao->rao_msg); + TAILQ_REMOVE(&ifi->ifi_ra_opt, rao, rao_next); + free(rao); + } + switch (ifi->state) { case IFS_DOWN: case IFS_TENTATIVE: /* interface_up returns 0 on success */ - flags = interface_up(ifinfo->ifname); + flags = interface_up(ifi->ifname); if (flags == 0) - ifinfo->state = IFS_DELAY; + ifi->state = IFS_DELAY; else if (flags == IFS_TENTATIVE) - ifinfo->state = IFS_TENTATIVE; + ifi->state = IFS_TENTATIVE; else - ifinfo->state = IFS_DOWN; + ifi->state = IFS_DOWN; break; case IFS_IDLE: { - int oldstatus = ifinfo->active; + int oldstatus = ifi->active; int probe = 0; - ifinfo->active = interface_status(ifinfo); + ifi->active = interface_status(ifi); - if (oldstatus != ifinfo->active) { + if (oldstatus != ifi->active) { warnmsg(LOG_DEBUG, __func__, "%s status is changed" " from %d to %d", - ifinfo->ifname, - oldstatus, ifinfo->active); + ifi->ifname, + oldstatus, ifi->active); probe = 1; - ifinfo->state = IFS_DELAY; - } else if (ifinfo->probeinterval && - (ifinfo->probetimer -= - ifinfo->timer.tv_sec) <= 0) { + ifi->state = IFS_DELAY; + } else if (ifi->probeinterval && + (ifi->probetimer -= + ifi->timer.tv_sec) <= 0) { /* probe timer expired */ - ifinfo->probetimer = - ifinfo->probeinterval; + ifi->probetimer = + ifi->probeinterval; probe = 1; - ifinfo->state = IFS_PROBE; + ifi->state = IFS_PROBE; } /* * If we need a probe, clear the previous * status wrt the "other" configuration. */ if (probe) - ifinfo->otherconfig = 0; + ifi->otherconfig = 0; if (probe && mobile_node) - defrouter_probe(ifinfo); + defrouter_probe(ifi); break; } case IFS_DELAY: - ifinfo->state = IFS_PROBE; - sendpacket(ifinfo); + ifi->state = IFS_PROBE; + sendpacket(ifi); break; case IFS_PROBE: - if (ifinfo->probes < MAX_RTR_SOLICITATIONS) - sendpacket(ifinfo); + if (ifi->probes < MAX_RTR_SOLICITATIONS) + sendpacket(ifi); else { warnmsg(LOG_INFO, __func__, "No answer after sending %d RSs", - ifinfo->probes); - ifinfo->probes = 0; - ifinfo->state = IFS_IDLE; + ifi->probes); + ifi->probes = 0; + ifi->state = IFS_IDLE; } break; } - rtsol_timer_update(ifinfo); - } + rtsol_timer_update(ifi); + } else { + /* Expiration check for RA options. */ + struct ra_opt *rao_tmp; + int expire = 0; - if (timercmp(&ifinfo->expire, &rtsol_timer, <)) - rtsol_timer = ifinfo->expire; + TAILQ_FOREACH_SAFE(rao, &ifi->ifi_ra_opt, rao_next, rao_tmp) { + warnmsg(LOG_DEBUG, __func__, + "RA expiration timer: " + "type=%d, msg=%s, timer=%ld:%08ld", + rao->rao_type, (char *)rao->rao_msg, + (long)rao->rao_expire.tv_sec, + (long)rao->rao_expire.tv_usec); + if (timercmp(&now, &rao->rao_expire, >=)) { + warnmsg(LOG_DEBUG, __func__, + "RA expiration timer: expired."); + TAILQ_REMOVE(&ifi->ifi_ra_opt, rao, rao_next); + expire = 1; + } + } + if (expire) + ra_opt_handler(ifi); + } + if (timercmp(&ifi->expire, &rtsol_timer, <)) + rtsol_timer = ifi->expire; } if (timercmp(&rtsol_timer, &tm_max, ==)) { warnmsg(LOG_DEBUG, __func__, "there is no timer"); - return(NULL); + return (NULL); } else if (timercmp(&rtsol_timer, &now, <)) /* this may occur when the interval is too small */ returnval.tv_sec = returnval.tv_usec = 0; else timersub(&rtsol_timer, &now, &returnval); if (dflag > 1) warnmsg(LOG_DEBUG, __func__, "New timer is %ld:%08ld", (long)returnval.tv_sec, (long)returnval.tv_usec); - return(&returnval); + return (&returnval); } void -rtsol_timer_update(struct ifinfo *ifinfo) +rtsol_timer_update(struct ifinfo *ifi) { #define MILLION 1000000 #define DADRETRY 10 /* XXX: adhoc */ long interval; struct timeval now; - bzero(&ifinfo->timer, sizeof(ifinfo->timer)); + bzero(&ifi->timer, sizeof(ifi->timer)); - switch (ifinfo->state) { + switch (ifi->state) { case IFS_DOWN: case IFS_TENTATIVE: - if (++ifinfo->dadcount > DADRETRY) { - ifinfo->dadcount = 0; - ifinfo->timer.tv_sec = PROBE_INTERVAL; + if (++ifi->dadcount > DADRETRY) { + ifi->dadcount = 0; + ifi->timer.tv_sec = PROBE_INTERVAL; } else - ifinfo->timer.tv_sec = 1; + ifi->timer.tv_sec = 1; break; case IFS_IDLE: if (mobile_node) { /* XXX should be configurable */ - ifinfo->timer.tv_sec = 3; + ifi->timer.tv_sec = 3; } else - ifinfo->timer = tm_max; /* stop timer(valid?) */ + ifi->timer = tm_max; /* stop timer(valid?) */ break; case IFS_DELAY: #ifndef HAVE_ARC4RANDOM interval = random() % (MAX_RTR_SOLICITATION_DELAY * MILLION); #else interval = arc4random_uniform(MAX_RTR_SOLICITATION_DELAY * MILLION); #endif - ifinfo->timer.tv_sec = interval / MILLION; - ifinfo->timer.tv_usec = interval % MILLION; + ifi->timer.tv_sec = interval / MILLION; + ifi->timer.tv_usec = interval % MILLION; break; case IFS_PROBE: - if (ifinfo->probes < MAX_RTR_SOLICITATIONS) - ifinfo->timer.tv_sec = RTR_SOLICITATION_INTERVAL; + if (ifi->probes < MAX_RTR_SOLICITATIONS) + ifi->timer.tv_sec = RTR_SOLICITATION_INTERVAL; else { /* * After sending MAX_RTR_SOLICITATIONS solicitations, * we're just waiting for possible replies; there * will be no more solicitation. Thus, we change * the timer value to MAX_RTR_SOLICITATION_DELAY based * on RFC 2461, Section 6.3.7. */ - ifinfo->timer.tv_sec = MAX_RTR_SOLICITATION_DELAY; + ifi->timer.tv_sec = MAX_RTR_SOLICITATION_DELAY; } break; default: warnmsg(LOG_ERR, __func__, "illegal interface state(%d) on %s", - ifinfo->state, ifinfo->ifname); + ifi->state, ifi->ifname); return; } /* reset the timer */ - if (timercmp(&ifinfo->timer, &tm_max, ==)) { - ifinfo->expire = tm_max; + if (timercmp(&ifi->timer, &tm_max, ==)) { + ifi->expire = tm_max; warnmsg(LOG_DEBUG, __func__, - "stop timer for %s", ifinfo->ifname); + "stop timer for %s", ifi->ifname); } else { gettimeofday(&now, NULL); - timeradd(&now, &ifinfo->timer, &ifinfo->expire); + timeradd(&now, &ifi->timer, &ifi->expire); if (dflag > 1) warnmsg(LOG_DEBUG, __func__, - "set timer for %s to %d:%d", ifinfo->ifname, - (int)ifinfo->timer.tv_sec, - (int)ifinfo->timer.tv_usec); + "set timer for %s to %d:%d", ifi->ifname, + (int)ifi->timer.tv_sec, + (int)ifi->timer.tv_usec); } #undef MILLION } /* timer related utility functions */ #define MILLION 1000000 #ifndef SMALL static void rtsold_set_dump_file(int sig __unused) { do_dump = 1; } #endif static void -usage(char *progname) +usage(void) { - if (progname && progname[strlen(progname) - 1] != 'd') { - fprintf(stderr, "usage: rtsol [-dDF] interfaces...\n"); - fprintf(stderr, "usage: rtsol [-dDF] -a\n"); - } else { - fprintf(stderr, "usage: rtsold [-adDfFm1] interfaces...\n"); - fprintf(stderr, "usage: rtsold [-dDfFm1] -a\n"); - } - exit(1); +#ifndef SMALL + fprintf(stderr, "usage: rtsold [-adDfFm1] [-O script-name] " + "[-P pidfile] [-R script-name] interfaces...\n"); + fprintf(stderr, "usage: rtsold [-dDfFm1] [-O script-name] " + "[-P pidfile] [-R script-name] -a\n"); +#else + fprintf(stderr, "usage: rtsol [-dDF] [-O script-name] " + "[-P pidfile] [-R script-name] interfaces...\n"); + fprintf(stderr, "usage: rtsol [-dDF] [-O script-name] " + "[-P pidfile] [-R script-name] -a\n"); +#endif } void -#if __STDC__ warnmsg(int priority, const char *func, const char *msg, ...) -#else -warnmsg(priority, func, msg, va_alist) - int priority; - const char *func; - const char *msg; - va_dcl -#endif { va_list ap; char buf[BUFSIZ]; va_start(ap, msg); if (fflag) { if (priority <= log_upto) { (void)vfprintf(stderr, msg, ap); (void)fprintf(stderr, "\n"); } } else { snprintf(buf, sizeof(buf), "<%s> %s", func, msg); msg = buf; vsyslog(priority, msg, ap); } va_end(ap); } /* * return a list of interfaces which is suitable to sending an RS. */ char ** autoifprobe(void) { static char **argv = NULL; static int n = 0; char **a; int s = 0, i, found; struct ifaddrs *ifap, *ifa, *target; struct in6_ndireq nd; /* initialize */ while (n--) free(argv[n]); if (argv) { free(argv); argv = NULL; } n = 0; if (getifaddrs(&ifap) != 0) - return NULL; + return (NULL); if (!Fflag && (s = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) { - err(1, "socket"); - /* NOTREACHED */ + warnmsg(LOG_ERR, __func__, "socket"); + exit(1); } target = NULL; /* find an ethernet */ for (ifa = ifap; ifa; ifa = ifa->ifa_next) { if ((ifa->ifa_flags & IFF_UP) == 0) continue; if ((ifa->ifa_flags & IFF_POINTOPOINT) != 0) continue; if ((ifa->ifa_flags & IFF_LOOPBACK) != 0) continue; if ((ifa->ifa_flags & IFF_MULTICAST) == 0) continue; if (ifa->ifa_addr->sa_family != AF_INET6) continue; found = 0; for (i = 0; i < n; i++) { if (strcmp(argv[i], ifa->ifa_name) == 0) { found++; break; } } if (found) continue; /* * Skip the interfaces which IPv6 and/or accepting RA * is disabled. */ if (!Fflag) { memset(&nd, 0, sizeof(nd)); strlcpy(nd.ifname, ifa->ifa_name, sizeof(nd.ifname)); if (ioctl(s, SIOCGIFINFO_IN6, (caddr_t)&nd) < 0) { - err(1, "ioctl(SIOCGIFINFO_IN6)"); - /* NOTREACHED */ + warnmsg(LOG_ERR, __func__, + "ioctl(SIOCGIFINFO_IN6)"); + exit(1); } if ((nd.ndi.flags & ND6_IFF_IFDISABLED)) continue; if (!(nd.ndi.flags & ND6_IFF_ACCEPT_RTADV)) continue; } /* if we find multiple candidates, just warn. */ if (n != 0 && dflag > 1) - warnx("multiple interfaces found"); + warnmsg(LOG_WARNING, __func__, + "multiple interfaces found"); a = (char **)realloc(argv, (n + 1) * sizeof(char **)); - if (a == NULL) - err(1, "realloc"); + if (a == NULL) { + warnmsg(LOG_ERR, __func__, "realloc"); + exit(1); + } argv = a; argv[n] = strdup(ifa->ifa_name); - if (!argv[n]) - err(1, "malloc"); + if (!argv[n]) { + warnmsg(LOG_ERR, __func__, "malloc"); + exit(1); + } n++; } if (n) { a = (char **)realloc(argv, (n + 1) * sizeof(char **)); - if (a == NULL) - err(1, "realloc"); + if (a == NULL) { + warnmsg(LOG_ERR, __func__, "realloc"); + exit(1); + } argv = a; argv[n] = NULL; if (dflag > 0) { for (i = 0; i < n; i++) - warnx("probing %s", argv[i]); + warnmsg(LOG_WARNING, __func__, "probing %s", + argv[i]); } } if (!Fflag) close(s); freeifaddrs(ifap); - return argv; + return (argv); } Index: projects/largeSMP/usr.sbin/rtsold/rtsold.h =================================================================== --- projects/largeSMP/usr.sbin/rtsold/rtsold.h (revision 222782) +++ projects/largeSMP/usr.sbin/rtsold/rtsold.h (revision 222783) @@ -1,104 +1,151 @@ /* $KAME: rtsold.h,v 1.19 2003/04/16 09:48:15 itojun Exp $ */ /* * 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. * * $FreeBSD$ */ +struct script_msg { + TAILQ_ENTRY(script_msg) sm_next; + + char *sm_msg; +}; + +struct ra_opt { + TAILQ_ENTRY(ra_opt) rao_next; + + u_int8_t rao_type; + struct timeval rao_expire; + size_t rao_len; + void *rao_msg; +}; + struct ifinfo { - struct ifinfo *next; /* pointer to the next interface */ + TAILQ_ENTRY(ifinfo) ifi_next; /* pointer to the next interface */ struct sockaddr_dl *sdl; /* link-layer address */ char ifname[IF_NAMESIZE]; /* interface name */ u_int32_t linkid; /* link ID of this interface */ int active; /* interface status */ int probeinterval; /* interval of probe timer (if necessary) */ int probetimer; /* rest of probe timer */ int mediareqok; /* wheter the IF supports SIOCGIFMEDIA */ int otherconfig; /* need a separate protocol for the "other" * configuration */ int state; int probes; int dadcount; struct timeval timer; struct timeval expire; int errors; /* # of errors we've got - detect wedge */ int racnt; /* total # of valid RAs it have got */ size_t rs_datalen; u_char *rs_data; + + TAILQ_HEAD(, ra_opt) ifi_ra_opt; }; /* per interface status */ #define IFS_IDLE 0 #define IFS_DELAY 1 #define IFS_PROBE 2 #define IFS_DOWN 3 #define IFS_TENTATIVE 4 +/* Interface list */ +extern TAILQ_HEAD(ifinfo_head_t, ifinfo) ifinfo_head; + +/* + * RFC 3542 API deprecates IPV6_PKTINFO in favor of + * IPV6_RECVPKTINFO + */ +#ifndef IPV6_RECVPKTINFO +#ifdef IPV6_PKTINFO +#define IPV6_RECVPKTINFO IPV6_PKTINFO +#endif +#endif +/* + * RFC 3542 API deprecates IPV6_HOPLIMIT in favor of + * IPV6_RECVHOPLIMIT + */ +#ifndef IPV6_RECVHOPLIMIT +#ifdef IPV6_HOPLIMIT +#define IPV6_RECVHOPLIMIT IPV6_HOPLIMIT +#endif +#endif + +#ifndef IN6ADDR_LINKLOCAL_ALLROUTERS_INIT +#define IN6ADDR_LINKLOCAL_ALLROUTERS_INIT \ + {{{ 0xff, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, \ + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02 }}} +#endif + /* rtsold.c */ extern struct timeval tm_max; extern int dflag; extern int aflag; extern int Fflag; -extern char *otherconf_script; +extern const char *otherconf_script; +extern const char *resolvconf_script; extern int ifconfig(char *); extern void iflist_init(void); struct ifinfo *find_ifinfo(int); void rtsol_timer_update(struct ifinfo *); extern void warnmsg(int, const char *, const char *, ...) __attribute__((__format__(__printf__, 3, 4))); extern char **autoifprobe(void); +extern int ra_opt_handler(struct ifinfo *); /* if.c */ extern int ifinit(void); extern int interface_up(char *); extern int interface_status(struct ifinfo *); extern int lladdropt_length(struct sockaddr_dl *); extern void lladdropt_fill(struct sockaddr_dl *, struct nd_opt_hdr *); extern struct sockaddr_dl *if_nametosdl(char *); extern int getinet6sysctl(int); extern int setinet6sysctl(int, int); /* rtsol.c */ extern int sockopen(void); extern void sendpacket(struct ifinfo *); extern void rtsol_input(int); /* probe.c */ extern int probe_init(void); extern void defrouter_probe(struct ifinfo *); /* dump.c */ extern void rtsold_dump_file(const char *); /* rtsock.c */ extern int rtsock_open(void); extern int rtsock_input(int); Index: projects/largeSMP/usr.sbin/tcpdrop/tcpdrop.c =================================================================== --- projects/largeSMP/usr.sbin/tcpdrop/tcpdrop.c (revision 222782) +++ projects/largeSMP/usr.sbin/tcpdrop/tcpdrop.c (revision 222783) @@ -1,322 +1,323 @@ /* $OpenBSD: tcpdrop.c,v 1.4 2004/05/22 23:55:22 deraadt Exp $ */ /*- * Copyright (c) 2009 Juli Mallett * Copyright (c) 2004 Markus Friedl * * Permission to use, copy, modify, and distribute this software for any * purpose with or without fee is hereby granted, provided that the above * copyright notice and this permission notice appear in all copies. * * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ #include __FBSDID("$FreeBSD$"); +#include #include #include #include #include -#include +#include #include #define TCPSTATES #include #include #include #include #include #include #include #include #include #define TCPDROP_FOREIGN 0 #define TCPDROP_LOCAL 1 struct host_service { char hs_host[NI_MAXHOST]; char hs_service[NI_MAXSERV]; }; static bool tcpdrop_list_commands = false; static struct xinpgen *getxpcblist(const char *); static void sockinfo(const struct sockaddr *, struct host_service *); static bool tcpdrop(const struct sockaddr *, const struct sockaddr *); static bool tcpdropall(void); static bool tcpdropbyname(const char *, const char *, const char *, const char *); static bool tcpdropconn(const struct in_conninfo *); static void usage(void); /* * Drop a tcp connection. */ int main(int argc, char *argv[]) { bool dropall; int ch; dropall = false; while ((ch = getopt(argc, argv, "al")) != -1) { switch (ch) { case 'a': dropall = true; break; case 'l': tcpdrop_list_commands = true; break; default: usage(); } } argc -= optind; argv += optind; if (dropall) { if (argc != 0) usage(); if (!tcpdropall()) exit(1); exit(0); } if (argc != 4 || tcpdrop_list_commands) usage(); if (!tcpdropbyname(argv[0], argv[1], argv[2], argv[3])) exit(1); exit(0); } static struct xinpgen * getxpcblist(const char *name) { struct xinpgen *xinp; size_t len; int rv; len = 0; rv = sysctlbyname(name, NULL, &len, NULL, 0); if (rv == -1) err(1, "sysctlbyname %s", name); if (len == 0) errx(1, "%s is empty", name); xinp = malloc(len); if (xinp == NULL) errx(1, "malloc failed"); rv = sysctlbyname(name, xinp, &len, NULL, 0); if (rv == -1) err(1, "sysctlbyname %s", name); return (xinp); } static void sockinfo(const struct sockaddr *sa, struct host_service *hs) { static const int flags = NI_NUMERICHOST | NI_NUMERICSERV; int rv; rv = getnameinfo(sa, sa->sa_len, hs->hs_host, sizeof hs->hs_host, hs->hs_service, sizeof hs->hs_service, flags); if (rv == -1) err(1, "getnameinfo"); } static bool tcpdrop(const struct sockaddr *lsa, const struct sockaddr *fsa) { struct host_service local, foreign; struct sockaddr_storage addrs[2]; int rv; memcpy(&addrs[TCPDROP_FOREIGN], fsa, fsa->sa_len); memcpy(&addrs[TCPDROP_LOCAL], lsa, lsa->sa_len); sockinfo(lsa, &local); sockinfo(fsa, &foreign); if (tcpdrop_list_commands) { printf("tcpdrop %s %s %s %s\n", local.hs_host, local.hs_service, foreign.hs_host, foreign.hs_service); return (true); } rv = sysctlbyname("net.inet.tcp.drop", NULL, NULL, &addrs, sizeof addrs); if (rv == -1) { warn("%s %s %s %s", local.hs_host, local.hs_service, foreign.hs_host, foreign.hs_service); return (false); } printf("%s %s %s %s: dropped\n", local.hs_host, local.hs_service, foreign.hs_host, foreign.hs_service); return (true); } static bool tcpdropall(void) { struct xinpgen *head, *xinp; struct xtcpcb *xpcb; struct tcpcb *tp; struct inpcb *inp; bool ok; ok = true; head = getxpcblist("net.inet.tcp.pcblist"); #define XINP_NEXT(xinp) \ ((struct xinpgen *)((uintptr_t)(xinp) + (xinp)->xig_len)) for (xinp = XINP_NEXT(head); xinp->xig_len > sizeof *xinp; xinp = XINP_NEXT(xinp)) { xpcb = (struct xtcpcb *)xinp; tp = &xpcb->xt_tp; inp = &xpcb->xt_inp; /* * XXX * Check protocol, support just v4 or v6, etc. */ /* Ignore PCBs which were freed during copyout. */ if (inp->inp_gencnt > head->xig_gen) continue; /* Skip listening sockets. */ if (tp->t_state == TCPS_LISTEN) continue; if (!tcpdropconn(&inp->inp_inc)) ok = false; } free(head); return (ok); } static bool tcpdropbyname(const char *lhost, const char *lport, const char *fhost, const char *fport) { static const struct addrinfo hints = { /* * Look for streams in all domains. */ .ai_family = AF_UNSPEC, .ai_socktype = SOCK_STREAM, }; struct addrinfo *ail, *local, *aif, *foreign; int error; bool ok, infamily; error = getaddrinfo(lhost, lport, &hints, &local); if (error != 0) errx(1, "getaddrinfo: %s port %s: %s", lhost, lport, gai_strerror(error)); error = getaddrinfo(fhost, fport, &hints, &foreign); if (error != 0) { freeaddrinfo(local); /* XXX gratuitous */ errx(1, "getaddrinfo: %s port %s: %s", lhost, lport, gai_strerror(error)); } ok = true; infamily = false; /* * Try every combination of local and foreign address pairs. */ for (ail = local; ail != NULL; ail = ail->ai_next) { for (aif = foreign; aif != NULL; aif = aif->ai_next) { if (ail->ai_family != aif->ai_family) continue; infamily = true; if (!tcpdrop(ail->ai_addr, aif->ai_addr)) ok = false; } } if (!infamily) { warnx("%s %s %s %s: different address families", lhost, lport, fhost, fport); ok = false; } freeaddrinfo(local); freeaddrinfo(foreign); return (ok); } static bool tcpdropconn(const struct in_conninfo *inc) { struct sockaddr *local, *foreign; struct sockaddr_in6 sin6[2]; struct sockaddr_in sin4[2]; if ((inc->inc_flags & INC_ISIPV6) != 0) { memset(sin6, 0, sizeof sin6); sin6[TCPDROP_LOCAL].sin6_len = sizeof sin6[TCPDROP_LOCAL]; sin6[TCPDROP_LOCAL].sin6_family = AF_INET6; sin6[TCPDROP_LOCAL].sin6_port = inc->inc_lport; memcpy(&sin6[TCPDROP_LOCAL].sin6_addr, &inc->inc6_laddr, sizeof inc->inc6_laddr); local = (struct sockaddr *)&sin6[TCPDROP_LOCAL]; sin6[TCPDROP_FOREIGN].sin6_len = sizeof sin6[TCPDROP_FOREIGN]; sin6[TCPDROP_FOREIGN].sin6_family = AF_INET6; sin6[TCPDROP_FOREIGN].sin6_port = inc->inc_fport; memcpy(&sin6[TCPDROP_FOREIGN].sin6_addr, &inc->inc6_faddr, sizeof inc->inc6_faddr); foreign = (struct sockaddr *)&sin6[TCPDROP_FOREIGN]; } else { memset(&sin4[TCPDROP_LOCAL], 0, sizeof sin4[TCPDROP_LOCAL]); sin4[TCPDROP_LOCAL].sin_len = sizeof sin4[TCPDROP_LOCAL]; sin4[TCPDROP_LOCAL].sin_family = AF_INET; sin4[TCPDROP_LOCAL].sin_port = inc->inc_lport; memcpy(&sin4[TCPDROP_LOCAL].sin_addr, &inc->inc_laddr, sizeof inc->inc_laddr); local = (struct sockaddr *)&sin4[TCPDROP_LOCAL]; sin4[TCPDROP_FOREIGN].sin_len = sizeof sin4[TCPDROP_FOREIGN]; sin4[TCPDROP_FOREIGN].sin_family = AF_INET; sin4[TCPDROP_FOREIGN].sin_port = inc->inc_fport; memcpy(&sin4[TCPDROP_FOREIGN].sin_addr, &inc->inc_faddr, sizeof inc->inc_faddr); foreign = (struct sockaddr *)&sin4[TCPDROP_FOREIGN]; } return (tcpdrop(local, foreign)); } static void usage(void) { fprintf(stderr, "usage: tcpdrop local-address local-port foreign-address foreign-port\n" " tcpdrop [-l] -a\n"); exit(1); } Index: projects/largeSMP/usr.sbin/zic =================================================================== --- projects/largeSMP/usr.sbin/zic (revision 222782) +++ projects/largeSMP/usr.sbin/zic (revision 222783) Property changes on: projects/largeSMP/usr.sbin/zic ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/usr.sbin/zic:r222698-222782 Index: projects/largeSMP =================================================================== --- projects/largeSMP (revision 222782) +++ projects/largeSMP (revision 222783) Property changes on: projects/largeSMP ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head:r222698-222782